ML_ST_tf_v2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "782e8f80-78a9-4e0b-9384-896f365185cf",
   "metadata": {},
   "source": [
    "# Stack Traces Project (TensorFlow)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8e1deae-812d-4af1-bdfc-bfe26372e904",
   "metadata": {
    "tags": []
   },
   "source": [
    "## Importing the necessary libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ce22b98e-8698-4c40-8ca0-245c1f8adf6d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import re\n",
    "from pathlib import Path\n",
    "from typing import Tuple, List, Optional, Union, Tuple\n",
    "from pprint import pprint\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.ticker import AutoMinorLocator\n",
    "from matplotlib import gridspec\n",
    "import string\n",
    "import seaborn as sns\n",
    "import datetime\n",
    "from itertools import permutations\n",
    "import heapq\n",
    "import csv\n",
    "import pickle\n",
    "import matplotlib.ticker as mtick\n",
    "import collections\n",
    "import os\n",
    "from tabulate import tabulate\n",
    "from openpyxl import load_workbook\n",
    "\n",
    "# pd.options.display.max_colwidth = 500\n",
    "# pd.options.display.max_columns = None\n",
    "# pd.options.display.max_rows = None\n",
    "pd.set_option('display.max_rows', 500)\n",
    "pd.set_option('display.max_columns', 500)\n",
    "pd.set_option('display.width', 1000)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "67fbe948-c066-476e-af2f-dff2bfc97539",
   "metadata": {},
   "source": [
    "## Setting up the project environment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "a33fcc34-2081-48b3-8cb7-817b87d3fe7c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# .csv configuration\n",
    "encoding = \"utf-8\"\n",
    "delimiter = None\n",
    "working_directory_path = \"../\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "491e6678-04ef-4571-8215-1f9b599e7af2",
   "metadata": {},
   "source": [
    "## Loading the question dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a4c9e24a-26aa-454f-a424-5a667d25817f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"question_tag.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_sotorrent = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "1401237b-1dd9-49ba-ad91-50929602772b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(163194, 11)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sotorrent.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "c616d9ac-5921-465c-aaf3-f2578e4864c3",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_sotorrent['Source'] = 'old'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b4d97d6d-cb88-42ca-a458-1d6528a4d591",
   "metadata": {},
   "source": [
    "Update the DB"
   ]
  },
  {
   "cell_type": "raw",
   "id": "24aea9fe-cdef-4c84-9175-8a744ae5ecbe",
   "metadata": {},
   "source": [
    "Ref: https://data.stackexchange.com/stackoverflow/query/edit/1769158\n",
    "\n",
    "SELECT \n",
    "      Id, \n",
    "      PostTypeId, \n",
    "      AcceptedAnswerId, \n",
    "      CreationDate, \n",
    "      ViewCount, \n",
    "      AnswerCount, \n",
    "      CommentCount, \n",
    "      Score,\n",
    "      Title,\n",
    "      Body,\n",
    "      Tags\n",
    "FROM Posts \n",
    "WHERE (tags like '%tensorflow%' OR\n",
    "      tags LIKE '%pytorch%' OR\n",
    "      tags LIKE '%scikit-learn%' OR\n",
    "      tags LIKE '%keras%' OR\n",
    "      tags LIKE '%nltk%' OR\n",
    "      tags LIKE '%huggingface%' OR\n",
    "      tags LIKE '%spark-ml%') AND\n",
    "      creationDate > '2021-01-25 00:00:00' AND\n",
    "      PostTypeId=1\n",
    "ORDER BY [creationDate] DESC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "48e538f7-e9e5-4dab-ac00-7c179f4892ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"proj1_update_qst.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_remained = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "cd5b336b-1fe2-4bc6-a99c-837f3f672283",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(49918, 11)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_remained.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "300bd414-ed1e-4bff-80f8-c3f6894c1295",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_remained['source'] = 'new'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0df7df97-57ab-49ad-af11-5ab770e837eb",
   "metadata": {},
   "source": [
    "Concat two data frames"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "96f69893-e004-497d-a407-0dc3cc3bb0dc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(213112, 13)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.concat([df_sotorrent, df_remained])\n",
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "8311d4f5-9748-465f-b6f4-023dde556a42",
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop_duplicates([\"Id\", \"PostTypeId\", \"AcceptedAnswerId\", \"ViewCount\", \"AnswerCount\", \"CommentCount\", \"Score\", \"Title\", 'Body', 'Tags'], ignore_index=False, inplace=True)\n",
    "df = df.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "bc3f69f7-987e-4076-b1e1-0e345ff45fb2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The total number of records (w/o considering specific tag set) : 171114\n"
     ]
    }
   ],
   "source": [
    "print(\"The total number of records (w/o considering specific tag set) :\", df.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "01b0b401-bbb9-4f7c-9411-e4c8f14a6c50",
   "metadata": {},
   "source": [
    "## Preparing and updating the all answers"
   ]
  },
  {
   "cell_type": "raw",
   "id": "23446440-e58a-4ac4-90ca-e4fa662975a4",
   "metadata": {},
   "source": [
    "with cte1 as (\n",
    "SELECT\n",
    "        p.Id, \n",
    "        p.PostTypeId, \n",
    "        p.AcceptedAnswerId, \n",
    "        p.CreationDate, \n",
    "        p.ViewCount, \n",
    "        p.AnswerCount, \n",
    "        p.CommentCount, \n",
    "        p.Score,\n",
    "        p.Title,\n",
    "        p.Body,\n",
    "        p.Tags,\n",
    "        p.parentid\n",
    "\n",
    "FROM Posts as p\n",
    "inner join posts q \n",
    "        on q.id = coalesce(p.parentid, p.id) -- only questions have tags\n",
    "\n",
    "WHERE p.CreationDate > '2019-01-01 00:00:00' AND      -- After this time\n",
    "      p.PostTypeId=1 AND                     -- Just answers\n",
    "      p.Tags LIKE '%tensorflow%' AND\n",
    "       -- p.Tags LIKE '%pytorch%' OR\n",
    "       -- p.Tags LIKE '%scikit-learn%' OR\n",
    "       -- p.Tags LIKE '%keras%' OR\n",
    "       -- p.Tags LIKE '%nltk%' OR\n",
    "       -- p.Tags LIKE '%huggingface%' AND\n",
    "       -- p.Tags LIKE '%spark-ml%')\n",
    "       p.Tags LIKE '%python%'\n",
    ")\n",
    "select\n",
    "        r.parentid as Q_id,\n",
    "        r.parentid as ParentId,\n",
    "        r.Id, \n",
    "        --r.AcceptedAnswerId, \n",
    "        --r.PostTypeId,\n",
    "        --cte1.Id, \n",
    "        --cte1.PostTypeId,\n",
    "        r.CreationDate\n",
    "        --cte1.CreationDate as Q_CreationDate\n",
    "        --r.Score,\n",
    "        --cte1.AcceptedAnswerId\n",
    "        --cte1.ViewCount, \n",
    "        --cte1.AnswerCount,\n",
    "        --cte1.CommentCount, \n",
    "        --cte1.Score\n",
    "        --cte1.Title,\n",
    "        --cte1.Body,\n",
    "        --cte1.Tags,\n",
    "from posts r \n",
    "inner join cte1\n",
    "       on cte1.id = r.parentid\n",
    "where\n",
    "      r.PostTypeId = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "3b71967b-0cbd-4003-8a2a-450723a8c126",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_part_1.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_ans_p1 = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "ef9dfb1a-f5a5-448f-8aef-f0b616defeb5",
   "metadata": {},
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_part_2.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_ans_p2 = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "d32200d2-8942-4efb-a7e3-fd26bb32de2f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54873, 4)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_tf_ans_all = pd.concat([df_ans_p1, df_ans_p2])\n",
    "df_tf_ans_all.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8e10dc18-a8ad-4e8e-b4fe-8e0312eb11cb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54873, 4)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_tf_ans_all.drop_duplicates([\"Q_id\", \"ParentId\", \"Id\", \"CreationDate\"], ignore_index=False, inplace=True)\n",
    "df_tf_ans_all = df_tf_ans_all.reset_index(drop=True)\n",
    "df_tf_ans_all.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "b7078a79-7572-46ed-a7b6-c96cb39190a7",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_all.csv\"\n",
    "path = Path(pure_data)\n",
    "df_tf_ans_all.to_csv(path, encoding=encoding, index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2e408693",
   "metadata": {},
   "source": [
    "## Answering the first research question (RQ1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db9a6e6e-46f4-432b-81c6-a5f646d6dd08",
   "metadata": {},
   "source": [
    "### Filtering the specific tags"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "eae1d276-5b01-482b-b244-cd89bef86bd7",
   "metadata": {},
   "outputs": [],
   "source": [
    "def tag_filter(pref_tags: List, tags: str) -> bool:\n",
    "    regex = \"\"\n",
    "    for tag in pref_tags:\n",
    "        regex += '(?=.*\\\\b'+ tag +'([+-]?([0-9]*[.])?[0-9]*)\\\\b)'\n",
    "    regex = r\"^\" + regex + \".*$\"\n",
    "    tags = tags.strip().lower()\n",
    "    match_result = re.match(regex, tags, re.MULTILINE | re.IGNORECASE)\n",
    "    if match_result is None:\n",
    "        return 0\n",
    "    else:\n",
    "        return 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "6b05dfaf-8a60-4a86-b341-b78d16a32226",
   "metadata": {},
   "outputs": [],
   "source": [
    "tags = [\"tensorflow\", \"python\"]\n",
    "df['HasPreferableTags'] = df['Tags'].apply(lambda row_tags: tag_filter(tags, row_tags))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "40b740ac-83e0-45f1-a2be-e8825609ee76",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags = df[df['HasPreferableTags']==True]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "3a2d3ed2-b450-4151-8856-352d7e8a2463",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags = df_w_tf_tags.drop(['HasPreferableTags'], axis='columns')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "32db8e0e-ad52-40ee-aab0-229a72267751",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54887, 13)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_w_tf_tags.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0ebb0623-d6d8-4261-90f3-c3eace22e119",
   "metadata": {},
   "source": [
    "We found there are some duplocated rows in our DB, so we eleminate those based on some columns "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "eb8b23cc-bffd-4fc3-a8c6-1fae9798336f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags.drop_duplicates([\"Id\", \"PostTypeId\", \"AcceptedAnswerId\", \"ViewCount\", \"AnswerCount\", \"CommentCount\", \"Score\", \"Title\"], ignore_index=False, inplace=True)\n",
    "df_w_tf_tags = df_w_tf_tags.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "b79a315e-28a4-4ae4-8fb7-dcfb510a291e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54887, 13)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_w_tf_tags.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "77c96c84-dedb-4f7f-a464-f8c8e8607970",
   "metadata": {
    "tags": []
   },
   "source": [
    "The new dataset that has a specific tag/s is reduced to:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "15fdfa92-2d73-4f6c-af4d-4a814b7d11d6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The orginal DB:  171114\n",
      "The new DB (sp):  54887\n",
      "The difference is:  116227\n"
     ]
    }
   ],
   "source": [
    "print(\"The orginal DB: \", df.shape[0])\n",
    "print(\"The new DB (sp): \", df_w_tf_tags.shape[0])\n",
    "print(\"The difference is: \", df.shape[0] - df_w_tf_tags.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "2050397c-ca76-4c95-843d-71fbd9940991",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Id</th>\n",
       "      <th>PostTypeId</th>\n",
       "      <th>AcceptedAnswerId</th>\n",
       "      <th>CreationDate</th>\n",
       "      <th>ViewCount</th>\n",
       "      <th>AnswerCount</th>\n",
       "      <th>CommentCount</th>\n",
       "      <th>Score</th>\n",
       "      <th>Title</th>\n",
       "      <th>Body</th>\n",
       "      <th>Tags</th>\n",
       "      <th>Source</th>\n",
       "      <th>source</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>43574928</td>\n",
       "      <td>1</td>\n",
       "      <td>43575038.0</td>\n",
       "      <td>2017-04-23 18:30:08</td>\n",
       "      <td>1118</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>How to use custom/non-default tf.Graph in Tens...</td>\n",
       "      <td>&lt;p&gt;I'm very new to &lt;code&gt;Tensorflow&lt;/code&gt; and...</td>\n",
       "      <td>&lt;python&gt;&lt;session&gt;&lt;graph&gt;&lt;parallel-processing&gt;&lt;...</td>\n",
       "      <td>old</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>52019226</td>\n",
       "      <td>1</td>\n",
       "      <td>52019631.0</td>\n",
       "      <td>2018-08-25 16:32:19</td>\n",
       "      <td>312</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>Keras w/ Tensorflow intermediate layer extract...</td>\n",
       "      <td>&lt;p&gt;I am currently trying to leverage an interm...</td>\n",
       "      <td>&lt;python&gt;&lt;tensorflow&gt;&lt;keras&gt;&lt;deep-learning&gt;&lt;bat...</td>\n",
       "      <td>old</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>48065685</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2018-01-02 18:02:51</td>\n",
       "      <td>461</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>Issue with train.next_batch in Tensorflow</td>\n",
       "      <td>&lt;p&gt;I am currently facing an issue with the nex...</td>\n",
       "      <td>&lt;python&gt;&lt;python-3.x&gt;&lt;tensorflow&gt;&lt;machine-learn...</td>\n",
       "      <td>old</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>64070871</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2020-09-25 20:19:10</td>\n",
       "      <td>45</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>How can i reshape TFRecord dataset to train RN...</td>\n",
       "      <td>&lt;p&gt;I am trying to feed an RNN model with &lt;code...</td>\n",
       "      <td>&lt;python&gt;&lt;tensorflow&gt;&lt;machine-learning&gt;&lt;keras&gt;&lt;...</td>\n",
       "      <td>old</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>45011900</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2017-07-10 12:21:24</td>\n",
       "      <td>2936</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>Loading a trained model from Python to C++ in ...</td>\n",
       "      <td>&lt;p&gt;I'm looking to run a basic fully-connected ...</td>\n",
       "      <td>&lt;python&gt;&lt;c++&gt;&lt;tensorflow&gt;</td>\n",
       "      <td>old</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Id  PostTypeId  AcceptedAnswerId         CreationDate  ViewCount  AnswerCount  CommentCount  Score                                              Title                                               Body                                               Tags Source source\n",
       "0  43574928           1        43575038.0  2017-04-23 18:30:08       1118            1             0      3  How to use custom/non-default tf.Graph in Tens...  <p>I'm very new to <code>Tensorflow</code> and...  <python><session><graph><parallel-processing><...    old    NaN\n",
       "1  52019226           1        52019631.0  2018-08-25 16:32:19        312            1             2      0  Keras w/ Tensorflow intermediate layer extract...  <p>I am currently trying to leverage an interm...  <python><tensorflow><keras><deep-learning><bat...    old    NaN\n",
       "2  48065685           1               NaN  2018-01-02 18:02:51        461            0             2      0          Issue with train.next_batch in Tensorflow  <p>I am currently facing an issue with the nex...  <python><python-3.x><tensorflow><machine-learn...    old    NaN\n",
       "3  64070871           1               NaN  2020-09-25 20:19:10         45            0             1      1  How can i reshape TFRecord dataset to train RN...  <p>I am trying to feed an RNN model with <code...  <python><tensorflow><machine-learning><keras><...    old    NaN\n",
       "4  45011900           1               NaN  2017-07-10 12:21:24       2936            0             7      3  Loading a trained model from Python to C++ in ...  <p>I'm looking to run a basic fully-connected ...                          <python><c++><tensorflow>    old    NaN"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_w_tf_tags.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "28990026-83cc-4af5-94f6-5b2a75eb41f7",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Extracting the code parts from body"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "23057f26-0779-428d-885f-7b166ea7ebca",
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_code_blocks(body: str, _id: int) -> List:\n",
    "    global index\n",
    "    # regex = r\"<pre><code>((.*?)|(\\n)*)*(<\\/code><\\/pre>|</pre></code>)\"\n",
    "    # regex = r\"(<pre>|(<pre((.*?)|(\\n)*)*><code>))((.*?)|(\\n)*)*(<\\/code><\\/pre>|</pre></code>|</pre>)\"\n",
    "    regex = r\"<pre(><code>|>|(((.*?)|(\\n)*)><code>)|((.*?)|(\\n)*)>)((.*?)|(\\n)*)*(<\\/code><\\/pre>|</pre></code>|</pre>)\"\n",
    "    matches = re.finditer(regex, body, re.MULTILINE | re.IGNORECASE)\n",
    "    result = []\n",
    "    \n",
    "    try:\n",
    "        for matchNum, match in enumerate(matches, start=1):\n",
    "            code = match.group()\n",
    "            code = re.sub('<pre(><code>|>|(((.*?)|(\\n)*)><code>)|((.*?)|(\\n)*)>)', '', code)\n",
    "            code = code.replace(\"<pre><code>\", \"\")\n",
    "            code = code.replace(\"</pre></code>\", \"\")\n",
    "            code = code.replace(\"</code></pre>\", \"\")\n",
    "            result.append(code)\n",
    "    except:\n",
    "        print(\"\\n Error(1): \", _id)\n",
    "        print(body)\n",
    "        return None\n",
    "    \n",
    "    return result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f87db6b2-b24a-45d5-a7d2-e234bc3f0f73",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_w_tf_tags['Code'] = df_w_tf_tags.apply(lambda row: extract_code_blocks(row.Body, row.Id), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "25534f29-f886-4366-9c74-366cb41becd3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(df_w_tags.iloc[7]['Body'])\n",
    "# print(type(extract_code_blocks(df_w_tags.iloc[7]['Body'])))\n",
    "# print(len(extract_code_blocks(df_w_tags.iloc[7]['Body'])))\n",
    "# print(extract_code_blocks(df_w_tags.iloc[7]['Body'])[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a8b1e27-61be-4359-9b1c-c80828413913",
   "metadata": {},
   "source": [
    "Reset the index of dataframe:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "e700b0aa-2f79-43c5-90e8-f969379e8f5d",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags = df_w_tf_tags.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4bdbd611",
   "metadata": {},
   "source": [
    "#### Finding the number of questions that have a code or not"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "717748e7",
   "metadata": {},
   "outputs": [],
   "source": [
    "count__question_w_code  = 0\n",
    "count__question_wo_code = 0\n",
    "count__num_codes = 0\n",
    "\n",
    "def counting_w_or_wo_code(row_code: List) -> bool:\n",
    "    global count__question_w_code, count__question_wo_code, count__num_codes\n",
    "    \n",
    "    if row_code:\n",
    "        count__question_w_code = count__question_w_code + 1\n",
    "        count__num_codes += len(row_code)\n",
    "        return True\n",
    "    else:\n",
    "        count__question_wo_code = count__question_wo_code + 1\n",
    "        return False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "9a723e24",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags['Has_code'] = df_w_tf_tags['Code'].apply(lambda row_code: counting_w_or_wo_code(row_code))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "d8547f91-ce63-4b34-a349-f13a0815fb18",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "We have 54887 records in our DB, including 45582 posts that have code block in themselve and 9305 w/o any code block.\n",
      "We have 102246 number of code blocks including stacktrace, snippet code, error message.\n"
     ]
    }
   ],
   "source": [
    "print(f\"We have {df_w_tf_tags.shape[0]} records in our DB, including {count__question_w_code} posts that have code block in themselve and {count__question_wo_code} w/o any code block.\")\n",
    "print(f\"We have {count__num_codes} number of code blocks including stacktrace, snippet code, error message.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ebd63ed7-eb63-42e3-b2a5-bc9641a5de68",
   "metadata": {},
   "source": [
    "### Define two lists for storing the information of the questions w or w/o Code Block"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "b52a3709-78cf-4cdc-8692-bffad767bdb4",
   "metadata": {},
   "outputs": [],
   "source": [
    "Question_w_code_info = []\n",
    "Question_wo_code_info = []\n",
    "\n",
    "def counting_w_or_wo_code(row_id: int, \n",
    "                           row_cr: object, \n",
    "                           row_vc: int, \n",
    "                           row_ac: int, \n",
    "                           row_cc: int, \n",
    "                           row_sc: int, \n",
    "                           row_ac_an_id: float, \n",
    "                           has_code: bool) -> None:\n",
    "    \n",
    "    global Question_w_code_info, Question_wo_code_info\n",
    "\n",
    "    if has_code:\n",
    "        Question_w_code_info.append((row_id, row_cr, row_vc, row_ac, row_cc, row_sc, row_ac_an_id))\n",
    "\n",
    "    else:\n",
    "        Question_wo_code_info.append((row_id, row_cr, row_vc, row_ac, row_cc, row_sc, row_ac_an_id))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "144b328e-9cf7-49a2-884f-b46b386b7a95",
   "metadata": {},
   "outputs": [],
   "source": [
    "_ = df_w_tf_tags.apply(lambda row: counting_w_or_wo_code(row.Id, \n",
    "                                                           row.CreationDate, \n",
    "                                                           row.ViewCount,\n",
    "                                                           row.AnswerCount,\n",
    "                                                           row.CommentCount,\n",
    "                                                           row.Score,\n",
    "                                                           row.AcceptedAnswerId,\n",
    "                                                           row.Has_code), axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eac096c8-8728-4cfb-a933-d7b8f61e10f7",
   "metadata": {},
   "source": [
    "Create dfs based on the two lists, Question_w_code_info and Question_wo_code_info and filled their answers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "42393cac-4d55-40c3-91b4-ff5b0c46e46a",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_q_w_code_v1 = pd.DataFrame(columns=['First_ans_time', 'First_acc_ans_time', 'Answers'])\n",
    "\n",
    "df_q_w_code_v1[\"Q_info\"] = Question_w_code_info\n",
    "\n",
    "df_q_w_code_v1[\"Q_id\"]          = df_q_w_code_v1['Q_info'].apply(lambda q_info: q_info[0])\n",
    "df_q_w_code_v1[\"Q_create_time\"] = df_q_w_code_v1['Q_info'].apply(lambda q_info: q_info[1])\n",
    "df_q_w_code_v1[\"View_count\"]    = df_q_w_code_v1['Q_info'].apply(lambda q_info: q_info[2])\n",
    "df_q_w_code_v1[\"Answer_count\"]  = df_q_w_code_v1['Q_info'].apply(lambda q_info: q_info[3])\n",
    "df_q_w_code_v1[\"Comment_count\"] = df_q_w_code_v1['Q_info'].apply(lambda q_info: q_info[4])\n",
    "df_q_w_code_v1[\"Score\"]         = df_q_w_code_v1['Q_info'].apply(lambda q_info: q_info[5])\n",
    "df_q_w_code_v1[\"Accepted_Answer_id\"] = df_q_w_code_v1['Q_info'].apply(lambda q_info: q_info[6])\n",
    "df_q_w_code_v1 = df_q_w_code_v1.drop(['Q_info'], axis='columns')\n",
    "df_q_w_code_v1[\"Q_create_time\"]      = pd.to_datetime(df_q_w_code_v1[\"Q_create_time\"])\n",
    "df_q_w_code_v1[\"First_acc_ans_time\"] = pd.to_datetime(df_q_w_code_v1[\"First_acc_ans_time\"])\n",
    "df_q_w_code_v1[\"First_ans_time\"]     = pd.to_datetime(df_q_w_code_v1[\"First_ans_time\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "9523776b-851e-4ccd-af23-a6d730b393e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Store\n",
    "df_q_w_code_v1[\"Q_id\"].to_csv('../code_output_csv/df_tf_q_w_code.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "raw",
   "id": "6986107d-0813-4c97-842f-9d39e842b4c7",
   "metadata": {},
   "source": [
    "SELECT df_tf_q_w_code.Q_id, all_results.Id, all_results.CreationDate, all_results.ParentId\n",
    "FROM df_tf_q_w_code\n",
    "INNER JOIN all_results\n",
    "ON df_tf_q_w_code.Q_id = all_results.ParentId;"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f234bf31-4ec3-482f-bfa6-97c39e6a55fa",
   "metadata": {},
   "source": [
    "The below csv file comes from diffrent query that introduced on top of the code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "1beff71e-9036-4dbb-9741-8073b69698f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Change 3\n",
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_all.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_night = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "32427453-eb31-4f84-953d-b250075634ae",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_w = pd.merge(df_q_w_code_v1, df_night, how='outer',left_on=['Q_id'], right_on=['ParentId']).reset_index(drop=True)\n",
    "\n",
    "pd_tmp_w[\"Answer_tup\"] = pd_tmp_w.apply(lambda x: (x.Id, x.CreationDate), axis=1)\n",
    "\n",
    "pd_tmp_new_w = pd_tmp_w.groupby([\"Q_id_x\", \"Q_create_time\", \"View_count\", \"Comment_count\", \"Score\", \"Answer_count\", \"Accepted_Answer_id\"], dropna=False)[\"Answer_tup\"].agg(list).reset_index()\n",
    "pd_tmp_new2_w = pd_tmp_new_w[pd_tmp_new_w.Answer_count != 0].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "cfdd6ce2-80be-4c95-817a-b84d3a5894a8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Warning!:  34220532 9.0 [(34220750.0, '2015-12-11 10:01:57'), (42785945.0, '2017-03-14 12:24:30'), (43580452.0, '2017-04-24 05:33:53'), (47272180.0, '2017-11-13 19:37:03'), (48489637.0, '2018-01-28 18:02:58'), (54226708.0, '2019-01-16 23:04:27'), (54467346.0, '2019-01-31 18:49:27'), (56544146.0, '2019-06-11 12:48:48')]\n",
      "Warning!:  34236252 4.0 [(34237210.0, '2015-12-12 06:55:01'), (43713062.0, '2017-05-01 01:19:23'), (47242212.0, '2017-11-11 19:58:05')]\n",
      "Warning!:  35628221 4.0 [(35630537.0, '2016-02-25 14:48:05'), (35631850.0, '2016-02-25 15:42:05'), (35642964.0, '2016-02-26 03:36:17')]\n",
      "Warning!:  37091899 5.0 [(37093142.0, '2016-05-07 20:03:53'), (41514176.0, '2017-01-06 20:54:32'), (52151134.0, '2018-09-03 13:52:01'), (53628537.0, '2018-12-05 09:01:21')]\n",
      "Warning!:  41835789 2.0 [(41836684.0, '2017-01-24 18:48:20')]\n",
      "Warning!:  46071998 2.0 [(46085544.0, '2017-09-06 23:07:24')]\n",
      "Warning!:  47525650 3.0 [(47525722.0, '2017-11-28 07:00:56'), (47525970.0, '2017-11-28 07:17:08')]\n",
      "Warning!:  47874461 2.0 [(51877820.0, '2018-08-16 13:06:27')]\n",
      "Warning!:  48556725 3.0 [(48563927.0, '2018-02-01 13:29:17'), (48658554.0, '2018-02-07 07:54:47')]\n",
      "Warning!:  48979426 4.0 [(56060197.0, '2019-05-09 13:09:32'), (59776666.0, '2020-01-16 19:38:27'), (62719871.0, '2020-07-03 17:07:31')]\n",
      "Warning!:  49237117 5.0 [(49238753.0, '2018-03-12 15:21:56'), (49238987.0, '2018-03-12 15:32:39'), (52917974.0, '2018-10-21 17:24:43'), (55747328.0, '2019-04-18 13:41:51')]\n",
      "Warning!:  50010929 4.0 [(50048350.0, '2018-04-26 16:53:14'), (50161767.0, '2018-05-03 18:32:52'), (55664372.0, '2019-04-13 10:16:30')]\n",
      "Warning!:  50376618 3.0 [(50376904.0, '2018-05-16 17:31:52'), (50378619.0, '2018-05-16 19:29:23')]\n",
      "Warning!:  50571245 2.0 [(50588762.0, '2018-05-29 16:16:23')]\n",
      "Warning!:  50715065 2.0 [(50717667.0, '2018-06-06 10:04:39')]\n",
      "Warning!:  50805706 2.0 [(51268152.0, '2018-07-10 14:43:01')]\n",
      "Warning!:  50895110 3.0 [(51896376.0, '2018-08-17 13:22:59'), (54888253.0, '2019-02-26 14:54:55')]\n",
      "Warning!:  50901790 2.0 [(50901811.0, '2018-06-18 01:48:53')]\n",
      "Warning!:  50977379 2.0 [(50977484.0, '2018-06-21 21:10:27')]\n",
      "Warning!:  51202602 3.0 [(51220960.0, '2018-07-07 07:25:14'), (60828438.0, '2020-03-24 09:43:05')]\n",
      "Warning!:  51225805 3.0 [(51229676.0, '2018-07-08 07:15:06'), (51230655.0, '2018-07-08 09:35:40')]\n",
      "Warning!:  52801571 2.0 [(53005953.0, '2018-10-26 09:45:38')]\n",
      "Warning!:  53366768 3.0 [(53380590.0, '2018-11-19 18:25:07'), (53405554.0, '2018-11-21 05:04:53')]\n",
      "Warning!:  54274107 2.0 [(54274774.0, '2019-01-20 08:30:57')]\n",
      "Warning!:  54805568 2.0 [(nan, nan)]\n",
      "Warning!:  55114425 2.0 [(nan, nan)]\n",
      "Warning!:  55116078 3.0 [(55168007.0, '2019-03-14 16:49:34'), (59809590.0, '2020-01-19 11:30:12')]\n",
      "Warning!:  55302893 3.0 [(55303437.0, '2019-03-22 15:54:44'), (55303491.0, '2019-03-22 15:57:23')]\n",
      "Warning!:  55481393 2.0 [(59642171.0, '2020-01-08 08:41:51')]\n",
      "Warning!:  55522240 3.0 [(55524029.0, '2019-04-04 20:01:20'), (63902982.0, '2020-09-15 13:31:03')]\n",
      "Warning!:  55533413 3.0 [(55543516.0, '2019-04-05 21:20:06'), (55827001.0, '2019-04-24 09:40:38')]\n",
      "Warning!:  55634133 3.0 [(55634686.0, '2019-04-11 14:16:31'), (55634744.0, '2019-04-11 14:19:33')]\n",
      "Warning!:  56460901 2.0 [(62140763.0, '2020-06-01 20:48:20')]\n",
      "Warning!:  56627371 2.0 [(58065919.0, '2019-09-23 15:43:37')]\n",
      "Warning!:  57062456 7.0 [(58726643.0, '2019-11-06 09:15:11'), (58900033.0, '2019-11-17 11:35:22'), (59639212.0, '2020-01-08 03:57:12'), (61002414.0, '2020-04-02 22:50:27'), (62379815.0, '2020-06-15 00:48:20'), (63280924.0, '2020-08-06 09:45:12')]\n",
      "Warning!:  57190095 3.0 [(57192370.0, '2019-07-24 23:12:19'), (57256143.0, '2019-07-29 14:41:48')]\n",
      "Warning!:  57381023 2.0 [(57382011.0, '2019-08-06 18:28:05')]\n",
      "Warning!:  57539273 2.0 [(57546746.0, '2019-08-18 17:14:25')]\n",
      "Warning!:  57542946 2.0 [(57546258.0, '2019-08-18 16:06:13')]\n",
      "Warning!:  58025069 3.0 [(58965019.0, '2019-11-20 23:33:45'), (59798962.0, '2020-01-18 08:56:05')]\n",
      "Warning!:  58270940 2.0 [(58284431.0, '2019-10-08 10:19:25')]\n",
      "Warning!:  58529511 2.0 [(58529805.0, '2019-10-23 19:30:04')]\n",
      "Warning!:  58683675 3.0 [(58684046.0, '2019-11-03 19:51:47'), (63273750.0, '2020-08-05 21:26:50')]\n",
      "Warning!:  58728086 3.0 [(59525939.0, '2019-12-30 05:04:12'), (64144594.0, '2020-09-30 19:36:06')]\n",
      "Warning!:  58856515 2.0 [(58875279.0, '2019-11-15 10:41:39')]\n",
      "Warning!:  58938886 4.0 [(58956985.0, '2019-11-20 14:42:41'), (59856857.0, '2020-01-22 09:52:46'), (63843401.0, '2020-09-11 08:23:09')]\n",
      "Warning!:  59123399 2.0 [(nan, nan)]\n",
      "Warning!:  59124918 2.0 [(61221420.0, '2020-04-15 04:37:45')]\n",
      "Warning!:  59380430 3.0 [(59381897.0, '2019-12-17 20:51:21'), (61188801.0, '2020-04-13 13:20:19')]\n",
      "Warning!:  59392084 3.0 [(63946065.0, '2020-09-17 20:54:18'), (64135699.0, '2020-09-30 10:31:56')]\n",
      "Warning!:  59459468 2.0 [(59464417.0, '2019-12-24 05:55:53')]\n",
      "Warning!:  59522643 2.0 [(59523463.0, '2019-12-29 21:29:23')]\n",
      "Warning!:  59673125 2.0 [(59677527.0, '2020-01-10 08:00:41')]\n",
      "Warning!:  59783361 2.0 [(59783618.0, '2020-01-17 08:28:11')]\n",
      "Warning!:  60058621 2.0 [(64217946.0, '2020-10-06 00:53:31')]\n",
      "Warning!:  60157335 5.0 [(60158209.0, '2020-02-10 20:50:30'), (60336262.0, '2020-02-21 10:03:50'), (62113067.0, '2020-05-31 07:42:36'), (64998855.0, '2020-11-25 05:12:13')]\n",
      "Warning!:  60199316 2.0 [(60199477.0, '2020-02-13 02:06:10')]\n",
      "Warning!:  60923982 2.0 [(60927524.0, '2020-03-30 09:25:27')]\n",
      "Warning!:  61319140 4.0 [(61321553.0, '2020-04-20 11:37:07'), (61321627.0, '2020-04-20 11:40:45')]\n",
      "Warning!:  61425296 9.0 [(61529308.0, '2020-04-30 17:27:54'), (61555448.0, '2020-05-02 06:40:07'), (61556348.0, '2020-05-02 08:17:31'), (61565801.0, '2020-05-02 20:22:28'), (61567062.0, '2020-05-02 22:20:37'), (61578364.0, '2020-05-03 16:58:12'), (61580287.0, '2020-05-03 19:16:59'), (61583254.0, '2020-05-04 00:07:44')]\n",
      "Warning!:  61630990 3.0 [(61633187.0, '2020-05-06 10:45:02'), (63635774.0, '2020-08-28 14:29:10')]\n",
      "Warning!:  61639263 4.0 [(61818969.0, '2020-05-15 12:14:10'), (62120235.0, '2020-05-31 17:46:28'), (64162264.0, '2020-10-01 19:46:12')]\n",
      "Warning!:  62092147 3.0 [(62092401.0, '2020-05-29 18:21:36'), (62202181.0, '2020-06-04 18:56:24')]\n",
      "Warning!:  62311222 3.0 [(62311595.0, '2020-06-10 19:26:28'), (62311811.0, '2020-06-10 19:39:12')]\n",
      "Warning!:  62482404 2.0 [(62502083.0, '2020-06-21 17:26:46')]\n",
      "Warning!:  62678401 2.0 [(62678585.0, '2020-07-01 14:00:47')]\n",
      "Warning!:  62836692 2.0 [(62837403.0, '2020-07-10 15:19:53')]\n",
      "Warning!:  63114255 2.0 [(63114571.0, '2020-07-27 11:41:19')]\n",
      "Warning!:  63561570 2.0 [(63561931.0, '2020-08-24 13:22:51')]\n",
      "Warning!:  63738681 3.0 [(63738887.0, '2020-09-04 09:56:50'), (63739202.0, '2020-09-04 10:17:29')]\n",
      "Warning!:  64434655 2.0 [(64438413.0, '2020-10-20 04:19:07')]\n",
      "Warning!:  64646247 2.0 [(64646350.0, '2020-11-02 13:18:18')]\n",
      "Warning!:  64993322 2.0 [(64994759.0, '2020-11-24 21:00:34')]\n"
     ]
    }
   ],
   "source": [
    "def checker_ans_and_ans_tup(q_id: int, ans_count: int, ans_list: list) -> None:\n",
    "    if len(ans_list) < ans_count:\n",
    "        print(\"Warning!: \", int(q_id), ans_count, ans_list)\n",
    "        \n",
    "_ = pd_tmp_new2_w.apply(lambda row: checker_ans_and_ans_tup(row.Q_id_x, row.Answer_count, row.Answer_tup), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "d0f3a96b-d724-4a6d-816d-21ba5ad1c65d",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new2_w.insert(len(pd_tmp_new2_w.columns), 'First_ans_time', np.nan)\n",
    "pd_tmp_new2_w.insert(len(pd_tmp_new2_w.columns), 'First_acc_ans_time', np.nan)\n",
    "pd_tmp_new3_w = pd_tmp_new2_w.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "c8a77995-f0dc-4c9e-b2ef-34b28ed28de2",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "for index1, row_status in pd_tmp_new3_w.iterrows():   \n",
    "    flag = 0\n",
    "    fr_time = pd.to_datetime(datetime.datetime.now())\n",
    "    acc_time = pd.to_datetime(datetime.datetime.now())\n",
    "        \n",
    "    if not row_status[\"Answer_tup\"]:\n",
    "        pd_tmp_new3_w.at[index1,'First_ans_time'] = np.nan\n",
    "        pd_tmp_new3_w.at[index1,'First_acc_ans_time'] = np.nan\n",
    "        continue\n",
    "\n",
    "    for answer in row_status[\"Answer_tup\"]:\n",
    "        \n",
    "        anwer_time = pd.to_datetime(answer[1])\n",
    "        \n",
    "        if flag == 0:\n",
    "            fr_time = anwer_time\n",
    "            flag = 1\n",
    "        \n",
    "        if fr_time > anwer_time:\n",
    "            fr_time = anwer_time\n",
    "        \n",
    "        if row_status[\"Accepted_Answer_id\"] == answer[0]:\n",
    "            acc_time = anwer_time\n",
    "\n",
    "    pd_tmp_new3_w.at[index1,'First_ans_time'] = fr_time\n",
    "        \n",
    "    if pd.isna(row_status[\"Accepted_Answer_id\"]):\n",
    "        acc_time = np.nan\n",
    "    \n",
    "    pd_tmp_new3_w.at[index1,'First_acc_ans_time'] = acc_time\n",
    "    \n",
    "pd_tmp_new3_w[\"Duration_ans\"] = pd_tmp_new3_w.apply(lambda row: (row.First_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_ans_time) else np.nan, axis=1)\n",
    "pd_tmp_new3_w[\"Duration_acc_ans\"] = pd_tmp_new3_w.apply(lambda row: (row.First_acc_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_acc_ans_time) else np.nan, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d8e3da07-0e29-48c5-9e23-ad299c1fe27b",
   "metadata": {},
   "source": [
    "We did the same behaviour for the Question_wo_code_info list."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "e48941aa-571c-4397-b3c7-3dd4c3c89f92",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_q_wo_code_v1 = pd.DataFrame(columns=['First_ans_time', 'First_acc_ans_time', 'Answers'])\n",
    "\n",
    "df_q_wo_code_v1[\"Q_info\"] = Question_wo_code_info\n",
    "\n",
    "df_q_wo_code_v1[\"Q_id\"]          = df_q_wo_code_v1['Q_info'].apply(lambda q_info: q_info[0])\n",
    "df_q_wo_code_v1[\"Q_create_time\"] = df_q_wo_code_v1['Q_info'].apply(lambda q_info: q_info[1])\n",
    "df_q_wo_code_v1[\"View_count\"]    = df_q_wo_code_v1['Q_info'].apply(lambda q_info: q_info[2])\n",
    "df_q_wo_code_v1[\"Answer_count\"]  = df_q_wo_code_v1['Q_info'].apply(lambda q_info: q_info[3])\n",
    "df_q_wo_code_v1[\"Comment_count\"] = df_q_wo_code_v1['Q_info'].apply(lambda q_info: q_info[4])\n",
    "df_q_wo_code_v1[\"Score\"]         = df_q_wo_code_v1['Q_info'].apply(lambda q_info: q_info[5])\n",
    "df_q_wo_code_v1[\"Accepted_Answer_id\"] = df_q_wo_code_v1['Q_info'].apply(lambda q_info: q_info[6])\n",
    "df_q_wo_code_v1 = df_q_wo_code_v1.drop(['Q_info'], axis='columns')\n",
    "df_q_wo_code_v1[\"Q_create_time\"] = pd.to_datetime(df_q_wo_code_v1[\"Q_create_time\"])\n",
    "df_q_wo_code_v1[\"First_acc_ans_time\"] = pd.to_datetime(df_q_wo_code_v1[\"First_acc_ans_time\"])\n",
    "df_q_wo_code_v1[\"First_ans_time\"]     = pd.to_datetime(df_q_wo_code_v1[\"First_ans_time\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "b6e57dad-db68-4579-bfb2-aaa08da7881c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Store\n",
    "df_q_wo_code_v1[\"Q_id\"].to_csv('../code_output_csv/df_tf_q_wo_code.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "raw",
   "id": "e3c41fc5-ade4-4b12-9368-7262b638af3f",
   "metadata": {
    "tags": []
   },
   "source": [
    "SELECT df_tf_q_wo_code.Q_id, all_results.Id, all_results.CreationDate, all_results.ParentId\n",
    "FROM df_tf_q_wo_code\n",
    "INNER JOIN all_results\n",
    "ON df_tf_q_wo_code.Q_id = all_results.ParentId;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "79c85a19-f954-4710-8bff-92bc0e3a308a",
   "metadata": {},
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_all.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_night_wo = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "9f685ef5-bff1-4bf2-9e2c-3329866a7248",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_night_wo = df_night_wo.reset_index(drop=True)\n",
    "df_q_wo_code_v1['Answers'] = df_q_wo_code_v1.apply(lambda x: [], axis=1)\n",
    "\n",
    "pd_tmp_wo = pd.merge(df_q_wo_code_v1, df_night_wo, how='left',left_on=['Q_id'],right_on=['ParentId']).reset_index(drop=True)\n",
    "\n",
    "pd_tmp_wo[\"Answer_tup\"] = pd_tmp_wo.apply(lambda x: (x.Id, x.CreationDate), axis=1)\n",
    "pd_tmp_new_wo = pd_tmp_wo.groupby([\"Q_id_x\", \"Q_create_time\", \"View_count\", \"Comment_count\", \"Score\", \"Answer_count\", \"Accepted_Answer_id\"], dropna=False)[\"Answer_tup\"].agg(list).reset_index()\n",
    "pd_tmp_new2_wo = pd_tmp_new_wo[pd_tmp_new_wo.Answer_count != 0].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "16ad1b6d-2358-431c-9b9c-a200c3119fdc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Warning!:  38164798 3 [(46519048.0, '2017-10-02 03:19:09'), (46589591.0, '2017-10-05 15:35:53')]\n",
      "Warning!:  38469494 2 [(39182361.0, '2016-08-27 14:43:44')]\n",
      "Warning!:  39691902 9 [(40295999.0, '2016-10-27 23:59:30'), (50698801.0, '2018-06-05 11:19:44'), (53881090.0, '2018-12-21 07:58:33'), (54554286.0, '2019-02-06 13:01:58'), (59001644.0, '2019-11-22 20:56:08'), (61759744.0, '2020-05-12 18:53:05'), (62806906.0, '2020-07-09 03:25:36'), (63051525.0, '2020-07-23 09:53:38')]\n",
      "Warning!:  44823397 2 [(nan, nan)]\n",
      "Warning!:  45004514 2 [(45008402.0, '2017-07-10 09:28:51')]\n",
      "Warning!:  45216886 2 [(54944891.0, '2019-03-01 12:40:29')]\n",
      "Warning!:  45874928 3 [(45877018.0, '2017-08-25 08:08:53'), (55431230.0, '2019-03-30 12:01:30')]\n",
      "Warning!:  45926945 2 [(45928692.0, '2017-08-28 23:34:56')]\n",
      "Warning!:  47161181 3 [(47161607.0, '2017-11-07 15:28:39'), (47217413.0, '2017-11-10 07:08:30')]\n",
      "Warning!:  47353479 2 [(47355283.0, '2017-11-17 16:34:02')]\n",
      "Warning!:  47946413 2 [(47996856.0, '2017-12-27 18:18:10')]\n",
      "Warning!:  48918896 2 [(63458214.0, '2020-08-17 20:30:00')]\n",
      "Warning!:  50474336 2 [(50474657.0, '2018-05-22 18:54:02')]\n",
      "Warning!:  51218869 2 [(51219305.0, '2018-07-07 01:53:47')]\n",
      "Warning!:  53974005 2 [(62697169.0, '2020-07-02 13:11:49')]\n",
      "Warning!:  55371035 2 [(59109801.0, '2019-11-29 18:55:38')]\n",
      "Warning!:  55514435 2 [(58683231.0, '2019-11-03 18:15:52')]\n",
      "Warning!:  55681502 3 [(nan, nan)]\n",
      "Warning!:  57610804 2 [(57614277.0, '2019-08-22 17:20:30')]\n",
      "Warning!:  58944060 6 [(58944091.0, '2019-11-19 22:54:17'), (58944276.0, '2019-11-19 23:11:45'), (59705945.0, '2020-01-12 16:40:36'), (60107059.0, '2020-02-07 04:24:08'), (61356393.0, '2020-04-22 02:40:47')]\n",
      "Warning!:  62073259 2 [(nan, nan)]\n",
      "Warning!:  62112161 2 [(62112692.0, '2020-05-31 06:57:59')]\n",
      "Warning!:  62906266 2 [(63099977.0, '2020-07-26 12:14:37')]\n",
      "Warning!:  64734203 2 [(64734388.0, '2020-11-08 02:53:34')]\n",
      "Warning!:  64738104 2 [(nan, nan)]\n"
     ]
    }
   ],
   "source": [
    "_ = pd_tmp_new2_wo.apply(lambda row: checker_ans_and_ans_tup(row.Q_id_x, row.Answer_count, row.Answer_tup), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "c1597218-7d80-46bd-9830-04e5a51ce61b",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new2_wo.insert(len(pd_tmp_new2_wo.columns), 'First_ans_time', np.nan)\n",
    "pd_tmp_new2_wo.insert(len(pd_tmp_new2_wo.columns), 'First_acc_ans_time', np.nan)\n",
    "pd_tmp_new3_wo = pd_tmp_new2_wo.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "7741c7cc-9ed9-4b7b-8af6-7c3c577d3717",
   "metadata": {},
   "outputs": [],
   "source": [
    "for index1, row_status in pd_tmp_new3_wo.iterrows():   \n",
    "    flag = 0\n",
    "    fr_time = pd.to_datetime(datetime.datetime.now())\n",
    "    acc_time = pd.to_datetime(datetime.datetime.now())\n",
    "        \n",
    "    if not row_status[\"Answer_tup\"]:\n",
    "        pd_tmp_new3_wo.at[index1,'First_ans_time'] = np.nan\n",
    "        pd_tmp_new3_wo.at[index1,'First_acc_ans_time'] = np.nan\n",
    "        continue\n",
    "\n",
    "    for answer in row_status[\"Answer_tup\"]:\n",
    "        \n",
    "        anwer_time = pd.to_datetime(answer[1])\n",
    "        \n",
    "        if flag == 0:\n",
    "            fr_time = anwer_time\n",
    "            flag = 1\n",
    "        \n",
    "        if fr_time > anwer_time:\n",
    "            fr_time = anwer_time\n",
    "        \n",
    "        if row_status[\"Accepted_Answer_id\"] == answer[0]:\n",
    "            acc_time = anwer_time\n",
    "\n",
    "    pd_tmp_new3_wo.at[index1,'First_ans_time'] = fr_time\n",
    "        \n",
    "    if pd.isna(row_status[\"Accepted_Answer_id\"]):\n",
    "        acc_time = np.nan\n",
    "    \n",
    "    pd_tmp_new3_wo.at[index1,'First_acc_ans_time'] = acc_time\n",
    "    \n",
    "pd_tmp_new3_wo[\"Duration_ans\"] = pd_tmp_new3_wo.apply(lambda row: (row.First_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_ans_time) else np.nan, axis=1)\n",
    "pd_tmp_new3_wo[\"Duration_acc_ans\"] = pd_tmp_new3_wo.apply(lambda row: (row.First_acc_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_acc_ans_time) else np.nan, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "86d8b748-6df2-44b3-92cb-638069ea6ce0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Store\n",
    "pd_tmp_new3_w.to_csv('./CSV_data/plt_tf_q_w_code.csv', encoding='utf-8')\n",
    "pd_tmp_new3_wo.to_csv('./CSV_data/plt_tf_q_wo_code.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "76009349",
   "metadata": {},
   "source": [
    "### Extracting the text parts from body"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "426b615b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_text_blocks(id: int, body: str) -> Optional[List]:\n",
    "    # It's bug\n",
    "    if id in [9286, 12769]: \n",
    "        body = body + \"</p>\"\n",
    "    \n",
    "    regex = r\"(<p>((.*?)|(\\n)*)*<\\/p>)|(<ul>((.*?)|(\\n)*)*<\\/ul>)|(<ol>((.*?)|(\\n)*)*<\\/ol>)\"\n",
    "    matches = re.finditer(regex, body, re.MULTILINE | re.IGNORECASE)\n",
    "    result = []\n",
    "    \n",
    "    # if id in [18397]:\n",
    "    #     return result\n",
    "\n",
    "    try:\n",
    "        for matchNum, match in enumerate(matches, start=1):\n",
    "            text = match.group()\n",
    "            text = text.replace(\"<p>\", \"\")\n",
    "            text = text.replace(\"<strong>\", \"\")\n",
    "            text = text.replace(\"<br>\", \"\")\n",
    "            text = text.replace(\"<ol>\", \"\")\n",
    "            text = text.replace(\"<ul>\", \"\")\n",
    "            text = text.replace(\"<li>\", \"\")\n",
    "            text = text.replace(\"</p>\", \"\")\n",
    "            text = text.replace(\"</strong>\", \"\")\n",
    "            text = text.replace(\"</ol>\", \"\")\n",
    "            text = text.replace(\"</ul>\", \"\")\n",
    "            text = text.replace(\"</li>\", \"\")\n",
    "            result.append(text)\n",
    "        return result\n",
    "    except:\n",
    "        print(\"Error(1): \", id)\n",
    "        print(body)\n",
    "        return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "96ff4ec3",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags['Text'] = df_w_tf_tags.apply(lambda row: extract_text_blocks(row.name, row.Body), axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec3bb294",
   "metadata": {},
   "source": [
    "#### Finding the number of body words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "18933061",
   "metadata": {},
   "outputs": [],
   "source": [
    "def find_q_text_words(text_list: List) -> int:\n",
    "    word_count = 0\n",
    "    for text in text_list:\n",
    "        word_count += sum([i.strip(string.punctuation).isalpha() for i in text.split()])\n",
    "    return word_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "15bc8431",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags['Q_text_words_num'] = df_w_tf_tags[\"Text\"].apply(lambda text_list: find_q_text_words(text_list))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b24d4e61-0992-43cf-bfc4-83cebb7d1bbf",
   "metadata": {},
   "source": [
    "Save the dataframe as csv format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "3d8f131a-19f7-46b8-b8e3-399ba5537607",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# df_w_Tens_tags.to_csv('./amin_result_v1.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4a5ff90-afcd-45a3-9d2d-a2b58de44dd9",
   "metadata": {},
   "source": [
    "### Find the Regular Expressions for Unix and Windows base Pathnames"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3406a54-9a09-4255-8287-d16fab93833d",
   "metadata": {},
   "source": [
    "#### 1) Absolute and Relative Pathnames in UNIX OS"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "20c90be3-66df-4290-972f-940233a71e7c",
   "metadata": {},
   "source": [
    "Stack/trace example: https://stackoverflow.com/questions/37337728/tensorflow-internalerror-blas-sgemm-launch-failed"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30e9d6c0-8adb-4939-9f06-329b036eb875",
   "metadata": {},
   "source": [
    "Find the restrictions and limitations related to the Unix pathnames: https://www.cyberciti.biz/faq/linuxunix-rules-for-naming-file-and-directory-names/"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe32c563-d1db-4f0a-9820-62fd42a6591e",
   "metadata": {},
   "source": [
    "Online regular expression environment for testing: https://regex101.com/r/ZyEx5u/4"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "76d9456a-4c9e-4a8f-bee5-86a5be847e82",
   "metadata": {},
   "source": [
    "> Regular Expression: \"[^\\n\\&\\:\\|\\>\\<\\/\\ \\\"\\;\\&]*(\\/[^\\n\\&\\:\\|\\>\\<\\/]+)+\\/([^\\n\\&\\:\\|\\>\\<\\/]+)(\\.pyc|\\.py)\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d6bb6875-32a5-4f52-b33a-9cb522a49f72",
   "metadata": {},
   "source": [
    "#### 1) Pathnames in Windows OS"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dcd9c86a-e55e-4200-85ff-ed3033b4ee64",
   "metadata": {},
   "source": [
    "Stack/trace example: https://stackoverflow.com/questions/49434031/tensorflow-on-windows-cpu-version-importerror-no-module-named-pywrap-tensorf"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "84dd40a8-d970-435c-a06d-6a32130d3055",
   "metadata": {},
   "source": [
    "Find the restrictions and limitations related to the Windows pathnames: <br /> \n",
    "https://docs.microsoft.com/en-us/dotnet/standard/io/file-path-formats <br />\n",
    "https://docs.microsoft.com/en-us/windows/win32/fileio/naming-a-file"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ced5ffa1-d9f8-4e8c-b9a8-8f438af95a1a",
   "metadata": {},
   "source": [
    "Online regular expression environment for testing: https://regex101.com/r/L6xmCa/1 , https://regex101.com/r/wz1WqW/1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2ec1be13-b32c-4d50-9ea2-490eafe5b139",
   "metadata": {},
   "source": [
    "> Regular Expression: \"[a-zA-Z]:\\\\?([^\\<\\>\\:\\\"\\\\\\/\\/\\|\\?\\*\\n]+\\\\)+([^\\<\\>\\:\\\"\\\\\\/\\/\\|\\?\\*\\n]+)(\\.pyc|\\.py)\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "bc401c12-44a2-402c-9e4a-1de67941e502",
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_pathnames_from_code_column(code_sec: List, _id: int) -> Tuple[str, List]: \n",
    "    try:\n",
    "        result_post_file_names = []\n",
    "        OS_flag = None\n",
    "        error_index = None\n",
    "        error_code = None\n",
    "        regex_unix    = r\"[^\\n\\&\\:\\|\\>\\<\\/\\ \\\"\\;\\&]*(\\/[^\\n\\&\\:\\|\\>\\<\\/]+)+\\/([^\\n\\&\\:\\|\\>\\<\\/\\\\]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d*)\\W*in\\s\\W*\\d*([a-zA-Z_$][a-zA-Z_$0-9]*)\"\n",
    "        regex_windows = r\"(([a-zA-Z]:)|\\~)\\\\?\\/?([^\\<\\>\\:\\\"\\\\\\|\\?\\*\\n]+\\\\)+([^\\<\\>\\:\\\"\\\\\\/\\/\\|\\?\\*\\n]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d*)\\W*in\\W+([a-zA-Z_$][a-zA-Z_$0-9]*)\"\n",
    "        \n",
    "        for idx, code in enumerate(code_sec):\n",
    "            pattern_unix  = re.compile(regex_unix)\n",
    "            pattern_windows  = re.compile(regex_windows)\n",
    "            if pattern_unix.search(code):\n",
    "                OS_flag = \"unix\"\n",
    "                break\n",
    "            elif pattern_windows.search(code):\n",
    "                OS_flag = \"windows\"\n",
    "                break\n",
    "                \n",
    "        if OS_flag == \"unix\":\n",
    "            for idx, code in enumerate(code_sec):\n",
    "                error_index = idx\n",
    "                code = code.replace(\"\\\\n\", \"\\n\")\n",
    "                code = code.replace('&lt;', '<')\n",
    "                code = code.replace('&gt;', '>')\n",
    "                code = code.replace('&quot;', '\"')\n",
    "                error_code = code\n",
    "                matches = re.finditer(regex_unix, code, re.MULTILINE)\n",
    "                file_names_for_each_code_part = []\n",
    "                \n",
    "                for matchNum, match in enumerate(matches, start=1):\n",
    "                    file_names_for_each_code_part.append((match.groups()[1].strip(), match.groups()[4].strip()))\n",
    "#                     file_names_for_each_code_part.append(match.groups()[1].strip())\n",
    "                    # print (\"Match {matchNum} was found at {start}-{end}: {match}\".format(matchNum = matchNum, start = match.start(), end = match.end(), match = match.group()))\n",
    "                    # for groupNum in range(0, len(match.groups())):\n",
    "                        # groupNum = groupNum + 1\n",
    "                        # print (\"Group {groupNum} found at {start}-{end}: {group}\".format(groupNum = groupNum, start = match.start(groupNum), end = match.end(groupNum), group = match.group(groupNum)))\n",
    "#                 print(file_names_for_each_code_part)\n",
    "#                 file_names_for_each_code_part = list(set(file_names_for_each_code_part))            # Create a unique list\n",
    "#                 print(file_names_for_each_code_part)\n",
    "                \n",
    "                if file_names_for_each_code_part:                                                   # Ignore the empty list\n",
    "                    result_post_file_names.append(file_names_for_each_code_part)\n",
    "                    \n",
    "        elif OS_flag == \"windows\":\n",
    "            for idx, code in enumerate(code_sec):\n",
    "                code = code.replace('&lt;', '<')\n",
    "                code = code.replace('&gt;', '>')\n",
    "                code = code.replace('&quot;', '\"')\n",
    "                error_code = code\n",
    "                matches = re.finditer(regex_windows, code, re.MULTILINE)                            \n",
    "                file_names_for_each_code_part = []\n",
    "\n",
    "                for matchNum, match in enumerate(matches, start=1):\n",
    "                    file_names_for_each_code_part.append((match.groups()[3].strip(), match.groups()[6].strip()))\n",
    "#                     file_names_for_each_code_part.append(match.groups()[1].strip())\n",
    "                    # print (\"Match {matchNum} was found at {start}-{end}: {match}\".format(matchNum = matchNum, start = match.start(), end = match.end(), match = match.group()))\n",
    "                    # for groupNum in range(0, len(match.groups())):\n",
    "                        # groupNum = groupNum + 1\n",
    "                        # print (\"Group {groupNum} found at {start}-{end}: {group}\".format(groupNum = groupNum, start = match.start(groupNum), end = match.end(groupNum), group = match.group(groupNum)))\n",
    "#                 file_names_for_each_code_part = [s.strip() for s in file_names_for_each_code_part]  # Strip the list\n",
    "#                 file_names_for_each_code_part = list(set(file_names_for_each_code_part))            # create a unique list\n",
    "                if file_names_for_each_code_part:                                                     # Ignore the empty list\n",
    "                    result_post_file_names.append(file_names_for_each_code_part)\n",
    "            \n",
    "    except:\n",
    "        print(\"\\n Error Id:\", _id, \" Error index:\", error_index, \"\\n\")\n",
    "        print(error_code)\n",
    "        return None, None\n",
    "\n",
    "    return OS_flag, result_post_file_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "fc583436-6446-47b3-8811-ee3b316c57cf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(None, [])"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Example for Unix path\n",
    "# https://stackoverflow.com/questions/51839415/tensorflow-valueerror-rank-mismatch\n",
    "# df_w_tf_tags.iloc[9][:]\n",
    "extract_pathnames_from_code_column(df_w_tf_tags[\"Code\"][51], 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "8642e202-ad70-4f13-9440-57eeeb3fda8a",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Example for Windows path\n",
    "# df_w_tf_tags['Code'] = df_w_tf_tags['Body'].apply(lambda row_body: extract_code_blocks(row_body))\n",
    "# print(df_w_tf_tags[\"Code\"][9])\n",
    "# extract_pathnames_from_code_column(df_w_tf_tags[\"Code\"][84])\n",
    "# extract_pathnames_from_code_column(df_w_tf_tags[\"Code\"][606])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "0c61970a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_w_tf_tags[\"Body\"][606]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "0dd38400-fb42-4dc7-8f9d-deb13a98c1a0",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_w_tf_tags['Bugy_py_files'] = df_w_tf_tags.apply(lambda row: extract_pathnames_from_code_column(row.Code, row.Id), axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "23e7d4de",
   "metadata": {},
   "source": [
    "Save the dataframe as csv format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "4e4ea318-37e3-4d03-b3a2-566723d22160",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# df_w_tf_tags.loc[df_w_PT_tags['Id'] == 46509039]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "4718e5db-2866-4f96-8a3a-2a09e559cb99",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_w_tf_tags[[\"Title\", \"Code\", \"Bugy_py_files\"]].head(100)\n",
    "# df_w_tf_tags.to_csv('./tf_Code_bug.csv', sep='\\n', encoding='utf-8')\n",
    "# df_w_tf_tags['Bugy_py_files'].to_csv('./tf_bugy_tuples.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "a7e0c3f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_has_trace_col(cal_tuple) -> bool:\n",
    "    if cal_tuple[0] == \"windows\" or cal_tuple[0] == \"unix\":\n",
    "        return True\n",
    "    elif cal_tuple[0] is None:\n",
    "        return False\n",
    "    else:\n",
    "        print(\"Error!\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "acd983ea",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags['Has_trace'] = df_w_tf_tags['Bugy_py_files'].apply(lambda cell_tuple: create_has_trace_col(cell_tuple))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e60aff5",
   "metadata": {},
   "source": [
    "#### Finding the number and type of LOC in the body of post:"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b33da7bf-c174-48ba-999c-766d91c33a7a",
   "metadata": {},
   "source": [
    "For example how many pairs did you find in a post and defining the OS of that stack trace"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "9c25a136",
   "metadata": {},
   "outputs": [],
   "source": [
    "def finding_line_of_code(code_list: List) -> Tuple[int, int, int]:\n",
    "    line_count_trace_win   = 0\n",
    "    line_count_trace_unix  = 0\n",
    "    line_count_simple_code = 0\n",
    "    \n",
    "    for code in code_list:\n",
    "        regex_unix    = r\"[^\\n\\&\\:\\|\\>\\<\\/\\ \\\"\\;\\&]*(\\/[^\\n\\&\\:\\|\\>\\<\\/]+)+\\/([^\\n\\&\\:\\|\\>\\<\\/\\\\]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d*)\\W*in\\s\\W*\\d*([a-zA-Z_$][a-zA-Z_$0-9]*)\"\n",
    "        regex_windows = r\"(([a-zA-Z]:)|\\~)\\\\?\\/?([^\\<\\>\\:\\\"\\\\\\|\\?\\*\\n]+\\\\)+([^\\<\\>\\:\\\"\\\\\\/\\/\\|\\?\\*\\n]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d*)\\W*in\\W+([a-zA-Z_$][a-zA-Z_$0-9]*)\"\n",
    "        # regex_unix    = r\"[^\\n\\&\\:\\|\\>\\<\\/\\ \\\"\\;\\&]*(\\/[^\\n\\&\\:\\|\\>\\<\\/]+)+\\/([^\\n\\&\\:\\|\\>\\<\\/]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d*)\\W*in\\W+([a-zA-Z_$][a-zA-Z_$0-9]*)\" version 2\n",
    "        # regex_unix    = r\"[^\\n\\&\\:\\|\\>\\<\\/\\ \\\"\\;\\&]*(\\/[^\\n\\&\\:\\|\\>\\<\\/]+)+\\/([^\\n\\&\\:\\|\\>\\<\\/]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d+)\\W*in\\W*([a-zA-Z_$][a-zA-Z_$0-9]*)\" version 1\n",
    "        # regex_windows = r\"[a-zA-Z]:\\\\?([^\\<\\>\\:\\\"\\\\\\/\\/\\|\\?\\*\\n]+\\\\)+([^\\<\\>\\:\\\"\\\\\\/\\/\\|\\?\\*\\n]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d+)\\W*in\\W*([a-zA-Z_$][a-zA-Z_$0-9]*)\"\n",
    "        pattern_unix  = re.compile(regex_unix)\n",
    "        pattern_windows = re.compile(regex_windows)\n",
    "        if pattern_unix.search(code):\n",
    "            OS_flag = \"unix\"\n",
    "        elif pattern_windows.search(code):\n",
    "            OS_flag = \"windows\"\n",
    "        else:\n",
    "            OS_flag = \"nothing\"\n",
    "            \n",
    "        code_wo_empty_line = os.linesep.join([s for s in code.splitlines() if s])\n",
    "            \n",
    "        if OS_flag == \"unix\":\n",
    "            line_count_trace_unix += len(code_wo_empty_line.splitlines())\n",
    "        elif OS_flag == \"windows\":\n",
    "            line_count_trace_win += len(code_wo_empty_line.splitlines())\n",
    "        elif OS_flag == \"nothing\":\n",
    "            line_count_simple_code += len(code_wo_empty_line.splitlines()) \n",
    "        else:\n",
    "            print(\"Error!\")\n",
    "\n",
    "    return (line_count_trace_unix, line_count_trace_win, line_count_simple_code)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "4054d8d7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0, 0, 6)"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "finding_line_of_code(df_w_tf_tags.loc[19, 'Code'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "c2f3958c",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags['Line_code_u_w_s'] = df_w_tf_tags['Code'].apply(lambda code_list: finding_line_of_code(code_list))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "b3e57594",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags['Line_code_uix'] = df_w_tf_tags['Line_code_u_w_s'].apply(lambda loc_tuple: loc_tuple[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "4a683106",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_w_tf_tags['Line_code_win'] = df_w_tf_tags['Line_code_u_w_s'].apply(lambda loc_tuple: loc_tuple[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "b99bf6ab",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Has a code (Python code) inside the post\n",
    "df_w_tf_tags['Line_code_simple_code'] = df_w_tf_tags['Line_code_u_w_s'].apply(lambda loc_tuple: loc_tuple[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "3534efd4",
   "metadata": {
    "scrolled": true,
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Id</th>\n",
       "      <th>PostTypeId</th>\n",
       "      <th>AcceptedAnswerId</th>\n",
       "      <th>CreationDate</th>\n",
       "      <th>ViewCount</th>\n",
       "      <th>AnswerCount</th>\n",
       "      <th>CommentCount</th>\n",
       "      <th>Score</th>\n",
       "      <th>Title</th>\n",
       "      <th>Body</th>\n",
       "      <th>Tags</th>\n",
       "      <th>Source</th>\n",
       "      <th>source</th>\n",
       "      <th>Code</th>\n",
       "      <th>Has_code</th>\n",
       "      <th>Text</th>\n",
       "      <th>Q_text_words_num</th>\n",
       "      <th>Bugy_py_files</th>\n",
       "      <th>Has_trace</th>\n",
       "      <th>Line_code_u_w_s</th>\n",
       "      <th>Line_code_uix</th>\n",
       "      <th>Line_code_win</th>\n",
       "      <th>Line_code_simple_code</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>43574928</td>\n",
       "      <td>1</td>\n",
       "      <td>43575038.0</td>\n",
       "      <td>2017-04-23 18:30:08</td>\n",
       "      <td>1118</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>How to use custom/non-default tf.Graph in Tens...</td>\n",
       "      <td>&lt;p&gt;I'm very new to &lt;code&gt;Tensorflow&lt;/code&gt; and...</td>\n",
       "      <td>&lt;python&gt;&lt;session&gt;&lt;graph&gt;&lt;parallel-processing&gt;&lt;...</td>\n",
       "      <td>old</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[t = np.linspace(0,2*np.pi)\\nfig, ax = plt.sub...</td>\n",
       "      <td>True</td>\n",
       "      <td>[I'm very new to &lt;code&gt;Tensorflow&lt;/code&gt; and I...</td>\n",
       "      <td>111</td>\n",
       "      <td>(None, [])</td>\n",
       "      <td>False</td>\n",
       "      <td>(0, 0, 40)</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>52019226</td>\n",
       "      <td>1</td>\n",
       "      <td>52019631.0</td>\n",
       "      <td>2018-08-25 16:32:19</td>\n",
       "      <td>312</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>Keras w/ Tensorflow intermediate layer extract...</td>\n",
       "      <td>&lt;p&gt;I am currently trying to leverage an interm...</td>\n",
       "      <td>&lt;python&gt;&lt;tensorflow&gt;&lt;keras&gt;&lt;deep-learning&gt;&lt;bat...</td>\n",
       "      <td>old</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[model = load_model('model.h5')\\ninp = model.i...</td>\n",
       "      <td>True</td>\n",
       "      <td>[I am currently trying to leverage an intermed...</td>\n",
       "      <td>77</td>\n",
       "      <td>(None, [])</td>\n",
       "      <td>False</td>\n",
       "      <td>(0, 0, 23)</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Id  PostTypeId  AcceptedAnswerId         CreationDate  ViewCount  AnswerCount  CommentCount  Score                                              Title                                               Body                                               Tags Source source                                               Code  Has_code                                               Text  Q_text_words_num Bugy_py_files  Has_trace Line_code_u_w_s  Line_code_uix  Line_code_win  Line_code_simple_code\n",
       "0  43574928           1        43575038.0  2017-04-23 18:30:08       1118            1             0      3  How to use custom/non-default tf.Graph in Tens...  <p>I'm very new to <code>Tensorflow</code> and...  <python><session><graph><parallel-processing><...    old    NaN  [t = np.linspace(0,2*np.pi)\\nfig, ax = plt.sub...      True  [I'm very new to <code>Tensorflow</code> and I...               111    (None, [])      False      (0, 0, 40)              0              0                     40\n",
       "1  52019226           1        52019631.0  2018-08-25 16:32:19        312            1             2      0  Keras w/ Tensorflow intermediate layer extract...  <p>I am currently trying to leverage an interm...  <python><tensorflow><keras><deep-learning><bat...    old    NaN  [model = load_model('model.h5')\\ninp = model.i...      True  [I am currently trying to leverage an intermed...                77    (None, [])      False      (0, 0, 23)              0              0                     23"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_w_tf_tags.head(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06fde1f5",
   "metadata": {},
   "source": [
    "#### Question Post's Length (body of the post): Defining the lists for plotting their comparisons"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "bd5dee83",
   "metadata": {},
   "outputs": [],
   "source": [
    "list_num_words_w_tra  = []\n",
    "list_num_words_wo_tra = []\n",
    "\n",
    "for index1, row in df_w_tf_tags.iterrows():\n",
    "    if row.Has_trace is True:\n",
    "        list_num_words_w_tra.append(row.Q_text_words_num)\n",
    "    elif row.Has_trace is False:\n",
    "        list_num_words_wo_tra.append(row.Q_text_words_num)\n",
    "    else:\n",
    "        print(\"Error!\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "684ae0b5-da13-4ad7-b853-44783e21104b",
   "metadata": {},
   "source": [
    "#### OS Stack Traces: Defining the number of stack straces the belongs to the Unix or Windows OS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "bd3530e5-831c-4db8-b250-089f05938c10",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "We found the 2426 of stackoverflow's posts that belong to the Windows OS.\n",
      "We found the 5403 of stackoverflow's posts that belong to the Unix OS.\n"
     ]
    }
   ],
   "source": [
    "counter_win  = 0\n",
    "counter_unix = 0\n",
    "# dim_win  = []\n",
    "# dim_unix = []\n",
    "\n",
    "for tp in df_w_tf_tags[\"Bugy_py_files\"]:\n",
    "    if tp[0] == \"windows\":\n",
    "        counter_win += 1\n",
    "#         np_array = np.array(tuple[1], dtype=object)\n",
    "#         dim_win.append(np_array.shape)\n",
    "        \n",
    "    elif tp[0] == \"unix\":\n",
    "        counter_unix += 1\n",
    "#         np_array = np.array(tuple[1], dtype=object)\n",
    "#         dim_unix.append(np_array.shape)\n",
    "\n",
    "print(f\"We found the {counter_win} of stackoverflow's posts that belong to the Windows OS.\")\n",
    "print(f\"We found the {counter_unix} of stackoverflow's posts that belong to the Unix OS.\")\n",
    "# print(\"The dimensions of Windows labels is: \", set(dim_win))\n",
    "# print(\"The dimensions of Unix labels is: \", set(dim_unix))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "862bb743-f63d-4f4f-9b62-e38bb4711f08",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Id                         int64\n",
       "PostTypeId                 int64\n",
       "AcceptedAnswerId         float64\n",
       "CreationDate              object\n",
       "ViewCount                  int64\n",
       "AnswerCount                int64\n",
       "CommentCount               int64\n",
       "Score                      int64\n",
       "Title                     object\n",
       "Body                      object\n",
       "Tags                      object\n",
       "Source                    object\n",
       "source                    object\n",
       "Code                      object\n",
       "Has_code                    bool\n",
       "Text                      object\n",
       "Q_text_words_num           int64\n",
       "Bugy_py_files             object\n",
       "Has_trace                   bool\n",
       "Line_code_u_w_s           object\n",
       "Line_code_uix              int64\n",
       "Line_code_win              int64\n",
       "Line_code_simple_code      int64\n",
       "dtype: object"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_w_tf_tags.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85e4c57c-d11e-4e1d-ac85-9c155d748630",
   "metadata": {},
   "source": [
    "#### Define two lists for storing the information of the questions w or w/o stack traces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "7a685f8b",
   "metadata": {},
   "outputs": [],
   "source": [
    "count_w_t  = 0\n",
    "count_wo_t = 0\n",
    "count_win  = 0\n",
    "count_unix = 0\n",
    "\n",
    "Question_with_trace_info = []\n",
    "Question_with_wo_trace_info = []\n",
    "\n",
    "def counting_w_or_wo_trace(row_id: int, \n",
    "                           row_cr: object, \n",
    "                           row_vc: int, \n",
    "                           row_ac: int, \n",
    "                           row_cc: int, \n",
    "                           row_sc: int, \n",
    "                           row_ac_an_id: float, \n",
    "                           row_t: object, \n",
    "                           has_code: bool) -> None:\n",
    "    \n",
    "    global count_w_t, count_wo_t, count_win, count_unix, Question_with_trace_info, Question_with_wo_trace_info\n",
    "\n",
    "    if has_code:\n",
    "        if row_t[0] is not None:\n",
    "            count_w_t = count_w_t + 1\n",
    "            Question_with_trace_info.append((row_id, row_cr, row_vc, row_ac, row_cc, row_sc, row_ac_an_id))\n",
    "            \n",
    "            if row_t[0] == 'unix':\n",
    "                count_unix = count_unix + 1\n",
    "            else:\n",
    "                count_win = count_win + 1\n",
    "        else:\n",
    "            count_wo_t = count_wo_t + 1\n",
    "            Question_with_wo_trace_info.append((row_id, row_cr, row_vc, row_ac, row_cc, row_sc, row_ac_an_id))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "ee29fdfd",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "_ = df_w_tf_tags.apply(lambda row: counting_w_or_wo_trace(row.Id, \n",
    "                                                               row.CreationDate, \n",
    "                                                               row.ViewCount,\n",
    "                                                               row.AnswerCount,\n",
    "                                                               row.CommentCount,\n",
    "                                                               row.Score,\n",
    "                                                               row.AcceptedAnswerId,\n",
    "                                                               row.Bugy_py_files, \n",
    "                                                               row.Has_code), axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "255f73aa-307c-4ca5-ad1f-9c80e0067c0f",
   "metadata": {},
   "source": [
    "#### Creating a string matrix of pairs"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ccfe9422-3706-4fc3-9135-8cbd87353226",
   "metadata": {
    "tags": []
   },
   "source": [
    "https://stackoverflow.com/questions/32037893/numpy-fix-array-with-rows-of-different-lengths-by-filling-the-empty-elements-wi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "3390d5ca-986e-4ddf-a48f-eb4e45ab46dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def numpy_fillna(data: List) -> np.ndarray:\n",
    "    # Get lengths of each row of data\n",
    "    lens = np.array([len(i) for i in data])\n",
    "\n",
    "    # Mask of valid places in each row\n",
    "    mask = np.arange(lens.max()) < lens[:, None]\n",
    "\n",
    "    # Setup output array and put elements from data into masked positions\n",
    "    out = np.zeros(mask.shape, dtype='object')\n",
    "    out[mask] = np.concatenate(data)\n",
    "    \n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "043411f3-4d8b-43b5-89e2-7a9b5d9e1385",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "_2D_array_with_str = []\n",
    "\n",
    "# The below list store the post Id for each instances (or row in _2D_array). \n",
    "# We have to mention that the index of _2D_array and _Id_array helps us to find the id pattern quickly.\n",
    "_Id_array = []\n",
    "\n",
    "for index1, row in df_w_tf_tags.iterrows():\n",
    "    row_tuple = row.Bugy_py_files\n",
    "    row_id = row.Id\n",
    "    \n",
    "    # row: (some rows have multiple patterns)\n",
    "    # ('unix', [[('estimator', 'train'), ('estimator', '_train_model'), \n",
    "    #            ('estimator', '_train_model_default'), ('estimator', '_call_model_fn'), ('nn_ops', 'sparse_softmax_cross_entropy_with_logits')]])\n",
    "    \n",
    "    if row_tuple[0] is None: continue\n",
    "    \n",
    "    for element in row_tuple[1]:\n",
    "        _Id_array.append(row_id)\n",
    "        element_tmp = element[:]\n",
    "        _2D_array_with_str.append(element_tmp)\n",
    "\n",
    "# _2D_array_pad = numpy_fillna(_2D_array)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0fffa51f-6992-42c2-9cf0-fd1888ae4047",
   "metadata": {},
   "source": [
    "#### Creating a numerical matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb2ab766-8031-4eca-9cef-455c127c6283",
   "metadata": {},
   "source": [
    "https://www.geeksforgeeks.org/python-pandas-factorize/"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e4cbea84-d201-4010-9759-c32f05b80aa9",
   "metadata": {},
   "source": [
    "##### Creating a dictionary based on pairs and assigning a unique number to that"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "8257c9af",
   "metadata": {},
   "outputs": [],
   "source": [
    "# dict pair : int\n",
    "\n",
    "dic = {} \n",
    "specific_val = 1\n",
    "for row in _2D_array_with_str:\n",
    "    for element in row:\n",
    "        if element not in dic:\n",
    "            dic[element] = specific_val\n",
    "            specific_val += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "82ffd8ee",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The number of unique pairs is:  10143\n"
     ]
    }
   ],
   "source": [
    "print(\"The number of unique pairs is: \", max(dic.values()))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dedfa8c9-9b6c-4643-b0b7-6b39e7fe40e4",
   "metadata": {},
   "source": [
    "##### Converting strings on the _2D_array into numbers "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "a2bf0be4-4828-4126-8816-81c1a4a3f4e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# _2D_array_with_str"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "7a556c82-ad7d-43c9-b7f1-c852da489f0e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# _2D_array = list()\n",
    "# _2D_array_str = _2D_array_with_str.copy()\n",
    "# _2D_array = _2D_array_with_str[:]\n",
    "_2D_array = _2D_array_with_str[:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "0ff5062e",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i, row in enumerate(_2D_array):\n",
    "    for j in range(len(row)):\n",
    "        _2D_array[i][j] = dic[_2D_array[i][j]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "b3971037-915c-4994-9107-5b36a9d62e33",
   "metadata": {},
   "outputs": [],
   "source": [
    "# _2D_array_with_str"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "ec5559a2",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# _2D_array[0:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "f97135b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# uniqe_dic int : pair\n",
    "uniqe_dic = dict([(value, key) for key, value in dic.items()])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a621bac-8178-438b-8464-73225fe218a7",
   "metadata": {},
   "source": [
    "By the below function we deleted duplicated rows in 2d_array:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "ca3c46e1-7f42-4369-8ea3-6a25dbadc990",
   "metadata": {},
   "outputs": [],
   "source": [
    "def find_2d_array_dup(_2d_array: list) -> Tuple[int, list]:\n",
    "    \n",
    "    dup_count = 0\n",
    "    _2d_array_tmp = _2d_array.copy()\n",
    "    _2d_array_tmp_next = []\n",
    "    \n",
    "    idx = 0\n",
    "    while idx != len(_2d_array_tmp):\n",
    "        \n",
    "        for indx_previous in range(0, idx+1):\n",
    "            _2d_array_tmp_next.append(_2d_array_tmp[indx_previous])\n",
    "        \n",
    "        for indx_next in range(idx+1, len(_2d_array_tmp)):\n",
    "#             print(idx, indx_next)\n",
    "#             print(_2d_array_tmp)\n",
    "#             print(_2d_array_tmp_next)\n",
    "\n",
    "            if _2d_array_tmp[idx] == _2d_array_tmp[indx_next]:\n",
    "                dup_count += 1\n",
    "            else:\n",
    "                _2d_array_tmp_next.append(_2d_array_tmp[indx_next])\n",
    "#         print(_2d_array_tmp_next)\n",
    "#         print(len(_2d_array_tmp_next))\n",
    "#         break\n",
    "        _2d_array_tmp = _2d_array_tmp_next\n",
    "        _2d_array_tmp_next = []\n",
    "        idx += 1\n",
    "    \n",
    "    return dup_count, _2d_array_tmp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "cd20b1db-3f52-4f69-ad94-f91eff4c5a81",
   "metadata": {},
   "outputs": [],
   "source": [
    "# duplication_row, _2D_array_new = find_2d_array_dup(_2D_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "038074a7-a09a-4288-9368-db126c18d2d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(f\"We decresed the number of patterns from {len(_2D_array)} to {len(_2D_array_new)}.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fa324359",
   "metadata": {},
   "source": [
    "## Answering the second research question (RQ2): Finding Patterns"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "31312935",
   "metadata": {},
   "source": [
    "## Contiguous Sequential Pattern Mining"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d2e4c26c",
   "metadata": {},
   "source": [
    "https://www.cc.gatech.edu/~hic/CS7616/pdf/lecture13.pdf"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c8ff498",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Approach (1):"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b6c9bae7",
   "metadata": {},
   "source": [
    "The shortest yet efficient implementation of the famous frequent sequential pattern mining algorithm PrefixSpan, the famous frequent closed sequential pattern mining algorithm BIDE (in closed.py), and the frequent generator sequential pattern mining algorithm FEAT (in generator.py), as a unified and holistic algorithm framework."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96570036",
   "metadata": {},
   "source": [
    "https://github.com/chuanconggao/PrefixSpan-py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "5eaa5c75",
   "metadata": {},
   "outputs": [],
   "source": [
    "# !pip3 install -U prefixspan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "35702190",
   "metadata": {},
   "outputs": [],
   "source": [
    "# from prefixspan import PrefixSpan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "c509adf6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# ps = PrefixSpan(_2D_array_pad)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "df9fc672",
   "metadata": {},
   "outputs": [],
   "source": [
    "# k = ps.frequent(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "b69e8fc1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# file = open(\"/home-students/amghad/sample.txt\", \"w\")\n",
    "# str_list = repr(k)\n",
    "# file.write(str_list)\n",
    "# file.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "34f2979a",
   "metadata": {},
   "source": [
    "### Approach (2): "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "566a8e0a",
   "metadata": {},
   "source": [
    "pymining is a small collection of data mining algorithms implemented in Python. I did not design any of the algorithms, but I use them in my own research so I thought other developers might be interested to use them as well."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "94392e0f",
   "metadata": {},
   "source": [
    "https://github.com/bartdag/pymining"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "411e8d57",
   "metadata": {},
   "source": [
    "### Approach (3): By this approach we simply check all possible status for each row (or pattern)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4ec37188-5c46-407b-b193-55c54b2e1841",
   "metadata": {},
   "source": [
    "We want to create a dict based on the vectors on the numerical matrix."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e2f60f00-0840-41a0-8398-8f863bc2bc68",
   "metadata": {},
   "source": [
    "By the below function, we tried to find and search the patterns for each vector based on the window_size"
   ]
  },
  {
   "cell_type": "raw",
   "id": "0d8aae18-d1d3-4282-9bf7-f7fe45ca5d6b",
   "metadata": {},
   "source": [
    "Input Vector: [1, 2, 6, 3]"
   ]
  },
  {
   "cell_type": "raw",
   "id": "822ce776-1a3b-476d-92e6-be663d91ff05",
   "metadata": {},
   "source": [
    "{0: [(2, [(1, 2), (2, 6), (6, 3)]),\n",
    "    (3, [(1, 2, 6), (2, 6, 3)]),\n",
    "    (4, [(1, 2, 6, 3)])]\n",
    " ...}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9233612b-f3b4-4328-88ed-7843b13d0957",
   "metadata": {},
   "source": [
    "0: The index of vector <br>\n",
    "list: Contains tuples with different windows_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "03fb9196",
   "metadata": {},
   "outputs": [],
   "source": [
    "def append_to_dicc_new(dicc: dict, vec_id: int, list_: list) -> dict:\n",
    "    if vec_id not in dicc:\n",
    "        # key not exist\n",
    "        dicc[vec_id] = list_\n",
    "    else:\n",
    "        # key exist\n",
    "        for tuple_ in list_:\n",
    "            for item in dicc[vec_id]:\n",
    "                if item[0] == tuple_[0]:\n",
    "                    item[1].append(tuple_[1][0])\n",
    "                    break\n",
    "    return dicc"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2a2a340-fad0-4464-a15b-01f01995d3b0",
   "metadata": {},
   "source": [
    "The \"low\" parameter can help us control the lower bound of the window_size."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "1c3380c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "low = 0    # <~~~~~~~ define the lower bound threshold\n",
    "up = len(uniqe_dic)\n",
    "dicc = {}     \n",
    "\n",
    "for i_v, vector in enumerate(_2D_array):    \n",
    "    for index, element in enumerate(vector):\n",
    "        list_ = []\n",
    "        for wind_size in range(low, len(vector)+1-index):\n",
    "            list_.append((wind_size, [(*vector[index:index+wind_size], )]))\n",
    "        dicc = append_to_dicc_new(dicc, i_v, list_)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "88e6b3a4-6f49-462d-aac4-afa5cb3d4725",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8223"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(_2D_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "ddd8bf1e-a7ad-45d8-99ab-a22e4b3f53c9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# dicc.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "f09aefad",
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_duplication(dup_list: list) -> list:\n",
    "    final_list = []\n",
    "    for item in dup_list:\n",
    "        if item not in final_list:\n",
    "            final_list.append(item)\n",
    "        else:\n",
    "            indx = final_list.index(item)\n",
    "            counter = final_list[indx][0] + 1\n",
    "            pair = final_list[indx][1]\n",
    "            final_list.remove(item)\n",
    "            final_list.append((counter, pair))\n",
    "    \n",
    "    return final_list"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a95a41ac-d183-4807-a0d1-fa9328e359cf",
   "metadata": {},
   "source": [
    "dic_count is a dictionary that key is window_size, and value is a list that contains tuples. Each tuple shows a pair and the number of pair's iteration on the matrix."
   ]
  },
  {
   "cell_type": "raw",
   "id": "4befab67-c057-41a7-84c5-aa6b89e363ef",
   "metadata": {},
   "source": [
    "{2: [(1, (5, 1)),\n",
    "  (1, (1, 6)),\n",
    "  (1, (1, 2)),\n",
    "  (1, (2, 6)),\n",
    "  (2, (6, 3)),\n",
    "  (1, (6, 1)),\n",
    "  (2, (1, 3)),\n",
    "  (3, (4, 5)),\n",
    "  (5, (3, 4)),\n",
    "  (2, (4, 3)),\n",
    "  (1, (3, 6)),\n",
    "  (1, (2, 3)),\n",
    "  (1, (3, 1)),\n",
    "  (2, (1, 4)),\n",
    "  (1, (4, 6))],\n",
    "  ...}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "id": "9ef3ffab",
   "metadata": {},
   "outputs": [],
   "source": [
    "dic_count = {}\n",
    "\n",
    "for vec_i, list_tuples in dicc.items():\n",
    "#     print(vec_i, list_tuples,\"\\n\")\n",
    "    for tuples_ in list_tuples:\n",
    "#         print(tuples_)\n",
    "        \n",
    "        if tuples_[0] not in dic_count:\n",
    "            lst_ = [(1, tuple_) for tuple_ in tuples_[1]] \n",
    "            \n",
    "            if len(lst_) != len(set(lst_)):\n",
    "                lst_ = remove_duplication(lst_)\n",
    "            \n",
    "            dic_count[tuples_[0]] = lst_  \n",
    "        else:\n",
    "            #(2, [(1, 3), (1, 4), (3, 4), (4, 5), (5, 1)]\n",
    "            #break\n",
    "            for tuple_ in tuples_[1]:\n",
    "                flag = 0 \n",
    "                for item in dic_count[tuples_[0]]:\n",
    "                    if item[1] == tuple_:\n",
    "                        counter = item[0] + 1\n",
    "                        dic_count[tuples_[0]].remove(item)\n",
    "                        dic_count[tuples_[0]].append((counter, item[1]))\n",
    "                        flag = 1\n",
    "                        break\n",
    "                # tuple is new\n",
    "                if flag == 0:\n",
    "                    dic_count[tuples_[0]].append((1, tuple_))\n",
    "#                 else: \n",
    "#                     print(\"Error100\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85825caa-0ea6-49f4-afb9-467ca6dcd2f8",
   "metadata": {},
   "source": [
    "dic_count is a dictionary that key is window_size, and value is a list that contains tuples. Each tuple shows a pair and the number of pair's iteration on the matrix."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ef1f87b4-36e2-4376-8f3f-ba55164a1b72",
   "metadata": {},
   "source": [
    "For each window_size we count the number of patterns that we found"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "id": "735ff20e-0a61-4acb-b2e4-3c9e4e9886cc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(20660, ()), (20660, ()), (20659, ())]"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dic_count.pop(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "id": "2db34be8-e770-4e6f-a243-1f70b1766461",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 600x400 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "inst_length = dic_count.keys()\n",
    "inst_count = [len(value) for key, value in dic_count.items()]\n",
    "comulative_inst_count = np.cumsum(inst_count)\n",
    "\n",
    "per_comulative_inst_count = []\n",
    "for element in comulative_inst_count:\n",
    "    per_comulative_inst_count.append((element*100)/comulative_inst_count[-1])\n",
    "\n",
    "fig = plt.figure(figsize=(6, 4), dpi=100)\n",
    "ax1 = fig.add_subplot(111)\n",
    "\n",
    "ax1.plot(inst_length, per_comulative_inst_count, '--o', color='#4b8262', linewidth=2)\n",
    "plt.xlabel('Instance\\'s length')\n",
    "plt.ylabel('Cumulative % of stack trace instances')\n",
    "ax1.set_title('tf RQ2')\n",
    "\n",
    "# plt.vlines(x=5, ymin=60, ymax=95, color='red')\n",
    "\n",
    "# plt.text(5.5, 65, 'sentence', rotation=0)\n",
    "ax1.axis(ymin=0, ymax=102)\n",
    "ax1.yaxis.set_major_formatter(mtick.PercentFormatter())\n",
    "\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "id": "42ac2495-df65-4f9b-a88d-c2fc13e51d38",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 700x420 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# import matplotlib.pyplot as plt; plt.rcdefaults()\n",
    "# import numpy as np\n",
    "# import matplotlib.pyplot as plt\n",
    "fig, ax = plt.subplots(figsize =(10, 6), dpi=70)\n",
    "ax.barh(list(inst_length), inst_count, align='center', alpha=0.5, log=True)\n",
    "\n",
    "for i in ax.patches:\n",
    "    plt.text(i.get_width()+0.2, i.get_y()+0.3,\n",
    "             str(round((i.get_width()), 2)),\n",
    "             fontsize = 10, fontweight ='bold',\n",
    "             color ='grey')\n",
    "\n",
    "# Add Plot Title\n",
    "ax.set_title('The frequency of stack trace intances for each instance\\'s length',loc ='center')\n",
    "# ax.set_yticks(y_pos, objects)\n",
    "ax.set_xlabel('Frequency')\n",
    "ax.set_ylabel('Instance\\'s length')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "baa8dc6b-c1d8-45fc-a851-184118ca1a3a",
   "metadata": {},
   "source": [
    "### Approach (4): CC-Span"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "87d8aea6-501b-4a56-98ea-2cdad58ed6dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def append_to_dicc_new(dicc, vec_id, list_):\n",
    "    if vec_id not in dicc:\n",
    "        # key not exist\n",
    "        dicc[vec_id] = list_\n",
    "    else:\n",
    "        # key exist\n",
    "        for tuple_ in list_:\n",
    "            for item in dicc[vec_id]:\n",
    "                if item[0] == tuple_[0]:\n",
    "                    item[1].append(tuple_[1][0])\n",
    "                    break\n",
    "    return dicc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "id": "b8570828-2fb9-4945-b5df-70c17a4ebf81",
   "metadata": {},
   "outputs": [],
   "source": [
    "# _2D_array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "id": "f9370071-2ec9-4fbb-8b33-8e66e38b8c48",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Searching Pattern\n",
    "low = 1     # <~~~~~~~ define the lower bound threshold\n",
    "up = len(uniqe_dic)\n",
    "dicc = {}     \n",
    "\n",
    "for i_v, vector in enumerate(_2D_array):    \n",
    "    for index, element in enumerate(vector):\n",
    "        list_ = []\n",
    "        for wind_size in range(low, len(vector)+1-index):\n",
    "            list_.append((wind_size, [(*vector[index:index+wind_size], )]))\n",
    "        dicc = append_to_dicc_new(dicc, i_v, list_)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "a64a3b0e-ffa9-4999-96cf-5bdc766df1a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Counting\n",
    "def remove_duplication(dup_list):\n",
    "    final_list = []\n",
    "    for item in dup_list:\n",
    "        if item not in final_list:\n",
    "            final_list.append(item)\n",
    "        else:\n",
    "            indx = final_list.index(item)\n",
    "            counter = final_list[indx][0] + 1\n",
    "            pair = final_list[indx][1]\n",
    "            final_list.remove(item)\n",
    "            final_list.append((counter, pair))\n",
    "    \n",
    "    return final_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "ca4b370c-9b01-4e16-a3a6-a7a58bc463e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "dic_count = {}\n",
    "\n",
    "for vec_i, list_tuples in dicc.items():\n",
    "#     print(vec_i, list_tuples,\"\\n\")\n",
    "    for tuples_ in list_tuples:\n",
    "#         print(tuples_)\n",
    "        \n",
    "        if tuples_[0] not in dic_count:\n",
    "            lst_ = [(1, tuple_) for tuple_ in tuples_[1]] \n",
    "            \n",
    "            if len(lst_) != len(set(lst_)):\n",
    "                lst_ = remove_duplication(lst_)\n",
    "            \n",
    "            dic_count[tuples_[0]] = lst_  \n",
    "        else:\n",
    "            #(2, [(1, 3), (1, 4), (3, 4), (4, 5), (5, 1)]\n",
    "            #break\n",
    "            for tuple_ in tuples_[1]:\n",
    "                flag = 0 \n",
    "                for item in dic_count[tuples_[0]]:\n",
    "                    if item[1] == tuple_:\n",
    "                        counter = item[0] + 1\n",
    "                        dic_count[tuples_[0]].remove(item)\n",
    "                        dic_count[tuples_[0]].append((counter, item[1]))\n",
    "                        flag = 1\n",
    "                        break\n",
    "                # tuple is new\n",
    "                if flag == 0:\n",
    "                    dic_count[tuples_[0]].append((1, tuple_))\n",
    "#                 else: \n",
    "#                     print(\"Error100\")\n",
    "#     print(\"salam\", dic_count[2], \"\\n\\n\" )\n",
    "#     if vec_i == 1:\n",
    "#         break\n",
    "\n",
    "# _2d_array = [[1, 3, 4, 3, 4, 5, 1, 4],\n",
    "#              [1, 6, 3, 4, 5],\n",
    "#              [1, 2, 6, 3],\n",
    "#              [6, 1, 3, 4, 5],\n",
    "#              [3, 4, 3, 6],\n",
    "#              [2, 3, 1, 4, 6]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "cb0c9fd7-b3c5-4e61-84bd-bd84f14b392a",
   "metadata": {},
   "outputs": [],
   "source": [
    "def is_subsequence(needle: list, haystack: list) -> bool:\n",
    "    # >>> is_subsequence([2, 3, 4], [1, 2, 3, 4, 5, 6])\n",
    "    return any(\n",
    "        haystack[i:i+len(needle)] == needle\n",
    "        for i in range(len(haystack) - len(needle) + 1)\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "777e6228-2296-47e7-bc0a-aa9f658346a8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def CloConSeqGen(dic_count: dict, threshold: int) -> list:\n",
    "    closed_list = []\n",
    "    for key, v_list in dic_count.items():\n",
    "        #print(key, v_list)\n",
    "        for item_tup in v_list:\n",
    "            tmp_flag = False\n",
    "            for key, v_list_search in dic_count.items():\n",
    "                for item_tup_search in v_list_search:\n",
    "                    #print(\"compare with: \", item_tup_search[1])\n",
    "                    if len(item_tup_search[1]) >= len(item_tup[1]) and is_subsequence(item_tup[1], item_tup_search[1]) and item_tup[1] != item_tup_search[1]:\n",
    "                        if item_tup[0] == item_tup_search[0]:\n",
    "                            tmp_flag = True\n",
    "            if not tmp_flag and item_tup[0] > threshold:\n",
    "\n",
    "                closed_list.append(item_tup)\n",
    "    return closed_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "id": "dfbd276c-0735-4163-910b-27bd57d50fa9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161])"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dic_count.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "id": "70d292da-2e31-4ae1-9590-255586e51611",
   "metadata": {},
   "outputs": [],
   "source": [
    "def CloConSeqGen_v2(dic_count: dict) -> list:\n",
    "    closed_list = []\n",
    "    flag_first_step = True\n",
    "    for win_size in list(reversed(list(dic_count))):\n",
    "        # Just add the last or the biggest window_size to the list \n",
    "        if flag_first_step:\n",
    "            flag_first_step = False\n",
    "            for item in dic_count[win_size]:\n",
    "                closed_list.append(item)\n",
    "            continue\n",
    "        \n",
    "        # For other window sizes\n",
    "        for item_tup_new in dic_count[win_size]:\n",
    "            flag_visit = False\n",
    "            for item_tup_old in closed_list:\n",
    "                if item_tup_new[0] == item_tup_old[0] and is_subsequence(item_tup_new[1], item_tup_old[1]):\n",
    "                    flag_visit = True\n",
    "                    break\n",
    "            if not flag_visit:\n",
    "                closed_list.append(item_tup_new)\n",
    "\n",
    "    return len(closed_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "id": "cbc205ce-b533-4d9e-81c6-623cc7d67941",
   "metadata": {},
   "outputs": [],
   "source": [
    "def CloConSeqGen_v3(dic_count: dict, threshold: int) -> list:\n",
    "    closed_list = []\n",
    "    removed_closed_list = []\n",
    "    flag_first_step = True\n",
    "    for win_size in list(reversed(list(dic_count))):\n",
    "        # Just add the last or the biggest window_size to the list \n",
    "        if flag_first_step:\n",
    "            for item in dic_count[win_size]:\n",
    "                if item[0] >= threshold:\n",
    "                    flag_first_step = False\n",
    "                    closed_list.append(item)\n",
    "                else:\n",
    "                    removed_closed_list.append(item)\n",
    "            continue\n",
    "        \n",
    "        # For other window sizes\n",
    "        for item_tup_new in dic_count[win_size]:\n",
    "            flag_visit = False\n",
    "            for item_tup_old in closed_list:\n",
    "                if item_tup_new[0] == item_tup_old[0] and is_subsequence(item_tup_new[1], item_tup_old[1]):\n",
    "                    flag_visit = True\n",
    "                    break  \n",
    "            if (not flag_visit) and (item_tup_new[0] >= threshold):\n",
    "                closed_list.append(item_tup_new)\n",
    "            else:\n",
    "                removed_closed_list.append(item_tup_new)\n",
    "\n",
    "    return closed_list, removed_closed_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "id": "57c0af36-4426-4796-9124-a3fbffa9354d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(CloConSeqGen_v2(dic_count))\n",
    "# print(CloConSeqGen_v3(dic_count, 1))\n",
    "# print(CloConSeqGen_v3(dic_count, 5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "227bf25d-7b45-42cf-818e-d417cf39afd9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(CloConSeqGen_v2(dic_count))\n",
    "# print(CloConSeqGen_v3(dic_count, threshold=1))\n",
    "# print(CloConSeqGen_v3(dic_count, threshold=1))\n",
    "# result, result_remove = CloConSeqGen_v3(dic_count, threshold=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3799f1c-fb15-41b4-b6ab-2aa842758c26",
   "metadata": {},
   "source": [
    "### Prepare data for support and pattern length when support is 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "id": "5ab26c3c-f9c0-489e-ad4c-e0fb63d2ab73",
   "metadata": {},
   "outputs": [],
   "source": [
    "result, result_remove = CloConSeqGen_v3(dic_count, threshold=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "id": "1e419114-043a-4723-87d3-ebd528fd14b0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "12729"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "id": "cdfc9b18-c42e-43cd-b139-327f2393d1c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Find the low and high boundries of CloConSeqGen_v3 result\n",
    "for element in result:\n",
    "    if result.index(element) == 0:\n",
    "        low = element[0] \n",
    "        high = element[0]\n",
    "        continue\n",
    "    if low > element[0]:\n",
    "        low = element[0]\n",
    "    elif high < element[0]:\n",
    "        high = element[0]\n",
    "        \n",
    "sorted_result = sorted(result, key=lambda tup: tup[0])\n",
    "\n",
    "sup_x = sorted(set([element[0] for element in result]))\n",
    "len_sup_y = []\n",
    "\n",
    "for sup in sup_x:\n",
    "    len_sup_y.append(len(sorted_result) - [y[0] for y in sorted_result].index(sup))\n",
    "    \n",
    "per_len_sup_y = [(x*100)/len_sup_y[0] for x in len_sup_y]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "id": "b265415b-7e33-4fc0-a345-ddf1311c318d",
   "metadata": {},
   "outputs": [],
   "source": [
    "dic_tf_sup = {\"x\":sup_x, \"y\":per_len_sup_y}\n",
    "pickle.dump(dic_tf_sup, open(\"./Pickle_data/dic_tf_sup.p\", \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "id": "349f7306-1cd3-4aab-9234-8d0fe2357663",
   "metadata": {},
   "outputs": [],
   "source": [
    "win_size_list = []\n",
    "for element in result:\n",
    "    win_size_list.append(len(element[1]))\n",
    "win_size_set = set(win_size_list)\n",
    "counter=collections.Counter(win_size_list)\n",
    "\n",
    "counter = dict(sorted(counter.items()))\n",
    "comulative_y = np.cumsum(list(counter.values()))\n",
    "\n",
    "per_list_patterns = []\n",
    "for element in comulative_y:\n",
    "    per_list_patterns.append((element*100)/comulative_y[-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "id": "8a80433a-3525-44d4-a1fc-77b8d29f6cff",
   "metadata": {},
   "outputs": [],
   "source": [
    "dic_tf_win = {\"x\":list(counter.keys()), \"y\":per_list_patterns}\n",
    "pickle.dump(dic_tf_win, open( \"./Pickle_data/dic_tf_win.p\", \"wb\" ) )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cb0f9a98-cb63-463f-98c7-d7a6eed48439",
   "metadata": {},
   "source": [
    "### Prepare data for support and pattern length when support is 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "id": "41cfdb45-6d41-4900-bf23-c26d2cfeee81",
   "metadata": {},
   "outputs": [],
   "source": [
    "result, result_remove = CloConSeqGen_v3(dic_count, threshold=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "id": "827b1c72-a704-40d2-9727-01c506a5fdc1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8090"
      ]
     },
     "execution_count": 292,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "id": "db59f4bc-ee0d-4955-9963-577bc063d2e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def convert_string_to_num(pattern: tuple, unique_dic: dict) -> tuple:\n",
    "    def getKey(dct, value):\n",
    "        return [key for key in dct if (dct[key] == value)]\n",
    "    num_tuple = ()\n",
    "    for item in pattern:\n",
    "        print(getKey(unique_dic, item))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "id": "c878ad83-8e8d-4db3-99d1-125ecc027389",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def convert_num_string(input_list: list, unique_dic: dict) -> list:\n",
    "    convert_lst = []\n",
    "    for pattern in input_list:\n",
    "        string_pattern_tuple = ()\n",
    "        for item in pattern[1]:\n",
    "            string_pattern_tuple = string_pattern_tuple + (unique_dic[item],)\n",
    "        convert_lst.append((pattern[0], string_pattern_tuple))\n",
    "    return convert_lst"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "id": "f64f434d-e2f5-4c48-84a7-921d0a597bcb",
   "metadata": {},
   "outputs": [],
   "source": [
    "def order_sublist(lst_big: list, lst_small: list) -> bool:\n",
    "    return lst_small in [lst_big[i:len(lst_small)+i] for i in range(len(lst_big))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "id": "4a1bc900-9e16-48bb-81e8-b69b954b937a",
   "metadata": {},
   "outputs": [],
   "source": [
    "result_str = convert_num_string(result, uniqe_dic)\n",
    "assert len(result) == len(result_str), \"It should be the same.\"\n",
    "pattern_str_w_qid = []\n",
    "\n",
    "for pattern in result_str:\n",
    "    for q_id in _Id_array:\n",
    "        # assert not df_w_sklearn_tags.loc[df_w_sklearn_tags['Id'] == q_id].Bugy_py_files.empty, \"It's not possible!\"\n",
    "        exit_flag_q_id = False\n",
    "        \n",
    "        for item in df_w_tf_tags['Bugy_py_files'].to_numpy()[df_w_tf_tags['Id'].to_numpy() == q_id].item()[1]:\n",
    "           \n",
    "            if order_sublist(item, list(pattern[1])):\n",
    "                # pattern_str_w_qid.append((pattern[0], pattern[1]))\n",
    "                \n",
    "                exist = False\n",
    "                for ele in pattern_str_w_qid:\n",
    "                    if ele[1] == pattern[0] and ele[2] == pattern[1]:\n",
    "                        exist = True\n",
    "                        ele[0].append(q_id)\n",
    "                        \n",
    "                        if len(ele[0]) ==  pattern[0]: exit_flag_q_id = True\n",
    "                if not exist:\n",
    "                    pattern_str_w_qid.append((list([q_id]), pattern[0], pattern[1]))\n",
    "                    if pattern[0] == 1: exit_flag_q_id = True\n",
    "        \n",
    "        if exit_flag_q_id: break\n",
    "            \n",
    "    # print(\"\\n\\n\")\n",
    "pickle.dump(pattern_str_w_qid, open(\"./Pickle_data/tf_ccspan_2_result_pure.p\", \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "id": "722f9e7d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_top_patterns(patterns_lst: list, top: int) -> list:\n",
    "    assert top != 0, \"The top value is wrong!\"\n",
    "    top_list = []\n",
    "    support_set = set()\n",
    "    \n",
    "    # Find the support list\n",
    "    for pattern in patterns_lst:\n",
    "        support_set.add(pattern[0])\n",
    "        \n",
    "    top_support_list = sorted(list(support_set))[0:top]\n",
    "    \n",
    "    for pattern in patterns_lst:\n",
    "        if pattern[0] in top_support_list:\n",
    "            top_list.append(pattern)\n",
    "            \n",
    "    return top_list"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a455b92-37e7-4d77-8ea1-4312b0a48faf",
   "metadata": {},
   "source": [
    "The below function based on the dic_count dictionary trys to find x top patterns for each window_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "id": "d07908ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "top_pattern_num = get_top_patterns(result, 100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "id": "3fa153b3-af93-4c74-907f-36fefcce2096",
   "metadata": {},
   "outputs": [],
   "source": [
    "# top_pattern_num"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "id": "7279c556",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def convert_num_string(input_list: list, unique_dic: dict) -> list:\n",
    "    convert_lst = []\n",
    "    for pattern in input_list:\n",
    "        string_pattern_tuple = ()\n",
    "        for item in pattern[1]:\n",
    "            string_pattern_tuple = string_pattern_tuple + (unique_dic[item],)\n",
    "        convert_lst.append((pattern[0], string_pattern_tuple))\n",
    "    return convert_lst"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bf82bce5-7af6-4782-b604-28cfe546b864",
   "metadata": {},
   "source": [
    "The below code converts the number pairs to the string pairs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "id": "d86e89e0-774f-4435-99cf-1b1f6cbbb396",
   "metadata": {},
   "outputs": [],
   "source": [
    "top_pattern_str = convert_num_string(top_pattern_num, uniqe_dic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "id": "f4c8d3ff-a0b4-4464-ba4e-80170401506d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# top_pattern_str"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ffdac07b-dca6-4c30-9d08-786f01535355",
   "metadata": {
    "tags": []
   },
   "source": [
    "#### Find the Id of a post based on its pattern:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 303,
   "id": "314c9c2e-1104-466f-ae09-ffb379fbc9d7",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "assert len(_2D_array_with_str) == len(_2D_array), \"Always should be the same.\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 304,
   "id": "f4fff9b0-4939-474b-9db2-03740e518292",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8223, 7896)"
      ]
     },
     "execution_count": 304,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(_Id_array), len(top_pattern_str)\n",
    "# , _Id_array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "id": "0be49c97-63db-4a8f-8488-7abe33c24ce9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def split(a, n):\n",
    "    k, m = divmod(len(a), n)\n",
    "    return (a[i*k+min(i, m):(i+1)*k+min(i+1, m)] for i in range(n))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 306,
   "id": "e11fdfaf-b53f-49e2-9703-78c2dd7dc831",
   "metadata": {},
   "outputs": [],
   "source": [
    "def find_q_id_based_on_str_pattern(total_IDs: list, pattern_str_lst: list) -> dict:\n",
    "    dict_pattern_and_qid = {}\n",
    "    \n",
    "    for pattern in pattern_str_lst: # 9k\n",
    "        for q_id in total_IDs:  # 5K\n",
    "            # assert not df_w_tf_tags.loc[df_w_tf_tags['Id'] == q_id].Bugy_py_files.empty, \"It's not possible!\"\n",
    "\n",
    "            for item in df_w_tf_tags['Bugy_py_files'].to_numpy()[df_w_tf_tags['Id'].to_numpy() == q_id].item()[1]:\n",
    "\n",
    "                if order_sublist(item, list(pattern[1])):\n",
    "                    pattern_tpl = (*pattern[1], )\n",
    "                    if pattern_tpl not in dict_pattern_and_qid.keys():\n",
    "                        dict_pattern_and_qid[pattern_tpl] = set({q_id})\n",
    "                    else:\n",
    "                        dict_pattern_and_qid[pattern_tpl].add(q_id)\n",
    "    return dict_pattern_and_qid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 307,
   "id": "5e53ed61-1196-48d0-a1ec-b2f84d41c2dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_q_number(dict_pattern_and_qid: dict) -> int:\n",
    "    q_total_set = set({})\n",
    "    \n",
    "    for key, value in dict_pattern_and_qid.items():\n",
    "        # print(value)\n",
    "        q_total_set.update(value)\n",
    "\n",
    "    return len(q_total_set), q_total_set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 308,
   "id": "d2500cf4-0c7d-46c3-a6fb-44154dd88900",
   "metadata": {},
   "outputs": [],
   "source": [
    "result, result_remove = CloConSeqGen_v3(dic_count, threshold=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 309,
   "id": "8f402c21-bec6-4e0b-8fc3-f9ec2706d6f8",
   "metadata": {},
   "outputs": [],
   "source": [
    "result_str = convert_num_string(result, uniqe_dic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 310,
   "id": "8a749e11-a331-4f16-b205-532651b372f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "split_number = 4\n",
    "result_str_split = list(split(result_str, split_number))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 311,
   "id": "bf161b64-0a3b-4984-b727-cafedc37b210",
   "metadata": {},
   "outputs": [],
   "source": [
    "assert len(result_str_split[0]) + len(result_str_split[1]) + len(result_str_split[2]) + len(result_str_split[3]) == len(result_str), \"Should be the same.\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 312,
   "id": "2e86ad8a-4de0-461d-a33e-b27d693069f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "split_result = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "id": "a683105c-75ef-4f92-b165-e2fc09facf84",
   "metadata": {},
   "outputs": [],
   "source": [
    "split_result.append(find_q_id_based_on_str_pattern(_Id_array, result_str_split[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 314,
   "id": "dbf43589-3c94-42e1-93e6-a934e4dab41f",
   "metadata": {},
   "outputs": [],
   "source": [
    "split_result.append(find_q_id_based_on_str_pattern(_Id_array, result_str_split[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 315,
   "id": "c642d477-742c-4000-a113-354abc68edef",
   "metadata": {},
   "outputs": [],
   "source": [
    "split_result.append(find_q_id_based_on_str_pattern(_Id_array, result_str_split[2]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 316,
   "id": "8b5c573a-1585-4e12-83d0-afe25fcc714f",
   "metadata": {},
   "outputs": [],
   "source": [
    "split_result.append(find_q_id_based_on_str_pattern(_Id_array, result_str_split[3]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 317,
   "id": "29632a03-1c9f-4910-be5c-e9c1e7eff078",
   "metadata": {},
   "outputs": [],
   "source": [
    "def mergeDictionary(dict_1, dict_2):\n",
    "    dict_3 = {**dict_1, **dict_2}\n",
    "    for key, value in dict_3.items():\n",
    "        if key in dict_1 and key in dict_2:\n",
    "            dict_3[key] = set.union(value , dict_1[key])\n",
    "            # dict_3[key] = [value , dict_1[key]]\n",
    "    return dict_3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 318,
   "id": "57ad0f84-e0e4-4f88-bf32-63e1bee7ca22",
   "metadata": {},
   "outputs": [],
   "source": [
    "merge_dict = {}\n",
    "for item in split_result:\n",
    "    merge_dict = mergeDictionary(merge_dict, item)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 319,
   "id": "bb4e94b5-596a-45f3-bb5e-b63845f2a9ec",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4"
      ]
     },
     "execution_count": 319,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(split_result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 323,
   "id": "56e48874-3d36-4d67-9b25-fe3f7fb9d87b",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7063"
      ]
     },
     "execution_count": 323,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dict_pattern_and_qid = merge_dict.copy()\n",
    "number, _ = get_q_number(merge_dict)\n",
    "number"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "id": "7c3911b8-9bc7-4f3f-9461-13a4a80afb76",
   "metadata": {},
   "outputs": [],
   "source": [
    "def order_sublist(lst_big: list, lst_small: list) -> bool:\n",
    "    return lst_small in [lst_big[i:len(lst_small)+i] for i in range(len(lst_big))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 325,
   "id": "740851d0-643f-4df4-bfbb-7210fa32266d",
   "metadata": {},
   "outputs": [],
   "source": [
    "lst_q_id_has_pattern = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 326,
   "id": "697f5c70-9188-4b28-912d-72995d2de04e",
   "metadata": {},
   "outputs": [],
   "source": [
    "top_pattern_str_split = list(split(top_pattern_str, 4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 327,
   "id": "4681bbae-51e8-4c12-8dc6-53ed15800d46",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7896"
      ]
     },
     "execution_count": 327,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(top_pattern_str_split[0]) + len(top_pattern_str_split[1]) + len(top_pattern_str_split[2]) + len(top_pattern_str_split[3]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 328,
   "id": "fb5074b1-ea39-4121-841c-b147b04e6cdc",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "for top_pattern in top_pattern_str_split[0]:\n",
    "    # print(\"Pattern(small set): \\t\", top_pattern[1])\n",
    "    for q_id in _Id_array:\n",
    "        # print(q_id)\n",
    "        # assert not df_w_tf_tags.loc[df_w_tf_tags['Id'] == q_id].Bugy_py_files.empty, \"It's pot possible!\"\n",
    "        \n",
    "        for item in df_w_tf_tags['Bugy_py_files'].to_numpy()[df_w_tf_tags['Id'].to_numpy() == q_id].item()[1]:\n",
    "            # print(\"Big set: \\t\", item)\n",
    "            if order_sublist(item, list(top_pattern[1])):\n",
    "                # print(\"Big Set: \\t\", item)\n",
    "                # print(\"Small Set: \\t\", list(top_pattern[1]))\n",
    "                # print(\"\\n\\n\")\n",
    "                lst_q_id_has_pattern.append(q_id)\n",
    "    # print(\"\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6cb54390-f7e5-4eba-ac84-dc5109a95998",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "for top_pattern in top_pattern_str_split[1]:\n",
    "    # print(\"Pattern(small set): \\t\", top_pattern[1])\n",
    "    for q_id in _Id_array:\n",
    "        # print(q_id)\n",
    "        # assert not df_w_tf_tags.loc[df_w_tf_tags['Id'] == q_id].Bugy_py_files.empty, \"It's pot possible!\"\n",
    "        \n",
    "        for item in df_w_tf_tags['Bugy_py_files'].to_numpy()[df_w_tf_tags['Id'].to_numpy() == q_id].item()[1]:\n",
    "            # print(\"Big set: \\t\", item)\n",
    "            if order_sublist(item, list(top_pattern[1])):\n",
    "                # print(\"Big Set: \\t\", item)\n",
    "                # print(\"Small Set: \\t\", list(top_pattern[1]))\n",
    "                # print(\"\\n\\n\")\n",
    "                lst_q_id_has_pattern.append(q_id)\n",
    "    # print(\"\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a7895f91-41d6-49c6-9082-d056bcfabe8d",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "for top_pattern in top_pattern_str_split[2]:\n",
    "    # print(\"Pattern(small set): \\t\", top_pattern[1])\n",
    "    for q_id in _Id_array:\n",
    "        # print(q_id)\n",
    "        # assert not df_w_tf_tags.loc[df_w_tf_tags['Id'] == q_id].Bugy_py_files.empty, \"It's pot possible!\"\n",
    "        \n",
    "        for item in df_w_tf_tags['Bugy_py_files'].to_numpy()[df_w_tf_tags['Id'].to_numpy() == q_id].item()[1]:\n",
    "            # print(\"Big set: \\t\", item)\n",
    "            if order_sublist(item, list(top_pattern[1])):\n",
    "                # print(\"Big Set: \\t\", item)\n",
    "                # print(\"Small Set: \\t\", list(top_pattern[1]))\n",
    "                # print(\"\\n\\n\")\n",
    "                lst_q_id_has_pattern.append(q_id)\n",
    "    # print(\"\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8e3dd412-8345-4e2d-946a-450c9c2f451a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "for top_pattern in top_pattern_str_split[3]:\n",
    "    # print(\"Pattern(small set): \\t\", top_pattern[1])\n",
    "    for q_id in _Id_array:\n",
    "        # print(q_id)\n",
    "        # assert not df_w_tf_tags.loc[df_w_tf_tags['Id'] == q_id].Bugy_py_files.empty, \"It's pot possible!\"\n",
    "        \n",
    "        for item in df_w_tf_tags['Bugy_py_files'].to_numpy()[df_w_tf_tags['Id'].to_numpy() == q_id].item()[1]:\n",
    "            # print(\"Big set: \\t\", item)\n",
    "            if order_sublist(item, list(top_pattern[1])):\n",
    "                # print(\"Big Set: \\t\", item)\n",
    "                # print(\"Small Set: \\t\", list(top_pattern[1]))\n",
    "                # print(\"\\n\\n\")\n",
    "                lst_q_id_has_pattern.append(q_id)\n",
    "    # print(\"\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4e23791a-9002-46e6-aeb8-2d6b65bb6d43",
   "metadata": {},
   "outputs": [],
   "source": [
    "assert lst_q_id_has_pattern, \"It not should be empty.\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6a60bb8a-137e-4bc5-89e3-6bc3c180c48d",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern = df_w_tf_tags.loc[df_w_tf_tags['Id'].isin(lst_q_id_has_pattern)].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e62807f4-5baf-4a4e-9b22-5e5a030e5f27",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern.rename(columns = {'Id':'Q_id', 'CreationDate':'Q_create_time'}, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7ffc07a5-f62c-488d-b873-b22c99368924",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_all.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_night = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fa58c7cb-163b-4c69-a8ef-d51ec98d694f",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_w = pd.merge(df_pattern, df_night, how='outer',left_on=['Q_id'], right_on=['ParentId']).reset_index(drop=True)\n",
    "pd_tmp_w[\"Answer_tup\"] = pd_tmp_w.apply(lambda x: (x.Id, x.CreationDate), axis=1)\n",
    "pd_tmp_new_w = pd_tmp_w.groupby([\"Q_id_x\", \"Q_create_time\", \"ViewCount\", \"CommentCount\", \"Score\", \"AnswerCount\", \"AcceptedAnswerId\"], dropna=False)[\"Answer_tup\"].agg(list).reset_index()\n",
    "df_pattern_ans = pd_tmp_new_w[pd_tmp_new_w.AnswerCount != 0].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cd2e8a96-1d56-4914-b24a-77278e0eb99e",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_ans = df_pattern_ans[df_pattern_ans.AcceptedAnswerId > 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a2e1b206-dfda-47ed-9469-06dc34e4d446",
   "metadata": {},
   "outputs": [],
   "source": [
    "def check_patt_ans(lst_ans_id_1, lst_ans_id_2):\n",
    "    assert set(lst_ans_id_1) == set(lst_ans_id_2), \"We couldn't find all answers.\"\n",
    "    assert len(lst_ans_id_1) == len(lst_ans_id_2), \"The length of answer lists are not the same.\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "73a6b524-d003-41d2-a51d-7b36e1306f17",
   "metadata": {},
   "outputs": [],
   "source": [
    "check_patt_ans(df_pattern_ans.AcceptedAnswerId, df_pattern[df_pattern.AcceptedAnswerId > 0].AcceptedAnswerId)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "73f97050-f8ad-4ac2-8339-7f1addc3f7cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_ans.insert(len(df_pattern_ans.columns), 'First_ans_time', np.nan)\n",
    "df_pattern_ans.insert(len(df_pattern_ans.columns), 'First_acc_ans_time', np.nan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "57cb3a82-8e38-4310-a2ed-1c89601e205d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def assign_the_answers(df: pd.DataFrame) -> pd.DataFrame:\n",
    "    for index1, row_status in df.iterrows():   \n",
    "        flag = 0\n",
    "        fr_time = pd.to_datetime(datetime.datetime.now())\n",
    "        acc_time = pd.to_datetime(datetime.datetime.now())\n",
    "\n",
    "        if not row_status[\"Answer_tup\"]:\n",
    "            pd_tmp_new3_w.at[index1,'First_ans_time'] = np.nan\n",
    "            pd_tmp_new3_w.at[index1,'First_acc_ans_time'] = np.nan\n",
    "            continue\n",
    "\n",
    "        for answer in row_status[\"Answer_tup\"]:\n",
    "\n",
    "            anwer_time = pd.to_datetime(answer[1])\n",
    "\n",
    "            if flag == 0:\n",
    "                fr_time = anwer_time\n",
    "                flag = 1\n",
    "\n",
    "            if fr_time > anwer_time:\n",
    "                fr_time = anwer_time\n",
    "\n",
    "            if row_status[\"AcceptedAnswerId\"] == answer[0]:\n",
    "                acc_time = anwer_time\n",
    "\n",
    "        df.at[index1,'First_ans_time'] = fr_time\n",
    "\n",
    "        if pd.isna(row_status[\"AcceptedAnswerId\"]):\n",
    "            acc_time = np.nan\n",
    "\n",
    "        df.at[index1,'First_acc_ans_time'] = acc_time\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "16ceb7d8-c5bc-44eb-815b-b3ebd36a92f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_ans = assign_the_answers(df_pattern_ans)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ce318475-189c-4af8-9566-054f69be6435",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_ans[\"Q_create_time\"]      = pd.to_datetime(df_pattern_ans[\"Q_create_time\"])\n",
    "df_pattern_ans[\"First_acc_ans_time\"] = pd.to_datetime(df_pattern_ans[\"First_acc_ans_time\"])\n",
    "df_pattern_ans[\"First_ans_time\"]     = pd.to_datetime(df_pattern_ans[\"First_ans_time\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d51bf41c-e69c-4e36-bbdd-b95db19b00b9",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_ans[\"Duration_ans\"] = df_pattern_ans.apply(lambda row: (row.First_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_ans_time) else np.nan, axis=1)\n",
    "df_pattern_ans[\"Duration_acc_ans\"] = df_pattern_ans.apply(lambda row: (row.First_acc_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_acc_ans_time) else np.nan, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "af4dc3e2-21f2-421d-8432-9acad1c4fc2e",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_ans.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5661abcc-2375-4a2d-ac90-bed22f468e54",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_result = pd.DataFrame(columns=['Index', 'Q_id', 'Title', 'OS', '# ST-Post', '# ST', 'Link'])\n",
    "\n",
    "for _indx, row in df_pattern.iterrows():\n",
    "    _c = 0\n",
    "    for item in row.Bugy_py_files[1]:\n",
    "        _c += len(item)\n",
    "    _link = \"https://stackoverflow.com/questions/\" + str(row.Q_id)\n",
    "    tmp_dic = {'Index': _indx+1, 'Q_id': row.Q_id, 'Title': row.Title, 'OS': row.Bugy_py_files[0], '# ST-Post': len(row.Bugy_py_files[1]), '# ST': _c, 'Link': _link}\n",
    "    df2 = pd.DataFrame(tmp_dic, index={len(tmp_dic)+1})\n",
    "    df_result = pd.concat([df_result, df2], ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c4cf331b-860e-4f6a-91bc-57c99ab10afb",
   "metadata": {},
   "outputs": [],
   "source": [
    "regex_unix    = r\"[^\\n\\&\\:\\|\\>\\<\\/\\ \\\"\\;\\&]*(\\/[^\\n\\&\\:\\|\\>\\<\\/]+)+\\/([^\\n\\&\\:\\|\\>\\<\\/\\\\]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d*)\\W*in\\s\\W*\\d*([a-zA-Z_$][a-zA-Z_$0-9]*)\"\n",
    "regex_windows = r\"(([a-zA-Z]:)|\\~)\\\\?\\/?([^\\<\\>\\:\\\"\\\\\\|\\?\\*\\n]+\\\\)+([^\\<\\>\\:\\\"\\\\\\/\\/\\|\\?\\*\\n]+)(\\.pyc|\\.py)[^\\d\\n]+(\\d*)\\W*in\\W+([a-zA-Z_$][a-zA-Z_$0-9]*)\"\n",
    "# https://regex101.com/r/5yCAdA/1   Unix\n",
    "# https://regex101.com/r/50Vcmo/1 Windows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b139b576-02c8-4019-9829-c5387d7464aa",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.options.display.max_colwidth = 500\n",
    "df_pattern.loc[df_pattern.Q_id == 54219826].Bugy_py_files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "024b8cd9-ab58-4448-a5f6-15b0fb6cb3f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern.loc[df_pattern.Q_id == 54219826]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5766c065-c694-422e-b157-19a55f52f36b",
   "metadata": {},
   "outputs": [],
   "source": [
    "lst_id_has_trace_wo_err_mess = []\n",
    "c_has_trace   = 0\n",
    "c_has_err_mes = 0\n",
    "\n",
    "def extract_error_messages(id: int, Has_code: bool, Has_trace: bool, code_sec: List) -> Optional[List]:\n",
    "    try:\n",
    "        global lst_id_has_trace_wo_err_mess, c_has_trace, c_has_err_mes\n",
    "        \n",
    "        if Has_trace: c_has_trace += 1\n",
    "        # if not (Has_trace and Has_code): print(f\"Error (2): It's not possiable. Has_code: {Has_code}, Has_trace: {Has_trace}\")\n",
    "            \n",
    "        result_total = []\n",
    "        \n",
    "        for code in code_sec:\n",
    "            # regex = r\"([A-Z]\\w+(Error|Warning|Exception))([^:]*):\\s(.*)$\"\n",
    "            # V3\n",
    "            regex = r\"^([A-Z]\\w+Error)(\\w|\\s|\\(|\\)|\\d)*:\\s(.*)$\"\n",
    "            # regex = r\"([A-Z]\\w+Error):\\s(.*)$\"\n",
    "            code = code.replace(\"\\\\n\", \"\\n\")\n",
    "            code = code.replace('&lt;', '<')\n",
    "            code = code.replace('&gt;', '>')\n",
    "            code = code.replace('&quot;', '\"')\n",
    "            matches = re.finditer(regex, code, re.MULTILINE)\n",
    "            result_each_code_part = []\n",
    "\n",
    "            for matchNum, match in enumerate(matches, start=1):\n",
    "                result_each_code_part.append((match.groups()[0].strip(), match.groups()[2].strip()))\n",
    "            \n",
    "            if result_each_code_part:   # Ignore the empty list\n",
    "                result_total.append(result_each_code_part)\n",
    "\n",
    "        if (not result_total) and Has_trace: \n",
    "            # print(\"Warning!: We have to check this is ID: \", id)\n",
    "            lst_id_has_trace_wo_err_mess.append(id)\n",
    "        \n",
    "        if result_total:\n",
    "            c_has_err_mes += len(result_total)\n",
    "            \n",
    "        return result_total\n",
    "    \n",
    "    except:\n",
    "        print(\"Error(1): \", id)\n",
    "        print(code_sec)\n",
    "        return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ea3133ee-8feb-4741-9663-10ba2d5aae6f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_pattern[\"Err_msg\"] = df_pattern.apply(lambda row: extract_error_messages(row.Q_id, row.Has_code, row.Has_trace, row.Code), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "dcbf43e9-7d3c-485a-9a1f-0a2d5da10956",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"We captured {c_has_trace} stack traces, however {c_has_trace-len(lst_id_has_trace_wo_err_mess)} stacktraces have a one type of error messages. I mean, {len(lst_id_has_trace_wo_err_mess)} stack traces didn't have any error message.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "96c34043-59ca-49ec-add3-929367024323",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_tmp = df_pattern[df_pattern['Err_msg'].map(lambda err_msg_list: len(err_msg_list)) > 0]\n",
    "df_tmp.reset_index(drop=True, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4609acf2-489e-42c5-bd10-dcebd3796e8b",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_err_msg = pd.DataFrame(columns=['Q_id', 'ErrorType', 'ErrorMessage'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f43bcfba-7be1-46f4-95b9-6a44c09d9194",
   "metadata": {},
   "outputs": [],
   "source": [
    "def message_extractor(msg_q_id: str, msg_list: list) -> None:\n",
    "    global df_pattern_err_msg\n",
    "    for msg_t in msg_list:\n",
    "        for message in msg_t:\n",
    "            df_pattern_err_msg = pd.concat([df_pattern_err_msg, pd.DataFrame.from_records([{'Q_id': msg_q_id, 'ErrorType': message[0], 'ErrorMessage': message[1]}])], ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "13526342-f7cf-43a0-b013-5acd281f3038",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "_ = df_tmp.apply(lambda row: message_extractor(row.Q_id, row.Err_msg), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2fbfe0ce-8a87-473e-bc51-b06e22ca6817",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_err_msg.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8b4d6c77-2c1a-462a-bf95-3e7aec229d15",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_err_msg.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b010615a-39a7-4a9b-b211-565e51094f9c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make sure all the elements are string\n",
    "df_pattern_err_msg = df_pattern_err_msg.applymap(str)\n",
    "\n",
    "# Notice: strip() function will remove leading and trailing whitespaces.\n",
    "df_pattern_err_msg['ErrorType'] = df_pattern_err_msg['ErrorType'].apply(lambda x: x.strip())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "898083b2-8efa-4536-be80-156b6925b254",
   "metadata": {},
   "outputs": [],
   "source": [
    "TOP = 10\n",
    "df_tmp = df_pattern_err_msg.groupby(['ErrorType'])['ErrorType'].count().reset_index(name='count').sort_values(['count'], ascending=False)\n",
    "df2 = df_tmp.iloc[:TOP]\n",
    "df2 = df2.append({'ErrorType': 'Other', 'count': df_tmp['count'].iloc[TOP:].sum()}, ignore_index=True)\n",
    "colors = ['tab:blue', 'tab:cyan', 'tab:gray', 'tab:orange', 'tab:red', 'burlywood', 'y']\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(8, 8))\n",
    "\n",
    "patches, texts, pcts =  ax.pie(df2.set_index('ErrorType')['count'],\n",
    "                               labels=df2['ErrorType'],\n",
    "                                autopct='%1.1f%%', \n",
    "                                pctdistance=0.7, \n",
    "                                labeldistance=1.1, \n",
    "                                # textprops={'fontsize': 12},\n",
    "                                textprops={'size': 'x-large'},\n",
    "                                colors=colors,\n",
    "                                wedgeprops={'linewidth': 3.0, 'edgecolor': 'white'},)\n",
    "\n",
    "s = df2.groupby('ErrorType')['count'].sum().map(lambda x : x).sort_values(ascending = False)\n",
    "labels = [f'{name}, {percentage*100:0.1f}%' for name, percentage in zip(s.index, s / s.sum())]\n",
    "ax.legend(bbox_to_anchor=(1.2, 1), \n",
    "           loc='upper left', \n",
    "           # labels=labels,\n",
    "           title=\"Error Types\",\n",
    "           # labelcolor=colors,\n",
    "           fontsize=14)\n",
    "\n",
    "\n",
    "plt.setp(pcts, color='white', fontweight='bold')\n",
    "ax.set_title(\"tf Error Distribution\", fontsize=18)\n",
    "for i, patch in enumerate(patches):\n",
    "    texts[i].set_color(patch.get_facecolor())\n",
    "    \n",
    "plt.setp(texts, fontweight=600)\n",
    "\n",
    "ax.axis('equal')  \n",
    "# plt.tight_layout()\n",
    "# ax.figure.savefig('piechart.png', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fd039f6a-c399-4022-b617-f3c029525e91",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get the frequency of each message type\n",
    "print(df_pattern_err_msg.groupby(['ErrorType'])['ErrorType'].count().reset_index(name='count').sort_values(['count'], ascending=False).nlargest(20, 'count'))\n",
    "df_pattern_err_msg.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bd469d64-2c15-4fa9-81df-790273d8e48a",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Score median: \", df_pattern.Score.median())\n",
    "print(\"View median: \", df_pattern.ViewCount.median())\n",
    "print(\"Answer_count median: \", df_pattern.AnswerCount.median())\n",
    "print(\"Comment median: \", df_pattern.CommentCount.median())\n",
    "print(\"# Accepted answer:\", df_pattern.AcceptedAnswerId.count(), \"  Total Accepted answer:\", df_pattern.AcceptedAnswerId.shape[0], \"  Percentage of Accepted answer:\", \"{:.2f}\".format(df_pattern.AcceptedAnswerId.count() *100 / df_pattern.AcceptedAnswerId.shape[0]))\n",
    "print(\"LOC median: \", (df_pattern[df_pattern.Has_trace == True].Line_code_win + df_pattern[df_pattern.Has_trace == True].Line_code_uix).median())\n",
    "print(\"LOP median: \", df_pattern.Q_text_words_num.median())\n",
    "print(\"Answer duration median: \", \"{:.2f}\".format(df_pattern_ans.Duration_ans.median()))\n",
    "print(\"Accepted Answer duration median: \", \"{:.2f}\".format(df_pattern_ans.Duration_acc_ans.median()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "130682a1-c135-4957-aac8-e4f9af3deed8",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_pattern_err_msg.Q_id = df_pattern_err_msg.Q_id.astype(int)\n",
    "df_result.Q_id = df_result.Q_id.astype(int)\n",
    "df_result_mrg = pd.merge(df_pattern_err_msg, df_result, on=[\"Q_id\"], how='right')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "84a25154-9b96-4755-a17c-b3ad44a30f13",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "from tabulate import tabulate\n",
    "# print(tabulate(df_pattern_err_msg, headers='keys', tablefmt='psql', showindex=False))\n",
    "# print(tabulate(df_result, headers='keys', tablefmt='psql', showindex=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "da919845-8275-4d7d-9f18-004ae0b6679e",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "from openpyxl import load_workbook\n",
    "\n",
    "\n",
    "def append_df_to_excel(filename, df, sheet_name='Sheet1', startrow=None,\n",
    "                       truncate_sheet=False, \n",
    "                       **to_excel_kwargs):\n",
    "    \"\"\"\n",
    "    Append a DataFrame [df] to existing Excel file [filename]\n",
    "    into [sheet_name] Sheet.\n",
    "    If [filename] doesn't exist, then this function will create it.\n",
    "\n",
    "    @param filename: File path or existing ExcelWriter\n",
    "                     (Example: '/path/to/file.xlsx')\n",
    "    @param df: DataFrame to save to workbook\n",
    "    @param sheet_name: Name of sheet which will contain DataFrame.\n",
    "                       (default: 'Sheet1')\n",
    "    @param startrow: upper left cell row to dump data frame.\n",
    "                     Per default (startrow=None) calculate the last row\n",
    "                     in the existing DF and write to the next row...\n",
    "    @param truncate_sheet: truncate (remove and recreate) [sheet_name]\n",
    "                           before writing DataFrame to Excel file\n",
    "    @param to_excel_kwargs: arguments which will be passed to `DataFrame.to_excel()`\n",
    "                            [can be a dictionary]\n",
    "    @return: None\n",
    "\n",
    "    Usage examples:\n",
    "\n",
    "    >>> append_df_to_excel('d:/temp/test.xlsx', df)\n",
    "\n",
    "    >>> append_df_to_excel('d:/temp/test.xlsx', df, header=None, index=False)\n",
    "\n",
    "    >>> append_df_to_excel('d:/temp/test.xlsx', df, sheet_name='Sheet2',\n",
    "                           index=False)\n",
    "\n",
    "    >>> append_df_to_excel('d:/temp/test.xlsx', df, sheet_name='Sheet2', \n",
    "                           index=False, startrow=25)\n",
    "\n",
    "    (c) [MaxU](https://stackoverflow.com/users/5741205/maxu?tab=profile)\n",
    "    \"\"\"\n",
    "    # Excel file doesn't exist - saving and exiting\n",
    "    if not os.path.isfile(filename):\n",
    "        df.to_excel(\n",
    "            filename,\n",
    "            sheet_name=sheet_name, \n",
    "            startrow=startrow if startrow is not None else 0, \n",
    "            **to_excel_kwargs)\n",
    "        return\n",
    "    \n",
    "    # ignore [engine] parameter if it was passed\n",
    "    if 'engine' in to_excel_kwargs:\n",
    "        to_excel_kwargs.pop('engine')\n",
    "\n",
    "    writer = pd.ExcelWriter(filename, engine='openpyxl', mode='a', if_sheet_exists='replace')\n",
    "\n",
    "    # try to open an existing workbook\n",
    "    writer.book = load_workbook(filename)\n",
    "    \n",
    "    # get the last row in the existing Excel sheet\n",
    "    # if it was not specified explicitly\n",
    "    if startrow is None and sheet_name in writer.book.sheetnames:\n",
    "        startrow = writer.book[sheet_name].max_row\n",
    "\n",
    "    # truncate sheet\n",
    "    if truncate_sheet and sheet_name in writer.book.sheetnames:\n",
    "        # index of [sheet_name] sheet\n",
    "        idx = writer.book.sheetnames.index(sheet_name)\n",
    "        # remove [sheet_name]\n",
    "        writer.book.remove(writer.book.worksheets[idx])\n",
    "        # create an empty sheet [sheet_name] using old index\n",
    "        writer.book.create_sheet(sheet_name, idx)\n",
    "    \n",
    "    # copy existing sheets\n",
    "    writer.sheets = {ws.title:ws for ws in writer.book.worksheets}\n",
    "\n",
    "    if startrow is None:\n",
    "        startrow = 0\n",
    "\n",
    "    # write out the new sheet\n",
    "    df.to_excel(writer, sheet_name, startrow=startrow, **to_excel_kwargs)\n",
    "\n",
    "    # save the workbook\n",
    "    writer.save()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "78596f0f-6163-442d-98ca-a7327e1a0c40",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# df_result_mrg.drop(\"Index\", axis=1, inplace=True)\n",
    "# df_result_mrg.to_excel(\"../Emp_st_p_1.xlsx\", sheet_name='TF') \n",
    "# df_result_mrg.drop(\"Index\", axis=1, inplace=True)\n",
    "append_df_to_excel(\"../Emp_st_p_1.xlsx\", df_result_mrg, sheet_name='tf')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a82bee9d-3046-4102-acbc-52007f116835",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Create covered question plot"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17bc50e6-8398-42c3-afbf-7a25ed41bb95",
   "metadata": {},
   "source": [
    "Create the x-axist: sorted based on the support"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6d60f337-4857-47f0-a31c-6054ad5655e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "result_str = convert_num_string(result, uniqe_dic)\n",
    "len(result_str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "683f3ca3-eb98-41a0-aba8-0c7b2565a49a",
   "metadata": {},
   "outputs": [],
   "source": [
    "result_str_rev_sorted = sorted(result_str, key=lambda tup: tup[0], reverse=True)\n",
    "x_number_of_patterns = list(range(1, len(result_str_rev_sorted)+1))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d07eac31-0c48-44b2-a98a-275ea953e486",
   "metadata": {},
   "source": [
    "Create the y-axist: % of coverd questions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1c2e304f-7c19-4486-ab46-f1e81fda28ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "number_of_unique_questions, set_of_unique_questions = get_q_number(dict_pattern_and_qid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5980fd60-ddce-4585-a887-f133e8886e2d",
   "metadata": {},
   "outputs": [],
   "source": [
    "visited_questions = set()\n",
    "y_axis_peresentage_of_coverage = []\n",
    "for support, pattern in result_str_rev_sorted:    \n",
    "    visited_questions.update(dict_pattern_and_qid[pattern])\n",
    "    if len(visited_questions) < len(set_of_unique_questions):\n",
    "        y_axis_peresentage_of_coverage.append(len(visited_questions)*100/len(set_of_unique_questions))\n",
    "    elif len(visited_questions) > len(set_of_unique_questions):\n",
    "        print(\"Error(100)!\")\n",
    "    elif len(visited_questions) == len(set_of_unique_questions):\n",
    "        y_axis_peresentage_of_coverage.append(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2340c6a3-f99b-4934-9d06-c325c77966e0",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, ax = plt.subplots(figsize=(8, 6), dpi=110)\n",
    "\n",
    "ax.plot(x_number_of_patterns, y_axis_peresentage_of_coverage, '-', color='green', linewidth=2)\n",
    "\n",
    "ax.set_title('TensorFlow', fontname=\"Times New Roman\", fontsize=16)\n",
    "\n",
    "ax.set_xlabel('The number of patterns', labelpad=10, fontname=\"Times New Roman\", fontsize=14)\n",
    "ax.set_xticks(range(1, max(x_number_of_patterns)+1, 200))\n",
    "\n",
    "ax.yaxis.set_major_formatter(mtick.PercentFormatter())\n",
    "ax.set_ylabel('The percentage of covered questions', labelpad=10, fontname=\"Times New Roman\", fontsize=14)\n",
    "ax.set_yticks(range(0, 101, 10))\n",
    "\n",
    "fig.tight_layout()\n",
    "# plt.savefig('sp_le_rq1.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e7e74428-a393-4200-8d1c-138654257da7",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt_dic_tf_covered_ques = {\"x\":x_number_of_patterns, \"y\":y_axis_peresentage_of_coverage}\n",
    "pickle.dump(plt_dic_tf_covered_ques, open(\"./Pickle_data/plt_dic_tf_covered_ques.p\", \"wb\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4b4bc342",
   "metadata": {},
   "source": [
    "## Answer time"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "448de367",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Loading Answer Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6b763dbb",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# pure_ans_data = working_directory_path + \"ansSample.csv\"\n",
    "# path_ans = Path(pure_ans_data)\n",
    "\n",
    "# if path_ans.suffix == \".csv\":\n",
    "#     df_ans = pd.read_csv(path_ans, encoding=encoding)\n",
    "# else:\n",
    "#     raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0daf259d-561a-4487-9908-da7e4da40191",
   "metadata": {},
   "source": [
    "## Create DF based on the questions contain stack traces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "31d5963c",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack = pd.DataFrame(columns=['First_ans_time', 'First_acc_ans_time', 'Answers'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "522f93f4",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack[\"Q_info\"] = Question_with_trace_info"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3f03ba0c",
   "metadata": {},
   "outputs": [],
   "source": [
    "Question_with_trace_info[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b3e2b596",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack[\"Q_id\"]          = df_status_w_stack['Q_info'].apply(lambda q_info: q_info[0])\n",
    "df_status_w_stack[\"Q_create_time\"] = df_status_w_stack['Q_info'].apply(lambda q_info: q_info[1])\n",
    "df_status_w_stack[\"View_count\"]    = df_status_w_stack['Q_info'].apply(lambda q_info: q_info[2])\n",
    "df_status_w_stack[\"Answer_count\"]  = df_status_w_stack['Q_info'].apply(lambda q_info: q_info[3])\n",
    "df_status_w_stack[\"Comment_count\"] = df_status_w_stack['Q_info'].apply(lambda q_info: q_info[4])\n",
    "df_status_w_stack[\"Score\"]         = df_status_w_stack['Q_info'].apply(lambda q_info: q_info[5])\n",
    "df_status_w_stack[\"Accepted_Answer_id\"] = df_status_w_stack['Q_info'].apply(lambda q_info: q_info[6])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "28deb778",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack = df_status_w_stack.drop(['Q_info'], axis='columns')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ba5793bb",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack[\"Q_create_time\"]      = pd.to_datetime(df_status_w_stack[\"Q_create_time\"])\n",
    "df_status_w_stack[\"First_acc_ans_time\"] = pd.to_datetime(df_status_w_stack[\"First_acc_ans_time\"])\n",
    "df_status_w_stack[\"First_ans_time\"]     = pd.to_datetime(df_status_w_stack[\"First_ans_time\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4dfd1bbb",
   "metadata": {},
   "source": [
    "#### Filling the Answers column: A list contains the id and time of answers"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ba7d931-c4d3-492a-8276-abc6de000ea3",
   "metadata": {},
   "source": [
    "We have to prepare and find answers that has sp parrentID, so first we catch the questions and store as table in DB. Before this job we have to create a table with all answrs and apply inner join to those tables:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f4d8093c-3c26-4d31-adab-ad1c4bd66959",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack[\"Q_id\"].to_csv('../code_output_csv/df_tf_w_stack.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "raw",
   "id": "61f7376f-6f99-46ba-ac0e-296c646279bd",
   "metadata": {},
   "source": [
    "SELECT df_tf_w_stack.Q_id, all_results.Id, all_results.CreationDate, all_results.ParentId\n",
    "FROM df_tf_w_stack\n",
    "INNER JOIN all_results\n",
    "ON df_tf_w_stack.Q_id = all_results.ParentId;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "510bf191",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_all.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_night = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fd4de2ff",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# df_status_w_stack['Answers'] = df_status_w_stack.apply(lambda x: [], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ba9770ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "# for index1, row_status in df_status_w_stack.iterrows():\n",
    "#     for index2, row_night in df_night.iterrows():\n",
    "#             if row_night[\"ParentId\"] == row_status[\"Q_id\"]:\n",
    "#                     row_status['Answers'].append((row_night[\"Id\"], row_night[\"CreationDate\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ba9e9b00-b12d-498a-a816-9284be6bf1b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack.head()\n",
    "df_status_w_stack.loc[df_status_w_stack[\"Q_id\"] == 64999781]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b3732d4c-0d2b-4cfa-9cd3-c1ad1472244a",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_w = pd.merge(df_status_w_stack, df_night, how='outer',left_on=['Q_id'], right_on=['ParentId']).reset_index(drop=True)\n",
    "pd_tmp_w[\"Answer_tup\"] = pd_tmp_w.apply(lambda x: (x.Id, x.CreationDate), axis=1)\n",
    "pd_tmp_new_w = pd_tmp_w.groupby([\"Q_id_x\", \"Q_create_time\", \"View_count\", \"Comment_count\", \"Score\", \"Answer_count\", \"Accepted_Answer_id\"], dropna=False)[\"Answer_tup\"].agg(list).reset_index()\n",
    "pd_tmp_new2_w = pd_tmp_new_w[pd_tmp_new_w.Answer_count != 0].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1e42a88a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_status_w_stack[\"Answers\"] = df_status_w_stack[\"Answers\"].apply(lambda answers_list: list(set(answers_list)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d6cbd12e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# def checker_1(i, answers_list, answers_count):\n",
    "#     global counter\n",
    "#     if len(answers_list) != answers_count:\n",
    "#         print(\"Error \", i, len(answers_list), answers_count)\n",
    "\n",
    "# def checker_ans_and_ans_tup(row_indx: int, ans_count: int, ans_list: list) -> None:\n",
    "#     if len(ans_list) != ans_count:\n",
    "#         print(\"Error!: \", row_indx, ans_count, ans_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bff2f339",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# df_status_w_stack.apply(lambda row: checker_1(row.name, row.Answers, row.Answer_count), axis=1)\n",
    "_ = pd_tmp_new2_w.apply(lambda row: checker_ans_and_ans_tup(row.Q_id_x, row.Answer_count, row.Answer_tup), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0735f1d5-381b-454f-b369-0968480dfa99",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new2_w.insert(len(pd_tmp_new2_w.columns), 'First_ans_time', np.nan)\n",
    "pd_tmp_new2_w.insert(len(pd_tmp_new2_w.columns), 'First_acc_ans_time', np.nan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "67d7019b-2ab6-4088-91ac-ad6ae20eceb9",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new3_w = pd_tmp_new2_w.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8a596f22",
   "metadata": {},
   "outputs": [],
   "source": [
    "for index1, row_status in pd_tmp_new3_w.iterrows():   \n",
    "    flag = 0\n",
    "    fr_time = pd.to_datetime(datetime.datetime.now())\n",
    "    acc_time = pd.to_datetime(datetime.datetime.now())\n",
    "        \n",
    "    if not row_status[\"Answer_tup\"]:\n",
    "        pd_tmp_new3_w.at[index1,'First_ans_time'] = np.nan\n",
    "        pd_tmp_new3_w.at[index1,'First_acc_ans_time'] = np.nan\n",
    "        continue\n",
    "\n",
    "    for answer in row_status[\"Answer_tup\"]:\n",
    "        \n",
    "        anwer_time = pd.to_datetime(answer[1])\n",
    "        \n",
    "        if flag == 0:\n",
    "            fr_time = anwer_time\n",
    "            flag = 1\n",
    "        \n",
    "        if fr_time > anwer_time:\n",
    "            fr_time = anwer_time\n",
    "        \n",
    "        if row_status[\"Accepted_Answer_id\"] == answer[0]:\n",
    "            acc_time = anwer_time\n",
    "\n",
    "    pd_tmp_new3_w.at[index1,'First_ans_time'] = fr_time\n",
    "        \n",
    "    if pd.isna(row_status[\"Accepted_Answer_id\"]):\n",
    "        acc_time = np.nan\n",
    "    \n",
    "    pd_tmp_new3_w.at[index1,'First_acc_ans_time'] = acc_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "638a57c5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_status_w_stack.info()\n",
    "# pd_tmp_new3_w.columns\n",
    "pd_tmp_new3_w.loc[pd_tmp_new3_w[\"Q_id_x\"] == 64999781]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "847107e7",
   "metadata": {},
   "source": [
    "Write df_status_w_stack to the CSV file:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4335bdcb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_status_w_stack.to_csv('./status_PT_df_w_stack_with_list_of_answers_col.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4253bbc",
   "metadata": {},
   "source": [
    "Read df_status_w_stack from the CSV file:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "29f760c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_status_w_stack = pd.read_csv('./status_df_w_stack_with_list_of_answers_col.csv', encoding=encoding)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8a424d7c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_status_w_stack[\"Duration_ans\"] = df_status_w_stack.apply(lambda row: (row.First_ans_time-row.Q_create_time).days, axis=1)\n",
    "# df_status_w_stack[\"Duration_acc_ans\"] = df_status_w_stack.apply(lambda row: (row.First_acc_ans_time-row.Q_create_time).days, axis=1)\n",
    "pd_tmp_new3_w[\"Duration_ans\"] = pd_tmp_new3_w.apply(lambda row: (row.First_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_ans_time) else np.nan, axis=1)\n",
    "pd_tmp_new3_w[\"Duration_acc_ans\"] = pd_tmp_new3_w.apply(lambda row: (row.First_acc_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_acc_ans_time) else np.nan, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "415a4f64",
   "metadata": {},
   "outputs": [],
   "source": [
    "# pd_tmp_new3_w['Duration_ans'] = pd_tmp_new3_w['Duration_ans'].fillna(0)\n",
    "# pd_tmp_new3_w['Duration_acc_ans'] = pd_tmp_new3_w['Duration_acc_ans'].fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fecf7b71",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_status_w_stack.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b85e4eaa",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new3_w.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e00d09fa",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "pd_tmp_new3_w.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f19800ec",
   "metadata": {},
   "source": [
    "## Create DF based on the questions do not have stack traces"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "57682814-ce03-41e4-88d5-1ed524f79811",
   "metadata": {
    "tags": []
   },
   "source": [
    "We have to prepare and find answers that has Pytorch parrentID, so first we catch the questions and store as table in DB. Before this job we have to create a table with all answrs and apply inner join to those tables:"
   ]
  },
  {
   "cell_type": "raw",
   "id": "ece49f68-9d28-4236-957e-0a98cd01b9f0",
   "metadata": {
    "tags": []
   },
   "source": [
    "SELECT df_tf_wo_stack.Q_id, all_results.Id, all_results.CreationDate, all_results.ParentId\n",
    "FROM df_tf_wo_stack\n",
    "INNER JOIN all_results\n",
    "ON df_tf_wo_stack.Q_id = all_results.ParentId;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45f26363",
   "metadata": {},
   "outputs": [],
   "source": [
    "pure_data = working_directory_path + \"db_results/\" + \"tf_ans_all.csv\"\n",
    "path = Path(pure_data)\n",
    "\n",
    "if path.suffix == \".csv\":\n",
    "    df_night_wo = pd.read_csv(path, encoding=encoding)\n",
    "else:\n",
    "    raise ValueError(\"{data_file_path.suffix} extensions are not supported\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e9c567d8-93cd-488a-9715-2c3b6a011077",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_night_wo.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9f2258f0-f3e4-4a6a-99a1-9413d666c046",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_night_wo.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6195bf92-32fc-45a2-8560-d3e654b35132",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# df_night_wo.drop_duplicates([\"Q_id\", \"Id\", \"ParentId\", \"CreationDate\"], ignore_index=False, inplace=True)\n",
    "df_night_wo = df_night_wo.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "927a0491",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_wo_stack = pd.DataFrame(columns=['First_ans_time', 'First_acc_ans_time', 'Answers'])\n",
    "\n",
    "df_status_wo_stack[\"Q_info\"] = Question_with_wo_trace_info\n",
    "\n",
    "df_status_wo_stack[\"Q_id\"]          = df_status_wo_stack['Q_info'].apply(lambda q_info: q_info[0])\n",
    "df_status_wo_stack[\"Q_create_time\"] = df_status_wo_stack['Q_info'].apply(lambda q_info: q_info[1])\n",
    "df_status_wo_stack[\"View_count\"]    = df_status_wo_stack['Q_info'].apply(lambda q_info: q_info[2])\n",
    "df_status_wo_stack[\"Answer_count\"]  = df_status_wo_stack['Q_info'].apply(lambda q_info: q_info[3])\n",
    "df_status_wo_stack[\"Comment_count\"] = df_status_wo_stack['Q_info'].apply(lambda q_info: q_info[4])\n",
    "df_status_wo_stack[\"Score\"]         = df_status_wo_stack['Q_info'].apply(lambda q_info: q_info[5])\n",
    "df_status_wo_stack[\"Accepted_Answer_id\"] = df_status_wo_stack['Q_info'].apply(lambda q_info: q_info[6])\n",
    "\n",
    "df_status_wo_stack = df_status_wo_stack.drop(['Q_info'], axis='columns')\n",
    "\n",
    "df_status_wo_stack[\"Q_create_time\"] = pd.to_datetime(df_status_wo_stack[\"Q_create_time\"])\n",
    "df_status_wo_stack[\"First_acc_ans_time\"] = pd.to_datetime(df_status_wo_stack[\"First_acc_ans_time\"])\n",
    "df_status_wo_stack[\"First_ans_time\"]     = pd.to_datetime(df_status_wo_stack[\"First_ans_time\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "43a896da-2513-472b-8562-a80b6b4c9ce1",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_status_wo_stack[\"Q_id\"].to_csv('../code_output_csv/df_tf_wo_stack.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bb4a6c56",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_night_wo.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "afe64a63",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_wo_stack.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "02caaef2",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_wo_stack['Answers'] = df_status_wo_stack.apply(lambda x: [], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "850dca9f-f50d-4844-8636-0c9163cb68dd",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_night_wo.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a8006ded-2978-4735-9e8d-1c6f818c5385",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_wo_stack.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "73ca48d7",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# # It is super time-consuming and never run!\n",
    "# for index1, row_status in df_status_wo_stack.iterrows():\n",
    "#     for index2, row_night in df_night_wo.iterrows():\n",
    "#         if row_night[\"ParentId\"] == row_status[\"Q_id\"]:\n",
    "#                 row_status['Answers'].append((row_night[\"Id\"], row_night[\"CreationDate\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "41fd65ef-8327-4d89-b46a-0b6c9e701df8",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_wo = pd.merge(df_status_wo_stack, df_night_wo, how='left',left_on=['Q_id'],right_on=['ParentId']).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c3a4bb1c-c844-4fb1-b291-4a243d589c2a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "pd_tmp_wo[\"Answer_tup\"] = pd_tmp_wo.apply(lambda x: (x.Id, x.CreationDate), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "399421d6-ecc5-41bf-872d-52d39610d1c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# pd_tmp_new = pd_tmp.groupby([\"First_ans_time\", \"First_acc_ans_time\", \"Q_id_x\", \"Q_create_time\", \"View_count\", \"Comment_count\", \"Score\", \"Answer_count\", \"Accepted_Answer_id\"])[\"Answer_tup\"].agg(list).reset_index()\n",
    "pd_tmp_new_wo = pd_tmp_wo.groupby([\"Q_id_x\", \"Q_create_time\", \"View_count\", \"Comment_count\", \"Score\", \"Answer_count\", \"Accepted_Answer_id\"], dropna=False)[\"Answer_tup\"].agg(list).reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d0cf8935-4cfb-4a8f-93ea-8a85965cf76d",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new2_wo = pd_tmp_new_wo[pd_tmp_new_wo.Answer_count != 0].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f5d2daa7-d193-4d18-b974-b71b7d697cfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "# def checker_ans_and_ans_tup(row_indx: int, ans_count: int, ans_list: list) -> None:\n",
    "#     if len(ans_list) != ans_count:\n",
    "#         print(\"Error!: \", row_indx, ans_count, ans_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b2018458-7fd9-4ca3-9da1-5714b03e3e6b",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# We checked all Q_id_x are unique\n",
    "_ = pd_tmp_new2_wo.apply(lambda row: checker_ans_and_ans_tup(row.Q_id_x, row.Answer_count, row.Answer_tup), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1abeea79",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_status_wo_stack[\"Answers\"] = df_status_wo_stack[\"Answers\"].apply(lambda answers_list: list(set(answers_list)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "21fdae5f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# df_status_wo_stack.apply(lambda row: checker_1(row.name, row.Answers, row.Answer_count), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "985c2fbc-1102-4751-bb8f-8d7d484182fd",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_wo_stack.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6f892ae5-e359-4a30-ae5e-255cc23ae8a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new2_wo.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "dcb04af7-e1a0-459f-a6a9-220ec4df1038",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new2_wo.insert(len(pd_tmp_new2_wo.columns), 'First_ans_time', np.nan)\n",
    "pd_tmp_new2_wo.insert(len(pd_tmp_new2_wo.columns), 'First_acc_ans_time', np.nan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "28955ecd-1434-4f76-bac1-6da683f872f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new3_wo = pd_tmp_new2_wo.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ad26bd44",
   "metadata": {},
   "outputs": [],
   "source": [
    "for index1, row_status in pd_tmp_new3_wo.iterrows():   \n",
    "    flag = 0\n",
    "    fr_time = pd.to_datetime(datetime.datetime.now())\n",
    "    acc_time = pd.to_datetime(datetime.datetime.now())\n",
    "        \n",
    "    if not row_status[\"Answer_tup\"]:\n",
    "        pd_tmp_new3_wo.at[index1,'First_ans_time'] = np.nan\n",
    "        pd_tmp_new3_wo.at[index1,'First_acc_ans_time'] = np.nan\n",
    "        continue\n",
    "\n",
    "    for answer in row_status[\"Answer_tup\"]:\n",
    "        \n",
    "        anwer_time = pd.to_datetime(answer[1])\n",
    "        \n",
    "        if flag == 0:\n",
    "            fr_time = anwer_time\n",
    "            flag = 1\n",
    "        \n",
    "        if fr_time > anwer_time:\n",
    "            fr_time = anwer_time\n",
    "        \n",
    "        if row_status[\"Accepted_Answer_id\"] == answer[0]:\n",
    "            acc_time = anwer_time\n",
    "\n",
    "    pd_tmp_new3_wo.at[index1,'First_ans_time'] = fr_time\n",
    "        \n",
    "    if pd.isna(row_status[\"Accepted_Answer_id\"]):\n",
    "        acc_time = np.nan\n",
    "    \n",
    "    pd_tmp_new3_wo.at[index1,'First_acc_ans_time'] = acc_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "59f34b9f",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new3_wo[\"Duration_ans\"] = pd_tmp_new3_wo.apply(lambda row: (row.First_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_ans_time) else np.nan, axis=1)\n",
    "pd_tmp_new3_wo[\"Duration_acc_ans\"] = pd_tmp_new3_wo.apply(lambda row: (row.First_acc_ans_time-row.Q_create_time).total_seconds()/3600 if not pd.isna(row.First_acc_ans_time) else np.nan, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4f1467f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# pd_tmp_new3_wo['Duration_ans'] = pd_tmp_new3_wo['Duration_ans'].fillna(0)\n",
    "# pd_tmp_new3_wo['Duration_acc_ans'] = pd_tmp_new3_wo['Duration_acc_ans'].fillna(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "daac7769",
   "metadata": {},
   "source": [
    "### Find the durarion of answers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6d6e7aaf-49d8-434f-93c2-1a5874830b9f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# We have to delete rows or posts that doesn't have answer, because their zero values affect the meadian value (or the box plot values):\n",
    "# df_status_w_stack_filtered = pd_tmp_new3_wo[pd_tmp_new3_wo[\"Answer_count\"] > 0]\n",
    "# pd_tmp_new3_filtered = pd_tmp_new3_w[pd_tmp_new3_w[\"Answer_count\"] > 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bf33ac9d-d93f-423f-bec9-f1c38644a4c2",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# pd_tmp_new3_w[[\"Q_create_time\", \"First_ans_time\", \"Answer_count\", \"Duration_ans\"]].head(500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d355b03d-d61b-4124-9ce2-7dddc08b253e",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new3_w[[\"Q_create_time\", \"First_ans_time\", \"Answer_count\", \"Duration_ans\"]].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7d041705-edf2-43bf-a88e-45f9d3036f4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_tmp_new3_w.to_csv('./CSV_data/plt_df_tf_w_ans_stack.csv', encoding='utf-8')\n",
    "pd_tmp_new3_wo.to_csv('./CSV_data/plt_df_tf_wo_ans_stack.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c005c133-bf61-4d79-acc3-19f170740a31",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Example:\n",
    "sns.set_theme(style=\"whitegrid\")\n",
    "\n",
    "# ###################################################################################\n",
    "df_1 = pd.DataFrame(columns=['Question Type', 'Hours'])\n",
    "df_1['Hours'] = pd_tmp_new3_w['Duration_ans']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Hours'])\n",
    "df_2['Hours'] = pd_tmp_new3_wo['Duration_ans']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "resultComment = pd.concat([df_1, df_2])\n",
    "resultComment['Type'] = \"First Answer\"\n",
    "# ###################################################################################\n",
    "df_1 = pd.DataFrame(columns=['Question Type', 'Hours'])\n",
    "df_1['Hours'] = pd_tmp_new3_w['Duration_acc_ans']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Hours'])\n",
    "df_2['Hours'] = pd_tmp_new3_wo['Duration_acc_ans']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "resultAnswer = pd.concat([df_1, df_2])\n",
    "resultAnswer['Type'] = \"First Accepted Answer\"\n",
    "# ###################################################################################\n",
    "\n",
    "result = pd.concat([resultComment, resultAnswer], ignore_index=True)\n",
    "\n",
    "fig = plt.figure(figsize=(10, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"Hours\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               # scale=\"count\",\n",
    "               # inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.5)\n",
    "\n",
    "plt.ylim(-5000,10000)\n",
    "# plt.xlim(-1,1)\n",
    "\n",
    "fig.suptitle('Answers Duration (tf Questions)')\n",
    "# fig.suptitle('TensorFlow Questions', prop={\"size\":10})\n",
    "# plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.legend(loc='upper left')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2cb65388",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_1 = pd.DataFrame(columns=['Questions (with stack traces)', 'Hours'])\n",
    "df_1['Hours'] = pd_tmp_new3_w['Duration_ans']\n",
    "df_1['Questions (with stack traces)'] = df_1['Questions (with stack traces)'].apply(lambda x: \"First Answer\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Questions (with stack traces)', 'Hours'])\n",
    "df_2['Hours'] = pd_tmp_new3_w['Duration_acc_ans']\n",
    "df_2['Questions (with stack traces)'] = df_2['Questions (with stack traces)'].apply(lambda x: \"First Accepted Answer\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "\n",
    "# Reset index\n",
    "result.reset_index(inplace=True)\n",
    "\n",
    "fig = plt.figure(figsize=(10, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=result[\"Questions (with stack traces)\"], y=result[\"Hours\"])\n",
    "\n",
    "# plt.ylim(-250,500)\n",
    "\n",
    "# fig = plt.figure(figsize=(10, 5))\n",
    "# gs = fig.add_gridspec(1, 2)\n",
    "\n",
    "# # min_ = min(min(df_status_w_stack['Duration_ans']), min(df_status_w_stack['Duration_acc_ans']))\n",
    "# # max_ = max(max(df_status_w_stack['Duration_ans']), max(df_status_w_stack['Duration_acc_ans']))\n",
    "\n",
    "# ax1 = fig.add_subplot(gs[0, 0])\n",
    "# sns.violinplot(data=df_status_w_stack['Duration_ans'])\n",
    "# ax1.set_xlabel(\"First Answer\")\n",
    "# ax1.set_ylabel(\"Day\")\n",
    "# # ax1 = plt.ylim(min_,max_)\n",
    "\n",
    "# ax2 = fig.add_subplot(gs[0, 1])\n",
    "# sns.violinplot(data=df_status_w_stack['Duration_acc_ans'])\n",
    "# ax2.set_xlabel(\"First Accepted Answer\")\n",
    "# ax2.set_ylabel(\"Day\")\n",
    "# # ax2 = plt.ylim(min_,max_)\n",
    "\n",
    "fig.suptitle('Answers Duration (with stack traces)')\n",
    "\n",
    "fig.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e5b540d9-50f5-4867-be9a-e13e9bd585ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_1 = pd.DataFrame(columns=['Questions (W/O stack traces)', 'Hours'])\n",
    "df_1['Hours'] = pd_tmp_new3_wo['Duration_ans']\n",
    "df_1['Questions (W/O stack traces)'] = df_1['Questions (W/O stack traces)'].apply(lambda x: \"First Answer\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Questions (W/O stack traces)', 'Hours'])\n",
    "df_2['Hours'] = pd_tmp_new3_wo['Duration_acc_ans']\n",
    "df_2['Questions (W/O stack traces)'] = df_2['Questions (W/O stack traces)'].apply(lambda x: \"First Accepted Answer\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "\n",
    "# Reset index\n",
    "result.reset_index(inplace=True)\n",
    "\n",
    "fig = plt.figure(figsize=(10, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=result[\"Questions (W/O stack traces)\"], y=result[\"Hours\"])\n",
    "\n",
    "# plt.ylim(-250,500)\n",
    "\n",
    "fig.suptitle('Answers Duration (W/O stack traces)')\n",
    "\n",
    "fig.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5fd473e",
   "metadata": {},
   "source": [
    "## Statistics Information"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "865c91d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Question Numbers with all ML tags: \", df.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1ac51430",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Question Numbers (with tf tags): \", df_w_tf_tags.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "817104d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Question Numbers (without tf tags): \", df.shape[0] - df_w_tf_tags.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0ada499a",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"tf Question Numbers (with code): \", count__question_w_code)\n",
    "print(\"tf Question Numbers (without code): \", count__question_wo_code)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "356dcb8a",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"tf Question Numbers (with code) have stack trace: \", count_w_t)  # 40 disappeared ?!\n",
    "print(\"tf Question Numbers (with code) that doesn't have stack trace: \",count_wo_t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1cebaf5d",
   "metadata": {
    "scrolled": true,
    "tags": []
   },
   "outputs": [],
   "source": [
    "print(\"tf Question Numbers (with stack trace:) on Unix based systems: \", count_unix)\n",
    "print(\"tf Question Numbers (with stack trace:) on Windows based systems: \", count_win)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3777daa8",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "line_count_unix   = 0\n",
    "line_count_win    = 0\n",
    "line_count_simple = 0\n",
    "\n",
    "for tuple in df_w_tf_tags[\"Line_code_u_w_s\"]:\n",
    "    line_count_unix   += tuple[0]\n",
    "    line_count_win    += tuple[1]\n",
    "    line_count_simple += tuple[2]\n",
    "        \n",
    "print(f\"The total number of line codes in the body part (w/o stack trace) is {line_count_simple}.\")\n",
    "print(f\"The total LOC amongst stack traces is {line_count_unix} (Unix-based reports).\")\n",
    "print(f\"The total LOC amongst stack traces is {line_count_win} (Windows-based reports).\")\n",
    "print(f\"The total LOC amongst stack traces is {line_count_unix+line_count_win}.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "13260028-8d85-4f82-8d22-ca47a7e5b65b",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f\"We found {max(dic.values())} unique pairs in stack traces posts related to the tf frameworks.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ef2d46d5-b096-40ca-83c2-f72897d7114e",
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "# print(\"Pst: \", '{:,}'.format(df_w_tf_tags.shape[0]), \"\\nPst_wc: \", math.ceil(count__question_w_code*100/(count__question_w_code+count__question_wo_code)), \"\\nPst_woc:\", 100-math.ceil(count__question_w_code*100/(count__question_w_code+count__question_wo_code)), \"\\nCB:\", '{:,}'.format(count__num_codes), \"\\nCB_ws:\", math.ceil(count_w_t*100/(count_w_t+count_wo_t)), \"\\nCB_wos:\", 100-(math.ceil(count_w_t*100/(count_w_t+count_wo_t))) , \"\\nST:\", '{:,}'.format(count_w_t), \"\\nU:\", math.ceil(count_unix*100/count_w_t), \"\\nW:\",100-math.ceil(count_unix*100/count_w_t))\n",
    "print(\"Pst: \", '{:,}'.format(df_w_tf_tags.shape[0]), \"\\nPst_wc: \", \n",
    "      '{0:0.2f}'.format(count__question_w_code*100/(count__question_w_code+count__question_wo_code)), \n",
    "      \"\\nPst_woc:\", '{0:0.2f}'.format(100-(count__question_w_code*100/(count__question_w_code+count__question_wo_code))), \n",
    "      \"\\nCB:\", '{:,}'.format(count__num_codes), \n",
    "      \"\\nCB_ws:\", '{0:0.2f}'.format(count_w_t*100/(count_w_t+count_wo_t)), \n",
    "      \"\\nCB_wos:\", '{0:0.2f}'.format(100-(count_w_t*100/(count_w_t+count_wo_t))) , \n",
    "      \"\\nST:\", '{:,}'.format(count_w_t), \n",
    "      \"\\nU:\", '{0:0.2f}'.format(count_unix*100/count_w_t), \n",
    "      \"\\nW:\",'{0:0.2f}'.format(100-(count_unix*100/count_w_t)))\n",
    "\n",
    "print( '{:,}'.format(df_w_tf_tags.shape[0]), \n",
    "      '& {0:0.1f}\\%'.format(count__question_w_code*100/(count__question_w_code+count__question_wo_code)), \n",
    "      '& {0:0.1f}\\%'.format(100-(count__question_w_code*100/(count__question_w_code+count__question_wo_code))), \n",
    "      '& {:,}'.format(count__num_codes), \n",
    "      '& {0:0.1f}\\%'.format(count_w_t*100/(count_w_t+count_wo_t)), \n",
    "      '& {0:0.1f}\\%'.format(100-(count_w_t*100/(count_w_t+count_wo_t))) , \n",
    "      '& \\multicolumn{1}{r:}{\\\\textbf{', count_w_t, '}}',\n",
    "      '& {0:0.1f}\\%'.format(count_w_t*100/df_w_tf_tags.shape[0]), \n",
    "      '& {0:0.1f}\\%'.format(count_unix*100/count_w_t), \n",
    "      '& {0:0.1f}\\%'.format(100-(count_unix*100/count_w_t)))\n",
    "\n",
    "print( '{:,}'.format(df_w_tf_tags.shape[0]), \n",
    "      '& {0:0.1f}\\%'.format(count__question_w_code*100/(count__question_w_code+count__question_wo_code)), \n",
    "      '& {:,}'.format(count__num_codes), \n",
    "      '& {0:0.1f}\\%'.format(count_w_t*100/(count_w_t+count_wo_t)), \n",
    "      '& \\multicolumn{1}{r:}{\\\\textbf{', count_w_t, '}}',\n",
    "      '& {0:0.1f}\\%'.format(count_w_t*100/df_w_tf_tags.shape[0]), \n",
    "      '& {0:0.1f}\\%'.format(count_unix*100/count_w_t), \n",
    "      '& {0:0.1f}\\%'.format(100-(count_unix*100/count_w_t)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3808f478-1be0-46aa-b0e3-81e90bbb5718",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"w-ST:\", '{:,}'.format(count_w_t), \"wo-ST:\", '{:,}'.format(count_wo_t))\n",
    "ACC_W, ACC_WO = 87 , 69\n",
    "print('P1: {0:0.4f}'.format(ACC_W/count_w_t))\n",
    "print('N1: {}'.format(count_w_t))\n",
    "print('P2: {0:0.4f}'.format(ACC_WO/count_wo_t))\n",
    "print('N2: {}'.format(count_wo_t))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f01a48e",
   "metadata": {},
   "source": [
    "## Graphical Results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "034f3688",
   "metadata": {},
   "source": [
    "### LOC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d2cdecdc",
   "metadata": {},
   "outputs": [],
   "source": [
    "code_wo_trace  = df_w_tf_tags[(df_w_tf_tags.Has_code == True) & (df_w_tf_tags.Has_trace == False)].Line_code_simple_code\n",
    "code_w_trace   = df_w_tf_tags[df_w_tf_tags.Has_trace == True].Line_code_win + df_w_tf_tags[df_w_tf_tags.Has_trace == True].Line_code_uix\n",
    "\n",
    "df_1 = pd.DataFrame(columns=['Question Type', 'LOC (in each question post)'])\n",
    "df_1['LOC (in each question post)'] = code_wo_trace\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"code blocks w/o stack trace\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'LOC (in each question post)'])\n",
    "df_2['LOC (in each question post)'] = code_w_trace\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"code blocks with stack trace\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "result['Type'] = \"Question Type\"\n",
    "\n",
    "result.to_csv('./CSV_data/plt_df_loc_tf.csv', encoding='utf-8', index=False)\n",
    "\n",
    "fig = plt.figure(figsize=(5, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"LOC (in each question post)\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               # scale=\"count\",\n",
    "               inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.7)\n",
    "\n",
    "plt.ylim(-15, 100)\n",
    "# plt.xlim(-1, 1)\n",
    "\n",
    "fig.suptitle('Comparing the LOC between regular code and stack trace')\n",
    "\n",
    "plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.plot(legend=None)\n",
    "plt.legend(loc='upper right', prop={\"size\":8})\n",
    "fig.tight_layout()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bdf06607",
   "metadata": {},
   "source": [
    "### Question Length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f198cd0c-00e6-4673-96e0-300b3dc0492c",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_1 = pd.DataFrame(columns=['Question Type', 'Words (in each question post)'])\n",
    "df_1['Words (in each question post)'] = list_num_words_wo_tra\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"w/o stack trace\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Words (in each question post)'])\n",
    "df_2['Words (in each question post)'] = list_num_words_w_tra\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"with stack trace\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "result['Type'] = \"Question Type\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "27aebb4b-37f8-4817-a1d3-a0d462cbad59",
   "metadata": {},
   "outputs": [],
   "source": [
    "result.to_csv('./CSV_data/plt_df_ques_len_tf.csv', encoding='utf-8', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "92381996",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig = plt.figure(figsize=(4, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"Words (in each question post)\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               scale=\"count\",\n",
    "               inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.7)\n",
    "\n",
    "plt.ylim(0, 300)\n",
    "# plt.xlim(-1, 1)\n",
    "\n",
    "fig.suptitle('Comparing the Question Length')\n",
    "\n",
    "plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.plot(legend=None)\n",
    "plt.legend(loc='upper right', prop={\"size\":8})\n",
    "fig.tight_layout()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4bb4fb36",
   "metadata": {},
   "source": [
    "### Plotting Score, View, Answer, Comment Counts "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9593359f-60ab-48a0-b48c-a93d7f61679a",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_status_w_stack.to_csv('./CSV_data/plt_df_tf_w_table3_stack.csv', encoding='utf-8')\n",
    "df_status_wo_stack.to_csv('./CSV_data/plt_df_tf_wo_table3_stack.csv', encoding='utf-8')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8a03529c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_sturges_rule_number(n_obser: int) -> int:\n",
    "    '''\n",
    "    Sturges’ Rule is the most common method for determining the optimal number of bins.\n",
    "\n",
    "    Parameters:\n",
    "    ----------\n",
    "    n_obser : int\n",
    "        The total number of observations in the dataset.\n",
    "\n",
    "    Returns:\n",
    "    ----------\n",
    "    Use ⌈ ⌉ for returning a value: Symbols that mean “ceiling” – i.e. round the answer up to\n",
    "    the nearest integer.\n",
    "    '''\n",
    "    # return round(1 + math.log2(n_obser))\n",
    "    return round(np.ceil(1 + (3.322 * np.log10(n_obser))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "53a4e98d",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=[25,15])\n",
    "plt.subplot(211)\n",
    "\n",
    "plt.title('View Count (With Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nView Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_w_stack['View_count']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_w_stack['View_count'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='purple', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(False)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=18)\n",
    "\n",
    "\n",
    "plt.subplot(212)\n",
    "\n",
    "plt.title('View Count (W/O Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nView Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_wo_stack['View_count']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_wo_stack['View_count'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='purple', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=15)\n",
    "\n",
    "plt.tight_layout(pad=4.0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cb5a9b0f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_1['Frequency'] = df_status_wo_stack['View_count']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_2['Frequency'] = df_status_w_stack['View_count']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "result['Type'] = \"Question Type\"\n",
    "\n",
    "fig = plt.figure(figsize=(5, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"Frequency\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               scale=\"count\",\n",
    "               inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.7)\n",
    "\n",
    "plt.ylim(-2500,13000)\n",
    "# plt.xlim(-1,1)\n",
    "\n",
    "fig.suptitle('View Count')\n",
    "plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.plot(legend=None)\n",
    "plt.legend(loc='upper right', prop={\"size\":10})\n",
    "fig.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2e8319c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=[25,15])\n",
    "\n",
    "plt.subplot(211)\n",
    "\n",
    "plt.title('Answer Count (With Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nAnswer Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_w_stack['Answer_count']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_w_stack['Answer_count'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='g', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=18)\n",
    "\n",
    "plt.subplot(212)\n",
    "\n",
    "plt.title('Answer Count (W/O Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nAnswer Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_wo_stack['Answer_count']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_wo_stack['Answer_count'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='g', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=18)\n",
    "\n",
    "plt.tight_layout(pad=4.0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0f44e9b9",
   "metadata": {},
   "outputs": [],
   "source": [
    "sns.set_theme(style=\"whitegrid\")\n",
    "df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_1['Frequency'] = df_status_wo_stack['Answer_count']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_2['Frequency'] = df_status_w_stack['Answer_count']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "result['Type'] = \"Question Type\"\n",
    "\n",
    "fig = plt.figure(figsize=(5, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"Frequency\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               scale=\"count\",\n",
    "               inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.5)\n",
    "\n",
    "plt.ylim(-.5,4.1)\n",
    "# plt.xlim(-1,1)\n",
    "\n",
    "fig.suptitle('Answer Count')\n",
    "plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.plot(legend=None)\n",
    "plt.legend(loc='upper right', prop={\"size\":10})\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a2f7989a",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=[25,15])\n",
    "\n",
    "plt.subplot(211)\n",
    "\n",
    "plt.title('Comment Count (With Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nComment Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_w_stack['Comment_count']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_w_stack['Comment_count'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='b', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=18)\n",
    "\n",
    "plt.subplot(212)\n",
    "\n",
    "plt.title('Comment Count (W/O Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nComment Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_wo_stack['Comment_count']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_wo_stack['Comment_count'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='b', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=18)\n",
    "\n",
    "plt.tight_layout(pad=4.0)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "46a20d74",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_1['Frequency'] = df_status_wo_stack['Comment_count']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_2['Frequency'] = df_status_w_stack['Comment_count']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "result['Type'] = \"Question Type\"\n",
    "\n",
    "fig = plt.figure(figsize=(5, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"Frequency\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               scale=\"count\",\n",
    "               inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.5)\n",
    "\n",
    "plt.ylim(-1.1,10)\n",
    "# plt.xlim(-1,1)\n",
    "\n",
    "fig.suptitle('Comment Count')\n",
    "plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.plot(legend=None)\n",
    "plt.legend(loc='upper right', prop={\"size\":10})\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5d464c39",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Example:\n",
    "\n",
    "sns.set_theme(style=\"whitegrid\")\n",
    "# ###################################################################################\n",
    "df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_1['Frequency'] = df_status_wo_stack['Comment_count']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_2['Frequency'] = df_status_w_stack['Comment_count']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "resultComment = pd.concat([df_1, df_2])\n",
    "resultComment['Type'] = \"Comment Count\"\n",
    "# ###################################################################################\n",
    "df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_1['Frequency'] = df_status_wo_stack['Answer_count']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_2['Frequency'] = df_status_w_stack['Answer_count']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "resultAnswer = pd.concat([df_1, df_2])\n",
    "resultAnswer['Type'] = \"Answer Count\"\n",
    "# ###################################################################################\n",
    "# df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "# df_1['Frequency'] = df_status_wo_stack['View_count']\n",
    "# df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "# df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "# df_2['Frequency'] = df_status_w_stack['View_count']\n",
    "# df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "# resultView = pd.concat([df_1, df_2])\n",
    "# resultView['Type'] = \"View Count\"\n",
    "# ###################################################################################\n",
    "# df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "# df_1['Frequency'] = df_status_wo_stack['Score']\n",
    "# df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "# df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "# df_2['Frequency'] = df_status_w_stack['Score']\n",
    "# df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "# resultScore = pd.concat([df_1, df_2])\n",
    "# resultScore['Type'] = \"Score Count\"\n",
    "# ###################################################################################\n",
    "result = pd.concat([resultComment, resultAnswer], ignore_index=True)\n",
    "\n",
    "fig = plt.figure(figsize=(10, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"Frequency\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               scale=\"count\",\n",
    "               inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.5)\n",
    "\n",
    "plt.ylim(-1,6)\n",
    "# plt.xlim(-1,1)\n",
    "\n",
    "fig.suptitle('tf Questions')\n",
    "# fig.suptitle('TensorFlow Questions', prop={\"size\":10})\n",
    "# plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.legend(loc='upper left')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1599a8c9",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=[25,15])\n",
    "\n",
    "plt.subplot(211)\n",
    "\n",
    "plt.title('Score Count (With Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nScore Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_w_stack['Score']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_w_stack['Score'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='orange', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=18)\n",
    "\n",
    "plt.subplot(212)\n",
    "\n",
    "plt.title('Score Count (W/O Stack Traces)', fontsize=25)\n",
    "plt.xlabel('\\nScore Range', fontsize=20)\n",
    "plt.ylabel('Frequency', fontsize=20)\n",
    "\n",
    "n_bins = get_sturges_rule_number(len(df_status_wo_stack['Score']))\n",
    "\n",
    "n, bins, patches = plt.hist(df_status_wo_stack['Score'], \n",
    "                             bins=n_bins, \n",
    "                             log=True, \n",
    "                             align='right', \n",
    "                             color='orange', \n",
    "                             edgecolor='black')\n",
    "\n",
    "# define minor ticks and draw a grid with them\n",
    "minor_locator = AutoMinorLocator(2)\n",
    "plt.gca().xaxis.set_minor_locator(minor_locator)\n",
    "# plt.grid(which='minor', color='white', lw = 0.5)\n",
    "# x ticks\n",
    "xticks = [(bins[idx+1] + value)/2 for idx, value in enumerate(bins[:-1])]\n",
    "dist = (xticks[1] - xticks[0])/2\n",
    "xticks = [e+dist for e in xticks]\n",
    "xticks_labels = [ \"{:.1f}\\nto\\n{:.1f}\".format(value, bins[idx+1]) for idx, value in enumerate(bins[:-1])]\n",
    "plt.xticks(xticks, labels = xticks_labels, fontsize=18)\n",
    "\n",
    "plt.tight_layout(pad=4.0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "beea8d59",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_1 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_1['Frequency'] = df_status_wo_stack['Score']\n",
    "df_1['Question Type'] = df_1['Question Type'].apply(lambda x: \"W/O Stack Traces\")\n",
    "\n",
    "df_2 = pd.DataFrame(columns=['Question Type', 'Frequency'])\n",
    "df_2['Frequency'] = df_status_w_stack['Score']\n",
    "df_2['Question Type'] = df_2['Question Type'].apply(lambda x: \"With Stack Traces\")\n",
    "\n",
    "result = pd.concat([df_1, df_2])\n",
    "result['Type'] = \"Question Type\"\n",
    "\n",
    "fig = plt.figure(figsize=(5, 5), dpi=100)\n",
    "\n",
    "sns.violinplot(x=\"Type\", \n",
    "               y=\"Frequency\", \n",
    "               data=result, \n",
    "               hue=\"Question Type\", \n",
    "               palette=\"Set2\", #colorblind\n",
    "               split=True, \n",
    "               scale=\"count\",\n",
    "               inner=\"quartile\",\n",
    "               scale_hue=False,\n",
    "               width=.5)\n",
    "\n",
    "plt.ylim(-7,15)\n",
    "# plt.xlim(-1,1)\n",
    "\n",
    "fig.suptitle('Score Count')\n",
    "plt.gca().axes.get_xaxis().set_visible(False)\n",
    "# plt.plot(legend=None)\n",
    "plt.legend(loc='upper right', prop={\"size\":10})\n",
    "fig.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "00877e55-5557-4103-b09f-a89538247ff4",
   "metadata": {},
   "outputs": [],
   "source": [
    "result2, result_remove2 = CloConSeqGen_v3(dic_count, threshold=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f5c70366-05e3-42b2-86cc-f11621064f88",
   "metadata": {},
   "outputs": [],
   "source": [
    "win_size_list_2 = []\n",
    "for element in result2:\n",
    "    win_size_list_2.append(len(element[1]))\n",
    "win_size_set_2 = set(win_size_list_2)\n",
    "cnt_2=collections.Counter(win_size_list_2)\n",
    "\n",
    "x = np.array(sorted(cnt_2.keys())) # the label locations\n",
    "width = 0.25  # the width of the bars\n",
    "\n",
    "patters_num_2 = []\n",
    "for key in sorted(cnt_2.keys()) :\n",
    "    patters_num_2.append(cnt_2[key])\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(8, 6), dpi=110)\n",
    "\n",
    "x = [str(item) for item in x] \n",
    "\n",
    "rects2 = ax.bar(x, patters_num_2, width, label='Support Threshold = 2')\n",
    "\n",
    "ax.set_title('Tensorflow', fontname=\"Times New Roman\", fontsize=16)\n",
    "ax.legend(prop={'family':'Times New Roman', 'size':10}, facecolor='w', framealpha=1)\n",
    "\n",
    "ax.set_xlabel('Pattern\\'s length', fontname=\"Times New Roman\", fontsize=14)\n",
    "ax.set_xticks(x)\n",
    "\n",
    "ax.set_ylabel('The number of stack trace patterns', labelpad=10, fontname=\"Times New Roman\", fontsize=14)\n",
    "ax.set_yticks(range(0, max(patters_num_2)+1, 50))\n",
    "\n",
    "ax.bar_label(rects2, padding=3)\n",
    "\n",
    "win_size_list = []\n",
    "for element in result2:\n",
    "    win_size_list.append(len(element[1]))\n",
    "win_size_set = set(win_size_list)\n",
    "counter=collections.Counter(win_size_list)\n",
    "\n",
    "counter = dict(sorted(counter.items()))\n",
    "comulative_y = np.cumsum(list(counter.values()))\n",
    "\n",
    "per_list_patterns = []\n",
    "for element in comulative_y:\n",
    "    per_list_patterns.append((element*100)/comulative_y[-1])\n",
    "\n",
    "x2 = counter.keys() \n",
    "x2 = [str(item) for item in x2] \n",
    "\n",
    "ax2 = ax.twinx()\n",
    "ax2.plot(x2, per_list_patterns, '-o', color='green', linewidth=2)\n",
    "ax2.yaxis.set_major_formatter(mtick.PercentFormatter())\n",
    "ax2.set_ylabel('Cumulative % of stack trace patterns', labelpad=10, fontname=\"Times New Roman\", fontsize=14)\n",
    "\n",
    "ax2.grid(axis = 'y', color='green', linestyle='--', linewidth=0.5)\n",
    "# ax2.grid(False)\n",
    "ax.grid(False)\n",
    "\n",
    "fig.tight_layout()\n",
    "# plt.savefig('tf_le_rq1.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bf1e532a-bd58-446d-a0eb-5ee0b8d81bf5",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt_dic_tf_plength_rq1 = {\"x\":x, \"y2\":per_list_patterns, \"y1\": patters_num_2}\n",
    "pickle.dump(plt_dic_tf_plength_rq1, open(\"./Pickle_data/plt_dic_tf_plength_rq1.p\", \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "756060e6-a3a2-4d09-8f92-b4fdbb0d518a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.rcParams[\"font.family\"] = \"sans-serif\"\n",
    "\n",
    "result2, result_remove2 = CloConSeqGen_v3(dic_count, threshold=2)\n",
    "      \n",
    "sorted_result2 = sorted(result2, key=lambda tup: tup[0])\n",
    "\n",
    "sup_x = sorted(set([element[0] for element in result2]))\n",
    "\n",
    "len_sup_y = []\n",
    "lst_supp = [y[0] for y in sorted_result2]\n",
    "\n",
    "cnt = collections.Counter(lst_supp)\n",
    "for item in sup_x:\n",
    "    len_sup_y.append(cnt[item])\n",
    "    \n",
    "comulative_y = np.cumsum(list(cnt.values()))\n",
    "\n",
    "per_len_sup_y = []\n",
    "for element in comulative_y:\n",
    "    per_len_sup_y.append((element*100)/comulative_y[-1])\n",
    "per_len_sup_y = [100-x for x in per_len_sup_y]\n",
    "per_len_sup_y.insert(0, 100)\n",
    "per_len_sup_y.pop()\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(40, 6), dpi=110)\n",
    "sup_x = [str(item) for item in sup_x] \n",
    "\n",
    "rects = ax.bar(sup_x, len_sup_y, width=0.5, label='Support Threshold = 2')\n",
    "\n",
    "ax.set_title('Tensorflow', fontname=\"Times New Roman\", fontsize=16)\n",
    "ax.legend(prop={'family':'Times New Roman', 'size':10}, facecolor='w', framealpha=1)\n",
    "\n",
    "ax.set_xlabel('Support', fontname=\"Times New Roman\", fontsize=14)\n",
    "ax.set_xticks(sup_x)\n",
    "plt.xlim([-1, len(sup_x)])\n",
    "\n",
    "ax.set_ylabel('The number of stack trace patterns', labelpad=10, fontname=\"Times New Roman\", fontsize=14)\n",
    "ax.set_yticks(range(0, max(len_sup_y)+1, 50))\n",
    "\n",
    "ax.bar_label(rects, padding=3)\n",
    "\n",
    "ax2 = ax.twinx()\n",
    "plt.plot(sup_x, per_len_sup_y, '--o', color='orange', linewidth=2)\n",
    "ax2.yaxis.set_major_formatter(mtick.PercentFormatter())\n",
    "ax2.set_ylabel('Cumulative % of stack trace patterns', labelpad=10, fontname=\"Times New Roman\", fontsize=14)\n",
    "\n",
    "ax2.grid(axis = 'y', color='orange', linestyle='--', linewidth=0.5)\n",
    "# ax2.grid(False)\n",
    "ax.grid(False)\n",
    "\n",
    "fig.tight_layout()\n",
    "# plt.savefig('tf_sup_rq1.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5f49fefc-c53a-4b58-9bfa-30ee2dd2207a",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt_dic_tf_sup_rq1 = {\"x\":sup_x, \"y2\":per_len_sup_y, \"y1\": len_sup_y}\n",
    "pickle.dump(plt_dic_tf_sup_rq1, open(\"./Pickle_data/plt_dic_tf_sup_rq1.p\", \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "79e624e0-5237-45af-aa0f-2e71e7b7c9fc",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  },
  "toc-autonumbering": true,
  "toc-showcode": false,
  "toc-showmarkdowntxt": false
 },
 "nbformat": 4,
 "nbformat_minor": 5
}