Merge pull request #2 from mcg1969/filter

Add filtering

project_inspect/__main__.py

@@ -27,9 +27,19 @@
     help="Deliver the output to the named file instead of standard output.",
     action="store")
 parser.add_argument(
-    "--root", '-r',
+    "--root",
     help="Specify the root directory of the project store.",
     action="store")
+parser.add_argument(
+    "--package", action="append",
+    help="""Limit the list/summary to the given package. The argument can
+simply be the name of a package, or a conda-style package/version spec;
+e.g., 'pandas<0.20'. Multiple --package arguments can be supplied.""")
+parser.add_argument(
+    "--package-file", action="store",
+    help="""Read a list of package specs from the given file. One spec
+should be supplied per line. The spec can simply be the name of a package,
+or a conda-style package/version spec; e.g., 'pandas<0.20'.""")
 parser.add_argument(
     "--summarize", '-s',
     help="""Optionally summarize the inventory. Choices include
@@ -64,6 +74,13 @@ def main(**kwargs):
         raise RuntimeError('Must supply --owner with --project')
     else:
         df = project.build_node_inventory(root)
+    packages = kwargs.get('package') or []
+    package_file = kwargs.get('package_file')
+    if package_file:
+        with open(package_file, 'rt') as fp:
+            packages.extend(spec for spec in map(str.strip, fp) if spec)
+    if packages:
+        df = project.filter_data(df, packages)
     summary = kwargs.get('summarize')
     if summary:
         df = project.summarize_data(df, summary)

project_inspect/imports.py

@@ -77,7 +77,7 @@ def find_python_imports(code):
     imports = set()
     for line in map(str.strip, code.splitlines()):
         if not line.startswith('#'):
-            imports.update(find_python_imports(line))
+            imports.update(find_python_imports(line.lstrip()))
     return imports
 
 

project_inspect/project.py

@@ -3,14 +3,17 @@
 from .imports import find_python_imports, find_r_imports
 
 from .utils import logger, load_file, shortpath, set_log_root
+from .version import VersionSpec
 
 from os.path import join, isdir, isfile, basename, dirname, exists, abspath
 from textwrap import TextWrapper
 from glob import glob
 
 import os
+import re
 import json
 import pandas as pd
+import numpy as np
 
 
 def visible_project_environments(project_home):
@@ -223,6 +226,24 @@ def _build_df(records):
     return df
 
 
+def filter_data(df, packages):
+    if not packages:
+        return df
+    mask = np.zeros(len(df), dtype=bool)
+    for package in packages:
+        spec = re.match(r'^([A-Za-z0-9-_.]+)\s*(.*)', package)
+        if not spec:
+            raise RuntimeError('Invalid package spec: {}'.format(package))
+        name, version = spec.groups()
+        t_mask = df['package'] == name
+        if t_mask.any() and version:
+            vspec = VersionSpec(version)
+            t2_mask = df['version'][t_mask].apply(vspec.match).values
+            t_mask[t_mask] = t2_mask
+        mask = mask | t_mask
+    return df[mask]
+
+
 def validate_summarize(level):
     sep = '_' if '_' in level else '/'
     parts = set(level.lower().split(sep))