From 442e55437c411ab3d048c82f38743219a3bd9df3 Mon Sep 17 00:00:00 2001
From: UsamaSadiq <usama7274@gmail.com>
Date: Fri, 27 Oct 2023 20:08:47 +0500
Subject: [PATCH] build: remove jenkins repo health job

---
 devops/jobs/CreateRepoHealthJobs.groovy    | 126 -------------
 devops/resources/run-repo-health-on-org.sh | 203 ---------------------
 2 files changed, 329 deletions(-)
 delete mode 100644 devops/jobs/CreateRepoHealthJobs.groovy
 delete mode 100644 devops/resources/run-repo-health-on-org.sh

diff --git a/devops/jobs/CreateRepoHealthJobs.groovy b/devops/jobs/CreateRepoHealthJobs.groovy
deleted file mode 100644
index d0f53776d..000000000
--- a/devops/jobs/CreateRepoHealthJobs.groovy
+++ /dev/null
@@ -1,126 +0,0 @@
-package devops.jobs
-import static org.edx.jenkins.dsl.Constants.common_logrotator
-import static org.edx.jenkins.dsl.Constants.common_wrappers
-import static org.edx.jenkins.dsl.DevopsTasks.common_read_permissions
-
-
-class CreateRepoHealthJobs{
-    public static def job = { dslFactory, extraVars ->
-        dslFactory.job(extraVars.get("FOLDER_NAME","RepoHealth") + "/org-repo-health-report") {
-
-            logRotator common_logrotator
-            wrappers common_wrappers
-
-            def access_control = extraVars.get('ACCESS_CONTROL',[])
-            access_control.each { acl ->
-                common_read_permissions.each { perm ->
-                    authorization {
-                        permission(perm,acl)
-                    }
-                }
-            }
-
-            def edx_repo_health_gitURL = 'git@github.com:edx/edx-repo-health.git'
-            def destination_repo_health_gitURL = "git@github.com:edx/repo-health-data.git"
-            def testeng_ci_gitURL = 'https://github.com/edx/testeng-ci.git'
-            def repo_tools_gitURL = 'git@github.com:edx/repo-tools.git'
-            
-            description('Generate a report listing repository structure standard compliance accross edX repos')
-            concurrentBuild(false)
-            parameters {
-                stringParam('EDX_REPO_HEALTH_BRANCH', 'master', 'Branch of the edx-repo-health repo to check out.')
-                stringParam('ONLY_CHECK_THIS_REPOSITORY', '', 'If you only want to run repo health on one repository, set this to org/name of said repository.')
-                stringParam('REPORT_DATE','','The date for which repo health data is required.(format: YYYY-MM-DD)')
-            }
-            multiscm {
-                git {
-                    remote {
-                        credentials('edx-secure')
-                        url(edx_repo_health_gitURL )
-                    }
-                    branch('$EDX_REPO_HEALTH_BRANCH')
-                    browser()
-                    extensions {
-                        cleanBeforeCheckout()
-                        relativeTargetDirectory('edx-repo-health')
-                    }
-                }
-                git {
-                    remote {
-                        credentials('edx-secure')
-                        url(testeng_ci_gitURL)
-                    }
-                    branch('master')
-                    extensions {
-                        cleanBeforeCheckout()
-                        relativeTargetDirectory('testeng-ci')
-                    }
-                }
-                git {
-                    remote {
-                        credentials('edx-secure')
-                        url(repo_tools_gitURL)
-                    }
-                    branch('master')
-                    extensions {
-                        cleanBeforeCheckout()
-                        relativeTargetDirectory('repo_tools')
-                    }
-                }
-                git {
-                    remote {
-                        credentials('edx-secure')
-                        url(destination_repo_health_gitURL )
-                    }
-                    branch('master')
-                    browser()
-                    extensions {
-                        wipeOutWorkspace()
-                        cleanBeforeCheckout()
-                        relativeTargetDirectory('repo-health-data')
-                    }
-                }
-            }
-            triggers {
-                cron('H H(4-11) * * 1-5')
-            }
-            wrappers {
-                timeout {
-                    absolute(90)
-                }
-                credentialsBinding {
-                    string('GITHUB_TOKEN', 'GITHUB_REPOHEALTH_STATUS_BOT_TOKEN')
-                    string('READTHEDOCS_API_KEY', 'READTHEDOCS_API_KEY')
-                    file('REPO_HEALTH_GOOGLE_CREDS_FILE', 'REPO_HEALTH_GOOGLE_CREDS_FILE')
-                }
-                timestamps()
-                sshAgent('edx-secure')
-            }
-            environmentVariables {
-                env('REPO_HEALTH_OWNERSHIP_SPREADSHEET_URL', extraVars.get('SPREADSHEET_URL'))
-                env('REPO_HEALTH_REPOS_WORKSHEET_ID', extraVars.get('WORKSHEET_ID'))
-                env('REPORT_DATE', '$REPORT_DATE')
-                env('EDX_REPO_HEALTH_BRANCH', '$EDX_REPO_HEALTH_BRANCH')
-                env('ONLY_CHECK_THIS_REPOSITORY', '$ONLY_CHECK_THIS_REPOSITORY')
-            }
-
-            steps{
-                shell(dslFactory.readFileFromWorkspace('devops/resources/run-repo-health-on-org.sh'))
-            }
-
-            publishers {
-                archiveArtifacts {
-                    pattern('repo-health-data/dashboards/*.csv')  // csv dashboards containing aggregated data
-                }
-                if (extraVars.get('NOTIFY_ON_FAILURE')){
-                        publishers {
-                            mailer(extraVars.get('NOTIFY_ON_FAILURE'), false, false)
-                        }
-                }
-            }
-            
-
-        }
-    }
-}
-
diff --git a/devops/resources/run-repo-health-on-org.sh b/devops/resources/run-repo-health-on-org.sh
deleted file mode 100644
index 4de24158c..000000000
--- a/devops/resources/run-repo-health-on-org.sh
+++ /dev/null
@@ -1,203 +0,0 @@
-#!/bin/bash
-set -e -v
-
-# Click requires this to work cause it interfaces weirdly with python 3 ASCII default
-export LC_ALL=C.UTF-8
-export LANG=C.UTF-8
-
-ORG_NAMES=("edx" "openedx")
-GITHUB_USER_EMAIL="admin+edxstatusbot@edx.org"
-# If the REPORT_DATE variable is set and not an empty string parse the date to standardize it.
-if [[ ! -z $REPORT_DATE ]]; then REPORT_DATE=$(date '+%Y-%m-%d' -d "$REPORT_DATE"); fi
-
-# for unknown reasons, repo-health-data does not come with correct refs
-# git checkout master is necessary to get the refs/heads/master
-cd "${WORKSPACE}/repo-health-data"
-git show-ref
-git checkout master
-git show-ref
-
-###############################
-# Get list of repos in edx org.
-###############################
-
-cd "${WORKSPACE}"
-virtualenv --python=/usr/bin/python3.8 venv -q --clear
-source venv/bin/activate
-
-cd "${WORKSPACE}/repo_tools"
-make install
-pip install -e .
-REPOSIORIES_URLS_FILE="${WORKSPACE}/repositories.txt"
-cat /dev/null > $REPOSIORIES_URLS_FILE
-for ORG_NAME in ${ORG_NAMES[@]}; do
-  get_org_repo_urls "${ORG_NAME}" --url_type https --forks --add_archived --output_file "${REPOSIORIES_URLS_FILE}" --username "${GITHUB_USER_EMAIL}" --token "${GITHUB_TOKEN}" --ignore-repo "clamps-ghsa-c4rq-qwgr-pj5h"
-done
-unset ORG_NAME
-
-deactivate
-
-############################
-# Run checks on repositories
-############################
-
-# Recreate new virtualenv just for testeng-ci script
-
-cd ${WORKSPACE}
-# delete previous venv to make room for new one
-rm -rf venv
-virtualenv --python=/usr/bin/python3.8 venv -q
-source venv/bin/activate
-
-# Install checks and dashboarding script, this should also install pytest-repo-health
-cd edx-repo-health
-pip install -r requirements/base.txt
-pip install -e .
-cd ${WORKSPACE}
-
-# data destination folder setup
-
-METADATA_FILE_DIST="${WORKSPACE}/repo-health-data/docs/checks_metadata.yaml"
-
-failed_repos=()
-
-OUTPUT_FILE_POSTFIX="_repo_health.yaml"
-
-echo
-echo
-# Git clone each repo in org and run checks on it
-input="${REPOSIORIES_URLS_FILE}"
-while IFS= read -r line
-do
-    cd "$WORKSPACE"
-
-    if [[ "$line" =~ ^(git@github\.com:|https://github\.com/)([a-zA-Z0-9_.-]+?)/([a-zA-Z0-9_.-]+?)\.git$ ]]; then
-        ORG_NAME="${BASH_REMATCH[2]}"
-        REPO_NAME="${BASH_REMATCH[3]}"
-        FULL_NAME="$ORG_NAME/$REPO_NAME"
-    else
-        echo "Skipping <$line>: Could not recognize as a GitHub URL in order to extract org and repo name."
-        continue
-    fi
-
-    if [[ "$REPO_NAME" = "edx-repo-health" ]]; then
-        echo "Skipping <$line>: edx-repo health"
-        continue
-    fi
-
-    if [[ -n "$ONLY_CHECK_THIS_REPOSITORY" && "$FULL_NAME" != "$ONLY_CHECK_THIS_REPOSITORY" ]]; then
-        echo "Skipping <$line>: ONLY_CHECK_THIS_REPOSITORY was set, and does not match"
-        continue
-    fi
-
-    echo
-    echo
-    echo "Processing repo: $FULL_NAME"
-
-    rm -rf target-repo
-    git clone -- "${line/https:\/\//https:\/\/$GITHUB_TOKEN@}" target-repo || {
-        failed_repos+=("$FULL_NAME")
-        continue
-    }
-
-    cd target-repo
-
-    # If the REPORT_DATE variable is set and not an empty string.
-    if [[ ! -z $REPORT_DATE ]]
-      then
-        # If a specific date is given for report
-        FIRST_COMMIT=$(git log --reverse --format="format:%ci" | sed -n 1p)
-        if [[ $REPORT_DATE > ${FIRST_COMMIT:0:10} ]]
-          then
-            git checkout `git rev-list -n 1 --before="${REPORT_DATE} 00:00" master`
-          else
-            echo "${REPO_NAME} doesn't have any commits prior to ${REPORT_DATE}"
-            failed_repos+=("$FULL_NAME")
-            continue
-        fi
-    fi
-
-    ORG_DATA_DIR="${WORKSPACE}/repo-health-data/individual_repo_data/${ORG_NAME}"
-    # make sure destination folder exists
-    mkdir -p "$ORG_DATA_DIR"
-
-    OUTPUT_FILE_NAME=${REPO_NAME}${OUTPUT_FILE_POSTFIX}
-    cd ${WORKSPACE}
-    if pytest --repo-health --repo-health-path edx-repo-health --repo-path target-repo --repo-health-metadata "${METADATA_FILE_DIST}" --output-path "${ORG_DATA_DIR}/${OUTPUT_FILE_NAME}" -o log_cli=true --exitfirst --noconftest -v -c /dev/null
-    then
-        true
-    elif pytest --repo-health --repo-health-path edx-repo-health --repo-path target-repo --repo-health-metadata "${METADATA_FILE_DIST}" --output-path "${ORG_DATA_DIR}/${OUTPUT_FILE_NAME}" -o log_cli=true --exitfirst --noconftest -v -c /dev/null
-    # rerun the same command if it fails once
-    then
-        true
-    else
-        failed_repos+=("$FULL_NAME") && continue
-    fi
-done < "$input"
-
-##############################
-# Recalculate aggregated data.
-##############################
-
-# Go into data repo, recalculate aggregate data, and push a PR
-IFS=,
-failed_repo_names=`echo "${failed_repos[*]}"`
-
-# Delete existing dashboards(sqlite) to re compile
-find "${WORKSPACE}/repo-health-data/dashboards" -type f -iname "dashboard*.sql" -delete
-
-# Compiling dashboard with latest yml files from both orgs
-echo "Pushing data"
-cd "${WORKSPACE}/repo-health-data/individual_repo_data"
-ls *
-repo_health_dashboard --data-dir . --configuration "${WORKSPACE}/edx-repo-health/repo_health_dashboard/configuration.yaml" --output-csv "${WORKSPACE}/repo-health-data/dashboards/dashboard"  --output-sqlite "${WORKSPACE}/repo-health-data/dashboards/dashboard"
-
-deactivate
-cd ${WORKSPACE}
-rm -rf venv
-
-# Only commit the data if running with master and no REPORT_DATE is set.
-if [[ ${EDX_REPO_HEALTH_BRANCH} == 'master' && -z ${REPORT_DATE} ]]
-then
-
-  ###########################################
-  # Commit files and push to repo-health-data
-  ###########################################
-  echo "Commit new files and push to master..."
-
-  commit_message="Update repo health data"
-
-  cd "${WORKSPACE}/repo-health-data"
-
-  if [[ ! "${failed_repos}" ]]
-  then
-     commit_message+="\nFollowing repos failed repo health checks\n ${failed_repo_names}"
-
-     for full_name in "${failed_repos[@]}"
-     do
-          OUTPUT_FILE_NAME="${full_name}${OUTPUT_FILE_POSTFIX}"
-          echo "reverting repo health data for ${OUTPUT_FILE_NAME}"
-          git checkout -- "individual_repo_data/${OUTPUT_FILE_NAME}"
-     done
-  fi
-
-  git add --all
-  git status
-  git diff-index --quiet HEAD || git commit -m ${commit_message}
-  git push origin master
-
-fi
-
-if [[ ${#failed_repos[@]} -ne 0 ]]; then
-  echo
-  echo
-  echo "TLDR Runbook(More detailed runbook: https://openedx.atlassian.net/wiki/spaces/AT/pages/3229057351/Repo+Health+Runbook ):"
-  echo "  To resolve, search the console output for 'ERRORS' (without the quotes), or search for any"
-  echo "  of the failed repo names listed below."
-  echo "The following repositories failed while executing pytest repo-health scripts causing the job to fail:"
-  echo
-  echo "    ${failed_repos[*]}"
-  echo
-  echo
-  exit 1
-fi