From 62d466df22feae6f80a6c8f985353ff2dd31748a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?David=20Ord=C3=A1s?=
 <3125580+davorpa@users.noreply.github.com>
Date: Fri, 24 Feb 2023 17:35:13 +0100
Subject: [PATCH] ci(check-urls): use a matrix to speed-up execution (#7060)

* ci(check-urls): use a matrix to speed-up execution

The matrix strategy creates 10 max-parallel workers with a disabled fail-fast parameter

Examples
- Before: https://github.com/davorpa/free-programming-books/actions/runs/2999590952
- After: https://github.com/davorpa/free-programming-books/actions/runs/3007199364

Running the workflow when -langs or -subject files are involved could exhaust the runner execution max-time.

Using this implementation... the elapsed time is of the longest execution time, normally no more than 30min on the worst of cases

* fix: parallel strategy when matrix files.length == 0

* security: apply EbookFoundation/free-programming-books#9287

::set-output --> GITHUB_OUTPUT env file
---
 .github/workflows/check-urls.yml | 100 ++++++++++++++++++++-----------
 1 file changed, 66 insertions(+), 34 deletions(-)

diff --git a/.github/workflows/check-urls.yml b/.github/workflows/check-urls.yml
index bd3152530..f6eb1dc91 100644
--- a/.github/workflows/check-urls.yml
+++ b/.github/workflows/check-urls.yml
@@ -5,6 +5,7 @@ on:
   pull_request:
 
 permissions:
+  # needed for checkout code
   contents: read
 
 # This allows a subsequently queued workflow run to interrupt/wait for previous runs
@@ -13,69 +14,100 @@ concurrency:
   cancel-in-progress: false         # true = interrupt, false = wait
 
 jobs:
-  check-urls:
-    runs-on: ubuntu-latest
-
-    outputs:
-      changed-files: ${{ steps.changed-files.outputs.all_changed_files }}
-
-    steps:
 
 # NOTE: tj-actions/changed-files.
 # For push events you need to include fetch-depth: 0 | 2 depending on your use case.
 #  0: retrieve all history for all branches and tags
 #  1: retrieve only current commit (by default)
 #  2: retrieve until the preceding commit
-      - name: Determine workflow parameters
-        id: init-params
+  get-changed-files:
+    name: Get changed files
+    runs-on: ubuntu-latest
+    outputs:
+      fetch-depth: ${{ steps.set-params.outputs.fetch-depth }}
+      files:       ${{ steps.set-files.outputs.files }}
+      files-len:   ${{ steps.set-files.outputs.files-len }}
+      matrix:      ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - name: Determine workflow params
+        id: set-params
         run: |
           echo "fetch_depth=0" >> $GITHUB_OUTPUT
           if [ "${{ github.event_name }}" == "pull_request" ]; then
             echo "fetch_depth=0" >> $GITHUB_OUTPUT
           fi
-
-      - uses: actions/checkout@v3
+      - name: Checkout
+        uses: actions/checkout@v3
         with:
-          fetch-depth: ${{ steps.init-params.outputs.fetch_depth }}
-
+          fetch-depth: ${{ steps.set-params.outputs.fetch-depth }}
       - name: Get changed files
         id: changed-files
         uses: tj-actions/changed-files@v35.5.5
         with:
           separator: " "
+          json: true
+      - id: set-files
+        run: |
+          echo "${{ steps.changed-files.outputs.all_changed_files }}"  \
+            | jq --raw-output '. | join(" ")'  \
+            | sed -e 's/^/files=/'  \
+            >> $GITHUB_OUTPUT
+          echo "${{ steps.changed-files.outputs.all_changed_files }}"  \
+            | jq --raw-output '. | length'  \
+            | sed -e 's/^/files-len=/'  \
+            >> $GITHUB_OUTPUT
+      - id: set-matrix
+        run: |
+          echo "{\"file\":${{ steps.changed-files.outputs.all_changed_files }}}"  \
+            | sed -e 's/^/matrix=/'  \
+            >> $GITHUB_OUTPUT
+
 
-      - uses: ruby/setup-ruby@v1
+  check-urls:
+    name: Check @ ${{ matrix.file }}
+    if: ${{ fromJSON(needs.get-changed-files.outputs.files-len) > 0 }}
+    needs: [get-changed-files]
+    runs-on: ubuntu-latest
+    strategy:
+      matrix: ${{ fromJSON(needs.get-changed-files.outputs.matrix) }}
+      max-parallel: 10
+      fail-fast: false
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: ${{ needs.get-changed-files.outputs.fetch-depth }}
+      - name: Setup Ruby v2.6
+        uses: ruby/setup-ruby@v1
         with:
           ruby-version: 2.6
-
-      - run: |
+      - name: Install awesome_bot
+        run: |
           gem install awesome_bot
-
-      - name: Check each changed file
+      - name: "Check URLs of file: ${{ matrix.file }}"
         run: |
-          # Set field separator
-          IFS=$' ';
-
-          # Processing loop
-          for file in ${{ steps.changed-files.outputs.all_changed_files }}; do
-            echo;
-            echo "::group::Processing file... $file";
-            awesome_bot "$file" --allow-redirect --allow-dupe --allow-ssl || true;
-            echo "::endgroup::";
-          done
-
-          # Unset field separator
-          unset IFS;
-
+          awesome_bot "${{ matrix.file }}" --allow-redirect --allow-dupe --allow-ssl || true;
       - uses: actions/upload-artifact@v3
         with:
           name: awesomebot-results
           path: ${{ github.workspace }}/ab-results-*.json
 
-      - name: Generate Summary Report using AwesomeBot results
+
+  reporter:
+    name: GitHub report
+    needs: [get-changed-files, check-urls]
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout  # for having the sources of the local action
+        uses: actions/checkout@v3
+        # download and unzip the ab-results-*.json generated by job-matrix: check-urls
+      - uses: actions/download-artifact@v3
+        with:
+          name: awesomebot-results
+      - name: Generate Summary Report
         uses: ./.github/actions/awesomebot-gh-summary-action
         with:
           ab-root: ${{ github.workspace }}
-          files: ${{ steps.changed-files.outputs.all_changed_files }}
+          files: ${{ needs.get-changed-files.outputs.files }}
           separator: " "
           append-heading: ${{ true }}