From 49e9524b32f742c0855e1eed4d6142b7fe268d6d Mon Sep 17 00:00:00 2001
From: "E. G. Patrick Bos" <egpbos@gmail.com>
Date: Sat, 28 Sep 2024 14:58:58 +0200
Subject: [PATCH] update PR link checker workflow

Use a different changed-files checker (the old one was no longer
maintained) and also make the link checking more fine-grained by only
checking added lines for broken links.

We now use a matrix to parallelize over changed files. This slows down
the complete workflow a bit in some cases, but it does offer a nicer
user experience, because the files that are checked will now be shown in
the GitHub UI itself, instead of only in the workflow run logs.

Lychee is installed directly from the GitHub nightly release. This is
really fast. The original reason, though, was because there is no Ubuntu
package for Lychee. Works out well, though.

Determining the base SHA is a bit convoluted, because GitHub does not
provide this in a usable way. See
https://www.kenmuse.com/blog/the-many-shas-of-a-github-pull-request/ for
a good overview of this problem. The git rev-parse command was suggested
there. The fetch depth must be at least two, otherwise the base SHA
cannot be used.
---
 .github/workflows/link-checker-pr.yml | 49 ++++++++++++++++-----------
 _sidebar.md                           |  2 +-
 2 files changed, 30 insertions(+), 21 deletions(-)

diff --git a/.github/workflows/link-checker-pr.yml b/.github/workflows/link-checker-pr.yml
index eff16d3c..4ea60d00 100644
--- a/.github/workflows/link-checker-pr.yml
+++ b/.github/workflows/link-checker-pr.yml
@@ -1,28 +1,37 @@
 name: Link Checker for Pull requests
 on: pull_request
 jobs:
+  changedFiles:
+    runs-on: ubuntu-latest
+    outputs:
+      files: ${{ steps.changed-markdown-files.outputs.all_changed_files }}
+    steps:
+      - uses: actions/checkout@v4
+      - name: Get changed markdown files
+        id: changed-markdown-files
+        uses: tj-actions/changed-files@v45
+        with:
+          # Avoid using single or double quotes for multiline patterns
+          files: |
+             **.md
+          matrix: true
+  
   linkChecker:
     runs-on: ubuntu-latest
+    needs: changedFiles
+    strategy:
+      matrix:
+        file: ${{ fromJSON(needs.changedFiles.outputs.files) }}
+      fail-fast: false
     steps:
-      - uses: actions/checkout@v3
-      - id: changed_files
-        uses: jitterbit/get-changed-files@v1
-      - name: Link Checker
-        uses: lycheeverse/lychee-action@v1.7.0
-        id: lychee
-        env:
-          GITHUB_TOKEN: ${{secrets.TOKEN_GITHUB}}        
+      - uses: actions/checkout@v4
         with:
-          args: ${{ steps.changed_files.outputs.all }}
-      - name: Count broken links
+          fetch-depth: 2
+      - name: download Lychee
+        run: |
+          wget https://github.com/lycheeverse/lychee/releases/download/nightly/lychee-x86_64-unknown-linux-gnu.tar.gz
+          tar xzf lychee-x86_64-unknown-linux-gnu.tar.gz
+      - name: Check all this file's additions for broken links
         run: |
-            broken_max=10
-            broken_count=$(printf "%d" $(grep "🚫 Errors" lychee/out.md | cut -d'|' -f3))
-            if [ "$broken_count" -gt "$broken_max" ]; then
-                echo "Number of broken links (${broken_count}) exceeds maximum allowed number (${broken_max})."
-                cat lychee/out.md
-                exit 1
-            else
-                echo "Number of broken links (${broken_count}) less than or equal to maximum allowed number (${broken_max})."
-                exit 0
-            fi
+          export base_sha=$(git rev-parse ${{ github.sha }}^)
+          git diff -U0 ${base_sha} ${{ github.event.pull_request.head.sha }} -- ${{ matrix.file }} | grep -v "+++" | grep "^+" | cut -c 2- | ./lychee -
diff --git a/_sidebar.md b/_sidebar.md
index df0e2c25..02ab9d1a 100644
--- a/_sidebar.md
+++ b/_sidebar.md
@@ -16,4 +16,4 @@
       * [R](/best_practices/language_guides/r.md)
       * [C and C++](/best_practices/language_guides/ccpp.md)
       * [Fortran](/best_practices/language_guides/fortran.md)
-* [Contributing to this Guide](/CONTRIBUTING.md)
+* [Contributing to this Guide](/CONTRIBUTING.md)
\ No newline at end of file