Stop lychee from causing so many PR failures (#2002)

Author: trask

.github/scripts/.lychee-relative.toml

@@ -0,0 +1,7 @@
+# Check link anchors
+include_fragments = true
+
+# For relative links only, we exclude all external URLs
+exclude = [
+  "^https?://.*",
+]

.lychee.toml renamed to .github/scripts/.lychee.toml

@@ -5,22 +5,51 @@ set -e
 export MSYS_NO_PATHCONV=1 # for Git Bash on Windows
 
 SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
-LYCHEE_CONFIG="$SCRIPT_DIR/../../.lychee.toml"
+ROOT_DIR="$SCRIPT_DIR/../.."
 DEPENDENCIES_DOCKERFILE="$SCRIPT_DIR/dependencies.dockerfile"
 
+# Parse command line arguments
+RELATIVE_ONLY=false
+MODIFIED_FILES=""
+
+while [[ $# -gt 0 ]]; do
+    case $1 in
+        --relative-only)
+            RELATIVE_ONLY=true
+            shift
+            ;;
+        *)
+            # Treat any other arguments as file paths
+            MODIFIED_FILES="$MODIFIED_FILES $1"
+            shift
+            ;;
+    esac
+done
+
 # Extract lychee version from dependencies.dockerfile
 LYCHEE_VERSION=$(grep "FROM lycheeverse/lychee:" "$DEPENDENCIES_DOCKERFILE" | sed 's/.*FROM lycheeverse\/lychee:\([^ ]*\).*/\1/')
 
+# Determine target files/directories and config file
+TARGET="."
+LYCHEE_CONFIG=".github/scripts/.lychee.toml"
+
+if [[ "$RELATIVE_ONLY" == "true" ]]; then
+    LYCHEE_CONFIG=".github/scripts/.lychee-relative.toml"
+fi
+
+if [[ -n "$MODIFIED_FILES" ]]; then
+    TARGET="$MODIFIED_FILES"
+fi
+
 # Build the lychee command with optional GitHub token
-CMD="lycheeverse/lychee:$LYCHEE_VERSION --verbose --config $(basename "$LYCHEE_CONFIG")"
+CMD="lycheeverse/lychee:$LYCHEE_VERSION --verbose --config $LYCHEE_CONFIG"
 
 # Add GitHub token if available
 if [[ -n "$GITHUB_TOKEN" ]]; then
     CMD="$CMD --github-token $GITHUB_TOKEN"
 fi
 
-# Add the target directory
-CMD="$CMD ."
+CMD="$CMD $TARGET"
 
 # Determine if we should allocate a TTY
 DOCKER_FLAGS="--rm --init"
@@ -32,4 +61,4 @@ fi
 
 # Run lychee with proper signal handling
 # shellcheck disable=SC2086
-exec docker run $DOCKER_FLAGS -v "$(dirname "$LYCHEE_CONFIG")":/data -w /data $CMD
+exec docker run $DOCKER_FLAGS -v "$ROOT_DIR":/data -w /data $CMD

.github/scripts/link-check.sh

@@ -8,6 +8,9 @@ on:
   pull_request:
   merge_group:
   workflow_dispatch:
+  schedule:
+    # Run daily at 7:30 AM UTC
+    - cron: '30 7 * * *'
 
 permissions:
   contents: read
@@ -199,3 +202,31 @@ jobs:
             )
           )
         run: exit 1 # fail
+
+  workflow-notification:
+    permissions:
+      contents: read
+      issues: write
+    if: github.event_name == 'schedule' && always()
+    needs:
+      - build
+      - test
+      - integration-test
+      - link-check
+      - markdown-lint-check
+      - misspell-check
+      - shell-script-check
+      - publish-snapshots
+    uses: ./.github/workflows/reusable-workflow-notification.yml
+    with:
+      success: >-
+        ${{
+          needs.build.result == 'success' &&
+          needs.test.result == 'success' &&
+          needs.integration-test.result == 'success' &&
+          needs.link-check.result == 'success' &&
+          needs.markdown-lint-check.result == 'success' &&
+          needs.misspell-check.result == 'success' &&
+          needs.shell-script-check.result == 'success' &&
+          needs.publish-snapshots.result == 'success'
+        }}

.github/workflows/build.yml

@@ -11,8 +11,35 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0 # needed for merge-base below
 
-      - name: Link check
+      - name: Link check - relative links (all files)
+        if: github.event_name == 'pull_request'
+        env:
+          GITHUB_TOKEN: ${{ github.token }}
+        run: ./.github/scripts/link-check.sh --relative-only
+
+      - name: Get modified files
+        if: github.event_name == 'pull_request'
+        id: modified-files
+        run: |
+          merge_base=$(git merge-base origin/${{ github.base_ref }} HEAD)
+          # Using lychee's default extension filter here to match when it runs against all files
+          modified_files=$(git diff --name-only $merge_base...${{ github.event.pull_request.head.sha }} \
+                            | grep -E '\.(md|mkd|mdx|mdown|mdwn|mkdn|mkdown|markdown|html|htm|txt)$' \
+                            | tr '\n' ' ' || true)
+          echo "files=$modified_files" >> $GITHUB_OUTPUT
+          echo "Modified files: $modified_files"
+
+      - name: Link check - all links (modified files only)
+        if: github.event_name == 'pull_request' && steps.modified-files.outputs.files != ''
+        env:
+          GITHUB_TOKEN: ${{ github.token }}
+        run: ./.github/scripts/link-check.sh ${{ steps.modified-files.outputs.files }}
+
+      - name: Link check - all links (all files)
+        if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
         env:
           GITHUB_TOKEN: ${{ github.token }}
         run: ./.github/scripts/link-check.sh