A homegrown check-files (#54875)

Author: hectorsector

.github/actions/get-changed-files/action.yml

@@ -0,0 +1,31 @@
+name: Get changed files
+description: Get a list of changed files
+
+inputs:
+  files:
+    description: 'Files or directories to check for changes'
+    required: false
+    default: '.'
+  head:
+    description: 'Head ref to check for changes against'
+    required: false
+
+outputs:
+  all_changed_files:
+    description: 'List of all changed files (unfiltered)'
+    value: ${{ steps.get_changes.outputs.all_changed_files }}
+  filtered_changed_files:
+    description: 'List of changed files matching the filter'
+    value: ${{ steps.get_changes.outputs.filtered_changed_files }}
+
+runs:
+  using: 'composite'
+  steps:
+    - name: Gather changed files
+      id: get_changes
+      env:
+        INPUT_FILES: ${{ inputs.files }}
+        PR: ${{ github.event.pull_request.number }}
+        HEAD: ${{ github.event.pull_request.head.ref || github.event.merge_group.head_ref || inputs.head || github.ref_name }}
+      shell: bash
+      run: ${{ github.action_path }}/get-changed-files.sh

.github/actions/get-changed-files/get-changed-files.sh

@@ -0,0 +1,72 @@
+#!/bin/bash
+
+# Required environment variables:
+# $INPUT_FILES: Pattern(s) to filter files by (e.g., "content/** data/**")
+# $FILTER: Derived from INPUT_FILES, defaults to "." if not provided
+# $PR: Pull request number (if running in PR context)
+# $HEAD: Current branch or SHA for git diff
+
+# Default value for files parameter if not provided
+FILTER=${INPUT_FILES:-.}
+
+# Print the filter
+echo "__ using filter: __"
+echo "$FILTER"
+
+# Find the file diff in the pull request or merge group
+# If its a pull request, use the faster call to the GitHub API
+# For push, workflow_dispatch, and merge_group, use git diff
+if [ -n "$PR" ]
+then
+  echo "__ running gh pr diff __"
+  DIFF=`gh pr diff $PR --name-only`
+  if [ -z "$DIFF" ]; then
+    echo "__ gh pr diff failed, falling back to git diff __"
+    HEAD=$(gh pr view $PR --json headRefName --jq .headRefName)
+  fi
+fi
+
+if [ -z "$DIFF" ]; then
+  echo "__ using branch name $HEAD __"
+  git fetch origin main --depth 1
+  echo "__ running git diff __"
+  DIFF=`git diff --name-only origin/main $HEAD`
+fi
+
+# So we can inspect the output
+echo "__ DIFF found __"
+echo "$DIFF"
+
+# Filter the DIFF to just the directories specified in the input files
+if [ "$FILTER" != "." ]; then
+  echo "__ filtering DIFF to only include $FILTER __"
+  FILTERED_DIFF=""
+  IFS=$'\n'
+  for file in $DIFF; do
+    while IFS= read -r pattern || [ -n "$pattern" ]; do
+      clean_pattern=${pattern%/}
+      if [[ $file == $clean_pattern || $file == $clean_pattern/* ]]; then
+        FILTERED_DIFF="$FILTERED_DIFF $file"
+        break
+      fi
+    done <<< "$FILTER"
+  done
+  unset IFS
+  DIFF=$FILTERED_DIFF
+  echo "__ filtered DIFF __"
+  echo "$DIFF"
+fi
+
+# Format the output
+echo "__ formatting output __"
+FORMATTED_DIFF=$(echo $DIFF | tr '\n' ' ' | tr -s ' ')
+echo "$FORMATTED_DIFF"
+
+# Set the output for GitHub Actions
+if [[ -n "$GITHUB_OUTPUT" ]]; then
+  echo "all_changed_files=$DIFF" >> "$GITHUB_OUTPUT"
+  echo "filtered_changed_files=$FORMATTED_DIFF" >> "$GITHUB_OUTPUT"
+else
+  echo "all_changed_files=$DIFF"
+  echo "filtered_changed_files=$FORMATTED_DIFF"
+fi

.github/workflows/content-lint-markdown.yml

@@ -7,7 +7,13 @@ name: 'Content Lint Markdown'
 on:
   pull_request:
   merge_group:
-
+  workflow_dispatch:
+    inputs:
+      branch:
+        description: The branch containing the changes we want to lint.
+        required: true
+        type: string
+        default: main
 permissions:
   contents: read
 
@@ -30,23 +36,19 @@ jobs:
 
       - name: Get changed content/data files
         id: changed-files
-        uses: tj-actions/changed-files@40853de9f8ce2d6cfdc73c1b96f14e22ba44aec4 # v45.0.0
+        uses: ./.github/actions/get-changed-files
         with:
-          # No need to escape the file names because we make the output of
-          # tj-actions/changed-files be set as an environment variable. Not
-          # as a direct input to the line of bash that uses it.
-          safe_output: false
           files: |
             content/**
             data/**
 
       - name: Print content linter annotations if changed content/data files
-        if: steps.changed-files.outputs.any_changed == 'true'
+        if: steps.changed-files.outputs.filtered_changed_files
         env:
           # Make it an environment variable so that its value doesn't need to be escaped.
           # See https://docs.github.com/en/actions/security-guides/security-hardening-for-github-actions#using-an-intermediate-environment-variable
           CHANGED_FILES: |-
-            ${{ steps.changed-files.outputs.all_changed_files }}
+            ${{ steps.changed-files.outputs.filtered_changed_files }}
         # If there are errors, using `--print-annotations` will make it
         # so it does *not* exit non-zero.
         # This is so that all warnings and errors are printed.
@@ -58,5 +60,5 @@ jobs:
           # Make it an environment variable so that its value doesn't need to be escaped.
           # See https://docs.github.com/en/actions/security-guides/security-hardening-for-github-actions#using-an-intermediate-environment-variable
           CHANGED_FILES: |-
-            ${{ steps.changed-files.outputs.all_changed_files }}
+            ${{ steps.changed-files.outputs.filtered_changed_files }}
         run: npm run lint-content -- --errors-only --paths $CHANGED_FILES