better?

Author: bmastbergen

.github/workflows/upstream-fix-check.yml

@@ -77,9 +77,6 @@ jobs:
 
             PR_COMMIT_MESSAGE=$(echo "$PR_COMMIT_DETAILS_JSON" | jq -r '.commit.message')
 
-            PR_COMMENT_BODY_ACCUMULATOR+="PR Commit Message:\n"
-            PR_COMMENT_BODY_ACCUMULATOR+="$(echo "$PR_COMMIT_MESSAGE" | sed 's/^/    /')\n"
-
             # Extract the upstream Linux commit hash from the PR commit message.
             UPSTREAM_LINUX_HASH=$(echo "$PR_COMMIT_MESSAGE" | grep -Eo "^commit [0-9a-f]{40}$" | awk '{print $2}')
 
@@ -99,7 +96,8 @@ jobs:
 
             FOUND_FIXING_COMMITS=()
             SEARCH_PAGE=1
-            MAX_SEARCH_PAGES=3 # Limit the search to prevent excessive API calls, e.g., up to 300 commits
+            MAX_SEARCH_PAGES=3 # Limit the search to prevent excessive API calls (up to 300 commits)
+            ALL_SEARCH_RESULTS_JSON="[]" # Initialize as an empty JSON array
 
             while [ "$SEARCH_PAGE" -le "$MAX_SEARCH_PAGES" ]; do
                 PR_COMMENT_BODY_ACCUMULATOR+="  Fetching search results (page $SEARCH_PAGE)...\n"
@@ -108,47 +106,52 @@ jobs:
                 # URL-encode the query string for safe API calls
                 ENCODED_SEARCH_QUERY=$(echo -n "$SEARCH_QUERY" | jq -sRr @uri)
 
-                SEARCH_RESULTS_JSON=$(curl -sS -H "Accept: application/vnd.github.v3+json" \
+                CURRENT_PAGE_RESULTS_JSON=$(curl -sS -H "Accept: application/vnd.github.v3+json" \
                                           -H "Authorization: token $GITHUB_TOKEN" \
                                           "https://api.github.com/search/commits?q=${ENCODED_SEARCH_QUERY}&per_page=100&page=$SEARCH_PAGE")
 
                 # Robustly check for an API error message from the search endpoint.
-                API_ERROR=$(echo "$SEARCH_RESULTS_JSON" | jq -r 'if type == "object" and .message then .message else null end')
+                API_ERROR=$(echo "$CURRENT_PAGE_RESULTS_JSON" | jq -r 'if type == "object" and .message then .message else null end')
                 if [ "$API_ERROR" != "null" ] && [ -n "$API_ERROR" ]; then
                     PR_COMMENT_BODY_ACCUMULATOR+="  ERROR: Failed to retrieve search results (page $SEARCH_PAGE) from GitHub Search API: $API_ERROR\n"
-                    PR_COMMENT_BODY_ACCUMULATOR+="  API Response snippet: $(echo "$SEARCH_RESULTS_JSON" | head -n 5)\n"
+                    PR_COMMENT_BODY_ACCUMULATOR+="  API Response snippet: $(echo "$CURRENT_PAGE_RESULTS_JSON" | head -n 5)\n"
                     exit 1
                 fi
 
                 # Check if there are any items on this page (search results are in the 'items' array)
-                CURRENT_PAGE_ITEMS=$(echo "$SEARCH_RESULTS_JSON" | jq '.items | length')
+                CURRENT_PAGE_ITEMS=$(echo "$CURRENT_PAGE_RESULTS_JSON" | jq '.items | length')
                 if [ "$CURRENT_PAGE_ITEMS" -eq 0 ]; then
                     PR_COMMENT_BODY_ACCUMULATOR+="  No more search results found on this page. Stopping search.\n"
                     break # No more commits to process
                 fi
 
-                # Iterate through each item (commit object) in the search results
-                echo "$SEARCH_RESULTS_JSON" | jq -c '.items[]' | while read -r commit_json_line; do
-                  COMMIT_SHA=$(echo "$commit_json_line" | jq -r '.sha')
-                  COMMIT_MESSAGE=$(echo "$commit_json_line" | jq -r '.commit.message') # Get the full message for reporting
-
-                  # Double-check the grep match for robustness to ensure it's a legitimate "Fixes:" tag
-                  # and matches the FULL upstream hash.
-                  if echo "$COMMIT_MESSAGE" | grep -Eq "^Fixes: [0-9a-fA-F]*${UPSTREAM_LINUX_HASH}.*"; then
-                    PR_COMMENT_BODY_ACCUMULATOR+="  Found upstream commit \`$COMMIT_SHA\` that fixes \`$UPSTREAM_LINUX_HASH\`.\n"
-                    PR_COMMENT_BODY_ACCUMULATOR+="  Fixing commit message snippet:\n"
-                    PR_COMMENT_BODY_ACCUMULATOR+="$(echo "$COMMIT_MESSAGE" | grep -Ei "^Fixes:.*" | sed 's/^/      /')\n"
-                    FOUND_FIXING_COMMITS+=("$COMMIT_SHA")
-                  fi
-                done # End of while read loop for search results
+                # Append current page's items to the accumulated results
+                ALL_SEARCH_RESULTS_JSON=$(echo "$ALL_SEARCH_RESULTS_JSON" "$CURRENT_PAGE_RESULTS_JSON" | jq -s '.[0] + .[1].items')
 
                 SEARCH_PAGE=$((SEARCH_PAGE + 1))
             done # End of while loop for search pages
 
+            # Process all accumulated search results with a single jq command for filtering
+            # This filters for the exact "Fixes: <FULL_SHA>" and extracts relevant info.
+            FILTERED_COMMITS=$(echo "$ALL_SEARCH_RESULTS_JSON" | jq -r --arg full_hash "$UPSTREAM_LINUX_HASH" '
+              .[] |
+              select(.commit.message | test("Fixes: [0-9a-fA-F]*\\Q" + $full_hash + "\\E.*"; "i")) | # "i" for case-insensitive
+              "\(.sha):\(.commit.message | split("\n")[] | select(test("Fixes: [0-9a-fA-F]*\\Q" + $full_hash + "\\E.*"; "i")))"
+            ')
+
+            if [ -n "$FILTERED_COMMITS" ]; then
+              IFS=$'\n' read -r -d '' -a FOUND_FIXING_COMMITS <<< "$FILTERED_COMMITS"
+            else
+              FOUND_FIXING_COMMITS=() # Ensure it's an empty array if no matches
+            fi
+
             if [ ${#FOUND_FIXING_COMMITS[@]} -gt 0 ]; then
               PR_COMMENT_BODY_ACCUMULATOR+="  SUCCESS: Found ${#FOUND_FIXING_COMMITS[@]} upstream Linux commit(s) that fix \`$UPSTREAM_LINUX_HASH\`:\n"
-              for FIX_COMMIT_SHA in "${FOUND_FIXING_COMMITS[@]}"; do
-                PR_COMMENT_BODY_ACCUMULATOR+="    - \`$FIX_COMMIT_SHA\`\n"
+              for FIX_COMMIT_ENTRY in "${FOUND_FIXING_COMMITS[@]}"; do
+                # Split SHA and message part for display
+                FIX_COMMIT_SHA=$(echo "$FIX_COMMIT_ENTRY" | cut -d ':' -f 1)
+                FIX_MESSAGE_SNIPPET=$(echo "$FIX_COMMIT_ENTRY" | cut -d ':' -f 2-)
+                PR_COMMENT_BODY_ACCUMULATOR+="    - \`$FIX_COMMIT_SHA\` (Fixes tag: \`$FIX_MESSAGE_SNIPPET\`)\n"
               done
             else
               PR_COMMENT_BODY_ACCUMULATOR+="  No upstream fixes found for \`$UPSTREAM_LINUX_HASH\`.\n"