get the pipeline fixed

Author: HarshCasper

.github/workflows/update-azure-coverage.yml

@@ -9,8 +9,6 @@ on:
         required: true
         type: string
         description: "Branch to checkout and compare against (e.g. harshmishra/doc-91)"
-  pull_request:
-    types: [opened, synchronize]
 
 jobs:
   update-azure-coverage:
@@ -25,7 +23,7 @@ jobs:
         with:
           fetch-depth: 0
           path: docs
-          ref: ${{ github.event.inputs.targetBranch }}
+          ref: ${{ github.event_name == 'workflow_dispatch' && github.event.inputs.targetBranch || 'main' }}
 
       - name: Set up system wide dependencies
         run: |
@@ -42,20 +40,25 @@ jobs:
         env:
           GITHUB_TOKEN: ${{ secrets.PRO_ACCESS_TOKEN }}
           REPOSITORY_NAME: localstack-pro
-          ARTIFACT_ID: implemented_features_python-amd64.csv
+          ARTIFACT_ID: implemented_features_python-amd64
           WORKFLOW: "Az / Build, Test, Push"
 
       - name: Generate Azure coverage JSON data
         working-directory: docs
         run: |
-          python3 scripts/create_azure_coverage.py -i target/implemented_features_python-amd64.csv/implemented_features.csv -o target/updated_azure_coverage
-          mv -f target/updated_azure_coverage/*.json src/data/azure-coverage/
+          python3 scripts/create_azure_coverage.py -i target/implemented_features.csv -o target/updated_azure_coverage
+          if ls target/updated_azure_coverage/*.json > /dev/null 2>&1; then
+            mv -f target/updated_azure_coverage/*.json src/data/azure-coverage/
+          else
+            echo "No JSON files generated in target/updated_azure_coverage."
+            exit 1
+          fi
 
       - name: Check for changes
         id: check-for-changes
         working-directory: docs
         env:
-          TARGET_BRANCH: ${{ github.event.inputs.targetBranch }}
+          TARGET_BRANCH: ${{ github.event_name == 'workflow_dispatch' && github.event.inputs.targetBranch || 'main' }}
         run: |
           mkdir -p resources
           (git diff --name-only origin/automated-azure-coverage-updates src/data/azure-coverage/ 2>/dev/null || git diff --name-only "origin/$TARGET_BRANCH" src/data/azure-coverage/ 2>/dev/null) | tee -a resources/diff-check.log

scripts/create_azure_coverage.py

@@ -63,6 +63,13 @@ def _load_csv(path: Path) -> dict[str, dict[str, dict[str, dict[str, Any]]]]:
         reader = csv.DictReader(file)
         if not reader.fieldnames:
             raise ValueError("Input CSV has no headers.")
+        required_headers = {"resource_provider", "service", "feature"}
+        if not required_headers.issubset(set(reader.fieldnames)):
+            raise ValueError(
+                "Unexpected CSV schema. Expected headers including "
+                f"{sorted(required_headers)}, got {reader.fieldnames}. "
+                "The downloaded artifact may contain an error payload instead of CSV data."
+            )
 
         for row in reader:
             provider = _normalize_provider(row.get("resource_provider", ""))
@@ -90,6 +97,12 @@ def _load_csv(path: Path) -> dict[str, dict[str, dict[str, dict[str, Any]]]]:
                 "pro": pro_only,
             }
 
+    if not coverage:
+        raise ValueError(
+            "No Azure coverage records were parsed from the input CSV. "
+            "Please verify the artifact content is valid and non-empty."
+        )
+
     return coverage
 
 

scripts/get_latest_github_metrics.sh

@@ -7,7 +7,7 @@ METRICS_ARTIFACTS_BRANCH=${2:-main}
 
 # env vars
 REPOSITORY_NAME=${REPOSITORY_NAME:-localstack-pro}
-ARTIFACT_ID=${ARTIFACT_ID:-implemented_features_python-amd64.csv}
+ARTIFACT_ID=${ARTIFACT_ID:-implemented_features_python-amd64}
 WORKFLOW=${WORKFLOW:-"Az / Build, Test, Push"}
 PREFIX_ARTIFACT=${PREFIX_ARTIFACT:-}
 FILTER_SUCCESS=${FILTER_SUCCESS:-1}
@@ -30,15 +30,28 @@ else
   SELECTOR='.[] | select(.status=="completed" and (.conclusion=="failure" or .conclusion=="success"))'
 fi
 
-RUN_IDS=$(gh run list --limit "$LIMIT" --branch "$METRICS_ARTIFACTS_BRANCH" --repo "$REPOSITORY_OWNER/$REPOSITORY_NAME" --workflow "$WORKFLOW" --json databaseId,conclusion,status --jq "$SELECTOR")
+RUN_IDS=()
+while IFS= read -r run_id; do
+  RUN_IDS+=("$run_id")
+done < <(
+  gh run list \
+    --limit "$LIMIT" \
+    --branch "$METRICS_ARTIFACTS_BRANCH" \
+    --repo "$REPOSITORY_OWNER/$REPOSITORY_NAME" \
+    --workflow "$WORKFLOW" \
+    --json databaseId,conclusion,status \
+    --jq "$SELECTOR | .databaseId"
+)
 
-if [ "$(echo "$RUN_IDS" | jq -rs '.[0].databaseId')" = "null" ]; then
-  echo "No matching workflow run found."
+if [ "${#RUN_IDS[@]}" -eq 0 ]; then
+  echo "No matching workflow runs found."
   exit 1
 fi
 
-for ((i=0; i<LIMIT; i++)); do
-  RUN_ID=$(echo "$RUN_IDS" | jq -rs ".[$i].databaseId")
+for RUN_ID in "${RUN_IDS[@]}"; do
+  if [ -z "$RUN_ID" ] || [ "$RUN_ID" = "null" ]; then
+    continue
+  fi
   echo "Trying run id: $RUN_ID"
 
   gh run download "$RUN_ID" --repo "$REPOSITORY_OWNER/$REPOSITORY_NAME" -p "$ARTIFACT_ID" -D "$TMP_FOLDER" || true
@@ -49,6 +62,11 @@ for ((i=0; i<LIMIT; i++)); do
   fi
 done
 
+if [ "$(ls -1 "$TMP_FOLDER" 2>/dev/null | wc -l)" -eq 0 ]; then
+  echo "Failed to download artifact '$ARTIFACT_ID' from the checked workflow runs."
+  exit 1
+fi
+
 echo "Moving artifact to $TARGET_FOLDER"
 mkdir -p "$TARGET_FOLDER"
 if [[ -z "${PREFIX_ARTIFACT}" ]]; then