chore(parsing): Allow parsing to continue on unrecognized file

scripts/metric_reporter/main.py

@@ -101,7 +101,9 @@ def main(
     except GCSClientError as error:
         logger.error(f"GCS client error: {error}")
     except ParserError as error:
-        logger.error(f"Parsing error: {error}")
+        # Log as warning since it's not a fatal situation.
+        # This allows the pipeline to continue on parsing other files.
+        logger.warning(f"Parsing error: {error}")
     except ReporterError as error:
         logger.error(f"Test Suite Reporter error: {error}")
     except Exception as error:

scripts/metric_reporter/parser/coverage_json_parser.py

@@ -204,15 +204,22 @@ def parse(self, artifact_file_names: list[str]) -> list[CoverageJsonGroup]:
         coverage_json_groups: list[CoverageJsonGroup] = []
         for artifact_file_name in artifact_file_names:
             self.logger.info(f"Parsing {artifact_file_name}")
-            file: ArtifactFile = self._parse_artifact_file_name(artifact_file_name)
-            group: CoverageJsonGroup = self._get_coverage_json_group(file, coverage_json_groups)
             try:
+                file: ArtifactFile = self._parse_artifact_file_name(artifact_file_name)
+                group: CoverageJsonGroup = self._get_coverage_json_group(
+                    file, coverage_json_groups
+                )
                 content: str = self._gcs_client.get_coverage_artifact_content(
                     file.repository, artifact_file_name
                 )
                 json_data: dict[str, Any] = json.loads(content)
                 coverage_json: CoverageJson = self._parse_json_data(file, json_data)
                 group.coverage_jsons.append(coverage_json)
+            except ParserError as error:
+                # We don't want to completely kill the pipeline if a file is a format that can't
+                # be parsed. So we log a warning and continue with the next file.
+                self.logger.warning("Skipping file %s: %s", artifact_file_name, error)
+                continue
             except (JSONDecodeError, ValidationError) as error:
                 error_mapping: dict[type, str] = {
                     JSONDecodeError: f"Invalid JSON format for file {artifact_file_name}",

scripts/metric_reporter/parser/junit_xml_parser.py

@@ -371,13 +371,20 @@ def parse(self, artifact_file_names: list[str]) -> list[JUnitXmlGroup]:
         junit_xml_groups: list[JUnitXmlGroup] = []
         for artifact_file_name in artifact_file_names:
             self.logger.info(f"Parsing {artifact_file_name}")
-            file: ArtifactFile = self._parse_artifact_file_name(artifact_file_name)
-            junit_xml: JUnitXmlJobTestSuites = self._get_junit_xml(file, junit_xml_groups)
             try:
+                file: ArtifactFile = self._parse_artifact_file_name(artifact_file_name)
+                junit_xml: JUnitXmlJobTestSuites = self._get_junit_xml(file, junit_xml_groups)
                 test_suites: JUnitXmlTestSuites = self._parse_test_suites(
                     file.repository, file.name
                 )
                 junit_xml.test_suites.append(test_suites)
+            except ParserError as error:
+                # We don't want to completely kill the pipeline if a file is a format that can't
+                # be parsed. So we log a warning and continue with the next file.
+                # Validation error is lower priority, but internally raises
+                # a ParserError so it needs to be second.
+                self.logger.warning("Skipping file %s: %s", artifact_file_name, error)
+                continue
             except ValidationError as error:
                 error_msg: str = f"Unexpected value or schema in file {artifact_file_name}"
                 self.logger.error(error_msg, exc_info=error)