Merge pull request #52 from hubmapconsortium/sunset666/bugfix_excepti…

…on_handling

Bugfix exception catching and tasks

src/ingest_validation_tests/fastq_validator_logic.py

@@ -167,6 +167,12 @@ def validate_fastq_file(self, fastq_file: Path) -> None:
         except IOError:
             self.errors.append(self._format_error(f"Unable to open FASTQ data file {fastq_file}."))
             return
+        except EOFError:
+            self.errors.append(self._format_error(f"EOF in FASTQ data file {fastq_file}."))
+        except Exception as e:
+            self.errors.append(
+                self._format_error(f"Unexpected error: {e} on data file {fastq_file}.")
+            )
         self._file_record_counts[str(fastq_file)] = records_read
 
     def validate_fastq_files_in_path(self, paths: List[Path], threads: int) -> None:
@@ -192,6 +198,9 @@ def validate_fastq_files_in_path(self, paths: List[Path], threads: int) -> None:
                 data_output = pool.imap_unordered(engine, file_list)
             except Exception as e:
                 _log(f"Error {e}")
+                pool.close()
+                pool.join()
+                data_found_one.append(f"Error {e}")
             else:
                 pool.close()
                 pool.join()
@@ -215,7 +224,8 @@ def _find_duplicates(self, dirs_and_files):
                 if len(filepaths) > 1:
                     self.errors.append(
                         _log(
-                            f"{filename} has been found multiple times during this validation. Locations of duplicates: {filepaths}."  # noqa: E501
+                            f"{filename} has been found multiple times during this validation. "
+                            f"Locations of duplicates: {filepaths}."  # noqa: E501
                         )
                     )
 

src/ingest_validation_tests/gz_validator.py

@@ -46,6 +46,9 @@ def collect_errors(self, **kwargs) -> List[str]:
             data_output = pool.imap_unordered(engine, file_list)
         except Exception as e:
             _log(f"Error {e}")
+            pool.close()
+            pool.join()
+            data_output2.extend(f"Error: {e}")
         else:
             pool.close()
             pool.join()