diff --git a/modelforge/curation/tmqm_curation.py b/modelforge/curation/tmqm_curation.py index fdfd75b6..956e79f5 100644 --- a/modelforge/curation/tmqm_curation.py +++ b/modelforge/curation/tmqm_curation.py @@ -441,7 +441,6 @@ def _process_downloaded( for name in snapshots_temp_dict.keys(): data_temp.append(snapshots_temp_dict[name]) - print(max_records, total_conformers) n_max = len(data_temp) if max_records is not None: n_max = max_records @@ -528,18 +527,20 @@ def process( ) from modelforge.utils.misc import list_files, ungzip_file - # list the files in the directory to examine - files = list_files( + # list the files in the directory that are gzipped + gzip_files = list_files( directory=f"{self.local_cache_dir}/tmqm_files/{self.extracted_filepath}", extension=".gz", ) - for file in files: + # ungzip the files + for file in gzip_files: ungzip_file( input_path_dir=f"{self.local_cache_dir}/tmqm_files/{self.extracted_filepath}", file_name=file, output_path_dir=f"{self.local_cache_dir}/tmqm_files/{self.extracted_filepath}", ) + # list the files in the directory by type xyz_files = list_files( directory=f"{self.local_cache_dir}/tmqm_files/{self.extracted_filepath}", extension=".xyz", diff --git a/modelforge/tests/data/test_file.txt.gz b/modelforge/tests/data/test_file.txt.gz new file mode 100644 index 00000000..3479ede5 Binary files /dev/null and b/modelforge/tests/data/test_file.txt.gz differ