From df548fa29ddfa2dfac3e67e475ce57f3707d04df Mon Sep 17 00:00:00 2001 From: 1kamma Date: Wed, 12 Apr 2023 12:34:56 +0300 Subject: [PATCH] update folders --- scrapping.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/scrapping.py b/scrapping.py index 46ec1b8..b5070f3 100644 --- a/scrapping.py +++ b/scrapping.py @@ -34,7 +34,8 @@ def _load_json_from_path(json_path: str) -> Dict: def get_raw_english_texts_of_project(project_dirname: str, oracc_site: str = 'oracc.museum.upenn.edu') -> List[Dict]: raw_jsons = list() - all_paths = glob.glob(f'jsons_unzipped/{project_dirname}/**/corpusjson/*.json', recursive=True) + all_paths = glob.glob(f'jsons_unzipped/{project_dirname}/**/corpusjson/*.json', recursive=True) + glob.glob( + f'jsons_unzipped/{project_dirname}/corpusjson/*.json', recursive=True) # path = Path(os.path.join(JSONS_DIR, project_dirname, 'catalogue.json')) # if not os.path.isfile(path): # return raw_jsons