From 5f955f2d3fdd5698c969d934756598f97c2e2def Mon Sep 17 00:00:00 2001 From: Arkadiy Shapkin Date: Wed, 19 Feb 2020 20:14:35 +0300 Subject: [PATCH] [whatsapp] Support loading txt files from subdirectories --- parsers/whatsapp.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/parsers/whatsapp.py b/parsers/whatsapp.py index 31e51b7..b48bab3 100644 --- a/parsers/whatsapp.py +++ b/parsers/whatsapp.py @@ -25,7 +25,7 @@ def infer_datetime_regex(f_path, max_messages=100): with open(f_path, 'r', encoding="utf8") as f: for c, line in enumerate(f): if c == max_messages: - break; + break matches = regex_message.search(line.upper()) if matches: pattern = "" @@ -67,7 +67,7 @@ def main(own_name, file_path, max_exported_messages, infer_datetime): global MAX_EXPORTED_MESSAGES MAX_EXPORTED_MESSAGES = max_exported_messages log.info('Parsing Whatsapp data...') - files = glob.glob(os.path.join(file_path, '*.txt')) + files = glob.glob(os.path.join(file_path, '*', '*.txt')) if len(files) == 0: log.error(f'No input files found under {file_path}') exit(0) @@ -91,7 +91,6 @@ def parse_messages(files, own_name, infer_datetime): data = [] for f_path in files: log.info(f'Reading {f_path}') - f_name = os.path.basename(f_path) conversation_id = uuid.uuid4().hex participants = set() conversation_data = []