Feed date scraper for Gemini (protocol)
description parsing: split at colon
| -rwxr-xr-x | test_zachwalk.py | 1 | ||||
| -rwxr-xr-x | zachwalk.py | 2 |
2 files changed, 2 insertions, 1 deletions
diff --git a/test_zachwalk.py b/test_zachwalk.py index 3cbf5b2..42efc43 100755 --- a/test_zachwalk.py +++ b/test_zachwalk.py @@ -8,6 +8,7 @@ def main(): assert zachwalk.gnd([b'=> path.gmi 2021-01-31 - my post']) == parse('2021-01-31').date() assert zachwalk.gnd([b'=> 2020/11/25/hello-gemini.gmi 2020-11-25 - Hello, Gemini!']) == parse('2020-11-25').date() assert zachwalk.gnd([b'=> gemini://drewdevault.com/2020/09/21/Gemini-TOFU.gmi September 21, 2020: TOFU recommendations for Gemini']) == parse('2020-09-21').date() + assert zachwalk.gnd(['=> gemini://drewdevault.com/2021/02/15/Status-update-February-2021.gmi February 15, 2021: Status update, February 2021']) == parse('2021-02-15').date() if __name__ == '__main__': main() diff --git a/zachwalk.py b/zachwalk.py index a270850..1a15ddc 100755 --- a/zachwalk.py +++ b/zachwalk.py @@ -34,7 +34,7 @@ def gnd(fp): if line.strip()[0:2] == '=>': try: desc =getdesc(line) - desc=desc.replace(':',' :') #quirk for parse on drew's capsule + desc=desc.split(':')[0] #this should only have 1 'datey' part date = parse(desc,fuzzy=True).date() # todo: read lots of these and compare them return date |