Feed date scraper for Gemini (protocol)
parsing: help out by adding space before colon
| -rwxr-xr-x | test_zachwalk.py | 1 | ||||
| -rwxr-xr-x | zachwalk.py | 4 |
2 files changed, 4 insertions, 1 deletions
diff --git a/test_zachwalk.py b/test_zachwalk.py index 1a551db..3cbf5b2 100755 --- a/test_zachwalk.py +++ b/test_zachwalk.py @@ -7,6 +7,7 @@ def main(): assert zachwalk.gnd([b'2021-01-31']) == zachwalk.DEFAULT 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() if __name__ == '__main__': main() diff --git a/zachwalk.py b/zachwalk.py index 7c8680e..2c2c3ea 100755 --- a/zachwalk.py +++ b/zachwalk.py @@ -29,7 +29,9 @@ def gnd(fp): line=line.decode('UTF-8') if line.strip()[0:2] == '=>': try: - date = parse(getdesc(line),fuzzy=True).date() + desc =getdesc(line) + desc=desc.replace(':',' :') #quirk for parse on drew's capsule + date = parse(desc,fuzzy=True).date() # todo: read lots of these and compare them return date except: |