setop/agenda_culturel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

WIP

Browse Source
This commit is contained in:
Jean-Marie Favreau 2024-10-12 21:26:18 +02:00
parent 51d71b748f
commit 10422d0499
1 changed files with 10 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
src/agenda_culturel/import_tasks/custom_extractors/acolab.py
View File

@ -75,6 +75,9 @@ class CExtractor(Extractor):
def find_timeslot(text): def find_timeslot(text):
text = re.sub(' +', ' ', text).split(' ') text = re.sub(' +', ' ', text).split(' ')
if len(text) < 3:
return None
day_name = text[0] day_name = text[0]
day_num = text[1] day_num = text[1]
hours = text[2] hours = text[2]
@ -88,6 +91,7 @@ class CExtractor(Extractor):
day_num = int(day_num[0]) day_num = int(day_num[0])
hours = CExtractor.find_hours(hours) hours = CExtractor.find_hours(hours)
if hours is None: if hours is None:
return None return None
@ -109,6 +113,7 @@ class CExtractor(Extractor):
for e in p.stripped_strings: for e in p.stripped_strings:
day = CExtractor.find_timeslot(e) day = CExtractor.find_timeslot(e)
if not day is None: if not day is None:
print('on a une date', day)
if not date is None and is_open: if not date is None and is_open:
# we reach a new day # we reach a new day
result.append((date, slot)) result.append((date, slot))
@ -124,12 +129,14 @@ class CExtractor(Extractor):
continue continue
if CExtractor.is_nickname(e): if CExtractor.is_nickname(e):
print('on a un nickname', e)
# we found a nickname # we found a nickname
is_open = True is_open = True
continue continue
hours = CExtractor.find_hours(e) hours = CExtractor.find_hours(e)
if not hours is None: if not hours is None:
print('on a une heure', hours)
if slot is None: if slot is None:
slot = hours slot = hours
else: else:
@ -137,6 +144,7 @@ class CExtractor(Extractor):
continue continue
if CExtractor.is_canceled(e): if CExtractor.is_canceled(e):
print('on a un cancel')
is_open = False is_open = False
continue continue
@ -166,10 +174,8 @@ class CExtractor(Extractor):
# annule # annule
# menage # menage
for p in description.select('p'): for p in description.select('p'):
CExtractor.find_timeslots(p) ts = CExtractor.find_timeslots(p)
print(ts, when, title)
if not '@' in p.text:
continue