def __init__(self,
date=None,
fuzzy_year_padding=1,
fuzzy_month_padding=1,
fuzzy_day_padding=1,
fuzzy_season_padding=12,
multiplier_if_uncertain=1,
multiplier_if_approximate=1,
multiplier_if_both=1):
super(ExtendedDateFormat, self).__init__()
self.orig_date = None
self.edtf = None
self.result_set = None
self.sortable_date = None
self.error = None
self.fuzzy_year_padding = int(fuzzy_year_padding)
self.fuzzy_month_padding = int(fuzzy_month_padding)
self.fuzzy_day_padding = int(fuzzy_day_padding)
self.fuzzy_season_padding = int(fuzzy_season_padding)
self.multiplier_if_uncertain = int(multiplier_if_uncertain)
self.multiplier_if_approximate = int(multiplier_if_approximate)
self.multiplier_if_both = int(multiplier_if_both)
try:
self.parse(date)
except Exception as e:
try:
self.parse(text_to_edtf(self.orig_date))
except Exception as err:
self.error = err
raise err
def __init__(self, date=None, fuzzy_year_padding=1, fuzzy_month_padding=1, fuzzy_day_padding=1,
fuzzy_season_padding=12, multiplier_if_uncertain=1, multiplier_if_approximate=1, multiplier_if_both=1):
super(ExtendedDateFormat, self).__init__()
self.orig_date = None
self.edtf = None
self.result_set = None
self.sortable_date = None
self.error = None
self.fuzzy_year_padding = int(fuzzy_year_padding)
self.fuzzy_month_padding = int(fuzzy_month_padding)
self.fuzzy_day_padding = int(fuzzy_day_padding)
self.fuzzy_season_padding = int(fuzzy_season_padding)
self.multiplier_if_uncertain = int(multiplier_if_uncertain)
self.multiplier_if_approximate = int(multiplier_if_approximate)
self.multiplier_if_both = int(multiplier_if_both)
try:
self.parse(date)
except:
try:
self.parse(text_to_edtf(self.orig_date))
except Exception as err:
self.error = err
raise err
def post_shows(api, scroll):
shows = []
url = 'https://studsterkel.wfmt.com/explore#t=date'
soup = get_url_as_soup(url)
ps = soup.find_all('p')
for p in ps:
show = {}
a = p.find('a')
if a is not None:
date = a.find('span')
if date is not None:
# Evil python mutates `a` object
[s.extract() for s in a('span')]
_edtf = parse_edtf(text_to_edtf(date.text))
title = a.text.strip()
person = get_person(title)
thumb = None
show = {
'when_happened':
struct_time_to_datetime(_edtf.upper_strict()),
'resolution':
len(str(_edtf)),
'when_original':
date.text,
'content_url':
'https://studsterkel.wfmt.com{}'.format(a.get('href')),
'title':
a.text.strip(),
'text':
'',
'with_thumbnail':
thumb,
'media_type':
'audio/mpeg',
'content_type':
'Oral histories',
'source_url':
'https://studsterkel.wfmt.com/',
'with_thumbnail':
api.cache_wiki_thumbnail(person)
}
resp = api.create_event(show, scroll)
pprint(resp.json())
def add_first_link(w):
if w is not None and 'event' in w:
e = w['event']
for tag in e(['sup', 'span']):
tag.decompose()
text = e.text.rstrip()
text = re.sub('^\s*\d+:\s*', '', text)
text = re.sub('^:\s*', '', text)
if w['context'] is not None and w['context'] != '':
text = '{}: {}'.format(w['context'], e.text.rstrip())
m = re.match('^(.+[a-z]{2,}\.\s+)(.*)', text)
if m is not None:
title = m.group(1)
text = m.group(2)
else:
title = text
text = ''
w['title'] = title
w['text'] = text
links = e.select('a')
if len(links) > 0 and links[0] is not None:
href = links[0].get('href')
w['content_url'] = 'https://en.wikipedia.org{}'.format(href, )
w['item'] = re.sub(r'/wiki/|/w/index.php\?title\=', '', href)
date_text = '{} {}'.format(w['date'], w['year'])
date_text = re.sub('–', '-', date_text)
try:
edtf_date_txt = text_to_edtf(date_text)
edtf_date = parse_edtf(edtf_date_txt)
iso_date = time.strftime('%Y-%m-%dT%H:%M:%SZ',
edtf_date.upper_fuzzy())
w['when_happened'] = iso_date
w['when_original'] = date_text
w['resolution'] = 10
del w['event']
del w['date']
del w['context']
if 'header' in w:
del w['header']
return w
except Exception:
pass
def getEDTF(datetext, letter_key=None):
''' datetext is a German. It needs be translated to English to be parsable
by the edtf library afterwards.
'''
try:
translated = simpleTranslate(prepareDate(datetext))
prepared_translated = prepareDateSets(translated)
if not isinstance(prepared_translated, str):
try:
if ' between ' in translated or (
'-' in translated or ' to ' in translated
) and not re.search(r'\d{4}\-\d{4}', translated) or (
'after ' in translated and 'before ' in translated
and '/' in translated):
edtf = '/'.join(map(text_to_edtf, prepared_translated))
elif ' and ' in translated:
edtf = '{' + ','.join(
map(text_to_edtf, prepared_translated)) + '}'
else:
edtf = '[' + ','.join(
map(text_to_edtf, prepared_translated)) + ']'
except Exception as e:
raise e
edtf = None
else:
try:
edtf = text_to_edtf(prepared_translated)
except Exception as e:
raise e
edtf = None
except Exception as e:
raise e
# run postprocessing of date
edtf = postprocessDate(translated, edtf)
log.debug(datetext)
log.debug((letter_key, edtf))
return edtf
def create_from_string(self, date_str):
edtf = str(text_to_edtf(date_str))
return ExtendedDate.objects.create(edtf_format=edtf)
def match_string(self, date_str):
return self.edtf_format == str(text_to_edtf(date_str))
