HTML-XSLT import: parse dates of events

1 files changed, 54 insertions, 0 deletions

diff --git a/chimere/utils.py b/chimere/utils.py
index 8c8624e..1985f89 100644
--- a/chimere/utils.py
+++ b/chimere/utils.py
@@ -845,6 +845,35 @@ RE_CLEANS = ((re.compile('(\n)*|^( )*(\n)*( )*|( )*(\n)*( )*$'), ''),
                         '<a href="%(base_url)s\\1"'),
              )
 
+from calendar import TimeEncoding, month_name
+
+def get_month_name(month_no, locale):
+    with TimeEncoding(locale) as encoding:
+        s = month_name[month_no]
+        if encoding is not None:
+            s = s.decode(encoding)
+        return s
+
+MONTH_NAMES = {locale:[get_month_name(no_month, locale+'.UTF-8')
+                        for no_month in xrange(1, 13)] for locale in ['fr_FR']}
+
+UNI_MONTH_NAMES = {locale:[m.decode('utf-8') for m in MONTH_NAMES[locale]]
+                                                 for locale in MONTH_NAMES}
+
+DATE_PARSINGS = {'fr_FR':[
+               re.compile(r'(?P<day1>\d{1,2}) '\
+                          r'(?P<month1>'+ '|'.join(UNI_MONTH_NAMES['fr_FR']) +') '\
+                          r'(?P<year1>\d{4})?[^\d]*'\
+                          r'(?P<day2>\d{1,2}) '\
+                          r'(?P<month2>'+ '|'.join(UNI_MONTH_NAMES['fr_FR']) +') *'\
+                          r'(?P<year2>\d{4})?.*'),
+               re.compile(r'(?P<day1>\d{1,2}) '\
+                          r'(?P<month1>'+ '|'.join(UNI_MONTH_NAMES['fr_FR']) +') *'\
+                          r'(?P<year1>\d{4})?')
+                         ]
+                }
+
+
 class HtmlXsltManager(ImportManager):
     def get(self):
         u"""
@@ -942,7 +971,32 @@ class HtmlXsltManager(ImportManager):
             else:
                 dct['point'] = self.importer_instance.default_localisation
             dct['description'] = item['description']
+            if 'date' in item:
+                has_dates = False
+                for locale in DATE_PARSINGS:
+                    if has_dates:
+                        break
+                    for r in DATE_PARSINGS[locale]:
+                        m = r.search(item['date'])
+                        if not m:
+                            continue
+                        has_dates = True
+                        values = m.groupdict()
+                        year1 = datetime.date.today().year if 'year1' not in values \
+                                else int(values['year1'])
+                        dct['start_date'] = datetime.date(year1,
+                              MONTH_NAMES[locale].index(values['month1'].encode('utf-8')) + 1,
+                              int(values['day1']))
+                        if 'day2' not in values:
+                            break
+                        year2 = datetime.date.today().year if 'year2' not in values \
+                                else int(values['year2'])
+                        dct['end_date'] = datetime.date(year2,
+                              MONTH_NAMES[locale].index(values['month2'].encode('utf-8')) + 1,
+                              int(values['day2']))
+                        break
             key = item['key']
+            print dct
             it, updated, created = self.create_or_update_item(cls, dct, key)
             if updated:
                 updated_item += 1