def parse(self, filename, provider=None):
try:
item = {}
self.set_item_defaults(item, filename)
with open(filename, 'r', encoding='windows-1252') as f:
# read the whole file into a single string
lines = f.read()
# Construct pattern for the regular expression
pattern = '(.*)\n'
for f in self.field_list:
pattern = pattern + f[0] + '(.*)\n'
m = re.match(pattern, ''.join(lines), re.MULTILINE | re.DOTALL)
if m:
for f in self.field_list:
if f[1] is not None:
item[f[1]] = m.group(f[2])
# fix the formatting
item[self.ITEM_VERSION_CREATED] = self.datetime(
item[self.ITEM_VERSION_CREATED])
item[self.ITEM_BODY_HTML] = '<p>' + html.escape(item[self.ITEM_BODY_HTML].strip()).replace('\n', '</p><p>')\
+ '</p>'
item.setdefault('word_count', get_word_count(item['body_html']))
return item
except Exception as ex:
raise AAPParserError.NewsBitesParserError(exception=ex,
provider=provider)
'qcode': '04000000',
'name': subject_codes['04000000']
}]
item[FORMAT] = FORMATS.HTML
def datetime(self, string):
"""
Convert the date string parsed from the source file to a datetime, assumes that the
time is local to Sydney Australia
:param string:
:return:
"""
# 06 June 2016 14:00:00
try:
local_dt = datetime.datetime.strptime(string, '%d %B %Y %H:%M:%S')
except ValueError:
local_dt = datetime.datetime.strptime(string, '%d %b %Y %H:%M:%S')
local_tz = pytz.timezone('Australia/Sydney')
aus_dt = local_tz.localize(local_dt, is_dst=None)
return aus_dt.astimezone(pytz.utc)
try:
register_feed_parser(NewsBitesFeedParser.NAME, NewsBitesFeedParser())
except AlreadyExistsError:
pass
register_feeding_service_error(
'file',
AAPParserError.NewsBitesParserError().get_error_description())
def set_item_defaults(self, item, filename):
item['guid'] = filename + ':' + str(uuid.uuid4())
item['urgency'] = 5
item['pubstatus'] = 'usable'
item['versioncreated'] = utcnow()
item[ITEM_TYPE] = CONTENT_TYPE.TEXT
item['anpa_category'] = [{'qcode': 'f'}]
item['subject'] = [{'qcode': '04000000', 'name': subject_codes['04000000']}]
item[FORMAT] = FORMATS.HTML
def datetime(self, string):
"""
Convert the date string parsed from the source file to a datetime, assumes that the
time is local to Sydney Australia
:param string:
:return:
"""
# 06 June 2016 14:00:00
local_dt = datetime.datetime.strptime(string, '%d %B %Y %H:%M:%S')
local_tz = pytz.timezone('Australia/Sydney')
aus_dt = local_tz.localize(local_dt, is_dst=None)
return aus_dt.astimezone(pytz.utc)
try:
register_feed_parser(NewsBitesFeedParser.NAME, NewsBitesFeedParser())
except AlreadyExistsError as ex:
pass
register_feeding_service_error('file', AAPParserError.NewsBitesParserError().get_error_description())