def parse_item(item_soup, stem_mapping):
"""Get information about a single HI episode.
@param item_soup: Soup containing information about a single HI episode.
@type item_soup: bs4.BeautifulSoup
@return: Dictionary describing the episode. Contains keys name (str value),
date (datetime.date), loc (url - str value), duration (seconds - int),
and orig_tags (tags applied to episode - list of str)
@rtype: dict
"""
item_date = common.interpret_2822_date(
item_soup.find('pubdate').contents[0]
)
duration_soup = item_soup.find('itunes:duration')
if duration_soup:
duration = common.interpret_duration(
item_soup.find('itunes:duration').contents[0]
)
else:
duration = 7200
title = item_soup.find('title').contents[0]
return {
'name': title,
'date': item_date,
'loc': '',
'duration': duration,
'orig_tags': get_item_tags(title, item_soup, stem_mapping)
}
def process_item(item_soup):
"""Parse information about a single podcast episode.
@param item_soup: Soup containing information about a single podcast
episode.
@type item_soup: bs4.BeautifulSoup
@return: Dictionary describing the episode. Contains keys name (str value),
date (datetime.date), loc (url - str value), duration (seconds - int),
and orig_tags (tags applied to episode - list of str)
@rtype: dict
"""
title = item_soup.find('title').contents[0].strip()
loc = item_soup.find('guid').contents[0]
pub_date_raw = item_soup.find('pubdate').contents[0]
pub_date = common.interpret_2822_date(pub_date_raw)
tags = map(
lambda x: x.contents[0],
item_soup.findAll('category')
)
duration_soup = item_soup.find('itunes:duration')
if duration_soup == None:
duration = 1800 if 'shorts' in tags else 3600
else:
duration_str = duration_soup.contents[0]
duration = common.interpret_duration(duration_str)
return {
'name': title,
'date': pub_date,
'tags': sorted(set(tags)),
'loc': loc,
'duration': duration
}
