4 from bs4 import BeautifulSoup
5 from constant import DETAIL_URL, SEARCH_URL
6 from hentai.logger import logger
9 def dojinshi_parser(id):
10 if not isinstance(id, (int, )) and (isinstance(id, (str, )) and not id.isdigit()):
11 raise Exception('Dojinshi id(%s) is not valid' % str(id))
13 logger.debug('Fetching dojinshi information of id %d' % id)
16 url = '%s/%d/' % (DETAIL_URL, id)
19 response = requests.get(url).content
21 logger.critical('%s%s' % tuple(e.message))
24 html = BeautifulSoup(response)
25 dojinshi_info = html.find('div', attrs={'id': 'info'})
27 title = dojinshi_info.find('h1').text
28 subtitle = dojinshi_info.find('h2')
30 dojinshi['name'] = title
31 dojinshi['subtitle'] = subtitle.text if subtitle else ''
33 dojinshi_cover = html.find('div', attrs={'id': 'cover'})
34 img_id = re.search('/galleries/([\d]+)/cover\.(jpg|png)$', dojinshi_cover.a.img['src'])
36 logger.critical('Tried yo get image id failed')
38 dojinshi['img_id'] = img_id.group(1)
39 dojinshi['ext'] = img_id.group(2)
42 for _ in dojinshi_info.find_all('div', class_=''):
43 pages = re.search('([\d]+) pages', _.text)
45 pages = pages.group(1)
47 dojinshi['pages'] = int(pages)
51 def search_parser(keyword, page):
52 logger.debug('Searching dojinshis of keyword %s' % keyword)
54 response = requests.get(SEARCH_URL, params={'q': keyword, 'page': page}).content
55 html = BeautifulSoup(response)
56 dojinshi_search_result = html.find_all('div', attrs={'class': 'preview-container'})
57 for dojinshi in dojinshi_search_result:
58 dojinshi_container = dojinshi.find('div', attrs={'class': 'caption'})
59 title = dojinshi_container.text.strip()
60 id_ = re.search('/g/(\d+)/', dojinshi.a['href']).group(1)
61 result.append({'id': id_, 'title': title})
65 def print_dojinshi(dojinshi_list):
68 logger.log(15, 'Print dojinshi list')
70 for dojinshi in dojinshi_list:
71 print dojinshi['id'], '-', dojinshi['title']
75 if __name__ == '__main__':
76 print dojinshi_parser(32271)