1 from __future__ import print_function
5 from bs4 import BeautifulSoup
6 from constant import DETAIL_URL, SEARCH_URL
7 from logger import logger
10 def dojinshi_parser(id):
11 if not isinstance(id, (int, )) and (isinstance(id, (str, )) and not id.isdigit()):
12 raise Exception('Dojinshi id(%s) is not valid' % str(id))
14 logger.debug('Fetching dojinshi information of id %d' % id)
17 url = '%s/%d/' % (DETAIL_URL, id)
20 response = requests.get(url).content
21 except Exception as e:
22 logger.critical('%s%s' % tuple(e.message))
25 html = BeautifulSoup(response)
26 dojinshi_info = html.find('div', attrs={'id': 'info'})
28 title = dojinshi_info.find('h1').text
29 subtitle = dojinshi_info.find('h2')
31 dojinshi['name'] = title
32 dojinshi['subtitle'] = subtitle.text if subtitle else ''
34 dojinshi_cover = html.find('div', attrs={'id': 'cover'})
35 img_id = re.search('/galleries/([\d]+)/cover\.(jpg|png)$', dojinshi_cover.a.img['src'])
37 logger.critical('Tried yo get image id failed')
39 dojinshi['img_id'] = img_id.group(1)
40 dojinshi['ext'] = img_id.group(2)
43 for _ in dojinshi_info.find_all('div', class_=''):
44 pages = re.search('([\d]+) pages', _.text)
46 pages = pages.group(1)
48 dojinshi['pages'] = int(pages)
52 def search_parser(keyword, page):
53 logger.debug('Searching dojinshis of keyword %s' % keyword)
55 response = requests.get(SEARCH_URL, params={'q': keyword, 'page': page}).content
56 html = BeautifulSoup(response)
57 dojinshi_search_result = html.find_all('div', attrs={'class': 'preview-container'})
58 for dojinshi in dojinshi_search_result:
59 dojinshi_container = dojinshi.find('div', attrs={'class': 'caption'})
60 title = dojinshi_container.text.strip()
61 id_ = re.search('/g/(\d+)/', dojinshi.a['href']).group(1)
62 result.append({'id': id_, 'title': title})
66 def print_dojinshi(dojinshi_list):
69 logger.log(15, 'Print dojinshi list')
71 for dojinshi in dojinshi_list:
72 print(dojinshi['id'], '-', dojinshi['title'])
76 if __name__ == '__main__':
77 print(dojinshi_parser(32271))