2 from __future__ import print_function
6 from bs4 import BeautifulSoup
8 from logger import logger
9 from tabulate import tabulate
12 def request(method, url, **kwargs):
13 if not hasattr(requests, method):
14 raise AttributeError('\'requests\' object has no attribute \'{}\''.format(method))
16 return requests.__dict__[method](url, proxies=constant.PROXY, **kwargs)
19 def doujinshi_parser(id_):
20 if not isinstance(id_, (int,)) and (isinstance(id_, (str,)) and not id_.isdigit()):
21 raise Exception('Doujinshi id({}) is not valid'.format(id_))
24 logger.log(15, 'Fetching doujinshi information of id {}'.format(id_))
27 url = '{}/{}/'.format(constant.DETAIL_URL, id_)
30 response = request('get', url).content
31 except Exception as e:
32 logger.critical(str(e))
35 html = BeautifulSoup(response)
36 doujinshi_info = html.find('div', attrs={'id': 'info'})
38 title = doujinshi_info.find('h1').text
39 subtitle = doujinshi_info.find('h2')
40 doujinshi['name'] = title
41 doujinshi['subtitle'] = subtitle.text if subtitle else ''
43 doujinshi_cover = html.find('div', attrs={'id': 'cover'})
44 img_id = re.search('/galleries/([\d]+)/cover\.(jpg|png)$', doujinshi_cover.a.img['src'])
46 logger.critical('Tried yo get image id failed')
48 doujinshi['img_id'] = img_id.group(1)
49 doujinshi['ext'] = img_id.group(2)
52 for _ in doujinshi_info.find_all('div', class_=''):
53 pages = re.search('([\d]+) pages', _.text)
55 pages = pages.group(1)
57 doujinshi['pages'] = int(pages)
61 def search_parser(keyword, page):
62 logger.debug('Searching doujinshis of keyword {}'.format(keyword))
65 response = request('get', url=constant.SEARCH_URL, params={'q': keyword, 'page': page}).content
66 except requests.ConnectionError as e:
68 logger.warn('If you are in China, please configure the proxy to fu*k GFW.')
71 html = BeautifulSoup(response)
72 doujinshi_search_result = html.find_all('div', attrs={'class': 'gallery'})
73 for doujinshi in doujinshi_search_result:
74 doujinshi_container = doujinshi.find('div', attrs={'class': 'caption'})
75 title = doujinshi_container.text.strip()
76 title = (title[:85] + '..') if len(title) > 85 else title
77 id_ = re.search('/g/(\d+)/', doujinshi.a['href']).group(1)
78 result.append({'id': id_, 'title': title})
82 def print_doujinshi(doujinshi_list):
83 if not doujinshi_list:
85 doujinshi_list = [i.values() for i in doujinshi_list]
86 headers = ['id', 'doujinshi']
87 logger.info('Search Result\n' +
88 tabulate(tabular_data=doujinshi_list, headers=headers, tablefmt='rst'))
90 if __name__ == '__main__':
91 print(doujinshi_parser("32271"))