2 from __future__ import print_function
6 from bs4 import BeautifulSoup
7 from constant import DETAIL_URL, SEARCH_URL
8 from logger import logger
9 from tabulate import tabulate
12 def doujinshi_parser(id):
13 if not isinstance(id, (int, )) and (isinstance(id, (str, )) and not id.isdigit()):
14 raise Exception('Doujinshi id(%s) is not valid' % str(id))
16 logger.debug('Fetching doujinshi information of id %d' % id)
19 url = '%s/%d/' % (DETAIL_URL, id)
22 response = requests.get(url).content
23 except Exception as e:
24 logger.critical('%s%s' % tuple(e.message))
27 html = BeautifulSoup(response)
28 doujinshi_info = html.find('div', attrs={'id': 'info'})
30 title = doujinshi_info.find('h1').text
31 subtitle = doujinshi_info.find('h2')
33 doujinshi['name'] = title
34 doujinshi['subtitle'] = subtitle.text if subtitle else ''
36 doujinshi_cover = html.find('div', attrs={'id': 'cover'})
37 img_id = re.search('/galleries/([\d]+)/cover\.(jpg|png)$', doujinshi_cover.a.img['src'])
39 logger.critical('Tried yo get image id failed')
41 doujinshi['img_id'] = img_id.group(1)
42 doujinshi['ext'] = img_id.group(2)
45 for _ in doujinshi_info.find_all('div', class_=''):
46 pages = re.search('([\d]+) pages', _.text)
48 pages = pages.group(1)
50 doujinshi['pages'] = int(pages)
54 def search_parser(keyword, page):
55 logger.debug('Searching doujinshis of keyword %s' % keyword)
57 response = requests.get(SEARCH_URL, params={'q': keyword, 'page': page}).content
58 html = BeautifulSoup(response)
59 doujinshi_search_result = html.find_all('div', attrs={'class': 'gallery'})
60 for doujinshi in doujinshi_search_result:
61 doujinshi_container = doujinshi.find('div', attrs={'class': 'caption'})
62 title = doujinshi_container.text.strip()
63 title = (title[:85] + '..') if len(title) > 85 else title
64 id_ = re.search('/g/(\d+)/', doujinshi.a['href']).group(1)
65 result.append({'id': id_, 'title': title})
69 def print_doujinshi(doujinshi_list):
70 if not doujinshi_list:
72 doujinshi_list = [i.values() for i in doujinshi_list]
73 headers = ['id', 'doujinshi']
74 logger.info('Search Result\n' +
75 tabulate(tabular_data=doujinshi_list, headers=headers, tablefmt='rst'))
77 if __name__ == '__main__':
78 print(doujinshi_parser("32271"))