1 from __future__ import print_function
5 from bs4 import BeautifulSoup
6 from constant import DETAIL_URL, SEARCH_URL
7 from logger import logger
8 from tabulate import tabulate
11 def dojinshi_parser(id):
12 if not isinstance(id, (int, )) and (isinstance(id, (str, )) and not id.isdigit()):
13 raise Exception('Dojinshi id(%s) is not valid' % str(id))
15 logger.debug('Fetching dojinshi information of id %d' % id)
18 url = '%s/%d/' % (DETAIL_URL, id)
21 response = requests.get(url).content
22 except Exception as e:
23 logger.critical('%s%s' % tuple(e.message))
26 html = BeautifulSoup(response)
27 dojinshi_info = html.find('div', attrs={'id': 'info'})
29 title = dojinshi_info.find('h1').text
30 subtitle = dojinshi_info.find('h2')
32 dojinshi['name'] = title
33 dojinshi['subtitle'] = subtitle.text if subtitle else ''
35 dojinshi_cover = html.find('div', attrs={'id': 'cover'})
36 img_id = re.search('/galleries/([\d]+)/cover\.(jpg|png)$', dojinshi_cover.a.img['src'])
38 logger.critical('Tried yo get image id failed')
40 dojinshi['img_id'] = img_id.group(1)
41 dojinshi['ext'] = img_id.group(2)
44 for _ in dojinshi_info.find_all('div', class_=''):
45 pages = re.search('([\d]+) pages', _.text)
47 pages = pages.group(1)
49 dojinshi['pages'] = int(pages)
53 def search_parser(keyword, page):
54 logger.debug('Searching dojinshis of keyword %s' % keyword)
56 response = requests.get(SEARCH_URL, params={'q': keyword, 'page': page}).content
57 html = BeautifulSoup(response)
58 dojinshi_search_result = html.find_all('div', attrs={'class': 'gallery'})
59 for dojinshi in dojinshi_search_result:
60 dojinshi_container = dojinshi.find('div', attrs={'class': 'caption'})
61 title = dojinshi_container.text.strip()
62 title = (title[:85] + '..') if len(title) > 85 else title
63 id_ = re.search('/g/(\d+)/', dojinshi.a['href']).group(1)
64 result.append({'id': id_, 'title': title})
68 def print_dojinshi(dojinshi_list):
71 dojinshi_list = [i.values() for i in dojinshi_list]
72 headers = ['id', 'dojinshi']
73 logger.info('Search Result\n' +
74 tabulate(tabular_data=dojinshi_list, headers=headers, tablefmt='rst'))
76 if __name__ == '__main__':
77 print(dojinshi_parser("32271"))