# coding: utf-8
-from __future__ import unicode_literals, print_function
import os
import re
-import threadpool
-import requests
import time
from bs4 import BeautifulSoup
from tabulate import tabulate
return result
-def favorites_parser():
+def favorites_parser(page=None):
result = []
html = BeautifulSoup(request('get', constant.FAV_URL).content, 'html.parser')
count = html.find('span', attrs={'class': 'count'})
return []
pages = int(count / 25)
- if pages:
- pages += 1 if count % (25 * pages) else 0
+ if page:
+ page_range_list = page
else:
- pages = 1
+ if pages:
+ pages += 1 if count % (25 * pages) else 0
+ else:
+ pages = 1
- logger.info('You have %d favorites in %d pages.' % (count, pages))
+ logger.info('You have %d favorites in %d pages.' % (count, pages))
- if os.getenv('DEBUG'):
- pages = 1
+ if os.getenv('DEBUG'):
+ pages = 1
- for page in range(1, pages + 1):
+ page_range_list = range(1, pages + 1)
+
+ for page in page_range_list:
try:
logger.info('Getting doujinshi ids of page %d' % page)
resp = request('get', constant.FAV_URL + '?page=%d' % page).content
response = request('get', url)
if response.status_code in (200, ):
response = response.content
+ elif response.status_code in (404,):
+ logger.error("Doujinshi with id {0} cannot be found".format(id_))
+ return []
else:
logger.debug('Slow down and retry ({}) ...'.format(id_))
time.sleep(1)
doujinshi['subtitle'] = subtitle.text if subtitle else ''
doujinshi_cover = html.find('div', attrs={'id': 'cover'})
- img_id = re.search('/galleries/([\d]+)/cover\.(jpg|png)$', doujinshi_cover.a.img.attrs['data-src'])
+ img_id = re.search('/galleries/([\d]+)/cover\.(jpg|png|gif)$', doujinshi_cover.a.img.attrs['data-src'])
ext = []
for i in html.find_all('div', attrs={'class': 'thumb-container'}):
doujinshi['img_id'] = img_id.group(1)
doujinshi['ext'] = ext
- pages = 0
- for _ in doujinshi_info.find_all('div', class_=''):
- pages = re.search('([\d]+) pages', _.text)
- if pages:
- pages = pages.group(1)
- break
+ for _ in doujinshi_info.find_all('div', class_='tag-container field-name'):
+ if re.search('Pages:', _.text):
+ pages = _.find('span', class_='name').string
doujinshi['pages'] = int(pages)
# gain information of the doujinshi
information_fields = doujinshi_info.find_all('div', attrs={'class': 'field-name'})
- needed_fields = ['Characters', 'Artists', 'Languages', 'Tags']
+ needed_fields = ['Characters', 'Artists', 'Languages', 'Tags', 'Parodies', 'Groups', 'Categories']
for field in information_fields:
field_name = field.contents[0].strip().strip(':')
if field_name in needed_fields:
- data = [sub_field.contents[0].strip() for sub_field in
+ data = [sub_field.find('span', attrs={'class': 'name'}).contents[0].strip() for sub_field in
field.find_all('a', attrs={'class': 'tag'})]
doujinshi[field_name.lower()] = ', '.join(data)
+ time_field = doujinshi_info.find('time')
+ if time_field.has_attr('datetime'):
+ doujinshi['date'] = time_field['datetime']
return doujinshi
-def search_parser(keyword, sorting='date', page=1):
+def old_search_parser(keyword, sorting='date', page=1):
logger.debug('Searching doujinshis of keyword {0}'.format(keyword))
response = request('get', url=constant.SEARCH_URL, params={'q': keyword, 'page': page, 'sort': sorting}).content
return
doujinshi_list = [(i['id'], i['title']) for i in doujinshi_list]
headers = ['id', 'doujinshi']
- logger.info('Search Result\n' +
+ logger.info('Search Result || Found %i doujinshis \n' % doujinshi_list.__len__() +
tabulate(tabular_data=doujinshi_list, headers=headers, tablefmt='rst'))
-def tag_parser(tag_name, sorting='date', max_page=1, index=0):
+def search_parser(keyword, sorting, page, is_page_all=False):
+ # keyword = '+'.join([i.strip().replace(' ', '-').lower() for i in keyword.split(',')])
result = []
- tag_name = tag_name.lower()
- if ',' in tag_name:
- tag_name = [i.strip().replace(' ', '-') for i in tag_name.split(',')]
- else:
- tag_name = tag_name.replace(' ', '-')
- if sorting == 'date':
- sorting = ''
-
- for p in range(1, max_page + 1):
- if isinstance(tag_name, str):
- logger.debug('Fetching page {0} for doujinshi with tag \'{1}\''.format(p, tag_name))
- response = request('get', url='%s/%s/%s?page=%d' % (constant.TAG_URL[index], tag_name, sorting, p)).content
- result += _get_title_and_id(response)
- else:
- for i in tag_name:
- logger.debug('Fetching page {0} for doujinshi with tag \'{1}\''.format(p, i))
- response = request('get',
- url='%s/%s/%s?page=%d' % (constant.TAG_URL[index], i, sorting, p)).content
- result += _get_title_and_id(response)
-
- if not result:
- logger.error('Cannot find doujinshi id of tag \'{0}\''.format(tag_name))
- return
-
- if not result:
- logger.warn('No results for tag \'{}\''.format(tag_name))
-
- return result
-
-
-def __api_suspended_search_parser(keyword, sorting, page):
- logger.debug('Searching doujinshis using keywords {0}'.format(keyword))
- result = []
- i = 0
- while i < 5:
- try:
- response = request('get', url=constant.SEARCH_URL, params={'query': keyword, 'page': page, 'sort': sorting}).json()
- except Exception as e:
- i += 1
- if not i < 5:
+ if not page:
+ page = [1]
+
+ if is_page_all:
+ url = request('get', url=constant.SEARCH_URL, params={'query': keyword}).url
+ init_response = request('get', url.replace('%2B', '+')).json()
+ page = range(1, init_response['num_pages']+1)
+
+ total = '/{0}'.format(page[-1]) if is_page_all else ''
+ for p in page:
+ i = 0
+
+ logger.info('Searching doujinshis using keywords "{0}" on page {1}{2}'.format(keyword, p, total))
+ while i < 3:
+ try:
+ url = request('get', url=constant.SEARCH_URL, params={'query': keyword,
+ 'page': p, 'sort': sorting}).url
+ response = request('get', url.replace('%2B', '+')).json()
+ except Exception as e:
logger.critical(str(e))
- logger.warn('If you are in China, please configure the proxy to fu*k GFW.')
- exit(1)
- continue
- break
-
- if 'result' not in response:
- raise Exception('No result in response')
-
- for row in response['result']:
- title = row['title']['english']
- title = title[:85] + '..' if len(title) > 85 else title
- result.append({'id': row['id'], 'title': title})
-
- if not result:
- logger.warn('No results for keywords {}'.format(keyword))
-
- return result
+ break
-def __api_suspended_tag_parser(tag_id, sorting, max_page=1):
- logger.info('Searching for doujinshi with tag id {0}'.format(tag_id))
- result = []
- response = request('get', url=constant.TAG_API_URL, params={'sort': sorting, 'tag_id': tag_id}).json()
- page = max_page if max_page <= response['num_pages'] else int(response['num_pages'])
-
- for i in range(1, page + 1):
- logger.info('Getting page {} ...'.format(i))
+ if 'result' not in response:
+ logger.warn('No result in response in page {}'.format(p))
+ break
- if page != 1:
- response = request('get', url=constant.TAG_API_URL,
- params={'sort': sorting, 'tag_id': tag_id}).json()
- for row in response['result']:
- title = row['title']['english']
- title = title[:85] + '..' if len(title) > 85 else title
- result.append({'id': row['id'], 'title': title})
+ for row in response['result']:
+ title = row['title']['english']
+ title = title[:85] + '..' if len(title) > 85 else title
+ result.append({'id': row['id'], 'title': title})
- if not result:
- logger.warn('No results for tag id {}'.format(tag_id))
+ if not result:
+ logger.warn('No results for keywords {}'.format(keyword))
return result
doujinshi['pages'] = len(response['images']['pages'])
# gain information of the doujinshi
- needed_fields = ['character', 'artist', 'language', 'tag']
+ needed_fields = ['character', 'artist', 'language', 'tag', 'parody', 'group', 'category']
for tag in response['tags']:
tag_type = tag['type']
if tag_type in needed_fields: