# coding: utf-8
from __future__ import unicode_literals, print_function
-import sys
import os
import re
import time
return result
-def favorites_parser(page_range=''):
+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
- page_range_list = range(1, pages + 1)
- if page_range:
- logger.info('page range is {0}'.format(page_range))
- page_range_list = page_range_parser(page_range, pages)
+ page_range_list = range(1, pages + 1)
for page in page_range_list:
try:
return result
-def page_range_parser(page_range, max_page_num):
- pages = set()
- ranges = str.split(page_range, ',')
- for range_str in ranges:
- idx = range_str.find('-')
- if idx == -1:
- try:
- page = int(range_str)
- if page <= max_page_num:
- pages.add(page)
- except ValueError:
- logger.error('page range({0}) is not valid'.format(page_range))
- else:
- try:
- left = int(range_str[:idx])
- right = int(range_str[idx + 1:])
- if right > max_page_num:
- right = max_page_num
- for page in range(left, right + 1):
- pages.add(page)
- except ValueError:
- logger.error('page range({0}) is not valid'.format(page_range))
-
- return list(pages)
-
-
def doujinshi_parser(id_):
if not isinstance(id_, (int,)) and (isinstance(id_, (str,)) and not id_.isdigit()):
raise Exception('Doujinshi id({0}) is not valid'.format(id_))
tabulate(tabular_data=doujinshi_list, headers=headers, tablefmt='rst'))
-def search_parser(keyword, sorting, page):
- logger.debug('Searching doujinshis using keywords {0}'.format(keyword))
+def search_parser(keyword, sorting, page, is_page_all=False):
# keyword = '+'.join([i.strip().replace(' ', '-').lower() for i in keyword.split(',')])
result = []
- i = 0
- while i < 5:
- try:
- url = request('get', url=constant.SEARCH_URL, params={'query': keyword, 'page': page, 'sort': sorting}).url
- response = request('get', url.replace('%2B', '+')).json()
- except Exception as e:
- logger.critical(str(e))
+ if not page:
+ page = [1]
- break
-
- if 'result' not in response:
- raise Exception('No result in response')
+ 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)
- 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})
+ total = '/{0}'.format(page[-1]) if is_page_all else ''
+ for p in page:
+ i = 0
- if not result:
- logger.warn('No results for keywords {}'.format(keyword))
-
- return result
-
-
-def search_parser_all(keyword):
- logger.debug('Searching doujinshis using keywords {0}'.format(keyword))
-
- result = []
-
- url = request('get', url=constant.SEARCH_URL, params={'query': keyword}).url
- init_response = request('get', url.replace('%2B', '+')).json()
-
- for page in range(init_response['num_pages']):
- try:
- url = request('get', url=constant.SEARCH_URL, params={'query': keyword, 'page': page+1}).url
- response = request('get', url.replace('%2B', '+')).json()
-
- print('Obtained %d / %d pages.' % (page+1, init_response['num_pages']), end='\r')
+ 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))
- except Exception as e:
- logger.critical(str(e))
+ break
if 'result' not in response:
- raise Exception('No result in response')
+ logger.warn('No result in response in page {}'.format(p))
+ break
for row in response['result']:
title = row['title']['english']