mirror of
				https://github.com/searxng/searxng.git
				synced 2025-10-31 02:27:06 -04:00 
			
		
		
		
	move meta information from comment to the about variable so the preferences, the documentation can show these information
		
			
				
	
	
		
			128 lines
		
	
	
		
			4.0 KiB
		
	
	
	
		
			Python
		
	
	
	
	
	
			
		
		
	
	
			128 lines
		
	
	
		
			4.0 KiB
		
	
	
	
		
			Python
		
	
	
	
	
	
| # SPDX-License-Identifier: AGPL-3.0-or-later
 | |
| """
 | |
|  Flickr (Images)
 | |
| """
 | |
| 
 | |
| from json import loads
 | |
| from time import time
 | |
| import re
 | |
| from urllib.parse import urlencode
 | |
| from searx.engines import logger
 | |
| from searx.utils import ecma_unescape, html_to_text
 | |
| 
 | |
| logger = logger.getChild('flickr-noapi')
 | |
| 
 | |
| # about
 | |
| about = {
 | |
|     "website": 'https://www.flickr.com',
 | |
|     "wikidata_id": 'Q103204',
 | |
|     "official_api_documentation": 'https://secure.flickr.com/services/api/flickr.photos.search.html',
 | |
|     "use_official_api": False,
 | |
|     "require_api_key": False,
 | |
|     "results": 'HTML',
 | |
| }
 | |
| 
 | |
| categories = ['images']
 | |
| 
 | |
| url = 'https://www.flickr.com/'
 | |
| search_url = url + 'search?{query}&page={page}'
 | |
| time_range_url = '&min_upload_date={start}&max_upload_date={end}'
 | |
| photo_url = 'https://www.flickr.com/photos/{userid}/{photoid}'
 | |
| modelexport_re = re.compile(r"^\s*modelExport:\s*({.*}),$", re.M)
 | |
| image_sizes = ('o', 'k', 'h', 'b', 'c', 'z', 'n', 'm', 't', 'q', 's')
 | |
| 
 | |
| paging = True
 | |
| time_range_support = True
 | |
| time_range_dict = {'day': 60 * 60 * 24,
 | |
|                    'week': 60 * 60 * 24 * 7,
 | |
|                    'month': 60 * 60 * 24 * 7 * 4,
 | |
|                    'year': 60 * 60 * 24 * 7 * 52}
 | |
| 
 | |
| 
 | |
| def build_flickr_url(user_id, photo_id):
 | |
|     return photo_url.format(userid=user_id, photoid=photo_id)
 | |
| 
 | |
| 
 | |
| def _get_time_range_url(time_range):
 | |
|     if time_range in time_range_dict:
 | |
|         return time_range_url.format(start=time(), end=str(int(time()) - time_range_dict[time_range]))
 | |
|     return ''
 | |
| 
 | |
| 
 | |
| def request(query, params):
 | |
|     params['url'] = (search_url.format(query=urlencode({'text': query}), page=params['pageno'])
 | |
|                      + _get_time_range_url(params['time_range']))
 | |
|     return params
 | |
| 
 | |
| 
 | |
| def response(resp):
 | |
|     results = []
 | |
| 
 | |
|     matches = modelexport_re.search(resp.text)
 | |
| 
 | |
|     if matches is None:
 | |
|         return results
 | |
| 
 | |
|     match = matches.group(1)
 | |
|     model_export = loads(match)
 | |
| 
 | |
|     if 'legend' not in model_export:
 | |
|         return results
 | |
| 
 | |
|     legend = model_export['legend']
 | |
| 
 | |
|     # handle empty page
 | |
|     if not legend or not legend[0]:
 | |
|         return results
 | |
| 
 | |
|     for index in legend:
 | |
|         photo = model_export['main'][index[0]][int(index[1])][index[2]][index[3]][int(index[4])]
 | |
|         author = ecma_unescape(photo.get('realname', ''))
 | |
|         source = ecma_unescape(photo.get('username', '')) + ' @ Flickr'
 | |
|         title = ecma_unescape(photo.get('title', ''))
 | |
|         content = html_to_text(ecma_unescape(photo.get('description', '')))
 | |
|         img_src = None
 | |
|         # From the biggest to the lowest format
 | |
|         for image_size in image_sizes:
 | |
|             if image_size in photo['sizes']:
 | |
|                 img_src = photo['sizes'][image_size]['url']
 | |
|                 img_format = 'jpg ' \
 | |
|                     + str(photo['sizes'][image_size]['width']) \
 | |
|                     + 'x' \
 | |
|                     + str(photo['sizes'][image_size]['height'])
 | |
|                 break
 | |
| 
 | |
|         if not img_src:
 | |
|             logger.debug('cannot find valid image size: {0}'.format(repr(photo)))
 | |
|             continue
 | |
| 
 | |
|         # For a bigger thumbnail, keep only the url_z, not the url_n
 | |
|         if 'n' in photo['sizes']:
 | |
|             thumbnail_src = photo['sizes']['n']['url']
 | |
|         elif 'z' in photo['sizes']:
 | |
|             thumbnail_src = photo['sizes']['z']['url']
 | |
|         else:
 | |
|             thumbnail_src = img_src
 | |
| 
 | |
|         if 'ownerNsid' not in photo:
 | |
|             # should not happen, disowned photo? Show it anyway
 | |
|             url = img_src
 | |
|         else:
 | |
|             url = build_flickr_url(photo['ownerNsid'], photo['id'])
 | |
| 
 | |
|         result = {
 | |
|             'url': url,
 | |
|             'img_src': img_src,
 | |
|             'thumbnail_src': thumbnail_src,
 | |
|             'source': source,
 | |
|             'img_format': img_format,
 | |
|             'template': 'images.html'
 | |
|         }
 | |
|         result['author'] = author.encode(errors='ignore').decode()
 | |
|         result['source'] = source.encode(errors='ignore').decode()
 | |
|         result['title'] = title.encode(errors='ignore').decode()
 | |
|         result['content'] = content.encode(errors='ignore').decode()
 | |
|         results.append(result)
 | |
| 
 | |
|     return results
 |