searx

commit: dae88d862b8de2a7cddf461045ef381b07819fe5
parent 334a286c18652423447aff589ec0f25406e4d8ea
Author: Thomas Pointhuber <thomas.pointhuber@gmx.at>
Date:   Tue,  2 Sep 2014 17:56:32 +0200

update piratebay engine and add comments

Diffstat:
M
searx/engines/piratebay.py
41
++++++++++++++++++++++++++++++++++-------
M
searx/settings.yml
1
-

2 files changed, 34 insertions(+), 8 deletions(-)
diff --git a/searx/engines/piratebay.py b/searx/engines/piratebay.py
@@ -1,39 +1,61 @@
+## Piratebay (Videos, Music, Files)
+# 
+# @website     https://thepiratebay.se
+# @provide-api no (nothing found)
+# 
+# @using-api   no
+# @results     HTML (using search portal)
+# @stable      yes (HTML can change)
+# @parse       url, title, content, seed, leech, magnetlink
+
 from urlparse import urljoin
 from cgi import escape
 from urllib import quote
 from lxml import html
 from operator import itemgetter
 
-categories = ['videos', 'music']
+# engine dependent config
+categories = ['videos', 'music', 'files']
+paging = True
 
+# search-url
 url = 'https://thepiratebay.se/'
 search_url = url + 'search/{search_term}/{pageno}/99/{search_type}'
-search_types = {'videos': '200',
+
+# piratebay specific type-definitions
+search_types = {'files': '0',                
                 'music': '100',
-                'files': '0'}
+                'videos': '200'}
 
+# specific xpath variables
 magnet_xpath = './/a[@title="Download this torrent using magnet"]'
 content_xpath = './/font[@class="detDesc"]//text()'
 
-paging = True
-
 
+# do search-request
 def request(query, params):
-    search_type = search_types.get(params['category'], '200')
+    search_type = search_types.get(params['category'], '0')
+
     params['url'] = search_url.format(search_term=quote(query),
                                       search_type=search_type,
                                       pageno=params['pageno'] - 1)
+
     return params
 
 
+# get response from search-request
 def response(resp):
     results = []
+
     dom = html.fromstring(resp.text)
+
     search_res = dom.xpath('//table[@id="searchResult"]//tr')
 
+    # return empty array if nothing is found
     if not search_res:
-        return results
+        return []
 
+    # parse results
     for result in search_res[1:]:
         link = result.xpath('.//div[@class="detName"]//a')[0]
         href = urljoin(url, link.attrib.get('href'))
@@ -41,17 +63,21 @@ def response(resp):
         content = escape(' '.join(result.xpath(content_xpath)))
         seed, leech = result.xpath('.//td[@align="right"]/text()')[:2]
 
+        # convert seed to int if possible
         if seed.isdigit():
             seed = int(seed)
         else:
             seed = 0
 
+        # convert leech to int if possible
         if leech.isdigit():
             leech = int(leech)
         else:
             leech = 0
 
         magnetlink = result.xpath(magnet_xpath)[0]
+
+        # append result
         results.append({'url': href,
                         'title': title,
                         'content': content,
@@ -60,4 +86,5 @@ def response(resp):
                         'magnetlink': magnetlink.attrib['href'],
                         'template': 'torrent.html'})
 
+    # return results sorted by seeder
     return sorted(results, key=itemgetter('seed'), reverse=True)
diff --git a/searx/settings.yml b/searx/settings.yml
@@ -82,7 +82,6 @@ engines:
 
   - name : piratebay
     engine : piratebay
-    categories : videos, music, files
     shortcut : tpb
 
   - name : soundcloud