commit: 090254feca463dcd2243c67971c7e24e4907739c
parent: d23dd7e276f37a37106d7b6a925d7760c4f0e9b4
Author: Adam Tauber <asciimoo@gmail.com>
Date: Sun, 14 Sep 2014 18:57:08 +0200
Merge pull request #99 from dalf/master
[enh] stick results from the same category and template and [fix] rewrite the google engine
Diffstat:
2 files changed, 120 insertions(+), 26 deletions(-)
diff --git a/searx/engines/google.py b/searx/engines/google.py
@@ -1,15 +1,17 @@
## Google (Web)
#
# @website https://www.google.com
-# @provide-api yes (https://developers.google.com/web-search/docs/), deprecated!
+# @provide-api yes (https://developers.google.com/custom-search/)
#
-# @using-api yes
-# @results JSON
-# @stable yes (but deprecated)
-# @parse url, title, content
+# @using-api no
+# @results HTML
+# @stable no (HTML can change)
+# @parse url, title, content, suggestion
from urllib import urlencode
-from json import loads
+from urlparse import unquote,urlparse,parse_qsl
+from lxml import html
+from searx.engines.xpath import extract_text, extract_url
# engine dependent config
categories = ['general']
@@ -17,21 +19,45 @@ paging = True
language_support = True
# search-url
-url = 'https://ajax.googleapis.com/'
-search_url = url + 'ajax/services/search/web?v=2.0&start={offset}&rsz=large&safe=off&filter=off&{query}&hl={language}' # noqa
+google_hostname = 'www.google.com'
+search_path = '/search'
+redirect_path = '/url'
+images_path = '/images'
+search_url = 'https://' + google_hostname + search_path + '?{query}&start={offset}&gbv=1'
+# specific xpath variables
+results_xpath= '//li[@class="g"]'
+url_xpath = './/h3/a/@href'
+title_xpath = './/h3'
+content_xpath = './/span[@class="st"]'
+suggestion_xpath = '//p[@class="_Bmc"]'
+
+images_xpath = './/div/a'
+image_url_xpath = './@href'
+image_img_src_xpath = './img/@src'
+
+# remove google-specific tracking-url
+def parse_url(url_string):
+ parsed_url = urlparse(url_string)
+ if parsed_url.netloc in [google_hostname, ''] and parsed_url.path==redirect_path:
+ query = dict(parse_qsl(parsed_url.query))
+ return query['q']
+ else:
+ return url_string
# do search-request
def request(query, params):
- offset = (params['pageno'] - 1) * 8
+ offset = (params['pageno'] - 1) * 10
- language = 'en-US'
- if params['language'] != 'all':
- language = params['language'].replace('_', '-')
+ if params['language'] == 'all':
+ language = 'en'
+ else:
+ language = params['language'].replace('_','-').lower()
params['url'] = search_url.format(offset=offset,
- query=urlencode({'q': query}),
- language=language)
+ query=urlencode({'q': query}))
+
+ params['headers']['Accept-Language'] = language
return params
@@ -40,18 +66,50 @@ def request(query, params):
def response(resp):
results = []
- search_res = loads(resp.text)
-
- # return empty array if there are no results
- if not search_res.get('responseData', {}).get('results'):
- return []
+ dom = html.fromstring(resp.text)
# parse results
- for result in search_res['responseData']['results']:
- # append result
- results.append({'url': result['unescapedUrl'],
- 'title': result['titleNoFormatting'],
- 'content': result['content']})
+ for result in dom.xpath(results_xpath):
+ title = extract_text(result.xpath(title_xpath)[0])
+ try:
+ url = parse_url(extract_url(result.xpath(url_xpath), search_url))
+ parsed_url = urlparse(url)
+ if parsed_url.netloc==google_hostname and parsed_url.path==search_path:
+ # remove the link to google news
+ continue
+
+ if parsed_url.netloc==google_hostname and parsed_url.path==images_path:
+ # images result
+ results = results + parse_images(result)
+ else:
+ # normal result
+ content = extract_text(result.xpath(content_xpath)[0])
+ # append result
+ results.append({'url': url,
+ 'title': title,
+ 'content': content})
+ except:
+ continue
+
+ # parse suggestion
+ for suggestion in dom.xpath(suggestion_xpath):
+ # append suggestion
+ results.append({'suggestion': extract_text(suggestion)})
# return results
return results
+
+def parse_images(result):
+ results = []
+ for image in result.xpath(images_xpath):
+ url = parse_url(extract_text(image.xpath(image_url_xpath)[0]))
+ img_src = extract_text(image.xpath(image_img_src_xpath)[0])
+
+ # append result
+ results.append({'url': url,
+ 'title': '',
+ 'content': '',
+ 'img_src': img_src,
+ 'template': 'images.html'})
+
+ return results
diff --git a/searx/search.py b/searx/search.py
@@ -49,7 +49,8 @@ def score_results(results):
flat_len = len(flat_res)
engines_len = len(results)
results = []
- # deduplication + scoring
+
+ # pass 1: deduplication + scoring
for i, res in enumerate(flat_res):
res['parsed_url'] = urlparse(res['url'])
@@ -90,7 +91,42 @@ def score_results(results):
else:
res['score'] = score
results.append(res)
- return sorted(results, key=itemgetter('score'), reverse=True)
+ results = sorted(results, key=itemgetter('score'), reverse=True)
+
+ # pass 2 : group results by category and template
+ gresults = []
+ categoryPositions = {}
+
+ for i, res in enumerate(results):
+ # FIXME : handle more than one category per engine
+ category = engines[res['engine']].categories[0] + ':' + '' if 'template' not in res else res['template']
+
+ current = None if category not in categoryPositions else categoryPositions[category]
+
+ # group with previous results using the same category if the group can accept more result and is not too far from the current position
+ if current != None and (current['count'] > 0) and (len(gresults) - current['index'] < 20):
+ # group with the previous results using the same category with this one
+ index = current['index']
+ gresults.insert(index, res)
+
+ # update every index after the current one (including the current one)
+ for k in categoryPositions:
+ v = categoryPositions[k]['index']
+ if v >= index:
+ categoryPositions[k]['index'] = v+1
+
+ # update this category
+ current['count'] -= 1
+
+ else:
+ # same category
+ gresults.append(res)
+
+ # update categoryIndex
+ categoryPositions[category] = { 'index' : len(gresults), 'count' : 8 }
+
+ # return gresults
+ return gresults
class Search(object):
