diff options
| author | Adam Tauber <asciimoo@gmail.com> | 2015-02-12 10:52:55 +0100 |
|---|---|---|
| committer | Adam Tauber <asciimoo@gmail.com> | 2015-02-12 10:52:55 +0100 |
| commit | f6db77d81ea87d99462b4c3cc40a8a27e0264724 (patch) | |
| tree | b26fb71a62082aeec81c7bb1bb3d7447d006aed3 /searx/engines/duckduckgo.py | |
| parent | 516105c570a920dadeb87b34ee5ee434ad5cb16f (diff) | |
| parent | f96154b7c454a3b02bf688f248b4471c2020c28f (diff) | |
Merge pull request #210 from Cqoicebordel/unit-tests
unit tests
Diffstat (limited to 'searx/engines/duckduckgo.py')
| -rw-r--r-- | searx/engines/duckduckgo.py | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/searx/engines/duckduckgo.py b/searx/engines/duckduckgo.py index 583e33f73..e35a6334c 100644 --- a/searx/engines/duckduckgo.py +++ b/searx/engines/duckduckgo.py @@ -15,7 +15,7 @@ from urllib import urlencode from lxml.html import fromstring -from searx.utils import html_to_text +from searx.engines.xpath import extract_text # engine dependent config categories = ['general'] @@ -28,8 +28,8 @@ url = 'https://duckduckgo.com/html?{query}&s={offset}' # specific xpath variables result_xpath = '//div[@class="results_links results_links_deep web-result"]' # noqa url_xpath = './/a[@class="large"]/@href' -title_xpath = './/a[@class="large"]//text()' -content_xpath = './/div[@class="snippet"]//text()' +title_xpath = './/a[@class="large"]' +content_xpath = './/div[@class="snippet"]' # do search-request @@ -64,8 +64,8 @@ def response(resp): if not res_url: continue - title = html_to_text(''.join(r.xpath(title_xpath))) - content = html_to_text(''.join(r.xpath(content_xpath))) + title = extract_text(r.xpath(title_xpath)) + content = extract_text(r.xpath(content_xpath)) # append result results.append({'title': title, |