11fc991 · searx · git.iko.soy

git.iko.soy/searx

on master help sign in

[enh] bing_images: use data from embedded JSON to improve results (e.g. real page title) (#1568)

11fc991Frank de Lange7 years ago

searx/engines/bing_images.py+33−34

compare with master

unifiedsplit

@@ -10,9 +10,6 @@

10 @stable no (HTML can change)

11 @parse url, title, img_src

13− @todo currently there are up to 35 images receive per page,

14− because bing does not parse count=10.

15− limited response to 10 images

16"""

13"""

18from lxml import html

15from lxml import html

@@ -28,10 +25,15 @@ safesearch = True

28time_range_support = True

25time_range_support = True

29language_support = True

26language_support = True

30supported_languages_url = 'https://www.bing.com/account/general'

27supported_languages_url = 'https://www.bing.com/account/general'

28+number_of_results = 28

32# search-url

30# search-url

33base_url = 'https://www.bing.com/'

31base_url = 'https://www.bing.com/'

34−search_string = 'images/search?{query}&count=10&first={offset}'

32+search_string = 'images/search'\

33+ '?{query}'\

34+ '&count={count}'\

35+ '&first={first}'\

36+ '&FORM=IBASEP'

35time_range_string = '&qft=+filterui:age-lt{interval}'

37time_range_string = '&qft=+filterui:age-lt{interval}'

36time_range_dict = {'day': '1440',

38time_range_dict = {'day': '1440',

37 'week': '10080',

39 'week': '10080',

@@ -44,16 +46,14 @@ safesearch_types = {2: 'STRICT',

44 0: 'OFF'}

46 0: 'OFF'}

47−_quote_keys_regex = re.compile('({|,)([a-z][a-z0-9]*):(")', re.I | re.U)

48−

49−

50# do search-request

49# do search-request

51def request(query, params):

50def request(query, params):

52− offset = (params['pageno'] - 1) * 10 + 1

51+ offset = ((params['pageno'] - 1) * number_of_results) + 1

54 search_path = search_string.format(

53 search_path = search_string.format(

55 query=urlencode({'q': query}),

54 query=urlencode({'q': query}),

55+ count=number_of_results,

56− offset=offset)

56+ first=offset)

58 language = match_language(params['language'], supported_languages, language_aliases).lower()

@@ -77,32 +77,31 @@ def response(resp):

77 dom = html.fromstring(resp.text)

79 # parse results

80− for result in dom.xpath('//div[@id="mmComponent_images_1"]/ul/li/div/div[@class="imgpt"]'):

80+ for result in dom.xpath('//div[@class="imgpt"]'):

81− link = result.xpath('./a')[0]

81+

82−

82+ img_format = result.xpath('./div[contains(@class, "img_info")]/span/text()')[0]

83− # TODO find actual title

83+ # Microsoft seems to experiment with this code so don't make the path too specific,

84− title = link.xpath('.//img/@alt')[0]

84+ # just catch the text section for the first anchor in img_info assuming this to be

85−

85+ # the originating site.

86− # parse json-data (it is required to add a space, to make it parsable)

86+ source = result.xpath('./div[contains(@class, "img_info")]//a/text()')[0]

87− json_data = loads(_quote_keys_regex.sub(r'\1"\2": \3', link.attrib.get('m')))

87+

88−

88+ try:

89− url = json_data.get('purl')

89+ m = loads(result.xpath('./a/@m')[0])

90− img_src = json_data.get('murl')

90+

91− thumbnail = json_data.get('turl')

91+ # strip 'Unicode private use area' highlighting, they render to Tux

92−

92+ # the Linux penguin and a standing diamond on my machine...

93− # append result

93+ title = m.get('t', '').replace(u'\ue000', '').replace(u'\ue001', '')

94− results.append({'template': 'images.html',

94+ results.append({'template': 'images.html',

95− 'url': url,

95+ 'url': m['purl'],

96− 'title': title,

97− 'content': '',

98− 'thumbnail_src': thumbnail,

96+ 'thumbnail_src': m['turl'],

99− 'img_src': img_src})

97+ 'img_src': m['murl'],

100−

98+ 'content': '',

101− # TODO stop parsing if 10 images are found

99+ 'title': title,

102− # if len(results) >= 10:

100+ 'source': source,

103− # break

101+ 'img_format': img_format})

102+ except:

103+ continue

104

105− # return results

106 return results

105 return results

107

106

108

107

tests/unit/engines/test_bing_images.py+18−4

@@ -53,17 +53,25 @@ class TestBingImagesEngine(SearxTestCase):

53 <li>

54 <div>

55 <div class="imgpt">

56− <a m='{"purl":"page_url","murl":"img_url","turl":"thumb_url"}'>

56+ <a m='{"purl":"page_url","murl":"img_url","turl":"thumb_url","t":"Page 1 title"}'>

57 <img src="" alt="alt text" />

58 </a>

59+ <div class="img_info">

60+ <span>1 x 1 - jpeg</span>

61+ <a>1.example.org</a>

62+ </div>

59 </div>

63 </div>

60 <div></div>

64 <div></div>

61 </div>

65 </div>

62 <div>

66 <div>

63 <div class="imgpt">

67 <div class="imgpt">

64− <a m='{"purl":"page_url2","murl":"img_url2","turl":"thumb_url2"}'>

68+ <a m='{"purl":"page_url2","murl":"img_url2","turl":"thumb_url2","t":"Page 2 title"}'>

65 <img src="" alt="alt text 2" />

69 <img src="" alt="alt text 2" />

66 </a>

70 </a>

71+ <div class="img_info">

72+ <span>2 x 2 - jpeg</span>

73+ <a>2.example.org</a>

74+ </div>

67 </div>

75 </div>

68 </div>

76 </div>

69 </li>

77 </li>

@@ -72,9 +80,13 @@ class TestBingImagesEngine(SearxTestCase):

72 <li>

80 <li>

73 <div>

81 <div>

74 <div class="imgpt">

82 <div class="imgpt">

75− <a m='{"purl":"page_url3","murl":"img_url3","turl":"thumb_url3"}'>

83+ <a m='{"purl":"page_url3","murl":"img_url3","turl":"thumb_url3","t":"Page 3 title"}'>

76 <img src="" alt="alt text 3" />

84 <img src="" alt="alt text 3" />

77 </a>

85 </a>

86+ <div class="img_info">

87+ <span>3 x 3 - jpeg</span>

88+ <a>3.example.org</a>

89+ </div>

78 </div>

90 </div>

79 </div>

91 </div>

80 </li>

92 </li>

@@ -86,11 +98,13 @@ class TestBingImagesEngine(SearxTestCase):

86 results = bing_images.response(response)

98 results = bing_images.response(response)

87 self.assertEqual(type(results), list)

99 self.assertEqual(type(results), list)

88 self.assertEqual(len(results), 3)

100 self.assertEqual(len(results), 3)

89− self.assertEqual(results[0]['title'], 'alt text')

101+ self.assertEqual(results[0]['title'], 'Page 1 title')

90 self.assertEqual(results[0]['url'], 'page_url')

102 self.assertEqual(results[0]['url'], 'page_url')

91 self.assertEqual(results[0]['content'], '')

103 self.assertEqual(results[0]['content'], '')

92 self.assertEqual(results[0]['thumbnail_src'], 'thumb_url')

104 self.assertEqual(results[0]['thumbnail_src'], 'thumb_url')

93 self.assertEqual(results[0]['img_src'], 'img_url')

105 self.assertEqual(results[0]['img_src'], 'img_url')

106+ self.assertEqual(results[0]['img_format'], '1 x 1 - jpeg')

107+ self.assertEqual(results[0]['source'], '1.example.org')

108

95 def test_fetch_supported_languages(self):

109 def test_fetch_supported_languages(self):

96 html = """

110 html = """