efb6eca · searx · git.iko.soy

git.iko.soy/searx

Merge branch 'master' of https://github.com/asciimoo/searx

commitefb6eca39e9dd851645f76bd72ac076f61a11c09 authorpw3t <romain@berthor.fr> dateThu 2 Jan 2014, 19:52 (13 years ago) parent6f2b8ac d2898b0

15 files changed+131−26

compare with master

unifiedsplit

README.md+1−1 engines.cfg_sample+5 searx/engines/dailymotion.py+32 searx/engines/flickr.py+1−1 searx/engines/piratebay.py+1−1 searx/engines/xpath.py+1−1 searx/settings.py+1−1 searx/static/js/searx.js+27 searx/templates/base.html+1 searx/templates/categories.html+5 searx/templates/index.html+1 searx/templates/preferences.html+19 searx/templates/results.html+4−3 searx/templates/search.html+1−5 searx/webapp.py+31−13

README.md+1−1

@@ -13,6 +13,7 @@ List of [running instances](https://github.com/asciimoo/searx/wiki/Searx-instanc

1313* Modular (see [examples](https://github.com/asciimoo/searx/blob/master/examples))

1414* Parallel queries

1515* Supports json output `curl https://searx.0x2a.tk/?format=json&q=[query]`

16+* Supports csv output `curl https://searx.0x2a.tk/?format=csv&q=[query]`

1617* Opensearch support (you can set as default search engine)

1718* Configurable search engines/categories

1819* User-agent forwarding

@@ -32,7 +33,6 @@ List of [running instances](https://github.com/asciimoo/searx/wiki/Searx-instanc

3233* Language support

3334* Documentation

3435* Pagination

35−* Search suggestions

3636* Tests

3737

3838

engines.cfg_sample+5

@@ -79,3 +79,8 @@ suggestion_xpath = //div[@id="satat"]//a

7979[youtube]

8080engine = youtube

8181categories = videos

82+

83+[dailymotion]

84+engine = dailymotion

85+categories = videos

86+

searx/engines/dailymotion.pynew file+32

@@ -0,0 +1,32 @@

1+from urllib import urlencode

2+from json import loads

3+from cgi import escape

5+categories = ['videos']

6+localization = 'en'

8+# see http://www.dailymotion.com/doc/api/obj-video.html

9+search_url = 'https://api.dailymotion.com/videos?fields=title,description,duration,url,thumbnail_360_url&sort=relevance&limit=25&page=1&{query}'

10+

11+def request(query, params):

12+ global search_url

13+ params['url'] = search_url.format(query=urlencode({'search': query, 'localization': localization }))

14+ return params

15+

16+

17+def response(resp):

18+ results = []

19+ search_res = loads(resp.text)

20+ if not 'list' in search_res:

21+ return results

22+ for res in search_res['list']:

23+ title = res['title']

24+ url = res['url']

25+ if res['thumbnail_360_url']:

26+ content = '<a href="{0}" title="{0}" ><img src="{1}" /></a><br />'.format(url, res['thumbnail_360_url'])

27+ else:

28+ content = ''

29+ if res['description']:

30+ content += escape(res['description'][:500])

31+ results.append({'url': url, 'title': title, 'content': content})

32+ return results

searx/engines/flickr.py+1−1

@@ -7,7 +7,7 @@ from urlparse import urljoin

77categories = ['images']

99url = 'https://secure.flickr.com/'

10−search_url = url+'search/?q={query}'

10+search_url = url+'search/?{query}'

1111

1212def request(query, params):

1313 params['url'] = search_url.format(query=urlencode({'q': query}))

searx/engines/piratebay.py+1−1

@@ -5,7 +5,7 @@ from urllib import quote

66categories = ['videos', 'music']

8−url = 'https://thepiratebay.sx/'

8+url = 'https://thepiratebay.se/'

99search_url = url + 'search/{search_term}/0/99/{search_type}'

1010search_types = {'videos': '200'

1111 ,'music' : '100'

searx/engines/xpath.py+1−1

@@ -28,7 +28,7 @@ def extract_url(xpath_results):

2828 url = xpath_results[0].attrib.get('href')

2929 else:

3030 url = xpath_results.attrib.get('href')

31− if not url.startswith('http://') or not url.startswith('https://'):

31+ if not url.startswith('http://') and not url.startswith('https://'):

3232 url = 'http://'+url

3333 parsed_url = urlparse(url)

3434 if not parsed_url.netloc:

searx/settings.py+1−1

@@ -13,4 +13,4 @@ blacklist = [] # search engine blacklist

1313

1414categories = {} # custom search engine categories

1515

16−hostname = None # domain name or None - if you want to rewrite the default HTTP host

16+base_url = None # "https://your.domain.tld/" or None (to use request parameters)

searx/static/js/searx.jsnew file+27

@@ -0,0 +1,27 @@

1+(function (w, d) {

2+ 'use strict';

3+ function addListener(el, type, fn) {

4+ if (el.addEventListener) {

5+ el.addEventListener(type, fn, false);

6+ } else {

7+ el.attachEvent('on' + type, fn);

8+ }

9+ }

10+

11+ function placeCursorAtEnd() {

12+ if (this.setSelectionRange) {

13+ var len = this.value.length * 2;

14+ this.setSelectionRange(len, len);

15+ }

16+ }

17+

18+ addListener(w, 'load', function () {

19+ var qinput = d.getElementById('q');

20+ if (qinput !== null) {

21+ addListener(qinput, 'focus', placeCursorAtEnd);

22+ qinput.focus();

23+ }

24+ });

25+

26+})(window, document);

27+

searx/templates/base.html+1

@@ -18,6 +18,7 @@

1818<div id="container">

1919{% block content %}

2020{% endblock %}

21+<script src="/static/js/searx.js" ></script>

2122</div>

2223</body>

2324</html>

searx/templates/categories.htmlnew file+5

@@ -0,0 +1,5 @@

1+{% for category in categories %}

2+ <div class="checkbox_container">

3+ <input type="checkbox" id="checkbox_{{ category|replace(' ', '_') }}" name="category_{{ category }}" {% if category in selected_categories %}checked="checked"{% endif %} /><label for="checkbox_{{ category|replace(' ', '_') }}" class="cb"></label><label for="checkbox_{{ category|replace(' ', '_') }}">{{ category }}</label>

4+ </div>

5+{% endfor %}

searx/templates/index.html+1

@@ -6,6 +6,7 @@

66 {% include 'search.html' %}

77 <p class="top_margin">

88 <a href="/about" class="hmarg">about</a>

9+ <a href="/preferences" class="hmarg">preferences</a>

910 </p>

1011</div>

1112{% endblock %}

searx/templates/preferences.htmlnew file+19

@@ -0,0 +1,19 @@

1+{% extends "base.html" %}

2+{% block head %} {% endblock %}

3+{% block content %}

4+<div class="row">

5+ <h2>Preferences</h2>

8+ <fieldset>

9+ <legend>Default categories</legend>

10+ <form method="post" action="/preferences" id="search_form">

11+ <p>

12+ {% include 'categories.html' %}

13+ </p>

14+ <input type="submit" value="save" />

15+ </form>

16+ </fieldset>

17+ <div class="right"><a href="/">back</a></div>

18+</div>

19+{% endblock %}

searx/templates/results.html+4−3

@@ -1,12 +1,13 @@

11{% extends "base.html" %}

22{% block title %}{{ q }} - {% endblock %}

33{% block content %}

4+<div class="right"><a href="/preferences">preferences</a></div>

45<div class="small">

56 {% include 'search.html' %}

67</div>

78<div id="results">

89 {% if suggestions %}

9− <div id="suggestions">Suggestions: {% for suggestion in suggestions %}<form method="post" action=""><input type="hidden" name="q" value="{{suggestion}}"><input type="submit" value="{{ suggestion }}" /></form>{% endfor %}</div>

10+ <div id="suggestions">Suggestions: {% for suggestion in suggestions %}<form method="post" action="/"><input type="hidden" name="q" value="{{suggestion}}"><input type="submit" value="{{ suggestion }}" /></form>{% endfor %}</div>

1011 {% endif %}

1112 <div>

1213 Number of results: {{ number_of_results }}

@@ -18,14 +19,14 @@

1819 {% include 'result_templates/default.html' %}

1920 {% endif %}

2021 {% endfor %}

21− <form method="post" action="">

22+ <form method="post" action="/">

2223 <div class="left">

2324 <input type="hidden" name="q" value="{{ q }}" />

2425 <input type="hidden" name="format" value="csv" />

2526 <input type="submit" value="download results in csv" />

2627 </div>

2728 </form>

28− <form method="post" action="">

29+ <form method="post" action="/">

2930 <div class="">

3031 <input type="hidden" name="q" value="{{ q }}" />

3132 <input type="hidden" name="format" value="json" />

searx/templates/search.html+1−5

@@ -4,10 +4,6 @@

44 <input type="submit" value="" id="search_submit" />

55</div>

66 <div>

7− {% for category in categories %}

8− <div class="checkbox_container">

9− <input type="checkbox" id="checkbox_{{ category|replace(' ', '_') }}" name="category_{{ category }}" {% if category in selected_categories %}checked="checked"{% endif %} /><label for="checkbox_{{ category|replace(' ', '_') }}" class="cb"></label><label for="checkbox_{{ category|replace(' ', '_') }}">{{ category }}</label>

10− </div>

11− {% endfor %}

7+ {% include 'categories.html' %}

128 </div>

139</form>

searx/webapp.py+31−13

@@ -22,7 +22,7 @@ if __name__ == "__main__":

2222 from sys import path

2323 path.append(os.path.realpath(os.path.dirname(os.path.realpath(__file__))+'/../'))

2424

25−from flask import Flask, request, render_template, url_for, Response, make_response

25+from flask import Flask, request, render_template, url_for, Response, make_response, redirect

2626from searx.engines import search, categories, engines, get_engines_stats

2727from searx import settings

2828import json

@@ -124,29 +124,46 @@ def index():

124124 response.headers.add('Content-Disposition', 'attachment;Filename=searx_-_{0}.csv'.format('_'.join(query.split())))

125125 return response

126126

127− template = render('results.html'

128− ,results=results

129− ,q=request_data['q']

130− ,selected_categories=selected_categories

131− ,number_of_results=len(results)

132− ,suggestions=suggestions

133− )

134− resp = make_response(template)

135− resp.set_cookie('categories', ','.join(selected_categories))

127+ return render('results.html'

128+ ,results=results

129+ ,q=request_data['q']

130+ ,selected_categories=selected_categories

131+ ,number_of_results=len(results)

132+ ,suggestions=suggestions

133+ )

136134

137− return resp

138135

139136@app.route('/about', methods=['GET'])

140137def about():

141138 global categories

142139 return render('about.html', categs=categories.items())

143140

141+

142+@app.route('/preferences', methods=['GET', 'POST'])

143+def preferences():

144+

145+ if request.method=='POST':

146+ selected_categories = []

147+ for pd_name,pd in request.form.items():

148+ if pd_name.startswith('category_'):

149+ category = pd_name[9:]

150+ if not category in categories:

151+ continue

152+ selected_categories.append(category)

153+ if selected_categories:

154+ resp = make_response(redirect('/'))

155+ resp.set_cookie('categories', ','.join(selected_categories))

156+ return resp

157+ return render('preferences.html')

158+

159+

144160@app.route('/stats', methods=['GET'])

145161def stats():

146162 global categories

147163 stats = get_engines_stats()

148164 return render('stats.html', stats=stats)

149165

166+

150167@app.route('/robots.txt', methods=['GET'])

151168def robots():

152169 return Response("""User-agent: *

@@ -155,6 +172,7 @@ Allow: /about

155172Disallow: /stats

156173""", mimetype='text/plain')

157174

175+

158176@app.route('/opensearch.xml', methods=['GET'])

159177def opensearch():

160178 global opensearch_xml

@@ -165,8 +183,8 @@ def opensearch():

165183 method = 'get'

166184 if request.is_secure:

167185 scheme = 'https'

168− if settings.hostname:

169− hostname = '{0}://{1}/'.format(scheme,settings.hostname)

186+ if settings.base_url:

187+ hostname = settings.base_url

170188 else:

171189 hostname = url_for('index', _external=True, _scheme=scheme)

172190 ret = opensearch_xml.format(method=method, host=hostname)