searx

commit: 4e2dae30f013d8ef9fd6d22e3f01e38f13f13c11
parent 29a526ff0e33030fc9d1416a9ccfeba93e376664
Author: Thomas Pointhuber <thomas.pointhuber@gmx.at>
Date:   Sat, 10 Jan 2015 16:42:57 +0100

[enh] add autocompletion for searx-specific strings

Diffstat:
M
searx/autocomplete.py
79
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
searx/query.py
2
+-
M
searx/webapp.py
7
+++++--

3 files changed, 85 insertions(+), 3 deletions(-)
diff --git a/searx/autocomplete.py b/searx/autocomplete.py
@@ -20,6 +20,85 @@ from lxml import etree
 from requests import get
 from json import loads
 from urllib import urlencode
+from searx.languages import language_codes
+from searx.engines import (
+    categories, engines, engine_shortcuts
+)
+
+
+def searx_bang(full_query):
+    '''check if the searchQuery contain a bang, and create fitting autocompleter results'''
+    # check if there is a query which can be parsed
+    if len(full_query.getSearchQuery()) == 0:
+        return []
+
+    results = []
+
+    # check if current query stats with !bang
+    if full_query.getSearchQuery()[0] == '!':
+        if len(full_query.getSearchQuery()) == 1:
+            # show some example queries
+            # TODO, check if engine is not avaliable
+            results.append("!images")
+            results.append("!wikipedia")
+            results.append("!osm")
+        else:
+            engine_query = full_query.getSearchQuery()[1:]
+
+            # check if query starts with categorie name
+            for categorie in categories:
+                if categorie.startswith(engine_query):
+                    results.append('!{categorie}'.format(categorie=categorie))
+
+            # check if query starts with engine name
+            for engine in engines:
+                if engine.startswith(engine_query):
+                    results.append('!{engine}'.format(engine=engine.replace(' ', '_')))
+
+            # check if query starts with engine shortcut
+            for engine_shortcut in engine_shortcuts:
+                if engine_shortcut.startswith(engine_query):
+                    results.append('!{engine_shortcut}'.format(engine_shortcut=engine_shortcut))
+
+    # check if current query stats with :bang
+    elif full_query.getSearchQuery()[0] == ':':
+        if len(full_query.getSearchQuery()) == 1:
+            # show some example queries
+            results.append(":en")
+            results.append(":en_us")
+            results.append(":english")
+            results.append(":united_kingdom")
+        else:
+            engine_query = full_query.getSearchQuery()[1:]
+
+            for lc in language_codes:
+                lang_id, lang_name, country = map(str.lower, lc)
+
+                # check if query starts with language-id
+                if lang_id.startswith(engine_query):
+                    if len(engine_query) <= 2:
+                        results.append(':{lang_id}'.format(lang_id=lang_id.split('_')[0]))
+                    else:
+                        results.append(':{lang_id}'.format(lang_id=lang_id))
+
+                # check if query starts with language name
+                if lang_name.startswith(engine_query):
+                    results.append(':{lang_name}'.format(lang_name=lang_name))
+
+                # check if query starts with country
+                if country.startswith(engine_query):
+                    results.append(':{country}'.format(country=country.replace(' ', '_')))
+
+    # remove duplicates
+    result_set = set(results)
+
+    # remove results which are already contained in the query
+    for query_part in full_query.query_parts:
+        if query_part in result_set:
+            result_set.remove(query_part)
+
+    # convert result_set back to list
+    return list(result_set)
 
 
 def dbpedia(query):
diff --git a/searx/query.py b/searx/query.py
@@ -77,7 +77,7 @@ class Query(object):
                     if lang == lang_id\
                        or lang_id.startswith(lang)\
                        or lang == lang_name\
-                       or lang == country:
+                       or lang.replace('_', ' ') == country:
                         parse_next = True
                         self.languages.append(lang)
                         break
diff --git a/searx/webapp.py b/searx/webapp.py
@@ -46,7 +46,7 @@ from searx.languages import language_codes
 from searx.https_rewrite import https_url_rewrite
 from searx.search import Search
 from searx.query import Query
-from searx.autocomplete import backends as autocomplete_backends
+from searx.autocomplete import searx_bang, backends as autocomplete_backends
 from searx import logger
 
 
@@ -352,8 +352,11 @@ def autocompleter():
     if not completer:
         return
 
+    # parse searx specific autocompleter results like !bang
+    raw_results = searx_bang(query)
+
     # run autocompletion
-    raw_results = completer(query.getSearchQuery())
+    raw_results.extend(completer(query.getSearchQuery()))
 
     # parse results (write :language and !engine back to result string)
     results = []