WIP search

cubicweb_elasticsearch/search_helpers.py

@@ -51,6 +51,7 @@ def compose_search(search, query=None, fields=(),
                                 must=Q('multi_match',
                                        query=query.split(char)[1],
                                        type="phrase",
+                                       slope=50,
                                        fields=fields))
     must = []
     must_not = []
@@ -61,6 +62,8 @@ def compose_search(search, query=None, fields=(),
     # proximity booster - phrase with slop=50
 
     # TODO find a better way to do this
+    print("phrase", phrase)
+    print("common", common)
     if not phrase and not common:  # invalid combination
         phrase = common = True
     if phrase:
@@ -68,11 +71,20 @@ def compose_search(search, query=None, fields=(),
                          query=query,
                          type="phrase",
                          slop=50,
-                         fields=fields)
+                         fields=fields,
+                         boost=10)
         should.append(phrase_query)
+    if fuzzy:
+        fuzzy_query = Q('multi_match',
+                        query=query,
+                        type="most_fields",
+                        fields=fields,
+                        fuzziness="auto:4,6")
+        should.append(fuzzy_query)
+
     # highfrequency/lowfrequency query
     # https://www.elastic.co/blog/stop-stopping-stop-words-a-look-at-common-terms-query
-    if common:
+    if False:
         common_query = dsl_query.Common(alltext={"query": query,
                                                  "cutoff_frequency": cutoff_frequency,
                                                  "low_freq_operator": "and",
@@ -83,21 +95,23 @@ def compose_search(search, query=None, fields=(),
 
     elements = query.split()
     elements_lowercase = [e.lower() for e in elements]
+    print("elements_lowercase", elements_lowercase)
     if 'or' in elements_lowercase and len(elements) >= 3:
         for element in query.split('or'):
             should.append(Q('multi_match',
                             query=element.strip(),
                             fields=fields))
         elements = []
+    # CAN NOT WORK
     for element in elements:
         if element.startswith('-'):
             must_not.append(Q('multi_match',
                               query=element[1:],
                               fields=fields))
 
-        else:
-            if fuzzy:
-                should.append(dsl_query.Fuzzy(alltext=element))
+        # else:
+        #     if fuzzy:
+        #         should.append(Q('fuzzy', alltext={"value": element, "fuzziness": "auto:4,5"}))
     bool_query = dsl_query.Bool(must=must,
                                 must_not=must_not,
                                 should=should,

cubicweb_elasticsearch/views.py

@@ -62,7 +62,7 @@ class CWFacetedSearch(FacetedSearch):
                  doc_types=None,
                  index=None,
                  form=None,
-                 track_total_hits=True,
+                 track_total_hits=True, # 10000,
                  **kwargs):
         if index:
             self.index = index
@@ -100,7 +100,7 @@ class CWFacetedSearch(FacetedSearch):
         Add custom highlighting
         """
         search = search.highlight(*(f if '^' not in f else f.split('^', 1)[0] for f in self.fields))
-        return search.highlight_options(fragment_size=150, encoder='html')
+        return search.highlight_options(fragment_size=450, encoder='html')
 
 
 class ElasticSearchView(StartupView):
@@ -138,10 +138,11 @@ class ElasticSearchView(StartupView):
         search = self.customize_search(query_string,
                                        facet_selections,
                                        start, stop)
+        import json
+        print(json.dumps(search._s.to_dict(), indent=2))
         if 'debug-es' in self._cw.form:
-            import json
             self.w(u'<pre>')
-            self.w(unicode(json.dumps(search._s.to_dict(), indent=2)))
+            self.w(json.dumps(search._s.to_dict(), indent=2))
             self.w(u'</pre>')
             self.w(u'<br/>')
         try: