Include raaga and taala aliases during solr indexing

related to #145 - change field attribute to multivalued - modify indexing of types with aliases ie taalas, raagas, taals, etc
MTG · Jul 24, 2014 · be4b5d1 · be4b5d1
1 parent 96ab581
commit be4b5d1
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 26 deletions.
diff --git a/carnatic/search.py b/carnatic/search.py
@@ -27,18 +27,11 @@ def search(name):
     query = "module_s:carnatic AND doctype_s:search AND title_t:(%s)" % name
     results = solr.search(query, rows=100)
     ret = collections.defaultdict(list)
-    klass_map = {"instrument": carnatic.models.Instrument,
-                 "raaga": carnatic.models.Raaga,
-                 "taala": carnatic.models.Taala,
-                 "concert": carnatic.models.Concert,
-                 "artist": carnatic.models.Artist,
-                 "work": carnatic.models.Work,
-                 "composer": carnatic.models.Composer}
     for d in results.docs:
         type = d["type_s"]
         id = d["object_id_i"]
         id = int(id)
-        klass = klass_map.get(type)
+        klass = get_klassmap().get(type)
         if klass:
             instance = klass.objects.get(pk=id)
             ret[type].append(instance)
@@ -49,14 +42,22 @@ def autocomplete(term):
     params['wt'] = 'json'
     params['q'] = term
     params['fq'] = "module_s:carnatic"
+    params['fl'] = "title_txt,type_s,object_id_i"
     path = 'suggest/?%s' % pysolr.safe_urlencode(params, True)
     response = solr._send_request('get', path)
     res = json.loads(response)
     check = res.get("response", {})
     docs = check.get("docs", [])
     ret = []
     for d in docs:
-        ret.append(d["title_t"])
+        type = d["type_s"]
+        id = int(d["object_id_i"])
+        if type in ["raaga", "taala"]:
+            cls = get_klassmap().get(type)
+            obj = cls.objects.get(pk=id)
+            ret.append(obj.name)
+        else :
+            ret.append(d["title_txt"][0])
     return ret[:5]
 
 def get_similar_concerts(works, raagas, taalas, artists):
@@ -111,3 +112,11 @@ def similar_recordings(mbid):
     else:
         return []
 
+def get_klassmap():
+    return {"instrument": carnatic.models.Instrument,
+                 "raaga": carnatic.models.Raaga,
+                 "taala": carnatic.models.Taala,
+                 "concert": carnatic.models.Concert,
+                 "artist": carnatic.models.Artist,
+                 "work": carnatic.models.Work,
+                 "composer": carnatic.models.Composer}
diff --git a/data/management/commands/solrimport.py b/data/management/commands/solrimport.py
@@ -27,13 +27,20 @@ class Command(BaseCommand):
     solr = pysolr.Solr(settings.SOLR_URL)
 
     def make_search_data(self, module, data, etype, namefield):
-        insert = [{"id": "%s_%s" % (etype, i.pk),
-                   "object_id_i": i.pk,
-                   "type_s": etype,
-                   "module_s": module,
-                   "title_t": getattr(i, namefield),
-                   "doctype_s": "search"
-                  } for i in data]
+        insert = []
+        for i in data:
+            doc = {     "id": "%s_%s" % (etype, i.pk),
+                        "object_id_i": i.pk,
+                        "type_s": etype,
+                        "module_s": module,
+                        "title_txt": getattr(i, namefield),
+                        "doctype_s": "search"
+                    }
+            if etype in ["raaga", "taala", "laya", "raag", "taal"]:
+                aliases = [alias.name for alias in i.aliases.all()]
+                aliases.append(doc.get("title_txt"))
+                doc.update({"title_txt": aliases})
+            insert.append(doc)
         return insert
 
     def create_carnatic_search_index(self):
@@ -46,14 +53,14 @@ def create_carnatic_search_index(self):
         concerts = carnatic.models.Concert.objects.all()
         raagas = carnatic.models.Raaga.objects.all()
         taalas = carnatic.models.Taala.objects.all()
-
+        
         insertinstr = self.make_search_data("carnatic", instruments, "instrument", "name")
         insertartist = self.make_search_data("carnatic", artists, "artist", "name")
         insertcomposer = self.make_search_data("carnatic", composers, "composer", "name")
         insertwork = self.make_search_data("carnatic", works, "work", "title")
         insertconcert = self.make_search_data("carnatic", concerts, "concert", "title")
-        insertraaga = self.make_search_data("carnatic", raagas, "raaga", "name")
-        inserttaala = self.make_search_data("carnatic", taalas, "taala", "name")
+        insertraaga = self.make_search_data("carnatic", raagas, "raaga", "common_name")
+        inserttaala = self.make_search_data("carnatic", taalas, "taala", "common_name")
 
         self.solr.add(insertinstr)
         self.solr.add(insertartist)

diff --git a/solr/collection1/conf/schema.xml b/solr/collection1/conf/schema.xml
@@ -15,13 +15,13 @@
 
    <dynamicField name="random_*" type="random" />
 
-   <field name="title_suggest" type="text_suggest" indexed="true" stored="false"/>
-   <field name="title_suggest_edge" type="text_suggest_edge" indexed="true" stored="false"/>
-   <field name="title_suggest_ngram" type="text_suggest_ngram" indexed="true" stored="false"/>
+   <field name="title_suggest" type="text_suggest" indexed="true" stored="false" multiValued="true"/>
+   <field name="title_suggest_edge" type="text_suggest_edge" indexed="true" stored="false" multiValued="true"/>
+   <field name="title_suggest_ngram" type="text_suggest_ngram" indexed="true" stored="false" multiValued="true"/>
 
-   <copyField source="title_t" dest="title_suggest" />
-   <copyField source="title_t" dest="title_suggest_edge" />
-   <copyField source="title_t" dest="title_suggest_ngram" />
+   <copyField source="title_txt" dest="title_suggest" />
+   <copyField source="title_txt" dest="title_suggest_edge" />
+   <copyField source="title_txt" dest="title_suggest_ngram" />
  </fields>
  <uniqueKey>id</uniqueKey>
 

diff --git a/solr/collection1/conf/solrconfig.xml b/solr/collection1/conf/solrconfig.xml
@@ -924,7 +924,7 @@
     <str name="echoParams">explicit</str>
     <str name="defType">edismax</str>
     <str name="rows">10</str>
-    <str name="fl">title_t,score</str>
+    <str name="fl">title_txt,score</str>
     <str name="qf">title_suggest^30 title_suggest_ngram^50.0 title_suggest_edge^20</str>
     <str name="pf">title_suggest_edge</str>
     <str name="sort">score desc</str>