Add fallback for non-supported languages & search in path and text

realms/config/__init__.py

@@ -102,6 +102,7 @@ ELASTICSEARCH_URL = 'http://127.0.0.1:9200'
 
 SEARCH_TYPE = 'whoosh'
 WHOOSH_INDEX = '/tmp/whoosh'
+WHOOSH_LANGUAGE = 'en'
 
 # Get ReCaptcha Keys for your domain here:
 # https://www.google.com/recaptcha/admin#whyrecaptcha

realms/modules/search/models.py

@@ -7,7 +7,7 @@ def simple(app):
 
 
 def whoosh(app):
-    return WhooshSearch(app.config['WHOOSH_INDEX'])
+    return WhooshSearch(app.config['WHOOSH_INDEX'], app.config['WHOOSH_LANGUAGE'])
 
 
 def elasticsearch(app):
@@ -47,38 +47,46 @@ class SimpleSearch(BaseSearch):
 
 
 class WhooshSearch(BaseSearch):
-    def __init__(self, index_path):
+    def __init__(self, index_path, language):
         from whoosh import index as whoosh_index
         from whoosh.fields import Schema, TEXT, ID
         from whoosh import qparser
         from whoosh.highlight import UppercaseFormatter
-        from whoosh.analysis import LanguageAnalyzer
+        from whoosh.analysis import SimpleAnalyzer, LanguageAnalyzer
+        from whoosh.lang import has_stemmer, has_stopwords
         import os.path
 
-        self.schema = Schema(path=ID(unique=True, stored=True), body=TEXT(analyzer=LanguageAnalyzer("de")))
+        if not has_stemmer(language) or not has_stopwords(language):
+            print("Language '%s' not supported by Whoosh, falling back to default analyzer." % (language))
+            analyzer = SimpleAnalyzer()
+        else:
+            analyzer = LanguageAnalyzer(language)
+
+        self.schema = Schema(path=ID(unique=True, stored=True), body=TEXT(analyzer=analyzer))
         self.formatter = UppercaseFormatter()
 
+        self.index_path = index_path
         if os.path.exists(index_path):
             self.search_index = whoosh_index.open_dir(index_path)
         else:
             os.mkdir(index_path)
             self.search_index = whoosh_index.create_in(index_path, self.schema)
 
-        self.query_parser = qparser.QueryParser("body", schema=self.schema)
+        self.query_parser = qparser.MultifieldParser(["body", "path"], schema=self.schema)
         self.query_parser.add_plugin(qparser.FuzzyTermPlugin())
 
     def index(self, index, doc_type, id_=None, body=None):
         writer = self.search_index.writer()
-        writer.update_document(path=id_.decode("utf-8"), body=body["content"])
+        writer.update_document(path=id_.decode("utf-8"), body=body["content"].decode("utf-8"))
         writer.commit()
 
     def index_wiki(self, name, body):
         self.index('wiki', 'page', id_=name, body=body)
 
     def delete_index(self, index):
-        writer = self.search_index.writer()
-        writer.delete_by_term('path', index)
-        writer.commit()
+        from whoosh import index as whoosh_index
+        self.search_index.close()
+        self.search_index = whoosh_index.create_in(self.index_path, schema=self.schema)
 
     def wiki(self, query):
         if not query: