geminispace.info

commit dc4a32429ccd4d547319e6aacb54e500dd04ff37
parent dd46f0e29bf49df3637eccb28477d409c6e698d7
Author: Natalie Pendragon <natpen@natpen.net>
Date:   Wed,  3 Jun 2020 10:47:33 -0400

[serve] Use "OR" as the default connector for queries

Diffstat:
M
gus/serve.py
 | 
16
+++++++++++++---

1 file changed, 13 insertions(+), 3 deletions(-)
diff --git a/gus/serve.py b/gus/serve.py
@@ -13,9 +13,8 @@ from urllib.parse import quote
 
 import jetforce
 from jetforce import Response, Status
-from whoosh import highlight
+from whoosh import highlight, qparser
 from whoosh.index import open_dir
-from whoosh.qparser import MultifieldParser
 
 from gus.crawl import run_crawl
 from gus.lib.index_statistics import compute_index_statistics, load_last_statistics_from_file
@@ -245,7 +244,7 @@ def index(request):
 
 
 def _search_index(query, requested_page):
-    query = MultifieldParser(["content", "url", "prompt"], ix.schema).parse(query)
+    query = query_parser.parse(query)
     results = searcher.search_page(query, requested_page, pagelen=10)
     return (
         len(results),
@@ -425,10 +424,21 @@ def main():
         hostname=args.hostname,
         app=app,
     )
+
     global ix
     ix = open_dir(INDEX_DIR)
+
+    global query_parser
+    or_group = qparser.OrGroup.factory(0.9)
+    query_parser = qparser.MultifieldParser(["content", "url", "prompt"], ix.schema, group=or_group)
+    query_parser.add_plugin(qparser.RegexPlugin())
+    query_parser.remove_plugin_class(qparser.WildcardPlugin)
+    query_parser.remove_plugin_class(qparser.BoostPlugin)
+    query_parser.remove_plugin_class(qparser.RangePlugin)
+
     global index_modification_time
     index_modification_time = _get_index_modification_time()
+
     global searcher
     with ix.searcher() as searcher:
         asyncio.run(server.run())