From: Michael Tremer <michael.tremer@ipfire.org>
Date: Tue, 4 Feb 2020 10:39:45 +0000 (+0000)
Subject: search: Use PostgreSQL's websearch_to_tsquery()
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=22e56c4a93383a45124bc42a4fe130079b784538;p=ipfire.org.git

search: Use PostgreSQL's websearch_to_tsquery()

This allows us to have a more native search like we know
it from larger search engines.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
---

diff --git a/src/backend/blog.py b/src/backend/blog.py
index 6d2e0ac3..b92b0276 100644
--- a/src/backend/blog.py
+++ b/src/backend/blog.py
@@ -11,7 +11,6 @@ import textile
 import unicodedata
 
 from . import misc
-from . import util
 from .decorators import *
 
 class Blog(misc.Object):
@@ -80,12 +79,10 @@ class Blog(misc.Object):
 			ORDER BY COALESCE(updated_at, created_at) DESC LIMIT %s", limit)
 
 	def search(self, query, limit=None):
-		query = util.parse_search_query(query)
-
 		return self._get_posts("SELECT blog.* FROM blog \
 			LEFT JOIN blog_search_index search_index ON blog.id = search_index.post_id \
-			WHERE search_index.document @@ to_tsquery('english', %s) \
-				ORDER BY ts_rank(search_index.document, to_tsquery('english', %s)) DESC \
+			WHERE search_index.document @@ websearch_to_tsquery('english', %s) \
+				ORDER BY ts_rank(search_index.document, websearch_to_tsquery('english', %s)) DESC \
 			LIMIT %s", query, query, limit)
 
 	def has_had_recent_activity(self, **kwargs):
diff --git a/src/backend/util.py b/src/backend/util.py
index a39abd9e..04bde507 100644
--- a/src/backend/util.py
+++ b/src/backend/util.py
@@ -10,25 +10,6 @@ import re
 import string
 import unicodedata
 
-def parse_search_query(query):
-	q = []
-	for word in query.split():
-		# Is this lexeme negated?
-		negated = word.startswith("!")
-
-		# Remove any special characters
-		word = re.sub(r"\W+", "", word, flags=re.UNICODE)
-		if not word:
-			continue
-
-		# Restore negation
-		if negated:
-			word = "!%s" % word
-
-		q.append(word)
-
-	return " & ".join(q)
-
 def format_size(s, max_unit=None):
 	units = ("B", "kB", "MB", "GB", "TB")
 
diff --git a/src/backend/wiki.py b/src/backend/wiki.py
index 02bb24ff..9bddc32b 100644
--- a/src/backend/wiki.py
+++ b/src/backend/wiki.py
@@ -124,12 +124,10 @@ class Wiki(misc.Object):
 		return ret
 
 	def search(self, query, account=None, limit=None):
-		query = util.parse_search_query(query)
-
 		res = self._get_pages("SELECT wiki.* FROM wiki_search_index search_index \
 			LEFT JOIN wiki ON search_index.wiki_id = wiki.id \
-			WHERE search_index.document @@ to_tsquery('english', %s) \
-				ORDER BY ts_rank(search_index.document, to_tsquery('english', %s)) DESC",
+			WHERE search_index.document @@ websearch_to_tsquery('english', %s) \
+				ORDER BY ts_rank(search_index.document, websearch_to_tsquery('english', %s)) DESC",
 			query, query)
 
 		pages = []