Merge pull request #183 from plone/fix/4205-munge-search-term

Use munge_search_term from plone.base, fix multi-word wildcard

Author: jensens

news/4205.bugfix

@@ -0,0 +1,3 @@
+Fix multi-word search so all word parts get wildcard prefix matching, not just the last one.
+Use ``munge_search_term`` from ``plone.base.utils``, deprecate old import path.
+@jensens

src/plone/app/querystring/querybuilder.py

@@ -4,70 +4,29 @@
 from plone.app.querystring.interfaces import IParsedQueryIndexModifier
 from plone.app.querystring.interfaces import IQueryModifier
 from plone.app.querystring.interfaces import IQuerystringRegistryReader
+from plone.base.utils import munge_search_term as _munge_search_term
 from plone.batching import Batch
 from plone.registry.interfaces import IRegistry
 from Products.CMFCore.utils import getToolByName
 from zope.component import getMultiAdapter
 from zope.component import getUtilitiesFor
 from zope.component import getUtility
+from zope.deferredimport import deprecated
 from zope.i18n import translate
 from zope.i18nmessageid import MessageFactory
 from zope.publisher.browser import BrowserView
 
 import json
 import logging
-import re
 
 logger = logging.getLogger("plone.app.querystring")
 _ = MessageFactory("plone")
 
-# We should accept both a simple space, unicode u'\u0020 but also a
-# multi-space, so called 'waji-kankaku', unicode u'\u3000'
-_MULTISPACE = "\u3000"
-_BAD_CHARS = ("?", "-", "+", "*", _MULTISPACE)
-
-
-def _quote_chars(s):
-    # We need to quote parentheses when searching text indices
-    if "(" in s:
-        s = s.replace("(", '"("')
-    if ")" in s:
-        s = s.replace(")", '")"')
-    if _MULTISPACE in s:
-        s = s.replace(_MULTISPACE, " ")
-    return s
-
-
-def _quote(term):
-    # The terms and, or and not must be wrapped in quotes to avoid
-    # being parsed as logical query atoms.
-    if term.lower() in ("and", "or", "not"):
-        term = '"%s"' % term
-    return _quote_chars(term)
-
-
-def munge_search_term(query):
-    original_query = query
-    for char in _BAD_CHARS:
-        query = query.replace(char, " ")
-
-    # extract quoted phrases first
-    quoted_phrases = re.findall(r'"([^"]*)"', query)
-    r = []
-    for qp in quoted_phrases:
-        # remove from original query
-        query = query.replace(f'"{qp}"', "")
-        # replace with cleaned leading/trailing whitespaces
-        # and skip empty phrases
-        clean_qp = qp.strip()
-        if not clean_qp:
-            continue
-        r.append(f'"{clean_qp}"')
-
-    r += map(_quote, query.strip().split())
-    r = " AND ".join(r)
-    r = r + ("*" if r and not original_query.endswith('"') else "")
-    return r
+deprecated(
+    "Moved to plone.base.utils. Import from there instead "
+    "(will be removed in Plone 7).",
+    munge_search_term="plone.base.utils:munge_search_term",
+)
 
 
 class ContentListingView(BrowserView):
@@ -267,7 +226,7 @@ def filter_query(self, query):
         return query
 
     def munge_search_term(self, q):
-        return munge_search_term(q)
+        return _munge_search_term(q)
 
 
 class RegistryConfiguration(BrowserView):

src/plone/app/querystring/tests/testQueryBuilder.py

@@ -333,80 +333,6 @@ def testQueryBuilderCustomQueryDoNotOverrideValues(self):
         self.assertEqual(len(results), 1)
         self.assertEqual(results[0].Title(), "Collectionstestpage 2")
 
-    def test_munge_search_term(self):
-        from plone.app.querystring.querybuilder import _BAD_CHARS
-        from plone.app.querystring.querybuilder import munge_search_term
-
-        search_term_tests = [
-            (
-                # search term
-                "spam ham",
-                "spam AND ham*",
-            ),
-            (
-                # quoted term
-                '"spam ham"',
-                '"spam ham"',
-            ),
-            (
-                # cleanup quoted terms
-                '" spam ham   "',
-                '"spam ham"',
-            ),
-            (
-                # quoted term with inner parenthesis
-                '"spam (ham)"',
-                '"spam (ham)"',
-            ),
-            (
-                # quoted term with inner parenthesis
-                '"spam" (ham)',
-                '"spam" AND "("ham")"*',
-            ),
-            (
-                # quoted term with inner parenthesis
-                '"(spam ham)"',
-                '"(spam ham)"',
-            ),
-            (
-                # mixed cases
-                "Spam hAm",
-                "Spam AND hAm*",
-            ),
-            (
-                # mix quoting and unquoted
-                'let\'s eat some "ham and eggs " without spam ',
-                '"ham and eggs" AND let\'s AND eat AND some ' "AND without AND spam*",
-            ),
-            (
-                'test "Welcome" to "Plone" retest',
-                '"Welcome" AND "Plone" AND test AND to AND retest*',
-            ),
-            (
-                # parentheses
-                "spam (ham)",
-                'spam AND "("ham")"*',
-            ),
-            (
-                # special keywords
-                "spam or not ham and eggs",
-                'spam AND "or" AND "not" AND ham AND "and" AND eggs*',
-            ),
-            (
-                # bad characters
-                " ".join(_BAD_CHARS),
-                "",
-            ),
-            (
-                # weird input
-                'test ""Welcome" to "Plone"" retest',
-                '"to" AND test AND WelcomePlone AND retest*',
-            ),
-        ]
-
-        for _in, _out in search_term_tests:
-            self.assertEqual(munge_search_term(_in), _out)
-
     def test_query_builder_unknown_sort(self):
         results = self.querybuilder(query=self.query, sort_on="unknown")
         self.assertEqual(len(results), 1)