fix: skip fuzzy search blend when raw query isn't tantivy-parseable

The fuzzy blend clause in parse_user_query() fed the raw, whoosh-syntax
query string directly to tantivy's own query parser. Since the
whoosh-compat migration, raw_query still contains whoosh grammar (date
keywords, whoosh-style ranges, bracket-class wildcards) that tantivy's
parser rejects with ValueError, which escaped parse_user_query and
turned into a generic HTTP 400 for the entire query whenever
ADVANCED_FUZZY_SEARCH_THRESHOLD was configured.

Deriving a clean plain-text-only extraction for the fuzzy clause was
ruled out: wc.parse() already expands unfielded terms into per-default-
field copies in the AST, so there's no "still unfielded" marker left to
walk without duplicating whoosh-compat's own expansion logic. Instead,
scope a narrow try/except ValueError around exactly the
index.parse_query() call and skip the fuzzy clause (logged at debug)
when it can't parse, leaving the exact/CJK clauses unaffected.

src/documents/search/_query.py

@@ -111,6 +111,40 @@ def _build_cjk_query(
         return None
 
 
+def _try_parse_fuzzy_query(
+    index: tantivy.Index,
+    raw_query: str,
+) -> tantivy.Query | None:
+    """Build the fuzzy blend clause from ``raw_query``, or None if it can't.
+
+    The fuzzy blend hands ``raw_query`` directly to tantivy's own query
+    parser (there's no clean AST-level fuzzy equivalent to whoosh-compat's
+    parse tree, and fuzzy matching was always an approximate, secondary,
+    0.1-boosted clause). But raw_query is whoosh grammar, not tantivy
+    grammar: it can contain date keywords (``today``), whoosh ranges
+    (``[2005 to 2009]``), or bracket-class wildcards (``202[0-1]*``) that
+    tantivy's parser rejects with a ValueError. Rather than let that escape
+    parse_user_query and fail the query's EXACT clause too (see paperless-
+    ngx's whoosh-compat migration regression), degrade gracefully: skip the
+    fuzzy clause and keep the exact/CJK clauses. Only ValueError is caught
+    — a broad except here would also hide real bugs.
+    """
+    try:
+        return index.parse_query(
+            raw_query,
+            DEFAULT_SEARCH_FIELDS,
+            field_boosts=_FIELD_BOOSTS,
+            fuzzy_fields={f: (True, 1, True) for f in DEFAULT_SEARCH_FIELDS},
+        )
+    except ValueError:
+        logger.debug(
+            "Skipping fuzzy search clause: raw query is not valid tantivy "
+            "query syntax: %r",
+            raw_query,
+        )
+        return None
+
+
 def build_permission_filter(
     schema: tantivy.Schema,
     user: AbstractBaseUser,
@@ -233,9 +267,13 @@ def parse_user_query(
     4. Optional fuzzy blend (ADVANCED_FUZZY_SEARCH_THRESHOLD) re-parses
        raw_query directly via index.parse_query — there's no clean AST-level
        fuzzy equivalent, and fuzzy matching was always an approximate,
-       secondary clause.
+       secondary clause. raw_query still carries whoosh grammar (date
+       keywords, bracket-class wildcards, etc.) that tantivy's own parser
+       cannot parse; when that happens the fuzzy clause is skipped rather
+       than letting the ValueError escape and fail the whole query (see
+       _try_parse_fuzzy_query).
     5. Optional CJK bigram clause — unchanged from before this migration,
-       never went through the old translate_query() either.
+       never went through the pre-whoosh-compat translation layer either.
     """
     registry = get_field_registry(settings.SEARCH_LANGUAGE)
     result = wc.parse(
@@ -265,13 +303,11 @@ def parse_user_query(
 
     threshold = settings.ADVANCED_FUZZY_SEARCH_THRESHOLD
     if threshold is not None:
-        fuzzy = index.parse_query(
-            raw_query,
-            DEFAULT_SEARCH_FIELDS,
-            field_boosts=_FIELD_BOOSTS,
-            fuzzy_fields={f: (True, 1, True) for f in DEFAULT_SEARCH_FIELDS},
-        )
-        clauses.append((tantivy.Occur.Should, tantivy.Query.boost_query(fuzzy, 0.1)))
+        fuzzy = _try_parse_fuzzy_query(index, raw_query)
+        if fuzzy is not None:
+            clauses.append(
+                (tantivy.Occur.Should, tantivy.Query.boost_query(fuzzy, 0.1)),
+            )
 
     if cjk_query is not None:
         clauses.append((tantivy.Occur.Should, cjk_query))

src/documents/tests/search/test_query.py

@@ -36,13 +36,28 @@ class TestParseUserQuery:
     def test_returns_tantivy_query(self, query_index: tantivy.Index) -> None:
         assert isinstance(parse_user_query(query_index, "invoice", UTC), tantivy.Query)
 
+    @pytest.mark.parametrize(
+        "raw_query",
+        [
+            pytest.param("invoice", id="plain_text"),
+            pytest.param("created:today", id="date_keyword"),
+            pytest.param("created:[2005 to 2009]", id="whoosh_date_range"),
+            pytest.param('added:"previous month"', id="quoted_date_phrase"),
+            pytest.param("title:202[0-1]*", id="bracket_class_wildcard"),
+        ],
+    )
     def test_fuzzy_mode_does_not_raise(
         self,
         query_index: tantivy.Index,
         settings,
+        raw_query: str,
     ) -> None:
+        # These are all valid whoosh grammar that tantivy's own query parser
+        # (used only by the fuzzy blend clause) cannot parse; the fuzzy
+        # clause must degrade gracefully instead of raising and failing the
+        # whole query. See _try_parse_fuzzy_query.
         settings.ADVANCED_FUZZY_SEARCH_THRESHOLD = 0.5
-        assert isinstance(parse_user_query(query_index, "invoice", UTC), tantivy.Query)
+        assert isinstance(parse_user_query(query_index, raw_query, UTC), tantivy.Query)
 
     def test_date_rewriting_applied_before_tantivy_parse(
         self,