fix: sanitize FTS5 special chars and interleave URL/title results

commit: ed73d491c3c87dc9cf969bd606f382a605000c4a
author: brice · 9mo ago · 2026-01-01 13:53 UTC
parent: 8ba0228
signature: GPG VerifiedCD88F9E564E0102646FB6077A1A07C7C7C3E349D · arch-gpg Verified for the committer email

1 file · +43 -24

M internal/infrastructure/persistence/sqlite/history_repo.go +43 -24


@@ -56,9 +56,10 @@ func (r *historyRepo) Search(ctx context.Context, query string, limit int) ([]en
56	56		}
57	57
58	58		// Build FTS5 query: "word1* word2*" (implicit AND between terms)
	59	+	// Sanitize each word to remove FTS5 special characters
59	60		parts := make([]string, len(words))
60	61		for i, word := range words {
61		-	parts[i] = word + "*"
	62	+	parts[i] = sanitizeFTS5Word(word) + "*"
62	63		}
63	64		ftsQuery := strings.Join(parts, " ")
64	65

@@ -81,40 +82,58 @@ func (r *historyRepo) Search(ctx context.Context, query string, limit int) ([]en
81	82		titleRows = []sqlc.History{}
82	83		}
83	84
84		-	// Merge results, deduplicating by ID
	85	+	// Merge results by interleaving URL and title matches for balanced results
85	86		seen := make(map[int64]bool)
86	87		var matches []entity.HistoryMatch
87	88
88		-	for i := range urlRows {
89		-	row := urlRows[i]
90		-	if !seen[row.ID] {
91		-	seen[row.ID] = true
92		-	matches = append(matches, entity.HistoryMatch{
93		-	Entry: historyFromRow(row),
94		-	Score: 1.0,
95		-	})
	89	+	i, j := 0, 0
	90	+	for len(matches) < limit && (i < len(urlRows) \|\| j < len(titleRows)) {
	91	+	// Take next URL match if available
	92	+	if i < len(urlRows) {
	93	+	row := urlRows[i]
	94	+	i++
	95	+	if !seen[row.ID] {
	96	+	seen[row.ID] = true
	97	+	matches = append(matches, entity.HistoryMatch{
	98	+	Entry: historyFromRow(row),
	99	+	Score: 1.0,
	100	+	})
	101	+	}
96	102		}
97		-	}
98	103
99		-	for i := range titleRows {
100		-	row := titleRows[i]
101		-	if !seen[row.ID] {
102		-	seen[row.ID] = true
103		-	matches = append(matches, entity.HistoryMatch{
104		-	Entry: historyFromRow(row),
105		-	Score: 1.0,
106		-	})
	104	+	// Take next title match if available
	105	+	if j < len(titleRows) && len(matches) < limit {
	106	+	row := titleRows[j]
	107	+	j++
	108	+	if !seen[row.ID] {
	109	+	seen[row.ID] = true
	110	+	matches = append(matches, entity.HistoryMatch{
	111	+	Entry: historyFromRow(row),
	112	+	Score: 1.0,
	113	+	})
	114	+	}
107	115		}
108	116		}
109	117
110		-	// Limit results
111		-	if len(matches) > limit {
112		-	matches = matches[:limit]
113		-	}
114		-
115	118		return matches, nil
116	119		}
117	120
	121	+	// sanitizeFTS5Word removes FTS5 special characters from a search word.
	122	+	// FTS5 special chars: " * ( ) : ^ - AND OR NOT NEAR
	123	+	func sanitizeFTS5Word(word string) string {
	124	+	// Remove characters that have special meaning in FTS5
	125	+	var result strings.Builder
	126	+	for _, r := range word {
	127	+	switch r {
	128	+	case '"', '*', '(', ')', ':', '^', '-':
	129	+	// Skip FTS5 special characters
	130	+	default:
	131	+	result.WriteRune(r)
	132	+	}
	133	+	}
	134	+	return result.String()
	135	+	}
	136	+
118	137		func (r historyRepo) GetRecent(ctx context.Context, limit, offset int) ([]entity.HistoryEntry, error) {
119	138		rows, err := r.queries.GetRecentHistory(ctx, sqlc.GetRecentHistoryParams{
120	139		Limit: int64(limit),

 	}
 	// Build FTS5 query: "word1* word2*" (implicit AND between terms)
+	// Sanitize each word to remove FTS5 special characters
 	parts := make([]string, len(words))
 	for i, word := range words {
-		parts[i] = word + "*"
+		parts[i] = sanitizeFTS5Word(word) + "*"
 	}
 	ftsQuery := strings.Join(parts, " ")
 		titleRows = []sqlc.History{}
 	}
-	// Merge results, deduplicating by ID
+	// Merge results by interleaving URL and title matches for balanced results
 	seen := make(map[int64]bool)
 	var matches []entity.HistoryMatch
-	for i := range urlRows {
+	i, j := 0, 0
-		row := urlRows[i]
+	for len(matches) < limit && (i < len(urlRows) || j < len(titleRows)) {
-		if !seen[row.ID] {
+		// Take next URL match if available
-			seen[row.ID] = true
+		if i < len(urlRows) {
-			matches = append(matches, entity.HistoryMatch{
+			row := urlRows[i]
-				Entry: historyFromRow(row),
+			i++
-				Score: 1.0,
+			if !seen[row.ID] {
-			})
+				seen[row.ID] = true
+				matches = append(matches, entity.HistoryMatch{
+					Entry: historyFromRow(row),
+					Score: 1.0,
+				})
+			}
 		}
-	}
-	for i := range titleRows {
+		// Take next title match if available
-		row := titleRows[i]
+		if j < len(titleRows) && len(matches) < limit {
-		if !seen[row.ID] {
+			row := titleRows[j]
-			seen[row.ID] = true
+			j++
-			matches = append(matches, entity.HistoryMatch{
+			if !seen[row.ID] {
-				Entry: historyFromRow(row),
+				seen[row.ID] = true
-				Score: 1.0,
+				matches = append(matches, entity.HistoryMatch{
-			})
+					Entry: historyFromRow(row),
+					Score: 1.0,
+				})
+			}
 		}
 	}
-	// Limit results
-	if len(matches) > limit {
-		matches = matches[:limit]
-	}
 	return matches, nil
 }
+// sanitizeFTS5Word removes FTS5 special characters from a search word.
+// FTS5 special chars: " * ( ) : ^ - AND OR NOT NEAR
+func sanitizeFTS5Word(word string) string {
+	// Remove characters that have special meaning in FTS5
+	var result strings.Builder
+	for _, r := range word {
+		switch r {
+		case '"', '*', '(', ')', ':', '^', '-':
+			// Skip FTS5 special characters
+		default:
+			result.WriteRune(r)
+		}
+	}
+	return result.String()
+}
 func (r *historyRepo) GetRecent(ctx context.Context, limit, offset int) ([]*entity.HistoryEntry, error) {
 	rows, err := r.queries.GetRecentHistory(ctx, sqlc.GetRecentHistoryParams{
 		Limit:  int64(limit),