feat(search): implement fuzzy matching on history title and URL

commit: 6990cc0a84ffe5a1246958b44346fff10337930b
author: brice · 10mo ago · 2025-12-16 18:24 UTC
parent: 83d87fb
signature: unsigned

1 file · +47 -3

M internal/application/usecase/search_history.go +47 -3


@@ -8,6 +8,7 @@ import (
8	8		"github.com/bnema/dumber/internal/domain/entity"
9	9		"github.com/bnema/dumber/internal/domain/repository"
10	10		"github.com/bnema/dumber/internal/logging"
	11	+	"github.com/sahilm/fuzzy"
11	12		)
12	13
13	14		// SearchHistoryUseCase handles history search and retrieval operations.

@@ -34,7 +35,10 @@ type SearchOutput struct {
34	35		}
35	36
36	37		// Search performs a fuzzy search on history entries.
	38	+	// Uses in-memory fuzzy matching on title + URL for token-based matching in any order.
37	39		func (uc SearchHistoryUseCase) Search(ctx context.Context, input SearchInput) (SearchOutput, error) {
	40	+	log := logging.FromContext(ctx)
	41	+
38	42		if input.Query == "" {
39	43		return &SearchOutput{Matches: []entity.HistoryMatch{}}, nil
40	44		}

@@ -44,12 +48,52 @@ func (uc *SearchHistoryUseCase) Search(ctx context.Context, input SearchInput) (
44	48		limit = 20 // Default limit
45	49		}
46	50
47		-	matches, err := uc.historyRepo.Search(ctx, input.Query, limit)
	51	+	// Fetch more candidates for fuzzy filtering (10x the limit)
	52	+	candidateLimit := limit * 10
	53	+	if candidateLimit < 100 {
	54	+	candidateLimit = 100
	55	+	}
	56	+
	57	+	candidates, err := uc.historyRepo.GetRecent(ctx, candidateLimit, 0)
48	58		if err != nil {
49		-	return nil, fmt.Errorf("failed to search history: %w", err)
	59	+	return nil, fmt.Errorf("failed to get history candidates: %w", err)
	60	+	}
	61	+
	62	+	if len(candidates) == 0 {
	63	+	return &SearchOutput{Matches: []entity.HistoryMatch{}}, nil
	64	+	}
	65	+
	66	+	// Build searchable strings: "title url" for each entry
	67	+	searchStrings := make([]string, len(candidates))
	68	+	for i, e := range candidates {
	69	+	searchStrings[i] = e.Title + " " + e.URL
	70	+	}
	71	+
	72	+	// Fuzzy match
	73	+	fuzzyMatches := fuzzy.Find(input.Query, searchStrings)
	74	+
	75	+	log.Debug().
	76	+	Str("query", input.Query).
	77	+	Int("candidates", len(candidates)).
	78	+	Int("matches", len(fuzzyMatches)).
	79	+	Msg("fuzzy search completed")
	80	+
	81	+	// Convert to HistoryMatch with scores, limited to requested limit
	82	+	resultLimit := limit
	83	+	if len(fuzzyMatches) < resultLimit {
	84	+	resultLimit = len(fuzzyMatches)
	85	+	}
	86	+
	87	+	results := make([]entity.HistoryMatch, 0, resultLimit)
	88	+	for i := 0; i < resultLimit; i++ {
	89	+	m := fuzzyMatches[i]
	90	+	results = append(results, entity.HistoryMatch{
	91	+	Entry: candidates[m.Index],
	92	+	Score: float64(m.Score),
	93	+	})
50	94		}
51	95
52		-	return &SearchOutput{Matches: matches}, nil
	96	+	return &SearchOutput{Matches: results}, nil
53	97		}
54	98
55	99		// GetRecent retrieves recent history entries with pagination.


@@ -8,6 +8,7 @@ import (
8		"github.com/bnema/dumber/internal/domain/entity"	8		"github.com/bnema/dumber/internal/domain/entity"
9		"github.com/bnema/dumber/internal/domain/repository"	9		"github.com/bnema/dumber/internal/domain/repository"
10		"github.com/bnema/dumber/internal/logging"	10		"github.com/bnema/dumber/internal/logging"
			11	+	"github.com/sahilm/fuzzy"
11		)	12		)
12			13
13		// SearchHistoryUseCase handles history search and retrieval operations.	14		// SearchHistoryUseCase handles history search and retrieval operations.

@@ -34,7 +35,10 @@ type SearchOutput struct {
34		}	35		}
35			36
36		// Search performs a fuzzy search on history entries.	37		// Search performs a fuzzy search on history entries.
			38	+	// Uses in-memory fuzzy matching on title + URL for token-based matching in any order.
37		func (uc SearchHistoryUseCase) Search(ctx context.Context, input SearchInput) (SearchOutput, error) {	39		func (uc SearchHistoryUseCase) Search(ctx context.Context, input SearchInput) (SearchOutput, error) {
			40	+	log := logging.FromContext(ctx)
			41	+
38		if input.Query == "" {	42		if input.Query == "" {
39		return &SearchOutput{Matches: []entity.HistoryMatch{}}, nil	43		return &SearchOutput{Matches: []entity.HistoryMatch{}}, nil
40		}	44		}

@@ -44,12 +48,52 @@ func (uc *SearchHistoryUseCase) Search(ctx context.Context, input SearchInput) (
44		limit = 20 // Default limit	48		limit = 20 // Default limit
45		}	49		}
46			50
47	-	matches, err := uc.historyRepo.Search(ctx, input.Query, limit)	51	+	// Fetch more candidates for fuzzy filtering (10x the limit)
			52	+	candidateLimit := limit * 10
			53	+	if candidateLimit < 100 {
			54	+	candidateLimit = 100
			55	+	}
			56	+
			57	+	candidates, err := uc.historyRepo.GetRecent(ctx, candidateLimit, 0)
48		if err != nil {	58		if err != nil {
49	-	return nil, fmt.Errorf("failed to search history: %w", err)	59	+	return nil, fmt.Errorf("failed to get history candidates: %w", err)
			60	+	}
			61	+
			62	+	if len(candidates) == 0 {
			63	+	return &SearchOutput{Matches: []entity.HistoryMatch{}}, nil
			64	+	}
			65	+
			66	+	// Build searchable strings: "title url" for each entry
			67	+	searchStrings := make([]string, len(candidates))
			68	+	for i, e := range candidates {
			69	+	searchStrings[i] = e.Title + " " + e.URL
			70	+	}
			71	+
			72	+	// Fuzzy match
			73	+	fuzzyMatches := fuzzy.Find(input.Query, searchStrings)
			74	+
			75	+	log.Debug().
			76	+	Str("query", input.Query).
			77	+	Int("candidates", len(candidates)).
			78	+	Int("matches", len(fuzzyMatches)).
			79	+	Msg("fuzzy search completed")
			80	+
			81	+	// Convert to HistoryMatch with scores, limited to requested limit
			82	+	resultLimit := limit
			83	+	if len(fuzzyMatches) < resultLimit {
			84	+	resultLimit = len(fuzzyMatches)
			85	+	}
			86	+
			87	+	results := make([]entity.HistoryMatch, 0, resultLimit)
			88	+	for i := 0; i < resultLimit; i++ {
			89	+	m := fuzzyMatches[i]
			90	+	results = append(results, entity.HistoryMatch{
			91	+	Entry: candidates[m.Index],
			92	+	Score: float64(m.Score),
			93	+	})
50		}	94		}
51			95
52	-	return &SearchOutput{Matches: matches}, nil	96	+	return &SearchOutput{Matches: results}, nil
53		}	97		}
54			98
55		// GetRecent retrieves recent history entries with pagination.	99		// GetRecent retrieves recent history entries with pagination.