fix(history): substring fuzzy search (#74)

closes #73 <img width="1434" height="302" alt="image" src="https://github.com/user-attachments/assets/87bb8431-40c7-4ee8-ad58-79c4fea6efa7" /> I used use prefix match imagine you have a long command ENOUGH that you used to use (like image), with fuzzy search + substring, it will always show suggestion about that command in the list but in the bottom and I see this very annoying but now I limit the score from fuzzy search, if the command is too long (which means fuzzy score is extremely weak), skip it right away <img width="1296" height="437" alt="image" src="https://github.com/user-attachments/assets/77d807a2-ddce-4c37-9f5b-a02cd44de325" /> it now has substring support
2026-07-30 21:11:34 +07:00
parent 0049256844
commit 37182f7212
3 changed files with 76 additions and 77 deletions
@@ -208,53 +208,34 @@ func SearchHistory(query string, aliases map[string]string) ([]HistResult, error
 	var results []HistResult
 	seenCmds := make(map[string]bool)

-	addMatches := func(q string, subcmdFilter bool) {
+	addMatches := func(q string) {
 		qLow := strings.ToLower(q)
-		queryFirstWord := ""
-		querySecondWord := ""
-		if strings.IndexByte(qLow, ' ') != -1 {
-			if fields := strings.Fields(qLow); len(fields) > 0 {
-				queryFirstWord = fields[0]
-				// find first non-flag token after the command as the subcommand
-				for _, f := range fields[1:] {
-					if !strings.HasPrefix(f, "-") {
-						querySecondWord = f
-						break
-					}
-				}
-			}
+
+		// extract pure substring matches (all words present) based strictly on recency order (historyCache is newest-first)
+		// this ensures that long commands with exact substrings are never truncated by the fuzzy searcher's limit
+		strictMatches := 0
+		words := strings.Fields(qLow)
+		if len(words) == 0 {
+			words = []string{qLow}
 		}
 		
-		// extract pure prefix matches based strictly on recency order (historyCache is newest-first)
-		strictMatches := 0
 		for _, cmd := range historyCache {
 			if seenCmds[cmd] {
 				continue
 			}
-			fields := strings.Fields(cmd)
-			firstWordLow := ""
-			if len(fields) > 0 {
-				firstWordLow = strings.ToLower(fields[0])
+			
+			cmdLow := strings.ToLower(cmd)
+			matchAll := true
+			for _, w := range words {
+				if !strings.Contains(cmdLow, w) {
+					matchAll = false
+					break
+				}
 			}
 			
-			if queryFirstWord != "" {
-				if firstWordLow != queryFirstWord {
+			if !matchAll {
 				continue
 			}
-				if subcmdFilter && querySecondWord != "" {
-					if len(fields) < 2 {
-						continue
-					}
-					secondWordLow := strings.ToLower(fields[1])
-					if !strings.HasPrefix(secondWordLow, querySecondWord) {
-						continue
-					}
-				}
-			} else {
-				if !strings.HasPrefix(firstWordLow, qLow) {
-					continue
-				}
-			}
 			
 			seenCmds[cmd] = true
 			results = append(results, HistResult{
@@ -274,33 +255,11 @@ func SearchHistory(query string, aliases map[string]string) ([]HistResult, error
 				continue
 			}
 			
-			// filter results by command name match
-			fields := strings.Fields(m.Str)
-			firstWord := m.Str
-			if len(fields) > 0 {
-				firstWord = fields[0]
-			}
-			firstWordLow := strings.ToLower(firstWord)
-
-			if queryFirstWord != "" {
-				if firstWordLow != queryFirstWord {
+			// filter out extremely weak fuzzy matches (e.g. random garbage typing that 
+			// loosely matches across a very long command)
+			if len(q) > 0 && m.Score/len(q) < 150 {
 				continue
 			}
-				// when query has a non-flag second token, filter by subcommand prefix
-				if subcmdFilter && querySecondWord != "" {
-					if len(fields) < 2 {
-						continue
-					}
-					secondWordLow := strings.ToLower(fields[1])
-					if !strings.HasPrefix(secondWordLow, querySecondWord) {
-						continue
-					}
-				}
-			} else {
-				if !strings.HasPrefix(firstWordLow, qLow) {
-					continue
-				}
-			}

 			seenCmds[m.Str] = true
 			results = append(results, HistResult{
@@ -311,18 +270,9 @@ func SearchHistory(query string, aliases map[string]string) ([]HistResult, error
 		}
 	}

-	addMatches(query, true)
+	addMatches(query)
 	for _, altQ := range alternativeQueries {
-		addMatches(altQ, true)
-	}
-
-	// fallback: if subcommand filter produced nothing, retry without it
-	// so typos like "git chckout" still surface fuzzy matches
-	if len(results) == 0 {
-		addMatches(query, false)
-		for _, altQ := range alternativeQueries {
-			addMatches(altQ, false)
-		}
+		addMatches(altQ)
 	}

 	getTier := func(cmd, q string) int {
@@ -335,7 +335,7 @@ func (f *FrecencyStore) QueryLocal(ctx context.Context, cwd, prefix string, limi
 	var rows *sql.Rows
 	var err error
 	if prefix != "" {
-		rows, err = f.db.QueryContext(ctxTimeout, `SELECT cmd, cwd, count, last_used FROM history_entries WHERE cwd = ? AND cmd LIKE ?`, cwd, prefix+"%")
+		rows, err = f.db.QueryContext(ctxTimeout, `SELECT cmd, cwd, count, last_used FROM history_entries WHERE cwd = ? AND cmd LIKE ?`, cwd, "%"+prefix+"%")
 	} else {
 		rows, err = f.db.QueryContext(ctxTimeout, `SELECT cmd, cwd, count, last_used FROM history_entries WHERE cwd = ?`, cwd)
 	}
@@ -397,7 +397,7 @@ func (f *FrecencyStore) QueryGlobal(ctx context.Context, prefix string, limit in
 	var rows *sql.Rows
 	var err error
 	if prefix != "" {
-		rows, err = f.db.QueryContext(ctxTimeout, `SELECT cmd, cwd, count, last_used FROM history_entries WHERE cmd LIKE ?`, prefix+"%")
+		rows, err = f.db.QueryContext(ctxTimeout, `SELECT cmd, cwd, count, last_used FROM history_entries WHERE cmd LIKE ?`, "%"+prefix+"%")
 	} else {
 		rows, err = f.db.QueryContext(ctxTimeout, `SELECT cmd, cwd, count, last_used FROM history_entries`)
 	}
@@ -145,6 +145,55 @@ func TestFrecencyStore_SQLiteConfigurationAndContext(t *testing.T) {
 	}
 }

+func TestFrecencyStore_SubstringMatch(t *testing.T) {
+	tmpDir := t.TempDir()
+	dbPath := filepath.Join(tmpDir, "history.db")
+	store, err := NewFrecencyStore(dbPath)
+	if err != nil {
+		t.Fatalf("NewFrecencyStore failed: %v", err)
+	}
+	defer store.Close()
+
+	cwd := "/home/user/project"
+	_ = store.Record(context.Background(), "sudo chown -R www-data:www-data /var/www/html", cwd, 0)
+	_ = store.Record(context.Background(), "sudo systemctl restart nginx", cwd, 0)
+	_ = store.Record(context.Background(), "chmod +x script.sh", cwd, 0)
+
+	_ = store.Record(context.Background(), "echo hello world", cwd, 0)
+
+	// Test QueryLocal with substring
+	entries, err := store.QueryLocal(context.Background(), cwd, "cho", 10)
+	if err != nil {
+		t.Fatalf("QueryLocal failed: %v", err)
+	}
+	if len(entries) != 2 { // should match 'sudo chown...' and 'echo...' (has 'cho')
+		t.Fatalf("expected 2 entries for 'cho', got %d", len(entries))
+	}
+
+	foundChown := false
+	for _, e := range entries {
+		if e.Cmd == "sudo chown -R www-data:www-data /var/www/html" {
+			foundChown = true
+			break
+		}
+	}
+	if !foundChown {
+		t.Errorf("expected 'sudo chown...' to be found with substring 'cho'")
+	}
+
+	// Test QueryGlobal with substring
+	entriesGlobal, err := store.QueryGlobal(context.Background(), "chown", 10)
+	if err != nil {
+		t.Fatalf("QueryGlobal failed: %v", err)
+	}
+	if len(entriesGlobal) != 1 {
+		t.Fatalf("expected 1 entry for 'chown', got %d", len(entriesGlobal))
+	}
+	if entriesGlobal[0].Cmd != "sudo chown -R www-data:www-data /var/www/html" {
+		t.Errorf("expected 'sudo chown...' to be found with substring 'chown', got %s", entriesGlobal[0].Cmd)
+	}
+}
+
 func TestFrecencyStore_NilReceiver(t *testing.T) {
 	var nilStore *FrecencyStore
 	if err := nilStore.Record(context.Background(), "cmd", "cwd", 0); err != nil {