feat: 기업 검색에 바이그램 퍼지 검색 폴백 추가 (closes #1)

'마켓컬리'처럼 브랜드명이 법인명과 다를 때 substring 검색에서
결과가 없으면 문자 바이그램 유사도(≥0.3) 기반 퍼지 검색으로
폴백하여 '컬리' 등 관련 기업을 찾아준다.

검색 우선순위: 종목코드 완전일치 → 법인코드 완전일치 →
이름 완전일치 → 부분문자열 → 바이그램 퍼지 (상위 10건)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: seapy

internal/cache/corpcode.go

@@ -10,6 +10,7 @@ import (
 	"net/http"
 	"net/url"
 	"os"
+	"sort"
 	"strings"
 	"time"
 
@@ -102,7 +103,7 @@ func Load(apiKey string) (*Store, bool, error) {
 }
 
 // Search finds corporations matching the query.
-// Returns exact match first; falls back to substring search.
+// Priority: exact stock/corp code → exact name → substring → bigram fuzzy.
 func (s *Store) Search(query string) []*CorpInfo {
 	// Try stock code (exact)
 	if c, ok := s.byStock[query]; ok {
@@ -124,7 +125,63 @@ func (s *Store) Search(query string) []*CorpInfo {
 			matches = append(matches, info)
 		}
 	}
-	return matches
+	if len(matches) > 0 {
+		return matches
+	}
+	// Fuzzy fallback: bigram similarity
+	return s.fuzzySearch(lower, 0.3, 10)
+}
+
+// fuzzySearch returns up to max corps whose name has bigram similarity ≥ threshold
+// with query, sorted by score descending.
+func (s *Store) fuzzySearch(query string, threshold float64, max int) []*CorpInfo {
+	type scored struct {
+		info  *CorpInfo
+		score float64
+	}
+	var results []scored
+	for _, info := range s.All {
+		score := bigramSim(query, strings.ToLower(info.CorpName))
+		if score >= threshold {
+			results = append(results, scored{info, score})
+		}
+	}
+	sort.Slice(results, func(i, j int) bool {
+		return results[i].score > results[j].score
+	})
+	if len(results) > max {
+		results = results[:max]
+	}
+	out := make([]*CorpInfo, len(results))
+	for i, r := range results {
+		out[i] = r.info
+	}
+	return out
+}
+
+// bigramSim returns the fraction of query's character bigrams that appear in target.
+// Uses rune-level bigrams for correct Korean handling.
+func bigramSim(query, target string) float64 {
+	qr := []rune(query)
+	tr := []rune(target)
+	if len(qr) < 2 {
+		return 0
+	}
+	// Build target bigram frequency map
+	tBig := make(map[[2]rune]int, len(tr))
+	for i := 0; i < len(tr)-1; i++ {
+		tBig[[2]rune{tr[i], tr[i+1]}]++
+	}
+	// Count how many query bigrams appear in target
+	matched := 0
+	for i := 0; i < len(qr)-1; i++ {
+		k := [2]rune{qr[i], qr[i+1]}
+		if tBig[k] > 0 {
+			matched++
+			tBig[k]--
+		}
+	}
+	return float64(matched) / float64(len(qr)-1)
 }
 
 // Status returns cache file info.

internal/cache/corpcode_test.go

@@ -0,0 +1,173 @@
+package cache
+
+import (
+	"testing"
+)
+
+// testStore returns a Store built from representative sample data.
+func testStore() *Store {
+	corps := []*CorpInfo{
+		{CorpCode: "00126380", CorpName: "삼성전자", StockCode: "005930"},
+		{CorpCode: "01153956", CorpName: "컬리"},
+		{CorpCode: "01494172", CorpName: "컬리넥스트마일"},
+		{CorpCode: "01547845", CorpName: "당근마켓"},
+		{CorpCode: "01717824", CorpName: "당근페이"},
+		{CorpCode: "01138364", CorpName: "더핑크퐁컴퍼니", StockCode: "403850"},
+		{CorpCode: "00293886", CorpName: "카카오"},
+		{CorpCode: "00000002", CorpName: "카카오뱅크"},
+		{CorpCode: "00000003", CorpName: "카카오페이"},
+		{CorpCode: "01154811", CorpName: "주식회사 오늘의집"},
+	}
+	return buildStore(corps)
+}
+
+func corpNames(corps []*CorpInfo) []string {
+	out := make([]string, len(corps))
+	for i, c := range corps {
+		out[i] = c.CorpName
+	}
+	return out
+}
+
+// --- exact match tests ---
+
+func TestSearch_ExactStockCode(t *testing.T) {
+	s := testStore()
+	results := s.Search("005930")
+	if len(results) != 1 || results[0].CorpName != "삼성전자" {
+		t.Fatalf("종목코드 완전일치 실패: got %v", corpNames(results))
+	}
+}
+
+func TestSearch_ExactCorpCode(t *testing.T) {
+	s := testStore()
+	results := s.Search("00126380")
+	if len(results) != 1 || results[0].CorpName != "삼성전자" {
+		t.Fatalf("법인코드 완전일치 실패: got %v", corpNames(results))
+	}
+}
+
+func TestSearch_ExactName(t *testing.T) {
+	s := testStore()
+	results := s.Search("삼성전자")
+	if len(results) != 1 || results[0].CorpName != "삼성전자" {
+		t.Fatalf("이름 완전일치 실패: got %v", corpNames(results))
+	}
+}
+
+// --- substring tests ---
+
+func TestSearch_Substring_당근(t *testing.T) {
+	s := testStore()
+	results := s.Search("당근")
+	if len(results) != 2 {
+		t.Fatalf("'당근' substring 검색: 2건 기대, got %d %v", len(results), corpNames(results))
+	}
+}
+
+func TestSearch_Substring_카카오(t *testing.T) {
+	// "카카오" 이름 완전일치가 존재하므로 정확히 1건만 반환 (substring 전에 리턴)
+	s := testStore()
+	results := s.Search("카카오")
+	if len(results) != 1 || results[0].CorpName != "카카오" {
+		t.Fatalf("'카카오' 이름 완전일치: 1건 기대, got %d %v", len(results), corpNames(results))
+	}
+}
+
+func TestSearch_Substring_카카오계열(t *testing.T) {
+	// "카카오뱅" 처럼 완전일치 없는 쿼리는 substring 으로 카카오뱅크를 찾음
+	s := testStore()
+	results := s.Search("카카오뱅")
+	if len(results) != 1 || results[0].CorpName != "카카오뱅크" {
+		t.Fatalf("'카카오뱅' substring 검색: 카카오뱅크 기대, got %v", corpNames(results))
+	}
+}
+
+func TestSearch_Substring_핑크퐁(t *testing.T) {
+	s := testStore()
+	results := s.Search("핑크퐁")
+	if len(results) != 1 || results[0].CorpName != "더핑크퐁컴퍼니" {
+		t.Fatalf("'핑크퐁' substring 검색 실패: got %v", corpNames(results))
+	}
+}
+
+// --- fuzzy tests ---
+
+func TestSearch_Fuzzy_마켓컬리(t *testing.T) {
+	s := testStore()
+	results := s.Search("마켓컬리")
+	if len(results) == 0 {
+		t.Fatal("'마켓컬리' fuzzy 검색: 결과 없음 (기대: '컬리' 포함)")
+	}
+	found := false
+	for _, r := range results {
+		if r.CorpName == "컬리" {
+			found = true
+		}
+	}
+	if !found {
+		t.Fatalf("'마켓컬리' fuzzy 결과에 '컬리' 없음: got %v", corpNames(results))
+	}
+	t.Logf("'마켓컬리' fuzzy 결과: %v", corpNames(results))
+}
+
+func TestSearch_Fuzzy_오늘의집(t *testing.T) {
+	// "오늘의집" → substring 으로도 찾힘, fuzzy 로도 찾혀야 함
+	s := testStore()
+	results := s.Search("오늘의집")
+	if len(results) == 0 {
+		t.Fatal("'오늘의집' 검색: 결과 없음")
+	}
+	found := false
+	for _, r := range results {
+		if r.CorpName == "주식회사 오늘의집" {
+			found = true
+		}
+	}
+	if !found {
+		t.Fatalf("'오늘의집' 결과에 '주식회사 오늘의집' 없음: got %v", corpNames(results))
+	}
+}
+
+func TestSearch_NoResult_완전엉뚱한검색어(t *testing.T) {
+	s := testStore()
+	results := s.Search("xyzxyz없는기업명zyx")
+	if len(results) != 0 {
+		t.Logf("'xyzxyz없는기업명zyx' 검색 결과 (fuzzy 허용): %v", corpNames(results))
+	}
+}
+
+// --- bigramSim unit tests ---
+
+func TestBigramSim_완전동일(t *testing.T) {
+	score := bigramSim("삼성전자", "삼성전자")
+	if score != 1.0 {
+		t.Fatalf("동일 문자열 bigramSim: 1.0 기대, got %f", score)
+	}
+}
+
+func TestBigramSim_부분겹침(t *testing.T) {
+	// "마켓컬리" bigrams: [마켓, 켓컬, 컬리]
+	// "컬리" bigrams: [컬리]  → 1 match / 3 query bigrams = 0.333
+	score := bigramSim("마켓컬리", "컬리")
+	if score < 0.3 {
+		t.Fatalf("'마켓컬리' vs '컬리': 0.3 이상 기대, got %f", score)
+	}
+}
+
+func TestBigramSim_핑크퐁(t *testing.T) {
+	// "핑크퐁" bigrams: [핑크, 크퐁]
+	// "더핑크퐁컴퍼니" bigrams: [더핑, 핑크, 크퐁, 퐁컴, 컴퍼, 퍼니] → 2 match / 2 = 1.0
+	score := bigramSim("핑크퐁", "더핑크퐁컴퍼니")
+	if score != 1.0 {
+		t.Fatalf("'핑크퐁' vs '더핑크퐁컴퍼니': 1.0 기대, got %f", score)
+	}
+}
+
+func TestBigramSim_단일문자(t *testing.T) {
+	// 단일 rune은 bigram 없음 → 0
+	score := bigramSim("가", "가나다라")
+	if score != 0 {
+		t.Fatalf("단일 문자 bigramSim: 0 기대, got %f", score)
+	}
+}