Fix the fucking regexp

cmd/record-indexer/postgres_json_test.go

@@ -0,0 +1,26 @@
+package main
+
+import (
+	"testing"
+)
+
+func TestPostgresFix(t *testing.T) {
+	type testCase struct{ input, want string }
+
+	cases := []testCase{
+		{`"a"`, `"a"`},
+		{`"\u0000"`, `"<0x00>"`},
+		{`"description":"\u0000"`, `"description":"<0x00>"`},
+		{`"\\u0000"`, `"\\u0000"`},
+		{`"\\\u0000"`, `"\\<0x00>"`},
+		{`\n\n\u0000\u0000 \u0000\u0000\u0000\u0000 \u0000\u0000\u0000\u0000\u0000`,
+			`\n\n<0x00><0x00> <0x00><0x00><0x00><0x00> <0x00><0x00><0x00><0x00><0x00>`},
+	}
+
+	for _, tc := range cases {
+		got := escapeNullCharForPostgres([]byte(tc.input))
+		if string(got) != tc.want {
+			t.Errorf("escapeNullCharForPostgres(%s) = %s, want %s", tc.input, string(got), tc.want)
+		}
+	}
+}

cmd/record-indexer/workerpool.go

@@ -132,10 +132,12 @@ func (p *WorkerPool) worker(ctx context.Context, signal chan struct{}) {
 	}
 }
 
-var postgresFixRegexp = regexp.MustCompile(`[^\\](\\\\)*(\\u0000)`)
+var postgresFixRegexp = regexp.MustCompile(`([^\\](\\\\)*)(\\u0000)+`)
 
 func escapeNullCharForPostgres(b []byte) []byte {
-	return postgresFixRegexp.ReplaceAll(b, []byte(`$1<0x00>`))
+	return postgresFixRegexp.ReplaceAllFunc(b, func(b []byte) []byte {
+		return bytes.ReplaceAll(b, []byte(`\u0000`), []byte(`<0x00>`))
+	})
 }
 
 func (p *WorkerPool) doWork(ctx context.Context, work WorkItem) error {
@@ -212,7 +214,6 @@ retry:
 			log.Warn().Msgf("Unexpected key format: %q", k)
 			continue
 		}
-		v = regexp.MustCompile(`[^\\](\\\\)*(\\u0000)`).ReplaceAll(v, []byte(`$1<0x00>`))
 		// lang, err := repo.GetLang(ctx, v)
 		// if err == nil {
 		// 	postsByLanguageIndexed.WithLabelValues(u.String(), lang).Inc()