Merge pull request #7426 from xlplbo/fix/openai-codex-turn-metadata-ascii

fix(openai): 改写 Codex turn metadata 时保留非 ASCII 转义

backend/internal/service/openai_codex_account_identity.go

@@ -154,7 +154,7 @@ func applyCodexAccountIdentityEmbeddedMetadata(values map[string]any, account *A
 	if !applyCodexAccountIdentityFields(metadata, account, apiKeyID) {
 		return false
 	}
-	rebuilt, err := json.Marshal(metadata)
+	rebuilt, err := marshalCodexTurnMetadata(metadata)
 	if err != nil {
 		return false
 	}
@@ -267,7 +267,7 @@ func applyCodexAccountIdentityHeaders(headers http.Header, account *Account, api
 	if raw := strings.TrimSpace(headers.Get(openAIWSTurnMetadataHeader)); raw != "" {
 		metadata := map[string]any{}
 		if err := json.Unmarshal([]byte(raw), &metadata); err == nil && metadata != nil && applyCodexAccountIdentityFields(metadata, account, apiKeyID) {
-			if rebuilt, err := json.Marshal(metadata); err == nil {
+			if rebuilt, err := marshalCodexTurnMetadata(metadata); err == nil {
 				headers.Set(openAIWSTurnMetadataHeader, string(rebuilt))
 			}
 		}

backend/internal/service/openai_codex_fingerprint.go

@@ -402,7 +402,7 @@ func rewriteCodexTurnMetadataFields(h http.Header, fields map[string]any) {
 	for k, v := range fields {
 		metadata[k] = v
 	}
-	rebuilt, err := json.Marshal(metadata)
+	rebuilt, err := marshalCodexTurnMetadata(metadata)
 	if err != nil {
 		return
 	}
@@ -595,7 +595,7 @@ func rewriteClientMetadataEmbeddedTurnMetadata(clientMetadata map[string]any, fi
 	for k, v := range fields {
 		metadata[k] = v
 	}
-	if rebuilt, err := json.Marshal(metadata); err == nil {
+	if rebuilt, err := marshalCodexTurnMetadata(metadata); err == nil {
 		clientMetadata["x-codex-turn-metadata"] = string(rebuilt)
 	}
 }

backend/internal/service/openai_codex_turn_metadata.go

@@ -0,0 +1,39 @@
+package service
+
+import (
+	"encoding/json"
+	"unicode/utf16"
+)
+
+// Turn metadata is also used as an HTTP header, including when carried in WS JSON.
+func marshalCodexTurnMetadata(metadata map[string]any) ([]byte, error) {
+	raw, err := json.Marshal(metadata)
+	if err != nil {
+		return nil, err
+	}
+	for i, b := range raw {
+		if b < 0x7f {
+			continue
+		}
+		out := make([]byte, 0, len(raw)+16)
+		out = append(out, raw[:i]...)
+		appendEscape := func(r rune) {
+			const hex = "0123456789abcdef"
+			out = append(out, '\\', 'u', hex[r>>12&15], hex[r>>8&15], hex[r>>4&15], hex[r&15])
+		}
+		for _, r := range string(raw[i:]) {
+			switch {
+			case r < 0x7f:
+				out = append(out, byte(r))
+			case r <= 0xffff:
+				appendEscape(r)
+			default:
+				high, low := utf16.EncodeRune(r)
+				appendEscape(high)
+				appendEscape(low)
+			}
+		}
+		return out, nil
+	}
+	return raw, nil
+}

backend/internal/service/openai_codex_turn_metadata_test.go

@@ -0,0 +1,88 @@
+package service
+
+import (
+	"encoding/json"
+	"net/http"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+	"github.com/tidwall/gjson"
+)
+
+func TestCodexTurnMetadataRewritePreservesHeaderSafeJSON(t *testing.T) {
+	account := newTestOAuthAccount(1, map[string]any{codexFingerprintModeExtraKey: "session"})
+	account.Credentials = map[string]any{"chatgpt_account_id": "test-account"}
+	ids := resolveCodexFingerprintIDsFromRequest(account, http.Header{})
+	require.NotNil(t, ids)
+
+	rewriters := map[string]func(*testing.T, string) string{
+		"account_header": func(t *testing.T, raw string) string {
+			h := make(http.Header)
+			h.Set(openAIWSTurnMetadataHeader, raw)
+			applyCodexAccountIdentityHeaders(h, account, 77)
+			return h.Get(openAIWSTurnMetadataHeader)
+		},
+		"account_body_map": func(t *testing.T, raw string) string {
+			cm := map[string]any{openAIWSTurnMetadataHeader: raw}
+			require.True(t, applyCodexAccountIdentityClientMetadataMap(map[string]any{"client_metadata": cm}, account, 77))
+			value, ok := cm[openAIWSTurnMetadataHeader].(string)
+			require.True(t, ok)
+			return value
+		},
+		"account_body_raw": func(t *testing.T, raw string) string {
+			body, err := json.Marshal(map[string]any{"client_metadata": map[string]any{openAIWSTurnMetadataHeader: raw}})
+			require.NoError(t, err)
+			updated, changed, err := applyCodexAccountIdentityClientMetadataRaw(body, account, 77)
+			require.NoError(t, err)
+			require.True(t, changed)
+			return gjson.GetBytes(updated, "client_metadata."+openAIWSTurnMetadataHeader).String()
+		},
+		"fingerprint_header": func(t *testing.T, raw string) string {
+			h := make(http.Header)
+			h.Set(openAIWSTurnMetadataHeader, raw)
+			applyCodexFingerprintHeaders(h, ids)
+			return h.Get(openAIWSTurnMetadataHeader)
+		},
+		"fingerprint_body_map": func(t *testing.T, raw string) string {
+			cm := map[string]any{openAIWSTurnMetadataHeader: raw}
+			require.True(t, applyCodexFingerprintClientMetadata(map[string]any{"client_metadata": cm}, ids))
+			value, ok := cm[openAIWSTurnMetadataHeader].(string)
+			require.True(t, ok)
+			return value
+		},
+		"fingerprint_body_raw": func(t *testing.T, raw string) string {
+			body, err := json.Marshal(map[string]any{"client_metadata": map[string]any{openAIWSTurnMetadataHeader: raw}})
+			require.NoError(t, err)
+			updated, changed, err := applyCodexFingerprintClientMetadataRaw(body, ids)
+			require.NoError(t, err)
+			require.True(t, changed)
+			return gjson.GetBytes(updated, "client_metadata."+openAIWSTurnMetadataHeader).String()
+		},
+	}
+
+	for _, raw := range []string{
+		`{"installation_id":"client-install","workspaces":{"C:\\work\\中文🚀":{}}}`,
+		`{"installation_id":"client-install","workspaces":{"C:\\work\\中文🚀":{"label":"café"}},"literal":"\\u4e2d","quote":"\"\\","controls":"\b\f\n\r\t\u0000\u007f","timestamp":1789302780858,"enabled":true}`,
+		`{"installation_id":"client-install","workspaces":{"C:\\work\\\u4e2d\u6587\ud83d\ude80":{"label":"caf\u00e9"}},"literal":"\\u4e2d","quote":"\"\\","controls":"\b\f\n\r\t\u0000\u007f","timestamp":1789302780858,"enabled":true}`,
+		`{"installation_id":"client-install","workspaces":{"C:\\work\\ascii":{}},"literal":"\\u4e2d"}`,
+	} {
+		var original map[string]any
+		require.NoError(t, json.Unmarshal([]byte(raw), &original))
+		for name, rewrite := range rewriters {
+			t.Run(name, func(t *testing.T) {
+				updated := rewrite(t, raw)
+				var decoded map[string]any
+				require.NoError(t, json.Unmarshal([]byte(updated), &decoded))
+				for key, value := range original {
+					if key != "installation_id" {
+						require.Equal(t, value, decoded[key], key)
+					}
+				}
+				require.NotEqual(t, original["installation_id"], decoded["installation_id"])
+				for _, b := range []byte(updated) {
+					require.True(t, b >= 0x20 && b < 0x7f, "non-printable ASCII header byte: 0x%02x", b)
+				}
+			})
+		}
+	}
+}