From 4e5632c3e32f9a8a5150c44e8a7f46efe7fc2688 Mon Sep 17 00:00:00 2001
From: smyyyyy <1298311250@qq.com>
Date: Thu, 10 Sep 2026 03:18:31 +0800
Subject: [PATCH] fix: validate Codex User-Agent before identity parsing

---
 backend/internal/pkg/openai/request.go        | 15 ++++++++
 .../pkg/openai/request_ua_validation_test.go  | 37 +++++++++++++++++++
 .../internal/service/openai_codex_identity.go |  9 +++--
 .../service/openai_ua_validation_test.go      | 29 +++++++++++++++
 4 files changed, 87 insertions(+), 3 deletions(-)
 create mode 100644 backend/internal/pkg/openai/request_ua_validation_test.go
 create mode 100644 backend/internal/service/openai_ua_validation_test.go

diff --git a/backend/internal/pkg/openai/request.go b/backend/internal/pkg/openai/request.go
index 172021b8ea..8ddf8d188d 100644
--- a/backend/internal/pkg/openai/request.go
+++ b/backend/internal/pkg/openai/request.go
@@ -3,6 +3,8 @@ package openai
 import (
 	"regexp"
 	"strings"
+
+	"golang.org/x/net/http/httpguts"
 )
 
 // CodexCLIUserAgentPrefixes matches Codex CLI User-Agent patterns
@@ -198,6 +200,10 @@ func matchCodexClientHeaderStrictPrefixes(value string, prefixes []string) bool
 //     UA 首段后配对，保留真实版本/OS/终端指纹；
 //  3. 均不命中 → ok=false，调用方应整体回退为默认官方身份。
 func PairCodexClientIdentity(userAgent string) (originator string, pairedUA string, ok bool) {
+	// Validate before trimming so control bytes cannot become a valid identity.
+	if !validCodexUserAgentValue(userAgent) {
+		return "", "", false
+	}
 	ua := strings.TrimSpace(userAgent)
 	slash := strings.IndexByte(ua, '/')
 	if slash <= 0 {
@@ -217,6 +223,15 @@ func PairCodexClientIdentity(userAgent string) (originator string, pairedUA stri
 	return "", "", false
 }
 
+func validCodexUserAgentValue(value string) bool {
+	if !httpguts.ValidHeaderFieldValue(value) {
+		return false
+	}
+	// httpguts follows the legacy field-value grammar and permits obs-fold.
+	// User-Agent is not an obs-folded header; reject CR/LF before forwarding.
+	return !strings.ContainsAny(value, "\r\n")
+}
+
 // codexOriginatorMaxLen 官方 clientInfo.name 均为短 ASCII 标识，远低于此上限。
 const codexOriginatorMaxLen = 64
 
diff --git a/backend/internal/pkg/openai/request_ua_validation_test.go b/backend/internal/pkg/openai/request_ua_validation_test.go
new file mode 100644
index 0000000000..a74c86f90a
--- /dev/null
+++ b/backend/internal/pkg/openai/request_ua_validation_test.go
@@ -0,0 +1,37 @@
+package openai
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestPairCodexClientIdentityRejectsInvalidHeaderBytes(t *testing.T) {
+	for name, ua := range map[string]string{
+		"crlf_suffix":        "codex-tui/0.146.0 (Linux)\r\nX-Injected: value",
+		"nul_suffix":         "codex-tui/0.146.0 (Linux\x00)",
+		"del_suffix":         "codex-tui/0.146.0 terminal\x7f",
+		"newline_prefix":     "\ncodex-tui/0.146.0",
+		"newline_suffix":     "codex-tui/0.146.0\n",
+		"control_in_trailer": "custom/0.146.0 bad\x01 (codex-tui; 0.146.0)",
+	} {
+		t.Run(name, func(t *testing.T) {
+			originator, paired, ok := PairCodexClientIdentity(ua)
+			require.False(t, ok)
+			require.Empty(t, originator)
+			require.Empty(t, paired)
+		})
+	}
+}
+
+func TestPairCodexClientIdentityPreservesValidSuffix(t *testing.T) {
+	for _, ua := range []string{
+		"codex-tui/0.146.0 (Mac OS X 14.0; arm64) iTerm",
+		"codex-tui/0.146.0 (Windows NT 10.0; x86_64) terminal",
+		"codex-tui/0.146.0 (Linux)\tterminal",
+	} {
+		_, paired, ok := PairCodexClientIdentity(ua)
+		require.True(t, ok)
+		require.Equal(t, ua, paired)
+	}
+}
diff --git a/backend/internal/service/openai_codex_identity.go b/backend/internal/service/openai_codex_identity.go
index 2cb8a45c40..09d1717c9a 100644
--- a/backend/internal/service/openai_codex_identity.go
+++ b/backend/internal/service/openai_codex_identity.go
@@ -114,7 +114,7 @@ func codexCanonicalUserAgent() string {
 	resolver := codexCanonicalUAResolver
 	codexCanonicalUAMu.RUnlock()
 	if resolver != nil {
-		if ua := strings.TrimSpace(resolver()); ua != "" {
+		if ua := resolver(); strings.TrimSpace(ua) != "" {
 			return ua
 		}
 	}
@@ -139,8 +139,11 @@ type codexOutboundIdentity struct {
 // 需要固定版本请填「Codex 客户端版本号」并关闭自动同步。
 func resolveCodexOutboundIdentity(candidateUA string) codexOutboundIdentity {
 	canonical := codexCanonicalUserAgent()
-	ua := strings.TrimSpace(candidateUA)
-	if ua == "" {
+	if _, _, ok := openai.PairCodexClientIdentity(canonical); !ok {
+		canonical = codexCLIUserAgent
+	}
+	ua := candidateUA
+	if strings.TrimSpace(ua) == "" {
 		ua = canonical
 	}
 	originator, pairedUA, ok := openai.PairCodexClientIdentity(ua)
diff --git a/backend/internal/service/openai_ua_validation_test.go b/backend/internal/service/openai_ua_validation_test.go
new file mode 100644
index 0000000000..08872f4059
--- /dev/null
+++ b/backend/internal/service/openai_ua_validation_test.go
@@ -0,0 +1,29 @@
+package service
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestCanonicalCodexIdentityValidatesBeforeTrimming(t *testing.T) {
+	codexCanonicalUAMu.RLock()
+	previous := codexCanonicalUAResolver
+	codexCanonicalUAMu.RUnlock()
+	t.Cleanup(func() { SetCodexCanonicalUserAgentResolver(previous) })
+
+	for _, raw := range []string{
+		"\ncodex-tui/0.200.1 (Windows)",
+		"codex-tui/0.200.1 (Windows)\n",
+		"codex-tui/0.200.1 bad\x00",
+	} {
+		SetCodexCanonicalUserAgentResolver(func() string { return raw })
+		identity := resolveCodexOutboundIdentity("")
+		require.Equal(t, codexCLIUserAgent, identity.userAgent)
+		require.Equal(t, codexCLIVersion, identity.version)
+	}
+
+	SetCodexCanonicalUserAgentResolver(nil)
+	identity := resolveCodexOutboundIdentity("codex_cli_rs/0.145.2 (Windows)\n")
+	require.Equal(t, codexCLIUserAgent, identity.userAgent)
+}