celery 升级优化V2 | 日志调整 | 前端BUG修复

video-gen-api/app/services/operation_log_service.py

@@ -21,7 +21,6 @@ MODULE_GENERATION_LOG_ROOT = os.path.join(LOG_BASE_DIR, "ModuleGeneration")
 AI_MODEL_LOG_ROOT = LOG_DIR
 SENSITIVE_KEY_PATTERNS = (
     "secret",
-    "token",
     "authorization",
     "cookie",
     "credential",
@@ -30,9 +29,28 @@ SENSITIVE_KEY_PATTERNS = (
     "access_key",
     "api_key",
     "apikey",
-    "security-token",
-    "x-tos-security-token",
+    "security_token",
 )
+SENSITIVE_TOKEN_KEYS = {
+    "token",
+    "access_token",
+    "refresh_token",
+    "bearer_token",
+    "security_token",
+    "x_tos_security_token",
+}
+FILE_BASE64_KEYS = {
+    "b64_json",
+    "file_data",
+    "file_base64",
+    "content_base64",
+    "image_base64",
+    "video_base64",
+    "audio_base64",
+}
+FILE_DATA_URI_MIME_PREFIXES = ("image/", "video/", "audio/")
+FILE_DATA_URI_MIME_TYPES = {"application/pdf", "application/octet-stream"}
+FILE_BASE64_PREVIEW_CHARS = 30
 
 
 def _safe_name(value: str | None, default: str = "unknown") -> str:
@@ -49,7 +67,41 @@ def _mask_string(value: str) -> str:
 
 def _is_sensitive_key(key: str) -> bool:
     lower = str(key).replace("-", "_").lower()
-    return lower == "sign" or any(pattern in lower for pattern in SENSITIVE_KEY_PATTERNS)
+    if lower == "sign" or lower in SENSITIVE_TOKEN_KEYS:
+        return True
+    return any(pattern in lower for pattern in SENSITIVE_KEY_PATTERNS)
+
+
+def _decoded_base64_size(value: str) -> int:
+    compact = "".join(value.split())
+    if not compact:
+        return 0
+    padding = 2 if compact.endswith("==") else (1 if compact.endswith("=") else 0)
+    return max(0, (len(compact) * 3) // 4 - padding)
+
+
+def _file_base64_preview(value: str, key_path: tuple[str, ...]) -> str | None:
+    data_uri = re.match(r"^data:([^;,]+);base64,(.*)$", value, flags=re.IGNORECASE | re.DOTALL)
+    prefix = ""
+    payload = value
+    is_file = False
+    if data_uri:
+        mime = str(data_uri.group(1) or "").lower()
+        is_file = mime.startswith(FILE_DATA_URI_MIME_PREFIXES) or mime in FILE_DATA_URI_MIME_TYPES
+        prefix = value[: value.find(",") + 1]
+        payload = data_uri.group(2)
+    elif key_path and key_path[-1].replace("-", "_").lower() in FILE_BASE64_KEYS:
+        # Raw base64 is treated as file content only for an explicit file field.
+        is_file = len(value) >= 64 and bool(re.fullmatch(r"[A-Za-z0-9+/=\s]+", value))
+    if not is_file:
+        return None
+
+    compact = "".join(payload.split())
+    preview = compact[:FILE_BASE64_PREVIEW_CHARS]
+    total_bytes = _decoded_base64_size(compact)
+    preview_bytes = min(total_bytes, (len(preview) * 3) // 4)
+    remaining_bytes = max(0, total_bytes - preview_bytes)
+    return f"{prefix}{preview}...<remaining_file_bytes:{remaining_bytes}>"
 
 
 def _sanitize_url(value: str) -> str:
@@ -65,10 +117,13 @@ def _sanitize_url(value: str) -> str:
         return value
 
 
-def sanitize_log_value(value: Any) -> Any:
+def sanitize_log_value(value: Any, *, key_path: tuple[str, ...] = ()) -> Any:
     if value is None:
         return None
     if isinstance(value, str):
+        file_preview = _file_base64_preview(value, key_path)
+        if file_preview is not None:
+            return file_preview
         text = _sanitize_url(value) if value.startswith(("http://", "https://")) else value
         if len(text) > MAX_LOG_FIELD_LENGTH:
             return text[:MAX_LOG_FIELD_LENGTH] + f"...<truncated:{len(text) - MAX_LOG_FIELD_LENGTH}>"
@@ -77,10 +132,14 @@ def sanitize_log_value(value: Any) -> Any:
         output: dict[str, Any] = {}
         for k, v in value.items():
             key = str(k)
-            output[key] = "***" if _is_sensitive_key(key) else sanitize_log_value(v)
+            output[key] = (
+                "***"
+                if _is_sensitive_key(key)
+                else sanitize_log_value(v, key_path=(*key_path, key))
+            )
         return output
-    if isinstance(value, list):
-        return [sanitize_log_value(v) for v in value]
+    if isinstance(value, (list, tuple)):
+        return [sanitize_log_value(v, key_path=(*key_path, str(index))) for index, v in enumerate(value)]
     return value
 
 
@@ -262,6 +321,13 @@ def log_ai_model_event(
     *,
     event_type: str,
     module: str | None = None,
+    step_code: str | None = None,
+    call_id: str | None = None,
+    event_phase: str | None = None,
+    owner_type: str | None = None,
+    owner_id: str | None = None,
+    generation_attempt_no: int | None = None,
+    latency_ms: int | None = None,
     event_status: str = "success",
     source: str | None = None,
     trace_id: str | None = None,
@@ -314,8 +380,19 @@ def log_ai_model_event(
     )
     entry.update(
         {
+            "call_id": call_id,
+            "event_phase": event_phase or event_type,
+            "step_code": step_code,
+            "owner_type": owner_type,
+            "owner_id": owner_id,
+            "generation_attempt_no": generation_attempt_no,
+            "latency_ms": latency_ms,
+            # Preserve the legacy fields for existing log readers while also
+            # exposing unambiguous configuration/provider model names.
             "model_name": model_config_name,
             "model_id": model_name,
+            "model_config_name": model_config_name,
+            "provider_model_name": model_name,
             "model_config_id": model_config_id,
             "provider": provider,
             "api_base": api_base,