Add grok-superimage server-side support, robust SSE streaming/errors, and proxy/imagine fixes

app/services/grok/services/image.py

@@ -546,33 +546,36 @@ async def process(self, response: AsyncIterable[dict]) -> AsyncGenerator[str, No
                     )
 
         if self.n == 1:
-            if self._target_id and self._target_id in images:
+            if (
+                self._target_id
+                and self._target_id in images
+                and images[self._target_id].get("is_final", False)
+            ):
                 selected = [(self._target_id, images[self._target_id])]
             else:
+                final_candidates = [
+                    item for item in images.items() if item[1].get("is_final", False)
+                ]
                 selected = (
-                    [
-                        max(
-                            images.items(),
-                            key=lambda x: (
-                                x[1].get("is_final", False),
-                                x[1].get("blob_size", 0),
-                            ),
-                        )
-                    ]
-                    if images
+                    [max(final_candidates, key=lambda x: x[1].get("blob_size", 0))]
+                    if final_candidates
                     else []
                 )
         else:
             selected = [
                 (image_id, images[image_id])
                 for image_id in self._index_map
-                if image_id in images
+                if image_id in images and images[image_id].get("is_final", False)
             ]
 
         for image_id, item in selected:
             if self.response_format == "url":
+                final_image_id = image_id
+                # Keep original imagine image name for superimage chat stream output.
+                if self.model != "grok-superimage-1.0":
+                    final_image_id = f"{image_id}-final"
                 output = await self._save_blob(
-                    f"{image_id}-final",
+                    final_image_id,
                     item.get("blob", ""),
                     item.get("is_final", False),
                     ext=item.get("ext"),
@@ -672,8 +675,8 @@ async def process(self, response: AsyncIterable[dict]) -> List[str]:
             images[image_id] = self._pick_best(images.get(image_id), item)
 
         selected = sorted(
-            images.values(),
-            key=lambda x: (x.get("is_final", False), x.get("blob_size", 0)),
+            [item for item in images.values() if item.get("is_final", False)],
+            key=lambda x: x.get("blob_size", 0),
             reverse=True,
         )
         if self.n:

app/services/reverse/ws_imagine.py

@@ -37,10 +37,9 @@ def _parse_image_url(self, url: str) -> tuple[Optional[str], Optional[str]]:
         return match.group(1), match.group(2).lower()
 
     def _is_final_image(self, url: str, blob_size: int, final_min_bytes: int) -> bool:
-        url_lower = (url or "").lower()
-        if url_lower.endswith((".jpg", ".jpeg")):
-            return True
-        return blob_size > final_min_bytes
+        # Final image must satisfy byte-size threshold to avoid tiny preview
+        # images being treated as final outputs.
+        return blob_size >= final_min_bytes
 
     def _classify_image(self, url: str, blob: str, final_min_bytes: int, medium_min_bytes: int) -> Optional[Dict[str, object]]:
         if not url or not blob:

readme.md

@@ -180,6 +180,7 @@ curl http://localhost:8000/v1/chat/completions \
 - 工具调用为**提示词模拟 + 客户端执行回填**：模型通过 `<tool_call>{...}</tool_call>` 输出调用请求，服务端解析为 `tool_calls`；不执行工具。
 - `grok-superimage-1.0` 与瀑布流 imagine 生成链路一致，可直接通过 `/v1/chat/completions` 调用；其 `n/size/response_format` 由服务端 `[superimage]` 统一控制。
 - `grok-superimage-1.0` 在 `/v1/chat/completions` 的流式输出仅返回最终成图，不返回中间预览图。
+- `grok-superimage-1.0` 流式 URL 出图会保持原始图片名（不追加 `-final` 后缀）。
 - `grok-imagine-1.0-edit` 必须提供图片，多图默认取**最后 3 张**与最后一个文本。
 - `grok-imagine-1.0-video` 支持文生视频与图生视频（通过 `image_url` 传参考图，**仅取第 1 张**）。
 - 除上述外的其他参数将自动丢弃并忽略。