lywsvip/flow2api
1
0
Fork 0
mirror of https://github.com/TheSmallHanCat/flow2api.git synced 2026-06-01 04:11:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix PR 133 merge integration issues

Browse Source
This commit is contained in:
genz27
2026-04-30 18:49:59 +08:00
parent 55431c93b2
commit abd0c0001f
19 changed files with 113 additions and 284 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
cat.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 7.6 KiB

BIN
cat_watercolor.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 866 KiB

6
docker-compose.yml
View File

@@ -4,14 +4,12 @@ services:
flow2api:
image: ghcr.io/thesmallhancat/flow2api:latest
container_name: flow2api
network_mode: host
ports:
- "38000:8000"
volumes:
- ./data:/app/data
- ./tmp:/app/tmp
- ./config/setting.toml:/app/config/setting.toml
- ./src/services/proxy_manager.py:/app/src/services/proxy_manager.py
- ./src/services/flow_client.py:/app/src/services/flow_client.py
- ./src/api/admin.py:/app/src/api/admin.py
environment:
- PYTHONUNBUFFERED=1
restart: unless-stopped

7
extension/background.js
View File

@@ -4,6 +4,7 @@ let heartbeatInterval = null;
const DEFAULT_SETTINGS = {
serverUrl: "ws://127.0.0.1:8000/captcha_ws",
apiKey: "",
routeKey: "",
clientLabel: ""
};
@@ -13,6 +14,7 @@ function getSettings() {
chrome.storage.local.get(DEFAULT_SETTINGS, (stored) => {
resolve({
serverUrl: (stored.serverUrl || DEFAULT_SETTINGS.serverUrl).trim(),
apiKey: (stored.apiKey || "").trim(),
routeKey: (stored.routeKey || "").trim(),
clientLabel: (stored.clientLabel || "").trim()
});
@@ -74,6 +76,9 @@ async function connectWS() {
const settings = await getSettings();
const url = new URL(settings.serverUrl || DEFAULT_SETTINGS.serverUrl);
if (settings.apiKey) {
url.searchParams.set("key", settings.apiKey);
}
if (settings.routeKey) {
url.searchParams.set("route_key", settings.routeKey);
}
@@ -227,7 +232,7 @@ async function handleGetToken(data) {
chrome.storage.onChanged.addListener((changes, areaName) => {
if (areaName !== "local") return;
if (changes.routeKey || changes.serverUrl || changes.clientLabel) {
if (changes.routeKey || changes.serverUrl || changes.apiKey || changes.clientLabel) {
console.log("[Flow2API] Extension settings changed, reconnecting WebSocket...");
closeSocket();
connectWS();

18
extension/manifest 2.json
View File

@@ -1,18 +0,0 @@
{
"manifest_version": 3,
"name": "Flow2API Captcha Worker",
"version": "1.0",
"permissions": ["activeTab", "scripting", "tabs", "storage"],
"host_permissions": ["https://labs.google/*"],
"options_page": "options.html",
"background": {
"service_worker": "background.js"
},
"content_scripts": [
{
"matches": ["https://labs.google/*"],
"js": ["content.js"],
"run_at": "document_end"
}
]
}

2
extension/manifest.json
View File

@@ -9,7 +9,7 @@
"scripting"
],
"host_permissions": [
"<all_urls>"
"https://labs.google/*"
],
"background": {
"service_worker": "background.js"

47
extension/options 2.js
View File

@@ -1,47 +0,0 @@
const DEFAULTS = {
routeKey: "",
clientLabel: "",
serverUrl: "ws://127.0.0.1:8000/captcha_ws"
};
function $(id) {
return document.getElementById(id);
}
function showStatus(message, isError = false) {
const status = $("status");
status.textContent = message;
status.style.color = isError ? "#b91c1c" : "#065f46";
}
function loadSettings() {
chrome.storage.local.get(DEFAULTS, (stored) => {
$("routeKey").value = stored.routeKey || "";
$("clientLabel").value = stored.clientLabel || "";
$("serverUrl").value = stored.serverUrl || DEFAULTS.serverUrl;
});
}
function saveSettings() {
const routeKey = $("routeKey").value.trim();
const clientLabel = $("clientLabel").value.trim();
const serverUrl = $("serverUrl").value.trim() || DEFAULTS.serverUrl;
if (!serverUrl.startsWith("ws://") && !serverUrl.startsWith("wss://")) {
showStatus("WebSocket URL 必须以 ws:// 或 wss:// 开头", true);
return;
}
chrome.storage.local.set({ routeKey, clientLabel, serverUrl }, () => {
if (chrome.runtime.lastError) {
showStatus(chrome.runtime.lastError.message || "保存失败", true);
return;
}
showStatus("已保存，后台连接会自动重连");
});
}
document.addEventListener("DOMContentLoaded", () => {
loadSettings();
$("saveBtn").addEventListener("click", saveSettings);
});

5
extension/options.html
View File

@@ -87,9 +87,12 @@
<label for="serverUrl">WebSocket URL</label>
<input id="serverUrl" type="text" placeholder="ws://127.0.0.1:8000/captcha_ws">
<label for="apiKey">Flow2API API Key</label>
<input id="apiKey" type="password" placeholder="管理后台里的 API Key">
<button id="saveBtn" type="button">保存</button>
<div class="status" id="status"></div>
<div class="hint">管理台 token 里填写同样的 <code>Route Key</code>，比如 9223 对 9223。</div>
<div class="hint">管理台 token 里填写同样的 <code>Route Key</code>，比如 9223 对 9223。后端会使用 API Key 校验 WebSocket 连接。</div>
</div>
</div>
<script src="options.js"></script>

8
extension/options.js
View File

@@ -1,5 +1,6 @@
const DEFAULT_SETTINGS = {
serverUrl: "ws://127.0.0.1:8000/captcha_ws",
apiKey: "",
routeKey: "",
clientLabel: ""
};
@@ -9,6 +10,7 @@ const $ = (id) => document.getElementById(id);
function normalizeSettings(values) {
return {
serverUrl: (values.serverUrl || DEFAULT_SETTINGS.serverUrl).trim(),
apiKey: (values.apiKey || "").trim(),
routeKey: (values.routeKey || "").trim(),
clientLabel: (values.clientLabel || "").trim()
};
@@ -33,6 +35,7 @@ function loadSettings() {
chrome.storage.local.get(DEFAULT_SETTINGS, (stored) => {
const settings = normalizeSettings(stored);
$("serverUrl").value = settings.serverUrl;
$("apiKey").value = settings.apiKey;
$("routeKey").value = settings.routeKey;
$("clientLabel").value = settings.clientLabel;
});
@@ -41,6 +44,7 @@ function loadSettings() {
function saveSettings() {
const settings = normalizeSettings({
serverUrl: $("serverUrl").value,
apiKey: $("apiKey").value,
routeKey: $("routeKey").value,
clientLabel: $("clientLabel").value
});
@@ -49,6 +53,10 @@ function saveSettings() {
setStatus("WebSocket URL 必须以 ws:// 或 wss:// 开头。", true);
return;
}
if (!settings.apiKey) {
setStatus("请填写 Flow2API API Key。", true);
return;
}
chrome.storage.local.set(settings, () => {
if (chrome.runtime.lastError) {

27
src/api/routes.py
View File

@@ -9,10 +9,10 @@ import re
from urllib.parse import urlparse
from curl_cffi.requests import AsyncSession
from fastapi import APIRouter, Depends, HTTPException, Query, WebSocket, WebSocketDisconnect
from fastapi import APIRouter, Depends, HTTPException, Query, Request, WebSocket, WebSocketDisconnect
from fastapi.responses import JSONResponse, StreamingResponse
from ..core.auth import verify_api_key_flexible
from ..core.auth import AuthManager, verify_api_key_flexible
from ..core.logger import debug_logger
from ..core.model_resolver import get_base_model_aliases, resolve_model_name
from ..core.models import (
@@ -23,7 +23,6 @@ from ..core.models import (
)
from ..services.generation_handler import MODEL_CONFIG, GenerationHandler
from ..services.browser_captcha_extension import ExtensionCaptchaService
from fastapi import WebSocket, WebSocketDisconnect
router = APIRouter()
@@ -486,6 +485,7 @@ async def _collect_non_stream_result(
model: str,
prompt: str,
images: List[bytes],
base_url_override: Optional[str] = None,
video_media_id: Optional[str] = None,
) -> str:
handler = _ensure_generation_handler()
@@ -495,6 +495,7 @@ async def _collect_non_stream_result(
prompt=prompt,
images=images if images else None,
stream=False,
base_url_override=base_url_override,
video_media_id=video_media_id,
):
result = chunk
@@ -723,6 +724,7 @@ async def _iterate_openai_stream(
prompt=normalized.prompt,
images=normalized.images if normalized.images else None,
stream=True,
base_url_override=base_url_override,
video_media_id=normalized.video_media_id,
):
if chunk.startswith("data: "):
@@ -746,6 +748,7 @@ async def _iterate_gemini_stream(
prompt=normalized.prompt,
images=normalized.images if normalized.images else None,
stream=True,
base_url_override=base_url_override,
video_media_id=normalized.video_media_id,
):
if chunk.startswith("data: "):
@@ -874,6 +877,7 @@ async def create_chat_completion(
normalized.model,
normalized.prompt,
normalized.images,
base_url_override=request_base_url,
video_media_id=normalized.video_media_id,
)
)
@@ -907,7 +911,8 @@ async def generate_content(
normalized.model,
normalized.prompt,
normalized.images,
request_base_url,
base_url_override=request_base_url,
video_media_id=normalized.video_media_id,
)
)
)
@@ -970,6 +975,20 @@ async def stream_generate_content(
@router.websocket("/captcha_ws")
async def captcha_websocket_endpoint(websocket: WebSocket):
from ..core.logger import debug_logger
api_key = (
websocket.query_params.get("key")
or websocket.query_params.get("api_key")
or websocket.headers.get("x-goog-api-key")
or ""
).strip()
authorization = (websocket.headers.get("authorization") or "").strip()
if authorization.lower().startswith("bearer "):
api_key = authorization[7:].strip()
if not api_key or not AuthManager.verify_api_key(api_key):
await websocket.close(code=1008)
return
service = await ExtensionCaptchaService.get_instance()
await service.connect(websocket)
try:

7
src/core/config.py
View File

@@ -12,8 +12,11 @@ class Config:
self._admin_password: Optional[str] = None
def _load_config(self) -> Dict[str, Any]:
"""Load configuration from setting.toml"""
config_path = Path(__file__).parent.parent.parent / "config" / "setting.toml"
"""Load configuration from setting.toml, falling back to the example file."""
config_dir = Path(__file__).parent.parent.parent / "config"
config_path = config_dir / "setting.toml"
if not config_path.exists():
config_path = config_dir / "setting_example.toml"
with open(config_path, "rb") as f:
return tomli.load(f)

10
src/core/database.py
View File

@@ -408,6 +408,7 @@ class Database:
("image_concurrency", "INTEGER DEFAULT -1"),
("video_concurrency", "INTEGER DEFAULT -1"),
("captcha_proxy_url", "TEXT"), # token级打码代理
("extension_route_key", "TEXT"), # extension 模式路由键
("ban_reason", "TEXT"), # 禁用原因
("banned_at", "TIMESTAMP"), # 禁用时间
]
@@ -567,6 +568,7 @@ class Database:
image_concurrency INTEGER DEFAULT -1,
video_concurrency INTEGER DEFAULT -1,
captcha_proxy_url TEXT,
extension_route_key TEXT,
ban_reason TEXT,
banned_at TIMESTAMP
)
@@ -845,13 +847,15 @@ class Database:
cursor = await db.execute("""
INSERT INTO tokens (st, at, at_expires, email, name, remark, is_active,
credits, user_paygate_tier, current_project_id, current_project_name,
image_enabled, video_enabled, image_concurrency, video_concurrency, captcha_proxy_url)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
image_enabled, video_enabled, image_concurrency, video_concurrency,
captcha_proxy_url, extension_route_key)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
""", (token.st, token.at, token.at_expires, token.email, token.name, token.remark,
token.is_active, token.credits, token.user_paygate_tier,
token.current_project_id, token.current_project_name,
token.image_enabled, token.video_enabled,
token.image_concurrency, token.video_concurrency, token.captcha_proxy_url))
token.image_concurrency, token.video_concurrency,
token.captcha_proxy_url, token.extension_route_key))
await db.commit()
token_id = cursor.lastrowid

1
src/core/models.py
View File

@@ -42,6 +42,7 @@ class Token(BaseModel):
# 打码代理（token 级，可覆盖全局浏览器打码代理）
captcha_proxy_url: Optional[str] = None
extension_route_key: Optional[str] = None
# 429禁用相关
ban_reason: Optional[str] = None # 禁用原因: "429_rate_limit" 或 None

17
src/services/browser_captcha_extension.py
View File

@@ -1,6 +1,7 @@
import asyncio
import json
import time
import uuid
from dataclasses import dataclass, field
from typing import Any, Dict, Optional
@@ -24,7 +25,7 @@ class ExtensionCaptchaService:
def __init__(self, db=None):
self.db = db
self.active_connections: list[ExtensionConnection] = []
self.pending_requests: dict[str, asyncio.Future] = {}
self.pending_requests: dict[str, tuple[asyncio.Future, WebSocket]] = {}
@classmethod
async def get_instance(cls, db=None) -> "ExtensionCaptchaService":
@@ -143,7 +144,10 @@ class ExtensionCaptchaService:
req_id = payload.get("req_id")
if req_id and req_id in self.pending_requests:
future = self.pending_requests[req_id]
future, owner_websocket = self.pending_requests[req_id]
if websocket is not owner_websocket:
debug_logger.log_warning(f"[Extension Captcha] Ignoring response from non-owner connection: {req_id}")
return
if not future.done():
future.set_result(payload)
except Exception as e:
@@ -169,9 +173,9 @@ class ExtensionCaptchaService:
f"Available route keys: {available}"
)
req_id = f"req_{int(time.time() * 1000)}_{id(self)}"
future = asyncio.Future()
self.pending_requests[req_id] = future
req_id = f"req_{uuid.uuid4().hex}"
future = asyncio.get_running_loop().create_future()
self.pending_requests[req_id] = (future, conn.websocket)
request_data = {
"type": "get_token",
@@ -194,16 +198,13 @@ class ExtensionCaptchaService:
error_msg = result.get("error")
debug_logger.log_error(f"[Extension Captcha] Error from extension: {error_msg}")
print(f"!!! EXTENSION ERROR: {error_msg} !!!")
return None
except asyncio.TimeoutError:
debug_logger.log_error(f"[Extension Captcha] Timeout waiting for token (req_id: {req_id})")
print("!!! EXTENSION TIMEOUT !!!")
return None
except Exception as e:
debug_logger.log_error(f"[Extension Captcha] Communication error: {e}")
print(f"!!! EXTENSION COMM ERROR: {e} !!!")
return None
finally:
self.pending_requests.pop(req_id, None)

62
src/services/flow_client.py
View File

@@ -213,14 +213,6 @@ class FlowClient:
headers["sec-ch-ua-platform"] = "\"Windows\""
headers["sec-ch-ua-mobile"] = "?0"
if 'aisandbox' in url:
print(f"[DEBUG-DEEP] API REQUEST to: {url[:80]}")
print(f"[DEBUG-DEEP] fingerprint: {fingerprint}")
print(f"[DEBUG-DEEP] proxy_url: {proxy_url}")
print(f"[DEBUG-DEEP] UA: {headers.get('User-Agent', '')[:100]}")
print(f"[DEBUG-DEEP] sec-ch-ua-platform: {headers.get('sec-ch-ua-platform', 'NOT SET')}")
print(f"[DEBUG-DEEP] sec-ch-ua-mobile: {headers.get('sec-ch-ua-mobile', 'NOT SET')}")
# Log request
if config.debug_enabled:
if isinstance(fingerprint, dict):
@@ -1946,7 +1938,6 @@ class FlowClient:
]
}
print(f"[CONCAT] 提交拼接: original={original_media_id}, extend={extend_media_id}", flush=True)
debug_logger.log_info(f"[CONCAT] 提交拼接任务: original={original_media_id[:12]}..., extend={extend_media_id[:12]}...")
result = await self._make_request(
@@ -1956,7 +1947,7 @@ class FlowClient:
use_at=True,
at_token=at
)
print(f"[CONCAT] 提交结果: {json.dumps(result, ensure_ascii=False)[:500]}", flush=True)
debug_logger.log_info(f"[CONCAT] 拼接任务已提交: {json.dumps(result, ensure_ascii=False)[:300]}")
return result
async def poll_concatenation_status(
@@ -2006,11 +1997,14 @@ class FlowClient:
ev_len = len(encoded_video) if encoded_video else 0
elapsed = int(time.time() - start_time)
all_keys = list(result.keys())
print(f"[CONCAT] 状态: {status}, outputUri: {'有:'+output_uri[:40] if output_uri else 'N/A'}, encodedVideo: {ev_len} chars, elapsed: {elapsed}s, keys: {all_keys}", flush=True)
debug_logger.log_info(
f"[CONCAT] 状态: {status}, outputUri={'yes' if output_uri else 'no'}, "
f"encodedVideo={ev_len} chars, elapsed={elapsed}s, keys={all_keys}"
)
# 优先检查 outputUri
if output_uri:
print(f"[CONCAT] 拼接完成 (outputUri): {output_uri[:120]}", flush=True)
debug_logger.log_info(f"[CONCAT] 拼接完成 (outputUri): {output_uri[:120]}")
return result
# Google API 返回 encodedVideo（base64 编码的 MP4）而不是 outputUri
@@ -2030,26 +2024,26 @@ class FlowClient:
# 构造 URL：FastAPI 挂载了 /tmp -> /app/tmp/
serve_url = f"/tmp/{video_filename}"
print(f"[CONCAT] 拼接完成 (encodedVideo): 保存 {len(video_bytes)} bytes -> {serve_url}", flush=True)
debug_logger.log_info(f"[CONCAT] 拼接完成 (encodedVideo): 保存 {len(video_bytes)} bytes -> {serve_url}")
result["outputUri"] = serve_url
result["local_file"] = save_path
return result
except Exception as e:
print(f"[CONCAT] 解码 encodedVideo 失败: {e}", flush=True)
debug_logger.log_error(f"[CONCAT] 解码 encodedVideo 失败: {e}")
raise Exception(f"解码拼接视频失败: {e}")
# SUCCESSFUL but neither outputUri nor encodedVideo
if "SUCCESSFUL" in status:
print(f"[CONCAT] SUCCESSFUL 但无 outputUri/encodedVideo，完整响应: {json.dumps(result, ensure_ascii=False)[:500]}", flush=True)
debug_logger.log_warning(f"[CONCAT] SUCCESSFUL 但无 outputUri/encodedVideo: {json.dumps(result, ensure_ascii=False)[:300]}")
if "FAILED" in status or "ERROR" in status:
print(f"[CONCAT] 失败: {status}, 响应: {json.dumps(result, ensure_ascii=False)[:300]}", flush=True)
debug_logger.log_error(f"[CONCAT] 失败: {status}, 响应: {json.dumps(result, ensure_ascii=False)[:300]}")
raise Exception(f"视频拼接失败: {status}")
await asyncio.sleep(poll_interval)
print(f"[CONCAT] 超时 ({timeout}s)，放弃拼接", flush=True)
debug_logger.log_error(f"[CONCAT] 超时 ({timeout}s)，放弃拼接")
raise Exception(f"视频拼接超时 ({timeout}s)")
# ========== 视频放大 (Video Upsampler) ==========
@@ -2662,13 +2656,12 @@ class FlowClient:
- 其他模式: browser_id 为 None
"""
captcha_method = config.captcha_method
print(f"[DEBUG] _get_recaptcha_token called: captcha_method={captcha_method}, project_id={project_id}, action={action}")
debug_logger.log_info(f"[reCAPTCHA] 开始获取 token: method={captcha_method}, project_id={project_id}, action={action}")
if captcha_method == "extension":
try:
from .browser_captcha_extension import ExtensionCaptchaService
service = await ExtensionCaptchaService.get_instance(self.db)
print(f"[DEBUG-EXTENSION] get_token requested, active_connections length: {len(service.active_connections)}")
extension_timeout = 45 if action == "VIDEO_GENERATION" else 25
token = await service.get_token(
project_id,
@@ -2680,7 +2673,6 @@ class FlowClient:
return token, None
except Exception as e:
debug_logger.log_error(f"[reCAPTCHA Extension] 错误: {str(e)}")
print(f"[DEBUG-EXTENSION] EXCEPTION IN get_token: {e}")
self._set_request_fingerprint(None)
return None, None
@@ -2695,8 +2687,6 @@ class FlowClient:
token = await service.get_token(project_id, action)
debug_logger.log_info(f"[reCAPTCHA] get_token 返回: {token[:50] if token else None}...")
fingerprint = service.get_last_fingerprint() if token else None
print(f"[DEBUG-DEEP] personal token obtained: {bool(token)}, token_prefix={str(token)[:40] if token else 'None'}")
print(f"[DEBUG-DEEP] personal fingerprint: {fingerprint}")
self._set_request_fingerprint(fingerprint if token else None)
return token, None
except RuntimeError as e:
@@ -2772,11 +2762,10 @@ class FlowClient:
proxy_url = await self.proxy_manager.get_request_proxy_url()
if proxy_url:
self._set_request_fingerprint({"proxy_url": proxy_url})
print(f"[DEBUG] API captcha using proxy: {proxy_url[:60]}...")
else:
self._set_request_fingerprint(None)
except Exception as e:
print(f"[DEBUG] Failed to get proxy for API captcha: {e}")
debug_logger.log_warning(f"[reCAPTCHA] Failed to get proxy for API captcha: {e}")
self._set_request_fingerprint(None)
else:
self._set_request_fingerprint(None)
@@ -2832,22 +2821,15 @@ class FlowClient:
if self.proxy_manager:
try:
proxy_url = await self.proxy_manager.get_request_proxy_url()
print(f"[DEBUG] Got proxy_url: {proxy_url[:60] if proxy_url else None}")
if proxy_url:
if proxy_url.startswith("socks5://"):
# curl_cffi 对 SOCKS5 使用 proxy 参数
proxy = proxy_url
print(f"[DEBUG] Using SOCKS5 proxy for CapSolver: {proxy[:60]}...")
else:
# HTTP/HTTPS 代理使用 proxies 字典
proxies = {"http": proxy_url, "https": proxy_url}
print(f"[DEBUG] Using HTTP proxy for CapSolver: {proxy_url[:60]}...")
except Exception as e:
print(f"[DEBUG] Failed to get proxy: {e}")
debug_logger.log_warning(f"[reCAPTCHA {method}] Failed to get proxy: {e}")
print(f"[DEBUG] CapSolver request: method={method}, project_id={project_id}, action={action}")
print(f"[DEBUG] CapSolver API Key: {client_key[:20]}...")
async with AsyncSession() as session:
create_url = f"{base_url}/createTask"
@@ -2861,21 +2843,15 @@ class FlowClient:
}
}
# 根据代理类型使用不同参数
print(f"[DEBUG] Sending createTask request to {create_url}")
if proxy:
print(f"[DEBUG] Using proxy param: {proxy[:60]}...")
result = await session.post(create_url, json=create_data, impersonate="chrome124", proxy=proxy)
else:
print(f"[DEBUG] Using proxies dict or no proxy")
result = await session.post(create_url, json=create_data, impersonate="chrome124", proxies=proxies)
print(f"[DEBUG] createTask response status: {result.status_code}")
debug_logger.log_info(f"[reCAPTCHA {method}] createTask response status: {result.status_code}")
result_json = result.json()
task_id = result_json.get('taskId')
print(f"[DEBUG] createTask response: {result_json}")
debug_logger.log_info(f"[reCAPTCHA {method}] created task_id: {task_id}, response: {result_json}")
if not task_id:
@@ -2896,29 +2872,21 @@ class FlowClient:
result = await session.post(get_url, json=get_data, impersonate="chrome124", proxies=proxies)
result_json = result.json()
print(f"[DEBUG] getTaskResult polling #{i+1}: status={result_json.get('status')}, response={result_json}")
debug_logger.log_info(f"[reCAPTCHA {method}] polling #{i+1}: {result_json}")
status = result_json.get('status')
if status == 'ready':
solution = result_json.get('solution', {})
response = solution.get('gRecaptchaResponse')
print(f"[DEBUG] Got solution: {solution}")
print(f"[DEBUG] Got gRecaptchaResponse: {response[:50] if response else None}")
if response:
print(f"[DEBUG] Token获取成功")
debug_logger.log_info(f"[reCAPTCHA {method}] Token获取成功")
return response
await asyncio.sleep(3)
print(f"[DEBUG] Timeout waiting for token after 40 retries")
debug_logger.log_error(f"[reCAPTCHA {method}] Timeout waiting for token")
return None
except Exception as e:
print(f"[DEBUG] Exception in _get_api_captcha_token: {type(e).__name__}: {str(e)}")
debug_logger.log_error(f"[reCAPTCHA {method}] error: {str(e)}")
import traceback
traceback.print_exc()
return None

30
src/services/generation_handler.py
View File

@@ -818,6 +818,7 @@ class GenerationHandler:
prompt: str,
images: Optional[List[bytes]] = None,
stream: bool = False,
base_url_override: Optional[str] = None,
video_media_id: Optional[str] = None,
) -> AsyncGenerator:
"""统一生成入口
@@ -1688,7 +1689,6 @@ class GenerationHandler:
debug_logger.log_info(f"[EXTEND] 续写视频: {video_media_id}")
if stream:
yield self._create_stream_chunk(f"视频续写任务提交中，源视频: {video_media_id[:8]}...\n")
print(f"[EXTEND-DEBUG] Calling generate_video_extend with video_media_id={video_media_id}")
result = await self.flow_client.generate_video_extend(
at=token.at,
project_id=project_id,
@@ -1755,7 +1755,17 @@ class GenerationHandler:
# 如果是 extend，传入源视频 media_id 用于后续拼接
extend_source_id = video_media_id if video_type == "extend" else None
async for chunk in self._poll_video_result(token, project_id, operations, stream, upsample_config, generation_result, request_log_state, extend_source_media_id=extend_source_id):
async for chunk in self._poll_video_result(
token,
project_id,
operations,
stream,
upsample_config,
generation_result,
response_state,
request_log_state,
extend_source_media_id=extend_source_id,
):
yield chunk
finally:
@@ -1769,6 +1779,7 @@ class GenerationHandler:
stream: bool,
upsample_config: Optional[Dict] = None,
generation_result: Optional[Dict[str, Any]] = None,
response_state: Optional[Dict[str, Any]] = None,
request_log_state: Optional[Dict[str, Any]] = None,
extend_source_media_id: Optional[str] = None,
) -> AsyncGenerator:
@@ -1826,7 +1837,6 @@ class GenerationHandler:
import re as _re
_uuid_match = _re.search(r'/video/([0-9a-f-]{36})', video_url or '')
video_media_id = _uuid_match.group(1) if _uuid_match else video_info.get("mediaGenerationId", "")
print(f"[POLL-DEBUG] fifeUrl={str(video_url)[:80]}, video_media_id={video_media_id}")
aspect_ratio = video_info.get("aspectRatio", "VIDEO_ASPECT_RATIO_LANDSCAPE")
if not video_url:
@@ -1862,7 +1872,14 @@ class GenerationHandler:
# 递归轮询放大结果（不再放大）
async for chunk in self._poll_video_result(
token, project_id, upsample_operations, stream, None, generation_result, response_state, request_log_state
token,
project_id,
upsample_operations,
stream,
None,
generation_result,
response_state,
request_log_state,
):
yield chunk
return
@@ -1875,12 +1892,10 @@ class GenerationHandler:
yield self._create_stream_chunk(f"⚠️ 放大失败: {str(e)}，返回原始视频\n")
# ========== Extend 视频拼接 ==========
print(f"[CONCAT-DEBUG] extend_source_media_id={extend_source_media_id}, video_media_id={video_media_id}", flush=True)
if extend_source_media_id and video_media_id:
try:
if stream:
yield self._create_stream_chunk("\n视频续写完成，正在拼接完整视频...\n")
print(f"[CONCAT] 开始拼接: original={extend_source_media_id[:40]}..., extend={video_media_id[:40]}...", flush=True)
debug_logger.log_info(f"[CONCAT] 开始拼接: original={extend_source_media_id[:12]}..., extend={video_media_id[:12]}...")
# 提交拼接任务
@@ -1926,9 +1941,8 @@ class GenerationHandler:
yield self._create_stream_chunk("⚠️ 拼接任务创建失败，返回续写片段\n")
except Exception as e:
import traceback
print(f"[CONCAT] ❌ 拼接异常: {str(e)}", flush=True)
print(f"[CONCAT] traceback: {traceback.format_exc()}", flush=True)
debug_logger.log_error(f"[CONCAT] 拼接失败: {str(e)}")
debug_logger.log_error(f"[CONCAT] traceback: {traceback.format_exc()}")
if stream:
yield self._create_stream_chunk(f"⚠️ 拼接失败: {str(e)}，返回续写片段\n")
# 拼接失败不影响返回，继续使用 extend 片段的 URL

11
src/services/token_manager.py
View File

@@ -209,7 +209,8 @@ class TokenManager:
video_enabled: bool = True,
image_concurrency: int = -1,
video_concurrency: int = -1,
captcha_proxy_url: Optional[str] = None
captcha_proxy_url: Optional[str] = None,
extension_route_key: Optional[str] = None,
) -> Token:
"""Add a new token and prepare its pooled projects."""
existing_token = await self.db.get_token_by_st(st)
@@ -284,7 +285,8 @@ class TokenManager:
video_enabled=video_enabled,
image_concurrency=image_concurrency,
video_concurrency=video_concurrency,
captcha_proxy_url=captcha_proxy_url
captcha_proxy_url=captcha_proxy_url,
extension_route_key=extension_route_key,
)
token_id = await self.db.add_token(token)
@@ -314,7 +316,8 @@ class TokenManager:
video_enabled: Optional[bool] = None,
image_concurrency: Optional[int] = None,
video_concurrency: Optional[int] = None,
captcha_proxy_url: Optional[str] = None
captcha_proxy_url: Optional[str] = None,
extension_route_key: Optional[str] = None,
):
"""Update token (支持修改project_id和project_name)
@@ -344,6 +347,8 @@ class TokenManager:
update_fields["video_concurrency"] = video_concurrency
if captcha_proxy_url is not None:
update_fields["captcha_proxy_url"] = captcha_proxy_url
if extension_route_key is not None:
update_fields["extension_route_key"] = extension_route_key
# 检查token是否因429被禁用，如果是且未过期，则清空429状态
token = await self.db.get_token(token_id)

135
test.py
View File

@@ -1,135 +0,0 @@
import base64
import json
import re
import urllib.request
import urllib.error
from pathlib import Path
from typing import Optional
# ── 配置 ──────────────────────────────────────────────
IMAGE_PATH = Path(__file__).parent / "cat.jpg"
API_URL = "http://localhost:8000/v1/chat/completions"
API_KEY = "han1234"
MODEL = "gemini-3.1-flash-image-landscape"
PROMPT = "将这张图片变成水彩画风格"
OUTPUT = Path(__file__).parent / "cat_watercolor.jpg"
# ───────────────────────────────────────────────────────
def read_image_base64(path: Path) -> str:
with open(path, "rb") as f:
return base64.b64encode(f.read()).decode()
def call_api(b64_image: str) -> str:
"""发送请求，收集所有 SSE chunk，返回拼接后的完整 content。"""
payload = json.dumps({
"model": MODEL,
"stream": True,
"messages": [
{
"role": "user",
"content": [
{"type": "text", "text": PROMPT},
{"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{b64_image}"}}
]
}
]
}).encode()
req = urllib.request.Request(
API_URL,
data=payload,
headers={
"Authorization": f"Bearer {API_KEY}",
"Content-Type": "application/json",
},
method="POST",
)
content = ""
reasoning = ""
with urllib.request.urlopen(req, timeout=300) as resp:
for raw_line in resp:
line = raw_line.decode("utf-8", errors="replace").strip()
if not line.startswith("data:"):
continue
data_str = line[5:].strip()
if data_str == "[DONE]":
break
try:
chunk = json.loads(data_str)
except json.JSONDecodeError:
continue
# 检查错误
if "error" in chunk:
raise RuntimeError(chunk["error"].get("message", str(chunk["error"])))
delta = chunk.get("choices", [{}])[0].get("delta", {})
if delta.get("reasoning_content"):
r = delta["reasoning_content"]
reasoning += r
print(r, end="", flush=True)
if delta.get("content"):
content += delta["content"]
print() # 换行
return content
def extract_image_url(content: str) -> Optional[str]:
"""从 markdown 或纯文本中提取图片 URL。"""
# ![...](url) 或直接 URL
patterns = [
r"!\[.*?\]\((https?://\S+?)\)",
r"(https?://\S+\.(?:jpg|jpeg|png|webp|gif)(?:\?\S*)?)",
r"(https?://\S+)",
]
for pat in patterns:
m = re.search(pat, content)
if m:
return m.group(1)
return None
def download_image(url: str, dest: Path):
print(f"📥 下载图片: {url}")
urllib.request.urlretrieve(url, dest)
print(f"✅ 已保存到: {dest}")
def main():
if not IMAGE_PATH.exists():
raise FileNotFoundError(f"找不到输入图片: {IMAGE_PATH}")
print(f"📷 读取图片: {IMAGE_PATH}")
b64 = read_image_base64(IMAGE_PATH)
print(f"🚀 发送请求到 {API_URL} ...")
print("-" * 50)
content = call_api(b64)
print("-" * 50)
if not content:
print("⚠️ 响应 content 为空，可能图片已在 reasoning_content 中描述，请查看上方日志。")
return
print(f"\n📝 响应内容:\n{content}\n")
url = extract_image_url(content)
if url:
download_image(url, OUTPUT)
else:
# 尝试 base64 内嵌
b64_match = re.search(r"data:image/[^;]+;base64,([A-Za-z0-9+/=]+)", content)
if b64_match:
img_data = base64.b64decode(b64_match.group(1))
OUTPUT.write_bytes(img_data)
print(f"✅ base64 图片已保存到: {OUTPUT}")
else:
print("⚠️ 未能从响应中提取图片 URL，请手动查看上方 content。")
if __name__ == "__main__":
main()

4
tests/test_veo_lite_support.py
View File

@@ -62,7 +62,7 @@ class VeoLiteFlowClientTests(unittest.IsolatedAsyncioTestCase):
async def test_generate_video_text_uses_v2_payload_for_lite(self):
captured = {}
async def fake_make_request(method, url, json_data, use_at, at_token):
async def fake_make_request(method, url, json_data, use_at, at_token, **kwargs):
captured["url"] = url
captured["json_data"] = json_data
return {"operations": [{"operation": {"name": "task-1"}}]}
@@ -92,7 +92,7 @@ class VeoLiteFlowClientTests(unittest.IsolatedAsyncioTestCase):
async def test_generate_video_start_end_uses_v2_payload_for_interpolation_lite(self):
captured = {}
async def fake_make_request(method, url, json_data, use_at, at_token):
async def fake_make_request(method, url, json_data, use_at, at_token, **kwargs):
captured["url"] = url
captured["json_data"] = json_data
return {"operations": [{"operation": {"name": "task-2"}}]}