636 lines
30 KiB
Python
636 lines
30 KiB
Python
"""
|
||
VLM服务 - 重构版本
|
||
提供视觉模型的图像分析功能
|
||
继承OpenAICompatibleService以复用通用逻辑
|
||
"""
|
||
|
||
import time
|
||
import asyncio
|
||
from typing import Optional, Dict, Any, List, Callable
|
||
from .openai_base import OpenAICompatibleService
|
||
from .thinking_filter import postprocess_model_output
|
||
from ..utils.common import (
|
||
format_api_error, preprocess_image, check_multi_image_support, ProgressBar,
|
||
log_complete, log_error,
|
||
PREFIX, PROCESS_PREFIX, WARN_PREFIX, ERROR_PREFIX, format_elapsed_time,
|
||
TASK_IMAGE_CAPTION, TASK_VIDEO_CAPTION
|
||
)
|
||
from .thinking_control import build_thinking_suppression, should_append_no_thinking_instruction
|
||
|
||
|
||
class VisionService(OpenAICompatibleService):
|
||
"""
|
||
视觉模型服务
|
||
支持单图和多图分析
|
||
"""
|
||
|
||
@staticmethod
|
||
def _get_config() -> Dict[str, Any]:
|
||
"""获取视觉模型配置"""
|
||
from ..config_manager import config_manager
|
||
config = config_manager.get_vision_config()
|
||
current_provider = config.get('provider')
|
||
|
||
if 'providers' in config and current_provider in config['providers']:
|
||
provider_config = config['providers'][current_provider]
|
||
return {
|
||
'provider': current_provider,
|
||
'model': provider_config.get('model', ''),
|
||
'base_url': provider_config.get('base_url', ''),
|
||
'api_key': provider_config.get('api_key', ''),
|
||
'temperature': provider_config.get('temperature', 0.7),
|
||
'top_p': provider_config.get('top_p', 0.9),
|
||
'max_tokens': provider_config.get('max_tokens', 2000),
|
||
'auto_unload': provider_config.get('auto_unload', True)
|
||
}
|
||
else:
|
||
return config
|
||
|
||
@staticmethod
|
||
async def _call_ollama_native_vision(
|
||
model: str,
|
||
system_prompt: str,
|
||
images_b64: List[str],
|
||
temperature: float,
|
||
top_p: float,
|
||
max_tokens: int,
|
||
base_url: str,
|
||
stream_callback: Optional[Callable[[str], None]] = None,
|
||
request_id: Optional[str] = None,
|
||
is_multi: bool = False,
|
||
auto_unload: bool = True,
|
||
enable_advanced_params: bool = False,
|
||
thinking_extra: Optional[Dict[str, Any]] = None,
|
||
filter_thinking_output: bool = True,
|
||
cancel_event: Optional[Any] = None,
|
||
task_type: str = None,
|
||
source: str = None
|
||
) -> Dict[str, Any]:
|
||
"""
|
||
调用Ollama原生视觉API (/api/chat)
|
||
支持单图和多图分析
|
||
|
||
参数:
|
||
enable_advanced_params: 是否发送高级参数(temperature/top_p/num_predict)
|
||
thinking_extra: 思维链控制参数
|
||
"""
|
||
from ..server import is_streaming_progress_enabled
|
||
|
||
try:
|
||
start_time = time.perf_counter()
|
||
|
||
_thinking_extra = thinking_extra # 使用传入的参数
|
||
_thinking_tag = "💭" if _thinking_extra else ""
|
||
|
||
# 计算基准 URL (确保移除 /v1 和末尾斜杠)
|
||
native_base = base_url.rstrip('/') if base_url else 'http://localhost:11434'
|
||
if native_base.endswith('/v1'):
|
||
native_base = native_base[:-3].rstrip('/')
|
||
|
||
# 动态计算num_ctx（根据图像数量）
|
||
# 每张图片约需要1024-2048 tokens
|
||
img_count = len(images_b64)
|
||
|
||
# 文本Token估算 (0.6系数)
|
||
prompt_ctx = int(len(system_prompt) * 0.6)
|
||
|
||
# 图像Token估算 (每张2048作为基准)
|
||
image_ctx = img_count * 2048
|
||
|
||
# --- 智能预留策略 (适配 Vision 模型) ---
|
||
# 关键点：Vision模型的思考过程同样占用大量 Output Token
|
||
|
||
is_safe_standard_model = False
|
||
if model:
|
||
m = model.lower()
|
||
if "instruct" in m or "chat" in m:
|
||
is_safe_standard_model = True
|
||
|
||
if _thinking_extra or is_safe_standard_model:
|
||
# 已关闭思维链 OR 标准指令模型 -> 极致节省模式
|
||
min_output = 512
|
||
# 单图允许进一步下探至 2048，多图保持 3072 起步以确保稳定
|
||
ctx_floor = 2048 if not is_multi else 3072
|
||
sys_buffer = 384
|
||
else:
|
||
# 未关闭思维链 -> 安全能够模式
|
||
min_output = 1024
|
||
# 单图下限从 4096 降至 2048 (适配 Ollama 显存分配优化)
|
||
ctx_floor = 2048 if not is_multi else 4096
|
||
sys_buffer = 384 if not is_multi else 1024
|
||
|
||
# 输出预留 (多图需更多)
|
||
# 如果是单图模式，预留 512 已足够描述；如果是多图，使用 min_output
|
||
base_reserve = (img_count * 512) if is_multi else 512
|
||
output_reserve = max(512 if not is_multi else min_output, base_reserve)
|
||
|
||
required_ctx = prompt_ctx + image_ctx + output_reserve + sys_buffer
|
||
|
||
# 范围: [ctx_floor, 65536]
|
||
num_ctx = max(ctx_floor, min(65536, required_ctx))
|
||
num_ctx = ((num_ctx + 1023) // 1024) * 1024
|
||
|
||
# [Debug] 输出多图请求信息
|
||
print(f"{PREFIX} 🐏 视觉请求 | 图片数量:{len(images_b64)} | num_ctx:{num_ctx} | 模型:{model}")
|
||
|
||
# 构建基础请求体
|
||
payload = {
|
||
"model": model,
|
||
"messages": [{"role": "user", "content": system_prompt, "images": images_b64}],
|
||
"stream": True
|
||
}
|
||
|
||
# ---构建 options---
|
||
# 基础参数：num_ctx（动态上下文窗口大小）
|
||
options = {
|
||
"num_ctx": num_ctx
|
||
}
|
||
|
||
# 高级参数：仅在用户启用时发送
|
||
# 参数说明（基于 Ollama 官方文档）：
|
||
# - temperature: 控制随机性，默认0.8，值越低输出越稳定
|
||
# - top_p: 核采样，默认0.9，限制候选词概率范围
|
||
# - num_predict: 最大生成Token数，默认-1（无限）
|
||
if enable_advanced_params:
|
||
options["temperature"] = temperature
|
||
options["top_p"] = top_p
|
||
options["num_predict"] = max_tokens
|
||
|
||
payload["options"] = options
|
||
|
||
# 添加思维链控制参数（如 think: true 或 think: false）
|
||
if _thinking_extra:
|
||
payload.update(_thinking_extra)
|
||
|
||
# 设置超时
|
||
# 基础读取超时60秒 + 每张图片增加30秒 + 上下文长度自适应
|
||
base_read_timeout = 60.0
|
||
per_image_read_timeout = 30.0
|
||
ctx_based_timeout = (num_ctx / 1000) * 2.0 # 每1000tokens增加2秒
|
||
|
||
calculated_read_timeout = base_read_timeout + (img_count * per_image_read_timeout) + ctx_based_timeout
|
||
|
||
# 最大读取超时限制为 10 分钟 (600s)
|
||
final_read_timeout = min(600.0, max(60.0, calculated_read_timeout))
|
||
|
||
# 创建统一进度条（自动处理等待→生成→完成的完整生命周期）
|
||
extra_info = f"Context:{num_ctx} | Timeout:{int(final_read_timeout)}s"
|
||
pbar = ProgressBar(
|
||
request_id=request_id,
|
||
service_name="Ollama",
|
||
extra_info=extra_info,
|
||
streaming=is_streaming_progress_enabled(),
|
||
task_type=task_type,
|
||
source=source
|
||
)
|
||
|
||
start_time = time.perf_counter()
|
||
|
||
try:
|
||
from .ollama_native import OllamaNativeAdapter
|
||
return await OllamaNativeAdapter.stream_chat(
|
||
model=model,
|
||
native_base=native_base,
|
||
payload=payload,
|
||
timeout=final_read_timeout,
|
||
pbar=pbar,
|
||
stream_callback=stream_callback,
|
||
cancel_event=cancel_event,
|
||
provider_label="Ollama(Vision)",
|
||
include_reasoning=not filter_thinking_output,
|
||
)
|
||
finally:
|
||
try:
|
||
from .llm import LLMService
|
||
await LLMService._unload_ollama_model(model, {"base_url": native_base, "auto_unload": auto_unload})
|
||
except:
|
||
pass
|
||
|
||
# 关键修复：单独捕获外层 CancelledError，确保 pbar 被正确停止
|
||
except asyncio.CancelledError:
|
||
if 'pbar' in locals() and pbar:
|
||
pbar.cancel(f"{WARN_PREFIX} 任务被外部取消 | 服务:Ollama(Vision)")
|
||
return {"success": False, "error": "任务被取消", "interrupted": True}
|
||
|
||
except Exception as e:
|
||
# 关键修复：确保 pbar 在异常时也被停止
|
||
if 'pbar' in locals() and pbar:
|
||
pbar.error(format_api_error(e, "Ollama"))
|
||
return {"success": False, "error": format_api_error(e, "Ollama")}
|
||
|
||
@staticmethod
|
||
async def analyze_image(
|
||
image_data: str,
|
||
request_id: Optional[str] = None,
|
||
stream_callback: Optional[Callable[[str], None]] = None,
|
||
prompt_content: Optional[str] = None,
|
||
custom_provider: Optional[str] = None,
|
||
custom_provider_config: Optional[Dict[str, Any]] = None,
|
||
cancel_event: Optional[Any] = None,
|
||
task_type: str = None,
|
||
source: Optional[str] = None
|
||
) -> Dict[str, Any]:
|
||
"""
|
||
使用视觉模型分析单张图像
|
||
|
||
参数:
|
||
image_data: 图像数据（Base64编码）
|
||
request_id: 请求ID
|
||
stream_callback: 流式输出回调
|
||
prompt_content: 自定义提示词
|
||
custom_provider: 自定义服务商
|
||
custom_provider_config: 自定义配置
|
||
|
||
返回:
|
||
Dict: {"success": bool, "data": {"description": str}, "error": str}
|
||
"""
|
||
try:
|
||
# 获取配置
|
||
if custom_provider and custom_provider_config:
|
||
provider = custom_provider
|
||
api_key = custom_provider_config.get('api_key')
|
||
model = custom_provider_config.get('model')
|
||
temperature = custom_provider_config.get('temperature', 0.7)
|
||
top_p = custom_provider_config.get('top_p', 0.9)
|
||
max_tokens = custom_provider_config.get('max_tokens', 2000)
|
||
base_url = custom_provider_config.get('base_url', '')
|
||
else:
|
||
config = VisionService._get_config()
|
||
provider = config.get('provider', 'unknown')
|
||
api_key = config.get('api_key')
|
||
model = config.get('model')
|
||
temperature = config.get('temperature', 0.7)
|
||
top_p = config.get('top_p', 0.9)
|
||
max_tokens = config.get('max_tokens', 2000)
|
||
base_url = config.get('base_url', '')
|
||
|
||
# 注：允许空API Key，支持无认证服务商
|
||
if not model:
|
||
return {"success": False, "error": "未配置模型名称"}
|
||
|
||
provider_display_name = VisionService.get_provider_display_name(provider)
|
||
|
||
from ..utils.common import REQUEST_PREFIX, PREFIX, format_model_with_thinking
|
||
|
||
# 检查服务配置以确定是否显示思维链标识
|
||
from ..config_manager import config_manager
|
||
service = config_manager.get_service(provider)
|
||
disable_thinking_enabled = service.get('disable_thinking', True) if service else True
|
||
# 只有当开关开启且模型支持时才显示标识
|
||
_thinking_check = build_thinking_suppression(provider, model) if disable_thinking_enabled else None
|
||
thinking_disabled = _thinking_check is not None
|
||
model_display = format_model_with_thinking(model, thinking_disabled)
|
||
|
||
# 预处理图像
|
||
processed_image = preprocess_image(image_data, request_id=request_id)
|
||
|
||
# 获取系统提示词
|
||
system_prompt = prompt_content or "请详细描述这张图片的内容，包括主要对象、场景、颜色、氛围等。"
|
||
provider_type = service.get('type', provider) if service else provider
|
||
if should_append_no_thinking_instruction(provider_type, model, disable_thinking_enabled):
|
||
system_prompt += " 请直接输出结果，不要包含任何思考过程、推理过程或 <think> 标签。"
|
||
|
||
# Ollama走原生API：/v1 地址保持 OpenAI-compatible 路径
|
||
is_native_ollama = False
|
||
if service and service.get('type') == 'ollama':
|
||
_url = base_url.rstrip('/') if base_url else ''
|
||
if not _url.endswith('/v1') and '/v1/' not in _url:
|
||
is_native_ollama = True
|
||
|
||
if is_native_ollama:
|
||
# 读取 Ollama 服务的配置
|
||
enable_advanced_params = service.get('enable_advanced_params', False)
|
||
filter_thinking_output = service.get('filter_thinking_output', True)
|
||
effective_filter_thinking_output = filter_thinking_output or disable_thinking_enabled
|
||
_ollama_thinking_extra = build_thinking_suppression(service.get('type', provider), model) if disable_thinking_enabled else None
|
||
|
||
# 提取纯base64
|
||
b64 = processed_image.split(',')[1] if ',' in processed_image else processed_image
|
||
|
||
# 提前计算auto_unload配置
|
||
native_base = base_url[:-3] if base_url and base_url.endswith('/v1') else (base_url or 'http://localhost:11434')
|
||
native_base = native_base.rstrip('/')
|
||
_cfg = {
|
||
'auto_unload': custom_provider_config.get('auto_unload', True) if custom_provider_config else config.get('auto_unload', True),
|
||
'base_url': native_base
|
||
}
|
||
auto_unload = _cfg['auto_unload']
|
||
|
||
result = await VisionService._call_ollama_native_vision(
|
||
model=model,
|
||
system_prompt=system_prompt,
|
||
images_b64=[b64],
|
||
temperature=temperature,
|
||
top_p=top_p,
|
||
max_tokens=max_tokens,
|
||
base_url=base_url,
|
||
stream_callback=stream_callback,
|
||
request_id=request_id,
|
||
is_multi=False,
|
||
auto_unload=auto_unload,
|
||
enable_advanced_params=enable_advanced_params,
|
||
thinking_extra=_ollama_thinking_extra,
|
||
filter_thinking_output=effective_filter_thinking_output,
|
||
cancel_event=cancel_event,
|
||
task_type=task_type or TASK_IMAGE_CAPTION,
|
||
source=source
|
||
)
|
||
|
||
if result["success"]:
|
||
success, content = postprocess_model_output(
|
||
result["content"],
|
||
filter_thinking_output=effective_filter_thinking_output,
|
||
)
|
||
if not success:
|
||
return {"success": False, "error": "API returned empty result after filtering reasoning content (Model only output thinking process)"}
|
||
|
||
return {
|
||
"success": True,
|
||
"data": {"description": content}
|
||
}
|
||
else:
|
||
return result
|
||
|
||
# 其他服务走HTTP直连
|
||
if not base_url:
|
||
base_url = VisionService.get_provider_base_url(provider, custom_provider_config if custom_provider else None)
|
||
|
||
# 构建消息（图像格式）
|
||
# 关键修复(BUG-01): system_prompt 独立作为 system role 发送
|
||
# Zhipu GLM-4V 等严格服务商要求 system/user role 严格分离
|
||
# user content array 仅包含简短的任务触发词和图片 URL
|
||
messages = []
|
||
if system_prompt:
|
||
messages.append({"role": "system", "content": system_prompt})
|
||
messages.append({
|
||
"role": "user",
|
||
"content": [
|
||
{"type": "text", "text": "Please analyze this image."},
|
||
{"type": "image_url", "image_url": {"url": processed_image}}
|
||
]
|
||
})
|
||
|
||
# 检查disable_thinking、enable_advanced_params和filter_thinking_output配置
|
||
from ..config_manager import config_manager
|
||
service = config_manager.get_service(provider)
|
||
disable_thinking_enabled = service.get('disable_thinking', True) if service else True
|
||
enable_advanced_params = service.get('enable_advanced_params', False) if service else False
|
||
filter_thinking_output = service.get('filter_thinking_output', True) if service else True
|
||
effective_filter_thinking_output = filter_thinking_output or disable_thinking_enabled
|
||
thinking_extra = build_thinking_suppression(provider, model) if disable_thinking_enabled else None
|
||
|
||
result = await VisionService._http_request_chat_completions(
|
||
base_url=base_url,
|
||
api_key=api_key,
|
||
model=model,
|
||
messages=messages,
|
||
temperature=temperature,
|
||
top_p=top_p,
|
||
max_tokens=max_tokens,
|
||
thinking_extra=thinking_extra,
|
||
enable_advanced_params=enable_advanced_params,
|
||
stream_callback=stream_callback,
|
||
request_id=request_id,
|
||
provider_display_name=provider_display_name,
|
||
cancel_event=cancel_event,
|
||
task_type=task_type or TASK_IMAGE_CAPTION,
|
||
source=source,
|
||
filter_thinking_output=effective_filter_thinking_output
|
||
)
|
||
|
||
if result["success"]:
|
||
success, content = postprocess_model_output(
|
||
result["content"],
|
||
filter_thinking_output=effective_filter_thinking_output,
|
||
)
|
||
if not success:
|
||
return {"success": False, "error": "API returned empty result after filtering reasoning content (Model only output thinking process)"}
|
||
return {
|
||
"success": True,
|
||
"data": {"description": content}
|
||
}
|
||
else:
|
||
return result
|
||
|
||
except Exception as e:
|
||
return {"success": False, "error": format_api_error(e, "VLM服务")}
|
||
|
||
@staticmethod
|
||
async def analyze_images(
|
||
images_data: List[str],
|
||
request_id: Optional[str] = None,
|
||
stream_callback: Optional[Callable[[str], None]] = None,
|
||
prompt_content: Optional[str] = None,
|
||
custom_provider: Optional[str] = None,
|
||
custom_provider_config: Optional[Dict[str, Any]] = None,
|
||
cancel_event: Optional[Any] = None,
|
||
task_type: str = None,
|
||
source: Optional[str] = None
|
||
) -> Dict[str, Any]:
|
||
"""
|
||
使用视觉模型分析多张图像
|
||
|
||
参数:
|
||
images_data: 图像数据列表（Base64编码）
|
||
request_id: 请求ID
|
||
stream_callback: 流式输出回调
|
||
prompt_content: 自定义提示词
|
||
custom_provider: 自定义服务商
|
||
custom_provider_config: 自定义配置
|
||
|
||
返回:
|
||
Dict: {"success": bool, "data": {"description": str}, "error": str}
|
||
"""
|
||
try:
|
||
# 获取配置
|
||
if custom_provider and custom_provider_config:
|
||
provider = custom_provider
|
||
api_key = custom_provider_config.get('api_key')
|
||
model = custom_provider_config.get('model')
|
||
temperature = custom_provider_config.get('temperature', 0.7)
|
||
top_p = custom_provider_config.get('top_p', 0.9)
|
||
max_tokens = custom_provider_config.get('max_tokens', 2000)
|
||
base_url = custom_provider_config.get('base_url', '')
|
||
else:
|
||
config = VisionService._get_config()
|
||
provider = config.get('provider', 'unknown')
|
||
api_key = config.get('api_key')
|
||
model = config.get('model')
|
||
temperature = config.get('temperature', 0.7)
|
||
top_p = config.get('top_p', 0.9)
|
||
max_tokens = config.get('max_tokens', 2000)
|
||
base_url = config.get('base_url', '')
|
||
|
||
# 注：允许空API Key，支持无认证服务商
|
||
if not model:
|
||
return {"success": False, "error": "未配置模型名称"}
|
||
|
||
provider_display_name = VisionService.get_provider_display_name(provider)
|
||
|
||
from ..utils.common import REQUEST_PREFIX, PREFIX, format_model_with_thinking
|
||
|
||
# 检查服务配置以确定是否显示思维链标识
|
||
from ..config_manager import config_manager
|
||
service = config_manager.get_service(provider)
|
||
disable_thinking_enabled = service.get('disable_thinking', True) if service else True
|
||
# 只有当开关开启且模型支持时才显示标识
|
||
_thinking_check = build_thinking_suppression(provider, model) if disable_thinking_enabled else None
|
||
thinking_disabled = _thinking_check is not None
|
||
model_display = format_model_with_thinking(model, thinking_disabled)
|
||
|
||
# 智能推断上限（节点层已做截断，此处作为服务层最后防线，静默处理）
|
||
from ..utils.common import get_model_max_images
|
||
max_images = get_model_max_images(model)
|
||
if len(images_data) > max_images:
|
||
images_data = images_data[:max_images]
|
||
|
||
# 预处理所有图像（智能压缩：根据图像数量动态调整质量）
|
||
img_count = len(images_data)
|
||
from ..utils.common import get_optimal_image_params
|
||
_, _, compression_level = get_optimal_image_params(img_count)
|
||
|
||
# 使用 ProgressBar 管理预处理进度
|
||
pbar = ProgressBar(request_id=request_id, service_name="图像预处理", streaming=False)
|
||
processed_images = []
|
||
for idx, img in enumerate(images_data, 1):
|
||
processed = preprocess_image(img, request_id=request_id, silent=True, image_count=img_count)
|
||
processed_images.append(processed)
|
||
|
||
pbar.done(f"{PREFIX} 🟡 预处理完成: {img_count}/{img_count} | 压缩:{compression_level}")
|
||
|
||
# 获取系统提示词
|
||
system_prompt = prompt_content or "请详细描述这些图片，分析它们之间的关系和差异。"
|
||
provider_type = service.get('type', provider) if service else provider
|
||
if should_append_no_thinking_instruction(provider_type, model, disable_thinking_enabled):
|
||
system_prompt += " 请直接输出结果，不要包含任何思考过程、推理过程或 <think> 标签。"
|
||
|
||
# 判断是否走原生 Ollama API：必须是 ollama 类型，且 base_url 不以 /v1 结尾或包含 /v1/
|
||
is_native_ollama = False
|
||
if service and service.get('type') == 'ollama':
|
||
# 兼容 "http://xxx:11434/v1/" 或 "http://xxx:11434/v1"
|
||
_url = base_url.rstrip('/')
|
||
if not _url.endswith('/v1') and '/v1/' not in base_url:
|
||
is_native_ollama = True
|
||
|
||
# Ollama走原生API
|
||
if is_native_ollama:
|
||
# 读取 Ollama 服务的配置
|
||
from ..config_manager import config_manager
|
||
# 此处保持类型判断，不再硬编码 ID 'ollama'
|
||
disable_thinking_enabled = service.get('disable_thinking', True)
|
||
enable_advanced_params = service.get('enable_advanced_params', False)
|
||
filter_thinking_output = service.get('filter_thinking_output', True)
|
||
effective_filter_thinking_output = filter_thinking_output or disable_thinking_enabled
|
||
_ollama_thinking_extra = build_thinking_suppression(service.get('type', provider), model) if disable_thinking_enabled else None
|
||
|
||
# 提前计算auto_unload配置
|
||
native_base = base_url[:-3] if base_url.endswith('/v1') else (base_url or 'http://localhost:11434')
|
||
native_base = native_base.rstrip('/')
|
||
_cfg = {
|
||
'auto_unload': custom_provider_config.get('auto_unload', True) if custom_provider_config else config.get('auto_unload', True),
|
||
'base_url': native_base
|
||
}
|
||
auto_unload = _cfg['auto_unload']
|
||
|
||
# 提取纯base64
|
||
b64_images = [img.split(',')[1] if ',' in img else img for img in processed_images]
|
||
|
||
result = await VisionService._call_ollama_native_vision(
|
||
model=model,
|
||
system_prompt=system_prompt,
|
||
images_b64=b64_images,
|
||
temperature=temperature,
|
||
top_p=top_p,
|
||
max_tokens=max_tokens,
|
||
base_url=base_url,
|
||
stream_callback=stream_callback,
|
||
request_id=request_id,
|
||
is_multi=True,
|
||
auto_unload=auto_unload,
|
||
enable_advanced_params=enable_advanced_params,
|
||
thinking_extra=_ollama_thinking_extra,
|
||
filter_thinking_output=effective_filter_thinking_output,
|
||
cancel_event=cancel_event,
|
||
task_type=task_type or TASK_VIDEO_CAPTION,
|
||
source=source
|
||
)
|
||
|
||
if result["success"]:
|
||
success, content = postprocess_model_output(
|
||
result["content"],
|
||
filter_thinking_output=effective_filter_thinking_output,
|
||
)
|
||
if not success:
|
||
return {"success": False, "error": "API returned empty result after filtering reasoning content (Model only output thinking process)"}
|
||
|
||
return {
|
||
"success": True,
|
||
"data": {"description": content}
|
||
}
|
||
else:
|
||
return result
|
||
|
||
# 其他服务走HTTP直连
|
||
if not base_url:
|
||
base_url = VisionService.get_provider_base_url(provider, custom_provider_config if custom_provider else None)
|
||
|
||
# 构建多图消息
|
||
# 关键修复(BUG-01): system_prompt 独立作为 system role 发送
|
||
# 多图 user content array 仅包含简短指令和所有图片 URL
|
||
messages = []
|
||
if system_prompt:
|
||
messages.append({"role": "system", "content": system_prompt})
|
||
|
||
multi_content = [{"type": "text", "text": "Please analyze these images."}]
|
||
for img in processed_images:
|
||
multi_content.append({"type": "image_url", "image_url": {"url": img}})
|
||
messages.append({"role": "user", "content": multi_content})
|
||
|
||
# 检查disable_thinking、enable_advanced_params和filter_thinking_output配置
|
||
from ..config_manager import config_manager
|
||
service = config_manager.get_service(provider)
|
||
disable_thinking_enabled = service.get('disable_thinking', True) if service else True
|
||
enable_advanced_params = service.get('enable_advanced_params', False) if service else False
|
||
filter_thinking_output = service.get('filter_thinking_output', True) if service else True
|
||
effective_filter_thinking_output = filter_thinking_output or disable_thinking_enabled
|
||
thinking_extra = build_thinking_suppression(provider, model) if disable_thinking_enabled else None
|
||
|
||
result = await VisionService._http_request_chat_completions(
|
||
base_url=base_url,
|
||
api_key=api_key,
|
||
model=model,
|
||
messages=messages,
|
||
temperature=temperature,
|
||
top_p=top_p,
|
||
max_tokens=max_tokens,
|
||
thinking_extra=thinking_extra,
|
||
enable_advanced_params=enable_advanced_params,
|
||
stream_callback=stream_callback,
|
||
request_id=request_id,
|
||
provider_display_name=provider_display_name,
|
||
cancel_event=cancel_event,
|
||
task_type=task_type or TASK_VIDEO_CAPTION,
|
||
source=source,
|
||
filter_thinking_output=effective_filter_thinking_output
|
||
)
|
||
|
||
if result["success"]:
|
||
success, content = postprocess_model_output(
|
||
result["content"],
|
||
filter_thinking_output=effective_filter_thinking_output,
|
||
)
|
||
if not success:
|
||
return {"success": False, "error": "API returned empty result after filtering reasoning content (Model only output thinking process)"}
|
||
return {
|
||
"success": True,
|
||
"data": {"description": content}
|
||
}
|
||
else:
|
||
return result
|
||
|
||
except Exception as e:
|
||
# 确保进度条在异常时被停止
|
||
if 'pbar' in locals() and pbar and not getattr(pbar, '_closed', False):
|
||
pbar.error(format_api_error(e, "VLM服务"))
|
||
return {"success": False, "error": format_api_error(e, "VLM服务")}
|