gardel/Mofox-Core
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

依旧修pyright喵喵喵~

Browse Source
This commit is contained in:
ikun-11451
2025-11-29 22:20:55 +08:00
parent 574c2384a2
commit acafc074b1
18 changed files with 106 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/chat/knowledge/embedding_store.py
View File

@@ -2,6 +2,7 @@ import asyncio
import math
import os
from dataclasses import dataclass
from typing import Any
# import tqdm
import aiofiles
@@ -121,7 +122,7 @@ class EmbeddingStore:
self.store = {}
self.faiss_index = None
self.faiss_index: Any = None
self.idx2hash = None
@staticmethod
@@ -158,6 +159,8 @@ class EmbeddingStore:
from src.config.config import model_config
from src.llm_models.utils_model import LLMRequest
assert model_config is not None
# 限制 chunk_size 和 max_workers 在合理范围内
chunk_size = max(MIN_CHUNK_SIZE, min(chunk_size, MAX_CHUNK_SIZE))
max_workers = max(MIN_WORKERS, min(max_workers, MAX_WORKERS))
@@ -402,6 +405,7 @@ class EmbeddingStore:
def build_faiss_index(self) -> None:
"""重新构建Faiss索引，以余弦相似度为度量"""
assert global_config is not None
# 获取所有的embedding
array = []
self.idx2hash = {}

5
src/chat/utils/utils_image.py
View File

@@ -344,14 +344,15 @@ class ImageManager:
# --- 新的帧选择逻辑：均匀抽取4帧 ---
num_frames = len(all_frames)
if num_frames <= 4:
# 如果总帧数小于等于4，则全部选中
# 如果总宽度小于等于4，则全部选中
selected_frames = all_frames
indices = list(range(num_frames))
else:
# 使用linspace计算4个均匀分布的索引
indices = np.linspace(0, num_frames - 1, 4, dtype=int)
selected_frames = [all_frames[i] for i in indices]
logger.debug(f"GIF Frame Analysis: Total frames={num_frames}, Selected indices={indices if num_frames > 4 else list(range(num_frames))}")
logger.debug(f"GIF Frame Analysis: Total frames={num_frames}, Selected indices={indices}")
# --- 帧选择逻辑结束 ---
# 如果选择后连一帧都没有（比如GIF只有一帧且后续处理失败？）或者原始GIF就没帧，也返回None

10
src/chat/utils/utils_video.py
View File

@@ -37,7 +37,7 @@ _locks_guard = asyncio.Lock()
logger = get_logger("utils_video")
from inkfox import video
from inkfox import video # type: ignore
class VideoAnalyzer:
@@ -123,7 +123,6 @@ class VideoAnalyzer:
# ---- 批量分析 ----
async def _analyze_batch(self, frames: list[tuple[str, float]], question: str | None) -> str:
from src.llm_models.payload_content.message import MessageBuilder, RoleType
from src.llm_models.utils_model import RequestType
prompt = self.batch_analysis_prompt.format(
personality_core=self.personality_core, personality_side=self.personality_side
@@ -139,12 +138,7 @@ class VideoAnalyzer:
for b64, _ in frames:
mb.add_image_content("jpeg", b64)
message = mb.build()
model_info, api_provider, client = self.video_llm._select_model()
resp = await self.video_llm._execute_request(
api_provider=api_provider,
client=client,
request_type=RequestType.RESPONSE,
model_info=model_info,
resp = await self.video_llm.execute_with_messages(
message_list=[message],
temperature=None,
max_tokens=None,

17
src/chat/utils/utils_video_legacy.py
View File

@@ -31,9 +31,9 @@ def _extract_frames_worker(
max_image_size: int,
frame_extraction_mode: str,
frame_interval_seconds: float | None,
) -> list[Any] | list[tuple[str, str]]:
) -> list[tuple[str, float]] | list[tuple[str, str]]:
"""线程池中提取视频帧的工作函数"""
frames = []
frames: list[tuple[str, float]] = []
try:
cap = cv2.VideoCapture(video_path)
fps = cap.get(cv2.CAP_PROP_FPS)
@@ -42,7 +42,7 @@ def _extract_frames_worker(
if frame_extraction_mode == "time_interval":
# 新模式：按时间间隔抽帧
time_interval = frame_interval_seconds
time_interval = frame_interval_seconds or 2.0
next_frame_time = 0.0
extracted_count = 0 # 初始化提取帧计数器
@@ -61,7 +61,7 @@ def _extract_frames_worker(
# 调整图像大小
if max(pil_image.size) > max_image_size:
ratio = max_image_size / max(pil_image.size)
new_size = tuple(int(dim * ratio) for dim in pil_image.size)
new_size = (int(pil_image.size[0] * ratio), int(pil_image.size[1] * ratio))
pil_image = pil_image.resize(new_size, Image.Resampling.LANCZOS)
# 转换为base64
@@ -240,6 +240,7 @@ class LegacyVideoAnalyzer:
estimated_frames = min(self.max_frames, total_frames // frame_interval + 1)
else:
estimated_frames = self.max_frames
frame_interval = 1
logger.info(f"计算得出帧间隔: {frame_interval} (将提取约{estimated_frames}帧)")
@@ -276,7 +277,7 @@ class LegacyVideoAnalyzer:
return await self._extract_frames_fallback(video_path)
logger.info(f"✅ 成功提取{len(frames)}帧 (线程池模式)")
return frames
return frames # type: ignore
except Exception as e:
logger.error(f"线程池帧提取失败: {e}")
@@ -315,7 +316,7 @@ class LegacyVideoAnalyzer:
# 调整图像大小
if max(pil_image.size) > self.max_image_size:
ratio = self.max_image_size / max(pil_image.size)
new_size = tuple(int(dim * ratio) for dim in pil_image.size)
new_size = (int(pil_image.size[0] * ratio), int(pil_image.size[1] * ratio))
pil_image = pil_image.resize(new_size, Image.Resampling.LANCZOS)
# 转换为base64
@@ -463,11 +464,11 @@ class LegacyVideoAnalyzer:
# logger.info(f"✅ 多帧消息构建完成，包含{len(frames)}张图片")
# 获取模型信息和客户端
model_info, api_provider, client = self.video_llm._select_model()
model_info, api_provider, client = self.video_llm._select_model() # type: ignore
# logger.info(f"使用模型: {model_info.name} 进行多帧分析")
# 直接执行多图片请求
api_response = await self.video_llm._execute_request(
api_response = await self.video_llm._execute_request( # type: ignore
api_provider=api_provider,
client=client,
request_type=RequestType.RESPONSE,