80 lines
2.3 KiB
YAML
80 lines
2.3 KiB
YAML
# 环境配置
|
||
environment: "development"
|
||
|
||
server:
|
||
port: 8087
|
||
host: "0.0.0.0"
|
||
|
||
log:
|
||
level: "info"
|
||
path: "logs"
|
||
# The number of log files to retain (default: 20)
|
||
retain_days: 20
|
||
|
||
# 文档解析配置
|
||
document_parser:
|
||
# 并发控制
|
||
max_concurrent: 5
|
||
queue_size: 1000
|
||
|
||
# 超时设置（统一超时配置）
|
||
download_timeout: 3600
|
||
processing_timeout: 3600 # 60分钟，用于MinerU和MarkItDown解析超时
|
||
|
||
# MinerU配置
|
||
mineru:
|
||
# 后端选择：pipeline|vlm-transformers|vlm-sglang-engine|vlm-sglang-client, 有cuda环境目前推荐: vlm-transformers
|
||
backend: "pipeline"
|
||
python_path: "./venv/bin/python" # 默认使用虚拟环境，如不存在则自动检测系统Python
|
||
# 这个是mineru的并发数，目前没用,占位使用,只需要小于document_parser.max_concurrent 即可
|
||
max_concurrent: 1
|
||
queue_size: 100
|
||
# timeout 已移除，统一使用 document_parser.processing_timeout (3600秒)
|
||
batch_size: 1
|
||
quality_level: "Balanced"
|
||
# 单进程最大GPU显存占用(GB)，仅对pipeline后端,这个是一个软上限
|
||
vram: 8
|
||
|
||
# MarkItDown配置
|
||
markitdown:
|
||
python_path: "./venv/bin/python" # 默认使用虚拟环境，如不存在则自动检测系统Python
|
||
# timeout 已移除，统一使用 document_parser.processing_timeout (3600秒)
|
||
enable_plugins: false
|
||
|
||
# 功能开关
|
||
features:
|
||
ocr: true
|
||
audio_transcription: true
|
||
azure_doc_intel: false
|
||
youtube_transcription: false
|
||
|
||
# 存储配置
|
||
storage:
|
||
# Sled数据库配置
|
||
sled:
|
||
path: "data/document_parser"
|
||
cache_capacity: 104857600 # 100MB
|
||
|
||
# OSS配置
|
||
oss:
|
||
endpoint: "oss-rg-china-mainland.aliyuncs.com"
|
||
# 公共bucket，用于存储公共文件，如文档文件
|
||
public_bucket: "nuwa-packages"
|
||
# 私有bucket，用于存储私有文件，如模型文件
|
||
private_bucket: "edu-nuwa-packages"
|
||
access_key_id: "${OSS_ACCESS_KEY_ID}" # 通过环境变量设置
|
||
access_key_secret: "${OSS_ACCESS_KEY_SECRET}" # 通过环境变量设置
|
||
region: "oss-rg-china-mainland"
|
||
upload_directory: "document_parser" # 上传文件的统一子目录前缀
|
||
|
||
# 全局文件大小配置
|
||
file_size_config:
|
||
max_file_size: "200MB"
|
||
large_document_threshold: "50MB"
|
||
|
||
# 外部集成配置
|
||
external_integration:
|
||
webhook_url: ""
|
||
api_key: ""
|
||
timeout: 30
|