Accueil Explorer Aide
Connexion
hewensong
/
Clerk2.5
1
0
Fork 0
Fichiers Tickets 0 Pull Requests 0 Wiki
Aborescence: 1cd66b8826
Branches Tags
Clerk2.5
/
pdf_converter_v2
/
config.py

config.py 2.4 KB
Historique Raw

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152 
  1. # Copyright (c) Opendatalab. All rights reserved.
    2. 

  2. 

  3. """
    4. 

  4. 配置文件 v2 - 从配置文件读取配置（不使用环境变量）
    5. 

  5. """
    6. 

  6. 

  7. import os
    8. 

  8. from .config_loader import get_config_loader
    9. 

  9. from .utils.device_env import detect_device_kind
    10. 

  10. 

  11. # 加载配置文件
    12. 

  12. _config = get_config_loader()
    13. 

  13. 

  14. # 设备环境：nvi（NVIDIA GPU）/ npu（华为昇腾 NPU）/ cpu，用于按环境选择 VLLM_PLUGINS、PADDLE_OCR_DEVICE 等
    15. 

  15. # 优先从配置文件读取，如果未配置则自动检测
    16. 

  16. _device_kind_from_config = _config.get_str("device_kind", "")
    17. 

  17. DEVICE_KIND = _device_kind_from_config if _device_kind_from_config else detect_device_kind()
    18. 

  18. 

  19. # 默认模型配置
    20. 

  20. DEFAULT_MODEL_NAME = "OpenDataLab/MinerU2.5-2509-1.2B"
    21. 

  21. DEFAULT_GPU_MEMORY_UTILIZATION = _config.get_float("default_gpu_memory_utilization", 0.9)
    22. 

  22. DEFAULT_DPI = _config.get_int("default_dpi", 200)
    23. 

  23. DEFAULT_MAX_PAGES = _config.get_int("default_max_pages", 10)
    24. 

  24. 

  25. # v2 特有配置（外部API相关）
    26. 

  26. DEFAULT_API_URL = _config.get_str("api_url", "http://127.0.0.1:5282")
    27. 

  27. DEFAULT_BACKEND = _config.get_str("backend", "vlm-vllm-async-engine")
    28. 

  28. DEFAULT_PARSE_METHOD = _config.get_str("parse_method", "auto")
    29. 

  29. DEFAULT_START_PAGE_ID = _config.get_int("start_page_id", 0)
    30. 

  30. DEFAULT_END_PAGE_ID = _config.get_int("end_page_id", 99999)
    31. 

  31. DEFAULT_LANGUAGE = _config.get_str("language", "ch")
    32. 

  32. DEFAULT_RESPONSE_FORMAT_ZIP = _config.get_bool("response_format_zip", True)
    33. 

  33. DEFAULT_RETURN_MIDDLE_JSON = _config.get_bool("return_middle_json", False)
    34. 

  34. DEFAULT_RETURN_MODEL_OUTPUT = _config.get_bool("return_model_output", True)
    35. 

  35. DEFAULT_RETURN_MD = _config.get_bool("return_md", True)
    36. 

  36. DEFAULT_RETURN_IMAGES = _config.get_bool("return_images", False)
    37. 

  37. DEFAULT_RETURN_CONTENT_LIST = _config.get_bool("return_content_list", False)
    38. 

  38. DEFAULT_SERVER_URL = os.getenv("SERVER_URL", "string")
    39. 

  39. 

  40. # PaddleOCR 配置
    41. 

  41. PADDLEOCR_CMD = _config.get_str("paddleocr_cmd", "paddleocr")
    42. 

  42. PADDLE_OCR_DEVICE = _config.get_str("paddle_ocr_device", "")
    43. 

  43. PADDLE_OCR_DEVICES = _config.get_str("paddle_ocr_devices", "")
    44. 

  44. PADDLE_DOC_PARSER_CMD = _config.get_str("paddle_doc_parser_cmd", "paddleocr")
    45. 

  45. VL_REC_BACKEND = _config.get_str("vl_rec_backend", "")
    46. 

  46. VL_REC_SERVER_URL = _config.get_str("vl_rec_server_url", "")
    47. 

  47. 

  48. # 日志配置（可选）
    49. 

  49. LOG_DIR = _config.get_str("log_dir", "./logs")
    50. 

  50. LOG_LEVEL = _config.get_str("log_level", "INFO")
    51. 

  51. LOG_TO_FILE = _config.get_bool("log_to_file", True)
    52. 

  52. LOG_TO_CONSOLE = _config.get_bool("log_to_console", True)
    53.