ran ruff format via uv

This commit is contained in:
yuisheaven
2025-10-05 02:16:42 +02:00
parent c9a687171a
commit 3ff6346c03
5 changed files with 87 additions and 67 deletions
+17 -15
View File
@@ -42,20 +42,20 @@ def setup_logging():
# Document Parsing Configuration
def get_unstructured_api_url() -> Optional[str]:
"""Get the Unstructured API URL from environment variables.
Returns:
The Unstructured API URL if parsing is enabled, None otherwise.
"""
enabled = os.getenv("ENABLE_UNSTRUCTURED_PARSING", "true").lower() == "true"
if not enabled:
return None
return os.getenv("UNSTRUCTURED_API_URL", "http://unstructured:8000")
def is_unstructured_parsing_enabled() -> bool:
"""Check if unstructured document parsing is enabled.
Returns:
True if enabled, False otherwise.
"""
@@ -64,36 +64,36 @@ def is_unstructured_parsing_enabled() -> bool:
def get_unstructured_strategy() -> str:
"""Get the parsing strategy for the Unstructured API.
Valid values are:
- 'auto': Automatically choose the best strategy (default)
- 'fast': Fast parsing without OCR
- 'hi_res': High-resolution parsing with OCR for better accuracy
Returns:
The parsing strategy to use.
"""
strategy = os.getenv("UNSTRUCTURED_STRATEGY", "auto").lower()
valid_strategies = ["auto", "fast", "hi_res"]
if strategy not in valid_strategies:
logging.warning(
f"Invalid UNSTRUCTURED_STRATEGY '{strategy}'. Using 'hi_res'. "
f"Valid options: {', '.join(valid_strategies)}"
)
return "hi_res"
return strategy
def get_unstructured_languages() -> list[str]:
"""Get the OCR languages for the Unstructured API.
Languages should be specified as ISO 639-3 codes (e.g., 'eng', 'deu', 'fra').
Multiple languages can be specified separated by commas.
Default languages: English (eng) and German (deu)
Common language codes:
- eng: English
- deu: German
@@ -106,17 +106,19 @@ def get_unstructured_languages() -> list[str]:
- zho: Chinese
- jpn: Japanese
- kor: Korean
Returns:
List of language codes for OCR processing.
"""
languages_str = os.getenv("UNSTRUCTURED_LANGUAGES", "eng,deu")
# Split by comma and clean up whitespace
languages = [lang.strip() for lang in languages_str.split(",") if lang.strip()]
if not languages:
logging.warning("No languages specified in UNSTRUCTURED_LANGUAGES. Using default: eng,deu")
logging.warning(
"No languages specified in UNSTRUCTURED_LANGUAGES. Using default: eng,deu"
)
return ["eng", "deu"]
return languages