49 lines
1.5 KiB
Python
49 lines
1.5 KiB
Python
# Copyright (c) 2025 Patrick Motsch
|
|
# All rights reserved.
|
|
from typing import Any, Dict, List
|
|
import xml.etree.ElementTree as ET
|
|
|
|
from modules.datamodels.datamodelExtraction import ContentPart
|
|
from ..subUtils import makeId
|
|
from ..subRegistry import Extractor
|
|
|
|
|
|
class XmlExtractor(Extractor):
|
|
"""
|
|
Extractor for XML files.
|
|
|
|
Supported formats:
|
|
- MIME types: application/xml
|
|
- File extensions: .xml, .rss, .atom
|
|
- Special handling: Uses ElementTree for parsing
|
|
"""
|
|
|
|
def detect(self, fileName: str, mimeType: str, headBytes: bytes) -> bool:
|
|
return mimeType == "application/xml" or (fileName or "").lower().endswith((".xml", ".rss", ".atom"))
|
|
|
|
def getSupportedExtensions(self) -> list[str]:
|
|
"""Return list of supported file extensions."""
|
|
return [".xml", ".rss", ".atom"]
|
|
|
|
def getSupportedMimeTypes(self) -> list[str]:
|
|
"""Return list of supported MIME types."""
|
|
return ["application/xml"]
|
|
|
|
def extract(self, fileBytes: bytes, context: Dict[str, Any]) -> List[ContentPart]:
|
|
mimeType = context.get("mimeType") or "application/xml"
|
|
text = fileBytes.decode("utf-8", errors="replace")
|
|
try:
|
|
ET.fromstring(text)
|
|
except Exception:
|
|
pass
|
|
return [ContentPart(
|
|
id=makeId(),
|
|
parentId=None,
|
|
label="main",
|
|
typeGroup="structure",
|
|
mimeType=mimeType,
|
|
data=text,
|
|
metadata={"size": len(fileBytes)}
|
|
)]
|
|
|
|
|