52 lines
1.8 KiB
Python
52 lines
1.8 KiB
Python
|
"""Knowledge document parsing for supported upload formats."""
|
||
|
|
from __future__ import annotations
|
||
|
|
|
||
|
|
from pathlib import Path
|
||
|
|
from xml.etree import ElementTree
|
||
|
|
from zipfile import BadZipFile, ZipFile
|
||
|
|
|
||
|
|
from pypdf import PdfReader
|
||
|
|
|
||
|
|
SUPPORTED_EXTENSIONS = {".txt", ".md", ".docx", ".pdf"}
|
||
|
|
_WORD_NS = "{http://schemas.openxmlformats.org/wordprocessingml/2006/main}"
|
||
|
|
|
||
|
|
|
||
|
|
def parse_document(path: str | Path) -> str:
|
||
|
|
"""Return document text for a supported knowledge-base upload."""
|
||
|
|
document_path = Path(path)
|
||
|
|
suffix = document_path.suffix.lower()
|
||
|
|
if suffix not in SUPPORTED_EXTENSIONS:
|
||
|
|
raise ValueError(f"Unsupported document type: {suffix or '<none>'}")
|
||
|
|
if not document_path.is_file():
|
||
|
|
raise FileNotFoundError(document_path)
|
||
|
|
|
||
|
|
if suffix in {".txt", ".md"}:
|
||
|
|
return document_path.read_text(encoding="utf-8-sig")
|
||
|
|
if suffix == ".pdf":
|
||
|
|
return _parse_pdf(document_path)
|
||
|
|
return _parse_docx(document_path)
|
||
|
|
|
||
|
|
|
||
|
|
def _parse_docx(path: Path) -> str:
|
||
|
|
try:
|
||
|
|
with ZipFile(path) as archive:
|
||
|
|
xml = archive.read("word/document.xml")
|
||
|
|
root = ElementTree.fromstring(xml)
|
||
|
|
except (BadZipFile, KeyError, ElementTree.ParseError) as exc:
|
||
|
|
raise ValueError(f"Invalid DOCX document: {path}") from exc
|
||
|
|
|
||
|
|
paragraphs = []
|
||
|
|
for paragraph in root.iter(f"{_WORD_NS}p"):
|
||
|
|
text = "".join(node.text or "" for node in paragraph.iter(f"{_WORD_NS}t"))
|
||
|
|
if text:
|
||
|
|
paragraphs.append(text)
|
||
|
|
return "\n".join(paragraphs)
|
||
|
|
|
||
|
|
|
||
|
|
def _parse_pdf(path: Path) -> str:
|
||
|
|
try:
|
||
|
|
reader = PdfReader(str(path))
|
||
|
|
return "\n".join(page.extract_text() or "" for page in reader.pages).strip()
|
||
|
|
except Exception as exc: # pypdf raises format-specific exceptions
|
||
|
|
raise ValueError(f"Invalid PDF document: {path}") from exc
|