FastAPIExperience2 / core /doc_processor.py
utkarsh134's picture
Upload 14 files
a88785b verified
Raw History Blame
2.45 kB
import pymupdf
from io import BytesIO
import docx2txt
from datetime import datetime
import zipfile
from xml.etree import ElementTree as ET
def extract_text_from_bytes(pdf_bytes):
try:
memory_stream = BytesIO(pdf_bytes)
doc = pymupdf.open(stream=memory_stream, filetype="pdf")
# Extract text
text = ""
for page in doc:
text += page.get_text()
# Get modification date
metadata = doc.metadata
mod_date = metadata.get("modDate", None)
doc.close()
return text, mod_date
except Exception as e:
return f"Error reading PDF: {str(e)}", None
def get_docx_last_modified(docx_stream):
try:
docx_stream.seek(0) # Reset stream position to start
with zipfile.ZipFile(docx_stream, 'r') as docx:
if 'docProps/core.xml' in docx.namelist():
with docx.open('docProps/core.xml') as f:
core_xml = f.read()
root = ET.fromstring(core_xml)
namespaces = {
'cp': 'http://schemas.openxmlformats.org/package/2006/metadata/core-properties',
'dcterms': 'http://purl.org/dc/terms/'
}
modified_element = root.find('.//dcterms:modified', namespaces)
if modified_element is not None:
modified_date_str = modified_element.text
return datetime.fromisoformat(modified_date_str.replace('Z', '+00:00'))
except Exception as e:
print(f"Error extracting DOCX metadata: {e}")
return None
def extract_text_from_docx(docx_bytes):
try:
docx_stream = BytesIO(docx_bytes)
text = docx2txt.process(docx_stream)
metadata_stream = BytesIO(docx_bytes)
mod_date = get_docx_last_modified(metadata_stream)
return text, mod_date
except Exception as e:
return f"Error reading DOCX: {str(e)}", None
def parse_pdf_date(pdf_date: str) -> datetime:
if not pdf_date or not pdf_date.startswith("D:"):
return datetime.now()
try:
date_str = pdf_date[2:14]
return datetime.strptime(date_str, "%Y%m%d%H%M")
except Exception:
return datetime.now()