"""
Document analysis tools for FastMCP server.
Extracts text and metadata from multiple document types.
"""
from fastmcp import FastMCP
from src.core.pdf import DocumentAnalyzer
from src.logging.logger import logger
from src.utils.markdown_formatter import format_document_markdown
def register_pdf_tools(mcp: FastMCP):
"""Register document analysis tools."""
analyzer = DocumentAnalyzer()
@mcp.tool
async def document_analysis(
url: str,
max_pages: int = 10,
extract_metadata: bool = True,
summary_length: int = 500
) -> str:
"""
Download and analyze documents of multiple types with OCR support.
Supports: PDF, Word (.docx), Text (.txt, .md), Images (.jpg, .png) with OCR.
Extracts text content and metadata without requiring authentication.
Automatically uses OCR for scanned PDFs and images.
Args:
url: URL of the document to analyze (PDF, Word, Text, etc.)
max_pages: Maximum pages to extract (for PDFs, default: 10)
extract_metadata: Whether to extract document metadata
summary_length: Length of text preview in output (default: 500)
"""
try:
logger.info(f"Document analysis for: {url}")
result = await analyzer.analyze_document(
url=url,
max_pages=max_pages,
extract_metadata=extract_metadata
)
return format_document_markdown(result, summary_length)
except Exception as e:
logger.error(f"Document analysis failed: {e}")
return f"# 📄 Document Analysis\n\n❌ **Error:** {str(e)}"
