LlamaParse Integration

LLAMAPARSE_API_KEY="llx_your_llamaparse_api_key_here"

pip install llama-cloud-services

from llama_cloud_services import LlamaParse
from config import LLAMAPARSE_API_KEY

# Initialize parser with configuration
parser = LlamaParse(
    api_key=LLAMAPARSE_API_KEY,
    num_workers=4,          # Parallel processing workers
    verbose=False,          # Disable verbose logging
    language="en",          # English language
    result_type="markdown"  # Output format
)

async def process_pdf_async(file_path: str) -> dict:
    """
    Async function to process a single PDF file and extract page-wise content.

    Args:
        file_path: Path to the PDF file (local or URL)

    Returns:
        Dictionary containing extracted text and status
    """
    try:
        # Parse the PDF asynchronously
        results = await parser.aparse(file_path)
        
        # Extract text from all pages
        text = ""
        for page in results.pages:
            text += page.md + "\n"  # Markdown content
            text += "=" * 80 + "\n"  # Page separator
        
        return {
            "text": text,
            "status": "success"
        }

    except Exception as e:
        return {
            "text": "",
            "status": "error",
            "error": str(e)
        }

from parsers.parse import process_pdf_async
from supabase_client import SupabaseCaseClient

supabase = SupabaseCaseClient()

async def process_lab_report(case_id: str, file_id: str, file_url: str):
    """
    Process a lab report PDF and store extracted text.
    """
    # Parse the PDF
    result = await process_pdf_async(file_url)
    
    if result["status"] == "success":
        # Store extracted text in file metadata
        await supabase.update_case_file_metadata(
            file_id=file_id,
            metadata={
                "extracted_text": result["text"],
                "parse_status": "completed"
            }
        )
        
        print(f"Successfully parsed lab report for case {case_id}")
        return result["text"]
    else:
        print(f"Error parsing PDF: {result['error']}")
        raise Exception(result["error"])

# Lab Report - Complete Blood Count (CBC)

**Patient**: John Doe
**Date**: 2024-03-04
**Lab**: Medical Center Laboratory

## Test Results

| Test | Result | Reference Range | Unit | Flag |
|------|--------|-----------------|------|------|
| WBC | 7.5 | 4.0-11.0 | K/uL | Normal |
| RBC | 4.8 | 4.5-5.5 | M/uL | Normal |
| Hemoglobin | 14.2 | 13.5-17.5 | g/dL | Normal |
| Hematocrit | 42 | 38-50 | % | Normal |
| Platelets | 250 | 150-400 | K/uL | Normal |

## Interpretation

All values within normal reference ranges.

================================================================================

# 1. Parse lab report
lab_text = await process_pdf_async(lab_report_url)

# 2. Combine with other data
context = {
    "doctor_notes": case.doctor_notes,
    "lab_data": lab_text,
    "radiology_data": radiology_summaries
}

# 3. Generate AI insights using Groq
insights = await medical_agent.generate_insights(context)

parser = LlamaParse(
    api_key=LLAMAPARSE_API_KEY,
    num_workers=8,                    # More workers for faster processing
    verbose=True,                     # Enable debugging
    language="en",
    result_type="markdown",
    
    # Advanced options
    parsing_instruction="Focus on extracting lab values and reference ranges",
    skip_diagonal_text=True,          # Skip watermarks
    invalidate_cache=False,           # Use cached results if available
    do_not_cache=False,               # Cache results for future use
    continuous_mode=True,             # Better for multi-page documents
)

async def process_multiple_pdfs(file_urls: list) -> list:
    """
    Process multiple PDFs in parallel.
    """
    tasks = [process_pdf_async(url) for url in file_urls]
    results = await asyncio.gather(*tasks)
    return results