PDF Extract Progressive Tools

# pdf-extract-orchestrator.py
# Implements the progressive tool fallback pattern

def extract_pdf_text(pdf_path):
    """
    Progressive PDF extraction following tool precedence:
    1. read_file (quick check)
    2. run_shell + pdftotext (primary extraction)
    3. execute_code_sandbox + PyMuPDF (final fallback)
    """
    extraction_log = []
    
    # Step 1: Try read_file
    print("Step 1: Attempting read_file...")
    try:
        content = read_file(pdf_path)
        if is_binary_or_image_data(content):
            extraction_log.append("read_file: Returned binary data")
            # Proceed to Step 2
        else:
            extraction_log.append("read_file: Success")
            return content, extraction_log
    except Exception as e:
        extraction_log.append(f"read_file: Failed - {e}")
    
    # Step 2: Try run_shell with pdftotext
    print("Step 2: Attempting run_shell + pdftotext...")
    try:
        run_shell(f"pdftotext {pdf_path} output.txt")
        content = read_file("output.txt")
        if content and len(content) > 100:
            extraction_log.append("run_shell/pdftotext: Success")
            return content, extraction_log
        else:
            extraction_log.append("run_shell/pdftotext: Empty extraction")
    except Exception as e:
        extraction_log.append(f"run_shell/pdftotext: Failed - {e}")
    
    # Step 3: Try execute_code_sandbox with PyMuPDF
    print("Step 3: Attempting execute_code_sandbox + PyMuPDF...")
    try:
        code = """
import fitz
doc = fitz.open("""" + pdf_path + """")
text = ""
for page in doc:
    text += page.get_text()
doc.close()
print(text[:1000])  # Preview
"""
        result = execute_code_sandbox(language="python", code=code)
        extraction_log.append("execute_code_sandbox/PyMuPDF: Success")
        return result, extraction_log
    except Exception as e:
        extraction_log.append(f"execute_code_sandbox/PyMuPDF: Failed - {e}")
    
    # Step 4: All methods failed
    extraction_log.append("ALL METHODS FAILED - Escalate to domain knowledge")
    return None, extraction_log

def is_binary_or_image_data(content):
    """Detect if content is binary/image data rather than extracted text"""
    if not content:
        return True
    # Check for PDF header without text extraction
    if content.startswith("%PDF-"):
        return True
    # Check for high ratio of non-printable characters
    non_printable = sum(1 for c in content if ord(c) < 32 and c not in '\n\r\t')
    if len(content) > 0 and non_printable / len(content) > 0.1:
        return True
    return False

                    ┌─────────────────┐
                    │  Start: PDF     │
                    │  Available?     │
                    └────────┬────────┘
                             │
                    ┌────────▼────────┐
                    │   Step 1:       │
                    │   read_file     │
                    └────────┬────────┘
                             │
              ┌──────────────┼──────────────┐
              │              │              │
        ┌─────▼─────┐  ┌─────▼─────┐  ┌─────▼─────┐
        │  Text     │  │  Binary   │  │  Error/   │
        │  Returned │  │  Data     │  │  Not Found│
        └─────┬─────┘  └─────┬─────┘  └─────┬─────┘
              │              │              │
              │         ┌────▼─────┐  ┌────▼─────┐
              │         │ Step 2:  │  │ Download │
              │         │ run_shell│  │ or Fix   │
              │         │ pdftotext│  │ Path     │
              │         └────┬─────┘  └──────────┘
              │              │
              │         ┌────▼─────┐
              │         │ Success? │
              │         └────┬─────┘
              │              │
        ┌─────▼─────┐  ┌─────▼─────┐
        │  Yes      │  │  No       │
        └─────┬─────┘  └─────┬─────┘
              │              │
              │         ┌────▼─────────┐
              │         │ Step 3:      │
              │         │ execute_     │
              │         │ code_sandbox │
              │         │ PyMuPDF      │
              │         └──────────────┘
              │
        ┌─────▼──────────────────┐
        │  Step 4: Quality Check │
        │  Step 5: Document      │
        │  Limitations           │
        └────────────────────────┘