# From the pdf-large-reader repository
cd /mnt/github/workspace-hub/pdf-large-reader
pip install -e .

# Or with extras
pip install -e ".[dev,progress]"

from pdf_large_reader import process_large_pdf, extract_text_only, extract_everything

# Simple text extraction
text = extract_text_only("large_document.pdf")
print(text)

# Process with automatic strategy selection
pages = process_large_pdf(
    "large_document.pdf",
    output_format="list",
    extract_images=True,
    extract_tables=True
)

# Memory-efficient streaming for very large files
for page in process_large_pdf("huge_file.pdf", output_format="generator"):
    print(f"Page {page.page_number}: {len(page.text)} characters")

from pdf_large_reader import extract_everything

# Extract everything including tables with robust error handling
pages = extract_everything("technical_standard.pdf")

for page in pages:
    if 'tables' in page.metadata:
        tables = page.metadata['tables']
        print(f"Page {page.page_number}: Found {len(tables)} tables")

        for i, table_df in enumerate(tables):
            print(f"  Table {i+1}: {table_df.shape[0]} rows x {table_df.shape[1]} cols")
            print(table_df.head())

# Extract text from PDF
pdf-large-reader document.pdf

# Save to file
pdf-large-reader document.pdf --output result.txt

# Extract with images and tables
pdf-large-reader document.pdf --extract-images --extract-tables

# Use generator format for large files
pdf-large-reader huge.pdf --output-format generator

# Verbose output
pdf-large-reader document.pdf --verbose

# Main entry point with automatic strategy
process_large_pdf(
    pdf_path,
    output_format="generator",    # "generator" (default), "list", or "text"
    extract_images=False,         # Extract images
    extract_tables=False,         # Extract tables with normalization
    chunk_size=None,              # Auto-calculated if None
    fallback_api_key=None,        # OpenAI API key for complex pages
    fallback_model="gpt-4.1",      # Model for fallback extraction
    progress_callback=None,       # Progress tracking function
    auto_strategy=True            # Enable automatic strategy selection
)

# Quick text extraction
extract_text_only(pdf_path) -> str

# Extract with images
extract_pages_with_images(pdf_path) -> List[PDFPage]

# Extract with tables
extract_pages_with_tables(pdf_path) -> List[PDFPage]

# Extract everything
extract_everything(pdf_path) -> List[PDFPage]

@dataclass
class PDFPage:
    page_number: int          # Page number (1-indexed)
    text: str                 # Extracted text from page
    images: List[dict]        # Extracted images with metadata
    metadata: dict            # Page metadata including tables