#!/usr/bin/env python3
"""Extract inline <style> blocks from HTML templates into external CSS files.

Replaces each <style>...</style> block with a <link rel="stylesheet"> tag.
This allows browsers to cache CSS across page navigations and dramatically
reduces HTML payload size.

Usage:
    python3 extract-inline-css.py
"""

import glob
import re
from pathlib import Path


def extract_css_from_template(html_path, css_output_dir, css_url_prefix):
    """Extract <style> blocks from an HTML file into an external CSS file.

    Returns the modified HTML content.
    """
    with open(html_path, 'r', encoding='utf-8') as f:
        html = f.read()

    style_blocks = list(re.finditer(
        r'<style[^>]*>(.*?)</style>',
        html,
        re.DOTALL | re.IGNORECASE
    ))

    if not style_blocks:
        return html  # No inline CSS to extract

    # Extract all CSS content
    css_parts = []
    for m in style_blocks:
        css = m.group(1).strip()
        if css:
            css_parts.append(css)

    if not css_parts:
        return html

    # Write external CSS file
    name = Path(html_path).stem
    css_filename = f"{name}.css"
    css_path = Path(css_output_dir) / css_filename
    css_path.parent.mkdir(parents=True, exist_ok=True)

    combined_css = '\n\n'.join(css_parts)
    with open(css_path, 'w', encoding='utf-8') as f:
        f.write(f"/* Extracted from {Path(html_path).name} */\n\n")
        f.write(combined_css)
        f.write('\n')

    # Replace first <style> block with <link>, remove the rest
    link_tag = f'<link rel="stylesheet" href="{css_url_prefix}{css_filename}">'
    new_html = html[:style_blocks[0].start()] + link_tag + html[style_blocks[0].end():]

    # Remove remaining style blocks (they were concatenated into the CSS file)
    for m in reversed(style_blocks[1:]):
        new_html = new_html[:m.start()] + new_html[m.end():]

    return new_html


def process_directory(html_dir, css_output_dir, css_url_prefix):
    """Process all HTML files in a directory."""
    templates = sorted(glob.glob(f"{html_dir}/*.html"))
    total_saved = 0

    for html_path in templates:
        new_html = extract_css_from_template(html_path, css_output_dir, css_url_prefix)

        # Write back the modified HTML
        with open(html_path, 'w', encoding='utf-8') as f:
            f.write(new_html)

        original_size = Path(html_path).stat().st_size
        total_saved += original_size - len(new_html.encode('utf-8'))
        name = Path(html_path).name
        print(f"  {name}: extracted CSS -> {css_url_prefix}{Path(html_path).stem}.css")

    return total_saved


def main():
    print("=== Extracting inline CSS from POS templates ===")
    saved_pos = process_directory(
        '/home/Autopartes/pos/templates',
        '/home/Autopartes/pos/static/css',
        '/pos/static/css/'
    )

    print("\n=== Extracting inline CSS from Dashboard templates ===")
    saved_dash = process_directory(
        '/home/Autopartes/dashboard',
        '/home/Autopartes/dashboard',
        ''  # Dashboard CSS is in the same directory
    )

    total_saved_kb = (saved_pos + saved_dash) / 1024
    print(f"\n=== Done ===")
    print(f"Total HTML payload reduced by ~{total_saved_kb:.0f} KB")
    print(f"POS templates: ~{saved_pos/1024:.0f} KB")
    print(f"Dashboard templates: ~{saved_dash/1024:.0f} KB")


if __name__ == '__main__':
    main()