markitect-main/tools/validate_js.py

#!/usr/bin/env python3
"""
JavaScript Validation Tool

Extracts JavaScript from HTML files and validates syntax.
Detects common issues like "Uncaught SyntaxError: unexpected token"
"""

import re
import subprocess
import sys
import tempfile
from pathlib import Path


def extract_javascript_from_html(html_file):
    """Extract all JavaScript code from an HTML file."""
    try:
        content = Path(html_file).read_text(encoding='utf-8')
    except Exception as e:
        print(f"❌ Failed to read {html_file}: {e}")
        return []

    # Find all <script> blocks
    script_pattern = r'<script[^>]*>(.*?)</script>'
    scripts = re.findall(script_pattern, content, re.DOTALL | re.IGNORECASE)

    # Filter out empty scripts and external script tags
    js_blocks = []
    for script in scripts:
        script = script.strip()
        if script and not script.startswith('http'):  # Skip external scripts
            js_blocks.append(script)

    return js_blocks


def validate_javascript_syntax(js_code):
    """Validate JavaScript syntax using Node.js."""
    try:
        # Create temporary JS file
        with tempfile.NamedTemporaryFile(mode='w', suffix='.js', delete=False) as f:
            f.write(js_code)
            temp_file = f.name

        # Try to parse with node --check
        result = subprocess.run(
            ['node', '--check', temp_file],
            capture_output=True,
            text=True
        )

        # Clean up
        Path(temp_file).unlink()

        if result.returncode == 0:
            return True, "✓ Syntax OK"
        else:
            return False, f"❌ Syntax Error: {result.stderr.strip()}"

    except FileNotFoundError:
        # Fallback: Try to detect obvious syntax errors
        return validate_javascript_basic(js_code)
    except Exception as e:
        return False, f"❌ Validation failed: {e}"


def validate_javascript_basic(js_code):
    """Basic JavaScript syntax validation without Node.js."""
    errors = []

    # Check for common syntax issues
    if js_code.count('{') != js_code.count('}'):
        errors.append("Mismatched curly braces")

    if js_code.count('(') != js_code.count(')'):
        errors.append("Mismatched parentheses")

    if js_code.count('[') != js_code.count(']'):
        errors.append("Mismatched square brackets")

    # Check for unescaped quotes in strings
    if re.search(r'["\']\s*["\']\s*["\']\s*["\']', js_code):
        errors.append("Possible quote escaping issue")

    # Check for Python-style string formatting leftover
    if '${' in js_code and '"}' in js_code:
        errors.append("Possible Python string template leftover")

    if errors:
        return False, f"❌ Potential issues: {', '.join(errors)}"
    else:
        return True, "✓ Basic validation passed (Node.js not available)"


def validate_html_js(html_file):
    """Validate all JavaScript in an HTML file."""
    print(f"🔍 Validating JavaScript in: {html_file}")

    js_blocks = extract_javascript_from_html(html_file)

    if not js_blocks:
        print("⚠️  No JavaScript found in HTML file")
        return True

    print(f"📝 Found {len(js_blocks)} JavaScript blocks")

    all_valid = True
    for i, js_block in enumerate(js_blocks, 1):
        print(f"\n--- Script Block {i} ({len(js_block)} chars) ---")

        # Show first few lines for context
        lines = js_block.split('\n')[:3]
        preview = '\n'.join(lines)
        if len(lines) < len(js_block.split('\n')):
            preview += "\n..."
        print(f"Preview:\n{preview}")

        is_valid, message = validate_javascript_syntax(js_block)
        print(message)

        if not is_valid:
            all_valid = False
            # Show more context around potential error
            if 'line' in message.lower():
                try:
                    line_num = re.search(r'line (\d+)', message, re.IGNORECASE)
                    if line_num:
                        line_no = int(line_num.group(1))
                        lines = js_block.split('\n')
                        start = max(0, line_no - 3)
                        end = min(len(lines), line_no + 2)
                        print("\nContext around error:")
                        for j in range(start, end):
                            marker = ">>> " if j == line_no - 1 else "    "
                            print(f"{marker}{j+1:3}: {lines[j]}")
                except:
                    pass

    print(f"\n{'✅' if all_valid else '❌'} Overall result: {'All JavaScript is valid' if all_valid else 'JavaScript validation failed'}")
    return all_valid


def main():
    if len(sys.argv) != 2:
        print("Usage: python validate_js.py <html_file>")
        print("Example: python validate_js.py /tmp/test-edit.html")
        sys.exit(1)

    html_file = sys.argv[1]

    if not Path(html_file).exists():
        print(f"❌ File not found: {html_file}")
        sys.exit(1)

    success = validate_html_js(html_file)
    sys.exit(0 if success else 1)


if __name__ == "__main__":
    main()