qt6-migration-pipeline/fix_export_headers_v4.py

#!/usr/bin/env python3
"""
Repariert Export-Header-Includes in Qt6 migrierten Dateien.

Nur für SHARED/MODULE Libraries - STATIC Libraries brauchen keine Export-Header.
"""

import os
import re
import sys
from pathlib import Path
from collections import defaultdict

def parse_cmake_file(cmake_path, root_dir):
    """
    Parse CMakeLists.txt und extrahiere Target-Informationen.
    
    Returns:
        targets: dict mapping target_name -> {'type': 'SHARED'|'STATIC'|'MODULE', 'sources': [files]}
        src_to_target: dict mapping source_file (relative path) -> target_name
    """
    targets = {}
    src_to_target = {}
    
    if not os.path.exists(cmake_path):
        return targets, src_to_target
    
    with open(cmake_path, 'r', encoding='utf-8', errors='ignore') as f:
        content = f.read()
    
    # Remove comments
    content = re.sub(r'#[^\n]*', '', content)
    
    # Parse SET() variables first
    set_vars = {}
    for match in re.finditer(r'SET\s*\(\s*(\w+)\s+([^)]+)\)', content, re.IGNORECASE):
        var_name = match.group(1)
        var_value = match.group(2).strip()
        # Split on whitespace
        files = [f.strip() for f in var_value.split() if f.strip()]
        set_vars[var_name] = files
    
    # Parse NGL_ADD_LIBRARY and NGL_ADD_LIBRARY2
    # Pattern: NGL_ADD_LIBRARY(... target_name SHARED|STATIC|MODULE ...)
    # or: NGL_ADD_LIBRARY2(target_name SHARED|STATIC|MODULE ...)
    
    for pattern in [r'NGL_ADD_LIBRARY\s*\(([^)]+)\)', r'NGL_ADD_LIBRARY2\s*\(([^)]+)\)']:
        for match in re.finditer(pattern, content, re.IGNORECASE | re.DOTALL):
            args_str = match.group(1)
            args = args_str.split()
            
            if len(args) < 3:
                continue
            
            # Skip flags like ALL, WS, AAR, SV, NO_TRANSLATION
            idx = 0
            while idx < len(args) and args[idx].upper() in ['ALL', 'WS', 'AAR', 'SV', 'NO_TRANSLATION']:
                idx += 1
            
            if idx >= len(args):
                continue
            
            target_name = args[idx].lower()  # Target names are lowercase in export headers
            idx += 1
            
            if idx >= len(args):
                continue
            
            lib_type = args[idx].upper()
            if lib_type not in ['SHARED', 'STATIC', 'MODULE']:
                continue
            
            idx += 1
            
            # Rest are source files (may include variable references like ${VAR})
            sources = []
            for arg in args[idx:]:
                if arg.startswith('${'):
                    # Variable reference - expand it
                    var_name = arg[2:-1]  # Remove ${ and }
                    if var_name in set_vars:
                        sources.extend(set_vars[var_name])
                else:
                    sources.append(arg)
            
            targets[target_name] = {
                'type': lib_type,
                'sources': [s for s in sources if not s.endswith('.h') or 'export' in s.lower()]
            }
            
            # Map source files to target (only .cpp/.h files, not export headers)
            cmake_dir = os.path.dirname(cmake_path)
            for src in sources:
                if src.endswith('_export.h'):
                    continue  # Skip export headers
                if os.path.splitext(src)[1] in ['.cpp', '.h', '.cc', '.cxx', '.hpp']:
                    # Store relative path from root_dir
                    abs_path = os.path.normpath(os.path.join(cmake_dir, src))
                    rel_path = os.path.relpath(abs_path, root_dir)
                    src_to_target[rel_path] = target_name
    
    return targets, src_to_target


def find_export_header_includes(content):
    """Find export header includes in file content."""
    # Match both "..." and <...> styles
    pattern = r'#include\s*[<"]([^>"\s]+_export\.h)[>"]'
    return re.findall(pattern, content)


def get_expected_export_header(target_name):
    """Get expected export header name for a target."""
    return f"{target_name}_export.h"


def fix_file(file_path, target_name, all_targets):
    """
    Fix export header include in a file.
    
    Returns:
        (old_include, new_include, reason) or None if no fix needed
    """
    with open(file_path, 'r', encoding='utf-8', errors='ignore') as f:
        content = f.read()
    
    current_includes = find_export_header_includes(content)
    
    if not current_includes:
        return None  # No export header include found
    
    # Check if target is SHARED or MODULE (STATIC doesn't need export headers)
    if target_name not in all_targets:
        return None  # Target not found
    
    target_info = all_targets[target_name]
    if target_info['type'] == 'STATIC':
        # STATIC libraries don't need export headers - remove the include
        return (current_includes[0], None, f"STATIC library '{target_name}' doesn't need export header")
    
    expected_header = get_expected_export_header(target_name)
    
    if len(current_includes) == 1 and current_includes[0] == expected_header:
        return None  # Already correct
    
    # Fix needed
    old_include = current_includes[0]
    return (old_include, expected_header, f"Target '{target_name}' ({target_info['type']})")


def main():
    if len(sys.argv) < 2:
        print("Usage: fix_export_headers_v4.py <base_dir>")
        sys.exit(1)
    
    base_dir = sys.argv[1]
    if not os.path.isdir(base_dir):
        print(f"Error: {base_dir} is not a directory")
        sys.exit(1)
    
    # Find all CMakeLists.txt files and parse them
    all_targets = {}
    all_src_to_target = {}
    
    for root, dirs, files in os.walk(base_dir):
        if 'CMakeLists.txt' in files:
            cmake_path = os.path.join(root, 'CMakeLists.txt')
            targets, src_to_target = parse_cmake_file(cmake_path, base_dir)
            all_targets.update(targets)
            all_src_to_target.update(src_to_target)
    
    print(f"Found {len(all_targets)} targets")
    shared_module_targets = [t for t, info in all_targets.items() if info['type'] in ['SHARED', 'MODULE']]
    print(f"SHARED/MODULE targets: {len(shared_module_targets)}")
    
    # Find all .h files with export header includes
    files_to_check = []
    for root, dirs, files in os.walk(base_dir):
        for file in files:
            if file.endswith('.h'):
                files_to_check.append(os.path.join(root, file))
    
    print(f"Checking {len(files_to_check)} header files...")
    
    fixed = []
    skipped = []
    errors = []
    
    for file_path in files_to_check:
        rel_path = os.path.relpath(file_path, base_dir)
        
        # Check if file has export header includes
        with open(file_path, 'r', encoding='utf-8', errors='ignore') as f:
            content = f.read()
        
        current_includes = find_export_header_includes(content)
        if not current_includes:
            skipped.append((file_path, "No export header include"))
            continue
        
        # Try to find target for this file
        target_name = None
        
        # Priority 1: Exact relative path match
        if rel_path in all_src_to_target:
            target_name = all_src_to_target[rel_path]
        
        # Priority 2: Filename match (for headers in same dir as CMakeLists.txt)
        if not target_name:
            filename = os.path.basename(file_path)
            for src_path, tgt in all_src_to_target.items():
                if os.path.basename(src_path) == filename:
                    target_name = tgt
                    break
        
        # Priority 3: Parse local CMakeLists.txt
        if not target_name:
            cmake_path = os.path.join(os.path.dirname(file_path), 'CMakeLists.txt')
            if os.path.exists(cmake_path):
                local_targets, local_mapping = parse_cmake_file(cmake_path, base_dir)
                # Check if file is in any local target
                for tgt_name, tgt_info in local_targets.items():
                    for src in tgt_info['sources']:
                        if src == os.path.basename(file_path):
                            target_name = tgt_name
                            break
        
        if not target_name:
            errors.append((file_path, "Could not determine target"))
            continue
        
        # Fix the file
        result = fix_file(file_path, target_name, all_targets)
        
        if result is None:
            skipped.append((file_path, "Already correct or STATIC"))
            continue
        
        old_include, new_include, reason = result
        
        if new_include is None:
            # Remove include for STATIC libraries
            # Find the line with the include and remove it
            lines = content.split('\n')
            new_lines = []
            for line in lines:
                if f'#include' in line and old_include in line:
                    continue  # Skip this line
                new_lines.append(line)
            new_content = '\n'.join(new_lines)
            
            with open(file_path, 'w', encoding='utf-8') as f:
                f.write(new_content)
            
            fixed.append((file_path, old_include, "REMOVED", reason))
        else:
            # Replace include
            # Use regex to handle both "..." and <...>
            pattern = rf'#include\s*[<"]{re.escape(old_include)}[>"]'
            replacement = f'#include "{new_include}"'
            new_content = re.sub(pattern, replacement, content)
            
            with open(file_path, 'w', encoding='utf-8') as f:
                f.write(new_content)
            
            fixed.append((file_path, old_include, new_include, reason))
    
    # Print summary
    print("\n" + "="*80)
    print(f"Summary:")
    print(f"  Fixed:   {len(fixed)}")
    print(f"  Skipped: {len(skipped)}")
    print(f"  Errors:  {len(errors)}")
    print("="*80)
    
    if fixed:
        print("\nFixed files:")
        for item in fixed[:20]:  # Show first 20
            if len(item) == 4:
                file_path, old, new, reason = item
                print(f"  FIX: {os.path.relpath(file_path, base_dir)}")
                print(f"       {old} -> {new}")
                print(f"       ({reason})")
        if len(fixed) > 20:
            print(f"  ... and {len(fixed) - 20} more")
    
    if errors:
        print("\nErrors:")
        for file_path, reason in errors[:20]:
            print(f"  ERROR: {os.path.relpath(file_path, base_dir)} - {reason}")
        if len(errors) > 20:
            print(f"  ... and {len(errors) - 20} more")


if __name__ == '__main__':
    main()