<?php
if ( ! defined( 'ABSPATH' ) ) { exit; }

class AF_Translate_Scanner {

    /**
     * Verifies that operations are strictly locked inside safe, valid system roots.
     */
    private static function is_path_safe( $path ) {
        if ( empty( $path ) ) {
            return false;
        }
        $normalized_path = wp_normalize_path( $path );
        $allowed_roots   = array(
            wp_normalize_path( WP_LANG_DIR ),
            wp_normalize_path( WP_PLUGIN_DIR ),
            wp_normalize_path( get_theme_root() )
        );

        foreach ( $allowed_roots as $root ) {
            if ( strpos( $normalized_path, $root ) === 0 ) {
                return true;
            }
        }
        return false;
    }

    /**
     * Scans the target bundle codebase and generates a standard .pot template file.
     * Hardened against path traversal, heavy execution timeouts, and breaking characters.
     *
     * @param string $slug The component text-domain/directory slug.
     * @param string $type Component type ('theme' or 'plugin').
     * @return true|WP_Error
     */
    public static function generate_pot( $slug, $type ) {
        $slug = sanitize_key( $slug );
        $type = sanitize_key( $type );

        // 1. Resolve and validate source directory path
        $base_dir    = ( 'theme' === $type ) ? get_theme_root() : WP_PLUGIN_DIR;
        $source_path = wp_normalize_path( trailingslashit( $base_dir ) . $slug );

        if ( ! self::is_path_safe( $source_path ) || ! is_dir( $source_path ) ) {
            return new WP_Error( 
                'invalid_source', 
                esc_html__( 'The target resource directory path does not exist or violates environment isolation.', 'af-translate' ) 
            );
        }

        // 2. Locate and secure target save path
        $dirs       = AF_Translate_Analyzer::locate_translation_dirs( $slug, $type );
        $target_dir = wp_normalize_path( $dirs['author'] );

        if ( ! self::is_path_safe( $target_dir ) ) {
            return new WP_Error( 
                'restricted_target', 
                esc_html__( 'The target directory path violates system boundary constraints.', 'af-translate' ) 
            );
        }

        if ( ! is_dir( $target_dir ) && ! wp_mkdir_p( $target_dir ) ) {
            return new WP_Error( 
                'dir_creation_failed', 
                esc_html__( 'Could not establish target languages destination framework.', 'af-translate' ) 
            );
        }

        $pot_file_path = trailingslashit( $target_dir ) . "{$slug}.pot";

        // 3. Scan files recursively for translatable strings
        $strings = self::extract_strings_from_codebase( $source_path, $slug );

        // 4. Construct POT File Schema Architecture
        $pot_content = self::build_pot_header( $slug );
        
        if ( empty( $strings ) ) {
            $pot_content .= "\n# " . esc_html__( 'No strings found matching text domain:', 'af-translate' ) . " {$slug}\n";
        } else {
            foreach ( $strings as $msgid => $references ) {
                foreach ( $references as $ref ) {
                    $pot_content .= "#: " . esc_html( $ref ) . "\n";
                }
                
                // Securely strip and normalize layout escapes matching native gettext token rules
                $clean_msgid   = str_replace( '\\"', '"', $msgid );
                $escaped_msgid = addcslashes( $clean_msgid, '"\\' );
                
                $pot_content .= "msgid \"" . $escaped_msgid . "\"\n";
                $pot_content .= "msgstr \"\"\n\n";
            }
        }

        // 5. Commit compiled raw text safely down to disk layer
        if ( false === file_put_contents( $pot_file_path, $pot_content ) ) {
            return new WP_Error( 
                'write_failed', 
                esc_html__( 'Failed to write generated POT file structure to disk.', 'af-translate' ) 
            );
        }

        return true;
    }

    /**
     * Recursively loops through the directory to grab strings matching the text domain.
     * Memory-optimized to prevent vendor folder memory crashes.
     */
    private static function extract_strings_from_codebase( $dir_path, $domain ) {
        $strings = array();
        
        try {
            $directory = new RecursiveDirectoryIterator( $dir_path, RecursiveDirectoryIterator::SKIP_DOTS );
            $iterator  = new RecursiveIteratorIterator( $directory );
        } catch ( Exception $e ) {
            return $strings;
        }

        // Directories to strictly skip during execution parsing loops
        $excluded_directories = array( 'node_modules', 'vendor', 'assets', 'bower_components', '.git', '.github' );

        // Regex explicitly targets: __, _e, _x, esc_html__, esc_html_e, esc_attr__, esc_attr_e
        $regex = '/\b(?:__|_e|_x|esc_html__|esc_html_e|esc_attr__|esc_attr_e)\(\s*([\'"])(.*?)\1\s*,\s*([\'"])' . preg_quote( $domain, '/' ) . '\3\s*\)/';

        foreach ( $iterator as $file ) {
            if ( ! $file->isFile() || 'php' !== $file->getExtension() ) {
                continue;
            }

            $file_path       = wp_normalize_path( $file->getPathname() );
            $path_segments   = explode( '/', trim( str_replace( wp_normalize_path( $dir_path ), '', $file_path ), '/' ) );
            
            // Check if file belongs to an excluded asset or dependency group
            $skip_file = false;
            foreach ( $excluded_directories as $exclude ) {
                if ( in_array( $exclude, $path_segments, true ) ) {
                    $skip_file = true;
                    break;
                }
            }
            
            if ( $skip_file ) {
                continue;
            }

            $content = file_get_contents( $file_path );
            if ( false === $content ) {
                continue;
            }

            if ( preg_match_all( $regex, $content, $matches, PREG_SET_ORDER ) ) {
                // Anonymize full system structure relative to standard workspace targets
                $relative_file = str_replace( wp_normalize_path( WP_CONTENT_DIR ), '', $file_path );
                
                foreach ( $matches as $match ) {
                    $msgid = $match;
                    
                    if ( empty( $msgid ) ) {
                        continue;
                    }
                    
                    if ( ! isset( $strings[$msgid] ) ) {
                        $strings[$msgid] = array();
                    }
                    
                    $strings[$msgid][] = ltrim( $relative_file, '/' );
                }
            }
        }

        // Deduplicate tracking tags cleanly across references
        foreach ( $strings as $msgid => $refs ) {
            $strings[$msgid] = array_unique( $refs );
        }

        return $strings;
    }

    /**
     * Generates a standard GNU gettext headers architecture.
     */
    private static function build_pot_header( $slug ) {
        $date = date_i18n( 'Y-m-d H:iO' );
        $year = date_i18n( 'Y' );
        
        return "# Copyright (C) {$year} AF Themes\n" .
               "# This file is distributed under the same license as the " . ucfirst( $slug ) . " package.\n" .
               "msgid \"\"\n" .
               "msgstr \"\"\n" .
               "\"Project-Id-Version: " . ucfirst( $slug ) . "\\n\"\n" .
               "\"POT-Creation-Date: {$date}\\n\"\n" .
               "\"PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\\n\"\n" .
               "\"Last-Translator: FULL NAME <EMAIL@ADDRESS>\\n\"\n" .
               "\"Language-Team: LANGUAGE <LL@li.org>\\n\"\n" .
               "\"MIME-Version: 1.0\\n\"\n" .
               "\"Content-Type: text/plain; charset=UTF-8\\n\"\n" .
               "\"Content-Transfer-Encoding: 8bit\\n\"\n" .
               "\"X-Generator: AF Translate Engine v1.0.0\\n\"\n\n";
    }
}