wp/wp-includes/l10n/class-wp-translation-file-mo.php
changeset 21 48c4eec2b7e6
child 22 8c2e4d02f4ef
equal deleted inserted replaced
20:7b1b88e27a20 21:48c4eec2b7e6
       
     1 <?php
       
     2 /**
       
     3  * I18N: WP_Translation_File_MO class.
       
     4  *
       
     5  * @package WordPress
       
     6  * @subpackage I18N
       
     7  * @since 6.5.0
       
     8  */
       
     9 
       
    10 /**
       
    11  * Class WP_Translation_File_MO.
       
    12  *
       
    13  * @since 6.5.0
       
    14  */
       
    15 class WP_Translation_File_MO extends WP_Translation_File {
       
    16 	/**
       
    17 	 * Endian value.
       
    18 	 *
       
    19 	 * V for little endian, N for big endian, or false.
       
    20 	 *
       
    21 	 * Used for unpack().
       
    22 	 *
       
    23 	 * @since 6.5.0
       
    24 	 * @var false|'V'|'N'
       
    25 	 */
       
    26 	protected $uint32 = false;
       
    27 
       
    28 	/**
       
    29 	 * The magic number of the GNU message catalog format.
       
    30 	 *
       
    31 	 * @since 6.5.0
       
    32 	 * @var int
       
    33 	 */
       
    34 	const MAGIC_MARKER = 0x950412de;
       
    35 
       
    36 	/**
       
    37 	 * Detects endian and validates file.
       
    38 	 *
       
    39 	 * @since 6.5.0
       
    40 	 *
       
    41 	 * @param string $header File contents.
       
    42 	 * @return false|'V'|'N' V for little endian, N for big endian, or false on failure.
       
    43 	 */
       
    44 	protected function detect_endian_and_validate_file( string $header ) {
       
    45 		$big = unpack( 'N', $header );
       
    46 
       
    47 		if ( false === $big ) {
       
    48 			return false;
       
    49 		}
       
    50 
       
    51 		$big = reset( $big );
       
    52 
       
    53 		if ( false === $big ) {
       
    54 			return false;
       
    55 		}
       
    56 
       
    57 		$little = unpack( 'V', $header );
       
    58 
       
    59 		if ( false === $little ) {
       
    60 			return false;
       
    61 		}
       
    62 
       
    63 		$little = reset( $little );
       
    64 
       
    65 		if ( false === $little ) {
       
    66 			return false;
       
    67 		}
       
    68 
       
    69 		// Force cast to an integer as it can be a float on x86 systems. See https://core.trac.wordpress.org/ticket/60678.
       
    70 		if ( (int) self::MAGIC_MARKER === $big ) {
       
    71 			return 'N';
       
    72 		}
       
    73 
       
    74 		// Force cast to an integer as it can be a float on x86 systems. See https://core.trac.wordpress.org/ticket/60678.
       
    75 		if ( (int) self::MAGIC_MARKER === $little ) {
       
    76 			return 'V';
       
    77 		}
       
    78 
       
    79 		$this->error = 'Magic marker does not exist';
       
    80 		return false;
       
    81 	}
       
    82 
       
    83 	/**
       
    84 	 * Parses the file.
       
    85 	 *
       
    86 	 * @since 6.5.0
       
    87 	 *
       
    88 	 * @return bool True on success, false otherwise.
       
    89 	 */
       
    90 	protected function parse_file(): bool {
       
    91 		$this->parsed = true;
       
    92 
       
    93 		$file_contents = file_get_contents( $this->file ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents
       
    94 
       
    95 		if ( false === $file_contents ) {
       
    96 			return false;
       
    97 		}
       
    98 
       
    99 		$file_length = strlen( $file_contents );
       
   100 
       
   101 		if ( $file_length < 24 ) {
       
   102 			$this->error = 'Invalid data';
       
   103 			return false;
       
   104 		}
       
   105 
       
   106 		$this->uint32 = $this->detect_endian_and_validate_file( substr( $file_contents, 0, 4 ) );
       
   107 
       
   108 		if ( false === $this->uint32 ) {
       
   109 			return false;
       
   110 		}
       
   111 
       
   112 		$offsets = substr( $file_contents, 4, 24 );
       
   113 
       
   114 		if ( false === $offsets ) {
       
   115 			return false;
       
   116 		}
       
   117 
       
   118 		$offsets = unpack( "{$this->uint32}rev/{$this->uint32}total/{$this->uint32}originals_addr/{$this->uint32}translations_addr/{$this->uint32}hash_length/{$this->uint32}hash_addr", $offsets );
       
   119 
       
   120 		if ( false === $offsets ) {
       
   121 			return false;
       
   122 		}
       
   123 
       
   124 		$offsets['originals_length']    = $offsets['translations_addr'] - $offsets['originals_addr'];
       
   125 		$offsets['translations_length'] = $offsets['hash_addr'] - $offsets['translations_addr'];
       
   126 
       
   127 		if ( $offsets['rev'] > 0 ) {
       
   128 			$this->error = 'Unsupported revision';
       
   129 			return false;
       
   130 		}
       
   131 
       
   132 		if ( $offsets['translations_addr'] > $file_length || $offsets['originals_addr'] > $file_length ) {
       
   133 			$this->error = 'Invalid data';
       
   134 			return false;
       
   135 		}
       
   136 
       
   137 		// Load the Originals.
       
   138 		$original_data     = str_split( substr( $file_contents, $offsets['originals_addr'], $offsets['originals_length'] ), 8 );
       
   139 		$translations_data = str_split( substr( $file_contents, $offsets['translations_addr'], $offsets['translations_length'] ), 8 );
       
   140 
       
   141 		foreach ( array_keys( $original_data ) as $i ) {
       
   142 			$o = unpack( "{$this->uint32}length/{$this->uint32}pos", $original_data[ $i ] );
       
   143 			$t = unpack( "{$this->uint32}length/{$this->uint32}pos", $translations_data[ $i ] );
       
   144 
       
   145 			if ( false === $o || false === $t ) {
       
   146 				continue;
       
   147 			}
       
   148 
       
   149 			$original    = substr( $file_contents, $o['pos'], $o['length'] );
       
   150 			$translation = substr( $file_contents, $t['pos'], $t['length'] );
       
   151 			// GlotPress bug.
       
   152 			$translation = rtrim( $translation, "\0" );
       
   153 
       
   154 			// Metadata about the MO file is stored in the first translation entry.
       
   155 			if ( '' === $original ) {
       
   156 				foreach ( explode( "\n", $translation ) as $meta_line ) {
       
   157 					if ( '' === $meta_line ) {
       
   158 						continue;
       
   159 					}
       
   160 
       
   161 					list( $name, $value ) = array_map( 'trim', explode( ':', $meta_line, 2 ) );
       
   162 
       
   163 					$this->headers[ strtolower( $name ) ] = $value;
       
   164 				}
       
   165 			} else {
       
   166 				/*
       
   167 				 * In MO files, the key normally contains both singular and plural versions.
       
   168 				 * However, this just adds the singular string for lookup,
       
   169 				 * which caters for cases where both __( 'Product' ) and _n( 'Product', 'Products' )
       
   170 				 * are used and the translation is expected to be the same for both.
       
   171 				 */
       
   172 				$parts = explode( "\0", (string) $original );
       
   173 
       
   174 				$this->entries[ $parts[0] ] = $translation;
       
   175 			}
       
   176 		}
       
   177 
       
   178 		return true;
       
   179 	}
       
   180 
       
   181 	/**
       
   182 	 * Exports translation contents as a string.
       
   183 	 *
       
   184 	 * @since 6.5.0
       
   185 	 *
       
   186 	 * @return string Translation file contents.
       
   187 	 */
       
   188 	public function export(): string {
       
   189 		// Prefix the headers as the first key.
       
   190 		$headers_string = '';
       
   191 		foreach ( $this->headers as $header => $value ) {
       
   192 			$headers_string .= "{$header}: $value\n";
       
   193 		}
       
   194 		$entries     = array_merge( array( '' => $headers_string ), $this->entries );
       
   195 		$entry_count = count( $entries );
       
   196 
       
   197 		if ( false === $this->uint32 ) {
       
   198 			$this->uint32 = 'V';
       
   199 		}
       
   200 
       
   201 		$bytes_for_entries = $entry_count * 4 * 2;
       
   202 		// Pair of 32bit ints per entry.
       
   203 		$originals_addr    = 28; /* header */
       
   204 		$translations_addr = $originals_addr + $bytes_for_entries;
       
   205 		$hash_addr         = $translations_addr + $bytes_for_entries;
       
   206 		$entry_offsets     = $hash_addr;
       
   207 
       
   208 		$file_header = pack(
       
   209 			$this->uint32 . '*',
       
   210 			// Force cast to an integer as it can be a float on x86 systems. See https://core.trac.wordpress.org/ticket/60678.
       
   211 			(int) self::MAGIC_MARKER,
       
   212 			0, /* rev */
       
   213 			$entry_count,
       
   214 			$originals_addr,
       
   215 			$translations_addr,
       
   216 			0, /* hash_length */
       
   217 			$hash_addr
       
   218 		);
       
   219 
       
   220 		$o_entries = '';
       
   221 		$t_entries = '';
       
   222 		$o_addr    = '';
       
   223 		$t_addr    = '';
       
   224 
       
   225 		foreach ( array_keys( $entries ) as $original ) {
       
   226 			$o_addr        .= pack( $this->uint32 . '*', strlen( $original ), $entry_offsets );
       
   227 			$entry_offsets += strlen( $original ) + 1;
       
   228 			$o_entries     .= $original . "\0";
       
   229 		}
       
   230 
       
   231 		foreach ( $entries as $translations ) {
       
   232 			$t_addr        .= pack( $this->uint32 . '*', strlen( $translations ), $entry_offsets );
       
   233 			$entry_offsets += strlen( $translations ) + 1;
       
   234 			$t_entries     .= $translations . "\0";
       
   235 		}
       
   236 
       
   237 		return $file_header . $o_addr . $t_addr . $o_entries . $t_entries;
       
   238 	}
       
   239 }