function _locale_import_read_po

Parses Gettext Portable Object file into an array

Parameters

$op: Storage operation type: db-store or mem-store.

$file: Drupal file object corresponding to the PO file to import.

$mode: Should existing translations be replaced LOCALE_IMPORT_KEEP or LOCALE_IMPORT_OVERWRITE.

$lang: Language code.

$group: Text group to import PO file into (eg. 'default' for interface translations).

Related topics

File

includes/locale.inc, line 723

Code

function _locale_import_read_po($op, $file, $mode = NULL, $lang = NULL, $group = 'default') {
    // The file will get closed by PHP on returning from this function.
    $fd = fopen($file->uri, 'rb');
    if (!$fd) {
        _locale_import_message('The translation import failed, because the file %filename could not be read.', $file);
        return FALSE;
    }
    
    /*
     * The parser context. Can be:
     *  - 'COMMENT' (#)
     *  - 'MSGID' (msgid)
     *  - 'MSGID_PLURAL' (msgid_plural)
     *  - 'MSGCTXT' (msgctxt)
     *  - 'MSGSTR' (msgstr or msgstr[])
     *  - 'MSGSTR_ARR' (msgstr_arg)
     */
    $context = 'COMMENT';
    // Current entry being read.
    $current = array();
    // Current plurality for 'msgstr[]'.
    $plural = 0;
    // Current line.
    $lineno = 0;
    while (!feof($fd)) {
        // Refresh the time limit every 10 parsed rows to ensure there is always
        // enough time to import the data for large PO files.
        if (!($lineno % 10)) {
            drupal_set_time_limit(30);
        }
        // A line should not be longer than 10 * 1024.
        $line = fgets($fd, 10 * 1024);
        if ($lineno == 0) {
            // The first line might come with a UTF-8 BOM, which should be removed.
            $line = str_replace("", '', $line);
        }
        $lineno++;
        // Trim away the linefeed.
        $line = trim(strtr($line, array(
            "\\\n" => "",
        )));
        if (!strncmp('#', $line, 1)) {
            // Lines starting with '#' are comments.
            if ($context == 'COMMENT') {
                // Already in comment token, insert the comment.
                $current['#'][] = substr($line, 1);
            }
            elseif ($context == 'MSGSTR' || $context == 'MSGSTR_ARR') {
                // We are currently in string token, close it out.
                _locale_import_one_string($op, $current, $mode, $lang, $file, $group);
                // Start a new entry for the comment.
                $current = array();
                $current['#'][] = substr($line, 1);
                $context = 'COMMENT';
            }
            else {
                // A comment following any other token is a syntax error.
                _locale_import_message('The translation file %filename contains an error: "msgstr" was expected but not found on line %line.', $file, $lineno);
                return FALSE;
            }
        }
        elseif (!strncmp('msgid_plural', $line, 12)) {
            // A plural form for the current message.
            if ($context != 'MSGID') {
                // A plural form cannot be added to anything else but the id directly.
                _locale_import_message('The translation file %filename contains an error: "msgid_plural" was expected but not found on line %line.', $file, $lineno);
                return FALSE;
            }
            // Remove 'msgid_plural' and trim away whitespace.
            $line = trim(substr($line, 12));
            // At this point, $line should now contain only the plural form.
            $quoted = _locale_import_parse_quoted($line);
            if ($quoted === FALSE) {
                // The plural form must be wrapped in quotes.
                _locale_import_message('The translation file %filename contains a syntax error on line %line.', $file, $lineno);
                return FALSE;
            }
            // Append the plural form to the current entry.
            $current['msgid'] .= "\x00" . $quoted;
            $context = 'MSGID_PLURAL';
        }
        elseif (!strncmp('msgid', $line, 5)) {
            // Starting a new message.
            if ($context == 'MSGSTR' || $context == 'MSGSTR_ARR') {
                // We are currently in a message string, close it out.
                _locale_import_one_string($op, $current, $mode, $lang, $file, $group);
                // Start a new context for the id.
                $current = array();
            }
            elseif ($context == 'MSGID') {
                // We are currently already in the context, meaning we passed an id with no data.
                _locale_import_message('The translation file %filename contains an error: "msgid" is unexpected on line %line.', $file, $lineno);
                return FALSE;
            }
            // Remove 'msgid' and trim away whitespace.
            $line = trim(substr($line, 5));
            // At this point, $line should now contain only the message id.
            $quoted = _locale_import_parse_quoted($line);
            if ($quoted === FALSE) {
                // The message id must be wrapped in quotes.
                _locale_import_message('The translation file %filename contains a syntax error on line %line.', $file, $lineno);
                return FALSE;
            }
            $current['msgid'] = $quoted;
            $context = 'MSGID';
        }
        elseif (!strncmp('msgctxt', $line, 7)) {
            // Starting a new context.
            if ($context == 'MSGSTR' || $context == 'MSGSTR_ARR') {
                // We are currently in a message, start a new one.
                _locale_import_one_string($op, $current, $mode, $lang, $file, $group);
                $current = array();
            }
            elseif (!empty($current['msgctxt'])) {
                // A context cannot apply to another context.
                _locale_import_message('The translation file %filename contains an error: "msgctxt" is unexpected on line %line.', $file, $lineno);
                return FALSE;
            }
            // Remove 'msgctxt' and trim away whitespaces.
            $line = trim(substr($line, 7));
            // At this point, $line should now contain the context.
            $quoted = _locale_import_parse_quoted($line);
            if ($quoted === FALSE) {
                // The context string must be quoted.
                _locale_import_message('The translation file %filename contains a syntax error on line %line.', $file, $lineno);
                return FALSE;
            }
            $current['msgctxt'] = $quoted;
            $context = 'MSGCTXT';
        }
        elseif (!strncmp('msgstr[', $line, 7)) {
            // A message string for a specific plurality.
            if ($context != 'MSGID' && $context != 'MSGCTXT' && $context != 'MSGID_PLURAL' && $context != 'MSGSTR_ARR') {
                // Message strings must come after msgid, msgxtxt, msgid_plural, or other msgstr[] entries.
                _locale_import_message('The translation file %filename contains an error: "msgstr[]" is unexpected on line %line.', $file, $lineno);
                return FALSE;
            }
            // Ensure the plurality is terminated.
            if (strpos($line, ']') === FALSE) {
                _locale_import_message('The translation file %filename contains a syntax error on line %line.', $file, $lineno);
                return FALSE;
            }
            // Extract the plurality.
            $frombracket = strstr($line, '[');
            $plural = substr($frombracket, 1, strpos($frombracket, ']') - 1);
            // Skip to the next whitespace and trim away any further whitespace, bringing $line to the message data.
            $line = trim(strstr($line, " "));
            $quoted = _locale_import_parse_quoted($line);
            if ($quoted === FALSE) {
                // The string must be quoted.
                _locale_import_message('The translation file %filename contains a syntax error on line %line.', $file, $lineno);
                return FALSE;
            }
            $current['msgstr'][$plural] = $quoted;
            $context = 'MSGSTR_ARR';
        }
        elseif (!strncmp("msgstr", $line, 6)) {
            // A string for the an id or context.
            if ($context != 'MSGID' && $context != 'MSGCTXT') {
                // Strings are only valid within an id or context scope.
                _locale_import_message('The translation file %filename contains an error: "msgstr" is unexpected on line %line.', $file, $lineno);
                return FALSE;
            }
            // Remove 'msgstr' and trim away away whitespaces.
            $line = trim(substr($line, 6));
            // At this point, $line should now contain the message.
            $quoted = _locale_import_parse_quoted($line);
            if ($quoted === FALSE) {
                // The string must be quoted.
                _locale_import_message('The translation file %filename contains a syntax error on line %line.', $file, $lineno);
                return FALSE;
            }
            $current['msgstr'] = $quoted;
            $context = 'MSGSTR';
        }
        elseif ($line != '') {
            // Anything that is not a token may be a continuation of a previous token.
            $quoted = _locale_import_parse_quoted($line);
            if ($quoted === FALSE) {
                // The string must be quoted.
                _locale_import_message('The translation file %filename contains a syntax error on line %line.', $file, $lineno);
                return FALSE;
            }
            // Append the string to the current context.
            if ($context == 'MSGID' || $context == 'MSGID_PLURAL') {
                $current['msgid'] .= $quoted;
            }
            elseif ($context == 'MSGCTXT') {
                $current['msgctxt'] .= $quoted;
            }
            elseif ($context == 'MSGSTR') {
                $current['msgstr'] .= $quoted;
            }
            elseif ($context == 'MSGSTR_ARR') {
                $current['msgstr'][$plural] .= $quoted;
            }
            else {
                // No valid context to append to.
                _locale_import_message('The translation file %filename contains an error: there is an unexpected string on line %line.', $file, $lineno);
                return FALSE;
            }
        }
    }
    // End of PO file, closed out the last entry.
    if ($context == 'MSGSTR' || $context == 'MSGSTR_ARR') {
        _locale_import_one_string($op, $current, $mode, $lang, $file, $group);
    }
    elseif ($context != 'COMMENT') {
        _locale_import_message('The translation file %filename ended unexpectedly at line %line.', $file, $lineno);
        return FALSE;
    }
}