/**
* Replace any current contents with header and entries from PO souce string
*
* @param string $source po formatted string to parse
*
* @return void
*
* @throws UnrecognizedInputException
*/
public function parsePoSource($source)
{
/**
* This is an incredibly ugly regex pattern that breaks a line of a po file into
* pieces that can be analyzed and acted upon.
*
* The matches array in preg_match will break out like this:
* [0] full string
* [1] mostly useless broad match of initial token, including trailing space
* [2] bare token, or full msgstr[n] clause
* [3] 'n' of a msgstr[n] line
* [4] '"' if a data line
* [5] remaining line
* [6] a bare or malformed comment
*/
$pattern = '/(^(#|#.|#;|#,|#\\||msgid|msgid_plural|msgctxt|msgstr|msgstr\\[([0-9]+)\\])\\s|(^"))(.+)|(^#.*)/';
$source_lines = explode("\n", $source);
$wsBreak = false;
$inHeader = true;
$headerEntry = new PoHeader();
$entry = $headerEntry;
$unrecognized = array();
$lastKey = '';
$currentPlural = 0;
foreach ($source_lines as $line => $s) {
$result = preg_match($pattern, $s, $matches);
if (!$result) {
$lastKey = '';
if ($s == '' || ctype_space($s)) {
if ($inHeader) {
$this->setHeaderEntry($headerEntry);
$entry = null;
$inHeader = false;
}
if (!$wsBreak) {
if (!($entry === null)) {
$this->addEntry($entry);
}
$entry = null;
$wsBreak = true;
}
} else {
$wsBreak = false;
$unrecognized[$line + 1] = $s;
}
} else {
if ($entry === null) {
$entry = new PoEntry();
}
$wsBreak = false;
$currentKey = $matches[2];
// will be used to set last key
switch ($matches[2]) {
case PoTokens::TRANSLATOR_COMMENTS:
case PoTokens::EXTRACTED_COMMENTS:
case PoTokens::REFERENCE:
case PoTokens::FLAG:
case PoTokens::OBSOLETE:
case PoTokens::PREVIOUS:
$entry->add($matches[2], $matches[5]);
break;
case PoTokens::CONTEXT:
case PoTokens::MESSAGE:
case PoTokens::PLURAL:
case PoTokens::TRANSLATED:
$entry->addQuoted($matches[2], $matches[5]);
break;
default:
if ($matches[4] == PoTokens::CONTINUED_DATA) {
$currentKey = $lastKey;
// keep the previous key
if ($currentKey == PoTokens::TRANSLATED_PLURAL) {
$entry->addQuotedAtPosition(PoTokens::TRANSLATED, $currentPlural, '"' . $matches[5]);
} else {
$entry->addQuoted($currentKey, '"' . $matches[5]);
}
} elseif (substr($matches[2], 0, 7) == PoTokens::TRANSLATED_PLURAL) {
$currentKey = PoTokens::TRANSLATED_PLURAL;
$currentPlural = $matches[3];
$entry->addQuotedAtPosition(PoTokens::TRANSLATED, $currentPlural, $matches[5]);
} elseif (isset($matches[6][0]) && $matches[6][0] == PoTokens::TRANSLATOR_COMMENTS) {
$value = substr($matches[6], 1);
$value = empty($value) ? '' : $value;
$entry->add(PoTokens::TRANSLATOR_COMMENTS, $value);
} else {
$unrecognized[$line + 1] = $s;
}
break;
}
$lastKey = $currentKey;
}
}
if (!($entry === null)) {
$this->addEntry($entry);
}
// throw at the very end, anything recognized has been processed
$this->unrecognizedInput = $unrecognized;
if (count($unrecognized)) {
throw new UnrecognizedInputException();
}
}
/**
* Check the supplied entry for sprintf directives and set php-format flag if found
*
* @param PoEntry $entry entry to check
*
* @return void
*/
public function checkPhpFormatFlag(PoEntry $entry)
{
if (preg_match('#(?<!%)%(?:\\d+\\$)?[+-]?(?:[ 0]|\'.{1})?-?\\d*(?:\\.\\d+)?[bcdeEufFgGosxX]#', $entry->get(PoTokens::MESSAGE) . $entry->get(PoTokens::PLURAL))) {
$entry->addFlag('php-format');
}
}
