function process($currentXml)
{
$currentDirectory = dirname($this->docvertTransformDirectory);
$documentType = detectDocumentType($currentXml);
switch ($documentType) {
case 'OpenDocument1.0':
//text-colon-section text-colon-style-name="Sect2" text-colon-name="Section2">
$styles = $this->getStyles();
$currentXml = preg_replace('/<office:document-content[^>]*?>/s', "\$0" . $styles, $currentXml);
//displayXmlString($currentXml);
$currentXml = xsltTransform($currentXml, $this->docvertTransformDirectory . 'fix-opendocument-content.xsl');
//displayXmlString($currentXml);
$currentXml = xsltTransform($currentXml, $this->docvertTransformDirectory . 'fix-opendocument-content-stage2.xsl');
//displayXmlString($currentXml);
$currentXml = xsltTransform($currentXml, $currentDirectory . DIRECTORY_SEPARATOR . 'lib' . DIRECTORY_SEPARATOR . 'sun-xsl' . DIRECTORY_SEPARATOR . 'opendocument-content-to-docbook.xsl');
//displayXmlString($this->docvertTransformDirectory.'opendocument-content-to-docbook.xsl');
//displayXmlString($currentXml);
$metaData = $this->getMetaData();
$currentXml = preg_replace('/<db:book[^>]*?>/s', "\$0" . $metaData, $currentXml);
//displayXmlString($currentXml);
$currentXml = xsltTransform($currentXml, $this->docvertTransformDirectory . 'fix-docbook.xsl');
//displayXmlString($currentXml);
break;
case 'OpenOffice1.x':
webServiceError('&error-process-transformopendocumenttodocbook-openofficefile;');
break;
default:
webServiceError('&error-process-transformopendocumenttodocbook-unsupported-file; [' . revealXml($currentXml) . ']', 500, array('documentType' => $documentType));
break;
}
return $currentXml;
}
function postToBloggerApiBlog($host, $port, $xmlRpcPath, $username, $password, $title, $body, $authorName, $authorEmail, $draft, $proxyUsername = null, $proxyPassword = null)
{
$postData = "<entry xmlns='http://www.w3.org/2005/Atom'>\n";
$postData .= "\t<title type='text'>" . revealXml($title) . "</title>\n";
if ($draft) {
$postData .= "\t<app:control xmlns:app='http://purl.org/atom/app#'>\n";
$postData .= "\t\t<app:draft>yes</app:draft>\n";
$postData .= "\t</app:control>\n";
}
$postData .= "\t<content type='xhtml'>\n";
$postData .= $body;
//must be valid XML
$postData .= "\t</content>\n";
$postData .= "\t<author>\n";
$postData .= "\t\t<name>" . revealXml($authorName) . "</name>\n";
$postData .= "\t\t<email>" . revealXml($authorEmail) . "</email>\n";
$postData .= "\t</author>\n";
$postData .= "</entry>";
$textArray['xmlrpc'] = $postData;
//TODO: how to do submit this to the xmlrpc interface?
return pullpage('POST', $host, $port, $xmlRpcPath, $username, $password, $proxyUsername, $proxyPassword, $textArray = null, $binaryArray = null, $debug = false, $timeOutInSeconds = null);
}
}
$pipeline = $unescapedPosts["pipeline"];
if (!preg_match($badCharactersPattern, $pipeline)) {
webServiceError('pipeline contains bad characters. Was "' . revealXml($pipeline) . '"');
}
$autopipeline = $unescapedPosts["autopipeline"];
if (!preg_match($badCharactersPattern, $autopipeline)) {
webServiceError('autopipeline contains bad characters. Was "' . revealXml($autopipeline) . '"');
}
$allhtml = '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en"><body>' . $allhtml . '</body></html>';
$transformDirectory = dirname(__FILE__) . DIRECTORY_SEPARATOR . 'core' . DIRECTORY_SEPARATOR . 'transform' . DIRECTORY_SEPARATOR;
$htmlPagesXsltPath = $transformDirectory . 'htmlpages-to-html.xsl';
$allhtml = characterEntityToNCR($allhtml);
$html = xsltTransform($allhtml, $htmlPagesXsltPath);
if (!$html) {
webServiceError("Unable to rebuild document, either because I was given invalid XML or there was a programming error.<hr />Document was:<blockquote>" . str_replace("\n", "<br />", str_replace(" ", ' ', revealXml($allhtml))) . "</blockquote>");
}
//displayXmlString($html);
$docbookBodyXsltPath = $transformDirectory . 'html-to-docbook-body.xsl';
$docbookBody = xsltTransform($html, $docbookBodyXsltPath);
$docbookBody = removeXmlDeclaration($docbookBody);
$docbookBody = preg_replace("/<docvert-remove-me[^>]*?>/", '', $docbookBody);
$docbookBody = preg_replace("/<\\/docvert-remove-me[^>]*?>/", '', $docbookBody);
//displayXmlString($docbookBody);
//displayXmlString(file_get_contents($docbookBodyXsltPath));
$allDocumentsPreviewDirectory = dirname(__FILE__) . DIRECTORY_SEPARATOR . 'writable' . DIRECTORY_SEPARATOR . $documentPathParts[0] . DIRECTORY_SEPARATOR;
$previewDirectory = $allDocumentsPreviewDirectory . $documentPathParts[1] . DIRECTORY_SEPARATOR;
$unitTestResults = $previewDirectory . 'test.html';
if (file_exists($unitTestResults)) {
silentlyUnlink($unitTestResults);
}
foreach ($filesToDisplay as $fileToDisplay) {
$possibleFile = getFirstByPattern($pathToUse . $fileToDisplay);
if ($possibleFile) {
$chosenFile = $possibleFile;
break;
}
}
if (!$chosenFile) {
$filesToDisplayAsString = null;
foreach ($filesToDisplay as $fileToDisplay) {
$filesToDisplayAsString .= '"' . $fileToDisplay . '", ';
}
$filesToDisplayAsString = trim($filesToDisplayAsString);
$filesToDisplayAsString = substr($filesToDisplayAsString, 0, strlen($filesToDisplayAsString) - 1);
$filesInPreviewDirectory = glob($pathToUse . '*');
die('Docvert or pipeline error: Unable to determine the file to preview. I searched for the filename patterns ' . $filesToDisplayAsString . ' were tested but do not exist. Was given pathToUse of <tt>"' . $pathToUse . '"</tt> which contained <pre>' . revealXml(print_r($filesInPreviewDirectory, true)) . '</pre>');
}
$chosenFile = str_replace('\\', '/', $chosenFile);
if ($thereIsAPreview) {
print '<html>';
print '<head>';
print '<frameset cols="80%, *" border="10" bordercolor="#333333" frameborder="15">';
print '<frame src="' . $chosenFile . '" id="contentFrame"/>';
print '<frame src ="' . dirname($chosenFile) . '/test.html"/>';
print '</frameset>';
print '</head>';
print '<body><noframes>Docvert requires a frames-compatible browser</noframes></body>';
print '</html>';
} else {
print '<html>';
print '<head>';
function setupOpenOfficeOrgServer()
{
$operatingSystemFamily = getOperatingSystemFamily();
if (!$this->allowedAdminAccess || $operatingSystemFamily != 'Unix') {
return;
}
$hideOpenOfficeOrgServerOption = getGlobalConfigItem('hideAdminOptionOpenOfficeOrgServer');
if ($hideOpenOfficeOrgServerOption == "true") {
return;
}
$userMessage = null;
$pidFile = '/tmp/openoffice.org-server.pid';
$output = '';
if (isset($_REQUEST['openofficeorg-server-on']) || isset($_REQUEST['openofficeorg-server-off'])) {
$unixConfigPath = dirname(__FILE__) . DIRECTORY_SEPARATOR . 'config' . DIRECTORY_SEPARATOR . 'unix-specific' . DIRECTORY_SEPARATOR;
$startStopBashScript = $unixConfigPath . 'openoffice.org-server-init.sh';
$startOrStop = null;
if (isset($_REQUEST['openofficeorg-server-on'])) {
if (file_exists($pidFile) && filesize($pidFile) > 0) {
$userMessage = 'existing-pid';
} else {
$startOrStop = 'start';
}
} elseif (isset($_REQUEST['openofficeorg-server-off'])) {
if (file_exists($pidFile) && filesize($pidFile) > 0) {
$startOrStop = 'stop';
} else {
$userMessage = 'no-existing-pid';
}
}
if ($startOrStop) {
include_once dirname(__FILE__) . '/config.php';
$runAsUser = '';
$sudo = '';
$startTime = 8;
$stopTime = 3;
$customUser = getGlobalConfigItem('runExternalApplicationAsUser');
if ($customUser) {
$runAsUser = $customUser;
}
$startStopScript = null;
$superUserPreference = getSuperUserPreference();
if ($superUserPreference == 'sudo') {
$startStopScript = $startStopBashScript;
$sudo = 'sudo -u {runAsUser}';
} elseif ($superUserPreference == 'nothing') {
$startStopScript = $startStopBashScript;
$sudo = '';
}
$commandTemplate = '{sudo} {startStopScript} {startOrStop} {runAsUser}';
$commandTemplate = str_replace('{sudo}', $sudo, $commandTemplate);
$commandTemplate = str_replace('{startStopScript}', $startStopScript, $commandTemplate);
$commandTemplate = str_replace('{startOrStop}', $startOrStop, $commandTemplate);
$commandTemplate = str_replace('{runAsUser}', $runAsUser, $commandTemplate);
$commandTemplate = trim($commandTemplate);
$output = shellCommand($commandTemplate, 0);
switch ($startOrStop) {
// due to OOo delay in startup/shutdown we'll just twiddle our thumbs for a bit
case 'start':
sleep($startTime);
break;
case 'stop':
sleep($stopTime);
break;
}
if (isset($_REQUEST['openofficeorg-server-on']) || isset($_REQUEST['openofficeorg-server-off'])) {
$output = '<blockquote><tt>' . revealXml($commandTemplate) . '</tt></blockquote><blockquote><tt>' . revealXml($output) . '</tt></blockquote>';
if (isset($_REQUEST['openofficeorg-server-on']) && !file_exists($pidFile)) {
webServiceError('&error-unable-to-start-ooo-server; ' . $output . ' ');
} elseif (isset($_REQUEST['openofficeorg-server-off']) && file_exists($pidFile)) {
webServiceError('&error-unable-to-stop-ooo-server; ' . $output . ' ');
}
}
}
}
if (!is_readable($pidFile)) {
$response = $this->getThemeFragment('admin-setupopenofficeorg-server-button-unknown.htmlf');
} elseif (file_exists($pidFile) && filesize($pidFile) > 0) {
$response = $this->getThemeFragment('admin-setupopenofficeorg-server-button-disabled.htmlf');
} else {
$response = $this->getThemeFragment('admin-setupopenofficeorg-server-button-enabled.htmlf');
}
//if($userMessage)
// {
// $response = $response.$this->getThemeFragment('error-setupopenofficeorg-server-'.$userMessage.'.htmlf');
// }
if ($output) {
$response = $response . '<p>Output:</p><pre style="padding:0px 0.5em;font-size:small;background:#cccccc">' . $output . '</pre>';
}
return $response;
}
function generateDocument($pages, $generatorPipeline)
{
if (preg_match('/.\\//s', $generatorPipeline)) {
webServiceError('&error-disallowed-characters;');
}
$userAgent = 'Mozilla/5.0 (X11; U; Linux i686; en-US; rv:bignumber) Docvert';
$httpContextOptions = array('http' => array('header' => 'User-Agent: ' . $userAgent));
$httpContext = stream_context_create($httpContextOptions);
$docvertDir = dirname(dirname(__FILE__)) . DIRECTORY_SEPARATOR;
$docvertWritableDir = getWritableDirectory();
$disallowDocumentGeneration = getGlobalConfigItem('doNotAllowDocumentGeneration');
if ($disallowDocumentGeneration == 'true') {
webServiceError('&document-generation-disabled;');
}
$pageXml = '<c:document xmlns="http://www.w3.org/1999/xhtml" xmlns:c="container">' . "\n";
$pageTemplate = "\n\t" . '<c:page url="{{url}}" {{baseUrl}}>{{page}}</c:page>' . "\n";
$config = array('indent' => true, 'output-xhtml' => true, 'wrap' => 200);
if (!class_exists('tidy')) {
webServiceError('&tidy-is-not-installed;');
}
$tidy = new tidy();
$baseTagPattern = "/<base[^>]*?href=([^>]*?)>/is";
foreach ($pages as $page) {
if (trim($page) != '' && (stringStartsWith($page, 'http://') || stringStartsWith($page, 'https://'))) {
$pageHtml = file_get_contents($page, null, $httpContext);
$tidy->parseString($pageHtml, $config, 'utf8');
$tidy->cleanRepair();
$thisPage = str_replace('{{url}}', $page, $pageTemplate);
$baseUrl = '';
//supporting that ugly old hack of <base>
preg_match($baseTagPattern, $pageHtml, $matches);
if (count($matches) > 0) {
$baseUrl = 'baseUrl="' . substr($matches[1], 1, -2) . '"';
}
$thisPage = str_replace('{{baseUrl}}', $baseUrl, $thisPage);
$tidiedPageContents = characterEntityToNCR(removeDoctype(removeXmlComments($tidy)));
$styleTagPattern = "/<style.*?<\\/style>/is";
$tidiedPageContents = preg_replace($styleTagPattern, '', $tidiedPageContents);
$scriptTagPattern = "/<script.*?<\\/script>/is";
$tidiedPageContents = preg_replace($scriptTagPattern, '', $tidiedPageContents);
$questionMarkPattern = "/<\\?.*?\\?>/is";
//as strangely used on news.yahoo.com
$tidiedPageContents = preg_replace($questionMarkPattern, '', $tidiedPageContents);
$thisPage = str_replace('{{page}}', $tidiedPageContents, $thisPage);
$pageXml .= $thisPage;
}
}
$pageXml .= '</c:document>';
$temporaryDirectory = getTemporaryDirectory();
$pipelineDirectory = dirname(dirname(__FILE__)) . DIRECTORY_SEPARATOR . 'generator-pipeline' . DIRECTORY_SEPARATOR . $generatorPipeline . DIRECTORY_SEPARATOR;
$pipelinePath = $pipelineDirectory . 'pipeline.xml';
if (!file_exists($pipelinePath)) {
webServiceError('&generation-pipeline-not-found; ' . revealXml($pipelinePath));
}
$pipelineString = file_get_contents($pipelinePath);
$pipelineString = substr($pipelineString, strpos($pipelineString, '<pipeline>') + 10);
$pipelineString = substr($pipelineString, 0, strpos($pipelineString, '</pipeline>'));
$pipelineStages = xmlStringToArray($pipelineString);
$pipelineSettings = array("pipeline" => $generatorPipeline, "autopipeline" => $generatorPipeline);
processAPipelineLevel($pipelineStages, $pageXml, $pipelineDirectory, $temporaryDirectory, $temporaryDirectory, $pipelineSettings);
$openDocumentPath = $temporaryDirectory . 'output.odt';
zipFiles($temporaryDirectory, $openDocumentPath);
header('Content-disposition: attachment; filename=' . basename($openDocumentPath));
header('Content-type: application/vnd.oasis.opendocument.text');
readfile($openDocumentPath);
}
function pullpage($method, $host, $port, $usepath, $username, $password, $proxyUsername, $proxyPassword, $textArray = null, $binaryArray = null, $debug = false, $timeOutInSeconds = null)
{
//Used as pullpage ("GET" || "POST", "localhost", "80", "/blah.php", $textArray['key'] = $value, $binaryArray['key'] = $value['type' || 'binary']);
//TODO: use parse_url rather than asking for url components
$output = null;
if ($binaryArray && $method == 'GET') {
die('Can\'t method=GET with binaries.');
}
$userAgent = "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:bignumber) Docvert";
if ($timeOutInSeconds == null) {
$timeOutInSeconds = 30;
}
$protocol = 'http';
if ($port == 443) {
$protocol = 'https';
}
$fp = fsockopen($host, $port, $errno, $errstr, $timeOutInSeconds);
stream_set_blocking($fp, false);
$response = null;
if (!$fp) {
return false;
//webServiceError($errstr.' ('.$errno.') at host "'.$host.'", port "'.$port.'", <br>'."\n", 502);
}
stream_set_timeout($fp, $timeOutInSeconds);
$request = null;
$request['head'] = null;
$request['body'] = null;
switch ($method) {
case 'GET':
case 'POST':
case 'PULL':
case 'MKCOL':
case 'PUT':
$request['head'] .= httpLine($method . ' ' . $usepath . ' HTTP/1.1');
break;
case 'HEAD':
$request['head'] .= httpLine($method . ' ' . $protocol . '://' . $host . $usepath . ' HTTP/1.1');
break;
default:
webServiceError('&error-http-unknown-method;', 500, array('method' => $method));
}
$request['head'] .= httpLine('User-Agent: ' . $userAgent);
$request['head'] .= httpLine('Accept: */*');
$request['head'] .= httpLine('Host: ' . $host);
// does this need ':'.$port
$request['head'] .= httpLine('Keep-Alive: 300');
$request['head'] .= httpLine('Connection: keep-alive');
if ($username . $password) {
$request['head'] .= httpLine('Authorization: Basic ' . base64_encode($username . ':' . $password));
}
if ($proxyUsername . $proxyPassword) {
$request['head'] .= httpLine('Proxy-Authorization: Basic ' . base64_encode($proxyUsername . ':' . $proxyPassword));
}
if ($binaryArray && $method == "POST") {
srand((double) microtime() * 1000000);
$boundary = "---------------------------" . substr(md5(rand(0, 32000)), 0, 10);
$request['body'] .= '--' . $boundary;
$request['head'] .= httpLine('Content-type: multipart/form-data; boundary=' . $boundary);
$arrayKeys = array_keys($binaryArray);
$formItemCount = 0;
foreach ($binaryArray as $key => $value) {
$formItemCount++;
$request['body'] .= "\r\n";
$request['body'] .= httpLine('Content-Disposition: form-data; name="' . $key . '"; filename="' . $value['name'] . '"');
$request['body'] .= httpLine('Content-Type: ' . $value['type']);
$request['body'] .= httpLine('Content-Transfer-Encoding: binary');
$request['body'] .= "\r\n";
$request['body'] .= $value['binary'];
$request['body'] .= "\r\n--" . $boundary;
}
$arrayKeys = array_keys($textArray);
foreach ($textArray as $key => $value) {
$request['body'] .= "\r\n";
$request['body'] .= httpLine('Content-Disposition: form-data; name="' . $key . '"');
$request['body'] .= "\r\n";
$request['body'] .= $value;
$request['body'] .= "\r\n--" . $boundary;
}
$request['body'] .= "\r\n\r\n";
}
if ($binaryArray && $method == "PUT") {
if (count($binaryArray) != 1) {
webServiceError("&error-http-put-multiple-files", 500, array('numberOfFiles' => count($binaryArray)));
}
$request['head'] .= httpLine('Content-type: application/octet-stream');
$request['body'] = $binaryArray[0];
} elseif ($textArray) {
$request['head'] .= httpLine('Content-type: application/x-www-form-urlencoded');
$arrayKeys = array_keys($textArray);
foreach ($textArray as $key => $value) {
if ($request['body'] != null) {
$request['body'] .= "&";
}
$request['body'] .= $key . '=' . urlencode($value);
}
}
$request['head'] .= httpLine('Content-length: ' . strlen($request['body']));
$request['head'] .= "\r\n";
$request['body'] .= "\r\n";
fputs($fp, $request['head'] . $request['body']);
$divisionString = "\r\n\r\n";
switch ($method) {
case 'MKCOL':
case 'PUT':
$exitAfterLoops = 0;
while (stripos($output, "\n") === False) {
$output .= fgets($fp, 8);
$exitAfterLoops++;
if ($exitAfterLoops > 30) {
break;
}
}
fclose($fp);
$output = trim($output);
$outputParts = explode(' ', $output);
$httpCode = $outputParts[1];
$httpMessage = '';
for ($i = 2; $i < count($outputParts); $i++) {
$httpMessage .= trim(' ' . $outputParts[$i]);
}
return $httpCode;
break;
//print $method.' = '.$httpCode.":".$httpMessage."<br />";
//print $method.' = '.$httpCode.":".$httpMessage."<br />";
case 'HEAD':
while (stripos($output, $divisionString) === False && !feof($fp)) {
$output .= fgets($fp, 8);
}
fclose($fp);
return $output;
break;
default:
if ($debug) {
print "Started feof at " . date(DATE_RFC822) . '<br />';
}
while (!feof($fp)) {
$info = stream_get_meta_data($fp);
if ($info['timed_out']) {
break;
}
$output .= fgets($fp, 4096);
}
if ($debug) {
print "Ended feof at " . date(DATE_RFC822) . '<br />';
}
}
fclose($fp);
if ($output) {
$divisionPosition = strpos($output, $divisionString);
if (!$divisionPosition) {
webServiceError("&error-http-segment;", 500, array('output' => revealXml($output)));
}
$httpHeader = substr($output, 0, $divisionPosition);
$httpHeader = explode("\n", $httpHeader);
$httpBody = substr($output, $divisionPosition + strlen($divisionString));
$response['head'] = $httpHeader;
$response['body'] = $httpBody;
}
if ($debug) {
print '<h1>Debug</h1>';
print '<p>Connection is ' . $method . ' to ' . $host . ':' . $port . $usepath . '. Scroll down to see request and response.</p>';
print '<h2>Request</h2>';
print '<div style="margin:0px 3%">';
print nl2br(textEncodingToHtmlEncoding($request['head'] . $request['body']));
print "</div>";
print '<h2>Response</h2>';
print '<h3>Head</h3>';
print '<div style="margin:0px 3%">';
if (is_array($response) && array_key_exists('head', $response)) {
foreach ($response['head'] as $eachHeader) {
print $eachHeader . "<br />";
}
} else {
print "No header in response.";
}
print "</div>";
print "<h3>Body</h3>";
print '<div style="margin:0px 3%">';
if (is_array($response) && array_key_exists('body', $response)) {
print nl2br(textEncodingToHtmlEncoding($response['body']));
} else {
print "No body in response.";
}
print "</div>";
die;
}
return $response;
}
