private function handleLogEvent(PhutilEvent $event)
{
$daemon = $this->getDaemon($event->getValue('id'));
// TODO: This is a bit awkward for historical reasons, clean it up after
// removing Conduit.
$message = $event->getValue('message');
$context = $event->getValue('context');
if (strlen($context) && $context !== $message) {
$message = "({$context}) {$message}";
}
$type = $event->getValue('type');
$message = phutil_utf8ize($message);
id(new PhabricatorDaemonLogEvent())->setLogID($daemon->getID())->setLogType($type)->setMessage((string) $message)->setEpoch(time())->save();
switch ($type) {
case 'WAIT':
$current_status = PhabricatorDaemonLog::STATUS_WAIT;
break;
default:
$current_status = PhabricatorDaemonLog::STATUS_RUNNING;
break;
}
if ($current_status !== $daemon->getStatus()) {
$daemon->setStatus($current_status)->save();
}
}
private static function truncateValue($value, $length)
{
if (is_string($value)) {
if (strlen($value) <= $length) {
return $value;
} else {
// NOTE: PhutilUTF8StringTruncator has huge runtime for giant strings.
return phutil_utf8ize(substr($value, 0, $length) . "\n<...>");
}
} else {
if (is_array($value)) {
foreach ($value as $key => $v) {
if ($length <= 0) {
$value['<...>'] = '<...>';
unset($value[$key]);
} else {
$v = self::truncateValue($v, $length);
$length -= strlen($v);
$value[$key] = $v;
}
}
return $value;
} else {
return $value;
}
}
}
public function testUTF8ize_owl_isCuteAndFerocious()
{
// This was once a ferocious owl when we used to use "?" as the replacement
// character instead of U+FFFD, but now he is sort of not as cute or
// ferocious.
$input = "M(o���o)M";
$expect = "M(o���o)M";
$result = phutil_utf8ize($input);
$this->assertEqual($expect, $result);
}
/**
* This method is kind of awkward here but both the SVN message and
* change parsers use it.
*/
protected function getSVNLogXMLObject($uri, $revision, $verbose = false)
{
if ($verbose) {
$verbose = '--verbose';
}
list($xml) = $this->repository->execxRemoteCommand("log --xml {$verbose} --limit 1 %s@%d", $uri, $revision);
// Subversion may send us back commit messages which won't parse because
// they have non UTF-8 garbage in them. Slam them into valid UTF-8.
$xml = phutil_utf8ize($xml);
return new SimpleXMLElement($xml);
}
public function __construct($status_code, $body)
{
// NOTE: Avoiding phutil_utf8_shorten() here because this isn't lazy
// and responses may be large.
if (strlen($body) > 512) {
$excerpt = substr($body, 0, 512) . '...';
} else {
$excerpt = $body;
}
$this->excerpt = phutil_utf8ize($excerpt);
parent::__construct($status_code);
}
private function parseSVNLogXML($xml)
{
$xml = phutil_utf8ize($xml);
$result = array();
$log = new SimpleXMLElement($xml);
foreach ($log->logentry as $entry) {
$commit = (int) $entry['revision'];
$epoch = (int) strtotime((string) $entry->date[0]);
$result[$commit] = $epoch;
}
return $result;
}
public function parseCommit(PhabricatorRepository $repository, PhabricatorRepositoryCommit $commit)
{
// NOTE: %B was introduced somewhat recently in git's history, so pull
// commit message information with %s and %b instead.
// Even though we pass --encoding here, git doesn't always succeed, so
// we try a little harder, since git *does* tell us what the actual encoding
// is correctly (unless it doesn't; encoding is sometimes empty).
list($info) = $repository->execxLocalCommand('log -n 1 --encoding=%s --format=%s %s --', 'UTF-8', implode('%x00', array('%e', '%cn', '%ce', '%an', '%ae', '%s%n%n%b')), $commit->getCommitIdentifier());
$parts = explode("", $info);
$encoding = array_shift($parts);
// See note above - git doesn't always convert the encoding correctly.
$do_convert = false;
if (strlen($encoding) && strtoupper($encoding) != 'UTF-8') {
if (function_exists('mb_convert_encoding')) {
$do_convert = true;
}
}
foreach ($parts as $key => $part) {
if ($do_convert) {
$parts[$key] = mb_convert_encoding($part, 'UTF-8', $encoding);
}
$parts[$key] = phutil_utf8ize($part);
}
$committer_name = $parts[0];
$committer_email = $parts[1];
$author_name = $parts[2];
$author_email = $parts[3];
$message = $parts[4];
if (strlen($author_email)) {
$author = "{$author_name} <{$author_email}>";
} else {
$author = "{$author_name}";
}
if (strlen($committer_email)) {
$committer = "{$committer_name} <{$committer_email}>";
} else {
$committer = "{$committer_name}";
}
if ($committer == $author) {
$committer = null;
}
$this->updateCommitData($author, $message, $committer);
if ($this->shouldQueueFollowupTasks()) {
$task = new PhabricatorWorkerTask();
$task->setTaskClass('PhabricatorRepositoryGitCommitChangeParserWorker');
$task->setData(array('commitID' => $commit->getID()));
$task->save();
}
}
public function parseCommit(PhabricatorRepository $repository, PhabricatorRepositoryCommit $commit)
{
list($stdout) = $repository->execxLocalCommand('log --template %s --rev %s', '{author}\\n{desc}', $commit->getCommitIdentifier());
list($author, $message) = explode("\n", $stdout, 2);
$author = phutil_utf8ize($author);
$message = phutil_utf8ize($message);
$message = trim($message);
$this->updateCommitData($author, $message);
if ($this->shouldQueueFollowupTasks()) {
$task = new PhabricatorWorkerTask();
$task->setTaskClass('PhabricatorRepositoryMercurialCommitChangeParserWorker');
$task->setData(array('commitID' => $commit->getID()));
$task->save();
}
}
private function loadSubversionCommitRef()
{
$repository = $this->getRepository();
list($xml) = $repository->execxRemoteCommand('log --xml --limit 1 %s', $repository->getSubversionPathURI(null, $this->identifier));
// Subversion may send us back commit messages which won't parse because
// they have non UTF-8 garbage in them. Slam them into valid UTF-8.
$xml = phutil_utf8ize($xml);
$log = new SimpleXMLElement($xml);
$entry = $log->logentry[0];
$author = (string) $entry->author;
$message = (string) $entry->msg;
list($author_name, $author_email) = $this->splitUserIdentifier($author);
// No hashes in Subversion.
$hashes = array();
return id(new DiffusionCommitRef())->setAuthorName($author_name)->setAuthorEmail($author_email)->setMessage($message)->setHashes($hashes);
}
public function parseCommit(PhabricatorRepository $repository, PhabricatorRepositoryCommit $commit)
{
// NOTE: %B was introduced somewhat recently in git's history, so pull
// commit message information with %s and %b instead.
list($info) = $repository->execxLocalCommand("log -n 1 --encoding='UTF-8' --pretty=format:%%an%%x00%%s%%n%%n%%b %s", $commit->getCommitIdentifier());
list($author, $message) = explode("", $info);
// Make sure these are valid UTF-8.
$author = phutil_utf8ize($author);
$message = phutil_utf8ize($message);
$message = trim($message);
$this->updateCommitData($author, $message);
if ($this->shouldQueueFollowupTasks()) {
$task = new PhabricatorWorkerTask();
$task->setTaskClass('PhabricatorRepositoryGitCommitChangeParserWorker');
$task->setData(array('commitID' => $commit->getID()));
$task->save();
}
}
public function __construct($status_code, $body, array $headers, $expect = null)
{
// NOTE: Avoiding PhutilUTF8StringTruncator here because this isn't lazy
// and responses may be large.
if (strlen($body) > 512) {
$excerpt = substr($body, 0, 512) . '...';
} else {
$excerpt = $body;
}
$content_type = BaseHTTPFuture::getHeader($headers, 'Content-Type');
$match = null;
if (preg_match('/;\\s*charset=([^;]+)/', $content_type, $match)) {
$encoding = trim($match[1], "\"'");
try {
$excerpt = phutil_utf8_convert($excerpt, 'UTF-8', $encoding);
} catch (Exception $ex) {
}
}
$this->excerpt = phutil_utf8ize($excerpt);
$this->expect = $expect;
parent::__construct($status_code);
}
/**
* This method is kind of awkward here but both the SVN message and
* change parsers use it.
*/
protected function getSVNLogXMLObject($uri, $revision, $verbose = false)
{
if ($verbose) {
$verbose = '--verbose';
}
try {
list($xml) = execx("svn log --xml {$verbose} --limit 1 --non-interactive %s@%d", $uri, $revision);
} catch (CommandException $ex) {
// HTTPS is generally faster and more reliable than svn+ssh, but some
// commit messages with non-UTF8 text can't be retrieved over HTTPS, see
// Facebook rE197184 for one example. Make an attempt to fall back to
// svn+ssh if we've failed outright to retrieve the message.
$fallback_uri = new PhutilURI($uri);
if ($fallback_uri->getProtocol() != 'https') {
throw $ex;
}
$fallback_uri->setProtocol('svn+ssh');
list($xml) = execx("svn log --xml {$verbose} --limit 1 --non-interactive %s@%d", $fallback_uri, $revision);
}
// Subversion may send us back commit messages which won't parse because
// they have non UTF-8 garbage in them. Slam them into valid UTF-8.
$xml = phutil_utf8ize($xml);
return new SimpleXMLElement($xml);
}
/**
* @phutil-external-symbol class PhabricatorStartup
*/
private function serveGitRequest(PhabricatorRepository $repository, PhabricatorUser $viewer)
{
$request = $this->getRequest();
$request_path = $this->getRequestDirectoryPath($repository);
$repository_root = $repository->getLocalPath();
// Rebuild the query string to strip `__magic__` parameters and prevent
// issues where we might interpret inputs like "service=read&service=write"
// differently than the server does and pass it an unsafe command.
// NOTE: This does not use getPassthroughRequestParameters() because
// that code is HTTP-method agnostic and will encode POST data.
$query_data = $_GET;
foreach ($query_data as $key => $value) {
if (!strncmp($key, '__', 2)) {
unset($query_data[$key]);
}
}
$query_string = http_build_query($query_data, '', '&');
// We're about to wipe out PATH with the rest of the environment, so
// resolve the binary first.
$bin = Filesystem::resolveBinary('git-http-backend');
if (!$bin) {
throw new Exception(pht('Unable to find `%s` in %s!', 'git-http-backend', '$PATH'));
}
$env = array('REQUEST_METHOD' => $_SERVER['REQUEST_METHOD'], 'QUERY_STRING' => $query_string, 'CONTENT_TYPE' => $request->getHTTPHeader('Content-Type'), 'HTTP_CONTENT_ENCODING' => $request->getHTTPHeader('Content-Encoding'), 'REMOTE_ADDR' => $_SERVER['REMOTE_ADDR'], 'GIT_PROJECT_ROOT' => $repository_root, 'GIT_HTTP_EXPORT_ALL' => '1', 'PATH_INFO' => $request_path, 'REMOTE_USER' => $viewer->getUsername()) + $this->getCommonEnvironment($viewer);
$input = PhabricatorStartup::getRawInput();
$command = csprintf('%s', $bin);
$command = PhabricatorDaemon::sudoCommandAsDaemonUser($command);
list($err, $stdout, $stderr) = id(new ExecFuture('%C', $command))->setEnv($env, true)->write($input)->resolve();
if ($err) {
if ($this->isValidGitShallowCloneResponse($stdout, $stderr)) {
// Ignore the error if the response passes this special check for
// validity.
$err = 0;
}
}
if ($err) {
return new PhabricatorVCSResponse(500, pht('Error %d: %s', $err, phutil_utf8ize($stderr)));
}
return id(new DiffusionGitResponse())->setGitData($stdout);
}
public function renderCow()
{
$width = 40;
$template = $this->template;
// Real ".cow" files are Perl scripts which define a variable called
// "$the_cow". We aren't going to interpret Perl, so strip all this stuff
// (and any comments in the file) away.
$template = phutil_split_lines($template, true);
$keep = array();
$is_perl_cowfile = false;
foreach ($template as $key => $line) {
if (preg_match('/^#/', $line)) {
continue;
}
if (preg_match('/^\\s*\\$the_cow/', $line)) {
$is_perl_cowfile = true;
continue;
}
if (preg_match('/^\\s*EOC\\s*$/', $line)) {
continue;
}
$keep[] = $line;
}
$template = implode('', $keep);
// Original .cow files are perl scripts which contain escaped sequences.
// We attempt to unescape here by replacing any character preceded by a
// backslash/escape with just that character.
if ($is_perl_cowfile) {
$template = preg_replace('/\\\\(.)/', '$1', $template);
}
$template = preg_replace_callback('/\\$([a-z]+)/', array($this, 'replaceTemplateVariable'), $template);
if ($template === false) {
throw new Exception(pht('Failed to replace template variables while rendering cow!'));
}
$lines = $this->text;
// TODO: It would be nice to use a utf8 soft wrap here instead, but we
// do not currently have one. Soft wrap first, then force to utf8.
$lines = wordwrap($lines, $width - 4, "\n", true);
$lines = phutil_split_lines($lines, false);
foreach ($lines as $key => $line) {
$lines[$key] = phutil_utf8ize($line);
}
if ($this->action == 'think') {
$borders = '((()))';
} else {
if (count($lines) == 1) {
$borders = '<<<>>>';
} else {
$borders = '/|\\\\|/';
}
}
$size = 0;
foreach ($lines as $line) {
$size = max(strlen($line), $size);
}
$balloon = array();
$balloon[] = ' ' . str_repeat('_', $size + 2);
$lines = array_values($lines);
$last = count($lines) - 1;
foreach ($lines as $idx => $line) {
if ($idx == 0) {
$l = $borders[0];
$r = $borders[3];
} else {
if ($idx == $last) {
$l = $borders[2];
$r = $borders[5];
} else {
$l = $borders[1];
$r = $borders[4];
}
}
$balloon[] = $l . ' ' . str_pad($line, $size) . ' ' . $r;
}
$balloon[] = ' ' . str_repeat('-', $size + 2);
$balloon = implode("\n", $balloon);
return rtrim($balloon . "\n" . $template);
}
/**
* Produce a human-readable explanation why a value can not be JSON-encoded.
*
* @param wild Value to validate.
* @param string Path within the object to provide context.
* @return string|null Explanation of why it can't be encoded, or null.
*/
function phutil_validate_json($value, $path = '')
{
if ($value === null) {
return;
}
if ($value === true) {
return;
}
if ($value === false) {
return;
}
if (is_int($value)) {
return;
}
if (is_float($value)) {
return;
}
if (is_array($value)) {
foreach ($value as $key => $subvalue) {
if (strlen($path)) {
$full_key = $path . ' > ';
} else {
$full_key = '';
}
if (!phutil_is_utf8($key)) {
$full_key = $full_key . phutil_utf8ize($key);
return pht('Dictionary key "%s" is not valid UTF8, and can not be JSON encoded.', $full_key);
}
$full_key .= $key;
$result = phutil_validate_json($subvalue, $full_key);
if ($result !== null) {
return $result;
}
}
}
if (is_string($value)) {
if (!phutil_is_utf8($value)) {
$display = substr($value, 0, 256);
$display = phutil_utf8ize($display);
if (!strlen($path)) {
return pht('String value is not valid UTF8, and can not be JSON encoded: %s', $display);
} else {
return pht('Dictionary value at key "%s" is not valid UTF8, and can not be ' . 'JSON encoded: %s', $path, $display);
}
}
}
return;
}
private function inlinePatch(PhabricatorMetaMTAMailBody $body, PhabricatorRepositoryCommit $commit)
{
if (!$this->getRawPatch()) {
return;
}
$inline_key = 'metamta.diffusion.inline-patches';
$inline_patches = PhabricatorEnv::getEnvConfig($inline_key);
if (!$inline_patches) {
return;
}
$repository = $commit->getRepository();
$raw_patch = $this->getRawPatch();
$result = null;
$len = substr_count($raw_patch, "\n");
if ($len <= $inline_patches) {
// We send email as utf8, so we need to convert the text to utf8 if
// we can.
$encoding = $repository->getDetail('encoding', 'UTF-8');
if ($encoding) {
$raw_patch = phutil_utf8_convert($raw_patch, 'UTF-8', $encoding);
}
$result = phutil_utf8ize($raw_patch);
}
if ($result) {
$result = "PATCH\n\n{$result}\n";
}
$body->addRawSection($result);
}
private function verifySubversionRoot(PhabricatorRepository $repository)
{
list($xml) = $repository->execxRemoteCommand('info --xml %s', $repository->getSubversionPathURI());
$xml = phutil_utf8ize($xml);
$xml = new SimpleXMLElement($xml);
$remote_root = (string) $xml->entry[0]->repository[0]->root[0];
$expect_root = $repository->getSubversionPathURI();
$normal_type_svn = PhabricatorRepositoryURINormalizer::TYPE_SVN;
$remote_normal = id(new PhabricatorRepositoryURINormalizer($normal_type_svn, $remote_root))->getNormalizedPath();
$expect_normal = id(new PhabricatorRepositoryURINormalizer($normal_type_svn, $expect_root))->getNormalizedPath();
if ($remote_normal != $expect_normal) {
throw new Exception(pht('Repository "%s" does not have a correctly configured remote URI. ' . 'The remote URI for a Subversion repository MUST point at the ' . 'repository root. The root for this repository is "%s", but the ' . 'configured URI is "%s". To resolve this error, set the remote URI ' . 'to point at the repository root. If you want to import only part ' . 'of a Subversion repository, use the "Import Only" option.', $repository->getDisplayName(), $remote_root, $expect_root));
}
}
protected function getUTF8StringFromStorage($string, $encoding)
{
if ($encoding == 'utf8') {
return $string;
}
if (function_exists('mb_detect_encoding')) {
if (strlen($encoding)) {
$try_encodings = array($encoding);
} else {
// TODO: This is pretty much a guess, and probably needs to be
// configurable in the long run.
$try_encodings = array('JIS', 'EUC-JP', 'SJIS', 'ISO-8859-1');
}
$guess = mb_detect_encoding($string, $try_encodings);
if ($guess) {
return mb_convert_encoding($string, 'UTF-8', $guess);
}
}
return phutil_utf8ize($string);
}
/**
* Sometimes, tab data includes binary information (like INSERT queries which
* write file data into the database). To successfully JSON encode it, we
* need to convert it to UTF-8.
*/
private function sanitizeForJSON($data)
{
if (is_object($data)) {
return '<object:' . get_class($data) . '>';
} else {
if (is_array($data)) {
foreach ($data as $key => $value) {
$data[$key] = $this->sanitizeForJSON($value);
}
return $data;
} else {
// Truncate huge strings. Since the data doesn't really matter much,
// just truncate bytes to avoid PhutilUTF8StringTruncator overhead.
$length = strlen($data);
$max = 4096;
if ($length > $max) {
$data = substr($data, 0, $max) . '...<' . $length . ' bytes>...';
}
return phutil_utf8ize($data);
}
}
}
/**
* Sometimes, tab data includes binary information (like INSERT queries which
* write file data into the database). To successfully JSON encode it, we
* need to convert it to UTF-8.
*/
private function sanitizeForJSON($data)
{
if (is_object($data)) {
return '<object:' . get_class($data) . '>';
} else {
if (is_array($data)) {
foreach ($data as $key => $value) {
$data[$key] = $this->sanitizeForJSON($value);
}
return $data;
} else {
return phutil_utf8ize($data);
}
}
}
private function getSVNLogXMLObject(PhabricatorRepository $repository, $uri, $revision)
{
list($xml) = $repository->execxRemoteCommand('log --xml --verbose --limit 1 %s@%d', $uri, $revision);
// Subversion may send us back commit messages which won't parse because
// they have non UTF-8 garbage in them. Slam them into valid UTF-8.
$xml = phutil_utf8ize($xml);
return new SimpleXMLElement($xml);
}
private function buildPatch(PhabricatorMetaMTAMail $template, PhabricatorRepository $repository, PhabricatorRepositoryCommit $commit)
{
$attach_key = 'metamta.diffusion.attach-patches';
$inline_key = 'metamta.diffusion.inline-patches';
$attach_patches = PhabricatorEnv::getEnvConfig($attach_key);
$inline_patches = PhabricatorEnv::getEnvConfig($inline_key);
if (!$attach_patches && !$inline_patches) {
return;
}
$encoding = $repository->getDetail('encoding', 'UTF-8');
$result = null;
$patch_error = null;
try {
$raw_patch = $this->loadRawPatchText($repository, $commit);
if ($attach_patches) {
$commit_name = $repository->formatCommitName($commit->getCommitIdentifier());
$template->addAttachment(new PhabricatorMetaMTAAttachment($raw_patch, $commit_name . '.patch', 'text/x-patch; charset=' . $encoding));
}
} catch (Exception $ex) {
phlog($ex);
$patch_error = 'Unable to generate: ' . $ex->getMessage();
}
if ($patch_error) {
$result = $patch_error;
} else {
if ($inline_patches) {
$len = substr_count($raw_patch, "\n");
if ($len <= $inline_patches) {
// We send email as utf8, so we need to convert the text to utf8 if
// we can.
if ($encoding) {
$raw_patch = phutil_utf8_convert($raw_patch, 'UTF-8', $encoding);
}
$result = phutil_utf8ize($raw_patch);
}
}
}
if ($result) {
$result = "PATCH\n\n{$result}\n";
}
return $result;
}
/**
* Find the words which are part of the query string, and bold them in a
* result string. This makes it easier for users to see why a result
* matched their query.
*/
private function emboldenQuery($str)
{
$query = $this->query->getParameter('query');
if (!strlen($query) || !strlen($str)) {
return $str;
}
// This algorithm is safe but not especially fast, so don't bother if
// we're dealing with a lot of data. This mostly prevents silly/malicious
// queries from doing anything bad.
if (strlen($query) + strlen($str) > 2048) {
return $str;
}
// Keep track of which characters we're going to make bold. This is
// byte oriented, but we'll make sure we don't put a bold in the middle
// of a character later.
$bold = array_fill(0, strlen($str), false);
// Split the query into words.
$parts = preg_split('/ +/', $query);
// Find all occurrences of each word, and mark them to be emboldened.
foreach ($parts as $part) {
$part = trim($part);
$part = trim($part, '"+');
if (!strlen($part)) {
continue;
}
$matches = null;
$has_matches = preg_match_all('/(?:^|\\b)(' . preg_quote($part, '/') . ')/i', $str, $matches, PREG_OFFSET_CAPTURE);
if (!$has_matches) {
continue;
}
// Flag the matching part of the range for boldening.
foreach ($matches[1] as $match) {
$offset = $match[1];
for ($ii = 0; $ii < strlen($match[0]); $ii++) {
$bold[$offset + $ii] = true;
}
}
}
// Split the string into ranges, applying bold styling as required.
$out = array();
$buf = '';
$pos = 0;
$is_bold = false;
// Make sure this is UTF8 because phutil_utf8v() will explode if it isn't.
$str = phutil_utf8ize($str);
foreach (phutil_utf8v($str) as $chr) {
if ($bold[$pos] != $is_bold) {
if (strlen($buf)) {
if ($is_bold) {
$out[] = phutil_tag('strong', array(), $buf);
} else {
$out[] = $buf;
}
$buf = '';
}
$is_bold = !$is_bold;
}
$buf .= $chr;
$pos += strlen($chr);
}
if (strlen($buf)) {
if ($is_bold) {
$out[] = phutil_tag('strong', array(), $buf);
} else {
$out[] = $buf;
}
}
return $out;
}
