/**
* Loop over range of wanted headers, insert headers into DB.
*
* @param array $groupMySQL The group info from mysql.
* @param int $first The oldest wanted header.
* @param int $last The newest wanted header.
* @param string $type Is this partrepair or update or backfill?
* @param null|array $missingParts If we are running in partrepair, the list of missing article numbers.
*
* @return array Empty on failure.
*/
public function scan($groupMySQL, $first, $last, $type = 'update', $missingParts = null)
{
// Start time of scan method and of fetching headers.
$startLoop = microtime(true);
// Check if MySQL tables exist, create if they do not, get their names at the same time.
$tableNames = $this->_groups->getCBPTableNames($this->_tablePerGroup, $groupMySQL['id']);
$returnArray = [];
$partRepair = $type === 'partrepair';
$addToPartRepair = $type === 'update' && $this->_partRepair;
// Download the headers.
if ($partRepair === true) {
// This is slower but possibly is better with missing headers.
$headers = $this->_nntp->getOverview($first . '-' . $last, true, false);
} else {
$headers = $this->_nntp->getXOVER($first . '-' . $last);
}
// If there was an error, try to reconnect.
if ($this->_nntp->isError($headers)) {
// Increment if part repair and return false.
if ($partRepair === true) {
$this->_pdo->queryExec(sprintf('UPDATE missed_parts SET attempts = attempts + 1 WHERE group_id = %d AND numberid %s', $groupMySQL['id'], $first == $last ? '= ' . $first : 'IN (' . implode(',', range($first, $last)) . ')'));
return $returnArray;
}
// This is usually a compression error, so try disabling compression.
$this->_nntp->doQuit();
if ($this->_nntp->doConnect(false) !== true) {
return $returnArray;
}
// Re-select group, download headers again without compression and re-enable compression.
$this->_nntp->selectGroup($groupMySQL['name']);
$headers = $this->_nntp->getXOVER($first . '-' . $last);
$this->_nntp->enableCompression();
// Check if the non-compression headers have an error.
if ($this->_nntp->isError($headers)) {
$message = $headers->code == 0 ? 'Unknown error' : $headers->message;
$this->log("Code {$headers->code}: {$message}\nSkipping group: {$groupMySQL['name']}", __FUNCTION__, Logger::LOG_WARNING, 'error');
return $returnArray;
}
}
// Start of processing headers.
$startCleaning = microtime(true);
// End of the getting data from usenet.
$timeHeaders = number_format($startCleaning - $startLoop, 2);
// Check if we got headers.
$msgCount = count($headers);
if ($msgCount < 1) {
return $returnArray;
}
// Get highest and lowest article numbers/dates.
$iterator1 = 0;
$iterator2 = $msgCount - 1;
while (true) {
if (!isset($returnArray['firstArticleNumber']) && isset($headers[$iterator1]['Number'])) {
$returnArray['firstArticleNumber'] = $headers[$iterator1]['Number'];
$returnArray['firstArticleDate'] = $headers[$iterator1]['Date'];
}
if (!isset($returnArray['lastArticleNumber']) && isset($headers[$iterator2]['Number'])) {
$returnArray['lastArticleNumber'] = $headers[$iterator2]['Number'];
$returnArray['lastArticleDate'] = $headers[$iterator2]['Date'];
}
// Break if we found non empty articles.
if (isset($returnArray['firstArticleNumber']) && isset($returnArray['lastArticleNumber'])) {
break;
}
// Break out if we couldn't find anything.
if ($iterator1++ >= $msgCount - 1 || $iterator2-- <= 0) {
break;
}
}
$headersRepaired = $articles = $rangeNotReceived = $collectionIDs = $binariesUpdate = $headersReceived = $headersNotInserted = [];
$notYEnc = $headersBlackListed = 0;
$partsQuery = $partsCheck = sprintf('INSERT IGNORE INTO %s (binaryid, number, messageid, partnumber, size) VALUES ', $tableNames['pname']);
$this->_pdo->beginTransaction();
// Loop articles, figure out files/parts.
foreach ($headers as $header) {
// Check if we got the article or not.
if (isset($header['Number'])) {
$headersReceived[] = $header['Number'];
} else {
if ($addToPartRepair) {
$rangeNotReceived[] = $header['Number'];
}
continue;
}
// If set we are running in partRepair mode.
if ($partRepair === true && !is_null($missingParts)) {
if (!in_array($header['Number'], $missingParts)) {
// If article isn't one that is missing skip it.
continue;
} else {
// We got the part this time. Remove article from part repair.
$headersRepaired[] = $header['Number'];
}
}
/*
* Find part / total parts. Ignore if no part count found.
*
* \s* Trims the leading space.
* (?!"Usenet Index Post) ignores these types of articles, they are useless.
* (.+) Fetches the subject.
* \s+ Trims trailing space after the subject.
* \((\d+)\/(\d+)\) Gets the part count.
* No ending ($) as there are cases of subjects with extra data after the part count.
*/
if (preg_match('/^\\s*(?!"Usenet Index Post)(.+)\\s+\\((\\d+)\\/(\\d+)\\)/', $header['Subject'], $matches)) {
// Add yEnc to subjects that do not have them, but have the part number at the end of the header.
if (!stristr($header['Subject'], 'yEnc')) {
$matches[1] .= ' yEnc';
}
} else {
if ($this->_showDroppedYEncParts === true && strpos($header['Subject'], '"Usenet Index Post') !== 0) {
file_put_contents(nZEDb_LOGS . 'not_yenc' . $groupMySQL['name'] . '.dropped.log', $header['Subject'] . PHP_EOL, FILE_APPEND);
}
$notYEnc++;
continue;
}
// Filter subject based on black/white list.
if ($this->isBlackListed($header, $groupMySQL['name'])) {
$headersBlackListed++;
continue;
}
if (!isset($header['Bytes'])) {
$header['Bytes'] = isset($header[':bytes']) ? $header[':bytes'] : 0;
}
$header['Bytes'] = (int) $header['Bytes'];
// Set up the info for inserting into parts/binaries/collections tables.
if (!isset($articles[$matches[1]])) {
// check whether file count should be ignored (XXX packs for now only).
$whitelistMatch = false;
if ($this->_ignoreFileCount($groupMySQL['name'], $matches[1])) {
$whitelistMatch = true;
$fileCount[1] = $fileCount[3] = 0;
}
// Attempt to find the file count. If it is not found, set it to 0.
if (!$whitelistMatch && !preg_match('/[[(\\s](\\d{1,5})(\\/|[\\s_]of[\\s_]|-)(\\d{1,5})[])\\s$:]/i', $matches[1], $fileCount)) {
$fileCount[1] = $fileCount[3] = 0;
if ($this->_showDroppedYEncParts === true) {
file_put_contents(nZEDb_LOGS . 'no_files' . $groupMySQL['name'] . '.log', $header['Subject'] . PHP_EOL, FILE_APPEND);
}
}
// Used to group articles together when forming the release/nzb.
$header['CollectionKey'] = $this->_collectionsCleaning->collectionsCleaner($matches[1], $groupMySQL['name']) . $header['From'] . $groupMySQL['id'] . $fileCount[3];
if (!isset($collectionIDs[$header['CollectionKey']])) {
/* Date from header should be a string this format:
* 31 Mar 2014 15:36:04 GMT or 6 Oct 1998 04:38:40 -0500
* Still make sure it's not unix time, convert it to unix time if it is.
*/
$header['Date'] = is_numeric($header['Date']) ? $header['Date'] : strtotime($header['Date']);
// Get the current unixtime from PHP.
$now = time();
$collectionID = $this->_pdo->queryInsert(sprintf("\n\t\t\t\t\t\t\tINSERT INTO %s (subject, fromname, date, xref, group_id,\n\t\t\t\t\t\t\t\ttotalfiles, collectionhash, dateadded)\n\t\t\t\t\t\t\tVALUES (%s, %s, FROM_UNIXTIME(%s), %s, %d, %d, '%s', NOW())\n\t\t\t\t\t\t\tON DUPLICATE KEY UPDATE dateadded = NOW(), noise = '%s'", $tableNames['cname'], $this->_pdo->escapeString(substr(utf8_encode($matches[1]), 0, 255)), $this->_pdo->escapeString(utf8_encode($header['From'])), is_numeric($header['Date']) ? $header['Date'] > $now ? $now : $header['Date'] : $now, $this->_pdo->escapeString(substr($header['Xref'], 0, 255)), $groupMySQL['id'], $fileCount[3], sha1($header['CollectionKey']), bin2hex(openssl_random_pseudo_bytes(16))));
if ($collectionID === false) {
if ($addToPartRepair) {
$headersNotInserted[] = $header['Number'];
}
$this->_pdo->Rollback();
$this->_pdo->beginTransaction();
continue;
}
$collectionIDs[$header['CollectionKey']] = $collectionID;
} else {
$collectionID = $collectionIDs[$header['CollectionKey']];
}
$binaryID = $this->_pdo->queryInsert(sprintf("\n\t\t\t\t\t\tINSERT INTO %s (binaryhash, name, collection_id, totalparts, currentparts, filenumber, partsize)\n\t\t\t\t\t\tVALUES (UNHEX('%s'), %s, %d, %d, 1, %d, %d)\n\t\t\t\t\t\tON DUPLICATE KEY UPDATE currentparts = currentparts + 1, partsize = partsize + %d", $tableNames['bname'], md5($matches[1] . $header['From'] . $groupMySQL['id']), $this->_pdo->escapeString(utf8_encode($matches[1])), $collectionID, $matches[3], $fileCount[1], $header['Bytes'], $header['Bytes']));
if ($binaryID === false) {
if ($addToPartRepair) {
$headersNotInserted[] = $header['Number'];
}
$this->_pdo->Rollback();
$this->_pdo->beginTransaction();
continue;
}
$binariesUpdate[$binaryID]['Size'] = 0;
$binariesUpdate[$binaryID]['Parts'] = 0;
$articles[$matches[1]]['CollectionID'] = $collectionID;
$articles[$matches[1]]['BinaryID'] = $binaryID;
} else {
$binaryID = $articles[$matches[1]]['BinaryID'];
$collectionID = $articles[$matches[1]]['CollectionID'];
$binariesUpdate[$binaryID]['Size'] += $header['Bytes'];
$binariesUpdate[$binaryID]['Parts']++;
}
// Strip the < and >, saves space in DB.
$header['Message-ID'][0] = "'";
$partsQuery .= '(' . $binaryID . ',' . $header['Number'] . ',' . rtrim($header['Message-ID'], '>') . "'," . $matches[2] . ',' . $header['Bytes'] . '),';
}
unset($headers);
// Reclaim memory.
// Start of inserting into SQL.
$startUpdate = microtime(true);
// End of processing headers.
$timeCleaning = number_format($startUpdate - $startCleaning, 2);
$binariesQuery = $binariesCheck = sprintf('INSERT INTO %s (id, partsize, currentparts) VALUES ', $tableNames['bname']);
foreach ($binariesUpdate as $binaryID => $binary) {
$binariesQuery .= '(' . $binaryID . ',' . $binary['Size'] . ',' . $binary['Parts'] . '),';
}
$binariesEnd = ' ON DUPLICATE KEY UPDATE partsize = VALUES(partsize) + partsize, currentparts = VALUES(currentparts) + currentparts';
$binariesQuery = rtrim($binariesQuery, ',') . $binariesEnd;
// Check if we got any binaries. If we did, try to insert them.
if (strlen($binariesCheck . $binariesEnd) === strlen($binariesQuery) ? true : $this->_pdo->queryExec($binariesQuery)) {
if ($this->_debug) {
$this->_colorCLI->doEcho($this->_colorCLI->debug('Sending ' . round(strlen($partsQuery) / 1024, 2) . ' KB of parts to MySQL'));
}
if (strlen($partsQuery) === strlen($partsCheck) ? true : $this->_pdo->queryExec(rtrim($partsQuery, ','))) {
$this->_pdo->Commit();
} else {
if ($addToPartRepair) {
$headersNotInserted += $headersReceived;
}
$this->_pdo->Rollback();
}
} else {
if ($addToPartRepair) {
$headersNotInserted += $headersReceived;
}
$this->_pdo->Rollback();
}
if ($this->_echoCLI && $partRepair === false) {
$this->_colorCLI->doEcho($this->_colorCLI->primary('Received ' . count($headersReceived) . ' articles of ' . number_format($last - $first + 1) . ' requested, ' . $headersBlackListed . ' blacklisted, ' . $notYEnc . ' not yEnc.'));
}
// Start of part repair.
$startPR = microtime(true);
// End of inserting.
$timeInsert = number_format($startPR - $startUpdate, 2);
if ($partRepair && count($headersRepaired) > 0) {
$this->removeRepairedParts($headersRepaired, $tableNames['prname'], $groupMySQL['id']);
}
if ($addToPartRepair) {
$notInsertedCount = count($headersNotInserted);
if ($notInsertedCount > 0) {
$this->addMissingParts($headersNotInserted, $tableNames['prname'], $groupMySQL['id']);
$this->log($notInsertedCount . ' articles failed to insert!', __FUNCTION__, Logger::LOG_WARNING, 'warning');
}
// Check if we have any missing headers.
if ($last - $first - $notYEnc - $headersBlackListed + 1 > count($headersReceived)) {
$rangeNotReceived = array_merge($rangeNotReceived, array_diff(range($first, $last), $headersReceived));
}
$notReceivedCount = count($rangeNotReceived);
if ($notReceivedCount > 0) {
$this->addMissingParts($rangeNotReceived, $tableNames['prname'], $groupMySQL['id']);
if ($this->_echoCLI) {
$this->_colorCLI->doEcho($this->_colorCLI->alternate('Server did not return ' . $notReceivedCount . ' articles from ' . $groupMySQL['name'] . '.'), true);
}
}
}
$currentMicroTime = microtime(true);
if ($this->_echoCLI) {
$this->_colorCLI->doEcho($this->_colorCLI->alternateOver($timeHeaders . 's') . $this->_colorCLI->primaryOver(' to download articles, ') . $this->_colorCLI->alternateOver($timeCleaning . 's') . $this->_colorCLI->primaryOver(' to process collections, ') . $this->_colorCLI->alternateOver($timeInsert . 's') . $this->_colorCLI->primaryOver(' to insert binaries/parts, ') . $this->_colorCLI->alternateOver(number_format($currentMicroTime - $startPR, 2) . 's') . $this->_colorCLI->primaryOver(' for part repair, ') . $this->_colorCLI->alternateOver(number_format($currentMicroTime - $startLoop, 2) . 's') . $this->_colorCLI->primary(' total.'));
}
return $returnArray;
}
