public void inject(Path crawlDb, Path urlDir) throws IOException {
SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
long start = System.currentTimeMillis();
if (LOG.isInfoEnabled()) {
LOG.info("Injector: starting at " + sdf.format(start));
LOG.info("Injector: crawlDb: " + crawlDb);
LOG.info("Injector: urlDir: " + urlDir);
}
Path tempDir =
new Path(
getConf().get("mapred.temp.dir", ".")
+ "/inject-temp-"
+ Integer.toString(new Random().nextInt(Integer.MAX_VALUE)));
// map text input file to a <url,CrawlDatum> file
if (LOG.isInfoEnabled()) {
LOG.info("Injector: Converting injected urls to crawl db entries.");
}
JobConf sortJob = new NutchJob(getConf());
sortJob.setJobName("inject " + urlDir);
FileInputFormat.addInputPath(sortJob, urlDir);
sortJob.setMapperClass(InjectMapper.class);
FileOutputFormat.setOutputPath(sortJob, tempDir);
sortJob.setOutputFormat(SequenceFileOutputFormat.class);
sortJob.setOutputKeyClass(Text.class);
sortJob.setOutputValueClass(CrawlDatum.class);
sortJob.setLong("injector.current.time", System.currentTimeMillis());
RunningJob mapJob = JobClient.runJob(sortJob);
long urlsInjected = mapJob.getCounters().findCounter("injector", "urls_injected").getValue();
long urlsFiltered = mapJob.getCounters().findCounter("injector", "urls_filtered").getValue();
LOG.info("Injector: total number of urls rejected by filters: " + urlsFiltered);
LOG.info(
"Injector: total number of urls injected after normalization and filtering: "
+ urlsInjected);
// merge with existing crawl db
if (LOG.isInfoEnabled()) {
LOG.info("Injector: Merging injected urls into crawl db.");
}
JobConf mergeJob = CrawlDb.createJob(getConf(), crawlDb);
FileInputFormat.addInputPath(mergeJob, tempDir);
mergeJob.setReducerClass(InjectReducer.class);
JobClient.runJob(mergeJob);
CrawlDb.install(mergeJob, crawlDb);
// clean up
FileSystem fs = FileSystem.get(getConf());
fs.delete(tempDir, true);
long end = System.currentTimeMillis();
LOG.info(
"Injector: finished at "
+ sdf.format(end)
+ ", elapsed: "
+ TimingUtil.elapsedTime(start, end));
}
// the method which actually copies the caches locally and unjars/unzips them
// and does chmod for the files
private static Path localizeCache(
Configuration conf, URI cache, long confFileStamp, CacheStatus cacheStatus, boolean isArchive)
throws IOException {
FileSystem fs = getFileSystem(cache, conf);
FileSystem localFs = FileSystem.getLocal(conf);
Path parchive = null;
if (isArchive) {
parchive =
new Path(
cacheStatus.localizedLoadPath, new Path(cacheStatus.localizedLoadPath.getName()));
} else {
parchive = cacheStatus.localizedLoadPath;
}
if (!localFs.mkdirs(parchive.getParent())) {
throw new IOException(
"Mkdirs failed to create directory " + cacheStatus.localizedLoadPath.toString());
}
String cacheId = cache.getPath();
fs.copyToLocalFile(new Path(cacheId), parchive);
if (isArchive) {
String tmpArchive = parchive.toString().toLowerCase();
File srcFile = new File(parchive.toString());
File destDir = new File(parchive.getParent().toString());
if (tmpArchive.endsWith(".jar")) {
RunJar.unJar(srcFile, destDir);
} else if (tmpArchive.endsWith(".zip")) {
FileUtil.unZip(srcFile, destDir);
} else if (isTarFile(tmpArchive)) {
FileUtil.unTar(srcFile, destDir);
}
// else will not do anyhting
// and copy the file into the dir as it is
}
long cacheSize = FileUtil.getDU(new File(parchive.getParent().toString()));
cacheStatus.size = cacheSize;
addCacheInfoUpdate(cacheStatus);
// do chmod here
try {
// Setting recursive permission to grant everyone read and execute
Path localDir = new Path(cacheStatus.localizedBaseDir, cacheStatus.uniqueParentDir);
LOG.info("Doing chmod on localdir :" + localDir);
FileUtil.chmod(localDir.toString(), "ugo+rx", true);
} catch (InterruptedException e) {
LOG.warn("Exception in chmod" + e.toString());
}
// update cacheStatus to reflect the newly cached file
cacheStatus.mtime = getTimestamp(conf, cache);
return cacheStatus.localizedLoadPath;
}
protected void setup(Context context) throws IOException, InterruptedException {
super.setup(context);
Path pt = new Path("/user/yao/query/query");
FileSystem fs = FileSystem.get(new Configuration());
BufferedReader br = new BufferedReader(new InputStreamReader(fs.open(pt)));
String line = br.readLine();
String[] keywords = line.split(",");
k0 = keywords[0];
k1 = keywords[1];
k2 = keywords[2];
br.close();
}
private int getNumSecrets(Path secretsPath) throws Exception {
FileSystem fileSystem = FileSystem.get(secretsPath.toUri(), getConf());
FSDataInputStream inputStream = fileSystem.open(secretsPath);
BufferedReader reader = new BufferedReader(new InputStreamReader(inputStream));
String currentLine;
int numLines = 0;
while ((currentLine = reader.readLine()) != null) {
if (!currentLine.isEmpty()) {
numLines++;
}
}
reader.close();
return numLines;
}
private void generateSaltIfNeeded(Path saltFilePath, Path secretsPath) throws Exception {
FileSystem fileSystem = FileSystem.get(saltFilePath.toUri(), getConf());
if (!fileSystem.exists(saltFilePath)) {
FSDataOutputStream outputStream = fileSystem.create(saltFilePath);
BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(outputStream));
int numSaltsToGenerate = getNumSecrets(secretsPath);
System.out.printf("Generating %d salts\n", numSaltsToGenerate);
for (int i = 0; i < numSaltsToGenerate; i++) {
writer.write(BCrypt.gensalt());
writer.newLine();
}
writer.close();
outputStream.close();
}
}
private static boolean hasFile(
dbutil db_util, FileSystem fs, String db_name, String file_id, String file_path)
throws Exception {
Get file_id_get = new Get(file_id.getBytes());
Result file_result = db_util.doGet(db_name, file_id_get);
KeyValue file_names = file_result.getColumnLatest("d".getBytes(), "filenames".getBytes());
if (file_names == null) {
return false;
}
String all_files = new String(file_names.getValue());
String[] files = all_files.split("\n");
for (String line : files) {
if (line.equals(file_path)) {
if (fs.globStatus(new Path(line + "*")).length == 0) {
Put new_put = new Put(file_id.getBytes());
new_put.add(
"d".getBytes(),
"filenames".getBytes(),
all_files.replace(file_path + "\n", "").getBytes());
db_util.doPut(db_name, new_put);
return false;
}
return true;
}
}
return false;
}
/**
* Clear the entire contents of the cache and delete the backing files. This should only be used
* when the server is reinitializing, because the users are going to lose their files.
*/
public static void purgeCache(Configuration conf, MRAsyncDiskService service) throws IOException {
synchronized (cachedArchives) {
LocalFileSystem localFs = FileSystem.getLocal(conf);
for (Map.Entry<String, CacheStatus> f : cachedArchives.entrySet()) {
try {
deleteLocalPath(service, localFs, f.getValue().localizedLoadPath);
} catch (IOException ie) {
LOG.debug("Error cleaning up cache", ie);
}
}
cachedArchives.clear();
}
}
public static void recursePath(Configuration conf, Path path, Job job) {
try {
FileSystem fs = path.getFileSystem(conf);
FileStatus[] fstats = fs.listStatus(path);
if (fstats != null) {
for (FileStatus f : fstats) {
Path p = f.getPath();
;
if (fs.isFile(p)) {
// connection times out otherwise
System.err.println("file:" + p.toString());
FileInputFormat.addInputPath(job, p);
} else {
System.err.println("dir:" + p.toString());
recursePath(conf, p, job);
}
}
}
} catch (IOException e) {
// shouldn't be here
throw new RuntimeException(e);
}
}
private Job configureJob(Path secretsPath, Path saltFilePath, Path inputPath, Path outputPath)
throws Exception {
Job job = Job.getInstance(getConf());
FileInputFormat.setInputPaths(job, inputPath);
FileOutputFormat.setOutputPath(job, outputPath);
job.getConfiguration().set(ObfuscateMapper.SECRET_WORDS_FILE_KEY, secretsPath.toString());
job.getConfiguration().set(ObfuscateMapper.SALT_FILE_KEY, saltFilePath.toString());
job.setInputFormatClass(TextInputFormat.class);
job.setMapperClass(ObfuscateMapper.class);
job.setNumReduceTasks(0);
job.setJarByClass(getClass());
FileSystem.get(outputPath.toUri(), getConf()).delete(outputPath, true);
return job;
}
/*
* Returns the relative path of the dir this cache will be localized in
* relative path that this cache will be localized in. For
* hdfs://hostname:port/absolute_path -- the relative path is
* hostname/absolute path -- if it is just /absolute_path -- then the
* relative path is hostname of DFS this mapred cluster is running
* on/absolute_path
*/
public static String makeRelative(URI cache, Configuration conf) throws IOException {
String host = cache.getHost();
if (host == null) {
host = cache.getScheme();
}
if (host == null) {
URI defaultUri = FileSystem.get(conf).getUri();
host = defaultUri.getHost();
if (host == null) {
host = defaultUri.getScheme();
}
}
String path = host + cache.getPath();
path = path.replace(":/", "/"); // remove windows device colon
return path;
}
private static Path checkCacheStatusValidity(
Configuration conf,
URI cache,
long confFileStamp,
CacheStatus cacheStatus,
FileStatus fileStatus,
boolean isArchive)
throws IOException {
FileSystem fs = FileSystem.get(cache, conf);
// Has to be
if (!ifExistsAndFresh(conf, fs, cache, confFileStamp, cacheStatus, fileStatus)) {
throw new IOException(
"Stale cache file: " + cacheStatus.localizedLoadPath + " for cache-file: " + cache);
}
LOG.info(
String.format(
"Using existing cache of %s->%s", cache.toString(), cacheStatus.localizedLoadPath));
return cacheStatus.localizedLoadPath;
}
private static void deleteCache(Configuration conf, MRAsyncDiskService asyncDiskService)
throws IOException {
List<CacheStatus> deleteSet = new LinkedList<CacheStatus>();
// try deleting cache Status with refcount of zero
synchronized (cachedArchives) {
for (Iterator<String> it = cachedArchives.keySet().iterator(); it.hasNext(); ) {
String cacheId = (String) it.next();
CacheStatus lcacheStatus = cachedArchives.get(cacheId);
if (lcacheStatus.refcount == 0) {
// delete this cache entry from the global list
// and mark the localized file for deletion
deleteSet.add(lcacheStatus);
it.remove();
}
}
}
// do the deletion asynchronously, after releasing the global lock
Thread cacheFileCleaner =
new Thread(new CacheFileCleanTask(asyncDiskService, FileSystem.getLocal(conf), deleteSet));
cacheFileCleaner.start();
}
private static FileSystem getFileSystem(URI cache, Configuration conf) throws IOException {
String fileSysName = getFileSysName(cache);
if (fileSysName != null) return FileSystem.getNamed(fileSysName, conf);
else return FileSystem.get(conf);
}
/**
* Returns the status of a given cache file on hdfs.
*
* @param conf configuration
* @param cache cache file
* @return FileStatus object of the file
* @throws IOException
*/
public static FileStatus getFileStatus(Configuration conf, URI cache) throws IOException {
FileSystem fileSystem = FileSystem.get(cache, conf);
Path filePath = new Path(cache.getPath());
return fileSystem.getFileStatus(filePath);
}
/**
* Returns mtime of a given cache file on hdfs.
*
* @param conf configuration
* @param cache cache file
* @return mtime of a given cache file on hdfs
* @throws IOException
*/
public static long getTimestamp(Configuration conf, URI cache) throws IOException {
FileSystem fileSystem = FileSystem.get(cache, conf);
Path filePath = new Path(cache.getPath());
return fileSystem.getFileStatus(filePath).getModificationTime();
}
// Information needed to get a single file:
// BASE_PATH, FILE_ID, TIMESTAMP_START, TIMESTAMP_STOP, SOURCE, FILESYSTEM
private static Vector<Path> getFile(
FileSystem fs, Hashtable<String, String> config, dbutil db_util) throws Exception {
Long latestVersion = latestVersion(config, db_util);
try {
config.put("timestamp_start", config.get("timestamp_start"));
config.put("timestamp_real", latestVersion.toString());
config.put("timestamp_stop", latestVersion.toString());
} catch (Exception E) {
logger.error("Tryign to get file that is impossible to generate: " + getFullPath(config));
return null;
}
if (Integer.parseInt(config.get("timestamp_start"))
> Integer.parseInt(config.get("timestamp_stop"))) {
return null;
}
logger.debug(
"Getting DB for timestamp "
+ config.get("timestamp_start")
+ " to "
+ config.get("timestamp_stop"));
String final_result = getFullPath(config);
String temp_path_base =
config.get("local_temp_path")
+ "_"
+ config.get("task_id")
+ "_"
+ config.get("run_id")
+ "/";
Path newPath = new Path(final_result + "*");
Vector<Path> ret_path = new Vector<Path>();
String lockName = lock(final_result.replaceAll("/", "_"));
if (fs.globStatus(newPath).length != 0) {
ret_path.add(newPath);
unlock(lockName);
config.put("full_file_name", final_result);
return ret_path;
} else {
if (!config.get("source").equals("local")) {
config.put("temp_path_base", temp_path_base);
config.put("timestamp_start", config.get("timestamp_start"));
config.put("timestamp_real", latestVersion.toString());
config.put("timestamp_stop", latestVersion.toString());
Class<?> sourceClass =
Class.forName("org.gestore.plugin.source." + config.get("source") + "Source");
Method process_data = sourceClass.getMethod("process", Hashtable.class, FileSystem.class);
Object processor = sourceClass.newInstance();
Object retVal;
try {
retVal = process_data.invoke(processor, config, fs);
} catch (InvocationTargetException E) {
Throwable exception = E.getTargetException();
logger.error("Unable to call method in child class: " + exception.toString());
exception.printStackTrace(System.out);
unlock(lockName);
return null;
}
FileStatus[] files = (FileStatus[]) retVal;
if (files == null) {
logger.error("Error getting files, no files returned");
return null;
}
for (FileStatus file : files) {
Path cur_file = file.getPath();
Path cur_local_path = new Path(temp_path_base + config.get("file_id"));
String suffix = getSuffix(config.get("file_id"), cur_file.getName());
cur_local_path = cur_local_path.suffix(suffix);
Path res_path = new Path(new String(final_result + suffix));
logger.debug("Moving file" + cur_file.toString() + " to " + res_path.toString());
if (config.get("copy").equals("true")) {
fs.moveFromLocalFile(cur_file, res_path);
} else {
fs.rename(cur_file, res_path);
}
}
config.put("full_file_name", final_result);
}
}
unlock(lockName);
return ret_path;
}
public int run(String[] args) throws Exception {
// printUsage();
/*
* SETUP
*/
Configuration argConf = getConf();
Hashtable<String, String> confArg = new Hashtable<String, String>();
setup(confArg, argConf);
Date currentTime = new Date();
Date endDate = new Date(new Long(confArg.get("timestamp_stop")));
Boolean full_run = confArg.get("intermediate").matches("(?i).*true.*");
Boolean quick_add = confArg.get("quick_add").matches("(?i).*true.*");
logger.info("Running GeStore");
// ZooKeeper setup
Configuration config = HBaseConfiguration.create();
zkWatcher = new ZooKeeperWatcher(config, "Testing", new HBaseAdmin(config));
zkInstance =
new ZooKeeper(
ZKConfig.getZKQuorumServersString(config),
config.getInt("zookeeper.session.timeout", -1),
zkWatcher);
if (!confArg.get("task_id").isEmpty()) {
confArg.put("temp_path", confArg.get("temp_path") + confArg.get("task_id"));
}
String lockRequest = confArg.get("file_id");
if (!confArg.get("run_id").isEmpty())
lockRequest = lockRequest + "_" + confArg.get("run_id") + "_";
if (!confArg.get("task_id").isEmpty())
lockRequest = lockRequest + "_" + confArg.get("task_id") + "_";
// Get type of movement
toFrom type_move = checkArgs(confArg);
if (type_move == toFrom.LOCAL2REMOTE && !confArg.get("format").equals("unknown")) {
List<String> arguments = new ArrayList<String>();
arguments.add("-Dinput=" + confArg.get("local_path"));
arguments.add("-Dtable=" + confArg.get("file_id"));
arguments.add("-Dtimestamp=" + confArg.get("timestamp_stop"));
arguments.add("-Dtype=" + confArg.get("format"));
arguments.add("-Dtarget_dir=" + confArg.get("base_path") + "_" + confArg.get("file_id"));
arguments.add("-Dtemp_hdfs_path=" + confArg.get("temp_path"));
arguments.add("-Drun_id=" + confArg.get("run_id"));
if (!confArg.get("run_id").isEmpty()) arguments.add("-Drun_id=" + confArg.get("run_id"));
if (!confArg.get("task_id").isEmpty()) arguments.add("-Dtask_id=" + confArg.get("task_id"));
if (quick_add) arguments.add("-Dquick_add=" + confArg.get("quick_add"));
String lockName = lock(lockRequest);
String[] argumentString = arguments.toArray(new String[arguments.size()]);
adddb.main(argumentString);
unlock(lockName);
System.exit(0);
}
// Database registration
dbutil db_util = new dbutil(config);
db_util.register_database(confArg.get("db_name_files"), true);
db_util.register_database(confArg.get("db_name_runs"), true);
db_util.register_database(confArg.get("db_name_updates"), true);
FileSystem hdfs = FileSystem.get(config);
FileSystem localFS = FileSystem.getLocal(config);
// Get source type
confArg.put("source", getSource(db_util, confArg.get("db_name_files"), confArg.get("file_id")));
confArg.put(
"database", isDatabase(db_util, confArg.get("db_name_files"), confArg.get("file_id")));
if (!confArg.get("source").equals("local")
&& type_move == toFrom.REMOTE2LOCAL
&& !confArg.get("timestamp_stop").equals(Integer.toString(Integer.MAX_VALUE))) {
confArg.put("timestamp_stop", Long.toString(latestVersion(confArg, db_util)));
}
/*
* Get previous timestamp
*/
Get run_id_get = new Get(confArg.get("run_id").getBytes());
Result run_get = db_util.doGet(confArg.get("db_name_runs"), run_id_get);
KeyValue run_file_prev =
run_get.getColumnLatest(
"d".getBytes(), (confArg.get("file_id") + "_db_timestamp").getBytes());
String last_timestamp = new String("0");
if (null != run_file_prev && !confArg.get("source").equals("local")) {
long last_timestamp_real = run_file_prev.getTimestamp();
Long current_timestamp = new Long(confArg.get("timestamp_real"));
if ((current_timestamp - last_timestamp_real) > 36000) {
last_timestamp = new String(run_file_prev.getValue());
Integer lastTimestamp = new Integer(last_timestamp);
lastTimestamp += 1;
last_timestamp = lastTimestamp.toString();
logger.info("Last timestamp: " + last_timestamp + " End data: " + endDate);
Date last_run = new Date(run_file_prev.getTimestamp());
if (last_run.before(endDate) && !full_run) {
confArg.put("timestamp_start", last_timestamp);
}
}
}
Integer tse = new Integer(confArg.get("timestamp_stop"));
Integer tss = new Integer(confArg.get("timestamp_start"));
if (tss > tse) {
logger.info("No new version of requested file.");
return 0;
}
/*
* Generate file
*/
String lockName = lock(lockRequest);
Get file_id_get = new Get(confArg.get("file_id").getBytes());
Result file_get = db_util.doGet(confArg.get("db_name_files"), file_id_get);
if (!file_get.isEmpty()) {
boolean found =
hasFile(
db_util,
hdfs,
confArg.get("db_name_files"),
confArg.get("file_id"),
getFullPath(confArg));
if (confArg.get("source").equals("fullfile")) {
found = false;
}
String filenames_put =
getFileNames(
db_util, confArg.get("db_name_files"), confArg.get("file_id"), getFullPath(confArg));
// Filename not found in file database
if (!found && type_move == toFrom.REMOTE2LOCAL) {
if (!confArg.get("source").equals("local")) {
// Generate intermediate file
if (getFile(hdfs, confArg, db_util) == null) {
unlock(lockName);
return 1;
}
// Put generated file into file database
if (!confArg.get("format").equals("fullfile")) {
putFileEntry(
db_util,
hdfs,
confArg.get("db_name_files"),
confArg.get("file_id"),
confArg.get("full_file_name"),
confArg.get("source"));
}
} else {
logger.warn("Remote file not found, and cannot be generated! File: " + confArg);
unlock(lockName);
return 1;
}
}
} else {
if (type_move == toFrom.REMOTE2LOCAL) {
logger.warn("Remote file not found, and cannot be generated.");
unlock(lockName);
return 1;
}
}
/*
* Copy file
* Update tables
*/
if (type_move == toFrom.LOCAL2REMOTE) {
if (!confArg.get("format").equals("fullfile")) {
putFileEntry(
db_util,
hdfs,
confArg.get("db_name_files"),
confArg.get("file_id"),
getFullPath(confArg),
confArg.get("source"));
}
putRunEntry(
db_util,
confArg.get("db_name_runs"),
confArg.get("run_id"),
confArg.get("file_id"),
confArg.get("type"),
confArg.get("timestamp_real"),
confArg.get("timestamp_stop"),
getFullPath(confArg),
confArg.get("delimiter"));
hdfs.copyFromLocalFile(new Path(confArg.get("local_path")), new Path(getFullPath(confArg)));
} else if (type_move == toFrom.REMOTE2LOCAL) {
FileStatus[] files = hdfs.globStatus(new Path(getFullPath(confArg) + "*"));
putRunEntry(
db_util,
confArg.get("db_name_runs"),
confArg.get("run_id"),
confArg.get("file_id"),
confArg.get("type"),
confArg.get("timestamp_real"),
confArg.get("timestamp_stop"),
getFullPath(confArg),
confArg.get("delimiter"));
unlock(lockName);
for (FileStatus file : files) {
Path cur_file = file.getPath();
Path cur_local_path =
new Path(new String(confArg.get("local_path") + confArg.get("file_id")));
String suffix = getSuffix(getFileName(confArg), cur_file.getName());
if (suffix.length() > 0) {
cur_local_path = cur_local_path.suffix(new String("." + suffix));
}
if (confArg.get("copy").equals("true")) {
String crc = hdfs.getFileChecksum(cur_file).toString();
if (checksumLocalTest(cur_local_path, crc)) {
continue;
} else {
hdfs.copyToLocalFile(cur_file, cur_local_path);
writeChecksum(cur_local_path, crc);
}
} else {
System.out.println(cur_local_path + "\t" + cur_file);
}
}
}
unlock(lockName);
return 0;
}