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));
}
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 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;
}
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();
}
}
/*
* 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 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();
}
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;
}
