19 package org.sleuthkit.autopsy.ingest;
21 import java.util.ArrayList;
22 import java.util.Collection;
23 import java.util.Comparator;
24 import java.util.HashSet;
25 import java.util.Iterator;
26 import java.util.List;
28 import java.util.TreeSet;
29 import java.util.concurrent.BlockingDeque;
30 import java.util.concurrent.LinkedBlockingDeque;
31 import java.util.concurrent.LinkedBlockingQueue;
32 import java.util.logging.Level;
33 import java.util.regex.Matcher;
34 import java.util.regex.Pattern;
46 final class IngestTasksScheduler {
48 private static final Logger logger = Logger.getLogger(IngestTasksScheduler.class.getName());
49 private static final int FAT_NTFS_FLAGS = TskData.
TSK_FS_TYPE_ENUM.
TSK_FS_TYPE_FAT12.getValue() | TskData.TSK_FS_TYPE_ENUM.TSK_FS_TYPE_FAT16.getValue() | TskData.TSK_FS_TYPE_ENUM.TSK_FS_TYPE_FAT32.getValue() | TskData.TSK_FS_TYPE_ENUM.TSK_FS_TYPE_NTFS.getValue();
50 private static IngestTasksScheduler instance;
58 private final LinkedBlockingQueue<DataSourceIngestTask> pendingDataSourceTasks;
59 private final DataSourceIngestTaskQueue dataSourceTasksDispenser;
84 private final TreeSet<FileIngestTask> rootDirectoryTasks;
85 private final List<FileIngestTask> directoryTasks;
86 private final BlockingDeque<FileIngestTask> pendingFileTasks;
87 private final FileIngestTaskQueue fileTasksDispenser;
100 private final Set<IngestTask> tasksInProgress;
105 synchronized static IngestTasksScheduler getInstance() {
106 if (IngestTasksScheduler.instance == null) {
107 IngestTasksScheduler.instance =
new IngestTasksScheduler();
109 return IngestTasksScheduler.instance;
115 private IngestTasksScheduler() {
116 this.pendingDataSourceTasks =
new LinkedBlockingQueue<>();
117 this.dataSourceTasksDispenser =
new DataSourceIngestTaskQueue();
118 this.rootDirectoryTasks =
new TreeSet<>(
new RootDirectoryTaskComparator());
119 this.directoryTasks =
new ArrayList<>();
120 this.pendingFileTasks =
new LinkedBlockingDeque<>();
121 this.fileTasksDispenser =
new FileIngestTaskQueue();
122 this.tasksInProgress =
new HashSet<>();
131 IngestTaskQueue getDataSourceIngestTaskQueue() {
132 return this.dataSourceTasksDispenser;
141 IngestTaskQueue getFileIngestTaskQueue() {
142 return this.fileTasksDispenser;
153 synchronized void scheduleIngestTasks(DataSourceIngestJob job) {
159 this.scheduleDataSourceIngestTask(job);
160 this.scheduleFileIngestTasks(job);
168 synchronized void scheduleDataSourceIngestTask(DataSourceIngestJob job) {
169 DataSourceIngestTask task =
new DataSourceIngestTask(job);
170 this.tasksInProgress.add(task);
172 this.pendingDataSourceTasks.put(task);
173 }
catch (InterruptedException ex) {
178 this.tasksInProgress.remove(task);
179 Thread.currentThread().interrupt();
188 synchronized void scheduleFileIngestTasks(DataSourceIngestJob job) {
191 List<AbstractFile> topLevelFiles = getTopLevelFiles(job.getDataSource());
192 for (AbstractFile firstLevelFile : topLevelFiles) {
193 FileIngestTask task =
new FileIngestTask(job, firstLevelFile);
194 if (IngestTasksScheduler.shouldEnqueueFileTask(task)) {
195 this.tasksInProgress.add(task);
196 this.rootDirectoryTasks.add(task);
199 shuffleFileTaskQueues();
208 synchronized void scheduleFileIngestTask(DataSourceIngestJob job, AbstractFile file) {
209 FileIngestTask task =
new FileIngestTask(job, file);
210 if (IngestTasksScheduler.shouldEnqueueFileTask(task)) {
211 this.tasksInProgress.add(task);
212 addToPendingFileTasksQueue(task);
222 synchronized void notifyTaskCompleted(IngestTask task) {
223 tasksInProgress.remove(task);
233 synchronized boolean tasksForJobAreCompleted(DataSourceIngestJob job) {
234 for (IngestTask task : tasksInProgress) {
235 if (task.getIngestJob().getId() == job.getId()) {
250 synchronized void cancelPendingTasksForIngestJob(DataSourceIngestJob job) {
251 long jobId = job.getId();
252 this.removeTasksForJob(this.rootDirectoryTasks, jobId);
253 this.removeTasksForJob(this.directoryTasks, jobId);
254 this.removeTasksForJob(this.pendingFileTasks, jobId);
255 this.removeTasksForJob(this.pendingDataSourceTasks, jobId);
256 this.shuffleFileTaskQueues();
267 private static List<AbstractFile> getTopLevelFiles(Content dataSource) {
268 List<AbstractFile> topLevelFiles =
new ArrayList<>();
269 Collection<AbstractFile> rootObjects = dataSource.accept(
new GetRootDirectoryVisitor());
270 if (rootObjects.isEmpty() && dataSource instanceof AbstractFile) {
272 topLevelFiles.add((AbstractFile) dataSource);
274 for (AbstractFile root : rootObjects) {
275 List<Content> children;
277 children = root.getChildren();
278 if (children.isEmpty()) {
281 topLevelFiles.add(root);
285 for (Content child : children) {
286 if (child instanceof AbstractFile) {
287 topLevelFiles.add((AbstractFile) child);
291 }
catch (TskCoreException ex) {
292 logger.log(Level.WARNING,
"Could not get children of root to enqueue: " + root.getId() +
": " + root.getName(), ex);
296 return topLevelFiles;
304 synchronized private void shuffleFileTaskQueues() {
311 if (!this.pendingFileTasks.isEmpty()) {
315 if (this.directoryTasks.isEmpty()) {
316 if (this.rootDirectoryTasks.isEmpty()) {
325 this.directoryTasks.add(this.rootDirectoryTasks.pollFirst());
331 FileIngestTask directoryTask = this.directoryTasks.remove(this.directoryTasks.size() - 1);
332 if (shouldEnqueueFileTask(directoryTask)) {
333 addToPendingFileTasksQueue(directoryTask);
335 this.tasksInProgress.remove(directoryTask);
340 final AbstractFile directory = directoryTask.getFile();
342 for (Content child : directory.getChildren()) {
343 if (child instanceof AbstractFile) {
344 AbstractFile file = (AbstractFile) child;
345 FileIngestTask childTask =
new FileIngestTask(directoryTask.getIngestJob(), file);
346 if (file.hasChildren()) {
352 this.tasksInProgress.add(childTask);
353 this.directoryTasks.add(childTask);
354 }
else if (shouldEnqueueFileTask(childTask)) {
357 this.tasksInProgress.add(childTask);
358 addToPendingFileTasksQueue(childTask);
362 }
catch (TskCoreException ex) {
363 String errorMessage = String.format(
"An error occurred getting the children of %s", directory.getName());
364 logger.log(Level.SEVERE, errorMessage, ex);
377 private static boolean shouldEnqueueFileTask(
final FileIngestTask task) {
378 final AbstractFile file = task.getFile();
382 if (!task.getIngestJob().shouldProcessUnallocatedSpace()
383 && file.getType().equals(TskData.TSK_DB_FILES_TYPE_ENUM.UNALLOC_BLOCKS)) {
389 String fileName = file.getName();
390 if (fileName.equals(
".") || fileName.equals(
"..")) {
400 TskData.TSK_FS_TYPE_ENUM fsType = TskData.TSK_FS_TYPE_ENUM.TSK_FS_TYPE_UNSUPP;
406 }
catch (TskCoreException ex) {
407 logger.log(Level.SEVERE,
"Error querying file system for " + f, ex);
411 if ((fsType.getValue() & FAT_NTFS_FLAGS) == 0) {
416 boolean isInRootDir =
false;
418 AbstractFile parent = f.getParentDirectory();
419 isInRootDir = parent.isRoot();
420 }
catch (TskCoreException ex) {
421 logger.log(Level.WARNING,
"Error querying parent directory for" + f.getName(), ex);
427 if (isInRootDir && f.getMetaAddr() < 32) {
428 String name = f.getName();
429 if (name.length() > 0 && name.charAt(0) ==
'$' && name.contains(
":")) {
443 synchronized private void addToPendingFileTasksQueue(FileIngestTask task) {
445 this.pendingFileTasks.putFirst(task);
446 }
catch (InterruptedException ex) {
451 this.tasksInProgress.remove(task);
452 Thread.currentThread().interrupt();
464 private void removeTasksForJob(Collection<? extends IngestTask> taskQueue,
long jobId) {
465 Iterator<? extends IngestTask> iterator = taskQueue.iterator();
466 while (iterator.hasNext()) {
467 IngestTask task = iterator.next();
468 if (task.getIngestJob().getId() == jobId) {
469 this.tasksInProgress.remove(task);
482 private static int countTasksForJob(Collection<? extends IngestTask> queue,
long jobId) {
483 Iterator<? extends IngestTask> iterator = queue.iterator();
485 while (iterator.hasNext()) {
486 IngestTask task = (IngestTask) iterator.next();
487 if (task.getIngestJob().getId() == jobId) {
501 synchronized IngestJobTasksSnapshot getTasksSnapshotForJob(
long jobId) {
502 return new IngestJobTasksSnapshot(jobId);
512 public int compare(FileIngestTask q1, FileIngestTask q2) {
516 return (
int) (q2.getFile().getId() - q1.getFile().getId());
518 return p2.ordinal() - p1.ordinal();
526 LAST, LOW, MEDIUM, HIGH
529 static final List<Pattern> LAST_PRI_PATHS =
new ArrayList<>();
531 static final List<Pattern> LOW_PRI_PATHS =
new ArrayList<>();
533 static final List<Pattern> MEDIUM_PRI_PATHS =
new ArrayList<>();
535 static final List<Pattern> HIGH_PRI_PATHS =
new ArrayList<>();
551 LAST_PRI_PATHS.add(Pattern.compile(
"^pagefile", Pattern.CASE_INSENSITIVE));
552 LAST_PRI_PATHS.add(Pattern.compile(
"^hiberfil", Pattern.CASE_INSENSITIVE));
555 LOW_PRI_PATHS.add(Pattern.compile(
"^\\$OrphanFiles", Pattern.CASE_INSENSITIVE));
556 LOW_PRI_PATHS.add(Pattern.compile(
"^Windows", Pattern.CASE_INSENSITIVE));
558 MEDIUM_PRI_PATHS.add(Pattern.compile(
"^Program Files", Pattern.CASE_INSENSITIVE));
560 HIGH_PRI_PATHS.add(Pattern.compile(
"^Users", Pattern.CASE_INSENSITIVE));
561 HIGH_PRI_PATHS.add(Pattern.compile(
"^Documents and Settings", Pattern.CASE_INSENSITIVE));
562 HIGH_PRI_PATHS.add(Pattern.compile(
"^home", Pattern.CASE_INSENSITIVE));
563 HIGH_PRI_PATHS.add(Pattern.compile(
"^ProgramData", Pattern.CASE_INSENSITIVE));
580 final String path = abstractFile.
getName();
584 for (Pattern p : HIGH_PRI_PATHS) {
585 Matcher m = p.matcher(path);
590 for (Pattern p : MEDIUM_PRI_PATHS) {
591 Matcher m = p.matcher(path);
596 for (Pattern p : LOW_PRI_PATHS) {
597 Matcher m = p.matcher(path);
602 for (Pattern p : LAST_PRI_PATHS) {
603 Matcher m = p.matcher(path);
625 return IngestTasksScheduler.this.pendingDataSourceTasks.take();
640 FileIngestTask task = IngestTasksScheduler.this.pendingFileTasks.takeFirst();
641 shuffleFileTaskQueues();
650 class IngestJobTasksSnapshot {
652 private final long jobId;
653 private final long rootQueueSize;
654 private final long dirQueueSize;
655 private final long fileQueueSize;
656 private final long dsQueueSize;
657 private final long runningListSize;
664 IngestJobTasksSnapshot(
long jobId) {
666 this.rootQueueSize = countTasksForJob(IngestTasksScheduler.this.rootDirectoryTasks, jobId);
667 this.dirQueueSize = countTasksForJob(IngestTasksScheduler.this.directoryTasks, jobId);
668 this.fileQueueSize = countTasksForJob(IngestTasksScheduler.this.pendingFileTasks, jobId);
669 this.dsQueueSize = countTasksForJob(IngestTasksScheduler.this.pendingDataSourceTasks, jobId);
670 this.runningListSize = countTasksForJob(IngestTasksScheduler.this.tasksInProgress, jobId);
689 long getRootQueueSize() {
690 return rootQueueSize;
699 long getDirectoryTasksQueueSize() {
703 long getFileQueueSize() {
704 return fileQueueSize;
707 long getDsQueueSize() {
711 long getRunningListSize() {
712 return runningListSize;
TskData.TSK_DB_FILES_TYPE_ENUM getType()
FileSystem getFileSystem()
int compare(FileIngestTask q1, FileIngestTask q2)
TskData.TSK_FS_TYPE_ENUM getFsType()