improved comments and formatting

This commit is contained in:
Szum123321 2022-11-29 22:31:04 +01:00
parent 8385044154
commit 472aeda184
10 changed files with 52 additions and 30 deletions

View File

@ -51,6 +51,8 @@ public class Globals {
private AwaitThread restoreAwaitThread = null;
private Path lockedPath = null;
private String combinedVersionString;
private Globals() {}
public ExecutorService getQueueExecutor() { return executorService; }
@ -107,4 +109,12 @@ public class Globals {
if(disableTMPFiles) log.error("Might cause: https://github.com/Szum123321/textile_backup/wiki/ZIP-Problems");
}
public String getCombinedVersionString() {
return combinedVersionString;
}
public void setCombinedVersionString(String combinedVersionString) {
this.combinedVersionString = combinedVersionString;
}
}

View File

@ -25,10 +25,9 @@ import java.io.InputStream;
import java.nio.file.Path;
import java.util.Optional;
public interface InputSupplier extends InputStreamSupplier {
InputStream getInputStream() throws IOException;
//If an entry is virtual (a.k.a there is no actual file to open, only input stream)
//If an entry is virtual (a.k.a. there is no actual file to open, only input stream)
Optional<Path> getPath();
String getName();

View File

@ -40,7 +40,7 @@ import java.util.concurrent.atomic.AtomicInteger;
import java.util.stream.Stream;
/**
* Basic abstract class representing directory compressor
* Basic abstract class representing directory compressor with all the bells and whistles
*/
public abstract class AbstractCompressor {
private final static TextileLogger log = new TextileLogger(TextileBackup.MOD_NAME);
@ -49,14 +49,14 @@ public abstract class AbstractCompressor {
Instant start = Instant.now();
FileTreeHashBuilder fileHashBuilder = new FileTreeHashBuilder();
BrokenFileHandler brokenFileHandler = new BrokenFileHandler();
BrokenFileHandler brokenFileHandler = new BrokenFileHandler(); //Basically a hashmap storing files and their respective exceptions
try (OutputStream outStream = Files.newOutputStream(outputFile);
BufferedOutputStream bufferedOutputStream = new BufferedOutputStream(outStream);
OutputStream arc = createArchiveOutputStream(bufferedOutputStream, ctx, coreLimit);
Stream<Path> fileStream = Files.walk(inputFile)) {
AtomicInteger fileCounter = new AtomicInteger(0);
AtomicInteger fileCounter = new AtomicInteger(0); //number of files to compress
var it = fileStream
.filter(path -> !Utilities.isBlacklisted(inputFile.relativize(path)))
@ -66,6 +66,7 @@ public abstract class AbstractCompressor {
log.info("File count: {}", fileCounter.get());
//will be used in conjunction with ParallelZip to avoid race condition
CountDownLatch latch = new CountDownLatch(fileCounter.get());
while(it.hasNext()) {
@ -83,6 +84,7 @@ public abstract class AbstractCompressor {
);
} catch (IOException e) {
brokenFileHandler.handle(file, e);
//In Permissive mode we allow partial backups
if(ConfigHelper.INSTANCE.get().errorErrorHandlingMode.isStrict()) throw e;
else log.sendErrorAL(ctx, "An exception occurred while trying to compress: {}",
inputFile.relativize(file).toString(), e
@ -90,6 +92,7 @@ public abstract class AbstractCompressor {
}
}
//wait for all the InputStreams to close/fail with InputSupplier
latch.await();
Instant now = Instant.now();

View File

@ -45,7 +45,7 @@ public class AbstractTarArchiver extends AbstractCompressor {
protected void addEntry(InputSupplier input, OutputStream arc) throws IOException {
try (InputStream fileInputStream = input.getInputStream()) {
TarArchiveEntry entry;
if(input.getPath().isEmpty()) {//Virtual entry
if(input.getPath().isEmpty()) { //Virtual entry
entry = new TarArchiveEntry(input.getName());
entry.setSize(input.size());
} else

View File

@ -22,16 +22,15 @@ import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.util.Arrays;
/*
This algorithm copies construction of SeaHash (https://ticki.github.io/blog/seahash-explained/) including its IV
What it differs in is that it uses Xoroshift64* instead of PCG. Although it might lower the output quality,
I don't think it matters that much, honestly. One advantage the xoroshift has is that it should be
easier to implement with AVX. Java should soon ship its vector api by default.
/**
* This algorithm copies construction of <a href="https://ticki.github.io/blog/seahash-explained/">SeaHash</a> including its IV.
* What it differs in is that it uses Xoroshift64* instead of PCG as its pseudo-random function. Although it might lower
* the output quality, I don't think it matters that much, honestly. One advantage of xoroshift is that it should be
* easier to implement with AVX. Java should soon ship its vector api by default.
*/
public class BalticHash implements Hash {
protected final static long[] IV = { 0x16f11fe89b0d677cL, 0xb480a793d8e6c86cL, 0x6fe2e5aaf078ebc9L, 0x14f994a4c5259381L};
//SeaHash IV
protected final static long[] IV = { 0x16f11fe89b0d677cL, 0xb480a793d8e6c86cL, 0x6fe2e5aaf078ebc9L, 0x14f994a4c5259381L };
private final long[] state = Arrays.copyOf(IV, IV.length);
protected final int buffer_limit = state.length * Long.BYTES;
protected final byte[] _byte_buffer = new byte[(state.length + 1) * Long.BYTES];

View File

@ -22,16 +22,16 @@ package net.szum123321.textile_backup.core.digest;
import net.szum123321.textile_backup.core.digest.BalticHash;
/*Mostly working XorSeaHash impl using SIMD. Should speed up calculation on most systems currently in use
/**
* Mostly working XorSeaHash impl using SIMD. Should speed up calculation on most systems currently in use
It's actually slower. I tested it by comparing runtimes while hashing a directly opened FileInputStream.
My cpu is AMD Ryzen 5 3500U
<br>...<br>
There are two reasons I can think of: either vector construction simply takes so much time or jvm auto-vectorizes better than me
It's still probably far from being the slowest part of code, so I don't expect any major slowdowns
I will keep this code here for future work perhaps
* It's actually slower. I tested it by comparing runtimes while hashing a directly opened FileInputStream.
* My cpu is AMD Ryzen 5 3500U
* There are two reasons I can think of: either vector construction simply takes so much time or jvm auto-vectorizes better than I.
* It's still probably far from being the slowest part of code, so I don't expect any major slowdowns
* I will keep this code here for future work perhaps
*/
public class BalticHashSIMD extends BalticHash {/*
public BalticHashSIMD() { throw new UnsupportedOperationException(); } //For safety

View File

@ -29,6 +29,10 @@ import java.nio.file.Files;
import java.nio.file.Path;
import java.util.concurrent.atomic.AtomicBoolean;
/**
* What this class does is it collects the hashed files and combines them into a single number,
* thus we can verify file tree integrity
*/
public class FileTreeHashBuilder {
private final static TextileLogger log = new TextileLogger(TextileBackup.MOD_NAME);
private final Object lock = new Object();

View File

@ -21,13 +21,12 @@ package net.szum123321.textile_backup.core.digest;
public interface Hash {
void update(byte b);
default void update(int b) {
update((byte)b);
}
default void update(int b) { update((byte)b); }
void update(long b);
default void update(byte[] b) {
update(b, 0, b.length);
}
default void update(byte[] b) { update(b, 0, b.length); }
void update(byte[] b, int off, int len);
long getValue();

View File

@ -27,8 +27,14 @@ import java.io.*;
import java.nio.file.Path;
import java.util.concurrent.CountDownLatch;
//This class calculates a hash of the file on the input stream, submits it to FileTreeHashBuilder.
//In case the underlying stream hasn't been read completely in, puts it into BrokeFileHandler
/**
* This class calculates a hash of the file on the input stream, submits it to FileTreeHashBuilder.
* In case the underlying stream hasn't been read completely in, puts it into BrokeFileHandler
* Futhermore, ParallelZip works by putting al the file requests into a queue and then compressing them
* with multiple threads. Thus, we have to make sure that all the files have been read before requesting the final value
* That is what CountDownLatch does
*/
public class HashingInputStream extends FilterInputStream {
private final Path path;
private final Hash hasher = Globals.CHECKSUM_SUPPLIER.get();
@ -36,7 +42,6 @@ public class HashingInputStream extends FilterInputStream {
private final BrokenFileHandler brokenFileHandler;
private final CountDownLatch latch;
public HashingInputStream(InputStream in, Path path, FileTreeHashBuilder hashBuilder, BrokenFileHandler brokenFileHandler, CountDownLatch latch) {
super(in);
this.path = path;

View File

@ -37,6 +37,9 @@ import java.nio.file.Path;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.FutureTask;
/**
* This class restores a file provided by RestoreContext.
*/
public class RestoreBackupRunnable implements Runnable {
private final static TextileLogger log = new TextileLogger(TextileBackup.MOD_NAME);
private final static ConfigHelper config = ConfigHelper.INSTANCE;