All Downloads are FREE. Search and download functionalities are using the official Maven repository.

jersey.repackaged.com.google.common.io.ByteSource Maven / Gradle / Ivy

There is a newer version: 2.26-b03
Show newest version
/*
 * Copyright (C) 2012 The Guava Authors
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package jersey.repackaged.com.google.common.io;

import static jersey.repackaged.com.google.common.base.Preconditions.checkArgument;
import static jersey.repackaged.com.google.common.base.Preconditions.checkNotNull;

import jersey.repackaged.com.google.common.annotations.Beta;
import jersey.repackaged.com.google.common.base.Ascii;
import jersey.repackaged.com.google.common.collect.ImmutableList;
import jersey.repackaged.com.google.common.hash.Funnels;
import jersey.repackaged.com.google.common.hash.HashCode;
import jersey.repackaged.com.google.common.hash.HashFunction;
import jersey.repackaged.com.google.common.hash.Hasher;

import java.io.BufferedInputStream;
import java.io.ByteArrayInputStream;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.io.OutputStream;
import java.io.Reader;
import java.nio.charset.Charset;
import java.util.Arrays;
import java.util.Iterator;

/**
 * A readable source of bytes, such as a file. Unlike an {@link InputStream}, a
 * {@code ByteSource} is not an open, stateful stream for input that can be read and closed.
 * Instead, it is an immutable supplier of {@code InputStream} instances.
 *
 * 

{@code ByteSource} provides two kinds of methods: *

    *
  • Methods that return a stream: These methods should return a new, independent * instance each time they are called. The caller is responsible for ensuring that the returned * stream is closed. *
  • Convenience methods: These are implementations of common operations that are * typically implemented by opening a stream using one of the methods in the first category, doing * something and finally closing the stream that was opened. *
* * @since 14.0 * @author Colin Decker */ public abstract class ByteSource { private static final int BUF_SIZE = 0x1000; // 4K /** * Constructor for use by subclasses. */ protected ByteSource() {} /** * Returns a {@link CharSource} view of this byte source that decodes bytes read from this source * as characters using the given {@link Charset}. */ public CharSource asCharSource(Charset charset) { return new AsCharSource(charset); } /** * Opens a new {@link InputStream} for reading from this source. This method should return a new, * independent stream each time it is called. * *

The caller is responsible for ensuring that the returned stream is closed. * * @throws IOException if an I/O error occurs in the process of opening the stream */ public abstract InputStream openStream() throws IOException; /** * Opens a new buffered {@link InputStream} for reading from this source. The returned stream is * not required to be a {@link BufferedInputStream} in order to allow implementations to simply * delegate to {@link #openStream()} when the stream returned by that method does not benefit * from additional buffering (for example, a {@code ByteArrayInputStream}). This method should * return a new, independent stream each time it is called. * *

The caller is responsible for ensuring that the returned stream is closed. * * @throws IOException if an I/O error occurs in the process of opening the stream * @since 15.0 (in 14.0 with return type {@link BufferedInputStream}) */ public InputStream openBufferedStream() throws IOException { InputStream in = openStream(); return (in instanceof BufferedInputStream) ? (BufferedInputStream) in : new BufferedInputStream(in); } /** * Returns a view of a slice of this byte source that is at most {@code length} bytes long * starting at the given {@code offset}. * * @throws IllegalArgumentException if {@code offset} or {@code length} is negative */ public ByteSource slice(long offset, long length) { return new SlicedByteSource(offset, length); } /** * Returns whether the source has zero bytes. The default implementation is to open a stream and * check for EOF. * * @throws IOException if an I/O error occurs * @since 15.0 */ public boolean isEmpty() throws IOException { Closer closer = Closer.create(); try { InputStream in = closer.register(openStream()); return in.read() == -1; } catch (Throwable e) { throw closer.rethrow(e); } finally { closer.close(); } } /** * Returns the size of this source in bytes. For most implementations, this is a heavyweight * operation that will open a stream, read (or {@link InputStream#skip(long) skip}, if possible) * to the end of the stream and return the total number of bytes that were read. * *

For some sources, such as a file, this method may use a more efficient implementation. Note * that in such cases, it is possible that this method will return a different number of * bytes than would be returned by reading all of the bytes (for example, some special files may * return a size of 0 despite actually having content when read). * *

In either case, if this is a mutable source such as a file, the size it returns may not be * the same number of bytes a subsequent read would return. * * @throws IOException if an I/O error occurs in the process of reading the size of this source */ public long size() throws IOException { Closer closer = Closer.create(); try { InputStream in = closer.register(openStream()); return countBySkipping(in); } catch (IOException e) { // skip may not be supported... at any rate, try reading } finally { closer.close(); } closer = Closer.create(); try { InputStream in = closer.register(openStream()); return countByReading(in); } catch (Throwable e) { throw closer.rethrow(e); } finally { closer.close(); } } /** * Counts the bytes in the given input stream using skip if possible. Returns SKIP_FAILED if the * first call to skip threw, in which case skip may just not be supported. */ private long countBySkipping(InputStream in) throws IOException { long count = 0; while (true) { // don't try to skip more than available() // things may work really wrong with FileInputStream otherwise long skipped = in.skip(Math.min(in.available(), Integer.MAX_VALUE)); if (skipped <= 0) { if (in.read() == -1) { return count; } else if (count == 0 && in.available() == 0) { // if available is still zero after reading a single byte, it // will probably always be zero, so we should countByReading throw new IOException(); } count++; } else { count += skipped; } } } private static final byte[] countBuffer = new byte[BUF_SIZE]; private long countByReading(InputStream in) throws IOException { long count = 0; long read; while ((read = in.read(countBuffer)) != -1) { count += read; } return count; } /** * Copies the contents of this byte source to the given {@code OutputStream}. Does not close * {@code output}. * * @throws IOException if an I/O error occurs in the process of reading from this source or * writing to {@code output} */ public long copyTo(OutputStream output) throws IOException { checkNotNull(output); Closer closer = Closer.create(); try { InputStream in = closer.register(openStream()); return ByteStreams.copy(in, output); } catch (Throwable e) { throw closer.rethrow(e); } finally { closer.close(); } } /** * Copies the contents of this byte source to the given {@code ByteSink}. * * @throws IOException if an I/O error occurs in the process of reading from this source or * writing to {@code sink} */ public long copyTo(ByteSink sink) throws IOException { checkNotNull(sink); Closer closer = Closer.create(); try { InputStream in = closer.register(openStream()); OutputStream out = closer.register(sink.openStream()); return ByteStreams.copy(in, out); } catch (Throwable e) { throw closer.rethrow(e); } finally { closer.close(); } } /** * Reads the full contents of this byte source as a byte array. * * @throws IOException if an I/O error occurs in the process of reading from this source */ public byte[] read() throws IOException { Closer closer = Closer.create(); try { InputStream in = closer.register(openStream()); return ByteStreams.toByteArray(in); } catch (Throwable e) { throw closer.rethrow(e); } finally { closer.close(); } } /** * Reads the contents of this byte source using the given {@code processor} to process bytes as * they are read. Stops when all bytes have been read or the consumer returns {@code false}. * Returns the result produced by the processor. * * @throws IOException if an I/O error occurs in the process of reading from this source or if * {@code processor} throws an {@code IOException} * @since 16.0 */ @Beta public T read(ByteProcessor processor) throws IOException { checkNotNull(processor); Closer closer = Closer.create(); try { InputStream in = closer.register(openStream()); return ByteStreams.readBytes(in, processor); } catch (Throwable e) { throw closer.rethrow(e); } finally { closer.close(); } } /** * Hashes the contents of this byte source using the given hash function. * * @throws IOException if an I/O error occurs in the process of reading from this source */ public HashCode hash(HashFunction hashFunction) throws IOException { Hasher hasher = hashFunction.newHasher(); copyTo(Funnels.asOutputStream(hasher)); return hasher.hash(); } /** * Checks that the contents of this byte source are equal to the contents of the given byte * source. * * @throws IOException if an I/O error occurs in the process of reading from this source or * {@code other} */ public boolean contentEquals(ByteSource other) throws IOException { checkNotNull(other); byte[] buf1 = new byte[BUF_SIZE]; byte[] buf2 = new byte[BUF_SIZE]; Closer closer = Closer.create(); try { InputStream in1 = closer.register(openStream()); InputStream in2 = closer.register(other.openStream()); while (true) { int read1 = ByteStreams.read(in1, buf1, 0, BUF_SIZE); int read2 = ByteStreams.read(in2, buf2, 0, BUF_SIZE); if (read1 != read2 || !Arrays.equals(buf1, buf2)) { return false; } else if (read1 != BUF_SIZE) { return true; } } } catch (Throwable e) { throw closer.rethrow(e); } finally { closer.close(); } } /** * Concatenates multiple {@link ByteSource} instances into a single source. Streams returned from * the source will contain the concatenated data from the streams of the underlying sources. * *

Only one underlying stream will be open at a time. Closing the concatenated stream will * close the open underlying stream. * * @param sources the sources to concatenate * @return a {@code ByteSource} containing the concatenated data * @since 15.0 */ public static ByteSource concat(Iterable sources) { return new ConcatenatedByteSource(sources); } /** * Concatenates multiple {@link ByteSource} instances into a single source. Streams returned from * the source will contain the concatenated data from the streams of the underlying sources. * *

Only one underlying stream will be open at a time. Closing the concatenated stream will * close the open underlying stream. * *

Note: The input {@code Iterator} will be copied to an {@code ImmutableList} when this * method is called. This will fail if the iterator is infinite and may cause problems if the * iterator eagerly fetches data for each source when iterated (rather than producing sources * that only load data through their streams). Prefer using the {@link #concat(Iterable)} * overload if possible. * * @param sources the sources to concatenate * @return a {@code ByteSource} containing the concatenated data * @throws NullPointerException if any of {@code sources} is {@code null} * @since 15.0 */ public static ByteSource concat(Iterator sources) { return concat(ImmutableList.copyOf(sources)); } /** * Concatenates multiple {@link ByteSource} instances into a single source. Streams returned from * the source will contain the concatenated data from the streams of the underlying sources. * *

Only one underlying stream will be open at a time. Closing the concatenated stream will * close the open underlying stream. * * @param sources the sources to concatenate * @return a {@code ByteSource} containing the concatenated data * @throws NullPointerException if any of {@code sources} is {@code null} * @since 15.0 */ public static ByteSource concat(ByteSource... sources) { return concat(ImmutableList.copyOf(sources)); } /** * Returns a view of the given byte array as a {@link ByteSource}. To view only a specific range * in the array, use {@code ByteSource.wrap(b).slice(offset, length)}. * * @since 15.0 (since 14.0 as {@code ByteStreams.asByteSource(byte[])}). */ public static ByteSource wrap(byte[] b) { return new ByteArrayByteSource(b); } /** * Returns an immutable {@link ByteSource} that contains no bytes. * * @since 15.0 */ public static ByteSource empty() { return EmptyByteSource.INSTANCE; } /** * A char source that reads bytes from this source and decodes them as characters using a * charset. */ private final class AsCharSource extends CharSource { private final Charset charset; private AsCharSource(Charset charset) { this.charset = checkNotNull(charset); } @Override public Reader openStream() throws IOException { return new InputStreamReader(ByteSource.this.openStream(), charset); } @Override public String toString() { return ByteSource.this.toString() + ".asCharSource(" + charset + ")"; } } /** * A view of a subsection of the containing byte source. */ private final class SlicedByteSource extends ByteSource { private final long offset; private final long length; private SlicedByteSource(long offset, long length) { checkArgument(offset >= 0, "offset (%s) may not be negative", offset); checkArgument(length >= 0, "length (%s) may not be negative", length); this.offset = offset; this.length = length; } @Override public InputStream openStream() throws IOException { return sliceStream(ByteSource.this.openStream()); } @Override public InputStream openBufferedStream() throws IOException { return sliceStream(ByteSource.this.openBufferedStream()); } private InputStream sliceStream(InputStream in) throws IOException { if (offset > 0) { try { ByteStreams.skipFully(in, offset); } catch (Throwable e) { Closer closer = Closer.create(); closer.register(in); try { throw closer.rethrow(e); } finally { closer.close(); } } } return ByteStreams.limit(in, length); } @Override public ByteSource slice(long offset, long length) { checkArgument(offset >= 0, "offset (%s) may not be negative", offset); checkArgument(length >= 0, "length (%s) may not be negative", length); long maxLength = this.length - offset; return ByteSource.this.slice(this.offset + offset, Math.min(length, maxLength)); } @Override public boolean isEmpty() throws IOException { return length == 0 || super.isEmpty(); } @Override public String toString() { return ByteSource.this.toString() + ".slice(" + offset + ", " + length + ")"; } } private static class ByteArrayByteSource extends ByteSource { protected final byte[] bytes; protected ByteArrayByteSource(byte[] bytes) { this.bytes = checkNotNull(bytes); } @Override public InputStream openStream() { return new ByteArrayInputStream(bytes); } @Override public InputStream openBufferedStream() throws IOException { return openStream(); } @Override public boolean isEmpty() { return bytes.length == 0; } @Override public long size() { return bytes.length; } @Override public byte[] read() { return bytes.clone(); } @Override public long copyTo(OutputStream output) throws IOException { output.write(bytes); return bytes.length; } @Override public T read(ByteProcessor processor) throws IOException { processor.processBytes(bytes, 0, bytes.length); return processor.getResult(); } @Override public HashCode hash(HashFunction hashFunction) throws IOException { return hashFunction.hashBytes(bytes); } // TODO(user): Possibly override slice() @Override public String toString() { return "ByteSource.wrap(" + Ascii.truncate(BaseEncoding.base16().encode(bytes), 30, "...") + ")"; } } private static final class EmptyByteSource extends ByteArrayByteSource { private static final EmptyByteSource INSTANCE = new EmptyByteSource(); private EmptyByteSource() { super(new byte[0]); } @Override public CharSource asCharSource(Charset charset) { checkNotNull(charset); return CharSource.empty(); } @Override public byte[] read() { return bytes; // length is 0, no need to clone } @Override public String toString() { return "ByteSource.empty()"; } } private static final class ConcatenatedByteSource extends ByteSource { private final Iterable sources; ConcatenatedByteSource(Iterable sources) { this.sources = checkNotNull(sources); } @Override public InputStream openStream() throws IOException { return new MultiInputStream(sources.iterator()); } @Override public boolean isEmpty() throws IOException { for (ByteSource source : sources) { if (!source.isEmpty()) { return false; } } return true; } @Override public long size() throws IOException { long result = 0L; for (ByteSource source : sources) { result += source.size(); } return result; } @Override public String toString() { return "ByteSource.concat(" + sources + ")"; } } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy