imageinfo.java

开源的HTML文本编辑器

/*
 * ImageInfo.java
 *
 * Version 1.9
 *
 * A Java class to determine image width, height and color depth for
 * a number of image file formats.
 *
 * Written by Marco Schmidt 
 *
 * Contributed to the Public Domain.
 */
package org.devlib.schmidt.imageinfo;

import java.io.DataInput;
import java.io.FileInputStream;
import java.io.InputStream;
import java.io.IOException;
import java.net.URL;
import java.util.Vector;

/**
 * Get file format, image resolution, number of bits per pixel and optionally 
 * number of images, comments and physical resolution from 
 * JPEG, GIF, BMP, PCX, PNG, IFF, RAS, PBM, PGM, PPM and PSD files 
 * (or input streams).
 * <p>
 * Use the class like this:
 * <pre>
 * ImageInfo ii = new ImageInfo();
 * ii.setInput(in); // in can be InputStream or RandomAccessFile
 * ii.setDetermineImageNumber(true); // default is false
 * ii.setCollectComments(true); // default is false
 * if (!ii.check()) {
 *   System.err.println("Not a supported image file format.");
 *   return;
 * }
 * System.out.println(ii.getFormatName() + ", " + ii.getMimeType() + 
 *   ", " + ii.getWidth() + " x " + ii.getHeight() + " pixels, " + 
 *   ii.getBitsPerPixel() + " bits per pixel, " + ii.getNumberOfImages() +
 *   " image(s), " + ii.getNumberOfComments() + " comment(s).");
 *  // there are other properties, check out the API documentation
 * </pre>
 * You can also use this class as a command line program.
 * Call it with a number of image file names and URLs as parameters:
 * <pre>
 *   java ImageInfo *.jpg *.png *.gif http://somesite.tld/image.jpg
 * </pre>
 * or call it without parameters and pipe data to it:
 * <pre>
 *   java ImageInfo &lt; image.jpg  
 * </pre>
 * <p>
 * Known limitations:
 * <ul>
 * <li>When the determination of the number of images is turned off, GIF bits 
 *  per pixel are only read from the global header.
 *  For some GIFs, local palettes change this to a typically larger
 *  value. To be certain to get the correct color depth, call
 *  setDetermineImageNumber(true) before calling check().
 *  The complete scan over the GIF file will take additional time.</li>
 * <li>Transparency information is not included in the bits per pixel count.
 *  Actually, it was my decision not to include those bits, so it's a feature! ;-)</li>
 * </ul>
 * <p>
 * Requirements:
 * <ul>
 * <li>Java 1.1 or higher</li>
 * </ul>
 * <p>
 * The latest version can be found at <a href="http://schmidt.devlib.org/image-info/">http://schmidt.devlib.org/image-info/</a>.
 * <p>
 * Written by Marco Schmidt.
 * <p>
 * This class is contributed to the Public Domain.
 * Use it at your own risk.
 * <p>
 * <a name="history">History</a>:
 * <ul>
 * <li><strong>2001-08-24</strong> Initial version.</li>
 * <li><strong>2001-10-13</strong> Added support for the file formats BMP and PCX.</li>
 * <li><strong>2001-10-16</strong> Fixed bug in read(int[], int, int) that returned
 * <li><strong>2002-01-22</strong> Added support for file formats Amiga IFF and Sun Raster (RAS).</li>
 * <li><strong>2002-01-24</strong> Added support for file formats Portable Bitmap / Graymap / Pixmap (PBM, PGM, PPM) and Adobe Photoshop (PSD).
 *   Added new method getMimeType() to return the MIME type associated with a particular file format.</li>
 * <li><strong>2002-03-15</strong> Added support to recognize number of images in file. Only works with GIF.
 *   Use {@link #setDetermineImageNumber} with <code>true</code> as argument to identify animated GIFs
 *   ({@link #getNumberOfImages()} will return a value larger than <code>1</code>).</li>
 * <li><strong>2002-04-10</strong> Fixed a bug in the feature 'determine number of images in animated GIF' introduced with version 1.1.
 *   Thanks to Marcelo P. Lima for sending in the bug report. 
 *   Released as 1.1.1.</li>
 * <li><strong>2002-04-18</strong> Added {@link #setCollectComments(boolean)}. 
 *  That new method lets the user specify whether textual comments are to be  
 *  stored in an internal list when encountered in an input image file / stream.
 *  Added two methods to return the physical width and height of the image in dpi: 
 *   {@link #getPhysicalWidthDpi()} and {@link #getPhysicalHeightDpi()}.
 *  If the physical resolution could not be retrieved, these methods return <code>-1</code>.
 *  </li>
 * <li><strong>2002-04-23</strong> Added support for the new properties physical resolution and
 *   comments for some formats. Released as 1.2.</li>
 * <li><strong>2002-06-17</strong> Added support for SWF, sent in by Michael Aird.
 *  Changed checkJpeg() so that other APP markers than APP0 will not lead to a failure anymore.
 *  Released as 1.3.</li>
 * <li><strong>2003-07-28</strong> Bug fix - skip method now takes return values into consideration.
 *  Less bytes than necessary may have been skipped, leading to flaws in the retrieved information in some cases.
 *  Thanks to Bernard Bernstein for pointing that out.
 *  Released as 1.4.</li>
 * <li><strong>2004-02-29</strong> Added support for recognizing progressive JPEG and
 *  interlaced PNG and GIF. A new method {@link #isProgressive()} returns whether ImageInfo
 *  has found that the storage type is progressive (or interlaced). 
 *  Thanks to Joe Germuska for suggesting the feature.
 *  Bug fix: BMP physical resolution is now correctly determined.
 *  Released as 1.5.</li>
 * <li><strong>2004-11-30</strong> Bug fix: recognizing progressive GIFs 
 * (interlaced in GIF terminology) did not work (thanks to Franz Jeitler for 
 *   pointing this out). Now it should work, but only if the number of images is determined.
 *  This is because information on interlacing is stored in a local image header.
 *  In theory, different images could be stored interlaced and non-interlaced in one 
 *  file. However, I think  that's unlikely. Right now, the last image in the GIF file 
 *  that is examined by ImageInfo is used for the "progressive" status.</li>
 * <li><strong>2005-01-02</strong> Some code clean up (unused methods and variables
 *  commented out, missing javadoc comments, etc.). Thanks to George Sexton for a long list.
 *  Removed usage of Boolean.toString because
 *  it's a Java 1.4+ feature (thanks to Gregor Dupont).
 *  Changed delimiter character in compact output from semicolon to tabulator
 * (for better integration with cut(1) and other Unix tools).
 *  Added some points to the <a href="http://schmidt.devlib.org/image-info/index.html#knownissues">'Known
 *  issues' section of the website</a>. 
 *  Released as 1.6.</li>
 * <li><strong>2005-07-26</strong> Removed code to identify Flash (SWF) files.
 *  Has repeatedly led to problems and support requests, and I don't know the
 *  format and don't have the time and interest to fix it myself.
 *  I repeatedly included fixes by others which didn't work for some people.
 *  I give up on SWF. Please do not contact me about it anymore.
 *  Set package of ImageInfo class to org.devlib.schmidt.imageinfo (a package
 *  was repeatedly requested by some users).
 *  Released as 1.7.</li>
 *  <li><strong>2006-02-23</strong> Removed Flash helper methods which weren't used elsewhere.
 *   Updated skip method which tries "read" whenever "skip(Bytes)" returns a result of 0.
 *   The old method didn't work with certain input stream types on truncated data streams.
 *   Thanks to Martin Leidig for reporting this and sending in code.
 *   Released as 1.8.</li>
 *  </li>
 *  <li><strong>2006-11-13</strong> Removed check that made ImageInfo report JPEG APPx
 *   markers smaller than 14 bytes as files in unknown format. Such JPEGs seem to be
 *   generated by Google's Picasa application. First reported with fix by 
 *   Karl von Randow. Released as 1.9.</li>  
 * </ul>
 * @author Marco Schmidt
 */
@SuppressWarnings("unchecked")
public class ImageInfo {
	/**
	 * Return value of {@link #getFormat()} for JPEG streams.
	 * ImageInfo can extract physical resolution and comments
	 * from JPEGs (only from APP0 headers).
	 * Only one image can be stored in a file.
	 * It is determined whether the JPEG stream is progressive 
	 * (see {@link #isProgressive()}).
	 */
	public static final int FORMAT_JPEG = 0;

	/**
	 * Return value of {@link #getFormat()} for GIF streams.
	 * ImageInfo can extract comments from GIFs and count the number
	 * of images (GIFs with more than one image are animations).
	 * It is determined whether the GIF stream is interlaced (see {@link #isProgressive()}).
	 */
	public static final int FORMAT_GIF = 1;

	/**
	 * Return value of {@link #getFormat()} for PNG streams.
	 * PNG only supports one image per file.
	 * Both physical resolution and comments can be stored with PNG,
	 * but ImageInfo is currently not able to extract those.
	 * It is determined whether the PNG stream is interlaced (see {@link #isProgressive()}).
	 */
	public static final int FORMAT_PNG = 2;

	/**
	 * Return value of {@link #getFormat()} for BMP streams.
	 * BMP only supports one image per file.
	 * BMP does not allow for comments.
	 * The physical resolution can be stored.
	 */
	public static final int FORMAT_BMP = 3;

	/**
	 * Return value of {@link #getFormat()} for PCX streams.
	 * PCX does not allow for comments or more than one image per file.
	 * However, the physical resolution can be stored.
	 */
	public static final int FORMAT_PCX = 4;

	/**
	 * Return value of {@link #getFormat()} for IFF streams.
	 */
	public static final int FORMAT_IFF = 5;

	/**
	 * Return value of {@link #getFormat()} for RAS streams.
	 * Sun Raster allows for one image per file only and is not able to
	 * store physical resolution or comments.
	 */
	public static final int FORMAT_RAS = 6;

	/** Return value of {@link #getFormat()} for PBM streams. */
	public static final int FORMAT_PBM = 7;

	/** Return value of {@link #getFormat()} for PGM streams. */
	public static final int FORMAT_PGM = 8;

	/** Return value of {@link #getFormat()} for PPM streams. */
	public static final int FORMAT_PPM = 9;

	/** Return value of {@link #getFormat()} for PSD streams. */
	public static final int FORMAT_PSD = 10;

/*	public static final int COLOR_TYPE_UNKNOWN = -1;
	public static final int COLOR_TYPE_TRUECOLOR_RGB = 0;
	public static final int COLOR_TYPE_PALETTED = 1;
	public static final int COLOR_TYPE_GRAYSCALE= 2;
	public static final int COLOR_TYPE_BLACK_AND_WHITE = 3;*/

	/**
	 * The names of all supported file formats.
	 * The FORMAT_xyz int constants can be used as index values for
	 * this array.
	 */
	private static final String[] FORMAT_NAMES =
		{"JPEG", "GIF", "PNG", "BMP", "PCX", 
		 "IFF", "RAS", "PBM", "PGM", "PPM", 
		 "PSD"};

	/**
	 * The names of the MIME types for all supported file formats.
	 * The FORMAT_xyz int constants can be used as index values for
	 * this array.
	 */
	private static final String[] MIME_TYPE_STRINGS =
		{"image/jpeg", "image/gif", "image/png", "image/bmp", "image/pcx", 
		 "image/iff", "image/ras", "image/x-portable-bitmap", "image/x-portable-graymap", "image/x-portable-pixmap", 
		 "image/psd"};

	private int width;
	private int height;
	private int bitsPerPixel;
	//private int colorType = COLOR_TYPE_UNKNOWN;
	private boolean progressive;
	private int format;
	private InputStream in;
	private DataInput din;
	private boolean collectComments = true;
	private Vector comments;
	private boolean determineNumberOfImages;
	private int numberOfImages;
	private int physicalHeightDpi;
	private int physicalWidthDpi;

	private void addComment(String s) {
		if (comments == null) {
			comments = new Vector();
		}
		comments.addElement(s);
	}

	/**
	 * Call this method after you have provided an input stream or file
	 * using {@link #setInput(InputStream)} or {@link #setInput(DataInput)}.
	 * If true is returned, the file format was known and information
	 * on the file's content can be retrieved using the various getXyz methods.
	 * @return if information could be retrieved from input
	 */
	public boolean check() {
		format = -1;
		width = -1;
		height = -1;
		bitsPerPixel = -1;
		numberOfImages = 1;
		physicalHeightDpi = -1;
		physicalWidthDpi = -1;
		comments = null;
		try {
			int b1 = read() & 0xff;
			int b2 = read() & 0xff;
			if (b1 == 0x47 && b2 == 0x49) {
				return checkGif();
			}
			else
			if (b1 == 0x89 && b2 == 0x50) {
				return checkPng();
			}
			else
			if (b1 == 0xff && b2 == 0xd8) {
				return checkJpeg();
			}
			else
			if (b1 == 0x42 && b2 == 0x4d) {
				return checkBmp();
			}
			else
			if (b1 == 0x0a && b2 < 0x06) {
				return checkPcx();
			}
			else
			if (b1 == 0x46 && b2 == 0x4f) {
				return checkIff();
			}
			else
			if (b1 == 0x59 && b2 == 0xa6) {
				return checkRas();
			}
			else
			if (b1 == 0x50 && b2 >= 0x31 && b2 <= 0x36) {
				return checkPnm(b2 - '0');
			}
			else
			if (b1 == 0x38 && b2 == 0x42) {
				return checkPsd();
			}
			else {
				return false;
			}
		} catch (IOException ioe) {
			return false;
		}
	}

	private boolean checkBmp() throws IOException {
		byte[] a = new byte[44];
		if (read(a) != a.length) {
			return false;
		}
		width = getIntLittleEndian(a, 16);
		height = getIntLittleEndian(a, 20);
		if (width < 1 || height < 1) {
			return false;
		}
		bitsPerPixel = getShortLittleEndian(a, 26);
		if (bitsPerPixel != 1 && bitsPerPixel != 4 &&
		    bitsPerPixel != 8 && bitsPerPixel != 16 &&
		    bitsPerPixel != 24 && bitsPerPixel != 32) {
		    return false;
		}
		int x = (int)(getIntLittleEndian(a, 36) * 0.0254);
		if (x > 0) {
			setPhysicalWidthDpi(x);
		}
		int y = (int)(getIntLittleEndian(a, 40) * 0.0254);
		if (y > 0) {
			setPhysicalHeightDpi(y);
		}
		format = FORMAT_BMP;
		return true;
	}

	private boolean checkGif() throws IOException {
		final byte[] GIF_MAGIC_87A = {0x46, 0x38, 0x37, 0x61};
		final byte[] GIF_MAGIC_89A = {0x46, 0x38, 0x39, 0x61};
		byte[] a = new byte[11]; // 4 from the GIF signature + 7 from the global header
		if (read(a) != 11) {
			return false;
		}
		if ((!equals(a, 0, GIF_MAGIC_89A, 0, 4)) &&
			(!equals(a, 0, GIF_MAGIC_87A, 0, 4))) {
			return false;
		}
		format = FORMAT_GIF;
		width = getShortLittleEndian(a, 4);
		height = getShortLittleEndian(a, 6);
		int flags = a[8] & 0xff;
		bitsPerPixel = ((flags >> 4) & 0x07) + 1;
		//progressive = (flags & 0x02) != 0;
		if (!determineNumberOfImages) {
			return true;
		}
		// skip global color palette
		if ((flags & 0x80) != 0) {
			int tableSize = (1 << ((flags & 7) + 1)) * 3;
			skip(tableSize);
		}
		numberOfImages = 0;
		int blockType;
		do
		{
			blockType = read();
			switch(blockType)
			{
				case(0x2c): // image separator
				{
					if (read(a, 0, 9) != 9) {
						return false;
					}
					flags = a[8] & 0xff;
					progressive = (flags & 0x40) != 0;
					/*int locWidth = getShortLittleEndian(a, 4);
					int locHeight = getShortLittleEndian(a, 6);
					System.out.println("LOCAL: " + locWidth + " x " + locHeight);*/
					int localBitsPerPixel = (flags & 0x07) + 1;
					if (localBitsPerPixel > bitsPerPixel) {
						bitsPerPixel = localBitsPerPixel;
					}
					if ((flags & 0x80) != 0) {
						skip((1 << localBitsPerPixel) * 3);
					}
					skip(1); // initial code length
					int n;
					do
					{
						n = read();
						if (n > 0) {