imgexam.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

/*
 * imgexam.c
 * Copyright (C) 2000-2004 A.J. van Os; Released under GNU GPL
 *
 * Description:
 * Functions to examine image headers
 *
 *================================================================
 * Part of this software is based on:
 * jpeg2ps - convert JPEG compressed images to PostScript Level 2
 * Copyright (C) 1994-99 Thomas Merz ([e-mail])
 *================================================================
 * The credit should go to him, but all the bugs are mine.
 */

#include <stdio.h>
#include <string.h>
#include <ctype.h>
#include "antiword.h"

/* BMP compression types */
#define BI_RGB		0
#define BI_RLE8		1
#define BI_RLE4		2

/* PNG colortype bits */
#define PNG_CB_PALETTE		0x01
#define PNG_CB_COLOR		0x02
#define PNG_CB_ALPHA		0x04

/* Instance signature */
#define MSOBI_WMF	0x0216
#define MSOBI_EMF	0x03d4
#define MSOBI_PICT	0x0542
#define MSOBI_PNG	0x06e0
#define MSOBI_JPEG	0x046a
#define MSOBI_DIB	0x07a8

/* The following enum is stolen from the IJG JPEG library */
typedef enum {		/* JPEG marker codes			*/
	M_SOF0	= 0xc0,	/* baseline DCT				*/
	M_SOF1	= 0xc1,	/* extended sequential DCT		*/
	M_SOF2	= 0xc2,	/* progressive DCT			*/
	M_SOF3	= 0xc3,	/* lossless (sequential)		*/

	M_SOF5	= 0xc5,	/* differential sequential DCT		*/
	M_SOF6	= 0xc6,	/* differential progressive DCT		*/
	M_SOF7	= 0xc7,	/* differential lossless		*/

	M_JPG	= 0xc8,	/* JPEG extensions			*/
	M_SOF9	= 0xc9,	/* extended sequential DCT		*/
	M_SOF10	= 0xca,	/* progressive DCT			*/
	M_SOF11	= 0xcb,	/* lossless (sequential)		*/

	M_SOF13	= 0xcd,	/* differential sequential DCT		*/
	M_SOF14	= 0xce,	/* differential progressive DCT		*/
	M_SOF15	= 0xcf,	/* differential lossless		*/

	M_DHT	= 0xc4,	/* define Huffman tables		*/

	M_DAC	= 0xcc,	/* define arithmetic conditioning table	*/

	M_RST0	= 0xd0,	/* restart				*/
	M_RST1	= 0xd1,	/* restart				*/
	M_RST2	= 0xd2,	/* restart				*/
	M_RST3	= 0xd3,	/* restart				*/
	M_RST4	= 0xd4,	/* restart				*/
	M_RST5	= 0xd5,	/* restart				*/
	M_RST6	= 0xd6,	/* restart				*/
	M_RST7	= 0xd7,	/* restart				*/

	M_SOI	= 0xd8,	/* start of image			*/
	M_EOI	= 0xd9,	/* end of image				*/
	M_SOS	= 0xda,	/* start of scan			*/
	M_DQT	= 0xdb,	/* define quantization tables		*/
	M_DNL	= 0xdc,	/* define number of lines		*/
	M_DRI	= 0xdd,	/* define restart interval		*/
	M_DHP	= 0xde,	/* define hierarchical progression	*/
	M_EXP	= 0xdf,	/* expand reference image(s)		*/

	M_APP0	= 0xe0,	/* application marker, used for JFIF	*/
	M_APP1	= 0xe1,	/* application marker			*/
	M_APP2	= 0xe2,	/* application marker			*/
	M_APP3	= 0xe3,	/* application marker			*/
	M_APP4	= 0xe4,	/* application marker			*/
	M_APP5	= 0xe5,	/* application marker			*/
	M_APP6	= 0xe6,	/* application marker			*/
	M_APP7	= 0xe7,	/* application marker			*/
	M_APP8	= 0xe8,	/* application marker			*/
	M_APP9	= 0xe9,	/* application marker			*/
	M_APP10	= 0xea,	/* application marker			*/
	M_APP11	= 0xeb,	/* application marker			*/
	M_APP12	= 0xec,	/* application marker			*/
	M_APP13	= 0xed,	/* application marker			*/
	M_APP14	= 0xee,	/* application marker, used by Adobe	*/
	M_APP15	= 0xef,	/* application marker			*/

	M_JPG0	= 0xf0,	/* reserved for JPEG extensions		*/
	M_JPG13	= 0xfd,	/* reserved for JPEG extensions		*/
	M_COM	= 0xfe,	/* comment				*/

	M_TEM	= 0x01	/* temporary use			*/
} JPEG_MARKER;


/*
 * bFillPaletteDIB - fill the palette part of the imagesdata
 *
 * returns TRUE if the images must be a color image, otherwise FALSE;
 */
static BOOL
bFillPaletteDIB(FILE *pFile, imagedata_type *pImg, BOOL bNewFormat)
{
	int	iIndex;
	BOOL	bIsColorPalette;

	fail(pFile == NULL);
	fail(pImg == NULL);

	if (pImg->uiBitsPerComponent > 8) {
		/* No palette, image uses more than 256 colors */
		return TRUE;
	}

	if (pImg->iColorsUsed <= 0) {
		/* Not specified, so compute the number of colors used */
		pImg->iColorsUsed = 1 << pImg->uiBitsPerComponent;
	}

	fail(pImg->iColorsUsed > 256);
	if (pImg->iColorsUsed > 256) {
		pImg->iColorsUsed = 256;
	}

	bIsColorPalette = FALSE;
	for (iIndex = 0; iIndex < pImg->iColorsUsed; iIndex++) {
		/* From BGR order to RGB order */
		pImg->aucPalette[iIndex][2] = (UCHAR)iNextByte(pFile);
		pImg->aucPalette[iIndex][1] = (UCHAR)iNextByte(pFile);
		pImg->aucPalette[iIndex][0] = (UCHAR)iNextByte(pFile);
		if (bNewFormat) {
			(void)iNextByte(pFile);
		}
		NO_DBG_PRINT_BLOCK(pImg->aucPalette[iIndex], 3);
		if (pImg->aucPalette[iIndex][0] !=
		     pImg->aucPalette[iIndex][1] ||
		    pImg->aucPalette[iIndex][1] !=
		     pImg->aucPalette[iIndex][2]) {
			bIsColorPalette = TRUE;
		}
	}

	return bIsColorPalette;
} /* end of bFillPaletteDIB */

/*
 * bExamineDIB - Examine a DIB header
 *
 * return TRUE if successful, otherwise FALSE
 */
static BOOL
bExamineDIB(FILE *pFile, imagedata_type *pImg)
{
	size_t	tHeaderSize;
	int	iPlanes, iCompression;

	tHeaderSize = (size_t)ulNextLong(pFile);
	switch (tHeaderSize) {
	case 12:
		pImg->iWidth = (int)usNextWord(pFile);
		pImg->iHeight = (int)usNextWord(pFile);
		iPlanes = (int)usNextWord(pFile);
		pImg->uiBitsPerComponent = (UINT)usNextWord(pFile);
		iCompression = BI_RGB;
		pImg->iColorsUsed = 0;
		break;
	case 40:
	case 64:
		pImg->iWidth = (int)ulNextLong(pFile);
		pImg->iHeight = (int)ulNextLong(pFile);
		iPlanes = (int)usNextWord(pFile);
		pImg->uiBitsPerComponent = (UINT)usNextWord(pFile);
		iCompression = (int)ulNextLong(pFile);
		(void)tSkipBytes(pFile, 12);
		pImg->iColorsUsed = (int)ulNextLong(pFile);
		(void)tSkipBytes(pFile, tHeaderSize - 36);
		break;
	default:
		DBG_DEC(tHeaderSize);
		return FALSE;
	}
	DBG_DEC(pImg->iWidth);
	DBG_DEC(pImg->iHeight);
	DBG_DEC(pImg->uiBitsPerComponent);
	DBG_DEC(iCompression);
	DBG_DEC(pImg->iColorsUsed);

	/* Do some sanity checks with the parameters */
	if (iPlanes != 1) {
		DBG_DEC(iPlanes);
		return FALSE;
	}
	if (pImg->iWidth <= 0 || pImg->iHeight <= 0) {
		DBG_DEC(pImg->iWidth);
		DBG_DEC(pImg->iHeight);
		return FALSE;
	}
	if (pImg->uiBitsPerComponent != 1 && pImg->uiBitsPerComponent != 4 &&
	    pImg->uiBitsPerComponent != 8 && pImg->uiBitsPerComponent != 24) {
		DBG_DEC(pImg->uiBitsPerComponent);
		return FALSE;
	}
	if (iCompression != BI_RGB &&
	    (pImg->uiBitsPerComponent == 1 || pImg->uiBitsPerComponent == 24)) {
		return FALSE;
	}
	if (iCompression == BI_RLE8 && pImg->uiBitsPerComponent == 4) {
		return FALSE;
	}
	if (iCompression == BI_RLE4 && pImg->uiBitsPerComponent == 8) {
		return FALSE;
	}

	switch (iCompression) {
	case BI_RGB:
		pImg->eCompression = compression_none;
		break;
	case BI_RLE4:
		pImg->eCompression = compression_rle4;
		break;
	case BI_RLE8:
		pImg->eCompression = compression_rle8;
		break;
	default:
		DBG_DEC(iCompression);
		return FALSE;
	}

	pImg->bColorImage = bFillPaletteDIB(pFile, pImg, tHeaderSize > 12);

	if (pImg->uiBitsPerComponent <= 8) {
		pImg->iComponents = 1;
	} else {
		pImg->iComponents = (int)(pImg->uiBitsPerComponent / 8);
	}

	return TRUE;
} /* end of bExamineDIB */

/*
 * iNextMarker - read the next JPEG marker
 */
static int
iNextMarker(FILE *pFile)
{
	int	iMarker;

	do {
		do {
			iMarker = iNextByte(pFile);
		} while (iMarker != 0xff && iMarker != EOF);
		if (iMarker == EOF) {
			return EOF;
		}
		do {
			iMarker = iNextByte(pFile);
		} while (iMarker == 0xff);
	} while (iMarker == 0x00);			/* repeat if ff/00 */

	return iMarker;
} /* end of iNextMarker */

/*
 * bExamineJPEG - Examine a JPEG header
 *
 * return TRUE if successful, otherwise FALSE
 */
static BOOL
bExamineJPEG(FILE *pFile, imagedata_type *pImg)
{
	size_t	tLength;
	int	iMarker, iIndex;
	char	appstring[10];
	BOOL	bSOFDone;

	tLength = 0;
	bSOFDone = FALSE;

	/* process JPEG markers */
	while (!bSOFDone && (iMarker = iNextMarker(pFile)) != (int)M_EOI) {
		switch (iMarker) {
		case EOF:
			DBG_MSG("Error: unexpected end of JPEG file");
			return FALSE;
	/* The following are not officially supported in PostScript level 2 */
		case M_SOF2:
		case M_SOF3:
		case M_SOF5:
		case M_SOF6:
		case M_SOF7:
		case M_SOF9:
		case M_SOF10:
		case M_SOF11:
		case M_SOF13:
		case M_SOF14:
		case M_SOF15:
			DBG_HEX(iMarker);
			return FALSE;
		case M_SOF0:
		case M_SOF1:
			tLength = (size_t)usNextWordBE(pFile);
			pImg->uiBitsPerComponent = (UINT)iNextByte(pFile);
			pImg->iHeight = (int)usNextWordBE(pFile);
			pImg->iWidth = (int)usNextWordBE(pFile);
			pImg->iComponents = iNextByte(pFile);
			bSOFDone = TRUE;
			break;
		case M_APP14:
		/*
		 * Check for Adobe application marker. It is known (per Adobe's
		 * TN5116) to contain the string "Adobe" at the start of the
		 * APP14 marker.
		 */
			tLength = (size_t)usNextWordBE(pFile);
			if (tLength < 12) {
				(void)tSkipBytes(pFile, tLength - 2);
			} else {
				for (iIndex = 0; iIndex < 5; iIndex++) {
					appstring[iIndex] =
							(char)iNextByte(pFile);
				}
				appstring[5] = '\0';
				if (STREQ(appstring, "Adobe")) {
					pImg->bAdobe = TRUE;
				}
				(void)tSkipBytes(pFile, tLength - 7);
			}
			break;
		case M_SOI:		/* ignore markers without parameters */
		case M_EOI:
		case M_TEM:
		case M_RST0:
		case M_RST1:
		case M_RST2:
		case M_RST3:
		case M_RST4:
		case M_RST5:
		case M_RST6:
		case M_RST7:
			break;
		default:		/* skip variable length markers */
			tLength = (size_t)usNextWordBE(pFile);
			(void)tSkipBytes(pFile, tLength - 2);
			break;
		}
	}

	DBG_DEC(pImg->iWidth);
	DBG_DEC(pImg->iHeight);
	DBG_DEC(pImg->uiBitsPerComponent);
	DBG_DEC(pImg->iComponents);

	/* Do some sanity checks with the parameters */
	if (pImg->iHeight <= 0 ||
	    pImg->iWidth <= 0 ||
	    pImg->iComponents <= 0) {
		DBG_DEC(pImg->iHeight);
		DBG_DEC(pImg->iWidth);
		DBG_DEC(pImg->iComponents);
		return FALSE;
	}

	/* Some broken JPEG files have this but they print anyway... */
	if (pImg->iComponents * 3 + 8 != (int)tLength) {
		DBG_MSG("Warning: SOF marker has incorrect length - ignored");
	}

	if (pImg->uiBitsPerComponent != 8) {
		DBG_DEC(pImg->uiBitsPerComponent);
		DBG_MSG("Not supported in PostScript level 2");
		return FALSE;
	}

	if (pImg->iComponents != 1 &&
	    pImg->iComponents != 3 &&
	    pImg->iComponents != 4) {
		DBG_DEC(pImg->iComponents);
		return FALSE;
	}

	pImg->bColorImage = pImg->iComponents >= 3;
	pImg->iColorsUsed = 0;
	pImg->eCompression = compression_jpeg;

	return TRUE;
} /* end of bExamineJPEG */

/*
 * bFillPalettePNG - fill the palette part of the imagesdata
 *
 * returns TRUE if sucessful, otherwise FALSE;
 */
static BOOL
bFillPalettePNG(FILE *pFile, imagedata_type *pImg, size_t tLength)
{
	int	iIndex, iEntries;

	fail(pFile == NULL);
	fail(pImg == NULL);

	if (pImg->uiBitsPerComponent > 8) {
		/* No palette, image uses more than 256 colors */
		return TRUE;
	}

	if (!pImg->bColorImage) {
		/* Only color images can have a palette */
		return FALSE;
	}

	if (tLength % 3 != 0) {
		/* Each palette entry takes three bytes */
		DBG_DEC(tLength);
		return FALSE;
	}

	iEntries = (int)(tLength / 3);
	DBG_DEC(iEntries);
	pImg->iColorsUsed = 1 << pImg->uiBitsPerComponent;
	DBG_DEC(pImg->iColorsUsed);

	if (iEntries > 256) {
		DBG_DEC(iEntries);
		return FALSE;
	}

	for (iIndex = 0; iIndex < iEntries; iIndex++) {
		pImg->aucPalette[iIndex][0] = (UCHAR)iNextByte(pFile);
		pImg->aucPalette[iIndex][1] = (UCHAR)iNextByte(pFile);
		pImg->aucPalette[iIndex][2] = (UCHAR)iNextByte(pFile);
		NO_DBG_PRINT_BLOCK(pImg->aucPalette[iIndex], 3);
	}
	for (;iIndex < pImg->iColorsUsed; iIndex++) {
		pImg->aucPalette[iIndex][0] = 0;
		pImg->aucPalette[iIndex][1] = 0;
		pImg->aucPalette[iIndex][2] = 0;
	}

	return TRUE;
} /* end of bFillPalettePNG */

/*
 * bExaminePNG - Examine a PNG header
 *
 * return TRUE if successful, otherwise FALSE
 */
static BOOL
bExaminePNG(FILE *pFile, imagedata_type *pImg)
{
	size_t		tLength;
	ULONG		ulLong1, ulLong2, ulName;
	int		iIndex, iTmp;
	int		iCompressionMethod, iFilterMethod, iInterlaceMethod;
	int		iColor, iIncrement;
	BOOL		bHasPalette, bHasAlpha;
	UCHAR	aucBuf[4];

	/* Check signature */
	ulLong1 = ulNextLongBE(pFile);
	ulLong2 = ulNextLongBE(pFile);
	if (ulLong1 != 0x89504e47UL || ulLong2 != 0x0d0a1a0aUL) {
		DBG_HEX(ulLong1);
		DBG_HEX(ulLong2);
		return FALSE;
	}

	ulName = 0x00;
	bHasPalette = FALSE;

	/* Examine chunks */
	while (ulName != PNG_CN_IEND) {
		tLength = (size_t)ulNextLongBE(pFile);
		ulName = 0x00;
		for (iIndex = 0; iIndex < (int)elementsof(aucBuf); iIndex++) {
			aucBuf[iIndex] = (UCHAR)iNextByte(pFile);
			if (!isalpha(aucBuf[iIndex])) {
				DBG_HEX(aucBuf[iIndex]);
				return FALSE;
			}
			ulName <<= 8;
			ulName |= aucBuf[iIndex];
		}

		switch (ulName) {
		case PNG_CN_IHDR:
			/* Header chunck */
			if (tLength < 13) {
				DBG_DEC(tLength);
				return FALSE;
			}
			pImg->iWidth = (int)ulNextLongBE(pFile);
			pImg->iHeight = (int)ulNextLongBE(pFile);
			pImg->uiBitsPerComponent = (UINT)iNextByte(pFile);
			iTmp = iNextByte(pFile);
			NO_DBG_HEX(iTmp);
			pImg->bColorImage = (iTmp & PNG_CB_COLOR) != 0;
			bHasPalette = (iTmp & PNG_CB_PALETTE) != 0;
			bHasAlpha = (iTmp & PNG_CB_ALPHA) != 0;
			if (bHasPalette && pImg->uiBitsPerComponent > 8) {
				/* This should not happen */
				return FALSE;
			}
			pImg->iComponents =
				(bHasPalette || !pImg->bColorImage) ? 1 : 3;
			if (bHasAlpha) {
				pImg->iComponents++;
			}
			iCompressionMethod = iNextByte(pFile);
			if (iCompressionMethod != 0) {
				DBG_DEC(iCompressionMethod);
				return FALSE;
			}
			iFilterMethod = iNextByte(pFile);