00001 /*
00002  * rdjpgcom.c
00003  *
00004  * Copyright (C) 1994-1997, Thomas G. Lane.
00005  * This file is part of the Independent JPEG Group's software.
00006  * For conditions of distribution and use, see the accompanying README file.
00007  *
00008  * This file contains a very simple stand-alone application that displays
00009  * the text in COM (comment) markers in a JFIF file.
00010  * This may be useful as an example of the minimum logic needed to parse
00011  * JPEG markers.
00012  */
00013 
00014 #define JPEG_CJPEG_DJPEG        /* to get the command-line config symbols */
00015 #include "jinclude.h"           /* get auto-config symbols, <stdio.h> */
00016 
00017 #include <ctype.h>              /* to declare isupper(), tolower() */
00018 #ifdef USE_SETMODE
00019 #include <fcntl.h>              /* to declare setmode()'s parameter macros */
00020 /* If you have setmode() but not <io.h>, just delete this line: */
00021 #include <io.h>                 /* to declare setmode() */
00022 #endif
00023 
00024 #ifdef USE_CCOMMAND             /* command-line reader for Macintosh */
00025 #ifdef __MWERKS__
00026 #include <SIOUX.h>              /* Metrowerks needs this */
00027 #include <console.h>            /* ... and this */
00028 #endif
00029 #ifdef THINK_C
00030 #include <console.h>            /* Think declares it here */
00031 #endif
00032 #endif
00033 
00034 #ifdef DONT_USE_B_MODE          /* define mode parameters for fopen() */
00035 #define READ_BINARY     "r"
00036 #else
00037 #ifdef VMS                      /* VMS is very nonstandard */
00038 #define READ_BINARY     "rb", "ctx=stm"
00039 #else                           /* standard ANSI-compliant case */
00040 #define READ_BINARY     "rb"
00041 #endif
00042 #endif
00043 
00044 #ifndef EXIT_FAILURE            /* define exit() codes if not provided */
00045 #define EXIT_FAILURE  1
00046 #endif
00047 #ifndef EXIT_SUCCESS
00048 #ifdef VMS
00049 #define EXIT_SUCCESS  1         /* VMS is very nonstandard */
00050 #else
00051 #define EXIT_SUCCESS  0
00052 #endif
00053 #endif
00054 
00055 
00056 /*
00057  * These macros are used to read the input file.
00058  * To reuse this code in another application, you might need to change these.
00059  */
00060 
00061 static FILE * infile;           /* input JPEG file */
00062 
00063 /* Return next input byte, or EOF if no more */
00064 #define NEXTBYTE()  getc(infile)
00065 
00066 
00067 /* Error exit handler */
00068 #define ERREXIT(msg)  (fprintf(stderr, "%s\n", msg), exit(EXIT_FAILURE))
00069 
00070 
00071 /* Read one byte, testing for EOF */
00072 static int
00073 read_1_byte (void)
00074 {
00075   int c;
00076 
00077   c = NEXTBYTE();
00078   if (c == EOF)
00079     ERREXIT("Premature EOF in JPEG file");
00080   return c;
00081 }
00082 
00083 /* Read 2 bytes, convert to unsigned int */
00084 /* All 2-byte quantities in JPEG markers are MSB first */
00085 static unsigned int
00086 read_2_bytes (void)
00087 {
00088   int c1, c2;
00089 
00090   c1 = NEXTBYTE();
00091   if (c1 == EOF)
00092     ERREXIT("Premature EOF in JPEG file");
00093   c2 = NEXTBYTE();
00094   if (c2 == EOF)
00095     ERREXIT("Premature EOF in JPEG file");
00096   return (((unsigned int) c1) << 8) + ((unsigned int) c2);
00097 }
00098 
00099 
00100 /*
00101  * JPEG markers consist of one or more 0xFF bytes, followed by a marker
00102  * code byte (which is not an FF).  Here are the marker codes of interest
00103  * in this program.  (See jdmarker.c for a more complete list.)
00104  */
00105 
00106 #define M_SOF0  0xC0            /* Start Of Frame N */
00107 #define M_SOF1  0xC1            /* N indicates which compression process */
00108 #define M_SOF2  0xC2            /* Only SOF0-SOF2 are now in common use */
00109 #define M_SOF3  0xC3
00110 #define M_SOF5  0xC5            /* NB: codes C4 and CC are NOT SOF markers */
00111 #define M_SOF6  0xC6
00112 #define M_SOF7  0xC7
00113 #define M_SOF9  0xC9
00114 #define M_SOF10 0xCA
00115 #define M_SOF11 0xCB
00116 #define M_SOF13 0xCD
00117 #define M_SOF14 0xCE
00118 #define M_SOF15 0xCF
00119 #define M_SOI   0xD8            /* Start Of Image (beginning of datastream) */
00120 #define M_EOI   0xD9            /* End Of Image (end of datastream) */
00121 #define M_SOS   0xDA            /* Start Of Scan (begins compressed data) */
00122 #define M_APP0  0xE0            /* Application-specific marker, type N */
00123 #define M_APP12 0xEC            /* (we don't bother to list all 16 APPn's) */
00124 #define M_COM   0xFE            /* COMment */
00125 
00126 
00127 /*
00128  * Find the next JPEG marker and return its marker code.
00129  * We expect at least one FF byte, possibly more if the compressor used FFs
00130  * to pad the file.
00131  * There could also be non-FF garbage between markers.  The treatment of such
00132  * garbage is unspecified; we choose to skip over it but emit a warning msg.
00133  * NB: this routine must not be used after seeing SOS marker, since it will
00134  * not deal correctly with FF/00 sequences in the compressed image data...
00135  */
00136 
00137 static int
00138 next_marker (void)
00139 {
00140   int c;
00141   int discarded_bytes = 0;
00142 
00143   /* Find 0xFF byte; count and skip any non-FFs. */
00144   c = read_1_byte();
00145   while (c != 0xFF) {
00146     discarded_bytes++;
00147     c = read_1_byte();
00148   }
00149   /* Get marker code byte, swallowing any duplicate FF bytes.  Extra FFs
00150    * are legal as pad bytes, so don't count them in discarded_bytes.
00151    */
00152   do {
00153     c = read_1_byte();
00154   } while (c == 0xFF);
00155 
00156   if (discarded_bytes != 0) {
00157     fprintf(stderr, "Warning: garbage data found in JPEG file\n");
00158   }
00159 
00160   return c;
00161 }
00162 
00163 
00164 /*
00165  * Read the initial marker, which should be SOI.
00166  * For a JFIF file, the first two bytes of the file should be literally
00167  * 0xFF M_SOI.  To be more general, we could use next_marker, but if the
00168  * input file weren't actually JPEG at all, next_marker might read the whole
00169  * file and then return a misleading error message...
00170  */
00171 
00172 static int
00173 first_marker (void)
00174 {
00175   int c1, c2;
00176 
00177   c1 = NEXTBYTE();
00178   c2 = NEXTBYTE();
00179   if (c1 != 0xFF || c2 != M_SOI)
00180     ERREXIT("Not a JPEG file");
00181   return c2;
00182 }
00183 
00184 
00185 /*
00186  * Most types of marker are followed by a variable-length parameter segment.
00187  * This routine skips over the parameters for any marker we don't otherwise
00188  * want to process.
00189  * Note that we MUST skip the parameter segment explicitly in order not to
00190  * be fooled by 0xFF bytes that might appear within the parameter segment;
00191  * such bytes do NOT introduce new markers.
00192  */
00193 
00194 static void
00195 skip_variable (void)
00196 /* Skip over an unknown or uninteresting variable-length marker */
00197 {
00198   unsigned int length;
00199 
00200   /* Get the marker parameter length count */
00201   length = read_2_bytes();
00202   /* Length includes itself, so must be at least 2 */
00203   if (length < 2)
00204     ERREXIT("Erroneous JPEG marker length");
00205   length -= 2;
00206   /* Skip over the remaining bytes */
00207   while (length > 0) {
00208     (void) read_1_byte();
00209     length--;
00210   }
00211 }
00212 
00213 
00214 /*
00215  * Process a COM marker.
00216  * We want to print out the marker contents as legible text;
00217  * we must guard against non-text junk and varying newline representations.
00218  */
00219 
00220 static void
00221 process_COM (void)
00222 {
00223   unsigned int length;
00224   int ch;
00225   int lastch = 0;
00226 
00227   /* Get the marker parameter length count */
00228   length = read_2_bytes();
00229   /* Length includes itself, so must be at least 2 */
00230   if (length < 2)
00231     ERREXIT("Erroneous JPEG marker length");
00232   length -= 2;
00233 
00234   while (length > 0) {
00235     ch = read_1_byte();
00236     /* Emit the character in a readable form.
00237      * Nonprintables are converted to \nnn form,
00238      * while \ is converted to \\.
00239      * Newlines in CR, CR/LF, or LF form will be printed as one newline.
00240      */
00241     if (ch == '\r') {
00242       printf("\n");
00243     } else if (ch == '\n') {
00244       if (lastch != '\r')
00245         printf("\n");
00246     } else if (ch == '\\') {
00247       printf("\\\\");
00248     } else if (isprint(ch)) {
00249       putc(ch, stdout);
00250     } else {
00251       printf("\\%03o", ch);
00252     }
00253     lastch = ch;
00254     length--;
00255   }
00256   printf("\n");
00257 }
00258 
00259 
00260 /*
00261  * Process a SOFn marker.
00262  * This code is only needed if you want to know the image dimensions...
00263  */
00264 
00265 static void
00266 process_SOFn (int marker)
00267 {
00268   unsigned int length;
00269   unsigned int image_height, image_width;
00270   int data_precision, num_components;
00271   const char * process;
00272   int ci;
00273 
00274   length = read_2_bytes();      /* usual parameter length count */
00275 
00276   data_precision = read_1_byte();
00277   image_height = read_2_bytes();
00278   image_width = read_2_bytes();
00279   num_components = read_1_byte();
00280 
00281   switch (marker) {
00282   case M_SOF0:  process = "Baseline";  break;
00283   case M_SOF1:  process = "Extended sequential";  break;
00284   case M_SOF2:  process = "Progressive";  break;
00285   case M_SOF3:  process = "Lossless";  break;
00286   case M_SOF5:  process = "Differential sequential";  break;
00287   case M_SOF6:  process = "Differential progressive";  break;
00288   case M_SOF7:  process = "Differential lossless";  break;
00289   case M_SOF9:  process = "Extended sequential, arithmetic coding";  break;
00290   case M_SOF10: process = "Progressive, arithmetic coding";  break;
00291   case M_SOF11: process = "Lossless, arithmetic coding";  break;
00292   case M_SOF13: process = "Differential sequential, arithmetic coding";  break;
00293   case M_SOF14: process = "Differential progressive, arithmetic coding"; break;
00294   case M_SOF15: process = "Differential lossless, arithmetic coding";  break;
00295   default:      process = "Unknown";  break;
00296   }
00297 
00298   printf("JPEG image is %uw * %uh, %d color components, %d bits per sample\n",
00299          image_width, image_height, num_components, data_precision);
00300   printf("JPEG process: %s\n", process);
00301 
00302   if (length != (unsigned int) (8 + num_components * 3))
00303     ERREXIT("Bogus SOF marker length");
00304 
00305   for (ci = 0; ci < num_components; ci++) {
00306     (void) read_1_byte();       /* Component ID code */
00307     (void) read_1_byte();       /* H, V sampling factors */
00308     (void) read_1_byte();       /* Quantization table number */
00309   }
00310 }
00311 
00312 
00313 /*
00314  * Parse the marker stream until SOS or EOI is seen;
00315  * display any COM markers.
00316  * While the companion program wrjpgcom will always insert COM markers before
00317  * SOFn, other implementations might not, so we scan to SOS before stopping.
00318  * If we were only interested in the image dimensions, we would stop at SOFn.
00319  * (Conversely, if we only cared about COM markers, there would be no need
00320  * for special code to handle SOFn; we could treat it like other markers.)
00321  */
00322 
00323 static int
00324 scan_JPEG_header (int verbose)
00325 {
00326   int marker;
00327 
00328   /* Expect SOI at start of file */
00329   if (first_marker() != M_SOI)
00330     ERREXIT("Expected SOI marker first");
00331 
00332   /* Scan miscellaneous markers until we reach SOS. */
00333   for (;;) {
00334     marker = next_marker();
00335     switch (marker) {
00336       /* Note that marker codes 0xC4, 0xC8, 0xCC are not, and must not be,
00337        * treated as SOFn.  C4 in particular is actually DHT.
00338        */
00339     case M_SOF0:                /* Baseline */
00340     case M_SOF1:                /* Extended sequential, Huffman */
00341     case M_SOF2:                /* Progressive, Huffman */
00342     case M_SOF3:                /* Lossless, Huffman */
00343     case M_SOF5:                /* Differential sequential, Huffman */
00344     case M_SOF6:                /* Differential progressive, Huffman */
00345     case M_SOF7:                /* Differential lossless, Huffman */
00346     case M_SOF9:                /* Extended sequential, arithmetic */
00347     case M_SOF10:               /* Progressive, arithmetic */
00348     case M_SOF11:               /* Lossless, arithmetic */
00349     case M_SOF13:               /* Differential sequential, arithmetic */
00350     case M_SOF14:               /* Differential progressive, arithmetic */
00351     case M_SOF15:               /* Differential lossless, arithmetic */
00352       if (verbose)
00353         process_SOFn(marker);
00354       else
00355         skip_variable();
00356       break;
00357 
00358     case M_SOS:                 /* stop before hitting compressed data */
00359       return marker;
00360 
00361     case M_EOI:                 /* in case it's a tables-only JPEG stream */
00362       return marker;
00363 
00364     case M_COM:
00365       process_COM();
00366       break;
00367 
00368     case M_APP12:
00369       /* Some digital camera makers put useful textual information into
00370        * APP12 markers, so we print those out too when in -verbose mode.
00371        */
00372       if (verbose) {
00373         printf("APP12 contains:\n");
00374         process_COM();
00375       } else
00376         skip_variable();
00377       break;
00378 
00379     default:                    /* Anything else just gets skipped */
00380       skip_variable();          /* we assume it has a parameter count... */
00381       break;
00382     }
00383   } /* end loop */
00384 }
00385 
00386 
00387 /* Command line parsing code */
00388 
00389 static const char * progname;   /* program name for error messages */
00390 
00391 
00392 static void
00393 usage (void)
00394 /* complain about bad command line */
00395 {
00396   fprintf(stderr, "rdjpgcom displays any textual comments in a JPEG file.\n");
00397 
00398   fprintf(stderr, "Usage: %s [switches] [inputfile]\n", progname);
00399 
00400   fprintf(stderr, "Switches (names may be abbreviated):\n");
00401   fprintf(stderr, "  -verbose    Also display dimensions of JPEG image\n");
00402 
00403   exit(EXIT_FAILURE);
00404 }
00405 
00406 
00407 static int
00408 keymatch (char * arg, const char * keyword, int minchars)
00409 /* Case-insensitive matching of (possibly abbreviated) keyword switches. */
00410 /* keyword is the constant keyword (must be lower case already), */
00411 /* minchars is length of minimum legal abbreviation. */
00412 {
00413   register int ca, ck;
00414   register int nmatched = 0;
00415 
00416   while ((ca = *arg++) != '\0') {
00417     if ((ck = *keyword++) == '\0')
00418       return 0;                 /* arg longer than keyword, no good */
00419     if (isupper(ca))            /* force arg to lcase (assume ck is already) */
00420       ca = tolower(ca);
00421     if (ca != ck)
00422       return 0;                 /* no good */
00423     nmatched++;                 /* count matched characters */
00424   }
00425   /* reached end of argument; fail if it's too short for unique abbrev */
00426   if (nmatched < minchars)
00427     return 0;
00428   return 1;                     /* A-OK */
00429 }
00430 
00431 
00432 /*
00433  * The main program.
00434  */
00435 
00436 int
00437 main (int argc, char **argv)
00438 {
00439   int argn;
00440   char * arg;
00441   int verbose = 0;
00442 
00443   /* On Mac, fetch a command line. */
00444 #ifdef USE_CCOMMAND
00445   argc = ccommand(&argv);
00446 #endif
00447 
00448   progname = argv[0];
00449   if (progname == NULL || progname[0] == 0)
00450     progname = "rdjpgcom";      /* in case C library doesn't provide it */
00451 
00452   /* Parse switches, if any */
00453   for (argn = 1; argn < argc; argn++) {
00454     arg = argv[argn];
00455     if (arg[0] != '-')
00456       break;                    /* not switch, must be file name */
00457     arg++;                      /* advance over '-' */
00458     if (keymatch(arg, "verbose", 1)) {
00459       verbose++;
00460     } else
00461       usage();
00462   }
00463 
00464   /* Open the input file. */
00465   /* Unix style: expect zero or one file name */
00466   if (argn < argc-1) {
00467     fprintf(stderr, "%s: only one input file\n", progname);
00468     usage();
00469   }
00470   if (argn < argc) {
00471     if ((infile = fopen(argv[argn], READ_BINARY)) == NULL) {
00472       fprintf(stderr, "%s: can't open %s\n", progname, argv[argn]);
00473       exit(EXIT_FAILURE);
00474     }
00475   } else {
00476     /* default input file is stdin */
00477 #ifdef USE_SETMODE              /* need to hack file mode? */
00478     setmode(fileno(stdin), O_BINARY);
00479 #endif
00480 #ifdef USE_FDOPEN               /* need to re-open in binary mode? */
00481     if ((infile = fdopen(fileno(stdin), READ_BINARY)) == NULL) {
00482       fprintf(stderr, "%s: can't open stdin\n", progname);
00483       exit(EXIT_FAILURE);
00484     }
00485 #else
00486     infile = stdin;
00487 #endif
00488   }
00489 
00490   /* Scan the JPEG headers. */
00491   (void) scan_JPEG_header(verbose);
00492 
00493   /* All done. */
00494   exit(EXIT_SUCCESS);
00495   return 0;                     /* suppress no-return-value warnings */
00496 }