1/*
2 * wrjpgcom.c
3 *
4 * This file was part of the Independent JPEG Group's software:
5 * Copyright (C) 1994-1997, Thomas G. Lane.
6 * libjpeg-turbo Modifications:
7 * Copyright (C) 2014, D. R. Commander
8 * For conditions of distribution and use, see the accompanying README file.
9 *
10 * This file contains a very simple stand-alone application that inserts
11 * user-supplied text as a COM (comment) marker in a JFIF file.
12 * This may be useful as an example of the minimum logic needed to parse
13 * JPEG markers.
14 */
15
16#define JPEG_CJPEG_DJPEG        /* to get the command-line config symbols */
17#include "jinclude.h"           /* get auto-config symbols, <stdio.h> */
18
19#ifndef HAVE_STDLIB_H           /* <stdlib.h> should declare malloc() */
20extern void * malloc ();
21#endif
22#include <ctype.h>              /* to declare isupper(), tolower() */
23#ifdef USE_SETMODE
24#include <fcntl.h>              /* to declare setmode()'s parameter macros */
25/* If you have setmode() but not <io.h>, just delete this line: */
26#include <io.h>                 /* to declare setmode() */
27#endif
28
29#ifdef USE_CCOMMAND             /* command-line reader for Macintosh */
30#ifdef __MWERKS__
31#include <SIOUX.h>              /* Metrowerks needs this */
32#include <console.h>            /* ... and this */
33#endif
34#ifdef THINK_C
35#include <console.h>            /* Think declares it here */
36#endif
37#endif
38
39#ifdef DONT_USE_B_MODE          /* define mode parameters for fopen() */
40#define READ_BINARY     "r"
41#define WRITE_BINARY    "w"
42#else
43#define READ_BINARY     "rb"
44#define WRITE_BINARY    "wb"
45#endif
46
47#ifndef EXIT_FAILURE            /* define exit() codes if not provided */
48#define EXIT_FAILURE  1
49#endif
50#ifndef EXIT_SUCCESS
51#define EXIT_SUCCESS  0
52#endif
53
54/* Reduce this value if your malloc() can't allocate blocks up to 64K.
55 * On DOS, compiling in large model is usually a better solution.
56 */
57
58#ifndef MAX_COM_LENGTH
59#define MAX_COM_LENGTH 65000L   /* must be <= 65533 in any case */
60#endif
61
62
63/*
64 * These macros are used to read the input file and write the output file.
65 * To reuse this code in another application, you might need to change these.
66 */
67
68static FILE * infile;           /* input JPEG file */
69
70/* Return next input byte, or EOF if no more */
71#define NEXTBYTE()  getc(infile)
72
73static FILE * outfile;          /* output JPEG file */
74
75/* Emit an output byte */
76#define PUTBYTE(x)  putc((x), outfile)
77
78
79/* Error exit handler */
80#define ERREXIT(msg)  (fprintf(stderr, "%s\n", msg), exit(EXIT_FAILURE))
81
82
83/* Read one byte, testing for EOF */
84static int
85read_1_byte (void)
86{
87  int c;
88
89  c = NEXTBYTE();
90  if (c == EOF)
91    ERREXIT("Premature EOF in JPEG file");
92  return c;
93}
94
95/* Read 2 bytes, convert to unsigned int */
96/* All 2-byte quantities in JPEG markers are MSB first */
97static unsigned int
98read_2_bytes (void)
99{
100  int c1, c2;
101
102  c1 = NEXTBYTE();
103  if (c1 == EOF)
104    ERREXIT("Premature EOF in JPEG file");
105  c2 = NEXTBYTE();
106  if (c2 == EOF)
107    ERREXIT("Premature EOF in JPEG file");
108  return (((unsigned int) c1) << 8) + ((unsigned int) c2);
109}
110
111
112/* Routines to write data to output file */
113
114static void
115write_1_byte (int c)
116{
117  PUTBYTE(c);
118}
119
120static void
121write_2_bytes (unsigned int val)
122{
123  PUTBYTE((val >> 8) & 0xFF);
124  PUTBYTE(val & 0xFF);
125}
126
127static void
128write_marker (int marker)
129{
130  PUTBYTE(0xFF);
131  PUTBYTE(marker);
132}
133
134static void
135copy_rest_of_file (void)
136{
137  int c;
138
139  while ((c = NEXTBYTE()) != EOF)
140    PUTBYTE(c);
141}
142
143
144/*
145 * JPEG markers consist of one or more 0xFF bytes, followed by a marker
146 * code byte (which is not an FF).  Here are the marker codes of interest
147 * in this program.  (See jdmarker.c for a more complete list.)
148 */
149
150#define M_SOF0  0xC0            /* Start Of Frame N */
151#define M_SOF1  0xC1            /* N indicates which compression process */
152#define M_SOF2  0xC2            /* Only SOF0-SOF2 are now in common use */
153#define M_SOF3  0xC3
154#define M_SOF5  0xC5            /* NB: codes C4 and CC are NOT SOF markers */
155#define M_SOF6  0xC6
156#define M_SOF7  0xC7
157#define M_SOF9  0xC9
158#define M_SOF10 0xCA
159#define M_SOF11 0xCB
160#define M_SOF13 0xCD
161#define M_SOF14 0xCE
162#define M_SOF15 0xCF
163#define M_SOI   0xD8            /* Start Of Image (beginning of datastream) */
164#define M_EOI   0xD9            /* End Of Image (end of datastream) */
165#define M_SOS   0xDA            /* Start Of Scan (begins compressed data) */
166#define M_COM   0xFE            /* COMment */
167
168
169/*
170 * Find the next JPEG marker and return its marker code.
171 * We expect at least one FF byte, possibly more if the compressor used FFs
172 * to pad the file.  (Padding FFs will NOT be replicated in the output file.)
173 * There could also be non-FF garbage between markers.  The treatment of such
174 * garbage is unspecified; we choose to skip over it but emit a warning msg.
175 * NB: this routine must not be used after seeing SOS marker, since it will
176 * not deal correctly with FF/00 sequences in the compressed image data...
177 */
178
179static int
180next_marker (void)
181{
182  int c;
183  int discarded_bytes = 0;
184
185  /* Find 0xFF byte; count and skip any non-FFs. */
186  c = read_1_byte();
187  while (c != 0xFF) {
188    discarded_bytes++;
189    c = read_1_byte();
190  }
191  /* Get marker code byte, swallowing any duplicate FF bytes.  Extra FFs
192   * are legal as pad bytes, so don't count them in discarded_bytes.
193   */
194  do {
195    c = read_1_byte();
196  } while (c == 0xFF);
197
198  if (discarded_bytes != 0) {
199    fprintf(stderr, "Warning: garbage data found in JPEG file\n");
200  }
201
202  return c;
203}
204
205
206/*
207 * Read the initial marker, which should be SOI.
208 * For a JFIF file, the first two bytes of the file should be literally
209 * 0xFF M_SOI.  To be more general, we could use next_marker, but if the
210 * input file weren't actually JPEG at all, next_marker might read the whole
211 * file and then return a misleading error message...
212 */
213
214static int
215first_marker (void)
216{
217  int c1, c2;
218
219  c1 = NEXTBYTE();
220  c2 = NEXTBYTE();
221  if (c1 != 0xFF || c2 != M_SOI)
222    ERREXIT("Not a JPEG file");
223  return c2;
224}
225
226
227/*
228 * Most types of marker are followed by a variable-length parameter segment.
229 * This routine skips over the parameters for any marker we don't otherwise
230 * want to process.
231 * Note that we MUST skip the parameter segment explicitly in order not to
232 * be fooled by 0xFF bytes that might appear within the parameter segment;
233 * such bytes do NOT introduce new markers.
234 */
235
236static void
237copy_variable (void)
238/* Copy an unknown or uninteresting variable-length marker */
239{
240  unsigned int length;
241
242  /* Get the marker parameter length count */
243  length = read_2_bytes();
244  write_2_bytes(length);
245  /* Length includes itself, so must be at least 2 */
246  if (length < 2)
247    ERREXIT("Erroneous JPEG marker length");
248  length -= 2;
249  /* Skip over the remaining bytes */
250  while (length > 0) {
251    write_1_byte(read_1_byte());
252    length--;
253  }
254}
255
256static void
257skip_variable (void)
258/* Skip over an unknown or uninteresting variable-length marker */
259{
260  unsigned int length;
261
262  /* Get the marker parameter length count */
263  length = read_2_bytes();
264  /* Length includes itself, so must be at least 2 */
265  if (length < 2)
266    ERREXIT("Erroneous JPEG marker length");
267  length -= 2;
268  /* Skip over the remaining bytes */
269  while (length > 0) {
270    (void) read_1_byte();
271    length--;
272  }
273}
274
275
276/*
277 * Parse the marker stream until SOFn or EOI is seen;
278 * copy data to output, but discard COM markers unless keep_COM is true.
279 */
280
281static int
282scan_JPEG_header (int keep_COM)
283{
284  int marker;
285
286  /* Expect SOI at start of file */
287  if (first_marker() != M_SOI)
288    ERREXIT("Expected SOI marker first");
289  write_marker(M_SOI);
290
291  /* Scan miscellaneous markers until we reach SOFn. */
292  for (;;) {
293    marker = next_marker();
294    switch (marker) {
295      /* Note that marker codes 0xC4, 0xC8, 0xCC are not, and must not be,
296       * treated as SOFn.  C4 in particular is actually DHT.
297       */
298    case M_SOF0:                /* Baseline */
299    case M_SOF1:                /* Extended sequential, Huffman */
300    case M_SOF2:                /* Progressive, Huffman */
301    case M_SOF3:                /* Lossless, Huffman */
302    case M_SOF5:                /* Differential sequential, Huffman */
303    case M_SOF6:                /* Differential progressive, Huffman */
304    case M_SOF7:                /* Differential lossless, Huffman */
305    case M_SOF9:                /* Extended sequential, arithmetic */
306    case M_SOF10:               /* Progressive, arithmetic */
307    case M_SOF11:               /* Lossless, arithmetic */
308    case M_SOF13:               /* Differential sequential, arithmetic */
309    case M_SOF14:               /* Differential progressive, arithmetic */
310    case M_SOF15:               /* Differential lossless, arithmetic */
311      return marker;
312
313    case M_SOS:                 /* should not see compressed data before SOF */
314      ERREXIT("SOS without prior SOFn");
315      break;
316
317    case M_EOI:                 /* in case it's a tables-only JPEG stream */
318      return marker;
319
320    case M_COM:                 /* Existing COM: conditionally discard */
321      if (keep_COM) {
322        write_marker(marker);
323        copy_variable();
324      } else {
325        skip_variable();
326      }
327      break;
328
329    default:                    /* Anything else just gets copied */
330      write_marker(marker);
331      copy_variable();          /* we assume it has a parameter count... */
332      break;
333    }
334  } /* end loop */
335}
336
337
338/* Command line parsing code */
339
340static const char * progname;   /* program name for error messages */
341
342
343static void
344usage (void)
345/* complain about bad command line */
346{
347  fprintf(stderr, "wrjpgcom inserts a textual comment in a JPEG file.\n");
348  fprintf(stderr, "You can add to or replace any existing comment(s).\n");
349
350  fprintf(stderr, "Usage: %s [switches] ", progname);
351#ifdef TWO_FILE_COMMANDLINE
352  fprintf(stderr, "inputfile outputfile\n");
353#else
354  fprintf(stderr, "[inputfile]\n");
355#endif
356
357  fprintf(stderr, "Switches (names may be abbreviated):\n");
358  fprintf(stderr, "  -replace         Delete any existing comments\n");
359  fprintf(stderr, "  -comment \"text\"  Insert comment with given text\n");
360  fprintf(stderr, "  -cfile name      Read comment from named file\n");
361  fprintf(stderr, "Notice that you must put quotes around the comment text\n");
362  fprintf(stderr, "when you use -comment.\n");
363  fprintf(stderr, "If you do not give either -comment or -cfile on the command line,\n");
364  fprintf(stderr, "then the comment text is read from standard input.\n");
365  fprintf(stderr, "It can be multiple lines, up to %u characters total.\n",
366          (unsigned int) MAX_COM_LENGTH);
367#ifndef TWO_FILE_COMMANDLINE
368  fprintf(stderr, "You must specify an input JPEG file name when supplying\n");
369  fprintf(stderr, "comment text from standard input.\n");
370#endif
371
372  exit(EXIT_FAILURE);
373}
374
375
376static int
377keymatch (char * arg, const char * keyword, int minchars)
378/* Case-insensitive matching of (possibly abbreviated) keyword switches. */
379/* keyword is the constant keyword (must be lower case already), */
380/* minchars is length of minimum legal abbreviation. */
381{
382  register int ca, ck;
383  register int nmatched = 0;
384
385  while ((ca = *arg++) != '\0') {
386    if ((ck = *keyword++) == '\0')
387      return 0;                 /* arg longer than keyword, no good */
388    if (isupper(ca))            /* force arg to lcase (assume ck is already) */
389      ca = tolower(ca);
390    if (ca != ck)
391      return 0;                 /* no good */
392    nmatched++;                 /* count matched characters */
393  }
394  /* reached end of argument; fail if it's too short for unique abbrev */
395  if (nmatched < minchars)
396    return 0;
397  return 1;                     /* A-OK */
398}
399
400
401/*
402 * The main program.
403 */
404
405int
406main (int argc, char **argv)
407{
408  int argn;
409  char * arg;
410  int keep_COM = 1;
411  char * comment_arg = NULL;
412  FILE * comment_file = NULL;
413  unsigned int comment_length = 0;
414  int marker;
415
416  /* On Mac, fetch a command line. */
417#ifdef USE_CCOMMAND
418  argc = ccommand(&argv);
419#endif
420
421  progname = argv[0];
422  if (progname == NULL || progname[0] == 0)
423    progname = "wrjpgcom";      /* in case C library doesn't provide it */
424
425  /* Parse switches, if any */
426  for (argn = 1; argn < argc; argn++) {
427    arg = argv[argn];
428    if (arg[0] != '-')
429      break;                    /* not switch, must be file name */
430    arg++;                      /* advance over '-' */
431    if (keymatch(arg, "replace", 1)) {
432      keep_COM = 0;
433    } else if (keymatch(arg, "cfile", 2)) {
434      if (++argn >= argc) usage();
435      if ((comment_file = fopen(argv[argn], "r")) == NULL) {
436        fprintf(stderr, "%s: can't open %s\n", progname, argv[argn]);
437        exit(EXIT_FAILURE);
438      }
439    } else if (keymatch(arg, "comment", 1)) {
440      if (++argn >= argc) usage();
441      comment_arg = argv[argn];
442      /* If the comment text starts with '"', then we are probably running
443       * under MS-DOG and must parse out the quoted string ourselves.  Sigh.
444       */
445      if (comment_arg[0] == '"') {
446        comment_arg = (char *) malloc((size_t) MAX_COM_LENGTH);
447        if (comment_arg == NULL)
448          ERREXIT("Insufficient memory");
449        if (strlen(argv[argn]) + 2 >= (size_t) MAX_COM_LENGTH) {
450          fprintf(stderr, "Comment text may not exceed %u bytes\n",
451                  (unsigned int) MAX_COM_LENGTH);
452          exit(EXIT_FAILURE);
453        }
454        strcpy(comment_arg, argv[argn]+1);
455        for (;;) {
456          comment_length = (unsigned int) strlen(comment_arg);
457          if (comment_length > 0 && comment_arg[comment_length-1] == '"') {
458            comment_arg[comment_length-1] = '\0'; /* zap terminating quote */
459            break;
460          }
461          if (++argn >= argc)
462            ERREXIT("Missing ending quote mark");
463          if (strlen(comment_arg) + strlen(argv[argn]) + 2 >=
464              (size_t) MAX_COM_LENGTH) {
465            fprintf(stderr, "Comment text may not exceed %u bytes\n",
466                    (unsigned int) MAX_COM_LENGTH);
467            exit(EXIT_FAILURE);
468          }
469          strcat(comment_arg, " ");
470          strcat(comment_arg, argv[argn]);
471        }
472      } else if (strlen(argv[argn]) >= (size_t) MAX_COM_LENGTH) {
473        fprintf(stderr, "Comment text may not exceed %u bytes\n",
474                (unsigned int) MAX_COM_LENGTH);
475        exit(EXIT_FAILURE);
476      }
477      comment_length = (unsigned int) strlen(comment_arg);
478    } else
479      usage();
480  }
481
482  /* Cannot use both -comment and -cfile. */
483  if (comment_arg != NULL && comment_file != NULL)
484    usage();
485  /* If there is neither -comment nor -cfile, we will read the comment text
486   * from stdin; in this case there MUST be an input JPEG file name.
487   */
488  if (comment_arg == NULL && comment_file == NULL && argn >= argc)
489    usage();
490
491  /* Open the input file. */
492  if (argn < argc) {
493    if ((infile = fopen(argv[argn], READ_BINARY)) == NULL) {
494      fprintf(stderr, "%s: can't open %s\n", progname, argv[argn]);
495      exit(EXIT_FAILURE);
496    }
497  } else {
498    /* default input file is stdin */
499#ifdef USE_SETMODE              /* need to hack file mode? */
500    setmode(fileno(stdin), O_BINARY);
501#endif
502#ifdef USE_FDOPEN               /* need to re-open in binary mode? */
503    if ((infile = fdopen(fileno(stdin), READ_BINARY)) == NULL) {
504      fprintf(stderr, "%s: can't open stdin\n", progname);
505      exit(EXIT_FAILURE);
506    }
507#else
508    infile = stdin;
509#endif
510  }
511
512  /* Open the output file. */
513#ifdef TWO_FILE_COMMANDLINE
514  /* Must have explicit output file name */
515  if (argn != argc-2) {
516    fprintf(stderr, "%s: must name one input and one output file\n",
517            progname);
518    usage();
519  }
520  if ((outfile = fopen(argv[argn+1], WRITE_BINARY)) == NULL) {
521    fprintf(stderr, "%s: can't open %s\n", progname, argv[argn+1]);
522    exit(EXIT_FAILURE);
523  }
524#else
525  /* Unix style: expect zero or one file name */
526  if (argn < argc-1) {
527    fprintf(stderr, "%s: only one input file\n", progname);
528    usage();
529  }
530  /* default output file is stdout */
531#ifdef USE_SETMODE              /* need to hack file mode? */
532  setmode(fileno(stdout), O_BINARY);
533#endif
534#ifdef USE_FDOPEN               /* need to re-open in binary mode? */
535  if ((outfile = fdopen(fileno(stdout), WRITE_BINARY)) == NULL) {
536    fprintf(stderr, "%s: can't open stdout\n", progname);
537    exit(EXIT_FAILURE);
538  }
539#else
540  outfile = stdout;
541#endif
542#endif /* TWO_FILE_COMMANDLINE */
543
544  /* Collect comment text from comment_file or stdin, if necessary */
545  if (comment_arg == NULL) {
546    FILE * src_file;
547    int c;
548
549    comment_arg = (char *) malloc((size_t) MAX_COM_LENGTH);
550    if (comment_arg == NULL)
551      ERREXIT("Insufficient memory");
552    comment_length = 0;
553    src_file = (comment_file != NULL ? comment_file : stdin);
554    while ((c = getc(src_file)) != EOF) {
555      if (comment_length >= (unsigned int) MAX_COM_LENGTH) {
556        fprintf(stderr, "Comment text may not exceed %u bytes\n",
557                (unsigned int) MAX_COM_LENGTH);
558        exit(EXIT_FAILURE);
559      }
560      comment_arg[comment_length++] = (char) c;
561    }
562    if (comment_file != NULL)
563      fclose(comment_file);
564  }
565
566  /* Copy JPEG headers until SOFn marker;
567   * we will insert the new comment marker just before SOFn.
568   * This (a) causes the new comment to appear after, rather than before,
569   * existing comments; and (b) ensures that comments come after any JFIF
570   * or JFXX markers, as required by the JFIF specification.
571   */
572  marker = scan_JPEG_header(keep_COM);
573  /* Insert the new COM marker, but only if nonempty text has been supplied */
574  if (comment_length > 0) {
575    write_marker(M_COM);
576    write_2_bytes(comment_length + 2);
577    while (comment_length > 0) {
578      write_1_byte(*comment_arg++);
579      comment_length--;
580    }
581  }
582  /* Duplicate the remainder of the source file.
583   * Note that any COM markers occuring after SOF will not be touched.
584   */
585  write_marker(marker);
586  copy_rest_of_file();
587
588  /* All done. */
589  exit(EXIT_SUCCESS);
590  return 0;                     /* suppress no-return-value warnings */
591}
592