source: nikanabo/current/xdelta/diy/xdelta3-main.h@ 541

Last change on this file since 541 was 185, checked in by geyser, 18 years ago
File size: 87.3 KB
RevLine 
[185]1/* xdelta 3 - delta compression tools and library
2 * Copyright (C) 2001, 2002, 2003, 2004, 2005, 2006, 2007,
3 * Joshua P. MacDonald
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18 */
19
20/* This is all the extra stuff you need for convenience to users in a command line
21 * application. It contains these major components:
22 *
23 * 1. VCDIFF tools
24 * 2. external compression support (this is POSIX-specific).
25 * 3. a general read/write loop that handles all of the Xdelta decode/encode/VCDIFF-print
26 * functions
27 * 4. command-line interpreter
28 * 5. an Xdelta application header which stores default filename, external compression settings
29 * 6. output/error printing
30 * 7. basic file support and OS interface
31 */
32
33/* TODO list:
34 * 1. do exact gzip-like filename, stdout handling. make a .vcdiff extension, refuse
35 * to encode to stdout without -cf, etc.
36 * 2. Allow the user to add a comment string to the app header without disturbing the default
37 * behavior.
38 * 3. "Source file must be seekable" is not actually true for encoding, given current
39 * behavior. Allow non-seekable sources? It would in theory let you use a fifo for
40 * the source.
41 */
42
43/* On error handling and printing:
44 *
45 * The xdelta library sets stream->msg to indicate what condition caused an internal
46 * failure, but many failures originate here and are printed here. The return convention
47 * is 0 for success, as throughout Xdelta code, but special attention is required here for
48 * the operating system calls with different error handling. See the main_file_* routines.
49 * All errors in this file have a message printed at the time of occurance. Since some of
50 * these calls occur within calls to the library, the error may end up being printed again
51 * with a more general error message.
52 */
53
54/******************************************************************************************/
55
56#ifndef XD3_POSIX
57#define XD3_POSIX 0
58#endif
59#ifndef XD3_STDIO
60#define XD3_STDIO 0
61#endif
62#ifndef XD3_WIN32
63#define XD3_WIN32 0
64#endif
65
66/* Combines xd3_strerror() and strerror() */
67const char* xd3_mainerror(int err_num);
68
69/* XPRINTX (used by main) prefixes an "xdelta3: " to the output. */
70#define XPR fprintf
71#define NT stderr, "xdelta3: "
72
73/* If none are set, default to posix. */
74#if (XD3_POSIX + XD3_STDIO + XD3_WIN32) == 0
75#undef XD3_POSIX
76#define XD3_POSIX 1
77#endif
78
79/* Handle externally-compressed inputs. */
80#ifndef EXTERNAL_COMPRESSION
81#define EXTERNAL_COMPRESSION 1
82#endif
83
84#define PRINTHDR_SPECIAL -4378291
85
86/* The number of soft-config variables. */
87#define XD3_SOFTCFG_VARCNT 7
88
89/* this is used as in XPR(NT XD3_LIB_ERRMSG (stream, ret)) to print an error message
90 * from the library. */
91#define XD3_LIB_ERRMSG(stream, ret) "%s: %s\n", xd3_errstring (stream), xd3_mainerror (ret)
92
93#include <stdio.h> /* fprintf */
94
95#if XD3_POSIX
96#include <unistd.h> /* close, read, write... */
97#include <sys/types.h>
98#include <fcntl.h>
99#endif
100
101#ifndef _WIN32
102#include <unistd.h> /* lots */
103#include <sys/time.h> /* gettimeofday() */
104#include <sys/stat.h> /* stat() and fstat() */
105#else
106#define strtoll _strtoi64
107#include <sys/types.h>
108#include <sys/stat.h>
109#ifndef WIFEXITED
110# define WIFEXITED(stat) (((*((int *) &(stat))) & 0xff) == 0)
111#endif
112#ifndef WEXITSTATUS
113# define WEXITSTATUS(stat) (((*((int *) &(stat))) >> 8) & 0xff)
114#endif
115#ifndef S_ISREG
116//# ifdef S_IFREG
117//# define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
118//# else
119# define S_ISREG(m) 1
120//# endif
121#endif /* !S_ISREG */
122
123// For standard input/output handles
124static STARTUPINFO winStartupInfo;
125#endif
126
127/******************************************************************************************
128 ENUMS and TYPES
129 ******************************************************************************************/
130
131/* These flags (mainly pertaining to main_read() operations) are set in the
132 * main_file->flags variable. All are related to with external decompression support.
133 *
134 * RD_FIRST causes the external decompression check when the input is first read.
135 *
136 * RD_NONEXTERNAL disables external decompression for reading a compressed input, in the
137 * case of Xdelta inputs. Note: Xdelta is supported as an external compression type,
138 * which makes is the reason for this flag. An example to justify this is: to create a
139 * delta between two files that are VCDIFF-compressed. Two external Xdelta decoders are
140 * run to supply decompressed source and target inputs to the Xdelta encoder. */
141typedef enum
142{
143 RD_FIRST = (1 << 0),
144 RD_NONEXTERNAL = (1 << 1),
145 RD_EXTERNAL_V1 = (1 << 2),
146} xd3_read_flags;
147
148/* main_file->mode values */
149typedef enum
150{
151 XO_READ = 0,
152 XO_WRITE = 1,
153} main_file_modes;
154
155/* Main commands. For example, CMD_PRINTHDR is the "xdelta printhdr" command. */
156typedef enum
157{
158 CMD_NONE = 0,
159 CMD_PRINTHDR,
160 CMD_PRINTHDRS,
161 CMD_PRINTDELTA,
162#if XD3_ENCODER
163 CMD_ENCODE,
164#endif
165 CMD_DECODE,
166 CMD_TEST,
167 CMD_CONFIG,
168} xd3_cmd;
169
170#if XD3_ENCODER
171#define CMD_DEFAULT CMD_ENCODE
172#define IS_ENCODE(cmd) (cmd == CMD_ENCODE)
173#else
174#define CMD_DEFAULT CMD_DECODE
175#define IS_ENCODE(cmd) (0)
176#endif
177
178typedef struct _main_file main_file;
179typedef struct _main_extcomp main_extcomp;
180typedef struct _main_blklru main_blklru;
181typedef struct _main_blklru_list main_blklru_list;
182
183/* The main_file object supports abstract system calls like open, close, read, write, seek,
184 * stat. The program uses these to represent both seekable files and non-seekable files.
185 * Source files must be seekable, but the target input and any output file do not require
186 * seekability.
187 */
188struct _main_file
189{
190#if XD3_STDIO
191 FILE *file;
192#elif XD3_POSIX
193 int file;
194#elif XD3_WIN32
195 HANDLE file;
196#endif
197
198 int mode; /* XO_READ and XO_WRITE */
199 const char *filename; /* File name or /dev/stdin, /dev/stdout, /dev/stderr. */
200 char *filename_copy; /* File name or /dev/stdin, /dev/stdout, /dev/stderr. */
201 const char *realname; /* File name or /dev/stdin, /dev/stdout, /dev/stderr. */
202 const main_extcomp *compressor; /* External compression struct. */
203 int flags; /* RD_FIRST, RD_NONEXTERNAL, ... */
204 xoff_t nread; /* for input position */
205 xoff_t nwrite; /* for output position */
206 uint8_t *snprintf_buf; /* internal snprintf() use */
207};
208
209/* Various strings and magic values used to detect and call external compression. See
210 * below for examples. */
211struct _main_extcomp
212{
213 const char *recomp_cmdname;
214 const char *recomp_options;
215
216 const char *decomp_cmdname;
217 const char *decomp_options;
218
219 const char *ident;
220 const char *magic;
221 int magic_size;
222 int flags;
223};
224
225/* This file implements a small LRU of source blocks. For encoding purposes,
226 * we prevent paging in blocks we've already scanned in the source (return
227 * XD3_NOTAVAIL). */
228struct _main_blklru_list
229{
230 main_blklru_list *next;
231 main_blklru_list *prev;
232};
233
234struct _main_blklru
235{
236 uint8_t *blk;
237 xoff_t blkno;
238 main_blklru_list link;
239};
240
241#define LRU_SIZE 32U
242#define XD3_MINSRCWINSZ XD3_ALLOCSIZE
243
244/* ... represented as a list (no cache index). */
245XD3_MAKELIST(main_blklru_list,main_blklru,link);
246
247// TODO:
248// struct _main_state
249// {
250
251/* Program options: various command line flags and options. */
252static int option_stdout = 0;
253static int option_force = 0;
254static int option_verbose = 0;
255static int option_quiet = 0;
256static int option_use_appheader = 1;
257static uint8_t* option_appheader = NULL;
258static int option_use_secondary = 0;
259static char* option_secondary = NULL;
260static int option_use_checksum = 1;
261static int option_use_altcodetable = 0;
262static char* option_smatch_config = NULL;
263static int option_no_compress = 0;
264static int option_no_output = 0; /* do not open or write output */
265static const char *option_source_filename = NULL;
266
267static int option_level = XD3_DEFAULT_LEVEL;
268static usize_t option_iopt_size = XD3_DEFAULT_IOPT_SIZE;
269static usize_t option_winsize = XD3_DEFAULT_WINSIZE;
270static usize_t option_srcwinsz = XD3_DEFAULT_SRCWINSZ;
271static usize_t option_sprevsz = XD3_DEFAULT_SPREVSZ;
272
273/* These variables are supressed to avoid their use w/o support. main() warns
274 * appropriately. */
275#if EXTERNAL_COMPRESSION
276static int option_decompress_inputs = 1;
277static int option_recompress_outputs = 1;
278#endif
279
280/* This is for comparing "printdelta" output without attention to
281 * copy-instruction modes. */
282#if VCDIFF_TOOLS
283static int option_print_cpymode = 1;
284#endif
285
286/* Static variables */
287IF_DEBUG(static int main_mallocs = 0;)
288
289static char* program_name = NULL;
290static uint8_t* appheader_used = NULL;
291static uint8_t* main_bdata = NULL;
292
293/* The LRU: obviously this is shared by all callers. */
294static int lru_size = 0;
295static main_blklru *lru = NULL; /* array of lru_size elts */
296static main_blklru_list lru_list;
297static main_blklru_list lru_free;
298static int do_not_lru = 0; /* set to avoid lru, instead discard oldest */
299
300static int lru_hits = 0;
301static int lru_misses = 0;
302static int lru_filled = 0;
303
304/* Hacks for VCDIFF tools */
305static int allow_fake_source = 0;
306
307/* This array of compressor types is compiled even if EXTERNAL_COMPRESSION is false just so
308 * the program knows the mapping of IDENT->NAME. */
309static main_extcomp extcomp_types[] =
310{
311 /* The entry for xdelta3 must be 0 because the program_name is set there. */
312 { "xdelta3", "-cfq", "xdelta3", "-dcfq", "X", "\xd6\xc3\xc4", 3, RD_NONEXTERNAL },
313 { "bzip2", "-cf", "bzip2", "-dcf", "B", "BZh", 3, 0 },
314 { "gzip", "-cf", "gzip", "-dcf", "G", "\037\213", 2, 0 },
315 { "compress", "-cf", "uncompress", "-cf", "Z", "\037\235", 2, 0 },
316
317 /* TODO: add commandline support for magic-less formats */
318 /*{ "lzma", "-cf", "lzma", "-dcf", "M", "]\000", 2, 0 },*/
319};
320
321// };
322
323static void main_get_appheader (xd3_stream *stream, main_file *ifile,
324 main_file *output, main_file *sfile);
325
326static int main_help (void);
327
328static int
329main_version (void)
330{
331 /* $Format: " DP(RINT \"VERSION=3.$Xdelta3Version$\\n\");" $ */
332 DP(RINT "VERSION=3.0q\n");
333 return EXIT_SUCCESS;
334}
335
336static int
337main_config (void)
338{
339 main_version ();
340
341 DP(RINT "EXTERNAL_COMPRESSION=%d\n", EXTERNAL_COMPRESSION);
342 DP(RINT "GENERIC_ENCODE_TABLES=%d\n", GENERIC_ENCODE_TABLES);
343 DP(RINT "GENERIC_ENCODE_TABLES_COMPUTE=%d\n", GENERIC_ENCODE_TABLES_COMPUTE);
344 DP(RINT "REGRESSION_TEST=%d\n", REGRESSION_TEST);
345 DP(RINT "SECONDARY_DJW=%d\n", SECONDARY_DJW);
346 DP(RINT "SECONDARY_FGK=%d\n", SECONDARY_FGK);
347 DP(RINT "VCDIFF_TOOLS=%d\n", VCDIFF_TOOLS);
348 DP(RINT "XD3_ALLOCSIZE=%d\n", XD3_ALLOCSIZE);
349 DP(RINT "XD3_DEBUG=%d\n", XD3_DEBUG);
350 DP(RINT "XD3_ENCODER=%d\n", XD3_ENCODER);
351 DP(RINT "XD3_POSIX=%d\n", XD3_POSIX);
352 DP(RINT "XD3_STDIO=%d\n", XD3_STDIO);
353 DP(RINT "XD3_WIN32=%d\n", XD3_WIN32);
354 DP(RINT "XD3_USE_LARGEFILE64=%d\n", XD3_USE_LARGEFILE64);
355 DP(RINT "XD3_DEFAULT_LEVEL=%d\n", XD3_DEFAULT_LEVEL);
356 DP(RINT "XD3_DEFAULT_IOPT_SIZE=%d\n", XD3_DEFAULT_IOPT_SIZE);
357 DP(RINT "XD3_DEFAULT_SPREVSZ=%d\n", XD3_DEFAULT_SPREVSZ);
358 DP(RINT "XD3_DEFAULT_SRCWINSZ=%d\n", XD3_DEFAULT_SRCWINSZ);
359 DP(RINT "XD3_DEFAULT_WINSIZE=%d\n", XD3_DEFAULT_WINSIZE);
360 DP(RINT "XD3_HARDMAXWINSIZE=%d\n", XD3_HARDMAXWINSIZE);
361 DP(RINT "XD3_NODECOMPRESSSIZE=%d\n", XD3_NODECOMPRESSSIZE);
362
363 return EXIT_SUCCESS;
364}
365
366static void
367reset_defaults(void)
368{
369 option_stdout = 0;
370 option_force = 0;
371 option_verbose = 0;
372 option_quiet = 0;
373 option_appheader = NULL;
374 option_use_secondary = 0;
375 option_secondary = NULL;
376 option_use_altcodetable = 0;
377 option_smatch_config = NULL;
378 option_no_compress = 0;
379 option_no_output = 0;
380 option_source_filename = NULL;
381 program_name = NULL;
382 appheader_used = NULL;
383 main_bdata = NULL;
384 lru_size = 0;
385 lru = NULL;
386 do_not_lru = 0;
387 lru_hits = 0;
388 lru_misses = 0;
389 lru_filled = 0;
390 allow_fake_source = 0;
391 option_smatch_config = NULL;
392
393 option_use_appheader = 1;
394 option_use_checksum = 1;
395#if EXTERNAL_COMPRESSION
396 option_decompress_inputs = 1;
397 option_recompress_outputs = 1;
398#endif
399#if VCDIFF_TOOLS
400 option_print_cpymode = 1;
401#endif
402 option_level = XD3_DEFAULT_LEVEL;
403 option_iopt_size = XD3_DEFAULT_IOPT_SIZE;
404 option_winsize = XD3_DEFAULT_WINSIZE;
405 option_srcwinsz = XD3_DEFAULT_SRCWINSZ;
406 option_sprevsz = XD3_DEFAULT_SPREVSZ;
407}
408
409static void*
410main_malloc1 (usize_t size)
411{
412 void* r = malloc (size);
413 if (r == NULL) { XPR(NT "malloc: %s\n", xd3_mainerror (ENOMEM)); }
414 else if (option_verbose > 3) { XPR(NT "malloc: %u: %p\n", size, r); }
415 return r;
416}
417
418static void*
419main_malloc (usize_t size)
420{
421 void *r = main_malloc1 (size);
422 if (r) { IF_DEBUG (main_mallocs += 1); }
423 return r;
424}
425
426static void*
427main_alloc (void *opaque,
428 usize_t items,
429 usize_t size)
430{
431 return main_malloc1 (items * size);
432}
433
434static void
435main_free1 (void *opaque, void *ptr)
436{
437 if (option_verbose > 3) { XPR(NT "free: %p\n", ptr); }
438 free (ptr);
439}
440
441static void
442main_free (void *ptr)
443{
444 if (ptr)
445 {
446 IF_DEBUG (main_mallocs -= 1);
447 main_free1 (NULL, ptr);
448 IF_DEBUG (XD3_ASSERT(main_mallocs >= 0));
449 }
450}
451
452/* This ensures that (ret = errno) always indicates failure, in case errno was
453 * accidentally not set. If this prints there's a bug somewhere. */
454static int
455get_errno (void)
456{
457#ifndef _WIN32
458 if (errno == 0)
459 {
460 XPR(NT "you found a bug: expected errno != 0\n");
461 errno = XD3_INTERNAL;
462 }
463 return errno;
464#else
465 DWORD errNum = GetLastError();
466 if (errNum == NO_ERROR) {
467 errNum = XD3_INTERNAL;
468 }
469 return errNum;
470#endif
471}
472
473const char*
474xd3_mainerror(int err_num) {
475#ifndef _WIN32
476 const char* x = xd3_strerror (err_num);
477 if (x != NULL) {
478 return x;
479 }
480 return strerror(err_num);
481#else
482 static char err_buf[256];
483 const char* x = xd3_strerror (err_num);
484 if (x != NULL) {
485 return x;
486 }
487 memset (err_buf, 0, 256);
488 FormatMessage (FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
489 NULL, err_num,
490 MAKELANGID (LANG_NEUTRAL, SUBLANG_DEFAULT),
491 err_buf, 256, NULL);
492 return err_buf;
493#endif
494}
495
496static long
497get_millisecs_now (void)
498{
499#ifndef _WIN32
500 struct timeval tv;
501
502 gettimeofday (& tv, NULL);
503
504 return (tv.tv_sec) * 1000L + (tv.tv_usec) / 1000;
505#else
506 // Found this in an example on www.codeproject.com
507 // It doesn't matter that the offset is Jan 1, 1601
508 // Result is the numbre of 100 nanosecond units
509 // 100ns * 10,000 = 1ms
510 SYSTEMTIME st;
511 FILETIME ft;
512 __int64 *pi = (__int64*)&ft;
513 GetLocalTime(&st);
514 SystemTimeToFileTime(&st, &ft);
515 return (long)((*pi) / 10000);
516#endif
517}
518
519/* Always >= 1 millisec, right? */
520static long
521get_millisecs_since (void)
522{
523 static long last = 0;
524 long now = get_millisecs_now();
525 long diff = now - last;
526 last = now;
527 return diff;
528}
529
530static char*
531main_format_bcnt (xoff_t r, char *buf)
532{
533 static const char* fmts[] = { "B", "KB", "MB", "GB" };
534 int i;
535
536 for (i = 0; i < SIZEOF_ARRAY(fmts); i += 1)
537 {
538 if (r <= (10 * 1024) || i == (-1 + (int)SIZEOF_ARRAY(fmts)))
539 {
540 sprintf (buf, "%"Q"u %s", r, fmts[i]);
541 break;
542 }
543 r /= 1024;
544 }
545 return buf;
546}
547
548static char*
549main_format_rate (xoff_t bytes, long millis, char *buf)
550{
551 xoff_t r = (xoff_t)(1.0 * bytes / (1.0 * millis / 1000.0));
552 static char lbuf[32];
553
554 main_format_bcnt (r, lbuf);
555 sprintf (buf, "%s/sec", lbuf);
556 return buf;
557}
558
559static char*
560main_format_millis (long millis, char *buf)
561{
562 if (millis < 1000) { sprintf (buf, "%lu ms", millis); }
563 else if (millis < 10000) { sprintf (buf, "%.1f sec", millis / 1000.0); }
564 else { sprintf (buf, "%lu sec", millis / 1000L); }
565 return buf;
566}
567
568/* A safe version of strtol for xoff_t. */
569static int
570main_strtoxoff (const char* s, xoff_t *xo, char which)
571{
572 char *e;
573 xoff_t x;
574
575 XD3_ASSERT(s && *s != 0);
576
577 {
578 /* Should check LONG_MIN, LONG_MAX, LLONG_MIN, LLONG_MAX? */
579#if SIZEOF_XOFF_T == 4
580 long xx = strtol (s, &e, 0);
581#else
582 long long xx = strtoll (s, &e, 0);
583#endif
584
585 if (xx < 0)
586 {
587 XPR(NT "-%c: negative integer: %s\n", which, s);
588 return EXIT_FAILURE;
589 }
590
591 x = xx;
592 }
593
594 if (*e != 0)
595 {
596 XPR(NT "-%c: invalid integer: %s\n", which, s);
597 return EXIT_FAILURE;
598 }
599
600 (*xo) = x;
601 return 0;
602}
603
604static int
605main_atou (const char* arg, usize_t *xo, usize_t low, usize_t high, char which)
606{
607 xoff_t x;
608 int ret;
609
610 if ((ret = main_strtoxoff (arg, & x, which))) { return ret; }
611
612 if (x < low)
613 {
614 XPR(NT "-%c: minimum value: %u\n", which, low);
615 return EXIT_FAILURE;
616 }
617 if (high == 0)
618 {
619 high = USIZE_T_MAX;
620 }
621 if (x > high)
622 {
623 XPR(NT "-%c: maximum value: %u\n", which, high);
624 return EXIT_FAILURE;
625 }
626 (*xo) = (usize_t)x;
627 return 0;
628}
629
630/******************************************************************************************
631 FILE BASICS
632 ******************************************************************************************/
633
634/* With all the variation in file system-call semantics, arguments, return values and
635 * error-handling for the POSIX and STDIO file APIs, the insides of these functions make
636 * me sick, which is why these wrappers exist. */
637
638#define XOPEN_OPNAME (xfile->mode == XO_READ ? "read" : "write")
639#define XOPEN_STDIO (xfile->mode == XO_READ ? "rb" : "wb")
640#define XOPEN_POSIX (xfile->mode == XO_READ ? O_RDONLY : O_WRONLY | O_CREAT | O_TRUNC)
641#define XOPEN_MODE (xfile->mode == XO_READ ? 0 : 0666)
642
643#define XF_ERROR(op, name, ret) \
644 XPR(NT "file %s failed: %s: %s: %s\n", (op), XOPEN_OPNAME, (name), xd3_mainerror (ret))
645
646#if XD3_STDIO
647#define XFNO(f) fileno(f->file)
648#define XSTDOUT_XF(f) { (f)->file = stdout; (f)->filename = "/dev/stdout"; }
649#define XSTDIN_XF(f) { (f)->file = stdin; (f)->filename = "/dev/stdin"; }
650
651#elif XD3_POSIX
652#define XFNO(f) f->file
653#define XSTDOUT_XF(f) { (f)->file = STDOUT_FILENO; (f)->filename = "/dev/stdout"; }
654#define XSTDIN_XF(f) { (f)->file = STDIN_FILENO; (f)->filename = "/dev/stdin"; }
655
656#elif XD3_WIN32
657#define XFNO(f) -1
658#define XSTDOUT_XF(f) { (f)->file = winStartupInfo.hStdOutput; (f)->filename = "(stdout)"; }
659#define XSTDIN_XF(f) { (f)->file = winStartupInfo.hStdInput; (f)->filename = "(stdin)"; }
660#endif
661
662static void
663main_file_init (main_file *xfile)
664{
665 memset (xfile, 0, sizeof (*xfile));
666
667#if XD3_POSIX
668 xfile->file = -1;
669#endif
670#if XD3_WIN32
671 xfile->file = INVALID_HANDLE_VALUE;
672#endif
673}
674
675static int
676main_file_isopen (main_file *xfile)
677{
678#if XD3_STDIO
679 return xfile->file != NULL;
680
681#elif XD3_POSIX
682 return xfile->file != -1;
683
684#elif XD3_WIN32
685 return xfile->file != INVALID_HANDLE_VALUE;
686#endif
687}
688
689static int
690main_file_close (main_file *xfile)
691{
692 int ret = 0;
693
694 if (! main_file_isopen (xfile))
695 {
696 return 0;
697 }
698
699#if XD3_STDIO
700 ret = fclose (xfile->file);
701 xfile->file = NULL;
702
703#elif XD3_POSIX
704 ret = close (xfile->file);
705 xfile->file = -1;
706
707#elif XD3_WIN32
708 if (!CloseHandle(xfile->file)) {
709 ret = get_errno ();
710 }
711 xfile->file = INVALID_HANDLE_VALUE;
712#endif
713
714 if (ret != 0) { XF_ERROR ("close", xfile->filename, ret = get_errno ()); }
715 return ret;
716}
717
718static void
719main_file_cleanup (main_file *xfile)
720{
721 if (main_file_isopen (xfile))
722 {
723 main_file_close (xfile);
724 }
725
726 if (xfile->snprintf_buf != NULL)
727 {
728 main_free(xfile->snprintf_buf);
729 xfile->snprintf_buf = NULL;
730 }
731
732 if (xfile->filename_copy != NULL)
733 {
734 main_free(xfile->filename_copy);
735 xfile->filename_copy = NULL;
736 }
737}
738
739static int
740main_file_open (main_file *xfile, const char* name, int mode)
741{
742 int ret = 0;
743
744 xfile->mode = mode;
745
746 XD3_ASSERT (name != NULL);
747 XD3_ASSERT (! main_file_isopen (xfile));
748 if (name[0] == 0)
749 {
750 XPR(NT "invalid file name: empty string\n");
751 return XD3_INVALID;
752 }
753
754#if XD3_STDIO
755 xfile->file = fopen (name, XOPEN_STDIO);
756
757 ret = (xfile->file == NULL) ? get_errno () : 0;
758
759#elif XD3_POSIX
760 if ((ret = open (name, XOPEN_POSIX, XOPEN_MODE)) < 0)
761 {
762 ret = get_errno ();
763 }
764 else
765 {
766 xfile->file = ret;
767 ret = 0;
768 }
769
770#elif XD3_WIN32
771 xfile->file = CreateFile(name,
772 (mode == XO_READ) ? GENERIC_READ : GENERIC_WRITE,
773 FILE_SHARE_READ,
774 NULL,
775 (mode == XO_READ) ? OPEN_EXISTING : (option_force ? CREATE_ALWAYS : CREATE_NEW),
776 FILE_ATTRIBUTE_NORMAL,
777 NULL);
778 if (xfile->file == INVALID_HANDLE_VALUE) {
779 ret = get_errno ();
780 }
781#endif
782 if (ret) { XF_ERROR ("open", name, ret); }
783 else { xfile->realname = name; xfile->nread = 0; }
784 return ret;
785}
786
787static int
788main_file_stat (main_file *xfile, xoff_t *size, int err_ifnoseek)
789{
790 int ret = 0;
791#if XD3_WIN32
792 LARGE_INTEGER li;
793 if (GetFileSizeEx(xfile->file, &li) == 0) {
794 ret = get_errno ();
795 } else {
796 *size = li.QuadPart;
797 }
798#else
799 struct stat sbuf;
800 if (fstat (XFNO (xfile), & sbuf) < 0)
801 {
802 ret = get_errno ();
803 if (err_ifnoseek) { XF_ERROR ("stat", xfile->filename, ret); }
804 return ret;
805 }
806
807 if (! S_ISREG (sbuf.st_mode))
808 {
809 if (err_ifnoseek) { XPR(NT "source file must be seekable: %s\n", xfile->filename); }
810 return ESPIPE;
811 }
812 (*size) = sbuf.st_size;
813#endif
814 return ret;
815}
816
817static int
818main_file_exists (main_file *xfile)
819{
820 struct stat sbuf;
821 return stat (xfile->filename, & sbuf) == 0 && S_ISREG (sbuf.st_mode);
822}
823
824#if (XD3_POSIX || EXTERNAL_COMPRESSION)
825/* POSIX-generic code takes a function pointer to read() or write(). This calls the
826 * function repeatedly until the buffer is full or EOF. The NREAD parameter is not
827 * set for write, NULL is passed. Return is signed, < 0 indicate errors, otherwise
828 * byte count. */
829typedef int (xd3_posix_func) (int fd, uint8_t *buf, usize_t size);
830
831static int
832xd3_posix_io (int fd, uint8_t *buf, usize_t size, xd3_posix_func *func, usize_t *nread)
833{
834 int ret;
835 usize_t nproc = 0;
836
837 while (nproc < size)
838 {
839 int result = (*func) (fd, buf + nproc, size - nproc);
840
841 if (result < 0)
842 {
843 ret = get_errno ();
844 if (ret != EAGAIN && ret != EINTR)
845 {
846 return ret;
847 }
848 result = 0;
849 }
850
851 if (nread != NULL && result == 0) { break; }
852
853 nproc += result;
854 }
855 if (nread != NULL) { (*nread) = nproc; }
856 return 0;
857}
858#endif
859
860/* POSIX is unbuffered, while STDIO is buffered. main_file_read() should always be called
861 * on blocks. */
862static int
863main_file_read (main_file *ifile,
864 uint8_t *buf,
865 usize_t size,
866 usize_t *nread,
867 const char *msg)
868{
869 int ret = 0;
870
871#if XD3_STDIO
872 usize_t result;
873
874 result = fread (buf, 1, size, ifile->file);
875
876 if (result < size && ferror (ifile->file))
877 {
878 ret = get_errno ();
879 }
880 else
881 {
882 *nread = result;
883 }
884
885#elif XD3_POSIX
886 ret = xd3_posix_io (ifile->file, buf, size, (xd3_posix_func*) &read, nread);
887
888#elif XD3_WIN32
889 DWORD nread2;
890 if (ReadFile (ifile->file, buf, size, &nread2, NULL) == 0) {
891 ret = get_errno();
892 } else {
893 *nread = (usize_t)nread2;
894 }
895#endif
896
897 if (ret)
898 {
899 XPR(NT "%s: %s: %s\n", msg, ifile->filename, xd3_mainerror (ret));
900 }
901 else
902 {
903 if (option_verbose > 3) { XPR(NT "main read: %s: %u\n", ifile->filename, (*nread)); }
904 ifile->nread += (*nread);
905 }
906
907 return ret;
908}
909
910static int
911main_file_write (main_file *ofile, uint8_t *buf, usize_t size, const char *msg)
912{
913 int ret = 0;
914
915#if XD3_STDIO
916 usize_t result;
917
918 result = fwrite (buf, 1, size, ofile->file);
919
920 if (result != size) { ret = get_errno (); }
921
922#elif XD3_POSIX
923 ret = xd3_posix_io (ofile->file, buf, size, (xd3_posix_func*) &write, NULL);
924
925#elif XD3_WIN32
926 DWORD nwrite;
927 if (WriteFile(ofile->file, buf, size, &nwrite, NULL) == 0) {
928 ret = get_errno ();
929 } else {
930 if (size != nwrite) {
931 XPR(NT "Incorrect write count");
932 ret = XD3_INTERNAL;
933 }
934 }
935#endif
936
937 if (ret)
938 {
939 XPR(NT "%s: %s: %s\n", msg, ofile->filename, xd3_mainerror (ret));
940 }
941 else
942 {
943 if (option_verbose > 3) { XPR(NT "main write: %s: %u\n", ofile->filename, size); }
944 ofile->nwrite += size;
945 }
946
947 return ret;
948}
949
950static int
951main_file_seek (main_file *xfile, xoff_t pos)
952{
953 int ret = 0;
954
955#if XD3_STDIO
956 if (fseek (xfile->file, pos, SEEK_SET) != 0) { ret = get_errno (); }
957
958#elif XD3_POSIX
959 if (lseek (xfile->file, pos, SEEK_SET) != pos) { ret = get_errno (); }
960
961#elif XD3_WIN32
962 LARGE_INTEGER move, out;
963 move.QuadPart = pos;
964 if (SetFilePointerEx(xfile->file, move, &out, FILE_BEGIN) == 0) {
965 ret = get_errno ();
966 }
967#endif
968
969 if (ret)
970 {
971 XPR(NT "seek failed: %s: %s\n", xfile->filename, xd3_mainerror (ret));
972 }
973
974 return ret;
975}
976
977/******************************************************************************************
978 VCDIFF TOOLS
979 ******************************************************************************************/
980
981#if VCDIFF_TOOLS
982
983/* The following macros let VCDIFF printing something printf-like with
984 * main_file_write(), e.g.,:
985 *
986 * VC(UT "trying to be portable: %d\n", x)VE;
987 */
988#define SNPRINTF_BUFSIZE 1024
989#define VC do { if (((ret = snprintf
990#define UT (char*)xfile->snprintf_buf, SNPRINTF_BUFSIZE,
991#define VE ) >= SNPRINTF_BUFSIZE \
992 && (ret = main_print_overflow(ret)) != 0) \
993 || (ret = main_file_write(xfile, xfile->snprintf_buf, \
994 ret, "print")) != 0) \
995 { return ret; } } while (0)
996
997#ifdef WIN32
998/* According to the internet, Windows vsnprintf() differs from most Unix
999 * implementations regarding the terminating 0 when the boundary condition
1000 * is met. It doesn't matter here, we don't rely on the trailing 0. */
1001#include <stdarg.h>
1002int
1003snprintf (char *str, int n, char *fmt, ...)
1004{
1005 va_list a;
1006 int ret;
1007 va_start (a, fmt);
1008 ret = vsnprintf (str, n, fmt, a);
1009 va_end (a);
1010 return ret;
1011}
1012#endif
1013
1014static int
1015main_print_overflow (int x)
1016{
1017 XPR(NT "internal print buffer overflow: %d bytes\n", x);
1018 return XD3_INTERNAL;
1019}
1020
1021/* This function prints a single VCDIFF window. */
1022static int
1023main_print_window (xd3_stream* stream, main_file *xfile)
1024{
1025 int ret;
1026 usize_t size = 0;
1027
1028 VC(UT " Offset Code Type1 Size1 @Addr1 + Type2 Size2 @Addr2\n")VE;
1029
1030 while (stream->inst_sect.buf < stream->inst_sect.buf_max)
1031 {
1032 uint code = stream->inst_sect.buf[0];
1033
1034 if ((ret = xd3_decode_instruction (stream))) { return ret; }
1035
1036 VC(UT " %06"Q"u %03u %s %3u", stream->dec_winstart + size, code,
1037 xd3_rtype_to_string (stream->dec_current1.type, option_print_cpymode),
1038 (usize_t) stream->dec_current1.size)VE;
1039
1040 if (stream->dec_current1.type != XD3_NOOP)
1041 {
1042 size += stream->dec_current1.size;
1043 if (stream->dec_current1.type >= XD3_CPY)
1044 {
1045 VC(UT " @%-6u", (usize_t)stream->dec_current1.addr)VE;
1046 }
1047 else
1048 {
1049 VC(UT " ")VE;
1050 }
1051 }
1052
1053 if (stream->dec_current2.type != XD3_NOOP)
1054 {
1055 size += stream->dec_current2.size;
1056 VC(UT " %s %3u",
1057 xd3_rtype_to_string (stream->dec_current2.type, option_print_cpymode),
1058 (usize_t)stream->dec_current2.size)VE;
1059
1060 if (stream->dec_current2.type >= XD3_CPY)
1061 {
1062 VC(UT " @%-6u", (usize_t)stream->dec_current2.addr)VE;
1063 }
1064 }
1065
1066 VC(UT "\n")VE;
1067 }
1068
1069 if (stream->dec_tgtlen != size && (stream->flags & XD3_SKIP_WINDOW) == 0)
1070 {
1071 XPR(NT "target window size inconsistency");
1072 return XD3_INTERNAL;
1073 }
1074
1075 if (stream->dec_position != stream->dec_maxpos)
1076 {
1077 XPR(NT "target window position inconsistency");
1078 return XD3_INTERNAL;
1079 }
1080
1081 if (stream->addr_sect.buf != stream->addr_sect.buf_max)
1082 {
1083 XPR(NT "address section inconsistency");
1084 return XD3_INTERNAL;
1085 }
1086
1087 return 0;
1088}
1089
1090static int
1091main_print_vcdiff_file (main_file *xfile, main_file *file, const char *type)
1092{
1093 int ret; /* Used by above macros */
1094 if (file->filename)
1095 {
1096 VC(UT "XDELTA filename (%s): %s\n", type, file->filename)VE;
1097 }
1098 if (file->compressor)
1099 {
1100 VC(UT "XDELTA ext comp (%s): %s\n", type, file->compressor->recomp_cmdname)VE;
1101 }
1102 return 0;
1103}
1104
1105/* This function prints a VCDIFF input, mainly for debugging purposes. */
1106static int
1107main_print_func (xd3_stream* stream, main_file *xfile)
1108{
1109 int ret;
1110
1111 if (xfile->snprintf_buf == NULL)
1112 {
1113 if ((xfile->snprintf_buf = main_malloc(SNPRINTF_BUFSIZE)) == NULL)
1114 {
1115 return ENOMEM;
1116 }
1117 }
1118
1119 if (stream->dec_winstart == 0)
1120 {
1121 VC(UT "VCDIFF version: 0\n")VE;
1122
1123 VC(UT "VCDIFF header size: %d\n", stream->dec_hdrsize)VE;
1124 VC(UT "VCDIFF header indicator: ")VE;
1125 if ((stream->dec_hdr_ind & VCD_SECONDARY) != 0) VC(UT "VCD_SECONDARY ")VE;
1126 if ((stream->dec_hdr_ind & VCD_CODETABLE) != 0) VC(UT "VCD_CODETABLE ")VE;
1127 if ((stream->dec_hdr_ind & VCD_APPHEADER) != 0) VC(UT "VCD_APPHEADER ")VE;
1128 if (stream->dec_hdr_ind == 0) VC(UT "none")VE;
1129 VC(UT "\n")VE;
1130
1131 IF_SEC(VC(UT "VCDIFF secondary compressor: %s\n",
1132 stream->sec_type ? stream->sec_type->name : "none")VE);
1133 IF_NSEC(VC(UT "VCDIFF secondary compressor: unsupported\n")VE);
1134
1135 if (stream->dec_hdr_ind & VCD_APPHEADER)
1136 {
1137 uint8_t *apphead;
1138 usize_t appheadsz;
1139 ret = xd3_get_appheader (stream, & apphead, & appheadsz);
1140
1141 if (ret == 0 && appheadsz > 0)
1142 {
1143 int sq = option_quiet;
1144 main_file i, o, s;
1145 XD3_ASSERT (apphead != NULL);
1146 VC(UT "VCDIFF application header: ")VE;
1147 if ((ret = main_file_write (xfile, apphead, appheadsz, "print")) != 0) { return ret; }
1148 VC(UT "\n")VE;
1149
1150 main_file_init (& i);
1151 main_file_init (& o);
1152 main_file_init (& s);
1153 option_quiet = 1;
1154 main_get_appheader (stream, &i, & o, & s);
1155 option_quiet = sq;
1156 if ((ret = main_print_vcdiff_file (xfile, & o, "output"))) { return ret; }
1157 if ((ret = main_print_vcdiff_file (xfile, & s, "source"))) { return ret; }
1158 main_file_cleanup (& i);
1159 main_file_cleanup (& o);
1160 main_file_cleanup (& s);
1161 }
1162 }
1163 }
1164 else
1165 {
1166 VC(UT "\n")VE;
1167 }
1168
1169 VC(UT "VCDIFF window number: %"Q"u\n", stream->current_window)VE;
1170 VC(UT "VCDIFF window indicator: ")VE;
1171 if ((stream->dec_win_ind & VCD_SOURCE) != 0) VC(UT "VCD_SOURCE ")VE;
1172 if ((stream->dec_win_ind & VCD_TARGET) != 0) VC(UT "VCD_TARGET ")VE;
1173 if ((stream->dec_win_ind & VCD_ADLER32) != 0) VC(UT "VCD_ADLER32 ")VE;
1174 if (stream->dec_win_ind == 0) VC(UT "none")VE;
1175 VC(UT "\n")VE;
1176
1177 if ((stream->dec_win_ind & VCD_ADLER32) != 0)
1178 {
1179 VC(UT "VCDIFF adler32 checksum: %08X\n", (usize_t)stream->dec_adler32)VE;
1180 }
1181
1182 if (stream->dec_del_ind != 0)
1183 {
1184 VC(UT "VCDIFF delta indicator: ")VE;
1185 if ((stream->dec_del_ind & VCD_DATACOMP) != 0) VC(UT "VCD_DATACOMP ")VE;
1186 if ((stream->dec_del_ind & VCD_INSTCOMP) != 0) VC(UT "VCD_INSTCOMP ")VE;
1187 if ((stream->dec_del_ind & VCD_ADDRCOMP) != 0) VC(UT "VCD_ADDRCOMP ")VE;
1188 if (stream->dec_del_ind == 0) VC(UT "none")VE;
1189 VC(UT "\n")VE;
1190 }
1191
1192 if (stream->dec_winstart != 0)
1193 {
1194 VC(UT "VCDIFF window at offset: %"Q"u\n", stream->dec_winstart)VE;
1195 }
1196
1197 if (SRCORTGT (stream->dec_win_ind))
1198 {
1199 VC(UT "VCDIFF copy window length: %u\n", (usize_t)stream->dec_cpylen)VE;
1200 VC(UT "VCDIFF copy window offset: %"Q"u\n", stream->dec_cpyoff)VE;
1201 }
1202
1203 VC(UT "VCDIFF delta encoding length: %u\n", (usize_t)stream->dec_enclen)VE;
1204 VC(UT "VCDIFF target window length: %u\n", (usize_t)stream->dec_tgtlen)VE;
1205
1206 VC(UT "VCDIFF data section length: %u\n", (usize_t)stream->data_sect.size)VE;
1207 VC(UT "VCDIFF inst section length: %u\n", (usize_t)stream->inst_sect.size)VE;
1208 VC(UT "VCDIFF addr section length: %u\n", (usize_t)stream->addr_sect.size)VE;
1209
1210 ret = 0;
1211 if ((stream->flags & XD3_JUST_HDR) != 0)
1212 {
1213 /* Print a header -- finished! */
1214 ret = PRINTHDR_SPECIAL;
1215 }
1216 else if ((stream->flags & XD3_SKIP_WINDOW) == 0)
1217 {
1218 ret = main_print_window (stream, xfile);
1219 }
1220
1221 return ret;
1222}
1223#endif /* VCDIFF_TOOLS */
1224
1225/******************************************************************************************
1226 Input decompression, output recompression
1227 ******************************************************************************************/
1228
1229#if EXTERNAL_COMPRESSION
1230/* This is tricky POSIX-specific code with lots of fork(), pipe(), dup(), waitpid(), and
1231 * exec() business. Most of this code originated in PRCS1, which did automatic
1232 * package-file decompression. It works with both XD3_POSIX and XD3_STDIO file
1233 * disciplines.
1234 *
1235 * To automatically detect compressed inputs requires a child process to reconstruct the
1236 * input stream, which was advanced in order to detect compression, because it may not be
1237 * seekable. In other words, the main program reads part of the input stream, and if it
1238 * detects a compressed input it then forks a pipe copier process, which copies the
1239 * first-read block out of the main-program's memory, then streams the remaining
1240 * compressed input into the input-decompression pipe.
1241 */
1242
1243#include <unistd.h>
1244#include <sys/stat.h>
1245#include <sys/wait.h>
1246
1247/* Remember which pipe FD is which. */
1248#define PIPE_READ_FD 0
1249#define PIPE_WRITE_FD 1
1250
1251static pid_t ext_subprocs[2];
1252static char* ext_tmpfile = NULL;
1253
1254/* Like write(), but makes repeated calls to empty the buffer. */
1255static int
1256main_pipe_write (int outfd, const uint8_t *exist_buf, usize_t remain)
1257{
1258 int ret;
1259
1260 if ((ret = xd3_posix_io (outfd, (uint8_t*) exist_buf, remain, (xd3_posix_func*) &write, NULL)))
1261 {
1262 XPR(NT "pipe write failed: %s", xd3_mainerror (ret));
1263 return ret;
1264 }
1265
1266 return 0;
1267}
1268
1269/* A simple error-reporting waitpid interface. */
1270static int
1271main_waitpid_check(pid_t pid)
1272{
1273 int status;
1274 int ret = 0;
1275
1276 if (waitpid (pid, & status, 0) < 0)
1277 {
1278 ret = get_errno ();
1279 XPR(NT "compression subprocess: wait: %s\n", xd3_mainerror (ret));
1280 }
1281 else if (! WIFEXITED (status))
1282 {
1283 ret = ECHILD;
1284 XPR(NT "compression subprocess: signal %d\n",
1285 WIFSIGNALED (status) ? WTERMSIG (status) : WSTOPSIG (status));
1286 }
1287 else if (WEXITSTATUS (status) != 0)
1288 {
1289 ret = ECHILD;
1290 XPR(NT "compression subprocess: exit %d\n", WEXITSTATUS (status));
1291 }
1292
1293 return ret;
1294}
1295
1296/* Wait for any existing child processes to check for abnormal exit. */
1297static int
1298main_external_compression_finish (void)
1299{
1300 int i;
1301 int ret;
1302
1303 for (i = 0; i < 2; i += 1)
1304 {
1305 if (! ext_subprocs[i]) { continue; }
1306
1307 if ((ret = main_waitpid_check (ext_subprocs[i])))
1308 {
1309 return ret;
1310 }
1311 }
1312
1313 return 0;
1314}
1315
1316/* This runs as a forked process of main_input_decompress_setup() to copy input to the
1317 * decompression process. First, the available input is copied out of the existing
1318 * buffer, then the buffer is reused to continue reading from the compressed input
1319 * file. */
1320static int
1321main_pipe_copier (uint8_t *pipe_buf,
1322 usize_t pipe_bufsize,
1323 usize_t nread,
1324 main_file *ifile,
1325 int outfd)
1326{
1327 int ret;
1328
1329 for (;;)
1330 {
1331 if (nread > 0 && (ret = main_pipe_write (outfd, pipe_buf, nread)))
1332 {
1333 return ret;
1334 }
1335
1336 if (nread < pipe_bufsize)
1337 {
1338 break;
1339 }
1340
1341 if ((ret = main_file_read (ifile, pipe_buf, pipe_bufsize, & nread, "pipe read failed")) < 0)
1342 {
1343 return ret;
1344 }
1345 }
1346
1347 return 0;
1348}
1349
1350/* This function is called after we have read some amount of data from the input file and
1351 * detected a compressed input. Here we start a decompression subprocess by forking
1352 * twice. The first process runs the decompression command, the second process copies
1353 * data to the input of the first. */
1354static int
1355main_input_decompress_setup (const main_extcomp *decomp,
1356 main_file *ifile,
1357 uint8_t *input_buf,
1358 usize_t input_bufsize,
1359 uint8_t *pipe_buf,
1360 usize_t pipe_bufsize,
1361 usize_t pipe_avail,
1362 usize_t *nread)
1363{
1364 int outpipefd[2], inpipefd[2]; /* The two pipes: input and output file descriptors. */
1365 int input_fd = -1; /* The resulting input_fd (output of decompression). */
1366 pid_t decomp_id, copier_id; /* The two subprocs. */
1367 int ret;
1368
1369 outpipefd[0] = outpipefd[1] = -1;
1370 inpipefd[0] = inpipefd[1] = -1;
1371
1372 if (pipe (outpipefd) || pipe (inpipefd))
1373 {
1374 XPR(NT "pipe failed: %s\n", xd3_mainerror (ret = get_errno ()));
1375 goto pipe_cleanup;
1376 }
1377
1378 if ((decomp_id = fork ()) < 0)
1379 {
1380 XPR(NT "fork failed: %s\n", xd3_mainerror (ret = get_errno ()));
1381 goto pipe_cleanup;
1382 }
1383
1384 /* The first child runs the decompression process: */
1385 if (decomp_id == 0)
1386 {
1387 /* Setup pipes: write to the outpipe, read from the inpipe. */
1388 if (dup2 (outpipefd[PIPE_WRITE_FD], STDOUT_FILENO) < 0 ||
1389 dup2 (inpipefd[PIPE_READ_FD], STDIN_FILENO) < 0 ||
1390 close (outpipefd[PIPE_READ_FD]) ||
1391 close (outpipefd[PIPE_WRITE_FD]) ||
1392 close (inpipefd[PIPE_READ_FD]) ||
1393 close (inpipefd[PIPE_WRITE_FD]) ||
1394 execlp (decomp->decomp_cmdname, decomp->decomp_cmdname, decomp->decomp_options, NULL))
1395 {
1396 XPR(NT "child process %s failed to execute: %s\n", decomp->decomp_cmdname, xd3_mainerror (get_errno ()));
1397 }
1398
1399 _exit (127);
1400 }
1401
1402 ext_subprocs[0] = decomp_id;
1403
1404 if ((copier_id = fork ()) < 0)
1405 {
1406 XPR(NT "fork failed: %s\n", xd3_mainerror (ret = get_errno ()));
1407 goto pipe_cleanup;
1408 }
1409
1410 /* The second child runs the copier process: */
1411 if (copier_id == 0)
1412 {
1413 int exitval = 0;
1414
1415 if (close (inpipefd[PIPE_READ_FD]) ||
1416 main_pipe_copier (pipe_buf, pipe_bufsize, pipe_avail, ifile, inpipefd[PIPE_WRITE_FD]) ||
1417 close (inpipefd[PIPE_WRITE_FD]))
1418 {
1419 XPR(NT "child copier process failed: %s\n", xd3_mainerror (get_errno ()));
1420 exitval = 1;
1421 }
1422
1423 _exit (exitval);
1424 }
1425
1426 ext_subprocs[1] = copier_id;
1427
1428 /* The parent closes both pipes after duplicating the output of compression. */
1429 input_fd = dup (outpipefd[PIPE_READ_FD]);
1430
1431 if (input_fd < 0 ||
1432 main_file_close (ifile) ||
1433 close (outpipefd[PIPE_READ_FD]) ||
1434 close (outpipefd[PIPE_WRITE_FD]) ||
1435 close (inpipefd[PIPE_READ_FD]) ||
1436 close (inpipefd[PIPE_WRITE_FD]))
1437 {
1438 XPR(NT "dup/close failed: %s\n", xd3_mainerror (ret = get_errno ()));
1439 goto pipe_cleanup;
1440 }
1441
1442#if XD3_STDIO
1443 /* Note: fdopen() acquires the fd, closes it when finished. */
1444 if ((ifile->file = fdopen (input_fd, "r")) == NULL)
1445 {
1446 XPR(NT "fdopen failed: %s\n", xd3_mainerror (ret = get_errno ()));
1447 goto pipe_cleanup;
1448 }
1449
1450#elif XD3_POSIX
1451 ifile->file = input_fd;
1452#endif
1453
1454 ifile->compressor = decomp;
1455
1456 /* Now the input file is decompressed. */
1457 return main_file_read (ifile, input_buf, input_bufsize, nread, "input decompression failed");
1458
1459 pipe_cleanup:
1460 close (input_fd);
1461 close (outpipefd[PIPE_READ_FD]);
1462 close (outpipefd[PIPE_WRITE_FD]);
1463 close (inpipefd[PIPE_READ_FD]);
1464 close (inpipefd[PIPE_WRITE_FD]);
1465 return ret;
1466}
1467
1468
1469/* This routine is called when the first buffer of input data is read by the main program
1470 * (unless input decompression is disabled by command-line option). If it recognizes the
1471 * magic number of a known input type it invokes decompression.
1472 *
1473 * Skips decompression if the decompression type or the file type is RD_NONEXTERNAL.
1474 *
1475 * Behaves exactly like main_file_read, otherwise.
1476 *
1477 * This function uses a separate buffer to read the first small block of input. If a
1478 * compressed input is detected, the separate buffer is passed to the pipe copier. This
1479 * avoids using the same size buffer in both cases. */
1480static int
1481main_decompress_input_check (main_file *ifile,
1482 uint8_t *input_buf,
1483 usize_t input_size,
1484 usize_t *nread)
1485{
1486 int i;
1487 int ret;
1488 uint8_t check_buf[XD3_ALLOCSIZE];
1489 usize_t check_nread;
1490
1491 if ((ret = main_file_read (ifile, check_buf, min (input_size, XD3_ALLOCSIZE), & check_nread, "input read failed")))
1492 {
1493 return ret;
1494 }
1495
1496 for (i = 0; i < SIZEOF_ARRAY (extcomp_types); i += 1)
1497 {
1498 const main_extcomp *decomp = & extcomp_types[i];
1499
1500 if ((check_nread > decomp->magic_size) &&
1501 /* The following expr skips decompression if we are trying to read a VCDIFF
1502 * input and that is the magic number. */
1503 !((decomp->flags & RD_NONEXTERNAL) && (ifile->flags & RD_NONEXTERNAL)) &&
1504 memcmp (check_buf, decomp->magic, decomp->magic_size) == 0)
1505 {
1506 if (! option_quiet)
1507 {
1508 XPR(NT "%s | %s %s\n",
1509 ifile->filename,
1510 decomp->decomp_cmdname,
1511 decomp->decomp_options);
1512 }
1513
1514 return main_input_decompress_setup (decomp, ifile,
1515 input_buf, input_size,
1516 check_buf, XD3_ALLOCSIZE,
1517 check_nread, nread);
1518 }
1519 }
1520
1521 /* Now read the rest of the input block. */
1522 (*nread) = 0;
1523
1524 if (check_nread == XD3_ALLOCSIZE)
1525 {
1526 ret = main_file_read (ifile, input_buf + XD3_ALLOCSIZE,
1527 input_size - XD3_ALLOCSIZE, nread,
1528 "input read failed");
1529 }
1530
1531 memcpy (input_buf, check_buf, check_nread);
1532
1533 (*nread) += check_nread;
1534
1535 return 0;
1536}
1537
1538/* This is called when the source file needs to be decompressed. We fork/exec a
1539 * decompression command with the proper input and output to a temporary file. */
1540static int
1541main_decompress_source (main_file *sfile, xd3_source *source)
1542{
1543 const main_extcomp *decomp = sfile->compressor;
1544 pid_t decomp_id; /* One subproc. */
1545 int input_fd = -1;
1546 int output_fd = -1;
1547 int ret;
1548 char *tmpname = NULL;
1549 char *tmpdir = getenv ("TMPDIR");
1550 static const char tmpl[] = "/xd3src.XXXXXX";
1551
1552 /* Make a template for mkstmp() */
1553 if (tmpdir == NULL) { tmpdir = "/tmp"; }
1554 if ((tmpname = main_malloc (strlen (tmpdir) + sizeof (tmpl) + 1)) == NULL) { return ENOMEM; }
1555 sprintf (tmpname, "%s%s", tmpdir, tmpl);
1556
1557 XD3_ASSERT (ext_tmpfile == NULL);
1558 ext_tmpfile = tmpname;
1559
1560 /* Open the output FD. */
1561 if ((output_fd = mkstemp (tmpname)) < 0)
1562 {
1563 XPR(NT "mkstemp failed: %s: %s", tmpname, xd3_mainerror (ret = get_errno ()));
1564 goto cleanup;
1565 }
1566
1567 /* Copy the input FD, reset file position. */
1568 XD3_ASSERT (main_file_isopen (sfile));
1569#if XD3_STDIO
1570 if ((input_fd = dup (fileno (sfile->file))) < 0)
1571 {
1572 XPR(NT "dup failed: %s", xd3_mainerror (ret = get_errno ()));
1573 goto cleanup;
1574 }
1575 main_file_close (sfile);
1576 sfile->file = NULL;
1577#elif XD3_POSIX
1578 input_fd = sfile->file;
1579 sfile->file = -1;
1580#endif
1581
1582 if ((ret = lseek (input_fd, SEEK_SET, 0)) != 0)
1583 {
1584 XPR(NT "lseek failed: : %s", xd3_mainerror (ret = get_errno ()));
1585 goto cleanup;
1586 }
1587
1588 if ((decomp_id = fork ()) < 0)
1589 {
1590 XPR(NT "fork failed: %s", xd3_mainerror (ret = get_errno ()));
1591 goto cleanup;
1592 }
1593
1594 /* The child runs the decompression process: */
1595 if (decomp_id == 0)
1596 {
1597 /* Setup pipes: write to the output file, read from the pipe. */
1598 if (dup2 (input_fd, STDIN_FILENO) < 0 ||
1599 dup2 (output_fd, STDOUT_FILENO) < 0 ||
1600 execlp (decomp->decomp_cmdname, decomp->decomp_cmdname, decomp->decomp_options, NULL))
1601 {
1602 XPR(NT "child process %s failed to execute: %s\n",
1603 decomp->decomp_cmdname, xd3_mainerror (get_errno ()));
1604 }
1605
1606 _exit (127);
1607 }
1608
1609 close (input_fd);
1610 close (output_fd);
1611 input_fd = -1;
1612 output_fd = -1;
1613
1614 /* Then wait for completion. */
1615 if ((ret = main_waitpid_check (decomp_id)))
1616 {
1617 goto cleanup;
1618 }
1619
1620 /* Open/stat the decompressed source file. */
1621 if ((ret = main_file_open (sfile, tmpname, XO_READ))) { goto cleanup; }
1622 if ((ret = main_file_stat (sfile, & source->size, 1))) { goto cleanup; }
1623 return 0;
1624
1625 cleanup:
1626 close (input_fd);
1627 close (output_fd);
1628 if (tmpname) { free (tmpname); }
1629 ext_tmpfile = NULL;
1630 return ret;
1631}
1632
1633/* Initiate re-compression of the output stream. This is easier than input decompression
1634 * because we know beforehand that the stream will be compressed, whereas the input has
1635 * already been read when we decide it should be decompressed. Thus, it only requires one
1636 * subprocess and one pipe. */
1637static int
1638main_recompress_output (main_file *ofile)
1639{
1640 pid_t recomp_id; /* One subproc. */
1641 int pipefd[2]; /* One pipe. */
1642 int output_fd = -1;
1643 int ret;
1644 const main_extcomp *recomp = ofile->compressor;
1645
1646 pipefd[0] = pipefd[1] = -1;
1647
1648 if (pipe (pipefd))
1649 {
1650 XPR(NT "pipe failed: %s\n", xd3_mainerror (ret = get_errno ()));
1651 goto pipe_cleanup;
1652 }
1653
1654 if ((recomp_id = fork ()) < 0)
1655 {
1656 XPR(NT "fork failed: %s\n", xd3_mainerror (ret = get_errno ()));
1657 goto pipe_cleanup;
1658 }
1659
1660 /* The child runs the recompression process: */
1661 if (recomp_id == 0)
1662 {
1663 /* Setup pipes: write to the output file, read from the pipe. */
1664 if (dup2 (XFNO (ofile), STDOUT_FILENO) < 0 ||
1665 dup2 (pipefd[PIPE_READ_FD], STDIN_FILENO) < 0 ||
1666 close (pipefd[PIPE_READ_FD]) ||
1667 close (pipefd[PIPE_WRITE_FD]) ||
1668 execlp (recomp->recomp_cmdname, recomp->recomp_cmdname, recomp->recomp_options, NULL))
1669 {
1670 XPR(NT "child process %s failed to execute: %s\n", recomp->recomp_cmdname, xd3_mainerror (get_errno ()));
1671 }
1672
1673 _exit (127);
1674 }
1675
1676 ext_subprocs[0] = recomp_id;
1677
1678 /* The parent closes both pipes after duplicating the output-fd for writing to the
1679 * compression pipe. */
1680 output_fd = dup (pipefd[PIPE_WRITE_FD]);
1681
1682 if (output_fd < 0 ||
1683 main_file_close (ofile) ||
1684 close (pipefd[PIPE_READ_FD]) ||
1685 close (pipefd[PIPE_WRITE_FD]))
1686 {
1687 XPR(NT "close failed: %s\n", xd3_mainerror (ret = get_errno ()));
1688 goto pipe_cleanup;
1689 }
1690
1691#if XD3_STDIO
1692 /* Note: fdopen() acquires the fd, closes it when finished. */
1693 if ((ofile->file = fdopen (output_fd, "w")) == NULL)
1694 {
1695 XPR(NT "fdopen failed: %s\n", xd3_mainerror (ret = get_errno ()));
1696 goto pipe_cleanup;
1697 }
1698
1699#elif XD3_POSIX
1700 ofile->file = output_fd;
1701#endif
1702
1703 /* Now the output file will be compressed. */
1704 return 0;
1705
1706 pipe_cleanup:
1707 close (output_fd);
1708 close (pipefd[PIPE_READ_FD]);
1709 close (pipefd[PIPE_WRITE_FD]);
1710 return ret;
1711}
1712#endif /* EXTERNAL_COMPRESSION */
1713
1714/* Identify the compressor that was used based on its ident string, which is passed in the
1715 * application header. */
1716static const main_extcomp*
1717main_ident_compressor (const char *ident)
1718{
1719 int i;
1720
1721 for (i = 0; i < SIZEOF_ARRAY (extcomp_types); i += 1)
1722 {
1723 if (strcmp (extcomp_types[i].ident, ident) == 0)
1724 {
1725 return & extcomp_types[i];
1726 }
1727 }
1728
1729 return NULL;
1730}
1731
1732/* Return the main_extcomp record to use for this identifier, if possible. */
1733static const main_extcomp*
1734main_get_compressor (const char *ident)
1735{
1736 const main_extcomp *ext = main_ident_compressor (ident);
1737
1738 if (ext == NULL)
1739 {
1740 if (! option_quiet)
1741 {
1742 XPR(NT "warning: cannot recompress output: "
1743 "unrecognized external compression ID: %s\n", ident);
1744 }
1745 return NULL;
1746 }
1747 else if (! EXTERNAL_COMPRESSION)
1748 {
1749 if (! option_quiet)
1750 {
1751 XPR(NT "warning: external support not compiled: "
1752 "original input was compressed: %s\n", ext->recomp_cmdname);
1753 }
1754 return NULL;
1755 }
1756 else
1757 {
1758 return ext;
1759 }
1760}
1761
1762/******************************************************************************************
1763 APPLICATION HEADER
1764 ******************************************************************************************/
1765
1766#if XD3_ENCODER
1767static const char*
1768main_apphead_string (const char* x)
1769{
1770 const char *y;
1771
1772 if (x == NULL) { return ""; }
1773
1774 if (strcmp (x, "/dev/stdin") == 0 ||
1775 strcmp (x, "/dev/stdout") == 0 ||
1776 strcmp (x, "/dev/stderr") == 0) { return "-"; }
1777
1778 // TODO: this is not portable
1779 return (y = strrchr (x, '/')) == NULL ? x : y + 1;
1780}
1781
1782static int
1783main_set_appheader (xd3_stream *stream, main_file *input, main_file *sfile)
1784{
1785 /* The user may disable the application header. Once the appheader is set, this
1786 * disables setting it again. */
1787 if (appheader_used || ! option_use_appheader) { return 0; }
1788
1789 /* The user may specify the application header, otherwise format the default header. */
1790 if (option_appheader)
1791 {
1792 appheader_used = option_appheader;
1793 }
1794 else
1795 {
1796 const char *iname;
1797 const char *icomp;
1798 const char *sname;
1799 const char *scomp;
1800 int len;
1801
1802 iname = main_apphead_string (input->filename);
1803 icomp = (input->compressor == NULL) ? "" : input->compressor->ident;
1804 len = strlen (iname) + strlen (icomp) + 2;
1805
1806 if (sfile->filename != NULL)
1807 {
1808 sname = main_apphead_string (sfile->filename);
1809 scomp = (sfile->compressor == NULL) ? "" : sfile->compressor->ident;
1810 len += strlen (sname) + strlen (scomp) + 2;
1811 }
1812 else
1813 {
1814 sname = scomp = "";
1815 }
1816
1817 if ((appheader_used = main_malloc (len)) == NULL)
1818 {
1819 return ENOMEM;
1820 }
1821
1822 if (sfile->filename == NULL)
1823 {
1824 sprintf ((char*)appheader_used, "%s/%s", iname, icomp);
1825 }
1826 else
1827 {
1828 sprintf ((char*)appheader_used, "%s/%s/%s/%s", iname, icomp, sname, scomp);
1829 }
1830 }
1831
1832 xd3_set_appheader (stream, appheader_used, strlen ((char*)appheader_used));
1833
1834 return 0;
1835}
1836#endif
1837
1838static void
1839main_get_appheader_params (main_file *file, char **parsed, int output, const char *type,
1840 main_file *other)
1841{
1842 /* Set the filename if it was not specified. If output, option_stdout (-c) overrides. */
1843 if (file->filename == NULL && ! (output && option_stdout) && strcmp (parsed[0], "-") != 0)
1844 {
1845 file->filename = parsed[0];
1846
1847 if (other->filename != NULL) {
1848 /* Take directory from the other file, if it has one. */
1849 /* TODO: This results in nonsense names like /dev/foo.tar.gz
1850 * and probably the filename-default logic interferes with
1851 * multi-file operation and the standard file extension?
1852 * Possibly the name header is bad, should be off by default.
1853 * Possibly we just want to remember external/compression
1854 * settings. */
1855 char *last_slash = strrchr(other->filename, '/');
1856
1857 if (last_slash != NULL) {
1858 int dlen = last_slash - other->filename;
1859
1860 XD3_ASSERT(file->filename_copy == NULL);
1861 file->filename_copy = main_malloc(dlen + 2 + strlen(file->filename));
1862
1863 strncpy(file->filename_copy, other->filename, dlen);
1864 file->filename_copy[dlen] = '/';
1865 strcpy(file->filename_copy + dlen + 1, parsed[0]);
1866
1867 file->filename = file->filename_copy;
1868 }
1869 }
1870
1871 if (! option_quiet)
1872 {
1873 XPR(NT "using default %s filename: %s\n", type, file->filename);
1874 }
1875 }
1876
1877 /* Set the compressor, initiate de/recompression later. */
1878 if (file->compressor == NULL && *parsed[1] != 0)
1879 {
1880 file->compressor = main_get_compressor (parsed[1]);
1881 }
1882}
1883
1884static void
1885main_get_appheader (xd3_stream *stream, main_file *ifile, main_file *output, main_file *sfile)
1886{
1887 uint8_t *apphead;
1888 usize_t appheadsz;
1889 int ret;
1890
1891 /* The user may disable the application header. Once the appheader is set, this
1892 * disables setting it again. */
1893 if (! option_use_appheader) { return; }
1894
1895 ret = xd3_get_appheader (stream, & apphead, & appheadsz);
1896
1897 /* Ignore failure, it only means we haven't received a header yet. */
1898 if (ret != 0) { return; }
1899
1900 if (appheadsz > 0)
1901 {
1902 char *start = (char*)apphead;
1903 char *slash;
1904 int place = 0;
1905 char *parsed[4];
1906
1907 memset (parsed, 0, sizeof (parsed));
1908
1909 while ((slash = strchr (start, '/')) != NULL)
1910 {
1911 *slash = 0;
1912 parsed[place++] = start;
1913 start = slash + 1;
1914 }
1915
1916 parsed[place++] = start;
1917
1918 /* First take the output parameters. */
1919 if (place == 2 || place == 4)
1920 {
1921 main_get_appheader_params (output, parsed, 1, "output", ifile);
1922 }
1923
1924 /* Then take the source parameters. */
1925 if (place == 4)
1926 {
1927 main_get_appheader_params (sfile, parsed+2, 0, "source", ifile);
1928 }
1929 }
1930
1931 option_use_appheader = 0;
1932 return;
1933}
1934
1935/******************************************************************************************
1936 Main I/O routines
1937 ******************************************************************************************/
1938
1939/* This function acts like the above except it may also try to recognize a compressed
1940 * input when the first buffer of data is read. The EXTERNAL_COMPRESSION code is called
1941 * to search for magic numbers. */
1942static int
1943main_read_primary_input (main_file *ifile,
1944 uint8_t *buf,
1945 usize_t size,
1946 usize_t *nread)
1947{
1948#if EXTERNAL_COMPRESSION
1949 if (option_decompress_inputs && ifile->flags & RD_FIRST)
1950 {
1951 ifile->flags &= ~RD_FIRST;
1952
1953 return main_decompress_input_check (ifile, buf, size, nread);
1954 }
1955#endif
1956
1957 return main_file_read (ifile, buf, size, nread, "input read failed");
1958}
1959
1960/* This function simply writes the stream output buffer, if there is any. This is used
1961 * for both encode and decode commands. (The VCDIFF tools use main_print_func()). */
1962static int
1963main_write_output (xd3_stream* stream, main_file *ofile)
1964{
1965 int ret;
1966
1967 if (stream->avail_out > 0 && (ret = main_file_write (ofile, stream->next_out, stream->avail_out, "write failed")))
1968 {
1969 return ret;
1970 }
1971
1972 return 0;
1973}
1974
1975/* Open the main output file, sets a default file name, initiate recompression. This
1976 * function is expected to fprint any error messages. */
1977static int
1978main_open_output (xd3_stream *stream, main_file *ofile)
1979{
1980 int ret;
1981
1982 if (ofile->filename == NULL)
1983 {
1984 XSTDOUT_XF (ofile);
1985
1986 if (option_verbose > 1) { XPR(NT "using standard output: %s\n", ofile->filename); }
1987 }
1988 else
1989 {
1990 /* Stat the file to check for overwrite. */
1991 if (option_force == 0 && main_file_exists (ofile))
1992 {
1993 XPR(NT "to overwrite output file specify -f: %s\n", ofile->filename);
1994 return EEXIST;
1995 }
1996
1997 if ((ret = main_file_open (ofile, ofile->filename, XO_WRITE)))
1998 {
1999 return ret;
2000 }
2001
2002 if (option_verbose > 1) { XPR(NT "output file: %s\n", ofile->filename); }
2003 }
2004
2005#if EXTERNAL_COMPRESSION
2006 /* Do output recompression. */
2007 if (ofile->compressor != NULL && option_recompress_outputs == 1)
2008 {
2009 if (! option_quiet)
2010 {
2011 XPR(NT "%s %s | %s\n",
2012 ofile->compressor->recomp_cmdname,
2013 ofile->compressor->recomp_options,
2014 ofile->filename);
2015 }
2016
2017 if ((ret = main_recompress_output (ofile)))
2018 {
2019 return ret;
2020 }
2021 }
2022#endif
2023
2024 return 0;
2025}
2026
2027/* This is called at different times for encoding and decoding. The encoder calls it
2028 * immediately, the decoder delays until the application header is received.
2029 * Stream may be NULL, in which case xd3_set_source is not called. */
2030static int
2031main_set_source (xd3_stream *stream, int cmd, main_file *sfile, xd3_source *source)
2032{
2033 int ret = 0, i;
2034 uint8_t *tmp_buf = NULL;
2035
2036 /* Open it, check for seekability, set required xd3_source fields. */
2037 if (allow_fake_source)
2038 {
2039 sfile->mode = XO_READ;
2040 sfile->realname = sfile->filename;
2041 sfile->nread = 0;
2042 source->size = UINT64_MAX;
2043 }
2044 else
2045 {
2046 if ((ret = main_file_open (sfile, sfile->filename, XO_READ)) ||
2047 (ret = main_file_stat (sfile, & source->size, 1)))
2048 {
2049 goto error;
2050 }
2051 }
2052
2053 source->name = sfile->filename;
2054 source->ioh = sfile;
2055 source->curblkno = (xoff_t) -1;
2056 source->curblk = NULL;
2057
2058#if EXTERNAL_COMPRESSION
2059 if (option_decompress_inputs)
2060 {
2061 /* If encoding, read the header to check for decompression. */
2062 if (IS_ENCODE (cmd))
2063 {
2064 usize_t nread;
2065 tmp_buf = main_malloc(XD3_ALLOCSIZE);
2066
2067 if ((ret = main_file_read (sfile, tmp_buf, XD3_ALLOCSIZE,
2068 & nread, "source read failed")))
2069 {
2070 goto error;
2071 }
2072
2073 /* Check known magic numbers. */
2074 for (i = 0; i < SIZEOF_ARRAY (extcomp_types); i += 1)
2075 {
2076 const main_extcomp *decomp = & extcomp_types[i];
2077
2078 if ((nread > decomp->magic_size) &&
2079 memcmp (tmp_buf, decomp->magic, decomp->magic_size) == 0)
2080 {
2081 sfile->compressor = decomp;
2082 break;
2083 }
2084 }
2085
2086 if (sfile->compressor == NULL)
2087 {
2088 if (option_verbose > 2)
2089 {
2090 XPR(NT "source block 0 read (not compressed)\n");
2091 }
2092 }
2093 }
2094
2095 /* In either the encoder or decoder, start decompression. */
2096 if (sfile->compressor)
2097 {
2098 xoff_t osize = source->size;
2099
2100 if (osize > XD3_NODECOMPRESSSIZE)
2101 {
2102 XPR(NT "source file too large for external decompression: %s: %"Q"u\n",
2103 sfile->filename, osize);
2104 ret = XD3_INTERNAL;
2105 goto error;
2106 }
2107
2108 if ((ret = main_decompress_source (sfile, source)))
2109 {
2110 goto error;
2111 }
2112
2113 if (! option_quiet)
2114 {
2115 char s1[32], s2[32];
2116 XPR(NT "%s | %s %s => %s %.1f%% [ %s , %s ]\n",
2117 sfile->filename,
2118 sfile->compressor->decomp_cmdname,
2119 sfile->compressor->decomp_options,
2120 sfile->realname,
2121 100.0 * source->size / osize,
2122 main_format_bcnt (osize, s1),
2123 main_format_bcnt (source->size, s2));
2124 }
2125 }
2126 }
2127#endif
2128
2129 /* At this point we know source->size.
2130 * Source buffer, blksize, LRU init. */
2131 if (source->size < option_srcwinsz)
2132 {
2133 /* Reduce sizes to actual source size, read whole file */
2134 option_srcwinsz = source->size;
2135 source->blksize = source->size;
2136 lru_size = 1;
2137 }
2138 else
2139 {
2140 option_srcwinsz = max(option_srcwinsz, XD3_MINSRCWINSZ);
2141
2142 source->blksize = (option_srcwinsz / LRU_SIZE);
2143 lru_size = LRU_SIZE;
2144 }
2145
2146 main_blklru_list_init (& lru_list);
2147 main_blklru_list_init (& lru_free);
2148
2149 if (option_verbose)
2150 {
2151 static char buf[32];
2152
2153 XPR(NT "source %s winsize %s size %"Q"u\n",
2154 sfile->filename, main_format_bcnt(option_srcwinsz, buf), source->size);
2155 }
2156
2157 if (option_verbose > 1)
2158 {
2159 XPR(NT "source block size: %u\n", source->blksize);
2160 }
2161
2162 if ((lru = main_malloc (sizeof (main_blklru) * lru_size)) == NULL)
2163 {
2164 ret = ENOMEM;
2165 goto error;
2166 }
2167
2168 for (i = 0; i < lru_size; i += 1)
2169 {
2170 lru[i].blkno = (xoff_t) -1;
2171
2172 if ((lru[i].blk = main_malloc (source->blksize)) == NULL)
2173 {
2174 ret = ENOMEM;
2175 goto error;
2176 }
2177
2178 main_blklru_list_push_back (& lru_free, & lru[i]);
2179 }
2180
2181 if (stream && (ret = xd3_set_source (stream, source)))
2182 {
2183 XPR(NT XD3_LIB_ERRMSG (stream, ret));
2184 goto error;
2185 }
2186
2187 error:
2188 if (tmp_buf != NULL)
2189 {
2190 main_free (tmp_buf);
2191 }
2192
2193 return ret;
2194}
2195
2196static void
2197main_set_winsize (main_file *ifile) {
2198 xoff_t file_size;
2199
2200 if (main_file_stat (ifile, &file_size, 0) == 0)
2201 {
2202 option_winsize = (usize_t) min(file_size, (xoff_t) option_winsize);
2203 }
2204
2205 option_winsize = max(option_winsize, XD3_ALLOCSIZE);
2206
2207 if (option_verbose > 1)
2208 {
2209 XPR(NT "input window size: %u\n", option_winsize);
2210 }
2211}
2212
2213/******************************************************************************************
2214 Source routines
2215 ******************************************************************************************/
2216
2217/* This is the callback for reading a block of source. This function is blocking and it
2218 * implements a small LRU.
2219 *
2220 * Note that it is possible for main_input() to handle getblk requests in a non-blocking
2221 * manner. If the callback is NULL then the caller of xd3_*_input() must handle the
2222 * XD3_GETSRCBLK return value and fill the source in the same way. See xd3_getblk for
2223 * details. To see an example of non-blocking getblk, see xdelta-test.h. */
2224static int
2225main_getblk_func (xd3_stream *stream,
2226 xd3_source *source,
2227 xoff_t blkno)
2228{
2229 xoff_t pos = blkno * source->blksize;
2230 main_file *sfile = (main_file*) source->ioh;
2231 main_blklru *blru = NULL;
2232 usize_t onblk = xd3_bytes_on_srcblk (source, blkno);
2233 usize_t nread;
2234 int ret;
2235 int i;
2236
2237 if (allow_fake_source)
2238 {
2239 source->curblkno = blkno;
2240 source->onblk = onblk;
2241 source->curblk = lru[0].blk;
2242 return 0;
2243 }
2244
2245 if (do_not_lru)
2246 {
2247 /* Direct lookup assumes sequential scan w/o skipping blocks. */
2248 int idx = blkno % lru_size;
2249 if (lru[idx].blkno == blkno)
2250 {
2251 source->curblkno = blkno;
2252 source->onblk = onblk;
2253 source->curblk = lru[idx].blk;
2254 lru_hits += 1;
2255 return 0;
2256 }
2257
2258 if (lru[idx].blkno != -1LL &&
2259 lru[idx].blkno != blkno - lru_size)
2260 {
2261 return XD3_TOOFARBACK;
2262 }
2263 }
2264 else
2265 {
2266 /* Sequential search through LRU. */
2267 for (i = 0; i < lru_size; i += 1)
2268 {
2269 if (lru[i].blkno == blkno)
2270 {
2271 main_blklru_list_remove (& lru[i]);
2272 main_blklru_list_push_back (& lru_list, & lru[i]);
2273
2274 source->curblkno = blkno;
2275 source->onblk = onblk;
2276 source->curblk = lru[i].blk;
2277 lru_hits += 1;
2278 return 0;
2279 }
2280 }
2281 }
2282
2283 if (! main_blklru_list_empty (& lru_free))
2284 {
2285 blru = main_blklru_list_pop_front (& lru_free);
2286 }
2287 else if (! main_blklru_list_empty (& lru_list))
2288 {
2289 if (do_not_lru) {
2290 blru = & lru[blkno % lru_size];
2291 main_blklru_list_remove(blru);
2292 } else {
2293 blru = main_blklru_list_pop_front (& lru_list);
2294 }
2295 lru_misses += 1;
2296 }
2297
2298 lru_filled += 1;
2299
2300 if ((ret = main_file_seek (sfile, pos)))
2301 {
2302 return ret;
2303 }
2304
2305 if ((ret = main_file_read (sfile, (uint8_t*) blru->blk, source->blksize,
2306 & nread, "source read failed")))
2307 {
2308 return ret;
2309 }
2310
2311 if (nread != onblk)
2312 {
2313 XPR(NT "source file size change: %s\n", sfile->filename);
2314 return XD3_INTERNAL;
2315 }
2316
2317 main_blklru_list_push_back (& lru_list, blru);
2318
2319 if (option_verbose > 3)
2320 {
2321 if (blru->blkno != -1LL)
2322 {
2323 XPR(NT "source block %"Q"u ejects %"Q"u (lru_hits=%u, lru_misses=%u, lru_filled=%u)\n",
2324 blkno, blru->blkno, lru_hits, lru_misses, lru_filled);
2325 }
2326 else
2327 {
2328 XPR(NT "source block %"Q"u read (lru_hits=%u, lru_misses=%u, lru_filled=%u)\n",
2329 blkno, lru_hits, lru_misses, lru_filled);
2330 }
2331 }
2332
2333 blru->blkno = blkno;
2334 source->curblk = blru->blk;
2335 source->curblkno = blkno;
2336 source->onblk = onblk;
2337
2338 return 0;
2339}
2340
2341/******************************************************************************************
2342 Main routines
2343 ******************************************************************************************/
2344
2345/* This is a generic input function. It calls the xd3_encode_input or xd3_decode_input
2346 * functions and makes calls to the various input handling routines above, which
2347 * coordinate external decompression.
2348 */
2349static int
2350main_input (xd3_cmd cmd,
2351 main_file *ifile,
2352 main_file *ofile,
2353 main_file *sfile)
2354{
2355 int ret;
2356 xd3_stream stream;
2357 usize_t nread;
2358 int stream_flags = 0;
2359 xd3_config config;
2360 xd3_source source;
2361 xoff_t last_total_in = 0;
2362 xoff_t last_total_out = 0;
2363 long start_time;
2364 int stdout_only = 0;
2365
2366 int (*input_func) (xd3_stream*);
2367 int (*output_func) (xd3_stream*, main_file *);
2368
2369 memset (& source, 0, sizeof (source));
2370 memset (& config, 0, sizeof (config));
2371
2372 config.alloc = main_alloc;
2373 config.freef = main_free1;
2374 config.sec_data.ngroups = 1;
2375 config.sec_addr.ngroups = 1;
2376 config.sec_inst.ngroups = 1;
2377 config.iopt_size = option_iopt_size;
2378 config.sprevsz = option_sprevsz;
2379
2380 do_not_lru = 0;
2381
2382 start_time = get_millisecs_now ();
2383
2384 /* main_input setup. */
2385 switch ((int) cmd)
2386 {
2387#if VCDIFF_TOOLS
2388 if (1) { case CMD_PRINTHDR: stream_flags = XD3_JUST_HDR; }
2389 else if (1) { case CMD_PRINTHDRS: stream_flags = XD3_SKIP_WINDOW; }
2390 else { case CMD_PRINTDELTA: stream_flags = XD3_SKIP_EMIT; }
2391 ifile->flags |= RD_NONEXTERNAL;
2392 input_func = xd3_decode_input;
2393 output_func = main_print_func;
2394 stream_flags |= XD3_ADLER32_NOVER;
2395 stdout_only = 1;
2396 break;
2397#endif
2398#if XD3_ENCODER
2399 case CMD_ENCODE:
2400 do_not_lru = 1;
2401 input_func = xd3_encode_input;
2402 output_func = main_write_output;
2403
2404 if (option_use_checksum) { stream_flags |= XD3_ADLER32; }
2405 if (option_use_secondary)
2406 {
2407 /* The default secondary compressor is DJW, if it's compiled, being used, etc. */
2408 if (option_secondary == NULL)
2409 {
2410 if (SECONDARY_DJW) { stream_flags |= XD3_SEC_DJW; }
2411 }
2412 else
2413 {
2414 if (strcmp (option_secondary, "fgk") == 0 && SECONDARY_FGK)
2415 {
2416 stream_flags |= XD3_SEC_FGK;
2417 }
2418 else if (strcmp (option_secondary, "djw") == 0 && SECONDARY_DJW)
2419 {
2420 stream_flags |= XD3_SEC_DJW;
2421 }
2422 else
2423 {
2424 XPR(NT "unrecognized secondary compressor type: %s\n", option_secondary);
2425 return EXIT_FAILURE;
2426 }
2427 }
2428 }
2429 if (option_no_compress) { stream_flags |= XD3_NOCOMPRESS; }
2430 if (option_use_altcodetable) { stream_flags |= XD3_ALT_CODE_TABLE; }
2431 if (option_smatch_config)
2432 {
2433 char *s = option_smatch_config, *e;
2434 int values[XD3_SOFTCFG_VARCNT];
2435 int got;
2436
2437 config.smatch_cfg = XD3_SMATCH_SOFT;
2438
2439 for (got = 0; got < XD3_SOFTCFG_VARCNT; got += 1, s = e + 1)
2440 {
2441 values[got] = strtol (s, &e, 10);
2442
2443 if ((values[got] < 0) ||
2444 (e == s) ||
2445 (got < XD3_SOFTCFG_VARCNT-1 && *e == 0) ||
2446 (got == XD3_SOFTCFG_VARCNT-1 && *e != 0))
2447 {
2448 XPR(NT "invalid string match specifier (-C) %d: %s\n",
2449 got, s);
2450 return EXIT_FAILURE;
2451 }
2452 }
2453
2454 config.smatcher_soft.large_look = values[0];
2455 config.smatcher_soft.large_step = values[1];
2456 config.smatcher_soft.small_look = values[2];
2457 config.smatcher_soft.small_chain = values[3];
2458 config.smatcher_soft.small_lchain = values[4];
2459 config.smatcher_soft.max_lazy = values[5];
2460 config.smatcher_soft.long_enough = values[6];
2461 }
2462 else
2463 {
2464 if (option_verbose > 1)
2465 {
2466 XPR(NT "compression level: %d\n", option_level);
2467 }
2468 if (option_level == 0)
2469 {
2470 stream_flags |= XD3_NOCOMPRESS;
2471 config.smatch_cfg = XD3_SMATCH_FASTEST;
2472 }
2473 else if (option_level == 1) { config.smatch_cfg = XD3_SMATCH_FASTEST; }
2474 else if (option_level <= 5) { config.smatch_cfg = XD3_SMATCH_FAST; }
2475 else if (option_level == 6) { config.smatch_cfg = XD3_SMATCH_DEFAULT; }
2476 else { config.smatch_cfg = XD3_SMATCH_SLOW; }
2477 }
2478 break;
2479#endif
2480 case CMD_DECODE:
2481 if (option_use_checksum == 0) { stream_flags |= XD3_ADLER32_NOVER; }
2482 stream_flags = 0;
2483 ifile->flags |= RD_NONEXTERNAL;
2484 input_func = xd3_decode_input;
2485 output_func = main_write_output;
2486 break;
2487 default:
2488 XPR(NT "internal error\n");
2489 return EXIT_FAILURE;
2490 }
2491
2492 main_set_winsize (ifile);
2493
2494 if ((main_bdata = main_malloc (option_winsize)) == NULL)
2495 {
2496 return EXIT_FAILURE;
2497 }
2498
2499 if (IS_ENCODE (cmd))
2500 {
2501 /* When encoding, open the source file, possibly decompress it. The decoder delays
2502 * this step until XD3_GOTHEADER. */
2503 if (sfile->filename != NULL && (ret = main_set_source (NULL, cmd, sfile, & source)))
2504 {
2505 return EXIT_FAILURE;
2506 }
2507 }
2508
2509 config.winsize = option_winsize;
2510 config.srcwin_maxsz = option_srcwinsz;
2511 config.getblk = main_getblk_func;
2512 config.flags = stream_flags;
2513
2514 if ((ret = xd3_config_stream (& stream, & config)))
2515 {
2516 XPR(NT XD3_LIB_ERRMSG (& stream, ret));
2517 return EXIT_FAILURE;
2518 }
2519
2520 if (IS_ENCODE (cmd) && sfile->filename != NULL &&
2521 (ret = xd3_set_source (& stream, & source)))
2522 {
2523 XPR(NT XD3_LIB_ERRMSG (& stream, ret));
2524 return EXIT_FAILURE;
2525 }
2526
2527 /* This times each window. */
2528 get_millisecs_since ();
2529
2530 /* Main input loop. */
2531 do
2532 {
2533 xoff_t input_offset;
2534 xoff_t input_remain;
2535 usize_t try_read;
2536
2537 input_offset = ifile->nread;
2538
2539 input_remain = XOFF_T_MAX - input_offset;
2540
2541 try_read = (usize_t) min ((xoff_t) config.winsize, input_remain);
2542
2543 if ((ret = main_read_primary_input (ifile, main_bdata, try_read, & nread)))
2544 {
2545 return EXIT_FAILURE;
2546 }
2547
2548 /* If we've reached EOF tell the stream to flush. */
2549 if (nread < try_read)
2550 {
2551 stream_flags |= XD3_FLUSH;
2552 xd3_set_flags (& stream, stream_flags);
2553 }
2554
2555#if XD3_ENCODER
2556 /* After the first main_read_primary_input completes, we know all the information
2557 * needed to encode the application header. */
2558 if (cmd == CMD_ENCODE && (ret = main_set_appheader (& stream, ifile, sfile)))
2559 {
2560 return EXIT_FAILURE;
2561 }
2562#endif
2563 xd3_avail_input (& stream, main_bdata, nread);
2564
2565 /* If we read zero bytes after encoding at least one window... */
2566 if (nread == 0 && stream.current_window > 0) {
2567 break;
2568 }
2569
2570 again:
2571 ret = input_func (& stream);
2572
2573 switch (ret)
2574 {
2575 case XD3_INPUT:
2576 continue;
2577
2578 case XD3_GOTHEADER:
2579 {
2580 XD3_ASSERT (stream.current_window == 0);
2581
2582 /* Need to process the appheader as soon as possible. It may contain a
2583 * suggested default filename/decompression routine for the ofile, and it may
2584 * contain default/decompression routine for the sources. */
2585 if (cmd == CMD_DECODE)
2586 {
2587 int have_src = sfile->filename != NULL;
2588 int need_src = xd3_decoder_needs_source (& stream);
2589 int recv_src;
2590
2591 /* May need to set the sfile->filename if none was given. */
2592 main_get_appheader (& stream, ifile, ofile, sfile);
2593
2594 recv_src = sfile->filename != NULL;
2595
2596 /* Check if the user expected a source to be required although it was not. */
2597 if (have_src && ! need_src && ! option_quiet)
2598 {
2599 XPR(NT "warning: output window %"Q"u does not copy source\n", stream.current_window);
2600 }
2601
2602 /* Check if we have no source name and need one. */
2603 /* TODO: this doesn't fire due to cpyblocks_ calculation check? */
2604 if (need_src && ! recv_src)
2605 {
2606 XPR(NT "input requires a source file, use -s\n");
2607 return EXIT_FAILURE;
2608 }
2609
2610 /* Now open the source file. */
2611 if (need_src && (ret = main_set_source (& stream, cmd, sfile, & source)))
2612 {
2613 return EXIT_FAILURE;
2614 }
2615 }
2616 else if (cmd == CMD_PRINTHDR ||
2617 cmd == CMD_PRINTHDRS ||
2618 cmd == CMD_PRINTDELTA)
2619 {
2620 if (xd3_decoder_needs_source (& stream) && sfile->filename == NULL)
2621 {
2622 allow_fake_source = 1;
2623 sfile->filename = "<placeholder>";
2624 main_set_source (& stream, cmd, sfile, & source);
2625 }
2626 }
2627 }
2628 /* FALLTHROUGH */
2629 case XD3_WINSTART:
2630 {
2631 /* e.g., set or unset XD3_SKIP_WINDOW. */
2632 /* xd3_set_flags (& stream, stream_flags); */
2633 goto again;
2634 }
2635
2636 case XD3_OUTPUT:
2637 {
2638 if (option_no_output == 0)
2639 {
2640 /* Defer opening the output file until the stream produces its first
2641 * output for both encoder and decoder, this way we delay long enough for
2642 * the decoder to receive the application header. (Or longer if there are
2643 * skipped windows, but I can't think of any reason not to delay open.) */
2644
2645 if (! main_file_isopen (ofile) && (ret = main_open_output (& stream, ofile)) != 0)
2646 {
2647 return EXIT_FAILURE;
2648 }
2649 if ((ret = output_func (& stream, ofile)) && (ret != PRINTHDR_SPECIAL))
2650 {
2651 return EXIT_FAILURE;
2652 }
2653 if (ret == PRINTHDR_SPECIAL)
2654 {
2655 xd3_abort_stream (& stream);
2656 ret = EXIT_SUCCESS;
2657 goto done;
2658 }
2659 ret = 0;
2660 }
2661
2662 xd3_consume_output (& stream);
2663 goto again;
2664 }
2665
2666 case XD3_WINFINISH:
2667 {
2668 if (IS_ENCODE (cmd) || cmd == CMD_DECODE)
2669 {
2670 if (! option_quiet && IS_ENCODE (cmd) && main_file_isopen (sfile))
2671 {
2672 /* Warn when no source copies are found */
2673 if (! xd3_encoder_used_source (& stream))
2674 {
2675 XPR(NT "warning: input window %"Q"u..%"Q"u has no source copies\n",
2676 stream.current_window * option_winsize,
2677 (stream.current_window+1) * option_winsize);
2678 }
2679
2680 /* Limited instruction buffer size affects source copies */
2681 if (option_verbose > 1 && stream.i_slots_used > stream.iopt_size)
2682 {
2683 XPR(NT "warning: input position %"Q"u overflowed instruction buffer, "
2684 "needed %u (vs. %u), consider raising -I\n",
2685 stream.current_window * option_winsize,
2686 stream.i_slots_used, stream.iopt_size);
2687 }
2688 }
2689
2690 if (option_verbose)
2691 {
2692 char rrateavg[32], wrateavg[32], tm[32];
2693 char rdb[32], wdb[32];
2694 char trdb[32], twdb[32];
2695 long millis = get_millisecs_since ();
2696 usize_t this_read = (usize_t)(stream.total_in - last_total_in);
2697 usize_t this_write = (usize_t)(stream.total_out - last_total_out);
2698 last_total_in = stream.total_in;
2699 last_total_out = stream.total_out;
2700
2701 if (option_verbose > 1)
2702 {
2703 XPR(NT "%"Q"u: in %s (%s): out %s (%s): total in %s: out %s: %s\n",
2704 stream.current_window,
2705 main_format_bcnt (this_read, rdb),
2706 main_format_rate (this_read, millis, rrateavg),
2707 main_format_bcnt (this_write, wdb),
2708 main_format_rate (this_write, millis, wrateavg),
2709 main_format_bcnt (stream.total_in, trdb),
2710 main_format_bcnt (stream.total_out, twdb),
2711 main_format_millis (millis, tm));
2712 }
2713 else
2714 {
2715 XPR(NT "%"Q"u: in %s: out %s: total in %s: out %s: %s\n",
2716 stream.current_window,
2717 main_format_bcnt (this_read, rdb),
2718 main_format_bcnt (this_write, wdb),
2719 main_format_bcnt (stream.total_in, trdb),
2720 main_format_bcnt (stream.total_out, twdb),
2721 main_format_millis (millis, tm));
2722 }
2723 }
2724 }
2725 goto again;
2726 }
2727
2728 default:
2729 /* input_func() error */
2730 XPR(NT XD3_LIB_ERRMSG (& stream, ret));
2731 return EXIT_FAILURE;
2732 }
2733 }
2734 while (nread == config.winsize);
2735done:
2736 /* Close the inputs. (ifile must be open, sfile may be open) */
2737 main_file_close (ifile);
2738 main_file_close (sfile);
2739
2740 /* If output file is not open yet because of delayed-open, it means we never encountered
2741 * a window in the delta, but it could have had a VCDIFF header? TODO: solve this
2742 * elsewhere. For now, it prints "nothing to output" below, but the check doesn't
2743 * happen in case of option_no_output. */
2744 if (! option_no_output)
2745 {
2746 if (!stdout_only && ! main_file_isopen (ofile))
2747 {
2748 XPR(NT "nothing to output: %s\n", ifile->filename);
2749 return EXIT_FAILURE;
2750 }
2751
2752 /* Have to close the output before calling main_external_compression_finish, or else it hangs. */
2753 if (main_file_close (ofile) != 0)
2754 {
2755 return EXIT_FAILURE;
2756 }
2757 }
2758
2759#if EXTERNAL_COMPRESSION
2760 if ((ret = main_external_compression_finish ()))
2761 {
2762 XPR(NT "external compression commands failed\n");
2763 return EXIT_FAILURE;
2764 }
2765#endif
2766
2767 if ((ret = xd3_close_stream (& stream)))
2768 {
2769 XPR(NT XD3_LIB_ERRMSG (& stream, ret));
2770 return EXIT_FAILURE;
2771 }
2772
2773 if (option_verbose > 1)
2774 {
2775 XPR(NT "scanner configuration: %s\n", stream.smatcher.name);
2776 XPR(NT "target hash table size: %u\n", stream.small_hash.size);
2777 if (sfile->filename != NULL)
2778 {
2779 XPR(NT "source hash table size: %u\n", stream.large_hash.size);
2780 }
2781 }
2782
2783 if (option_verbose > 2)
2784 {
2785 XPR(NT "source copies: %"Q"u (%"Q"u bytes)\n", stream.n_scpy, stream.l_scpy);
2786 XPR(NT "target copies: %"Q"u (%"Q"u bytes)\n", stream.n_tcpy, stream.l_tcpy);
2787 XPR(NT "adds: %"Q"u (%"Q"u bytes)\n", stream.n_add, stream.l_add);
2788 XPR(NT "runs: %"Q"u (%"Q"u bytes)\n", stream.n_run, stream.l_run);
2789 }
2790
2791 xd3_free_stream (& stream);
2792
2793 if (option_verbose)
2794 {
2795 char tm[32];
2796 long end_time = get_millisecs_now ();
2797 XPR(NT "finished in %s; input %"Q"u output %"Q"u bytes (%0.2f%%)\n",
2798 main_format_millis (end_time - start_time, tm),
2799 ifile->nread, ofile->nwrite,
2800 100.0 * ofile->nwrite / ifile->nread);
2801 }
2802
2803 return EXIT_SUCCESS;
2804}
2805
2806/* free memory before exit, reset single-use variables. */
2807static void
2808main_cleanup (void)
2809{
2810 int i;
2811
2812 if (appheader_used != NULL &&
2813 appheader_used != option_appheader)
2814 {
2815 main_free (appheader_used);
2816 appheader_used = NULL;
2817 }
2818
2819 main_free (main_bdata);
2820 main_bdata = NULL;
2821
2822#if EXTERNAL_COMPRESSION
2823 main_free (ext_tmpfile);
2824 ext_tmpfile = NULL;
2825#endif
2826
2827 for (i = 0; lru && i < lru_size; i += 1)
2828 {
2829 main_free (lru[i].blk);
2830 }
2831
2832 main_free (lru);
2833 lru = NULL;
2834
2835 lru_hits = 0;
2836 lru_misses = 0;
2837 lru_filled = 0;
2838
2839 XD3_ASSERT (main_mallocs == 0);
2840}
2841
2842static void
2843setup_environment (int argc,
2844 char **argv,
2845 int *argc_out,
2846 char ***argv_out,
2847 char ***argv_free,
2848 char **env_free)
2849{
2850 int n, i, i0;
2851 char *p, *v = getenv("XDELTA");
2852 if (v == NULL) {
2853 (*argc_out) = argc;
2854 (*argv_out) = argv;
2855 (*argv_free) = NULL;
2856 (*env_free) = NULL;
2857 return;
2858 }
2859
2860 (*env_free) = main_malloc(strlen(v) + 1);
2861 strcpy(*env_free, v);
2862
2863 /* Space needed for extra args, at least # of spaces */
2864 n = argc + 1;
2865 for (p = *env_free; *p != 0; ) {
2866 if (*p++ == ' ') {
2867 n++;
2868 }
2869 }
2870
2871 (*argv_free) = main_malloc(sizeof(char*) * (n + 1));
2872 (*argv_out) = (*argv_free);
2873 (*argv_out)[0] = argv[0];
2874 (*argv_out)[n] = NULL;
2875
2876 i = 1;
2877 for (p = *env_free; *p != 0; ) {
2878 (*argv_out)[i++] = p;
2879 while (*p != ' ' && *p != 0) {
2880 p++;
2881 }
2882 while (*p == ' ') {
2883 *p++ = 0;
2884 }
2885 }
2886
2887 for (i0 = 1; i0 < argc; i0++) {
2888 (*argv_out)[i++] = argv[i0];
2889 }
2890
2891 /* Counting spaces is an upper bound, argv stays NULL terminated. */
2892 (*argc_out) = i;
2893 while (i <= n) {
2894 (*argv_out)[i++] = NULL;
2895 }
2896}
2897
2898int
2899#if PYTHON_MODULE || SWIG_MODULE
2900xd3_main_cmdline (int argc, char **argv)
2901#else
2902main (int argc, char **argv)
2903#endif
2904{
2905 xd3_cmd cmd;
2906 main_file ifile;
2907 main_file ofile;
2908 main_file sfile;
2909 static const char *flags = "0123456789cdefhnqvDJNORTVs:B:C:E:F:I:L:O:M:P:W:A::S::";
2910 int my_optind;
2911 char *my_optarg;
2912 char *my_optstr;
2913 char *sfilename;
2914 int env_argc;
2915 char **env_argv;
2916 char **free_argv; /* malloc() in setup_environment() */
2917 char *free_value; /* malloc() in setup_environment() */
2918 int ret;
2919
2920#ifdef _WIN32
2921 GetStartupInfo(&winStartupInfo);
2922 setvbuf(stderr, NULL, _IONBF, 0); /* Do not buffer stderr */
2923#endif
2924
2925 main_file_init (& ifile);
2926 main_file_init (& ofile);
2927 main_file_init (& sfile);
2928
2929 reset_defaults();
2930
2931 free_argv = NULL;
2932 free_value = NULL;
2933 setup_environment(argc, argv, &env_argc, &env_argv, &free_argv, &free_value);
2934 cmd = CMD_NONE;
2935 sfilename = NULL;
2936 my_optind = 1;
2937 argv = env_argv;
2938 argc = env_argc;
2939 program_name = env_argv[0];
2940 extcomp_types[0].recomp_cmdname = program_name;
2941 extcomp_types[0].decomp_cmdname = program_name;
2942 takearg:
2943 my_optarg = NULL;
2944 my_optstr = argv[my_optind];
2945 /* This doesn't use getopt() because it makes trouble for -P & python which reenter
2946 * main() and thus care about freeing all memory. I never had much trust for getopt
2947 * anyway, it's too opaque. This implements a fairly standard non-long-option getopt
2948 * with support for named operations (e.g., "xdelta3 [encode|decode|printhdr...] < in >
2949 * out"). */
2950 if (my_optstr)
2951 {
2952 if (*my_optstr == '-') { my_optstr += 1; }
2953 else if (cmd == CMD_NONE) { goto nonflag; }
2954 else { my_optstr = NULL; }
2955 }
2956 while (my_optstr)
2957 {
2958 char *s;
2959 my_optarg = NULL;
2960 if ((ret = *my_optstr++) == 0) { my_optind += 1; goto takearg; }
2961
2962 /* Option handling: first check for one ':' following the option in flags, then
2963 * check for two. The syntax allows:
2964 *
2965 * 1. -Afoo defines optarg="foo"
2966 * 2. -A foo defines optarg="foo"
2967 * 3. -A "" defines optarg="" (allows optional empty-string)
2968 * 4. -A [EOA or -moreargs] error (mandatory case)
2969 * 5. -A [EOA -moreargs] defines optarg=NULL (optional case)
2970 * 6. -A=foo defines optarg="foo"
2971 * 7. -A= defines optarg="" (mandatory case)
2972 * 8. -A= defines optarg=NULL (optional case)
2973 *
2974 * See tests in test_command_line_arguments().
2975 */
2976 s = strchr (flags, ret);
2977 if (s && s[1] && s[1] == ':')
2978 {
2979 int eqcase = 0;
2980 int option = s[2] && s[2] == ':';
2981
2982 /* Case 1, set optarg to the remaining characters. */
2983 my_optarg = my_optstr;
2984 my_optstr = "";
2985
2986 /* Case 2-5 */
2987 if (*my_optarg == 0)
2988 {
2989 /* Condition 4-5 */
2990 int have_arg = my_optind < (argc - 1) && *argv[my_optind+1] != '-';
2991
2992 if (! have_arg)
2993 {
2994 if (! option)
2995 {
2996 /* Case 4 */
2997 XPR(NT "-%c: requires an argument\n", ret);
2998 ret = EXIT_FAILURE;
2999 goto cleanup;
3000 }
3001 /* Case 5. */
3002 my_optarg = NULL;
3003 }
3004 else
3005 {
3006 /* Case 2-3. */
3007 my_optarg = argv[++my_optind];
3008 }
3009 }
3010 /* Case 6-8. */
3011 else if (*my_optarg == '=')
3012 {
3013 /* Remove the = in all cases. */
3014 my_optarg += 1;
3015 eqcase = 1;
3016
3017 if (option && *my_optarg == 0)
3018 {
3019 /* Case 8. */
3020 my_optarg = NULL;
3021 }
3022 }
3023 }
3024
3025 switch (ret)
3026 {
3027 /* case: if no '-' was found, maybe check for a command name. */
3028 nonflag:
3029 if (strcmp (my_optstr, "decode") == 0) { cmd = CMD_DECODE; }
3030 else if (strcmp (my_optstr, "encode") == 0)
3031 {
3032#if XD3_ENCODER
3033 cmd = CMD_ENCODE;
3034#else
3035 XPR(NT "encoder support not compiled\n");
3036 return EXIT_FAILURE;
3037#endif
3038 }
3039 else if (strcmp (my_optstr, "config") == 0) { cmd = CMD_CONFIG; }
3040#if REGRESSION_TEST
3041 else if (strcmp (my_optstr, "test") == 0) { cmd = CMD_TEST; }
3042#endif
3043#if VCDIFF_TOOLS
3044 else if (strcmp (my_optstr, "printhdr") == 0) { cmd = CMD_PRINTHDR; }
3045 else if (strcmp (my_optstr, "printhdrs") == 0) { cmd = CMD_PRINTHDRS; }
3046 else if (strcmp (my_optstr, "printdelta") == 0) { cmd = CMD_PRINTDELTA; }
3047#endif
3048
3049 /* If no option was found and still no command, let the default command be
3050 * encode. The remaining args are treated as filenames. */
3051 if (cmd == CMD_NONE)
3052 {
3053 cmd = CMD_DEFAULT;
3054 my_optstr = NULL;
3055 break;
3056 }
3057 else
3058 {
3059 /* But if we find a command name, continue the getopt loop. */
3060 my_optind += 1;
3061 goto takearg;
3062 }
3063
3064 /* gzip-like options */
3065 case '0': case '1': case '2': case '3': case '4':
3066 case '5': case '6': case '7': case '8': case '9':
3067 option_level = ret - '0';
3068 break;
3069 case 'f': option_force = 1; break;
3070 case 'v': option_verbose += 1; option_quiet = 0; break;
3071 case 'q': option_quiet = 1; option_verbose = 0; break;
3072 case 'c': option_stdout = 1; break;
3073 case 'd':
3074 if (cmd == CMD_NONE) { cmd = CMD_DECODE; }
3075 else { ret = main_help (); goto exit; }
3076 break;
3077 case 'e':
3078#if XD3_ENCODER
3079 if (cmd == CMD_NONE) { cmd = CMD_ENCODE; }
3080 else { ret = main_help (); goto exit; }
3081 break;
3082#else
3083 XPR(NT "encoder support not compiled\n");
3084 return EXIT_FAILURE;
3085#endif
3086
3087 case 'n': option_use_checksum = 0; break;
3088 case 'N': option_no_compress = 1; break;
3089 case 'T': option_use_altcodetable = 1; break;
3090 case 'C': option_smatch_config = my_optarg; break;
3091 case 'J': option_no_output = 1; break;
3092 case 'S': if (my_optarg == NULL) { option_use_secondary = 0; }
3093 else { option_use_secondary = 1; option_secondary = my_optarg; } break;
3094 case 'A': if (my_optarg == NULL) { option_use_appheader = 0; }
3095 else { option_appheader = (uint8_t*) my_optarg; } break;
3096 case 'B':
3097 if ((ret = main_atou (my_optarg, & option_srcwinsz, XD3_MINSRCWINSZ,
3098 0, 'B')))
3099 {
3100 goto exit;
3101 }
3102 break;
3103 case 'I':
3104 if ((ret = main_atou (my_optarg, & option_iopt_size, 0,
3105 0, 'I')))
3106 {
3107 goto exit;
3108 }
3109 break;
3110 case 'P':
3111 if ((ret = main_atou (my_optarg, & option_sprevsz, 0,
3112 0, 'P')))
3113 {
3114 goto exit;
3115 }
3116 break;
3117 case 'W':
3118 if ((ret = main_atou (my_optarg, & option_winsize, XD3_ALLOCSIZE,
3119 XD3_HARDMAXWINSIZE, 'W')))
3120 {
3121 goto exit;
3122 }
3123 break;
3124 case 'D':
3125#if EXTERNAL_COMPRESSION == 0
3126 if (option_verbose > 0)
3127 {
3128 XPR(NT "warning: -D option ignored, "
3129 "external compression support was not compiled\n");
3130 }
3131#else
3132 option_decompress_inputs = 0;
3133#endif
3134 break;
3135 case 'R':
3136#if EXTERNAL_COMPRESSION == 0
3137 if (option_verbose > 0)
3138 {
3139 XPR(NT "warning: -R option ignored, "
3140 "external compression support was not compiled\n");
3141 }
3142#else
3143 option_recompress_outputs = 0;
3144#endif
3145 break;
3146 case 's':
3147 if (sfilename != NULL)
3148 {
3149 XPR(NT "specify only one source file\n");
3150 goto cleanup;
3151 }
3152
3153 sfilename = my_optarg;
3154 break;
3155
3156 case 'V':
3157 ret = main_version (); goto exit;
3158 default:
3159 ret = main_help (); goto exit;
3160 }
3161 }
3162
3163 option_source_filename = sfilename;
3164
3165 /* In case there were no arguments, set the default command. */
3166 if (cmd == CMD_NONE) { cmd = CMD_DEFAULT; }
3167
3168 argc -= my_optind;
3169 argv += my_optind;
3170
3171 /* There may be up to two more arguments. */
3172 if (argc > 2)
3173 {
3174 XPR(NT "too many filenames: %s ...\n", argv[2]);
3175 ret = EXIT_FAILURE;
3176 goto cleanup;
3177 }
3178
3179 ifile.flags = RD_FIRST;
3180 sfile.flags = RD_FIRST;
3181 sfile.filename = option_source_filename;
3182
3183 /* The infile takes the next argument, if there is one. But if not, infile is set to
3184 * stdin. */
3185 if (argc > 0)
3186 {
3187 ifile.filename = argv[0];
3188
3189 if ((ret = main_file_open (& ifile, ifile.filename, XO_READ)))
3190 {
3191 goto cleanup;
3192 }
3193 }
3194 else
3195 {
3196 XSTDIN_XF (& ifile);
3197 }
3198
3199 /* The ofile takes the following argument, if there is one. But if not, it is left NULL
3200 * until the application header is processed. It will be set in main_open_output. */
3201 if (argc > 1)
3202 {
3203 /* Check for conflicting arguments. */
3204 if (option_stdout && ! option_quiet)
3205 {
3206 XPR(NT "warning: -c option overrides output filename: %s\n", argv[1]);
3207 }
3208
3209 if (! option_stdout) { ofile.filename = argv[1]; }
3210 }
3211
3212 switch (cmd)
3213 {
3214 case CMD_PRINTHDR:
3215 case CMD_PRINTHDRS:
3216 case CMD_PRINTDELTA:
3217#if XD3_ENCODER
3218 case CMD_ENCODE:
3219#endif
3220 case CMD_DECODE:
3221 ret = main_input (cmd, & ifile, & ofile, & sfile);
3222 break;
3223
3224#if REGRESSION_TEST
3225 case CMD_TEST:
3226 ret = xd3_selftest ();
3227 break;
3228#endif
3229
3230 case CMD_CONFIG:
3231 ret = main_config ();
3232 break;
3233
3234 default:
3235 ret = main_help ();
3236 break;
3237 }
3238
3239#if EXTERNAL_COMPRESSION
3240 if (ext_tmpfile != NULL)
3241 {
3242 unlink (ext_tmpfile);
3243 }
3244#endif
3245
3246 if (0)
3247 {
3248 cleanup:
3249 ret = EXIT_FAILURE;
3250 exit:
3251 (void)0;
3252 }
3253
3254 main_file_cleanup (& ifile);
3255 main_file_cleanup (& ofile);
3256 main_file_cleanup (& sfile);
3257
3258 main_free (free_argv);
3259 main_free (free_value);
3260
3261 main_cleanup ();
3262
3263 fflush (stdout);
3264 fflush (stderr);
3265 return ret;
3266}
3267
3268static int
3269main_help (void)
3270{
3271 /* Note: update wiki when command-line features change */
3272 main_version ();
3273 DP(RINT "usage: xdelta3 [command/options] [input [output]]\n");
3274 DP(RINT "special command names:\n");
3275 DP(RINT " config prints xdelta3 configuration\n");
3276 DP(RINT " decode decompress the input\n");
3277 DP(RINT " encode compress the input%s\n", XD3_ENCODER ? "" : " [Not compiled]");
3278#if REGRESSION_TEST
3279 DP(RINT " test run the builtin tests\n");
3280#endif
3281#if VCDIFF_TOOLS
3282 DP(RINT "special commands for VCDIFF inputs:\n");
3283 DP(RINT " printdelta print information about the entire delta\n");
3284 DP(RINT " printhdr print information about the first window\n");
3285 DP(RINT " printhdrs print information about all windows\n");
3286#endif
3287 DP(RINT "standard options:\n");
3288 DP(RINT " -0 .. -9 compression level\n");
3289 DP(RINT " -c use stdout\n");
3290 DP(RINT " -d decompress\n");
3291 DP(RINT " -e compress%s\n", XD3_ENCODER ? "" : " [Not compiled]");
3292 DP(RINT " -f force overwrite\n");
3293 DP(RINT " -h show help\n");
3294 DP(RINT " -q be quiet\n");
3295 DP(RINT " -v be verbose (max 2)\n");
3296 DP(RINT " -V show version\n");
3297
3298 DP(RINT "memory options:\n");
3299 DP(RINT " -B bytes source window size\n");
3300 DP(RINT " -W bytes input window size\n");
3301 DP(RINT " -P size compression duplicates window\n");
3302 DP(RINT " -I size instruction buffer size (0 = unlimited)\n");
3303
3304 DP(RINT "compression options:\n");
3305 DP(RINT " -s source source file to copy from (if any)\n");
3306 DP(RINT " -S [djw|fgk] enable/disable secondary compression\n");
3307 DP(RINT " -N disable small string-matching compression\n");
3308 DP(RINT " -D disable external decompression (encode/decode)\n");
3309 DP(RINT " -R disable external recompression (decode)\n");
3310 DP(RINT " -n disable checksum (encode/decode)\n");
3311 DP(RINT " -C soft config (encode, undocumented)\n");
3312 DP(RINT " -A [apphead] disable/provide application header (encode)\n");
3313
3314#if XD3_DEBUG > 0
3315 DP(RINT "developer options:\n");
3316 DP(RINT " -J disable output (check/compute only)\n");
3317 DP(RINT " -P repeat count (for profiling)\n");
3318 DP(RINT " -T use alternate code table\n");
3319#endif
3320
3321 DP(RINT "the XDELTA environment variable may contain extra args:\n");
3322 DP(RINT " XDELTA=\"-s source-x.y.tar.gz\" \\\n");
3323 DP(RINT " tar --use-compress-program=xdelta3 \\\n");
3324 DP(RINT " -cf target-x.z.tar.gz.vcdiff target-x.y/\n");
3325 return EXIT_FAILURE;
3326}
3327
Note: See TracBrowser for help on using the repository browser.