1 /* Various utility functions.
2 Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
3 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
5 This file is part of GNU Wget.
7 GNU Wget is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3 of the License, or
10 (at your option) any later version.
12 GNU Wget is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with Wget. If not, see <http://www.gnu.org/licenses/>.
20 Additional permission under GNU GPL version 3 section 7
22 If you modify this program, or any covered work, by linking or
23 combining it with the OpenSSL project's OpenSSL library (or a
24 modified version of that library), containing parts covered by the
25 terms of the OpenSSL or SSLeay licenses, the Free Software Foundation
26 grants you additional permission to convey the resulting work.
27 Corresponding Source for a non-source form of such a combination
28 shall include the source code for the parts of OpenSSL used as well
29 as that of the covered work. */
31 #define USE_GNULIB_ALLOC
39 #ifdef HAVE_SYS_TIME_H
40 # include <sys/time.h>
46 # include <sys/mman.h>
49 # include <process.h> /* getpid() */
54 #ifdef HAVE_SYS_UTIME_H
55 # include <sys/utime.h>
63 /* For TIOCGWINSZ and friends: */
64 #ifdef HAVE_SYS_IOCTL_H
65 # include <sys/ioctl.h>
71 /* Needed for Unix version of run_with_timeout. */
75 #ifndef HAVE_SIGSETJMP
76 /* If sigsetjmp is a macro, configure won't pick it up. */
78 # define HAVE_SIGSETJMP
82 #if defined HAVE_SIGSETJMP || defined HAVE_SIGBLOCK
83 # define USE_SIGNAL_TIMEOUT
94 memfatal (const char *context, long attempted_size)
96 /* Make sure we don't try to store part of the log line, and thus
98 log_set_save_context (false);
100 /* We have different log outputs in different situations:
101 1) output without bytes information
102 2) output with bytes information */
103 if (attempted_size == UNKNOWN_ATTEMPTED_SIZE)
105 logprintf (LOG_ALWAYS,
106 _("%s: %s: Failed to allocate enough memory; memory exhausted.\n"),
111 logprintf (LOG_ALWAYS,
112 _("%s: %s: Failed to allocate %ld bytes; memory exhausted.\n"),
113 exec_name, context, attempted_size);
119 /* Utility function: like xstrdup(), but also lowercases S. */
122 xstrdup_lower (const char *s)
124 char *copy = xstrdup (s);
131 /* Copy the string formed by two pointers (one on the beginning, other
132 on the char after the last char) to a new, malloc-ed location.
135 strdupdelim (const char *beg, const char *end)
137 char *res = xmalloc (end - beg + 1);
138 memcpy (res, beg, end - beg);
139 res[end - beg] = '\0';
143 /* Parse a string containing comma-separated elements, and return a
144 vector of char pointers with the elements. Spaces following the
145 commas are ignored. */
147 sepstring (const char *s)
161 res = xrealloc (res, (i + 2) * sizeof (char *));
162 res[i] = strdupdelim (p, s);
165 /* Skip the blanks following the ','. */
166 while (c_isspace (*s))
173 res = xrealloc (res, (i + 2) * sizeof (char *));
174 res[i] = strdupdelim (p, s);
179 /* Like sprintf, but prints into a string of sufficient size freshly
180 allocated with malloc, which is returned. If unable to print due
181 to invalid format, returns NULL. Inability to allocate needed
182 memory results in abort, as with xmalloc. This is in spirit
183 similar to the GNU/BSD extension asprintf, but somewhat easier to
186 Internally the function either calls vasprintf or loops around
187 vsnprintf until the correct size is found. Since Wget also ships a
188 fallback implementation of vsnprintf, this should be portable. */
190 /* Constant is using for limits memory allocation for text buffer.
191 Applicable in situation when: vasprintf is not available in the system
192 and vsnprintf return -1 when long line is truncated (in old versions of
193 glibc and in other system where C99 doesn`t support) */
195 #define FMT_MAX_LENGTH 1048576
198 aprintf (const char *fmt, ...)
200 #if defined HAVE_VASPRINTF && !defined DEBUG_MALLOC
205 va_start (args, fmt);
206 ret = vasprintf (&str, fmt, args);
208 if (ret < 0 && errno == ENOMEM)
209 memfatal ("aprintf", UNKNOWN_ATTEMPTED_SIZE); /* for consistency
210 with xmalloc/xrealloc */
214 #else /* not HAVE_VASPRINTF */
216 /* vasprintf is unavailable. snprintf into a small buffer and
217 resize it as necessary. */
219 char *str = xmalloc (size);
221 /* #### This code will infloop and eventually abort in xrealloc if
222 passed a FMT that causes snprintf to consistently return -1. */
229 va_start (args, fmt);
230 n = vsnprintf (str, size, fmt, args);
233 /* If the printing worked, return the string. */
234 if (n > -1 && n < size)
237 /* Else try again with a larger buffer. */
238 if (n > -1) /* C99 */
239 size = n + 1; /* precisely what is needed */
240 else if (size >= FMT_MAX_LENGTH) /* We have a huge buffer, */
241 { /* maybe we have some wrong
243 logprintf (LOG_ALWAYS,
244 _("%s: aprintf: text buffer is too big (%ld bytes), "
246 exec_name, size); /* printout a log message */
247 abort (); /* and abort... */
251 /* else, we continue to grow our
252 * buffer: Twice the old size. */
255 str = xrealloc (str, size);
257 #endif /* not HAVE_VASPRINTF */
260 /* Concatenate the NULL-terminated list of string arguments into
261 freshly allocated space. */
264 concat_strings (const char *str0, ...)
267 int saved_lengths[5]; /* inspired by Apache's apr_pstrcat */
270 const char *next_str;
271 int total_length = 0;
274 /* Calculate the length of and allocate the resulting string. */
277 va_start (args, str0);
278 for (next_str = str0; next_str != NULL; next_str = va_arg (args, char *))
280 int len = strlen (next_str);
281 if (argcount < countof (saved_lengths))
282 saved_lengths[argcount++] = len;
286 p = ret = xmalloc (total_length + 1);
288 /* Copy the strings into the allocated space. */
291 va_start (args, str0);
292 for (next_str = str0; next_str != NULL; next_str = va_arg (args, char *))
295 if (argcount < countof (saved_lengths))
296 len = saved_lengths[argcount++];
298 len = strlen (next_str);
299 memcpy (p, next_str, len);
308 /* Format the provided time according to the specified format. The
309 format is a string with format elements supported by strftime. */
312 fmttime (time_t t, const char *fmt)
314 static char output[32];
315 struct tm *tm = localtime(&t);
318 if (!strftime(output, sizeof(output), fmt, tm))
323 /* Return pointer to a static char[] buffer in which zero-terminated
324 string-representation of TM (in form hh:mm:ss) is printed.
326 If TM is NULL, the current time will be used. */
331 return fmttime(t, "%H:%M:%S");
334 /* Like the above, but include the date: YYYY-MM-DD hh:mm:ss. */
337 datetime_str (time_t t)
339 return fmttime(t, "%Y-%m-%d %H:%M:%S");
342 /* The Windows versions of the following two functions are defined in
343 mswindows.c. On MSDOS this function should never be called. */
345 #if !defined(WINDOWS) && !defined(MSDOS)
347 fork_to_background (void)
350 /* Whether we arrange our own version of opt.lfilename here. */
351 bool logfile_changed = false;
353 if (!opt.lfilename && (!opt.quiet || opt.server_response))
355 /* We must create the file immediately to avoid either a race
356 condition (which arises from using unique_name and failing to
357 use fopen_excl) or lying to the user about the log file name
358 (which arises from using unique_name, printing the name, and
359 using fopen_excl later on.) */
360 FILE *new_log_fp = unique_create (DEFAULT_LOGFILE, false, &opt.lfilename);
363 logfile_changed = true;
376 /* parent, no error */
377 printf (_("Continuing in background, pid %d.\n"), (int) pid);
379 printf (_("Output will be written to %s.\n"), quote (opt.lfilename));
380 exit (0); /* #### should we use _exit()? */
383 /* child: give up the privileges and keep running. */
385 freopen ("/dev/null", "r", stdin);
386 freopen ("/dev/null", "w", stdout);
387 freopen ("/dev/null", "w", stderr);
389 #endif /* !WINDOWS && !MSDOS */
391 /* "Touch" FILE, i.e. make its mtime ("modified time") equal the time
392 specified with TM. The atime ("access time") is set to the current
396 touch (const char *file, time_t tm)
398 #ifdef HAVE_STRUCT_UTIMBUF
399 struct utimbuf times;
407 times.actime = time (NULL);
408 if (utime (file, ×) == -1)
409 logprintf (LOG_NOTQUIET, "utime(%s): %s\n", file, strerror (errno));
412 /* Checks if FILE is a symbolic link, and removes it if it is. Does
413 nothing under MS-Windows. */
415 remove_link (const char *file)
420 if (lstat (file, &st) == 0 && S_ISLNK (st.st_mode))
422 DEBUGP (("Unlinking %s (symlink).\n", file));
425 logprintf (LOG_VERBOSE, _("Failed to unlink symlink %s: %s\n"),
426 quote (file), strerror (errno));
431 /* Does FILENAME exist? This is quite a lousy implementation, since
432 it supplies no error codes -- only a yes-or-no answer. Thus it
433 will return that a file does not exist if, e.g., the directory is
434 unreadable. I don't mind it too much currently, though. The
435 proper way should, of course, be to have a third, error state,
436 other than true/false, but that would introduce uncalled-for
437 additional complexity to the callers. */
439 file_exists_p (const char *filename)
442 return access (filename, F_OK) >= 0;
445 return stat (filename, &buf) >= 0;
449 /* Returns 0 if PATH is a directory, 1 otherwise (any kind of file).
450 Returns 0 on error. */
452 file_non_directory_p (const char *path)
455 /* Use lstat() rather than stat() so that symbolic links pointing to
456 directories can be identified correctly. */
457 if (lstat (path, &buf) != 0)
459 return S_ISDIR (buf.st_mode) ? false : true;
462 /* Return the size of file named by FILENAME, or -1 if it cannot be
463 opened or seeked into. */
465 file_size (const char *filename)
467 #if defined(HAVE_FSEEKO) && defined(HAVE_FTELLO)
469 /* We use fseek rather than stat to determine the file size because
470 that way we can also verify that the file is readable without
471 explicitly checking for permissions. Inspired by the POST patch
473 FILE *fp = fopen (filename, "rb");
476 fseeko (fp, 0, SEEK_END);
482 if (stat (filename, &st) < 0)
488 /* stat file names named PREFIX.1, PREFIX.2, etc., until one that
489 doesn't exist is found. Return a freshly allocated copy of the
493 unique_name_1 (const char *prefix)
496 int plen = strlen (prefix);
497 char *template = (char *)alloca (plen + 1 + 24);
498 char *template_tail = template + plen;
500 memcpy (template, prefix, plen);
501 *template_tail++ = '.';
504 number_to_string (template_tail, count++);
505 while (file_exists_p (template));
507 return xstrdup (template);
510 /* Return a unique file name, based on FILE.
512 More precisely, if FILE doesn't exist, it is returned unmodified.
513 If not, FILE.1 is tried, then FILE.2, etc. The first FILE.<number>
514 file name that doesn't exist is returned.
516 The resulting file is not created, only verified that it didn't
517 exist at the point in time when the function was called.
518 Therefore, where security matters, don't rely that the file created
519 by this function exists until you open it with O_EXCL or
522 If ALLOW_PASSTHROUGH is 0, it always returns a freshly allocated
523 string. Otherwise, it may return FILE if the file doesn't exist
524 (and therefore doesn't need changing). */
527 unique_name (const char *file, bool allow_passthrough)
529 /* If the FILE itself doesn't exist, return it without
531 if (!file_exists_p (file))
532 return allow_passthrough ? (char *)file : xstrdup (file);
534 /* Otherwise, find a numeric suffix that results in unused file name
536 return unique_name_1 (file);
539 /* Create a file based on NAME, except without overwriting an existing
540 file with that name. Providing O_EXCL is correctly implemented,
541 this function does not have the race condition associated with
542 opening the file returned by unique_name. */
545 unique_create (const char *name, bool binary, char **opened_name)
547 /* unique file name, based on NAME */
548 char *uname = unique_name (name, false);
550 while ((fp = fopen_excl (uname, binary)) == NULL && errno == EEXIST)
553 uname = unique_name (name, false);
555 if (opened_name && fp != NULL)
558 *opened_name = uname;
570 /* Open the file for writing, with the addition that the file is
571 opened "exclusively". This means that, if the file already exists,
572 this function will *fail* and errno will be set to EEXIST. If
573 BINARY is set, the file will be opened in binary mode, equivalent
576 If opening the file fails for any reason, including the file having
577 previously existed, this function returns NULL and sets errno
581 fopen_excl (const char *fname, bool binary)
585 int flags = O_WRONLY | O_CREAT | O_EXCL;
590 fd = open (fname, flags, 0666);
593 return fdopen (fd, binary ? "wb" : "w");
594 #else /* not O_EXCL */
595 /* Manually check whether the file exists. This is prone to race
596 conditions, but systems without O_EXCL haven't deserved
598 if (file_exists_p (fname))
603 return fopen (fname, binary ? "wb" : "w");
604 #endif /* not O_EXCL */
607 /* Create DIRECTORY. If some of the pathname components of DIRECTORY
608 are missing, create them first. In case any mkdir() call fails,
609 return its error status. Returns 0 on successful completion.
611 The behaviour of this function should be identical to the behaviour
612 of `mkdir -p' on systems where mkdir supports the `-p' option. */
614 make_directory (const char *directory)
616 int i, ret, quit = 0;
619 /* Make a copy of dir, to be able to write to it. Otherwise, the
620 function is unsafe if called with a read-only char *argument. */
621 STRDUP_ALLOCA (dir, directory);
623 /* If the first character of dir is '/', skip it (and thus enable
624 creation of absolute-pathname directories. */
625 for (i = (*dir == '/'); 1; ++i)
627 for (; dir[i] && dir[i] != '/'; i++)
632 /* Check whether the directory already exists. Allow creation of
633 of intermediate directories to fail, as the initial path components
634 are not necessarily directories! */
635 if (!file_exists_p (dir))
636 ret = mkdir (dir, 0777);
647 /* Merge BASE with FILE. BASE can be a directory or a file name, FILE
648 should be a file name.
650 file_merge("/foo/bar", "baz") => "/foo/baz"
651 file_merge("/foo/bar/", "baz") => "/foo/bar/baz"
652 file_merge("foo", "bar") => "bar"
654 In other words, it's a simpler and gentler version of uri_merge. */
657 file_merge (const char *base, const char *file)
660 const char *cut = (const char *)strrchr (base, '/');
663 return xstrdup (file);
665 result = xmalloc (cut - base + 1 + strlen (file) + 1);
666 memcpy (result, base, cut - base);
667 result[cut - base] = '/';
668 strcpy (result + (cut - base) + 1, file);
673 /* Like fnmatch, but performs a case-insensitive match. */
676 fnmatch_nocase (const char *pattern, const char *string, int flags)
679 /* The FNM_CASEFOLD flag started as a GNU extension, but it is now
680 also present on *BSD platforms, and possibly elsewhere. */
681 return fnmatch (pattern, string, flags | FNM_CASEFOLD);
683 /* Turn PATTERN and STRING to lower case and call fnmatch on them. */
684 char *patcopy = (char *) alloca (strlen (pattern) + 1);
685 char *strcopy = (char *) alloca (strlen (string) + 1);
687 for (p = patcopy; *pattern; pattern++, p++)
688 *p = c_tolower (*pattern);
690 for (p = strcopy; *string; string++, p++)
691 *p = c_tolower (*string);
693 return fnmatch (patcopy, strcopy, flags);
697 static bool in_acclist (const char *const *, const char *, bool);
699 /* Determine whether a file is acceptable to be followed, according to
700 lists of patterns to accept/reject. */
702 acceptable (const char *s)
706 while (l && s[l] != '/')
713 return (in_acclist ((const char *const *)opt.accepts, s, true)
714 && !in_acclist ((const char *const *)opt.rejects, s, true));
716 return in_acclist ((const char *const *)opt.accepts, s, true);
718 else if (opt.rejects)
719 return !in_acclist ((const char *const *)opt.rejects, s, true);
723 /* Check if D2 is a subdirectory of D1. E.g. if D1 is `/something', subdir_p()
724 will return true if and only if D2 begins with `/something/' or is exactly
727 subdir_p (const char *d1, const char *d2)
731 if (!opt.ignore_case)
732 for (; *d1 && *d2 && (*d1 == *d2); ++d1, ++d2)
735 for (; *d1 && *d2 && (c_tolower (*d1) == c_tolower (*d2)); ++d1, ++d2)
738 return *d1 == '\0' && (*d2 == '\0' || *d2 == '/');
741 /* Iterate through DIRLIST (which must be NULL-terminated), and return the
742 first element that matches DIR, through wildcards or front comparison (as
745 dir_matches_p (char **dirlist, const char *dir)
748 int (*matcher) (const char *, const char *, int)
749 = opt.ignore_case ? fnmatch_nocase : fnmatch;
751 for (x = dirlist; *x; x++)
753 /* Remove leading '/' */
754 char *p = *x + (**x == '/');
755 if (has_wildcards_p (p))
757 if (matcher (p, dir, FNM_PATHNAME) == 0)
762 if (subdir_p (p, dir))
767 return *x ? true : false;
770 /* Returns whether DIRECTORY is acceptable for download, wrt the
771 include/exclude lists.
773 The leading `/' is ignored in paths; relative and absolute paths
774 may be freely intermixed. */
777 accdir (const char *directory)
779 /* Remove starting '/'. */
780 if (*directory == '/')
784 if (!dir_matches_p (opt.includes, directory))
789 if (dir_matches_p (opt.excludes, directory))
795 /* Return true if STRING ends with TAIL. For instance:
797 match_tail ("abc", "bc", false) -> 1
798 match_tail ("abc", "ab", false) -> 0
799 match_tail ("abc", "abc", false) -> 1
801 If FOLD_CASE is true, the comparison will be case-insensitive. */
804 match_tail (const char *string, const char *tail, bool fold_case)
808 /* We want this to be fast, so we code two loops, one with
809 case-folding, one without. */
813 for (i = strlen (string), j = strlen (tail); i >= 0 && j >= 0; i--, j--)
814 if (string[i] != tail[j])
819 for (i = strlen (string), j = strlen (tail); i >= 0 && j >= 0; i--, j--)
820 if (c_tolower (string[i]) != c_tolower (tail[j]))
824 /* If the tail was exhausted, the match was succesful. */
831 /* Checks whether string S matches each element of ACCEPTS. A list
832 element are matched either with fnmatch() or match_tail(),
833 according to whether the element contains wildcards or not.
835 If the BACKWARD is false, don't do backward comparison -- just compare
838 in_acclist (const char *const *accepts, const char *s, bool backward)
840 for (; *accepts; accepts++)
842 if (has_wildcards_p (*accepts))
844 int res = opt.ignore_case
845 ? fnmatch_nocase (*accepts, s, 0) : fnmatch (*accepts, s, 0);
846 /* fnmatch returns 0 if the pattern *does* match the string. */
854 if (match_tail (s, *accepts, opt.ignore_case))
859 int cmp = opt.ignore_case
860 ? strcasecmp (s, *accepts) : strcmp (s, *accepts);
869 /* Return the location of STR's suffix (file extension). Examples:
870 suffix ("foo.bar") -> "bar"
871 suffix ("foo.bar.baz") -> "baz"
872 suffix ("/foo/bar") -> NULL
873 suffix ("/foo.bar/baz") -> NULL */
875 suffix (const char *str)
879 for (i = strlen (str); i && str[i] != '/' && str[i] != '.'; i--)
883 return (char *)str + i;
888 /* Return true if S contains globbing wildcards (`*', `?', `[' or
892 has_wildcards_p (const char *s)
895 if (*s == '*' || *s == '?' || *s == '[' || *s == ']')
900 /* Return true if FNAME ends with a typical HTML suffix. The
901 following (case-insensitive) suffixes are presumed to be HTML
906 ?html (`?' matches one character)
908 #### CAVEAT. This is not necessarily a good indication that FNAME
909 refers to a file that contains HTML! */
911 has_html_suffix_p (const char *fname)
915 if ((suf = suffix (fname)) == NULL)
917 if (!strcasecmp (suf, "html"))
919 if (!strcasecmp (suf, "htm"))
921 if (suf[0] && !strcasecmp (suf + 1, "html"))
926 /* Read a line from FP and return the pointer to freshly allocated
927 storage. The storage space is obtained through malloc() and should
928 be freed with free() when it is no longer needed.
930 The length of the line is not limited, except by available memory.
931 The newline character at the end of line is retained. The line is
932 terminated with a zero character.
934 After end-of-file is encountered without anything being read, NULL
935 is returned. NULL is also returned on error. To distinguish
936 between these two cases, use the stdio function ferror(). */
939 read_whole_line (FILE *fp)
943 char *line = xmalloc (bufsize);
945 while (fgets (line + length, bufsize - length, fp))
947 length += strlen (line + length);
949 /* Possible for example when reading from a binary file where
950 a line begins with \0. */
953 if (line[length - 1] == '\n')
956 /* fgets() guarantees to read the whole line, or to use up the
957 space we've given it. We can double the buffer
960 line = xrealloc (line, bufsize);
962 if (length == 0 || ferror (fp))
967 if (length + 1 < bufsize)
968 /* Relieve the memory from our exponential greediness. We say
969 `length + 1' because the terminating \0 is not included in
970 LENGTH. We don't need to zero-terminate the string ourselves,
971 though, because fgets() does that. */
972 line = xrealloc (line, length + 1);
976 /* Read FILE into memory. A pointer to `struct file_memory' are
977 returned; use struct element `content' to access file contents, and
978 the element `length' to know the file length. `content' is *not*
979 zero-terminated, and you should *not* read or write beyond the [0,
980 length) range of characters.
982 After you are done with the file contents, call read_file_free to
985 Depending on the operating system and the type of file that is
986 being read, read_file() either mmap's the file into memory, or
987 reads the file into the core using read().
989 If file is named "-", fileno(stdin) is used for reading instead.
990 If you want to read from a real file named "-", use "./-" instead. */
993 read_file (const char *file)
996 struct file_memory *fm;
998 bool inhibit_close = false;
1000 /* Some magic in the finest tradition of Perl and its kin: if FILE
1001 is "-", just use stdin. */
1004 fd = fileno (stdin);
1005 inhibit_close = true;
1006 /* Note that we don't inhibit mmap() in this case. If stdin is
1007 redirected from a regular file, mmap() will still work. */
1010 fd = open (file, O_RDONLY);
1013 fm = xnew (struct file_memory);
1018 if (fstat (fd, &buf) < 0)
1020 fm->length = buf.st_size;
1021 /* NOTE: As far as I know, the callers of this function never
1022 modify the file text. Relying on this would enable us to
1023 specify PROT_READ and MAP_SHARED for a marginal gain in
1024 efficiency, but at some cost to generality. */
1025 fm->content = mmap (NULL, fm->length, PROT_READ | PROT_WRITE,
1026 MAP_PRIVATE, fd, 0);
1027 if (fm->content == (char *)MAP_FAILED)
1037 /* The most common reason why mmap() fails is that FD does not point
1038 to a plain file. However, it's also possible that mmap() doesn't
1039 work for a particular type of file. Therefore, whenever mmap()
1040 fails, we just fall back to the regular method. */
1041 #endif /* HAVE_MMAP */
1044 size = 512; /* number of bytes fm->contents can
1045 hold at any given time. */
1046 fm->content = xmalloc (size);
1050 if (fm->length > size / 2)
1052 /* #### I'm not sure whether the whole exponential-growth
1053 thing makes sense with kernel read. On Linux at least,
1054 read() refuses to read more than 4K from a file at a
1055 single chunk anyway. But other Unixes might optimize it
1056 better, and it doesn't *hurt* anything, so I'm leaving
1059 /* Normally, we grow SIZE exponentially to make the number
1060 of calls to read() and realloc() logarithmic in relation
1061 to file size. However, read() can read an amount of data
1062 smaller than requested, and it would be unreasonable to
1063 double SIZE every time *something* was read. Therefore,
1064 we double SIZE only when the length exceeds half of the
1065 entire allocated size. */
1067 fm->content = xrealloc (fm->content, size);
1069 nread = read (fd, fm->content + fm->length, size - fm->length);
1071 /* Successful read. */
1072 fm->length += nread;
1082 if (size > fm->length && fm->length != 0)
1083 /* Due to exponential growth of fm->content, the allocated region
1084 might be much larger than what is actually needed. */
1085 fm->content = xrealloc (fm->content, fm->length);
1092 xfree (fm->content);
1097 /* Release the resources held by FM. Specifically, this calls
1098 munmap() or xfree() on fm->content, depending whether mmap or
1099 malloc/read were used to read in the file. It also frees the
1100 memory needed to hold the FM structure itself. */
1103 read_file_free (struct file_memory *fm)
1108 munmap (fm->content, fm->length);
1113 xfree (fm->content);
1118 /* Free the pointers in a NULL-terminated vector of pointers, then
1119 free the pointer itself. */
1121 free_vec (char **vec)
1132 /* Append vector V2 to vector V1. The function frees V2 and
1133 reallocates V1 (thus you may not use the contents of neither
1134 pointer after the call). If V1 is NULL, V2 is returned. */
1136 merge_vecs (char **v1, char **v2)
1146 /* To avoid j == 0 */
1151 for (i = 0; v1[i]; i++)
1154 for (j = 0; v2[j]; j++)
1156 /* Reallocate v1. */
1157 v1 = xrealloc (v1, (i + j + 1) * sizeof (char **));
1158 memcpy (v1 + i, v2, (j + 1) * sizeof (char *));
1163 /* Append a freshly allocated copy of STR to VEC. If VEC is NULL, it
1164 is allocated as needed. Return the new value of the vector. */
1167 vec_append (char **vec, const char *str)
1169 int cnt; /* count of vector elements, including
1170 the one we're about to append */
1173 for (cnt = 0; vec[cnt]; cnt++)
1179 /* Reallocate the array to fit the new element and the NULL. */
1180 vec = xrealloc (vec, (cnt + 1) * sizeof (char *));
1181 /* Append a copy of STR to the vector. */
1182 vec[cnt - 1] = xstrdup (str);
1187 /* Sometimes it's useful to create "sets" of strings, i.e. special
1188 hash tables where you want to store strings as keys and merely
1189 query for their existence. Here is a set of utility routines that
1190 makes that transparent. */
1193 string_set_add (struct hash_table *ht, const char *s)
1195 /* First check whether the set element already exists. If it does,
1196 do nothing so that we don't have to free() the old element and
1197 then strdup() a new one. */
1198 if (hash_table_contains (ht, s))
1201 /* We use "1" as value. It provides us a useful and clear arbitrary
1202 value, and it consumes no memory -- the pointers to the same
1203 string "1" will be shared by all the key-value pairs in all `set'
1205 hash_table_put (ht, xstrdup (s), "1");
1208 /* Synonym for hash_table_contains... */
1211 string_set_contains (struct hash_table *ht, const char *s)
1213 return hash_table_contains (ht, s);
1216 /* Convert the specified string set to array. ARRAY should be large
1217 enough to hold hash_table_count(ht) char pointers. */
1219 void string_set_to_array (struct hash_table *ht, char **array)
1221 hash_table_iterator iter;
1222 for (hash_table_iterate (ht, &iter); hash_table_iter_next (&iter); )
1223 *array++ = iter.key;
1226 /* Free the string set. This frees both the storage allocated for
1227 keys and the actual hash table. (hash_table_destroy would only
1228 destroy the hash table.) */
1231 string_set_free (struct hash_table *ht)
1233 hash_table_iterator iter;
1234 for (hash_table_iterate (ht, &iter); hash_table_iter_next (&iter); )
1236 hash_table_destroy (ht);
1239 /* Utility function: simply call xfree() on all keys and values of HT. */
1242 free_keys_and_values (struct hash_table *ht)
1244 hash_table_iterator iter;
1245 for (hash_table_iterate (ht, &iter); hash_table_iter_next (&iter); )
1252 /* Get digit grouping data for thousand separors by calling
1253 localeconv(). The data includes separator string and grouping info
1254 and is cached after the first call to the function.
1256 In locales that don't set a thousand separator (such as the "C"
1257 locale), this forces it to be ",". We are now only showing
1258 thousand separators in one place, so this shouldn't be a problem in
1262 get_grouping_data (const char **sep, const char **grouping)
1264 static const char *cached_sep;
1265 static const char *cached_grouping;
1266 static bool initialized;
1269 /* Get the grouping info from the locale. */
1270 struct lconv *lconv = localeconv ();
1271 cached_sep = lconv->thousands_sep;
1272 cached_grouping = lconv->grouping;
1273 #if ! USE_NLS_PROGRESS_BAR
1274 /* We can't count column widths, so ensure that the separator
1275 * is single-byte only (let check below determine what byte). */
1276 if (strlen(cached_sep) > 1)
1281 /* Many locales (such as "C" or "hr_HR") don't specify
1282 grouping, which we still want to use it for legibility.
1283 In those locales set the sep char to ',', unless that
1284 character is used for decimal point, in which case set it
1286 if (*lconv->decimal_point != ',')
1290 cached_grouping = "\x03";
1295 *grouping = cached_grouping;
1298 /* Return a printed representation of N with thousand separators.
1299 This should respect locale settings, with the exception of the "C"
1300 locale which mandates no separator, but we use one anyway.
1302 Unfortunately, we cannot use %'d (in fact it would be %'j) to get
1303 the separators because it's too non-portable, and it's hard to test
1304 for this feature at configure time. Besides, it wouldn't display
1305 separators in the "C" locale, still used by many Unix users. */
1308 with_thousand_seps (wgint n)
1310 static char outbuf[48];
1311 char *p = outbuf + sizeof outbuf;
1313 /* Info received from locale */
1314 const char *grouping, *sep;
1317 /* State information */
1318 int i = 0, groupsize;
1319 const char *atgroup;
1321 bool negative = n < 0;
1323 /* Initialize grouping data. */
1324 get_grouping_data (&sep, &grouping);
1325 seplen = strlen (sep);
1327 groupsize = *atgroup++;
1329 /* This would overflow on WGINT_MIN, but printing negative numbers
1330 is not an important goal of this fuinction. */
1334 /* Write the number into the buffer, backwards, inserting the
1335 separators as necessary. */
1339 *--p = n % 10 + '0';
1343 /* Prepend SEP to every groupsize'd digit and get new groupsize. */
1344 if (++i == groupsize)
1349 memcpy (p -= seplen, sep, seplen);
1352 groupsize = *atgroup++;
1361 /* N, a byte quantity, is converted to a human-readable abberviated
1362 form a la sizes printed by `ls -lh'. The result is written to a
1363 static buffer, a pointer to which is returned.
1365 Unlike `with_thousand_seps', this approximates to the nearest unit.
1366 Quoting GNU libit: "Most people visually process strings of 3-4
1367 digits effectively, but longer strings of digits are more prone to
1368 misinterpretation. Hence, converting to an abbreviated form
1369 usually improves readability."
1371 This intentionally uses kilobyte (KB), megabyte (MB), etc. in their
1372 original computer-related meaning of "powers of 1024". We don't
1373 use the "*bibyte" names invented in 1998, and seldom used in
1374 practice. Wikipedia's entry on "binary prefix" discusses this in
1378 human_readable (HR_NUMTYPE n)
1380 /* These suffixes are compatible with those of GNU `ls -lh'. */
1381 static char powers[] =
1383 'K', /* kilobyte, 2^10 bytes */
1384 'M', /* megabyte, 2^20 bytes */
1385 'G', /* gigabyte, 2^30 bytes */
1386 'T', /* terabyte, 2^40 bytes */
1387 'P', /* petabyte, 2^50 bytes */
1388 'E', /* exabyte, 2^60 bytes */
1393 /* If the quantity is smaller than 1K, just print it. */
1396 snprintf (buf, sizeof (buf), "%d", (int) n);
1400 /* Loop over powers, dividing N with 1024 in each iteration. This
1401 works unchanged for all sizes of wgint, while still avoiding
1402 non-portable `long double' arithmetic. */
1403 for (i = 0; i < countof (powers); i++)
1405 /* At each iteration N is greater than the *subsequent* power.
1406 That way N/1024.0 produces a decimal number in the units of
1408 if ((n / 1024) < 1024 || i == countof (powers) - 1)
1410 double val = n / 1024.0;
1411 /* Print values smaller than 10 with one decimal digits, and
1412 others without any decimals. */
1413 snprintf (buf, sizeof (buf), "%.*f%c",
1414 val < 10 ? 1 : 0, val, powers[i]);
1419 return NULL; /* unreached */
1422 /* Count the digits in the provided number. Used to allocate space
1423 when printing numbers. */
1426 numdigit (wgint number)
1430 ++cnt; /* accomodate '-' */
1431 while ((number /= 10) != 0)
1436 #define PR(mask) *p++ = n / (mask) + '0'
1438 /* DIGITS_<D> is used to print a D-digit number and should be called
1439 with mask==10^(D-1). It prints n/mask (the first digit), reducing
1440 n to n%mask (the remaining digits), and calling DIGITS_<D-1>.
1441 Recursively this continues until DIGITS_1 is invoked. */
1443 #define DIGITS_1(mask) PR (mask)
1444 #define DIGITS_2(mask) PR (mask), n %= (mask), DIGITS_1 ((mask) / 10)
1445 #define DIGITS_3(mask) PR (mask), n %= (mask), DIGITS_2 ((mask) / 10)
1446 #define DIGITS_4(mask) PR (mask), n %= (mask), DIGITS_3 ((mask) / 10)
1447 #define DIGITS_5(mask) PR (mask), n %= (mask), DIGITS_4 ((mask) / 10)
1448 #define DIGITS_6(mask) PR (mask), n %= (mask), DIGITS_5 ((mask) / 10)
1449 #define DIGITS_7(mask) PR (mask), n %= (mask), DIGITS_6 ((mask) / 10)
1450 #define DIGITS_8(mask) PR (mask), n %= (mask), DIGITS_7 ((mask) / 10)
1451 #define DIGITS_9(mask) PR (mask), n %= (mask), DIGITS_8 ((mask) / 10)
1452 #define DIGITS_10(mask) PR (mask), n %= (mask), DIGITS_9 ((mask) / 10)
1454 /* DIGITS_<11-20> are only used on machines with 64-bit wgints. */
1456 #define DIGITS_11(mask) PR (mask), n %= (mask), DIGITS_10 ((mask) / 10)
1457 #define DIGITS_12(mask) PR (mask), n %= (mask), DIGITS_11 ((mask) / 10)
1458 #define DIGITS_13(mask) PR (mask), n %= (mask), DIGITS_12 ((mask) / 10)
1459 #define DIGITS_14(mask) PR (mask), n %= (mask), DIGITS_13 ((mask) / 10)
1460 #define DIGITS_15(mask) PR (mask), n %= (mask), DIGITS_14 ((mask) / 10)
1461 #define DIGITS_16(mask) PR (mask), n %= (mask), DIGITS_15 ((mask) / 10)
1462 #define DIGITS_17(mask) PR (mask), n %= (mask), DIGITS_16 ((mask) / 10)
1463 #define DIGITS_18(mask) PR (mask), n %= (mask), DIGITS_17 ((mask) / 10)
1464 #define DIGITS_19(mask) PR (mask), n %= (mask), DIGITS_18 ((mask) / 10)
1466 /* Shorthand for casting to wgint. */
1469 /* Print NUMBER to BUFFER in base 10. This is equivalent to
1470 `sprintf(buffer, "%lld", (long long) number)', only typically much
1471 faster and portable to machines without long long.
1473 The speedup may make a difference in programs that frequently
1474 convert numbers to strings. Some implementations of sprintf,
1475 particularly the one in some versions of GNU libc, have been known
1476 to be quite slow when converting integers to strings.
1478 Return the pointer to the location where the terminating zero was
1479 printed. (Equivalent to calling buffer+strlen(buffer) after the
1482 BUFFER should be large enough to accept as many bytes as you expect
1483 the number to take up. On machines with 64-bit wgints the maximum
1484 needed size is 24 bytes. That includes the digits needed for the
1485 largest 64-bit number, the `-' sign in case it's negative, and the
1486 terminating '\0'. */
1489 number_to_string (char *buffer, wgint number)
1494 int last_digit_char = 0;
1496 #if (SIZEOF_WGINT != 4) && (SIZEOF_WGINT != 8)
1497 /* We are running in a very strange environment. Leave the correct
1498 printing to sprintf. */
1499 p += sprintf (buf, "%j", (intmax_t) (n));
1500 #else /* (SIZEOF_WGINT == 4) || (SIZEOF_WGINT == 8) */
1506 /* n = -n would overflow because -n would evaluate to a
1507 wgint value larger than WGINT_MAX. Need to make n
1508 smaller and handle the last digit separately. */
1509 int last_digit = n % 10;
1510 /* The sign of n%10 is implementation-defined. */
1512 last_digit_char = '0' - last_digit;
1514 last_digit_char = '0' + last_digit;
1515 /* After n is made smaller, -n will not overflow. */
1523 /* Use the DIGITS_ macro appropriate for N's number of digits. That
1524 way printing any N is fully open-coded without a loop or jump.
1525 (Also see description of DIGITS_*.) */
1527 if (n < 10) DIGITS_1 (1);
1528 else if (n < 100) DIGITS_2 (10);
1529 else if (n < 1000) DIGITS_3 (100);
1530 else if (n < 10000) DIGITS_4 (1000);
1531 else if (n < 100000) DIGITS_5 (10000);
1532 else if (n < 1000000) DIGITS_6 (100000);
1533 else if (n < 10000000) DIGITS_7 (1000000);
1534 else if (n < 100000000) DIGITS_8 (10000000);
1535 else if (n < 1000000000) DIGITS_9 (100000000);
1536 #if SIZEOF_WGINT == 4
1537 /* wgint is 32 bits wide: no number has more than 10 digits. */
1538 else DIGITS_10 (1000000000);
1540 /* wgint is 64 bits wide: handle numbers with 9-19 decimal digits.
1541 Constants are constructed by compile-time multiplication to avoid
1542 dealing with different notations for 64-bit constants
1543 (nL/nLL/nI64, depending on the compiler and architecture). */
1544 else if (n < 10*(W)1000000000) DIGITS_10 (1000000000);
1545 else if (n < 100*(W)1000000000) DIGITS_11 (10*(W)1000000000);
1546 else if (n < 1000*(W)1000000000) DIGITS_12 (100*(W)1000000000);
1547 else if (n < 10000*(W)1000000000) DIGITS_13 (1000*(W)1000000000);
1548 else if (n < 100000*(W)1000000000) DIGITS_14 (10000*(W)1000000000);
1549 else if (n < 1000000*(W)1000000000) DIGITS_15 (100000*(W)1000000000);
1550 else if (n < 10000000*(W)1000000000) DIGITS_16 (1000000*(W)1000000000);
1551 else if (n < 100000000*(W)1000000000) DIGITS_17 (10000000*(W)1000000000);
1552 else if (n < 1000000000*(W)1000000000) DIGITS_18 (100000000*(W)1000000000);
1553 else DIGITS_19 (1000000000*(W)1000000000);
1556 if (last_digit_char)
1557 *p++ = last_digit_char;
1560 #endif /* (SIZEOF_WGINT == 4) || (SIZEOF_WGINT == 8) */
1567 #undef SPRINTF_WGINT
1590 /* Print NUMBER to a statically allocated string and return a pointer
1591 to the printed representation.
1593 This function is intended to be used in conjunction with printf.
1594 It is hard to portably print wgint values:
1595 a) you cannot use printf("%ld", number) because wgint can be long
1596 long on 32-bit machines with LFS.
1597 b) you cannot use printf("%lld", number) because NUMBER could be
1598 long on 32-bit machines without LFS, or on 64-bit machines,
1599 which do not require LFS. Also, Windows doesn't support %lld.
1600 c) you cannot use printf("%j", (int_max_t) number) because not all
1601 versions of printf support "%j", the most notable being the one
1603 d) you cannot #define WGINT_FMT to the appropriate format and use
1604 printf(WGINT_FMT, number) because that would break translations
1605 for user-visible messages, such as printf("Downloaded: %d
1608 What you should use instead is printf("%s", number_to_static_string
1611 CAVEAT: since the function returns pointers to static data, you
1612 must be careful to copy its result before calling it again.
1613 However, to make it more useful with printf, the function maintains
1614 an internal ring of static buffers to return. That way things like
1615 printf("%s %s", number_to_static_string (num1),
1616 number_to_static_string (num2)) work as expected. Three buffers
1617 are currently used, which means that "%s %s %s" will work, but "%s
1618 %s %s %s" won't. If you need to print more than three wgints,
1619 bump the RING_SIZE (or rethink your message.) */
1622 number_to_static_string (wgint number)
1624 static char ring[RING_SIZE][24];
1626 char *buf = ring[ringpos];
1627 number_to_string (buf, number);
1628 ringpos = (ringpos + 1) % RING_SIZE;
1632 /* Determine the width of the terminal we're running on. If that's
1633 not possible, return 0. */
1636 determine_screen_width (void)
1638 /* If there's a way to get the terminal size using POSIX
1639 tcgetattr(), somebody please tell me. */
1644 if (opt.lfilename != NULL)
1647 fd = fileno (stderr);
1648 if (ioctl (fd, TIOCGWINSZ, &wsz) < 0)
1649 return 0; /* most likely ENOTTY */
1652 #elif defined(WINDOWS)
1653 CONSOLE_SCREEN_BUFFER_INFO csbi;
1654 if (!GetConsoleScreenBufferInfo (GetStdHandle (STD_ERROR_HANDLE), &csbi))
1656 return csbi.dwSize.X;
1657 #else /* neither TIOCGWINSZ nor WINDOWS */
1659 #endif /* neither TIOCGWINSZ nor WINDOWS */
1662 /* Whether the rnd system (either rand or [dl]rand48) has been
1664 static int rnd_seeded;
1666 /* Return a random number between 0 and MAX-1, inclusive.
1668 If the system does not support lrand48 and MAX is greater than the
1669 value of RAND_MAX+1 on the system, the returned value will be in
1670 the range [0, RAND_MAX]. This may be fixed in a future release.
1671 The random number generator is seeded automatically the first time
1674 This uses lrand48 where available, rand elsewhere. DO NOT use it
1675 for cryptography. It is only meant to be used in situations where
1676 quality of the random numbers returned doesn't really matter. */
1679 random_number (int max)
1684 srand48 ((long) time (NULL) ^ (long) getpid ());
1687 return lrand48 () % max;
1688 #else /* not HAVE_DRAND48 */
1694 srand ((unsigned) time (NULL) ^ (unsigned) getpid ());
1699 /* Like rand() % max, but uses the high-order bits for better
1700 randomness on architectures where rand() is implemented using a
1701 simple congruential generator. */
1703 bounded = (double) max * rnd / (RAND_MAX + 1.0);
1704 return (int) bounded;
1706 #endif /* not HAVE_DRAND48 */
1709 /* Return a random uniformly distributed floating point number in the
1710 [0, 1) range. Uses drand48 where available, and a really lame
1711 kludge elsewhere. */
1719 srand48 ((long) time (NULL) ^ (long) getpid ());
1723 #else /* not HAVE_DRAND48 */
1724 return ( random_number (10000) / 10000.0
1725 + random_number (10000) / (10000.0 * 10000.0)
1726 + random_number (10000) / (10000.0 * 10000.0 * 10000.0)
1727 + random_number (10000) / (10000.0 * 10000.0 * 10000.0 * 10000.0));
1728 #endif /* not HAVE_DRAND48 */
1731 /* Implementation of run_with_timeout, a generic timeout-forcing
1732 routine for systems with Unix-like signal handling. */
1734 #ifdef USE_SIGNAL_TIMEOUT
1735 # ifdef HAVE_SIGSETJMP
1736 # define SETJMP(env) sigsetjmp (env, 1)
1738 static sigjmp_buf run_with_timeout_env;
1741 abort_run_with_timeout (int sig)
1743 assert (sig == SIGALRM);
1744 siglongjmp (run_with_timeout_env, -1);
1746 # else /* not HAVE_SIGSETJMP */
1747 # define SETJMP(env) setjmp (env)
1749 static jmp_buf run_with_timeout_env;
1752 abort_run_with_timeout (int sig)
1754 assert (sig == SIGALRM);
1755 /* We don't have siglongjmp to preserve the set of blocked signals;
1756 if we longjumped out of the handler at this point, SIGALRM would
1757 remain blocked. We must unblock it manually. */
1758 int mask = siggetmask ();
1759 mask &= ~sigmask (SIGALRM);
1762 /* Now it's safe to longjump. */
1763 longjmp (run_with_timeout_env, -1);
1765 # endif /* not HAVE_SIGSETJMP */
1767 /* Arrange for SIGALRM to be delivered in TIMEOUT seconds. This uses
1768 setitimer where available, alarm otherwise.
1770 TIMEOUT should be non-zero. If the timeout value is so small that
1771 it would be rounded to zero, it is rounded to the least legal value
1772 instead (1us for setitimer, 1s for alarm). That ensures that
1773 SIGALRM will be delivered in all cases. */
1776 alarm_set (double timeout)
1779 /* Use the modern itimer interface. */
1780 struct itimerval itv;
1782 itv.it_value.tv_sec = (long) timeout;
1783 itv.it_value.tv_usec = 1000000 * (timeout - (long)timeout);
1784 if (itv.it_value.tv_sec == 0 && itv.it_value.tv_usec == 0)
1785 /* Ensure that we wait for at least the minimum interval.
1786 Specifying zero would mean "wait forever". */
1787 itv.it_value.tv_usec = 1;
1788 setitimer (ITIMER_REAL, &itv, NULL);
1789 #else /* not ITIMER_REAL */
1790 /* Use the old alarm() interface. */
1791 int secs = (int) timeout;
1793 /* Round TIMEOUTs smaller than 1 to 1, not to zero. This is
1794 because alarm(0) means "never deliver the alarm", i.e. "wait
1795 forever", which is not what someone who specifies a 0.5s
1796 timeout would expect. */
1799 #endif /* not ITIMER_REAL */
1802 /* Cancel the alarm set with alarm_set. */
1808 struct itimerval disable;
1810 setitimer (ITIMER_REAL, &disable, NULL);
1811 #else /* not ITIMER_REAL */
1813 #endif /* not ITIMER_REAL */
1816 /* Call FUN(ARG), but don't allow it to run for more than TIMEOUT
1817 seconds. Returns true if the function was interrupted with a
1818 timeout, false otherwise.
1820 This works by setting up SIGALRM to be delivered in TIMEOUT seconds
1821 using setitimer() or alarm(). The timeout is enforced by
1822 longjumping out of the SIGALRM handler. This has several
1823 advantages compared to the traditional approach of relying on
1824 signals causing system calls to exit with EINTR:
1826 * The callback function is *forcibly* interrupted after the
1827 timeout expires, (almost) regardless of what it was doing and
1828 whether it was in a syscall. For example, a calculation that
1829 takes a long time is interrupted as reliably as an IO
1832 * It works with both SYSV and BSD signals because it doesn't
1833 depend on the default setting of SA_RESTART.
1835 * It doesn't require special handler setup beyond a simple call
1836 to signal(). (It does use sigsetjmp/siglongjmp, but they're
1839 The only downside is that, if FUN allocates internal resources that
1840 are normally freed prior to exit from the functions, they will be
1841 lost in case of timeout. */
1844 run_with_timeout (double timeout, void (*fun) (void *), void *arg)
1854 signal (SIGALRM, abort_run_with_timeout);
1855 if (SETJMP (run_with_timeout_env) != 0)
1857 /* Longjumped out of FUN with a timeout. */
1858 signal (SIGALRM, SIG_DFL);
1861 alarm_set (timeout);
1864 /* Preserve errno in case alarm() or signal() modifies it. */
1865 saved_errno = errno;
1867 signal (SIGALRM, SIG_DFL);
1868 errno = saved_errno;
1873 #else /* not USE_SIGNAL_TIMEOUT */
1876 /* A stub version of run_with_timeout that just calls FUN(ARG). Don't
1877 define it under Windows, because Windows has its own version of
1878 run_with_timeout that uses threads. */
1881 run_with_timeout (double timeout, void (*fun) (void *), void *arg)
1886 #endif /* not WINDOWS */
1887 #endif /* not USE_SIGNAL_TIMEOUT */
1891 /* Sleep the specified amount of seconds. On machines without
1892 nanosleep(), this may sleep shorter if interrupted by signals. */
1895 xsleep (double seconds)
1897 #ifdef HAVE_NANOSLEEP
1898 /* nanosleep is the preferred interface because it offers high
1899 accuracy and, more importantly, because it allows us to reliably
1900 restart receiving a signal such as SIGWINCH. (There was an
1901 actual Debian bug report about --limit-rate malfunctioning while
1902 the terminal was being resized.) */
1903 struct timespec sleep, remaining;
1904 sleep.tv_sec = (long) seconds;
1905 sleep.tv_nsec = 1000000000 * (seconds - (long) seconds);
1906 while (nanosleep (&sleep, &remaining) < 0 && errno == EINTR)
1907 /* If nanosleep has been interrupted by a signal, adjust the
1908 sleeping period and return to sleep. */
1910 #elif defined(HAVE_USLEEP)
1911 /* If usleep is available, use it in preference to select. */
1914 /* On some systems, usleep cannot handle values larger than
1915 1,000,000. If the period is larger than that, use sleep
1916 first, then add usleep for subsecond accuracy. */
1918 seconds -= (long) seconds;
1920 usleep (seconds * 1000000);
1921 #else /* fall back select */
1922 /* Note that, although Windows supports select, it can't be used to
1923 implement sleeping because Winsock's select doesn't implement
1924 timeout when it is passed NULL pointers for all fd sets. (But it
1925 does under Cygwin, which implements Unix-compatible select.) */
1926 struct timeval sleep;
1927 sleep.tv_sec = (long) seconds;
1928 sleep.tv_usec = 1000000 * (seconds - (long) seconds);
1929 select (0, NULL, NULL, NULL, &sleep);
1930 /* If select returns -1 and errno is EINTR, it means we were
1931 interrupted by a signal. But without knowing how long we've
1932 actually slept, we can't return to sleep. Using gettimeofday to
1933 track sleeps is slow and unreliable due to clock skew. */
1937 #endif /* not WINDOWS */
1939 /* Encode the octets in DATA of length LENGTH to base64 format,
1940 storing the result to DEST. The output will be zero-terminated,
1941 and must point to a writable buffer of at least
1942 1+BASE64_LENGTH(length) bytes. The function returns the length of
1943 the resulting base64 data, not counting the terminating zero.
1945 This implementation does not emit newlines after 76 characters of
1949 base64_encode (const void *data, int length, char *dest)
1951 /* Conversion table. */
1952 static const char tbl[64] = {
1953 'A','B','C','D','E','F','G','H','I','J','K','L','M','N','O','P',
1954 'Q','R','S','T','U','V','W','X','Y','Z','a','b','c','d','e','f',
1955 'g','h','i','j','k','l','m','n','o','p','q','r','s','t','u','v',
1956 'w','x','y','z','0','1','2','3','4','5','6','7','8','9','+','/'
1958 /* Access bytes in DATA as unsigned char, otherwise the shifts below
1959 don't work for data with MSB set. */
1960 const unsigned char *s = data;
1961 /* Theoretical ANSI violation when length < 3. */
1962 const unsigned char *end = (const unsigned char *) data + length - 2;
1965 /* Transform the 3x8 bits to 4x6 bits, as required by base64. */
1966 for (; s < end; s += 3)
1968 *p++ = tbl[s[0] >> 2];
1969 *p++ = tbl[((s[0] & 3) << 4) + (s[1] >> 4)];
1970 *p++ = tbl[((s[1] & 0xf) << 2) + (s[2] >> 6)];
1971 *p++ = tbl[s[2] & 0x3f];
1974 /* Pad the result if necessary... */
1978 *p++ = tbl[s[0] >> 2];
1979 *p++ = tbl[(s[0] & 3) << 4];
1984 *p++ = tbl[s[0] >> 2];
1985 *p++ = tbl[((s[0] & 3) << 4) + (s[1] >> 4)];
1986 *p++ = tbl[((s[1] & 0xf) << 2)];
1990 /* ...and zero-terminate it. */
1996 /* Store in C the next non-whitespace character from the string, or \0
1997 when end of string is reached. */
1998 #define NEXT_CHAR(c, p) do { \
1999 c = (unsigned char) *p++; \
2000 } while (c_isspace (c))
2002 #define IS_ASCII(c) (((c) & 0x80) == 0)
2004 /* Decode data from BASE64 (a null-terminated string) into memory
2005 pointed to by DEST. DEST is assumed to be large enough to
2006 accomodate the decoded data, which is guaranteed to be no more than
2009 Since DEST is assumed to contain binary data, it is not
2010 NUL-terminated. The function returns the length of the data
2011 written to TO. -1 is returned in case of error caused by malformed
2014 This function originates from Free Recode. */
2017 base64_decode (const char *base64, void *dest)
2019 /* Table of base64 values for first 128 characters. Note that this
2020 assumes ASCII (but so does Wget in other places). */
2021 static const signed char base64_char_to_value[128] =
2023 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 0- 9 */
2024 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 10- 19 */
2025 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 20- 29 */
2026 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 30- 39 */
2027 -1, -1, -1, 62, -1, -1, -1, 63, 52, 53, /* 40- 49 */
2028 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, /* 50- 59 */
2029 -1, -1, -1, -1, -1, 0, 1, 2, 3, 4, /* 60- 69 */
2030 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, /* 70- 79 */
2031 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, /* 80- 89 */
2032 25, -1, -1, -1, -1, -1, -1, 26, 27, 28, /* 90- 99 */
2033 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, /* 100-109 */
2034 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, /* 110-119 */
2035 49, 50, 51, -1, -1, -1, -1, -1 /* 120-127 */
2037 #define BASE64_CHAR_TO_VALUE(c) ((int) base64_char_to_value[c])
2038 #define IS_BASE64(c) ((IS_ASCII (c) && BASE64_CHAR_TO_VALUE (c) >= 0) || c == '=')
2040 const char *p = base64;
2046 unsigned long value;
2048 /* Process first byte of a quadruplet. */
2052 if (c == '=' || !IS_BASE64 (c))
2053 return -1; /* illegal char while decoding base64 */
2054 value = BASE64_CHAR_TO_VALUE (c) << 18;
2056 /* Process second byte of a quadruplet. */
2059 return -1; /* premature EOF while decoding base64 */
2060 if (c == '=' || !IS_BASE64 (c))
2061 return -1; /* illegal char while decoding base64 */
2062 value |= BASE64_CHAR_TO_VALUE (c) << 12;
2065 /* Process third byte of a quadruplet. */
2068 return -1; /* premature EOF while decoding base64 */
2070 return -1; /* illegal char while decoding base64 */
2076 return -1; /* premature EOF while decoding base64 */
2078 return -1; /* padding `=' expected but not found */
2082 value |= BASE64_CHAR_TO_VALUE (c) << 6;
2083 *q++ = 0xff & value >> 8;
2085 /* Process fourth byte of a quadruplet. */
2088 return -1; /* premature EOF while decoding base64 */
2092 return -1; /* illegal char while decoding base64 */
2094 value |= BASE64_CHAR_TO_VALUE (c);
2095 *q++ = 0xff & value;
2098 #undef BASE64_CHAR_TO_VALUE
2100 return q - (char *) dest;
2106 /* Simple merge sort for use by stable_sort. Implementation courtesy
2107 Zeljko Vrba with additional debugging by Nenad Barbutov. */
2110 mergesort_internal (void *base, void *temp, size_t size, size_t from, size_t to,
2111 int (*cmpfun) (const void *, const void *))
2113 #define ELT(array, pos) ((char *)(array) + (pos) * size)
2117 size_t mid = (to + from) / 2;
2118 mergesort_internal (base, temp, size, from, mid, cmpfun);
2119 mergesort_internal (base, temp, size, mid + 1, to, cmpfun);
2122 for (k = from; (i <= mid) && (j <= to); k++)
2123 if (cmpfun (ELT (base, i), ELT (base, j)) <= 0)
2124 memcpy (ELT (temp, k), ELT (base, i++), size);
2126 memcpy (ELT (temp, k), ELT (base, j++), size);
2128 memcpy (ELT (temp, k++), ELT (base, i++), size);
2130 memcpy (ELT (temp, k++), ELT (base, j++), size);
2131 for (k = from; k <= to; k++)
2132 memcpy (ELT (base, k), ELT (temp, k), size);
2137 /* Stable sort with interface exactly like standard library's qsort.
2138 Uses mergesort internally, allocating temporary storage with
2142 stable_sort (void *base, size_t nmemb, size_t size,
2143 int (*cmpfun) (const void *, const void *))
2147 void *temp = alloca (nmemb * size * sizeof (void *));
2148 mergesort_internal (base, temp, size, 0, nmemb - 1, cmpfun);
2152 /* Print a decimal number. If it is equal to or larger than ten, the
2153 number is rounded. Otherwise it is printed with one significant
2154 digit without trailing zeros and with no more than three fractional
2155 digits total. For example, 0.1 is printed as "0.1", 0.035 is
2156 printed as "0.04", 0.0091 as "0.009", and 0.0003 as simply "0".
2158 This is useful for displaying durations because it provides
2159 order-of-magnitude information without unnecessary clutter --
2160 long-running downloads are shown without the fractional part, and
2161 short ones still retain one significant digit. */
2164 print_decimal (double number)
2166 static char buf[32];
2167 double n = number >= 0 ? number : -number;
2170 /* Cut off at 9.95 because the below %.1f would round 9.96 to
2171 "10.0" instead of "10". OTOH 9.94 will print as "9.9". */
2172 snprintf (buf, sizeof buf, "%.0f", number);
2174 snprintf (buf, sizeof buf, "%.1f", number);
2175 else if (n >= 0.001)
2176 snprintf (buf, sizeof buf, "%.1g", number);
2177 else if (n >= 0.0005)
2178 /* round [0.0005, 0.001) to 0.001 */
2179 snprintf (buf, sizeof buf, "%.3f", number);
2181 /* print numbers close to 0 as 0, not 0.000 */
2198 { "/somedir", "/somedir", true },
2199 { "/somedir", "/somedir/d2", true },
2200 { "/somedir/d1", "/somedir", false },
2203 for (i = 0; i < countof(test_array); ++i)
2205 bool res = subdir_p (test_array[i].d1, test_array[i].d2);
2207 mu_assert ("test_subdir_p: wrong result",
2208 res == test_array[i].result);
2215 test_dir_matches_p()
2223 { { "/somedir", "/someotherdir", NULL }, "somedir", true },
2224 { { "/somedir", "/someotherdir", NULL }, "anotherdir", false },
2225 { { "/somedir", "/*otherdir", NULL }, "anotherdir", true },
2226 { { "/somedir/d1", "/someotherdir", NULL }, "somedir/d1", true },
2227 { { "*/*d1", "/someotherdir", NULL }, "somedir/d1", true },
2228 { { "/somedir/d1", "/someotherdir", NULL }, "d1", false },
2229 { { "!COMPLETE", NULL, NULL }, "!COMPLETE", true },
2230 { { "*COMPLETE", NULL, NULL }, "!COMPLETE", true },
2231 { { "*/!COMPLETE", NULL, NULL }, "foo/!COMPLETE", true },
2232 { { "*COMPLETE", NULL, NULL }, "foo/!COMPLETE", false },
2233 { { "*/*COMPLETE", NULL, NULL }, "foo/!COMPLETE", true },
2234 { { "/dir with spaces", NULL, NULL }, "dir with spaces", true },
2235 { { "/dir*with*spaces", NULL, NULL }, "dir with spaces", true },
2238 for (i = 0; i < countof(test_array); ++i)
2240 bool res = dir_matches_p (test_array[i].dirlist, test_array[i].dir);
2242 mu_assert ("test_dir_matches_p: wrong result",
2243 res == test_array[i].result);
2249 #endif /* TESTING */