1 /* Various utility functions.
2 Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
3 2005, 2006, 2007, 2008, 2009 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. */
37 #ifdef HAVE_SYS_TIME_H
38 # include <sys/time.h>
44 # include <sys/mman.h>
47 # include <process.h> /* getpid() */
52 #ifdef HAVE_SYS_UTIME_H
53 # include <sys/utime.h>
61 /* For TIOCGWINSZ and friends: */
62 #ifdef HAVE_SYS_IOCTL_H
63 # include <sys/ioctl.h>
69 /* Needed for Unix version of run_with_timeout. */
73 #ifndef HAVE_SIGSETJMP
74 /* If sigsetjmp is a macro, configure won't pick it up. */
76 # define HAVE_SIGSETJMP
80 #if defined HAVE_SIGSETJMP || defined HAVE_SIGBLOCK
81 # define USE_SIGNAL_TIMEOUT
89 #endif /* def __VMS */
96 memfatal (const char *context, long attempted_size)
98 /* Make sure we don't try to store part of the log line, and thus
100 log_set_save_context (false);
102 /* We have different log outputs in different situations:
103 1) output without bytes information
104 2) output with bytes information */
105 if (attempted_size == UNKNOWN_ATTEMPTED_SIZE)
107 logprintf (LOG_ALWAYS,
108 _("%s: %s: Failed to allocate enough memory; memory exhausted.\n"),
113 logprintf (LOG_ALWAYS,
114 _("%s: %s: Failed to allocate %ld bytes; memory exhausted.\n"),
115 exec_name, context, attempted_size);
121 /* Character property table for (re-)escaping VMS ODS5 extended file
122 names. Note that this table ignores Unicode.
124 ODS2 valid characters: 0-9 A-Z a-z $ - _ ~
126 ODS5 Invalid characters:
127 C0 control codes (0x00 to 0x1F inclusive)
131 ODS5 Invalid characters only in VMS V7.2 (which no one runs, right?):
132 Double quotation marks (")
135 Left angle bracket (<)
136 Right angle bracket (>)
140 Characters escaped by "^":
141 SP ! # % & ' ( ) + , . ; = @ [ ] ^ ` { } ~
143 Either "^_" or "^ " is accepted as a space. Period (.) is a special
144 case. Note that un-escaped < and > can also confuse a directory
147 Characters put out as ^xx:
149 80-9F (C1 control characters)
150 A0 (nonbreaking space)
151 FF (Latin small letter y diaeresis)
154 Unicode: "^Uxxxx", where "xxxx" is four hex digits.
156 Property table values:
166 unsigned char char_prop[ 256] = {
168 /* NUL SOH STX ETX EOT ENQ ACK BEL BS HT LF VT FF CR SO SI */
169 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
171 /* DLE DC1 DC2 DC3 DC4 NAK SYN ETB CAN EM SUB ESC FS GS RS US */
172 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
174 /* SP ! " # $ % & ' ( ) * + , - . / */
175 2, 1, 0, 1, 16, 1, 1, 1, 1, 1, 0, 1, 1, 16, 4, 0,
177 /* 0 1 2 3 4 5 6 7 8 9 : ; < = > ? */
178 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 0, 1, 1, 1, 1, 1,
180 /* @ A B C D E F G H I J K L M N O */
181 1, 80, 80, 80, 80, 80, 80, 16, 16, 16, 16, 16, 16, 16, 16, 16,
183 /* P Q R S T U V W X Y Z [ \ ] ^ _ */
184 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 1, 0, 1, 1, 16,
186 /* ` a b c d e f g h i j k l m n o */
187 1, 96, 96, 96, 96, 96, 96, 32, 32, 32, 32, 32, 32, 32, 32, 32,
189 /* p q r s t u v w x y z { | } ~ DEL */
190 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 1, 0, 1, 17, 8,
192 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
193 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
194 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
195 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
196 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
197 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
198 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
199 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8
202 /* Utility function: like xstrdup(), but also lowercases S. */
205 xstrdup_lower (const char *s)
207 char *copy = xstrdup (s);
214 /* Copy the string formed by two pointers (one on the beginning, other
215 on the char after the last char) to a new, malloc-ed location.
218 strdupdelim (const char *beg, const char *end)
220 char *res = xmalloc (end - beg + 1);
221 memcpy (res, beg, end - beg);
222 res[end - beg] = '\0';
226 /* Parse a string containing comma-separated elements, and return a
227 vector of char pointers with the elements. Spaces following the
228 commas are ignored. */
230 sepstring (const char *s)
244 res = xrealloc (res, (i + 2) * sizeof (char *));
245 res[i] = strdupdelim (p, s);
248 /* Skip the blanks following the ','. */
249 while (c_isspace (*s))
256 res = xrealloc (res, (i + 2) * sizeof (char *));
257 res[i] = strdupdelim (p, s);
262 /* Like sprintf, but prints into a string of sufficient size freshly
263 allocated with malloc, which is returned. If unable to print due
264 to invalid format, returns NULL. Inability to allocate needed
265 memory results in abort, as with xmalloc. This is in spirit
266 similar to the GNU/BSD extension asprintf, but somewhat easier to
269 Internally the function either calls vasprintf or loops around
270 vsnprintf until the correct size is found. Since Wget also ships a
271 fallback implementation of vsnprintf, this should be portable. */
273 /* Constant is using for limits memory allocation for text buffer.
274 Applicable in situation when: vasprintf is not available in the system
275 and vsnprintf return -1 when long line is truncated (in old versions of
276 glibc and in other system where C99 doesn`t support) */
278 #define FMT_MAX_LENGTH 1048576
281 aprintf (const char *fmt, ...)
283 #if defined HAVE_VASPRINTF && !defined DEBUG_MALLOC
288 va_start (args, fmt);
289 ret = vasprintf (&str, fmt, args);
291 if (ret < 0 && errno == ENOMEM)
292 memfatal ("aprintf", UNKNOWN_ATTEMPTED_SIZE); /* for consistency
293 with xmalloc/xrealloc */
297 #else /* not HAVE_VASPRINTF */
299 /* vasprintf is unavailable. snprintf into a small buffer and
300 resize it as necessary. */
302 char *str = xmalloc (size);
304 /* #### This code will infloop and eventually abort in xrealloc if
305 passed a FMT that causes snprintf to consistently return -1. */
312 va_start (args, fmt);
313 n = vsnprintf (str, size, fmt, args);
316 /* If the printing worked, return the string. */
317 if (n > -1 && n < size)
320 /* Else try again with a larger buffer. */
321 if (n > -1) /* C99 */
322 size = n + 1; /* precisely what is needed */
323 else if (size >= FMT_MAX_LENGTH) /* We have a huge buffer, */
324 { /* maybe we have some wrong
326 logprintf (LOG_ALWAYS,
327 _("%s: aprintf: text buffer is too big (%ld bytes), "
329 exec_name, size); /* printout a log message */
330 abort (); /* and abort... */
334 /* else, we continue to grow our
335 * buffer: Twice the old size. */
338 str = xrealloc (str, size);
340 #endif /* not HAVE_VASPRINTF */
343 /* Concatenate the NULL-terminated list of string arguments into
344 freshly allocated space. */
347 concat_strings (const char *str0, ...)
350 int saved_lengths[5]; /* inspired by Apache's apr_pstrcat */
353 const char *next_str;
354 int total_length = 0;
357 /* Calculate the length of and allocate the resulting string. */
360 va_start (args, str0);
361 for (next_str = str0; next_str != NULL; next_str = va_arg (args, char *))
363 int len = strlen (next_str);
364 if (argcount < countof (saved_lengths))
365 saved_lengths[argcount++] = len;
369 p = ret = xmalloc (total_length + 1);
371 /* Copy the strings into the allocated space. */
374 va_start (args, str0);
375 for (next_str = str0; next_str != NULL; next_str = va_arg (args, char *))
378 if (argcount < countof (saved_lengths))
379 len = saved_lengths[argcount++];
381 len = strlen (next_str);
382 memcpy (p, next_str, len);
391 /* Format the provided time according to the specified format. The
392 format is a string with format elements supported by strftime. */
395 fmttime (time_t t, const char *fmt)
397 static char output[32];
398 struct tm *tm = localtime(&t);
401 if (!strftime(output, sizeof(output), fmt, tm))
406 /* Return pointer to a static char[] buffer in which zero-terminated
407 string-representation of TM (in form hh:mm:ss) is printed.
409 If TM is NULL, the current time will be used. */
414 return fmttime(t, "%H:%M:%S");
417 /* Like the above, but include the date: YYYY-MM-DD hh:mm:ss. */
420 datetime_str (time_t t)
422 return fmttime(t, "%Y-%m-%d %H:%M:%S");
425 /* The Windows versions of the following two functions are defined in
426 mswindows.c. On MSDOS this function should never be called. */
431 fork_to_background (void)
436 #else /* def __VMS */
438 #if !defined(WINDOWS) && !defined(MSDOS)
440 fork_to_background (void)
443 /* Whether we arrange our own version of opt.lfilename here. */
444 bool logfile_changed = false;
446 if (!opt.lfilename && (!opt.quiet || opt.server_response))
448 /* We must create the file immediately to avoid either a race
449 condition (which arises from using unique_name and failing to
450 use fopen_excl) or lying to the user about the log file name
451 (which arises from using unique_name, printing the name, and
452 using fopen_excl later on.) */
453 FILE *new_log_fp = unique_create (DEFAULT_LOGFILE, false, &opt.lfilename);
456 logfile_changed = true;
469 /* parent, no error */
470 printf (_("Continuing in background, pid %d.\n"), (int) pid);
472 printf (_("Output will be written to %s.\n"), quote (opt.lfilename));
473 exit (0); /* #### should we use _exit()? */
476 /* child: give up the privileges and keep running. */
478 freopen ("/dev/null", "r", stdin);
479 freopen ("/dev/null", "w", stdout);
480 freopen ("/dev/null", "w", stderr);
482 #endif /* !WINDOWS && !MSDOS */
484 #endif /* def __VMS [else] */
487 /* "Touch" FILE, i.e. make its mtime ("modified time") equal the time
488 specified with TM. The atime ("access time") is set to the current
492 touch (const char *file, time_t tm)
494 #ifdef HAVE_STRUCT_UTIMBUF
495 struct utimbuf times;
503 times.actime = time (NULL);
504 if (utime (file, ×) == -1)
505 logprintf (LOG_NOTQUIET, "utime(%s): %s\n", file, strerror (errno));
508 /* Checks if FILE is a symbolic link, and removes it if it is. Does
509 nothing under MS-Windows. */
511 remove_link (const char *file)
516 if (lstat (file, &st) == 0 && S_ISLNK (st.st_mode))
518 DEBUGP (("Unlinking %s (symlink).\n", file));
521 logprintf (LOG_VERBOSE, _("Failed to unlink symlink %s: %s\n"),
522 quote (file), strerror (errno));
527 /* Does FILENAME exist? This is quite a lousy implementation, since
528 it supplies no error codes -- only a yes-or-no answer. Thus it
529 will return that a file does not exist if, e.g., the directory is
530 unreadable. I don't mind it too much currently, though. The
531 proper way should, of course, be to have a third, error state,
532 other than true/false, but that would introduce uncalled-for
533 additional complexity to the callers. */
535 file_exists_p (const char *filename)
538 return access (filename, F_OK) >= 0;
541 return stat (filename, &buf) >= 0;
545 /* Returns 0 if PATH is a directory, 1 otherwise (any kind of file).
546 Returns 0 on error. */
548 file_non_directory_p (const char *path)
551 /* Use lstat() rather than stat() so that symbolic links pointing to
552 directories can be identified correctly. */
553 if (lstat (path, &buf) != 0)
555 return S_ISDIR (buf.st_mode) ? false : true;
558 /* Return the size of file named by FILENAME, or -1 if it cannot be
559 opened or seeked into. */
561 file_size (const char *filename)
563 #if defined(HAVE_FSEEKO) && defined(HAVE_FTELLO)
565 /* We use fseek rather than stat to determine the file size because
566 that way we can also verify that the file is readable without
567 explicitly checking for permissions. Inspired by the POST patch
569 FILE *fp = fopen (filename, "rb");
572 fseeko (fp, 0, SEEK_END);
578 if (stat (filename, &st) < 0)
585 If no UNIQ_SEP is defined (as on VMS), have unique_name() return the
586 original name. With the VMS file systems' versioning, everything
587 should be fine, and appending ".NN" just causes trouble.
592 /* stat file names named PREFIX.1, PREFIX.2, etc., until one that
593 doesn't exist is found. Return a freshly allocated copy of the
597 unique_name_1 (const char *prefix)
600 int plen = strlen (prefix);
601 char *template = (char *)alloca (plen + 1 + 24);
602 char *template_tail = template + plen;
604 memcpy (template, prefix, plen);
605 *template_tail++ = UNIQ_SEP;
608 number_to_string (template_tail, count++);
609 while (file_exists_p (template));
611 return xstrdup (template);
614 /* Return a unique file name, based on FILE.
616 More precisely, if FILE doesn't exist, it is returned unmodified.
617 If not, FILE.1 is tried, then FILE.2, etc. The first FILE.<number>
618 file name that doesn't exist is returned.
620 2005-02-19 SMS. "." is now UNIQ_SEP, and may be different.
622 The resulting file is not created, only verified that it didn't
623 exist at the point in time when the function was called.
624 Therefore, where security matters, don't rely that the file created
625 by this function exists until you open it with O_EXCL or
628 If ALLOW_PASSTHROUGH is 0, it always returns a freshly allocated
629 string. Otherwise, it may return FILE if the file doesn't exist
630 (and therefore doesn't need changing). */
633 unique_name (const char *file, bool allow_passthrough)
635 /* If the FILE itself doesn't exist, return it without
637 if (!file_exists_p (file))
638 return allow_passthrough ? (char *)file : xstrdup (file);
640 /* Otherwise, find a numeric suffix that results in unused file name
642 return unique_name_1 (file);
645 #else /* def UNIQ_SEP */
647 /* Dummy unique_name() for VMS. Return the original name as easily as
651 unique_name (const char *file, bool allow_passthrough)
653 /* Return the FILE itself, without modification, irregardful. */
654 return allow_passthrough ? (char *)file : xstrdup (file);
657 #endif /* def UNIQ_SEP [else] */
659 /* Create a file based on NAME, except without overwriting an existing
660 file with that name. Providing O_EXCL is correctly implemented,
661 this function does not have the race condition associated with
662 opening the file returned by unique_name. */
665 unique_create (const char *name, bool binary, char **opened_name)
667 /* unique file name, based on NAME */
668 char *uname = unique_name (name, false);
670 while ((fp = fopen_excl (uname, binary)) == NULL && errno == EEXIST)
673 uname = unique_name (name, false);
675 if (opened_name && fp != NULL)
678 *opened_name = uname;
690 /* Open the file for writing, with the addition that the file is
691 opened "exclusively". This means that, if the file already exists,
692 this function will *fail* and errno will be set to EEXIST. If
693 BINARY is set, the file will be opened in binary mode, equivalent
696 If opening the file fails for any reason, including the file having
697 previously existed, this function returns NULL and sets errno
701 fopen_excl (const char *fname, int binary)
707 VMS lacks O_BINARY, but makes up for it in weird and wonderful ways.
708 It also has file versions which obviate all the O_EXCL effort.
709 O_TRUNC (something of a misnomer) requests a new version.
712 /* Common open() optional arguments:
713 sequential access only, access callback function.
715 # define OPEN_OPT_ARGS "fop=sqo", "acc", acc_cb, &open_id
718 int flags = O_WRONLY | O_CREAT | O_TRUNC;
723 fd = open( fname, /* File name. */
725 0777, /* Mode for default protection. */
726 "ctx=bin,stm", /* Binary, stream access. */
727 "rfm=stmlf", /* Stream_LF. */
728 OPEN_OPT_ARGS); /* Access callback. */
733 fd = open( fname, /* File name. */
735 0777, /* Mode for default protection. */
736 "ctx=bin,stm", /* Binary, stream access. */
737 "rfm=fix", /* Fixed-length, */
738 "mrs=512", /* 512-byte records. */
739 OPEN_OPT_ARGS); /* Access callback. */
744 fd = open( fname, /* File name. */
746 0777, /* Mode for default protection.
748 "rfm=stmlf", /* Stream_LF. */
749 OPEN_OPT_ARGS); /* Access callback. */
751 # else /* def __VMS */
752 int flags = O_WRONLY | O_CREAT | O_EXCL;
757 fd = open (fname, flags, 0666);
758 # endif /* def __VMS [else] */
762 return fdopen (fd, binary ? "wb" : "w");
763 #else /* not O_EXCL */
764 /* Manually check whether the file exists. This is prone to race
765 conditions, but systems without O_EXCL haven't deserved
767 if (file_exists_p (fname))
772 return fopen (fname, binary ? "wb" : "w");
773 #endif /* not O_EXCL */
776 /* Create DIRECTORY. If some of the pathname components of DIRECTORY
777 are missing, create them first. In case any mkdir() call fails,
778 return its error status. Returns 0 on successful completion.
780 The behaviour of this function should be identical to the behaviour
781 of `mkdir -p' on systems where mkdir supports the `-p' option. */
783 make_directory (const char *directory)
785 int i, ret, quit = 0;
788 /* Make a copy of dir, to be able to write to it. Otherwise, the
789 function is unsafe if called with a read-only char *argument. */
790 STRDUP_ALLOCA (dir, directory);
792 /* If the first character of dir is '/', skip it (and thus enable
793 creation of absolute-pathname directories. */
794 for (i = (*dir == '/'); 1; ++i)
796 for (; dir[i] && dir[i] != '/'; i++)
801 /* Check whether the directory already exists. Allow creation of
802 of intermediate directories to fail, as the initial path components
803 are not necessarily directories! */
804 if (!file_exists_p (dir))
805 ret = mkdir (dir, 0777);
816 /* Merge BASE with FILE. BASE can be a directory or a file name, FILE
817 should be a file name.
819 file_merge("/foo/bar", "baz") => "/foo/baz"
820 file_merge("/foo/bar/", "baz") => "/foo/bar/baz"
821 file_merge("foo", "bar") => "bar"
823 In other words, it's a simpler and gentler version of uri_merge. */
826 file_merge (const char *base, const char *file)
829 const char *cut = (const char *)strrchr (base, '/');
832 return xstrdup (file);
834 result = xmalloc (cut - base + 1 + strlen (file) + 1);
835 memcpy (result, base, cut - base);
836 result[cut - base] = '/';
837 strcpy (result + (cut - base) + 1, file);
842 /* Like fnmatch, but performs a case-insensitive match. */
845 fnmatch_nocase (const char *pattern, const char *string, int flags)
848 /* The FNM_CASEFOLD flag started as a GNU extension, but it is now
849 also present on *BSD platforms, and possibly elsewhere. */
850 return fnmatch (pattern, string, flags | FNM_CASEFOLD);
852 /* Turn PATTERN and STRING to lower case and call fnmatch on them. */
853 char *patcopy = (char *) alloca (strlen (pattern) + 1);
854 char *strcopy = (char *) alloca (strlen (string) + 1);
856 for (p = patcopy; *pattern; pattern++, p++)
857 *p = c_tolower (*pattern);
859 for (p = strcopy; *string; string++, p++)
860 *p = c_tolower (*string);
862 return fnmatch (patcopy, strcopy, flags);
866 static bool in_acclist (const char *const *, const char *, bool);
868 /* Determine whether a file is acceptable to be followed, according to
869 lists of patterns to accept/reject. */
871 acceptable (const char *s)
875 while (l && s[l] != '/')
882 return (in_acclist ((const char *const *)opt.accepts, s, true)
883 && !in_acclist ((const char *const *)opt.rejects, s, true));
885 return in_acclist ((const char *const *)opt.accepts, s, true);
887 else if (opt.rejects)
888 return !in_acclist ((const char *const *)opt.rejects, s, true);
892 /* Check if D2 is a subdirectory of D1. E.g. if D1 is `/something', subdir_p()
893 will return true if and only if D2 begins with `/something/' or is exactly
896 subdir_p (const char *d1, const char *d2)
900 if (!opt.ignore_case)
901 for (; *d1 && *d2 && (*d1 == *d2); ++d1, ++d2)
904 for (; *d1 && *d2 && (c_tolower (*d1) == c_tolower (*d2)); ++d1, ++d2)
907 return *d1 == '\0' && (*d2 == '\0' || *d2 == '/');
910 /* Iterate through DIRLIST (which must be NULL-terminated), and return the
911 first element that matches DIR, through wildcards or front comparison (as
914 dir_matches_p (char **dirlist, const char *dir)
917 int (*matcher) (const char *, const char *, int)
918 = opt.ignore_case ? fnmatch_nocase : fnmatch;
920 for (x = dirlist; *x; x++)
922 /* Remove leading '/' */
923 char *p = *x + (**x == '/');
924 if (has_wildcards_p (p))
926 if (matcher (p, dir, FNM_PATHNAME) == 0)
931 if (subdir_p (p, dir))
936 return *x ? true : false;
939 /* Returns whether DIRECTORY is acceptable for download, wrt the
940 include/exclude lists.
942 The leading `/' is ignored in paths; relative and absolute paths
943 may be freely intermixed. */
946 accdir (const char *directory)
948 /* Remove starting '/'. */
949 if (*directory == '/')
953 if (!dir_matches_p (opt.includes, directory))
958 if (dir_matches_p (opt.excludes, directory))
964 /* Return true if STRING ends with TAIL. For instance:
966 match_tail ("abc", "bc", false) -> 1
967 match_tail ("abc", "ab", false) -> 0
968 match_tail ("abc", "abc", false) -> 1
970 If FOLD_CASE is true, the comparison will be case-insensitive. */
973 match_tail (const char *string, const char *tail, bool fold_case)
977 /* We want this to be fast, so we code two loops, one with
978 case-folding, one without. */
982 for (i = strlen (string), j = strlen (tail); i >= 0 && j >= 0; i--, j--)
983 if (string[i] != tail[j])
988 for (i = strlen (string), j = strlen (tail); i >= 0 && j >= 0; i--, j--)
989 if (c_tolower (string[i]) != c_tolower (tail[j]))
993 /* If the tail was exhausted, the match was succesful. */
1000 /* Checks whether string S matches each element of ACCEPTS. A list
1001 element are matched either with fnmatch() or match_tail(),
1002 according to whether the element contains wildcards or not.
1004 If the BACKWARD is false, don't do backward comparison -- just compare
1007 in_acclist (const char *const *accepts, const char *s, bool backward)
1009 for (; *accepts; accepts++)
1011 if (has_wildcards_p (*accepts))
1013 int res = opt.ignore_case
1014 ? fnmatch_nocase (*accepts, s, 0) : fnmatch (*accepts, s, 0);
1015 /* fnmatch returns 0 if the pattern *does* match the string. */
1023 if (match_tail (s, *accepts, opt.ignore_case))
1028 int cmp = opt.ignore_case
1029 ? strcasecmp (s, *accepts) : strcmp (s, *accepts);
1038 /* Return the location of STR's suffix (file extension). Examples:
1039 suffix ("foo.bar") -> "bar"
1040 suffix ("foo.bar.baz") -> "baz"
1041 suffix ("/foo/bar") -> NULL
1042 suffix ("/foo.bar/baz") -> NULL */
1044 suffix (const char *str)
1048 for (i = strlen (str); i && str[i] != '/' && str[i] != '.'; i--)
1051 if (str[i++] == '.')
1052 return (char *)str + i;
1057 /* Return true if S contains globbing wildcards (`*', `?', `[' or
1061 has_wildcards_p (const char *s)
1064 if (*s == '*' || *s == '?' || *s == '[' || *s == ']')
1069 /* Return true if FNAME ends with a typical HTML suffix. The
1070 following (case-insensitive) suffixes are presumed to be HTML
1075 ?html (`?' matches one character)
1077 #### CAVEAT. This is not necessarily a good indication that FNAME
1078 refers to a file that contains HTML! */
1080 has_html_suffix_p (const char *fname)
1084 if ((suf = suffix (fname)) == NULL)
1086 if (!strcasecmp (suf, "html"))
1088 if (!strcasecmp (suf, "htm"))
1090 if (suf[0] && !strcasecmp (suf + 1, "html"))
1095 /* Read a line from FP and return the pointer to freshly allocated
1096 storage. The storage space is obtained through malloc() and should
1097 be freed with free() when it is no longer needed.
1099 The length of the line is not limited, except by available memory.
1100 The newline character at the end of line is retained. The line is
1101 terminated with a zero character.
1103 After end-of-file is encountered without anything being read, NULL
1104 is returned. NULL is also returned on error. To distinguish
1105 between these two cases, use the stdio function ferror(). */
1108 read_whole_line (FILE *fp)
1112 char *line = xmalloc (bufsize);
1114 while (fgets (line + length, bufsize - length, fp))
1116 length += strlen (line + length);
1118 /* Possible for example when reading from a binary file where
1119 a line begins with \0. */
1122 if (line[length - 1] == '\n')
1125 /* fgets() guarantees to read the whole line, or to use up the
1126 space we've given it. We can double the buffer
1129 line = xrealloc (line, bufsize);
1131 if (length == 0 || ferror (fp))
1136 if (length + 1 < bufsize)
1137 /* Relieve the memory from our exponential greediness. We say
1138 `length + 1' because the terminating \0 is not included in
1139 LENGTH. We don't need to zero-terminate the string ourselves,
1140 though, because fgets() does that. */
1141 line = xrealloc (line, length + 1);
1145 /* Read FILE into memory. A pointer to `struct file_memory' are
1146 returned; use struct element `content' to access file contents, and
1147 the element `length' to know the file length. `content' is *not*
1148 zero-terminated, and you should *not* read or write beyond the [0,
1149 length) range of characters.
1151 After you are done with the file contents, call read_file_free to
1154 Depending on the operating system and the type of file that is
1155 being read, read_file() either mmap's the file into memory, or
1156 reads the file into the core using read().
1158 If file is named "-", fileno(stdin) is used for reading instead.
1159 If you want to read from a real file named "-", use "./-" instead. */
1161 struct file_memory *
1162 read_file (const char *file)
1165 struct file_memory *fm;
1167 bool inhibit_close = false;
1169 /* Some magic in the finest tradition of Perl and its kin: if FILE
1170 is "-", just use stdin. */
1173 fd = fileno (stdin);
1174 inhibit_close = true;
1175 /* Note that we don't inhibit mmap() in this case. If stdin is
1176 redirected from a regular file, mmap() will still work. */
1179 fd = open (file, O_RDONLY);
1182 fm = xnew (struct file_memory);
1187 if (fstat (fd, &buf) < 0)
1189 fm->length = buf.st_size;
1190 /* NOTE: As far as I know, the callers of this function never
1191 modify the file text. Relying on this would enable us to
1192 specify PROT_READ and MAP_SHARED for a marginal gain in
1193 efficiency, but at some cost to generality. */
1194 fm->content = mmap (NULL, fm->length, PROT_READ | PROT_WRITE,
1195 MAP_PRIVATE, fd, 0);
1196 if (fm->content == (char *)MAP_FAILED)
1206 /* The most common reason why mmap() fails is that FD does not point
1207 to a plain file. However, it's also possible that mmap() doesn't
1208 work for a particular type of file. Therefore, whenever mmap()
1209 fails, we just fall back to the regular method. */
1210 #endif /* HAVE_MMAP */
1213 size = 512; /* number of bytes fm->contents can
1214 hold at any given time. */
1215 fm->content = xmalloc (size);
1219 if (fm->length > size / 2)
1221 /* #### I'm not sure whether the whole exponential-growth
1222 thing makes sense with kernel read. On Linux at least,
1223 read() refuses to read more than 4K from a file at a
1224 single chunk anyway. But other Unixes might optimize it
1225 better, and it doesn't *hurt* anything, so I'm leaving
1228 /* Normally, we grow SIZE exponentially to make the number
1229 of calls to read() and realloc() logarithmic in relation
1230 to file size. However, read() can read an amount of data
1231 smaller than requested, and it would be unreasonable to
1232 double SIZE every time *something* was read. Therefore,
1233 we double SIZE only when the length exceeds half of the
1234 entire allocated size. */
1236 fm->content = xrealloc (fm->content, size);
1238 nread = read (fd, fm->content + fm->length, size - fm->length);
1240 /* Successful read. */
1241 fm->length += nread;
1251 if (size > fm->length && fm->length != 0)
1252 /* Due to exponential growth of fm->content, the allocated region
1253 might be much larger than what is actually needed. */
1254 fm->content = xrealloc (fm->content, fm->length);
1261 xfree (fm->content);
1266 /* Release the resources held by FM. Specifically, this calls
1267 munmap() or xfree() on fm->content, depending whether mmap or
1268 malloc/read were used to read in the file. It also frees the
1269 memory needed to hold the FM structure itself. */
1272 read_file_free (struct file_memory *fm)
1277 munmap (fm->content, fm->length);
1282 xfree (fm->content);
1287 /* Free the pointers in a NULL-terminated vector of pointers, then
1288 free the pointer itself. */
1290 free_vec (char **vec)
1301 /* Append vector V2 to vector V1. The function frees V2 and
1302 reallocates V1 (thus you may not use the contents of neither
1303 pointer after the call). If V1 is NULL, V2 is returned. */
1305 merge_vecs (char **v1, char **v2)
1315 /* To avoid j == 0 */
1320 for (i = 0; v1[i]; i++)
1323 for (j = 0; v2[j]; j++)
1325 /* Reallocate v1. */
1326 v1 = xrealloc (v1, (i + j + 1) * sizeof (char **));
1327 memcpy (v1 + i, v2, (j + 1) * sizeof (char *));
1332 /* Append a freshly allocated copy of STR to VEC. If VEC is NULL, it
1333 is allocated as needed. Return the new value of the vector. */
1336 vec_append (char **vec, const char *str)
1338 int cnt; /* count of vector elements, including
1339 the one we're about to append */
1342 for (cnt = 0; vec[cnt]; cnt++)
1348 /* Reallocate the array to fit the new element and the NULL. */
1349 vec = xrealloc (vec, (cnt + 1) * sizeof (char *));
1350 /* Append a copy of STR to the vector. */
1351 vec[cnt - 1] = xstrdup (str);
1356 /* Sometimes it's useful to create "sets" of strings, i.e. special
1357 hash tables where you want to store strings as keys and merely
1358 query for their existence. Here is a set of utility routines that
1359 makes that transparent. */
1362 string_set_add (struct hash_table *ht, const char *s)
1364 /* First check whether the set element already exists. If it does,
1365 do nothing so that we don't have to free() the old element and
1366 then strdup() a new one. */
1367 if (hash_table_contains (ht, s))
1370 /* We use "1" as value. It provides us a useful and clear arbitrary
1371 value, and it consumes no memory -- the pointers to the same
1372 string "1" will be shared by all the key-value pairs in all `set'
1374 hash_table_put (ht, xstrdup (s), "1");
1377 /* Synonym for hash_table_contains... */
1380 string_set_contains (struct hash_table *ht, const char *s)
1382 return hash_table_contains (ht, s);
1385 /* Convert the specified string set to array. ARRAY should be large
1386 enough to hold hash_table_count(ht) char pointers. */
1388 void string_set_to_array (struct hash_table *ht, char **array)
1390 hash_table_iterator iter;
1391 for (hash_table_iterate (ht, &iter); hash_table_iter_next (&iter); )
1392 *array++ = iter.key;
1395 /* Free the string set. This frees both the storage allocated for
1396 keys and the actual hash table. (hash_table_destroy would only
1397 destroy the hash table.) */
1400 string_set_free (struct hash_table *ht)
1402 hash_table_iterator iter;
1403 for (hash_table_iterate (ht, &iter); hash_table_iter_next (&iter); )
1405 hash_table_destroy (ht);
1408 /* Utility function: simply call xfree() on all keys and values of HT. */
1411 free_keys_and_values (struct hash_table *ht)
1413 hash_table_iterator iter;
1414 for (hash_table_iterate (ht, &iter); hash_table_iter_next (&iter); )
1421 /* Get digit grouping data for thousand separors by calling
1422 localeconv(). The data includes separator string and grouping info
1423 and is cached after the first call to the function.
1425 In locales that don't set a thousand separator (such as the "C"
1426 locale), this forces it to be ",". We are now only showing
1427 thousand separators in one place, so this shouldn't be a problem in
1431 get_grouping_data (const char **sep, const char **grouping)
1433 static const char *cached_sep;
1434 static const char *cached_grouping;
1435 static bool initialized;
1438 /* Get the grouping info from the locale. */
1439 struct lconv *lconv = localeconv ();
1440 cached_sep = lconv->thousands_sep;
1441 cached_grouping = lconv->grouping;
1442 #if ! USE_NLS_PROGRESS_BAR
1443 /* We can't count column widths, so ensure that the separator
1444 * is single-byte only (let check below determine what byte). */
1445 if (strlen(cached_sep) > 1)
1450 /* Many locales (such as "C" or "hr_HR") don't specify
1451 grouping, which we still want to use it for legibility.
1452 In those locales set the sep char to ',', unless that
1453 character is used for decimal point, in which case set it
1455 if (*lconv->decimal_point != ',')
1459 cached_grouping = "\x03";
1464 *grouping = cached_grouping;
1467 /* Return a printed representation of N with thousand separators.
1468 This should respect locale settings, with the exception of the "C"
1469 locale which mandates no separator, but we use one anyway.
1471 Unfortunately, we cannot use %'d (in fact it would be %'j) to get
1472 the separators because it's too non-portable, and it's hard to test
1473 for this feature at configure time. Besides, it wouldn't display
1474 separators in the "C" locale, still used by many Unix users. */
1477 with_thousand_seps (wgint n)
1479 static char outbuf[48];
1480 char *p = outbuf + sizeof outbuf;
1482 /* Info received from locale */
1483 const char *grouping, *sep;
1486 /* State information */
1487 int i = 0, groupsize;
1488 const char *atgroup;
1490 bool negative = n < 0;
1492 /* Initialize grouping data. */
1493 get_grouping_data (&sep, &grouping);
1494 seplen = strlen (sep);
1496 groupsize = *atgroup++;
1498 /* This would overflow on WGINT_MIN, but printing negative numbers
1499 is not an important goal of this fuinction. */
1503 /* Write the number into the buffer, backwards, inserting the
1504 separators as necessary. */
1508 *--p = n % 10 + '0';
1512 /* Prepend SEP to every groupsize'd digit and get new groupsize. */
1513 if (++i == groupsize)
1518 memcpy (p -= seplen, sep, seplen);
1521 groupsize = *atgroup++;
1530 /* N, a byte quantity, is converted to a human-readable abberviated
1531 form a la sizes printed by `ls -lh'. The result is written to a
1532 static buffer, a pointer to which is returned.
1534 Unlike `with_thousand_seps', this approximates to the nearest unit.
1535 Quoting GNU libit: "Most people visually process strings of 3-4
1536 digits effectively, but longer strings of digits are more prone to
1537 misinterpretation. Hence, converting to an abbreviated form
1538 usually improves readability."
1540 This intentionally uses kilobyte (KB), megabyte (MB), etc. in their
1541 original computer-related meaning of "powers of 1024". We don't
1542 use the "*bibyte" names invented in 1998, and seldom used in
1543 practice. Wikipedia's entry on "binary prefix" discusses this in
1547 human_readable (HR_NUMTYPE n)
1549 /* These suffixes are compatible with those of GNU `ls -lh'. */
1550 static char powers[] =
1552 'K', /* kilobyte, 2^10 bytes */
1553 'M', /* megabyte, 2^20 bytes */
1554 'G', /* gigabyte, 2^30 bytes */
1555 'T', /* terabyte, 2^40 bytes */
1556 'P', /* petabyte, 2^50 bytes */
1557 'E', /* exabyte, 2^60 bytes */
1562 /* If the quantity is smaller than 1K, just print it. */
1565 snprintf (buf, sizeof (buf), "%d", (int) n);
1569 /* Loop over powers, dividing N with 1024 in each iteration. This
1570 works unchanged for all sizes of wgint, while still avoiding
1571 non-portable `long double' arithmetic. */
1572 for (i = 0; i < countof (powers); i++)
1574 /* At each iteration N is greater than the *subsequent* power.
1575 That way N/1024.0 produces a decimal number in the units of
1577 if ((n / 1024) < 1024 || i == countof (powers) - 1)
1579 double val = n / 1024.0;
1580 /* Print values smaller than 10 with one decimal digits, and
1581 others without any decimals. */
1582 snprintf (buf, sizeof (buf), "%.*f%c",
1583 val < 10 ? 1 : 0, val, powers[i]);
1588 return NULL; /* unreached */
1591 /* Count the digits in the provided number. Used to allocate space
1592 when printing numbers. */
1595 numdigit (wgint number)
1599 ++cnt; /* accomodate '-' */
1600 while ((number /= 10) != 0)
1605 #define PR(mask) *p++ = n / (mask) + '0'
1607 /* DIGITS_<D> is used to print a D-digit number and should be called
1608 with mask==10^(D-1). It prints n/mask (the first digit), reducing
1609 n to n%mask (the remaining digits), and calling DIGITS_<D-1>.
1610 Recursively this continues until DIGITS_1 is invoked. */
1612 #define DIGITS_1(mask) PR (mask)
1613 #define DIGITS_2(mask) PR (mask), n %= (mask), DIGITS_1 ((mask) / 10)
1614 #define DIGITS_3(mask) PR (mask), n %= (mask), DIGITS_2 ((mask) / 10)
1615 #define DIGITS_4(mask) PR (mask), n %= (mask), DIGITS_3 ((mask) / 10)
1616 #define DIGITS_5(mask) PR (mask), n %= (mask), DIGITS_4 ((mask) / 10)
1617 #define DIGITS_6(mask) PR (mask), n %= (mask), DIGITS_5 ((mask) / 10)
1618 #define DIGITS_7(mask) PR (mask), n %= (mask), DIGITS_6 ((mask) / 10)
1619 #define DIGITS_8(mask) PR (mask), n %= (mask), DIGITS_7 ((mask) / 10)
1620 #define DIGITS_9(mask) PR (mask), n %= (mask), DIGITS_8 ((mask) / 10)
1621 #define DIGITS_10(mask) PR (mask), n %= (mask), DIGITS_9 ((mask) / 10)
1623 /* DIGITS_<11-20> are only used on machines with 64-bit wgints. */
1625 #define DIGITS_11(mask) PR (mask), n %= (mask), DIGITS_10 ((mask) / 10)
1626 #define DIGITS_12(mask) PR (mask), n %= (mask), DIGITS_11 ((mask) / 10)
1627 #define DIGITS_13(mask) PR (mask), n %= (mask), DIGITS_12 ((mask) / 10)
1628 #define DIGITS_14(mask) PR (mask), n %= (mask), DIGITS_13 ((mask) / 10)
1629 #define DIGITS_15(mask) PR (mask), n %= (mask), DIGITS_14 ((mask) / 10)
1630 #define DIGITS_16(mask) PR (mask), n %= (mask), DIGITS_15 ((mask) / 10)
1631 #define DIGITS_17(mask) PR (mask), n %= (mask), DIGITS_16 ((mask) / 10)
1632 #define DIGITS_18(mask) PR (mask), n %= (mask), DIGITS_17 ((mask) / 10)
1633 #define DIGITS_19(mask) PR (mask), n %= (mask), DIGITS_18 ((mask) / 10)
1635 /* Shorthand for casting to wgint. */
1638 /* Print NUMBER to BUFFER in base 10. This is equivalent to
1639 `sprintf(buffer, "%lld", (long long) number)', only typically much
1640 faster and portable to machines without long long.
1642 The speedup may make a difference in programs that frequently
1643 convert numbers to strings. Some implementations of sprintf,
1644 particularly the one in some versions of GNU libc, have been known
1645 to be quite slow when converting integers to strings.
1647 Return the pointer to the location where the terminating zero was
1648 printed. (Equivalent to calling buffer+strlen(buffer) after the
1651 BUFFER should be large enough to accept as many bytes as you expect
1652 the number to take up. On machines with 64-bit wgints the maximum
1653 needed size is 24 bytes. That includes the digits needed for the
1654 largest 64-bit number, the `-' sign in case it's negative, and the
1655 terminating '\0'. */
1658 number_to_string (char *buffer, wgint number)
1663 int last_digit_char = 0;
1665 #if (SIZEOF_WGINT != 4) && (SIZEOF_WGINT != 8)
1666 /* We are running in a very strange environment. Leave the correct
1667 printing to sprintf. */
1668 p += sprintf (buf, "%j", (intmax_t) (n));
1669 #else /* (SIZEOF_WGINT == 4) || (SIZEOF_WGINT == 8) */
1675 /* n = -n would overflow because -n would evaluate to a
1676 wgint value larger than WGINT_MAX. Need to make n
1677 smaller and handle the last digit separately. */
1678 int last_digit = n % 10;
1679 /* The sign of n%10 is implementation-defined. */
1681 last_digit_char = '0' - last_digit;
1683 last_digit_char = '0' + last_digit;
1684 /* After n is made smaller, -n will not overflow. */
1692 /* Use the DIGITS_ macro appropriate for N's number of digits. That
1693 way printing any N is fully open-coded without a loop or jump.
1694 (Also see description of DIGITS_*.) */
1696 if (n < 10) DIGITS_1 (1);
1697 else if (n < 100) DIGITS_2 (10);
1698 else if (n < 1000) DIGITS_3 (100);
1699 else if (n < 10000) DIGITS_4 (1000);
1700 else if (n < 100000) DIGITS_5 (10000);
1701 else if (n < 1000000) DIGITS_6 (100000);
1702 else if (n < 10000000) DIGITS_7 (1000000);
1703 else if (n < 100000000) DIGITS_8 (10000000);
1704 else if (n < 1000000000) DIGITS_9 (100000000);
1705 #if SIZEOF_WGINT == 4
1706 /* wgint is 32 bits wide: no number has more than 10 digits. */
1707 else DIGITS_10 (1000000000);
1709 /* wgint is 64 bits wide: handle numbers with 9-19 decimal digits.
1710 Constants are constructed by compile-time multiplication to avoid
1711 dealing with different notations for 64-bit constants
1712 (nL/nLL/nI64, depending on the compiler and architecture). */
1713 else if (n < 10*(W)1000000000) DIGITS_10 (1000000000);
1714 else if (n < 100*(W)1000000000) DIGITS_11 (10*(W)1000000000);
1715 else if (n < 1000*(W)1000000000) DIGITS_12 (100*(W)1000000000);
1716 else if (n < 10000*(W)1000000000) DIGITS_13 (1000*(W)1000000000);
1717 else if (n < 100000*(W)1000000000) DIGITS_14 (10000*(W)1000000000);
1718 else if (n < 1000000*(W)1000000000) DIGITS_15 (100000*(W)1000000000);
1719 else if (n < 10000000*(W)1000000000) DIGITS_16 (1000000*(W)1000000000);
1720 else if (n < 100000000*(W)1000000000) DIGITS_17 (10000000*(W)1000000000);
1721 else if (n < 1000000000*(W)1000000000) DIGITS_18 (100000000*(W)1000000000);
1722 else DIGITS_19 (1000000000*(W)1000000000);
1725 if (last_digit_char)
1726 *p++ = last_digit_char;
1729 #endif /* (SIZEOF_WGINT == 4) || (SIZEOF_WGINT == 8) */
1736 #undef SPRINTF_WGINT
1759 /* Print NUMBER to a statically allocated string and return a pointer
1760 to the printed representation.
1762 This function is intended to be used in conjunction with printf.
1763 It is hard to portably print wgint values:
1764 a) you cannot use printf("%ld", number) because wgint can be long
1765 long on 32-bit machines with LFS.
1766 b) you cannot use printf("%lld", number) because NUMBER could be
1767 long on 32-bit machines without LFS, or on 64-bit machines,
1768 which do not require LFS. Also, Windows doesn't support %lld.
1769 c) you cannot use printf("%j", (int_max_t) number) because not all
1770 versions of printf support "%j", the most notable being the one
1772 d) you cannot #define WGINT_FMT to the appropriate format and use
1773 printf(WGINT_FMT, number) because that would break translations
1774 for user-visible messages, such as printf("Downloaded: %d
1777 What you should use instead is printf("%s", number_to_static_string
1780 CAVEAT: since the function returns pointers to static data, you
1781 must be careful to copy its result before calling it again.
1782 However, to make it more useful with printf, the function maintains
1783 an internal ring of static buffers to return. That way things like
1784 printf("%s %s", number_to_static_string (num1),
1785 number_to_static_string (num2)) work as expected. Three buffers
1786 are currently used, which means that "%s %s %s" will work, but "%s
1787 %s %s %s" won't. If you need to print more than three wgints,
1788 bump the RING_SIZE (or rethink your message.) */
1791 number_to_static_string (wgint number)
1793 static char ring[RING_SIZE][24];
1795 char *buf = ring[ringpos];
1796 number_to_string (buf, number);
1797 ringpos = (ringpos + 1) % RING_SIZE;
1801 /* Determine the width of the terminal we're running on. If that's
1802 not possible, return 0. */
1805 determine_screen_width (void)
1807 /* If there's a way to get the terminal size using POSIX
1808 tcgetattr(), somebody please tell me. */
1813 if (opt.lfilename != NULL)
1816 fd = fileno (stderr);
1817 if (ioctl (fd, TIOCGWINSZ, &wsz) < 0)
1818 return 0; /* most likely ENOTTY */
1821 #elif defined(WINDOWS)
1822 CONSOLE_SCREEN_BUFFER_INFO csbi;
1823 if (!GetConsoleScreenBufferInfo (GetStdHandle (STD_ERROR_HANDLE), &csbi))
1825 return csbi.dwSize.X;
1826 #else /* neither TIOCGWINSZ nor WINDOWS */
1828 #endif /* neither TIOCGWINSZ nor WINDOWS */
1831 /* Whether the rnd system (either rand or [dl]rand48) has been
1833 static int rnd_seeded;
1835 /* Return a random number between 0 and MAX-1, inclusive.
1837 If the system does not support lrand48 and MAX is greater than the
1838 value of RAND_MAX+1 on the system, the returned value will be in
1839 the range [0, RAND_MAX]. This may be fixed in a future release.
1840 The random number generator is seeded automatically the first time
1843 This uses lrand48 where available, rand elsewhere. DO NOT use it
1844 for cryptography. It is only meant to be used in situations where
1845 quality of the random numbers returned doesn't really matter. */
1848 random_number (int max)
1853 srand48 ((long) time (NULL) ^ (long) getpid ());
1856 return lrand48 () % max;
1857 #else /* not HAVE_DRAND48 */
1863 srand ((unsigned) time (NULL) ^ (unsigned) getpid ());
1868 /* Like rand() % max, but uses the high-order bits for better
1869 randomness on architectures where rand() is implemented using a
1870 simple congruential generator. */
1872 bounded = (double) max * rnd / (RAND_MAX + 1.0);
1873 return (int) bounded;
1875 #endif /* not HAVE_DRAND48 */
1878 /* Return a random uniformly distributed floating point number in the
1879 [0, 1) range. Uses drand48 where available, and a really lame
1880 kludge elsewhere. */
1888 srand48 ((long) time (NULL) ^ (long) getpid ());
1892 #else /* not HAVE_DRAND48 */
1893 return ( random_number (10000) / 10000.0
1894 + random_number (10000) / (10000.0 * 10000.0)
1895 + random_number (10000) / (10000.0 * 10000.0 * 10000.0)
1896 + random_number (10000) / (10000.0 * 10000.0 * 10000.0 * 10000.0));
1897 #endif /* not HAVE_DRAND48 */
1900 /* Implementation of run_with_timeout, a generic timeout-forcing
1901 routine for systems with Unix-like signal handling. */
1903 #ifdef USE_SIGNAL_TIMEOUT
1904 # ifdef HAVE_SIGSETJMP
1905 # define SETJMP(env) sigsetjmp (env, 1)
1907 static sigjmp_buf run_with_timeout_env;
1910 abort_run_with_timeout (int sig)
1912 assert (sig == SIGALRM);
1913 siglongjmp (run_with_timeout_env, -1);
1915 # else /* not HAVE_SIGSETJMP */
1916 # define SETJMP(env) setjmp (env)
1918 static jmp_buf run_with_timeout_env;
1921 abort_run_with_timeout (int sig)
1923 assert (sig == SIGALRM);
1924 /* We don't have siglongjmp to preserve the set of blocked signals;
1925 if we longjumped out of the handler at this point, SIGALRM would
1926 remain blocked. We must unblock it manually. */
1927 int mask = siggetmask ();
1928 mask &= ~sigmask (SIGALRM);
1931 /* Now it's safe to longjump. */
1932 longjmp (run_with_timeout_env, -1);
1934 # endif /* not HAVE_SIGSETJMP */
1936 /* Arrange for SIGALRM to be delivered in TIMEOUT seconds. This uses
1937 setitimer where available, alarm otherwise.
1939 TIMEOUT should be non-zero. If the timeout value is so small that
1940 it would be rounded to zero, it is rounded to the least legal value
1941 instead (1us for setitimer, 1s for alarm). That ensures that
1942 SIGALRM will be delivered in all cases. */
1945 alarm_set (double timeout)
1948 /* Use the modern itimer interface. */
1949 struct itimerval itv;
1951 itv.it_value.tv_sec = (long) timeout;
1952 itv.it_value.tv_usec = 1000000 * (timeout - (long)timeout);
1953 if (itv.it_value.tv_sec == 0 && itv.it_value.tv_usec == 0)
1954 /* Ensure that we wait for at least the minimum interval.
1955 Specifying zero would mean "wait forever". */
1956 itv.it_value.tv_usec = 1;
1957 setitimer (ITIMER_REAL, &itv, NULL);
1958 #else /* not ITIMER_REAL */
1959 /* Use the old alarm() interface. */
1960 int secs = (int) timeout;
1962 /* Round TIMEOUTs smaller than 1 to 1, not to zero. This is
1963 because alarm(0) means "never deliver the alarm", i.e. "wait
1964 forever", which is not what someone who specifies a 0.5s
1965 timeout would expect. */
1968 #endif /* not ITIMER_REAL */
1971 /* Cancel the alarm set with alarm_set. */
1977 struct itimerval disable;
1979 setitimer (ITIMER_REAL, &disable, NULL);
1980 #else /* not ITIMER_REAL */
1982 #endif /* not ITIMER_REAL */
1985 /* Call FUN(ARG), but don't allow it to run for more than TIMEOUT
1986 seconds. Returns true if the function was interrupted with a
1987 timeout, false otherwise.
1989 This works by setting up SIGALRM to be delivered in TIMEOUT seconds
1990 using setitimer() or alarm(). The timeout is enforced by
1991 longjumping out of the SIGALRM handler. This has several
1992 advantages compared to the traditional approach of relying on
1993 signals causing system calls to exit with EINTR:
1995 * The callback function is *forcibly* interrupted after the
1996 timeout expires, (almost) regardless of what it was doing and
1997 whether it was in a syscall. For example, a calculation that
1998 takes a long time is interrupted as reliably as an IO
2001 * It works with both SYSV and BSD signals because it doesn't
2002 depend on the default setting of SA_RESTART.
2004 * It doesn't require special handler setup beyond a simple call
2005 to signal(). (It does use sigsetjmp/siglongjmp, but they're
2008 The only downside is that, if FUN allocates internal resources that
2009 are normally freed prior to exit from the functions, they will be
2010 lost in case of timeout. */
2013 run_with_timeout (double timeout, void (*fun) (void *), void *arg)
2023 signal (SIGALRM, abort_run_with_timeout);
2024 if (SETJMP (run_with_timeout_env) != 0)
2026 /* Longjumped out of FUN with a timeout. */
2027 signal (SIGALRM, SIG_DFL);
2030 alarm_set (timeout);
2033 /* Preserve errno in case alarm() or signal() modifies it. */
2034 saved_errno = errno;
2036 signal (SIGALRM, SIG_DFL);
2037 errno = saved_errno;
2042 #else /* not USE_SIGNAL_TIMEOUT */
2045 /* A stub version of run_with_timeout that just calls FUN(ARG). Don't
2046 define it under Windows, because Windows has its own version of
2047 run_with_timeout that uses threads. */
2050 run_with_timeout (double timeout, void (*fun) (void *), void *arg)
2055 #endif /* not WINDOWS */
2056 #endif /* not USE_SIGNAL_TIMEOUT */
2060 /* Sleep the specified amount of seconds. On machines without
2061 nanosleep(), this may sleep shorter if interrupted by signals. */
2064 xsleep (double seconds)
2066 #ifdef HAVE_NANOSLEEP
2067 /* nanosleep is the preferred interface because it offers high
2068 accuracy and, more importantly, because it allows us to reliably
2069 restart receiving a signal such as SIGWINCH. (There was an
2070 actual Debian bug report about --limit-rate malfunctioning while
2071 the terminal was being resized.) */
2072 struct timespec sleep, remaining;
2073 sleep.tv_sec = (long) seconds;
2074 sleep.tv_nsec = 1000000000 * (seconds - (long) seconds);
2075 while (nanosleep (&sleep, &remaining) < 0 && errno == EINTR)
2076 /* If nanosleep has been interrupted by a signal, adjust the
2077 sleeping period and return to sleep. */
2079 #elif defined(HAVE_USLEEP)
2080 /* If usleep is available, use it in preference to select. */
2083 /* On some systems, usleep cannot handle values larger than
2084 1,000,000. If the period is larger than that, use sleep
2085 first, then add usleep for subsecond accuracy. */
2087 seconds -= (long) seconds;
2089 usleep (seconds * 1000000);
2090 #else /* fall back select */
2091 /* Note that, although Windows supports select, it can't be used to
2092 implement sleeping because Winsock's select doesn't implement
2093 timeout when it is passed NULL pointers for all fd sets. (But it
2094 does under Cygwin, which implements Unix-compatible select.) */
2095 struct timeval sleep;
2096 sleep.tv_sec = (long) seconds;
2097 sleep.tv_usec = 1000000 * (seconds - (long) seconds);
2098 select (0, NULL, NULL, NULL, &sleep);
2099 /* If select returns -1 and errno is EINTR, it means we were
2100 interrupted by a signal. But without knowing how long we've
2101 actually slept, we can't return to sleep. Using gettimeofday to
2102 track sleeps is slow and unreliable due to clock skew. */
2106 #endif /* not WINDOWS */
2108 /* Encode the octets in DATA of length LENGTH to base64 format,
2109 storing the result to DEST. The output will be zero-terminated,
2110 and must point to a writable buffer of at least
2111 1+BASE64_LENGTH(length) bytes. The function returns the length of
2112 the resulting base64 data, not counting the terminating zero.
2114 This implementation does not emit newlines after 76 characters of
2118 base64_encode (const void *data, int length, char *dest)
2120 /* Conversion table. */
2121 static const char tbl[64] = {
2122 'A','B','C','D','E','F','G','H','I','J','K','L','M','N','O','P',
2123 'Q','R','S','T','U','V','W','X','Y','Z','a','b','c','d','e','f',
2124 'g','h','i','j','k','l','m','n','o','p','q','r','s','t','u','v',
2125 'w','x','y','z','0','1','2','3','4','5','6','7','8','9','+','/'
2127 /* Access bytes in DATA as unsigned char, otherwise the shifts below
2128 don't work for data with MSB set. */
2129 const unsigned char *s = data;
2130 /* Theoretical ANSI violation when length < 3. */
2131 const unsigned char *end = (const unsigned char *) data + length - 2;
2134 /* Transform the 3x8 bits to 4x6 bits, as required by base64. */
2135 for (; s < end; s += 3)
2137 *p++ = tbl[s[0] >> 2];
2138 *p++ = tbl[((s[0] & 3) << 4) + (s[1] >> 4)];
2139 *p++ = tbl[((s[1] & 0xf) << 2) + (s[2] >> 6)];
2140 *p++ = tbl[s[2] & 0x3f];
2143 /* Pad the result if necessary... */
2147 *p++ = tbl[s[0] >> 2];
2148 *p++ = tbl[(s[0] & 3) << 4];
2153 *p++ = tbl[s[0] >> 2];
2154 *p++ = tbl[((s[0] & 3) << 4) + (s[1] >> 4)];
2155 *p++ = tbl[((s[1] & 0xf) << 2)];
2159 /* ...and zero-terminate it. */
2165 /* Store in C the next non-whitespace character from the string, or \0
2166 when end of string is reached. */
2167 #define NEXT_CHAR(c, p) do { \
2168 c = (unsigned char) *p++; \
2169 } while (c_isspace (c))
2171 #define IS_ASCII(c) (((c) & 0x80) == 0)
2173 /* Decode data from BASE64 (a null-terminated string) into memory
2174 pointed to by DEST. DEST is assumed to be large enough to
2175 accomodate the decoded data, which is guaranteed to be no more than
2178 Since DEST is assumed to contain binary data, it is not
2179 NUL-terminated. The function returns the length of the data
2180 written to TO. -1 is returned in case of error caused by malformed
2183 This function originates from Free Recode. */
2186 base64_decode (const char *base64, void *dest)
2188 /* Table of base64 values for first 128 characters. Note that this
2189 assumes ASCII (but so does Wget in other places). */
2190 static const signed char base64_char_to_value[128] =
2192 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 0- 9 */
2193 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 10- 19 */
2194 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 20- 29 */
2195 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* 30- 39 */
2196 -1, -1, -1, 62, -1, -1, -1, 63, 52, 53, /* 40- 49 */
2197 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, /* 50- 59 */
2198 -1, -1, -1, -1, -1, 0, 1, 2, 3, 4, /* 60- 69 */
2199 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, /* 70- 79 */
2200 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, /* 80- 89 */
2201 25, -1, -1, -1, -1, -1, -1, 26, 27, 28, /* 90- 99 */
2202 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, /* 100-109 */
2203 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, /* 110-119 */
2204 49, 50, 51, -1, -1, -1, -1, -1 /* 120-127 */
2206 #define BASE64_CHAR_TO_VALUE(c) ((int) base64_char_to_value[c])
2207 #define IS_BASE64(c) ((IS_ASCII (c) && BASE64_CHAR_TO_VALUE (c) >= 0) || c == '=')
2209 const char *p = base64;
2215 unsigned long value;
2217 /* Process first byte of a quadruplet. */
2221 if (c == '=' || !IS_BASE64 (c))
2222 return -1; /* illegal char while decoding base64 */
2223 value = BASE64_CHAR_TO_VALUE (c) << 18;
2225 /* Process second byte of a quadruplet. */
2228 return -1; /* premature EOF while decoding base64 */
2229 if (c == '=' || !IS_BASE64 (c))
2230 return -1; /* illegal char while decoding base64 */
2231 value |= BASE64_CHAR_TO_VALUE (c) << 12;
2234 /* Process third byte of a quadruplet. */
2237 return -1; /* premature EOF while decoding base64 */
2239 return -1; /* illegal char while decoding base64 */
2245 return -1; /* premature EOF while decoding base64 */
2247 return -1; /* padding `=' expected but not found */
2251 value |= BASE64_CHAR_TO_VALUE (c) << 6;
2252 *q++ = 0xff & value >> 8;
2254 /* Process fourth byte of a quadruplet. */
2257 return -1; /* premature EOF while decoding base64 */
2261 return -1; /* illegal char while decoding base64 */
2263 value |= BASE64_CHAR_TO_VALUE (c);
2264 *q++ = 0xff & value;
2267 #undef BASE64_CHAR_TO_VALUE
2269 return q - (char *) dest;
2275 /* Simple merge sort for use by stable_sort. Implementation courtesy
2276 Zeljko Vrba with additional debugging by Nenad Barbutov. */
2279 mergesort_internal (void *base, void *temp, size_t size, size_t from, size_t to,
2280 int (*cmpfun) (const void *, const void *))
2282 #define ELT(array, pos) ((char *)(array) + (pos) * size)
2286 size_t mid = (to + from) / 2;
2287 mergesort_internal (base, temp, size, from, mid, cmpfun);
2288 mergesort_internal (base, temp, size, mid + 1, to, cmpfun);
2291 for (k = from; (i <= mid) && (j <= to); k++)
2292 if (cmpfun (ELT (base, i), ELT (base, j)) <= 0)
2293 memcpy (ELT (temp, k), ELT (base, i++), size);
2295 memcpy (ELT (temp, k), ELT (base, j++), size);
2297 memcpy (ELT (temp, k++), ELT (base, i++), size);
2299 memcpy (ELT (temp, k++), ELT (base, j++), size);
2300 for (k = from; k <= to; k++)
2301 memcpy (ELT (base, k), ELT (temp, k), size);
2306 /* Stable sort with interface exactly like standard library's qsort.
2307 Uses mergesort internally, allocating temporary storage with
2311 stable_sort (void *base, size_t nmemb, size_t size,
2312 int (*cmpfun) (const void *, const void *))
2316 void *temp = alloca (nmemb * size * sizeof (void *));
2317 mergesort_internal (base, temp, size, 0, nmemb - 1, cmpfun);
2321 /* Print a decimal number. If it is equal to or larger than ten, the
2322 number is rounded. Otherwise it is printed with one significant
2323 digit without trailing zeros and with no more than three fractional
2324 digits total. For example, 0.1 is printed as "0.1", 0.035 is
2325 printed as "0.04", 0.0091 as "0.009", and 0.0003 as simply "0".
2327 This is useful for displaying durations because it provides
2328 order-of-magnitude information without unnecessary clutter --
2329 long-running downloads are shown without the fractional part, and
2330 short ones still retain one significant digit. */
2333 print_decimal (double number)
2335 static char buf[32];
2336 double n = number >= 0 ? number : -number;
2339 /* Cut off at 9.95 because the below %.1f would round 9.96 to
2340 "10.0" instead of "10". OTOH 9.94 will print as "9.9". */
2341 snprintf (buf, sizeof buf, "%.0f", number);
2343 snprintf (buf, sizeof buf, "%.1f", number);
2344 else if (n >= 0.001)
2345 snprintf (buf, sizeof buf, "%.1g", number);
2346 else if (n >= 0.0005)
2347 /* round [0.0005, 0.001) to 0.001 */
2348 snprintf (buf, sizeof buf, "%.3f", number);
2350 /* print numbers close to 0 as 0, not 0.000 */
2367 { "/somedir", "/somedir", true },
2368 { "/somedir", "/somedir/d2", true },
2369 { "/somedir/d1", "/somedir", false },
2372 for (i = 0; i < countof(test_array); ++i)
2374 bool res = subdir_p (test_array[i].d1, test_array[i].d2);
2376 mu_assert ("test_subdir_p: wrong result",
2377 res == test_array[i].result);
2384 test_dir_matches_p()
2392 { { "/somedir", "/someotherdir", NULL }, "somedir", true },
2393 { { "/somedir", "/someotherdir", NULL }, "anotherdir", false },
2394 { { "/somedir", "/*otherdir", NULL }, "anotherdir", true },
2395 { { "/somedir/d1", "/someotherdir", NULL }, "somedir/d1", true },
2396 { { "*/*d1", "/someotherdir", NULL }, "somedir/d1", true },
2397 { { "/somedir/d1", "/someotherdir", NULL }, "d1", false },
2398 { { "!COMPLETE", NULL, NULL }, "!COMPLETE", true },
2399 { { "*COMPLETE", NULL, NULL }, "!COMPLETE", true },
2400 { { "*/!COMPLETE", NULL, NULL }, "foo/!COMPLETE", true },
2401 { { "*COMPLETE", NULL, NULL }, "foo/!COMPLETE", false },
2402 { { "*/*COMPLETE", NULL, NULL }, "foo/!COMPLETE", true },
2403 { { "/dir with spaces", NULL, NULL }, "dir with spaces", true },
2404 { { "/dir*with*spaces", NULL, NULL }, "dir with spaces", true },
2405 { { "/Tmp/has", NULL, NULL }, "/Tmp/has space", false },
2406 { { "/Tmp/has", NULL, NULL }, "/Tmp/has,comma", false },
2409 for (i = 0; i < countof(test_array); ++i)
2411 bool res = dir_matches_p (test_array[i].dirlist, test_array[i].dir);
2413 mu_assert ("test_dir_matches_p: wrong result",
2414 res == test_array[i].result);
2420 #endif /* TESTING */