#include <fcntl.h>
#include <assert.h>
+/* For TIOCGWINSZ and friends: */
+#ifdef HAVE_SYS_IOCTL_H
+# include <sys/ioctl.h>
+#endif
+#ifdef HAVE_TERMIOS_H
+# include <termios.h>
+#endif
+
+/* Needed for run_with_timeout. */
+#undef USE_SIGNAL_TIMEOUT
+#ifdef HAVE_SIGNAL_H
+# include <signal.h>
+#endif
+#ifdef HAVE_SETJMP_H
+# include <setjmp.h>
+#endif
+/* If sigsetjmp is a macro, configure won't pick it up. */
+#ifdef sigsetjmp
+# define HAVE_SIGSETJMP
+#endif
+#ifdef HAVE_SIGNAL
+# ifdef HAVE_SIGSETJMP
+# define USE_SIGNAL_TIMEOUT
+# endif
+# ifdef HAVE_SIGBLOCK
+# define USE_SIGNAL_TIMEOUT
+# endif
+#endif
+
#include "wget.h"
#include "utils.h"
#include "fnmatch.h"
static void
memfatal (const char *what)
{
- /* HACK: expose save_log_p from log.c, so we can turn it off in
- order to prevent saving the log. Saving the log is dangerous
- because logprintf() and logputs() can call malloc(), so this
- could infloop. When logging is turned off, infloop can no longer
- happen.
-
- #### This is no longer really necessary because the new routines
- in log.c cons only if the line exceeds eighty characters. But
- this can come at the end of a line, so it's OK to be careful.
-
- On a more serious note, it would be good to have a
- log_forced_shutdown() routine that exposes this cleanly. */
- extern int save_log_p;
-
- save_log_p = 0;
+ /* Make sure we don't try to store part of the log line, and thus
+ call malloc. */
+ log_set_save_context (0);
logprintf (LOG_ALWAYS, _("%s: %s: Not enough memory.\n"), exec_name, what);
exit (1);
}
#endif /* DEBUG_MALLOC */
\f
+/* Utility function: like xstrdup(), but also lowercases S. */
+
+char *
+xstrdup_lower (const char *s)
+{
+ char *copy = xstrdup (s);
+ char *p = copy;
+ for (; *p; p++)
+ *p = TOLOWER (*p);
+ return copy;
+}
+
+/* Return a count of how many times CHR occurs in STRING. */
+
+int
+count_char (const char *string, char chr)
+{
+ const char *p;
+ int count = 0;
+ for (p = string; *p; p++)
+ if (*p == chr)
+ ++count;
+ return count;
+}
+
/* Copy the string formed by two pointers (one on the beginning, other
on the char after the last char) to a new, malloc-ed location.
0-terminate it. */
ptm->tm_hour, ptm->tm_min, ptm->tm_sec);
return output;
}
-
-/* Returns an error message for ERRNUM. #### This requires more work.
- This function, as well as the whole error system, is very
- ill-conceived. */
-const char *
-uerrmsg (uerr_t errnum)
-{
- switch (errnum)
- {
- case URLUNKNOWN:
- return _("Unknown/unsupported protocol");
- break;
- case URLBADPORT:
- return _("Invalid port specification");
- break;
- case URLBADHOST:
- return _("Invalid host name");
- break;
- default:
- abort ();
- /* $@#@#$ compiler. */
- return NULL;
- }
-}
\f
/* The Windows versions of the following two functions are defined in
mswindows.c. */
else if (pid != 0)
{
/* parent, no error */
- printf (_("Continuing in background.\n"));
+ printf (_("Continuing in background, pid %d.\n"), (int)pid);
if (changedp)
printf (_("Output will be written to `%s'.\n"), opt.lfilename);
- exit (0);
- }
- /* child: keep running */
-}
-#endif /* not WINDOWS */
-\f
-/* Canonicalize PATH, and return a new path. The new path differs from PATH
- in that:
- Multple `/'s are collapsed to a single `/'.
- Leading `./'s and trailing `/.'s are removed.
- Trailing `/'s are removed.
- Non-leading `../'s and trailing `..'s are handled by removing
- portions of the path.
-
- E.g. "a/b/c/./../d/.." will yield "a/b". This function originates
- from GNU Bash.
-
- Changes for Wget:
- Always use '/' as stub_char.
- Don't check for local things using canon_stat.
- Change the original string instead of strdup-ing.
- React correctly when beginning with `./' and `../'. */
-void
-path_simplify (char *path)
-{
- register int i, start, ddot;
- char stub_char;
-
- if (!*path)
- return;
-
- /*stub_char = (*path == '/') ? '/' : '.';*/
- stub_char = '/';
-
- /* Addition: Remove all `./'-s preceding the string. If `../'-s
- precede, put `/' in front and remove them too. */
- i = 0;
- ddot = 0;
- while (1)
- {
- if (path[i] == '.' && path[i + 1] == '/')
- i += 2;
- else if (path[i] == '.' && path[i + 1] == '.' && path[i + 2] == '/')
- {
- i += 3;
- ddot = 1;
- }
- else
- break;
+ exit (0); /* #### should we use _exit()? */
}
- if (i)
- strcpy (path, path + i - ddot);
-
- /* Replace single `.' or `..' with `/'. */
- if ((path[0] == '.' && path[1] == '\0')
- || (path[0] == '.' && path[1] == '.' && path[2] == '\0'))
- {
- path[0] = stub_char;
- path[1] = '\0';
- return;
- }
- /* Walk along PATH looking for things to compact. */
- i = 0;
- while (1)
- {
- if (!path[i])
- break;
- while (path[i] && path[i] != '/')
- i++;
-
- start = i++;
-
- /* If we didn't find any slashes, then there is nothing left to do. */
- if (!path[start])
- break;
-
- /* Handle multiple `/'s in a row. */
- while (path[i] == '/')
- i++;
-
- if ((start + 1) != i)
- {
- strcpy (path + start + 1, path + i);
- i = start + 1;
- }
-
- /* Check for trailing `/'. */
- if (start && !path[i])
- {
- zero_last:
- path[--i] = '\0';
- break;
- }
-
- /* Check for `../', `./' or trailing `.' by itself. */
- if (path[i] == '.')
- {
- /* Handle trailing `.' by itself. */
- if (!path[i + 1])
- goto zero_last;
-
- /* Handle `./'. */
- if (path[i + 1] == '/')
- {
- strcpy (path + i, path + i + 1);
- i = (start < 0) ? 0 : start;
- continue;
- }
-
- /* Handle `../' or trailing `..' by itself. */
- if (path[i + 1] == '.' &&
- (path[i + 2] == '/' || !path[i + 2]))
- {
- while (--start > -1 && path[start] != '/');
- strcpy (path + start + 1, path + i + 2);
- i = (start < 0) ? 0 : start;
- continue;
- }
- } /* path == '.' */
- } /* while */
-
- if (!*path)
- {
- *path = stub_char;
- path[1] = '\0';
- }
+ /* child: give up the privileges and keep running. */
+ setsid ();
+ freopen ("/dev/null", "r", stdin);
+ freopen ("/dev/null", "w", stdout);
+ freopen ("/dev/null", "w", stderr);
}
+#endif /* not WINDOWS */
\f
/* "Touch" FILE, i.e. make its atime and mtime equal to the time
specified with TM. */
{
int quit = 0;
int i;
+ int ret = 0;
char *dir;
/* Make a copy of dir, to be able to write to it. Otherwise, the
if (!dir[i])
quit = 1;
dir[i] = '\0';
- /* Check whether the directory already exists. */
+ /* Check whether the directory already exists. Allow creation of
+ of intermediate directories to fail, as the initial path components
+ are not necessarily directories! */
if (!file_exists_p (dir))
- {
- if (mkdir (dir, 0777) < 0)
- return -1;
- }
+ ret = mkdir (dir, 0777);
+ else
+ ret = 0;
if (quit)
break;
else
dir[i] = '/';
}
- return 0;
+ return ret;
+}
+
+/* Merge BASE with FILE. BASE can be a directory or a file name, FILE
+ should be a file name.
+
+ file_merge("/foo/bar", "baz") => "/foo/baz"
+ file_merge("/foo/bar/", "baz") => "/foo/bar/baz"
+ file_merge("foo", "bar") => "bar"
+
+ In other words, it's a simpler and gentler version of uri_merge_1. */
+
+char *
+file_merge (const char *base, const char *file)
+{
+ char *result;
+ const char *cut = (const char *)strrchr (base, '/');
+
+ if (!cut)
+ return xstrdup (file);
+
+ result = (char *)xmalloc (cut - base + 1 + strlen (file) + 1);
+ memcpy (result, base, cut - base);
+ result[cut - base] = '/';
+ strcpy (result + (cut - base) + 1, file);
+
+ return result;
}
\f
static int in_acclist PARAMS ((const char *const *, const char *, int));
match_backwards ("abc", "bc") -> 1
match_backwards ("abc", "ab") -> 0
match_backwards ("abc", "abc") -> 1 */
-static int
-match_backwards (const char *string, const char *pattern)
+int
+match_tail (const char *string, const char *pattern)
{
int i, j;
}
/* Checks whether string S matches each element of ACCEPTS. A list
- element are matched either with fnmatch() or match_backwards(),
+ element are matched either with fnmatch() or match_tail(),
according to whether the element contains wildcards or not.
If the BACKWARD is 0, don't do backward comparison -- just compare
{
if (backward)
{
- if (match_backwards (s, *accepts))
+ if (match_tail (s, *accepts))
return 1;
}
else
return 0;
}
-/* Return the malloc-ed suffix of STR. For instance:
+/* Return the location of STR's suffix (file extension). Examples:
suffix ("foo.bar") -> "bar"
suffix ("foo.bar.baz") -> "baz"
suffix ("/foo/bar") -> NULL
{
int i;
- for (i = strlen (str); i && str[i] != '/' && str[i] != '.'; i--);
+ for (i = strlen (str); i && str[i] != '/' && str[i] != '.'; i--)
+ ;
+
if (str[i++] == '.')
- return xstrdup (str + i);
+ return (char *)str + i;
else
return NULL;
}
-/* Read a line from FP. The function reallocs the storage as needed
- to accomodate for any length of the line. Reallocs are done
- storage exponentially, doubling the storage after each overflow to
- minimize the number of calls to realloc() and fgets(). The newline
- character at the end of line is retained.
+/* Return non-zero if FNAME ends with a typical HTML suffix. The
+ following (case-insensitive) suffixes are presumed to be HTML files:
+
+ html
+ htm
+ ?html (`?' matches one character)
+
+ #### CAVEAT. This is not necessarily a good indication that FNAME
+ refers to a file that contains HTML! */
+int
+has_html_suffix_p (const char *fname)
+{
+ char *suf;
+
+ if ((suf = suffix (fname)) == NULL)
+ return 0;
+ if (!strcasecmp (suf, "html"))
+ return 1;
+ if (!strcasecmp (suf, "htm"))
+ return 1;
+ if (suf[0] && !strcasecmp (suf + 1, "html"))
+ return 1;
+ return 0;
+}
+
+/* Read a line from FP and return the pointer to freshly allocated
+ storage. The stoarage space is obtained through malloc() and
+ should be freed with free() when it is no longer needed.
+
+ The length of the line is not limited, except by available memory.
+ The newline character at the end of line is retained. The line is
+ terminated with a zero character.
After end-of-file is encountered without anything being read, NULL
is returned. NULL is also returned on error. To distinguish
read_whole_line (FILE *fp)
{
int length = 0;
- int bufsize = 81;
+ int bufsize = 82;
char *line = (char *)xmalloc (bufsize);
while (fgets (line + length, bufsize - length, fp))
{
length += strlen (line + length);
- assert (length > 0);
+ if (length == 0)
+ /* Possible for example when reading from a binary file where
+ a line begins with \0. */
+ continue;
+
if (line[length - 1] == '\n')
break;
+
/* fgets() guarantees to read the whole line, or to use up the
space we've given it. We can double the buffer
unconditionally. */
{
char inbuf[24];
/* Print the number into the buffer. */
- long_to_string (inbuf, l);
+ number_to_string (inbuf, l);
return legible_1 (inbuf);
}
/* Count the digits in a (long) integer. */
int
-numdigit (long a)
+numdigit (long number)
{
- int res = 1;
- if (a < 0)
+ int cnt = 1;
+ if (number < 0)
{
- a = -a;
- ++res;
+ number = -number;
+ ++cnt;
}
- while ((a /= 10) != 0)
- ++res;
- return res;
+ while ((number /= 10) > 0)
+ ++cnt;
+ return cnt;
}
#define ONE_DIGIT(figure) *p++ = n / (figure) + '0'
#define DIGITS_18(figure) ONE_DIGIT_ADVANCE (figure); DIGITS_17 ((figure) / 10)
#define DIGITS_19(figure) ONE_DIGIT_ADVANCE (figure); DIGITS_18 ((figure) / 10)
-/* Print NUMBER to BUFFER in base 10. This is completely equivalent
- to `sprintf(buffer, "%ld", number)', only much faster.
+/* Print NUMBER to BUFFER in base 10. This should be completely
+ equivalent to `sprintf(buffer, "%ld", number)', only much faster.
The speedup may make a difference in programs that frequently
convert numbers to strings. Some implementations of sprintf,
particularly the one in GNU libc, have been known to be extremely
slow compared to this function.
- BUFFER should accept as many bytes as you expect the number to take
- up. On machines with 64-bit longs the maximum needed size is 24
- bytes. That includes the worst-case digits, the optional `-' sign,
- and the trailing \0. */
+ Return the pointer to the location where the terminating zero was
+ printed. (Equivalent to calling buffer+strlen(buffer) after the
+ function is done.)
-void
-long_to_string (char *buffer, long number)
+ BUFFER should be big enough to accept as many bytes as you expect
+ the number to take up. On machines with 64-bit longs the maximum
+ needed size is 24 bytes. That includes the digits needed for the
+ largest 64-bit number, the `-' sign in case it's negative, and the
+ terminating '\0'. */
+
+char *
+number_to_string (char *buffer, long number)
{
char *p = buffer;
long n = number;
/* We are running in a strange or misconfigured environment. Let
sprintf cope with it. */
sprintf (buffer, "%ld", n);
+ p += strlen (buffer);
#else /* (SIZEOF_LONG == 4) || (SIZEOF_LONG == 8) */
if (n < 0)
*p = '\0';
#endif /* (SIZEOF_LONG == 4) || (SIZEOF_LONG == 8) */
+
+ return p;
}
#undef ONE_DIGIT
*p = '\0';
return res;
}
+
+/* Determine the width of the terminal we're running on. If that's
+ not possible, return 0. */
+
+int
+determine_screen_width (void)
+{
+ /* If there's a way to get the terminal size using POSIX
+ tcgetattr(), somebody please tell me. */
+#ifndef TIOCGWINSZ
+ return 0;
+#else /* TIOCGWINSZ */
+ int fd;
+ struct winsize wsz;
+
+ if (opt.lfilename != NULL)
+ return 0;
+
+ fd = fileno (stderr);
+ if (ioctl (fd, TIOCGWINSZ, &wsz) < 0)
+ return 0; /* most likely ENOTTY */
+
+ return wsz.ws_col;
+#endif /* TIOCGWINSZ */
+}
+
+/* Return a random number between 0 and MAX-1, inclusive.
+
+ If MAX is greater than the value of RAND_MAX+1 on the system, the
+ returned value will be in the range [0, RAND_MAX]. This may be
+ fixed in a future release.
+
+ The random number generator is seeded automatically the first time
+ it is called.
+
+ This uses rand() for portability. It has been suggested that
+ random() offers better randomness, but this is not required for
+ Wget, so I chose to go for simplicity and use rand
+ unconditionally. */
+
+int
+random_number (int max)
+{
+ static int seeded;
+ double bounded;
+ int rnd;
+
+ if (!seeded)
+ {
+ srand (time (NULL));
+ seeded = 1;
+ }
+ rnd = rand ();
+
+ /* On systems that don't define RAND_MAX, assume it to be 2**15 - 1,
+ and enforce that assumption by masking other bits. */
+#ifndef RAND_MAX
+# define RAND_MAX 32767
+ rnd &= RAND_MAX;
+#endif
+
+ /* This is equivalent to rand() % max, but uses the high-order bits
+ for better randomness on architecture where rand() is implemented
+ using a simple congruential generator. */
+
+ bounded = (double)max * rnd / (RAND_MAX + 1.0);
+ return (int)bounded;
+}
+
+#if 0
+/* A debugging function for checking whether an MD5 library works. */
+
+#include "gen-md5.h"
+
+char *
+debug_test_md5 (char *buf)
+{
+ unsigned char raw[16];
+ static char res[33];
+ unsigned char *p1;
+ char *p2;
+ int cnt;
+ ALLOCA_MD5_CONTEXT (ctx);
+
+ gen_md5_init (ctx);
+ gen_md5_update ((unsigned char *)buf, strlen (buf), ctx);
+ gen_md5_finish (ctx, raw);
+
+ p1 = raw;
+ p2 = res;
+ cnt = 16;
+ while (cnt--)
+ {
+ *p2++ = XDIGIT_TO_xchar (*p1 >> 4);
+ *p2++ = XDIGIT_TO_xchar (*p1 & 0xf);
+ ++p1;
+ }
+ *p2 = '\0';
+
+ return res;
+}
+#endif
+\f
+/* Implementation of run_with_timeout, a generic timeout handler for
+ systems with Unix-like signal handling. */
+#ifdef HAVE_SIGSETJMP
+#define SETJMP(env) sigsetjmp (env, 1)
+
+static sigjmp_buf run_with_timeout_env;
+
+static RETSIGTYPE
+abort_run_with_timeout (int sig)
+{
+ assert (sig == SIGALRM);
+ siglongjmp (run_with_timeout_env, -1);
+}
+#else /* not HAVE_SIGSETJMP */
+#define SETJMP(env) setjmp (env)
+
+static jmp_buf run_with_timeout_env;
+
+static RETSIGTYPE
+abort_run_with_timeout (int sig)
+{
+ assert (sig == SIGALRM);
+ /* We don't have siglongjmp to preserve the set of blocked signals;
+ if we longjumped out of the handler at this point, SIGALRM would
+ remain blocked. We must unblock it manually. */
+ int mask = siggetmask ();
+ mask &= ~sigmask(SIGALRM);
+ sigsetmask (mask);
+
+ /* Now it's safe to longjump. */
+ longjmp (run_with_timeout_env, -1);
+}
+#endif /* not HAVE_SIGSETJMP */
+
+int
+run_with_timeout (long timeout, void (*fun) (void *), void *arg)
+{
+#ifndef USE_SIGNAL_TIMEOUT
+ fun (arg);
+ return 0;
+#else
+ int saved_errno;
+
+ if (timeout == 0)
+ {
+ fun (arg);
+ return 0;
+ }
+
+ signal (SIGALRM, abort_run_with_timeout);
+ if (SETJMP (run_with_timeout_env) != 0)
+ {
+ /* Longjumped out of FUN with a timeout. */
+ signal (SIGALRM, SIG_DFL);
+ return 1;
+ }
+ alarm (timeout);
+ fun (arg);
+
+ /* Preserve errno in case alarm() or signal() modifies it. */
+ saved_errno = errno;
+ alarm (0);
+ signal (SIGALRM, SIG_DFL);
+ errno = saved_errno;
+
+ return 0;
+#endif
+}
+