/* File retrieval.
- Copyright (C) 1995, 1996, 1997, 1998 Free Software Foundation, Inc.
+ Copyright (C) 1995, 1996, 1997, 1998, 2000 Free Software Foundation, Inc.
This file is part of Wget.
{
uerr_t result;
char *url;
- int location_changed, already_redirected, dummy;
+ int location_changed, dummy;
int local_use_proxy;
char *mynewloc, *proxy;
struct urlinfo *u;
-
+ slist *redirections;
/* If dt is NULL, just ignore it. */
if (!dt)
*newloc = NULL;
if (file)
*file = NULL;
- already_redirected = 0;
- again:
+ redirections = NULL;
+
u = newurl ();
- /* Parse the URL. RFC2068 requires `Location' to contain an
- absoluteURI, but many sites break this requirement. #### We
- should be liberal and accept a relative location, too. */
- result = parseurl (url, u, already_redirected);
+ /* Parse the URL. */
+ result = parseurl (url, u, 0);
if (result != URLOK)
{
- freeurl (u, 1);
logprintf (LOG_NOTQUIET, "%s: %s.\n", url, uerrmsg (result));
+ freeurl (u, 1);
+ free_slist (redirections);
+ free (url);
return result;
}
+ redirected:
+
/* Set the referer. */
if (refurl)
u->referer = xstrdup (refurl);
{
logputs (LOG_NOTQUIET, _("Could not find proxy host.\n"));
freeurl (u, 1);
+ free_slist (redirections);
+ free (url);
return PROXERR;
}
/* Parse the proxy URL. */
else
logprintf (LOG_NOTQUIET, _("Proxy %s: Must be HTTP.\n"), proxy);
freeurl (u, 1);
+ free_slist (redirections);
+ free (url);
return PROXERR;
}
u->proto = URLHTTP;
retrieval, so we save recursion to oldrec, and restore it
later. */
int oldrec = opt.recursive;
- if (already_redirected)
+ if (redirections)
opt.recursive = 0;
result = ftp_loop (u, dt);
opt.recursive = oldrec;
#### All of this is, of course, crap. These types should be
determined through mailcap. */
- if (already_redirected && u->local && (u->proto == URLFTP ))
+ if (redirections && u->local && (u->proto == URLFTP ))
{
char *suf = suffix (u->local);
if (suf && (!strcasecmp (suf, "html") || !strcasecmp (suf, "htm")))
location_changed = (result == NEWLOCATION);
if (location_changed)
{
- /* Check for redirection to oneself. */
- if (url_equal (url, mynewloc))
+ char *construced_newloc;
+ uerr_t newloc_result;
+ struct urlinfo *newloc_struct;
+
+ assert (mynewloc != NULL);
+
+ /* The HTTP specs only allow absolute URLs to appear in
+ redirects, but a ton of boneheaded webservers and CGIs out
+ there break the rules and use relative URLs, and popular
+ browsers are lenient about this, so wget should be too. */
+ construced_newloc = url_concat (url, mynewloc);
+ free (mynewloc);
+ mynewloc = construced_newloc;
+
+ /* Now, see if this new location makes sense. */
+ newloc_struct = newurl ();
+ newloc_result = parseurl (mynewloc, newloc_struct, 1);
+ if (newloc_result != URLOK)
+ {
+ logprintf (LOG_NOTQUIET, "%s: %s.\n", mynewloc, uerrmsg (newloc_result));
+ freeurl (newloc_struct, 1);
+ freeurl (u, 1);
+ free_slist (redirections);
+ free (url);
+ free (mynewloc);
+ return result;
+ }
+
+ /* Now mynewloc will become newloc_struct->url, because if the
+ Location contained relative paths like .././something, we
+ don't want that propagating as url. */
+ free (mynewloc);
+ mynewloc = xstrdup (newloc_struct->url);
+
+ /* Check for redirection to back to itself. */
+ if (!strcmp (u->url, newloc_struct->url))
{
logprintf (LOG_NOTQUIET, _("%s: Redirection to itself.\n"),
mynewloc);
+ freeurl (newloc_struct, 1);
+ freeurl (u, 1);
+ free_slist (redirections);
+ free (url);
+ free (mynewloc);
return WRONGCODE;
}
- if (mynewloc)
+
+ /* The new location is OK. Let's check for redirection cycle by
+ peeking through the history of redirections. */
+ if (in_slist (redirections, newloc_struct->url))
{
- /* The HTTP specs only allow absolute URLs to appear in redirects, but
- a ton of boneheaded webservers and CGIs out there break the rules
- and use relative URLs, and popular browsers are lenient about this,
- so wget should be too. */
- if (strstr(mynewloc, "://") == NULL)
- /* Doesn't look like an absolute URL (this check will incorrectly
- think that rare relative URLs containing "://" later in the
- string are absolute). */
- {
- char *temp = malloc(strlen(url) + strlen(mynewloc) + 1);
-
- if (mynewloc[0] == '/')
- /* "Hostless absolute" URL. Convert to absolute. */
- sprintf(temp,"%s%s", url, mynewloc);
- else
- /* Relative URL. Convert to absolute. */
- sprintf(temp,"%s/%s", url, mynewloc);
-
- free(mynewloc);
- mynewloc = temp;
- }
-
+ logprintf (LOG_NOTQUIET, _("%s: Redirection cycle detected.\n"),
+ mynewloc);
+ freeurl (newloc_struct, 1);
+ freeurl (u, 1);
+ free_slist (redirections);
free (url);
- url = mynewloc;
+ free (mynewloc);
+ return WRONGCODE;
}
+
+ redirections = add_slist (redirections, newloc_struct->url, NOSORT);
+
+ free (url);
+ url = mynewloc;
freeurl (u, 1);
- already_redirected = 1;
- goto again;
+ u = newloc_struct;
+ goto redirected;
}
+
if (file)
{
if (u->local)
*file = NULL;
}
freeurl (u, 1);
+ free_slist (redirections);
if (newloc)
*newloc = url;
char *filename, *new_file;
int dt;
- if (opt.quota && opt.downloaded > opt.quota)
+ if (downloaded_exceeds_quota ())
{
status = QUOTEXC;
break;
{
logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
}
+
+/* Increment opt.downloaded by BY_HOW_MUCH. If an overflow occurs,
+ set opt.downloaded_overflow to 1. */
+void
+downloaded_increase (unsigned long by_how_much)
+{
+ VERY_LONG_TYPE old;
+ if (opt.downloaded_overflow)
+ return;
+ old = opt.downloaded;
+ opt.downloaded += by_how_much;
+ if (opt.downloaded < old) /* carry flag, where are you when I
+ need you? */
+ {
+ /* Overflow. */
+ opt.downloaded_overflow = 1;
+ opt.downloaded = ~((VERY_LONG_TYPE)0);
+ }
+}
+
+/* Return non-zero if the downloaded amount of bytes exceeds the
+ desired quota. If quota is not set or if the amount overflowed, 0
+ is returned. */
+int
+downloaded_exceeds_quota (void)
+{
+ if (!opt.quota)
+ return 0;
+ if (opt.downloaded_overflow)
+ /* We don't really know. (Wildly) assume not. */
+ return 0;
+
+ return opt.downloaded > opt.quota;
+}
projects / wget / blobdiff