/* Handling of recursive HTTP retrieving.
- Copyright (C) 1996-2005 Free Software Foundation, Inc.
+ Copyright (C) 1996-2006 Free Software Foundation, Inc.
This file is part of GNU Wget.
#include "hash.h"
#include "res.h"
#include "convert.h"
-\f
+#include "html-url.h"
+#include "css-url.h"
+#include "spider.h"
+
/* Functions for maintaining the URL queue. */
struct queue_element {
int depth; /* the depth */
bool html_allowed; /* whether the document is allowed to
be treated as HTML. */
-
+ bool css_allowed; /* whether the document is allowed to
+ be treated as CSS. */
struct queue_element *next; /* next element in queue */
};
static void
url_enqueue (struct url_queue *queue,
- const char *url, const char *referer, int depth, bool html_allowed)
+ const char *url, const char *referer, int depth,
+ bool html_allowed, bool css_allowed)
{
struct queue_element *qel = xnew (struct queue_element);
qel->url = url;
qel->referer = referer;
qel->depth = depth;
qel->html_allowed = html_allowed;
+ qel->css_allowed = css_allowed;
qel->next = NULL;
++queue->count;
static bool
url_dequeue (struct url_queue *queue,
const char **url, const char **referer, int *depth,
- bool *html_allowed)
+ bool *html_allowed, bool *css_allowed)
{
struct queue_element *qel = queue->head;
*referer = qel->referer;
*depth = qel->depth;
*html_allowed = qel->html_allowed;
+ *css_allowed = qel->css_allowed;
--queue->count;
/* Enqueue the starting URL. Use start_url_parsed->url rather than
just URL so we enqueue the canonical form of the URL. */
- url_enqueue (queue, xstrdup (start_url_parsed->url), NULL, 0, true);
+ url_enqueue (queue, xstrdup (start_url_parsed->url), NULL, 0, true, false);
string_set_add (blacklist, start_url_parsed->url);
while (1)
bool descend = false;
char *url, *referer, *file = NULL;
int depth;
- bool html_allowed;
+ bool html_allowed, css_allowed;
+ bool is_css = false;
bool dash_p_leaf_HTML = false;
if (opt.quota && total_downloaded_bytes > opt.quota)
if (!url_dequeue (queue,
(const char **)&url, (const char **)&referer,
- &depth, &html_allowed))
+ &depth, &html_allowed, &css_allowed))
break;
/* ...and download it. Note that this download is in most cases
DEBUGP (("Already downloaded \"%s\", reusing it from \"%s\".\n",
url, file));
+ /* this sucks, needs to be combined! */
if (html_allowed
&& downloaded_html_set
&& string_set_contains (downloaded_html_set, file))
- descend = true;
+ {
+ descend = true;
+ is_css = false;
+ }
+ if (css_allowed
+ && downloaded_css_set
+ && string_set_contains (downloaded_css_set, file))
+ {
+ descend = 1;
+ is_css = true;
+ }
}
else
{
int dt = 0;
char *redirected = NULL;
- bool oldrec = opt.recursive;
- opt.recursive = false;
- status = retrieve_url (url, &file, &redirected, referer, &dt);
- opt.recursive = oldrec;
+ status = retrieve_url (url, &file, &redirected, referer, &dt, false);
if (html_allowed && file && status == RETROK
&& (dt & RETROKF) && (dt & TEXTHTML))
- descend = true;
+ {
+ descend = true;
+ is_css = false;
+ }
+
+ /* a little different, css_allowed can override content type
+ lots of web servers serve css with an incorrect content type
+ */
+ if (file && status == RETROK
+ && (dt & RETROKF) &&
+ ((dt & TEXTCSS) || css_allowed))
+ {
+ descend = true;
+ is_css = false;
+ }
if (redirected)
{
}
}
+ if (opt.spider)
+ {
+ visited_url (url, referer);
+ }
+
if (descend
&& depth >= opt.reclevel && opt.reclevel != INFINITE_RECURSION)
{
}
}
- /* If the downloaded document was HTML, parse it and enqueue the
+ /* If the downloaded document was HTML or CSS, parse it and enqueue the
links it contains. */
if (descend)
{
bool meta_disallow_follow = false;
struct urlpos *children
- = get_urls_html (file, url, &meta_disallow_follow);
+ = is_css ? get_urls_css_file (file, url) :
+ get_urls_html (file, url, &meta_disallow_follow);
if (opt.use_robots && meta_disallow_follow)
{
{
url_enqueue (queue, xstrdup (child->url->url),
xstrdup (url), depth + 1,
- child->link_expect_html);
+ child->link_expect_html,
+ child->link_expect_css);
/* We blacklist the URL we have enqueued, because we
don't want to enqueue (and hence download) the
same URL twice. */
}
}
- if (opt.delete_after || (file && !acceptable (file)))
+ if (file
+ && (opt.delete_after
+ || opt.spider /* opt.recursive is implicitely true */
+ || !acceptable (file)))
{
/* Either --delete-after was specified, or we loaded this
- otherwise rejected (e.g. by -R) HTML file just so we
- could harvest its hyperlinks -- in either case, delete
- the local file. */
+ (otherwise unneeded because of --spider or rejected by -R)
+ HTML file just to harvest its hyperlinks -- in either case,
+ delete the local file. */
DEBUGP (("Removing file due to %s in recursive_retrieve():\n",
opt.delete_after ? "--delete-after" :
- "recursive rejection criteria"));
+ (opt.spider ? "--spider" :
+ "recursive rejection criteria")));
logprintf (LOG_VERBOSE,
- (opt.delete_after
+ (opt.delete_after || opt.spider
? _("Removing %s.\n")
: _("Removing %s since it should be rejected.\n")),
file);
if (unlink (file))
logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
+ logputs (LOG_VERBOSE, "\n");
register_delete_file (file);
}
{
char *d1, *d2;
int d3;
- bool d4;
+ bool d4, d5;
while (url_dequeue (queue,
- (const char **)&d1, (const char **)&d2, &d3, &d4))
+ (const char **)&d1, (const char **)&d2, &d3, &d4, &d5))
{
xfree (d1);
xfree_null (d2);
if (string_set_contains (blacklist, url))
{
+ if (opt.spider)
+ {
+ char *referrer = url_string (parent, true);
+ DEBUGP (("download_child_p: parent->url is: `%s'\n", parent->url));
+ visited_url (url, referrer);
+ xfree (referrer);
+ }
DEBUGP (("Already on the black list.\n"));
goto out;
}
&& u->port == start_url_parsed->port
&& !(opt.page_requisites && upos->link_inline_p))
{
- if (!frontcmp (start_url_parsed->dir, u->dir))
+ if (!subdir_p (start_url_parsed->dir, u->dir))
{
DEBUGP (("Going to \"%s\" would escape \"%s\" with no_parent on.\n",
u->dir, start_url_parsed->dir));
projects / wget / blobdiff