void
location_goto(struct document_view *doc_view, unsigned char *url)
{
unsigned char *new_abs_url;
struct uri *new_uri;
struct delayed_goto *deg;
/* Workaround for bug 611. Does not crash, but may lead to infinite loop.*/
if (!doc_view) return;
new_abs_url = join_urls(doc_view->document->uri,
trim_chars(url, ' ', 0));
if (!new_abs_url)
return;
new_uri = get_uri(new_abs_url, 0);
mem_free(new_abs_url);
if (!new_uri)
return;
deg = mem_calloc(1, sizeof(*deg));
if (!deg) {
done_uri(new_uri);
return;
}
assert(doc_view->vs);
deg->vs = doc_view->vs;
deg->uri = new_uri;
/* It does not seem to be very safe inside of frames to
* call goto_uri() right away. */
register_bottom_half(delayed_goto, deg);
}
void
html_a(struct html_context *html_context, unsigned char *a,
unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
unsigned char *href;
href = get_url_val(a, (unsigned char *)"href", html_context->doc_cp);
if (href) {
unsigned char *target;
mem_free_set(&format.link,
join_urls(html_context->base_href,
trim_chars(href, ' ', 0)));
mem_free(href);
target = get_target(html_context->options, a);
if (target) {
mem_free_set(&format.target, target);
} else {
mem_free_set(&format.target, stracpy(html_context->base_target));
}
if (0) {
; /* Shut up compiler */
#ifdef CONFIG_GLOBHIST
} else if (get_global_history_item(format.link)) {
format.style.color.foreground = format.color.vlink;
html_top->pseudo_class &= ~ELEMENT_LINK;
html_top->pseudo_class |= ELEMENT_VISITED;
#endif
#ifdef CONFIG_BOOKMARKS
} else if (get_bookmark(format.link)) {
format.style.color.foreground = format.color.bookmark_link;
html_top->pseudo_class &= ~ELEMENT_VISITED;
/* XXX: Really set ELEMENT_LINK? --pasky */
html_top->pseudo_class |= ELEMENT_LINK;
#endif
} else {
format.style.color.foreground = format.color.clink;
html_top->pseudo_class &= ~ELEMENT_VISITED;
html_top->pseudo_class |= ELEMENT_LINK;
}
mem_free_set(&format.title,
get_attr_val(a, (unsigned char *)"title", html_context->doc_cp));
html_focusable(html_context, a);
} else {
pop_html_element(html_context);
}
set_fragment_identifier(html_context, a, (unsigned char *)"name");
}
void
html_frame(struct html_context *html_context, unsigned char *a,
unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
unsigned char *name, *src, *url;
src = get_url_val(a, (unsigned char *)"src", html_context->doc_cp);
if (!src) {
url = stracpy((const unsigned char *)"about:blank");
} else {
url = join_urls(html_context->base_href, src);
mem_free(src);
}
if (!url) return;
name = get_attr_val(a, (unsigned char *)"name", html_context->doc_cp);
if (!name) {
name = stracpy(url);
} else if (!name[0]) {
/* When name doesn't have a value */
mem_free(name);
name = stracpy(url);
}
if (!name) return;
if (!html_context->options->frames || !html_top->frameset) {
html_focusable(html_context, a);
put_link_line((unsigned char *)"Frame: ", name, url, (unsigned char *)"", html_context);
} else {
if (html_context->special_f(html_context, SP_USED, NULL)) {
html_context->special_f(html_context, SP_FRAME,
html_top->frameset, name, url);
}
}
mem_free(name);
mem_free(url);
}
void
html_base(struct html_context *html_context, unsigned char *a,
unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
unsigned char *al;
al = get_url_val(a, (unsigned char *)"href", html_context->doc_cp);
if (al) {
unsigned char *base = join_urls(html_context->base_href, al);
struct uri *uri = base ? get_uri(base, 0) : NULL;
mem_free(al);
mem_free_if(base);
if (uri) {
done_uri(html_context->base_href);
html_context->base_href = uri;
}
}
al = get_target(html_context->options, a);
if (al) mem_free_set(&html_context->base_target, al);
}
void end_dump(struct status *stat, void *p)
{
struct cache_entry *ce = stat->ce;
int oh = get_output_handle();
if (oh == -1) return;
if (ce && ce->redirect && dump_red_count++ < MAX_REDIRECTS) {
unsigned char *u, *p;
if (stat->state >= 0) change_connection(stat, NULL, PRI_CANCEL);
u = join_urls(ce->url, ce->redirect);
if (!http_bugs.bug_302_redirect) if (!ce->redirect_get && (p = strchr(ce->url, POST_CHAR))) add_to_strn(&u, p);
load_url(u, stat, PRI_MAIN, 0);
mem_free(u);
return;
}
if (stat->state >= 0 && stat->state < S_TRANS) return;
if (stat->state >= S_TRANS && dmp != D_SOURCE) return;
if (dmp == D_SOURCE) {
if (ce) {
struct fragment *frag;
nextfrag:
foreach(frag, ce->frag) if (frag->offset <= dump_pos && frag->offset + frag->length > dump_pos) {
int l = frag->length - (dump_pos - frag->offset);
int w = hard_write(oh, frag->data + dump_pos - frag->offset, l);
if (w != l) {
detach_connection(stat, dump_pos);
if (w < 0) fprintf(stderr, "Error writing to stdout: %s.\n", strerror(errno));
else fprintf(stderr, "Can't write to stdout.\n");
retval = RET_ERROR;
goto terminate;
}
dump_pos += w;
detach_connection(stat, dump_pos);
goto nextfrag;
}
}
if (stat->state >= 0) return;
} else if (ce) {
/* prefix can have entities in it, but linkname cannot. */
void
put_link_line(unsigned char *prefix, unsigned char *linkname,
unsigned char *link, unsigned char *target,
struct html_context *html_context)
{
html_context->has_link_lines = 1;
html_stack_dup(html_context, ELEMENT_KILLABLE);
ln_break(html_context, 1);
mem_free_set(&format.link, NULL);
mem_free_set(&format.target, NULL);
mem_free_set(&format.title, NULL);
format.form = NULL;
put_chrs(html_context, prefix, strlen((const char *)prefix));
format.link = join_urls(html_context->base_href, link);
format.target = stracpy(target);
format.style.color.foreground = format.color.clink;
/* linkname typically comes from get_attr_val, which
* has already expanded character entity references.
* Tell put_chrs not to expand them again. */
format.style.attr |= AT_NO_ENTITIES;
put_chrs(html_context, linkname, strlen((const char *)linkname));
ln_break(html_context, 1);
pop_html_element(html_context);
}
static void
html_img_do(unsigned char *a, unsigned char *object_src,
struct html_context *html_context)
{
int ismap, usemap = 0;
int add_brackets = 0;
unsigned char *src = NULL;
unsigned char *label = NULL;
unsigned char *usemap_attr;
struct document_options *options = html_context->options;
int display_style = options->image_link.display_style;
/* Note about display_style:
* 0 means always display IMG
* 1 means always display filename
* 2 means display alt/title attribute if possible, IMG if not
* 3 means display alt/title attribute if possible, filename if not */
usemap_attr = get_attr_val(a, (unsigned char *)"usemap", html_context->doc_cp);
if (usemap_attr) {
unsigned char *joined_urls = join_urls(html_context->base_href,
usemap_attr);
unsigned char *map_url;
mem_free(usemap_attr);
if (!joined_urls) return;
map_url = straconcat((unsigned char *)"MAP@", joined_urls,
(unsigned char *) NULL);
mem_free(joined_urls);
if (!map_url) return;
html_stack_dup(html_context, ELEMENT_KILLABLE);
mem_free_set(&format.link, map_url);
format.form = NULL;
format.style.attr |= AT_BOLD;
usemap = 1;
}
ismap = format.link
&& has_attr(a, (unsigned char *)"ismap", html_context->doc_cp)
&& !usemap;
if (display_style == 2 || display_style == 3) {
label = get_attr_val(a, (unsigned char *)"alt", html_context->doc_cp);
if (!label)
label = get_attr_val(a, (unsigned char *)"title", html_context->doc_cp);
/* Little hack to preserve rendering of [ ], in directory listings,
* but we still want to drop extra spaces in alt or title attribute
* to limit display width on certain websites. --Zas */
if (label && strlen((const char *)label) > 5) clr_spaces(label);
}
src = null_or_stracpy(object_src);
if (!src) src = get_url_val(a, (unsigned char *)"src", html_context->doc_cp);
if (!src) src = get_url_val(a, (unsigned char *)"dynsrc", html_context->doc_cp);
/* If we have no label yet (no title or alt), so
* just use default ones, or image filename. */
if (!label || !*label) {
mem_free_set(&label, NULL);
/* Do we want to display images with no alt/title and with no
* link on them ?
* If not, just exit now. */
if (!options->images && !format.link) {
mem_free_if(src);
if (usemap) pop_html_element(html_context);
return;
}
add_brackets = 1;
if (usemap) {
label = stracpy((const unsigned char *)"USEMAP");
} else if (ismap) {
label = stracpy((const unsigned char *)"ISMAP");
} else {
if (display_style == 3)
label = get_image_filename_from_src(options->image_link.filename_maxlen, src);
}
} else {
label = get_image_label(options->image_link.label_maxlen, label);
}
if (!label || !*label) {
mem_free_set(&label, NULL);
add_brackets = 1;
if (display_style == 1)
label = get_image_filename_from_src(options->image_link.filename_maxlen, src);
if (!label || !*label)
mem_free_set(&label, stracpy((const unsigned char *)"IMG"));
}
mem_free_set(&format.image, NULL);
mem_free_set(&format.title, NULL);
if (label) {
int img_link_tag = options->image_link.tagging;
if (img_link_tag && (img_link_tag == 2 || add_brackets)) {
unsigned char *img_link_prefix = options->image_link.prefix;
unsigned char *img_link_suffix = options->image_link.suffix;
unsigned char *new_label = straconcat(img_link_prefix, label, img_link_suffix, (unsigned char *) NULL);
if (new_label) mem_free_set(&label, new_label);
}
if (!options->image_link.show_any_as_links) {
put_image_label(a, label, html_context);
} else {
if (src) {
format.image = join_urls(html_context->base_href, src);
}
format.title = get_attr_val(a, (unsigned char *)"title", html_context->doc_cp);
if (ismap) {
unsigned char *new_link;
html_stack_dup(html_context, ELEMENT_KILLABLE);
new_link = straconcat(format.link, (unsigned char *)"?0,0", (unsigned char *) NULL);
if (new_link)
mem_free_set(&format.link, new_link);
}
put_image_label(a, label, html_context);
if (ismap) pop_html_element(html_context);
mem_free_set(&format.image, NULL);
mem_free_set(&format.title, NULL);
}
mem_free(label);
}
mem_free_if(src);
if (usemap) pop_html_element(html_context);
}
void
html_form(struct html_context *html_context, unsigned char *a,
unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
unsigned char *al;
struct form *form;
html_context->was_br = 1;
form = init_form();
if (!form) return;
form->method = FORM_METHOD_GET;
form->form_num = a - html_context->startf;
al = get_attr_val(a, (unsigned char *)"method", html_context->doc_cp);
if (al) {
if (!c_strcasecmp((const char *)al, "post")) {
unsigned char *enctype;
enctype = get_attr_val(a, (unsigned char *)"enctype",
html_context->doc_cp);
form->method = FORM_METHOD_POST;
if (enctype) {
if (!c_strcasecmp((const char *)enctype, "multipart/form-data"))
form->method = FORM_METHOD_POST_MP;
else if (!c_strcasecmp((const char *)enctype, "text/plain"))
form->method = FORM_METHOD_POST_TEXT_PLAIN;
mem_free(enctype);
}
}
mem_free(al);
}
form->onsubmit = get_attr_val(a, (unsigned char *)"onsubmit", html_context->doc_cp);
al = get_attr_val(a, (unsigned char *)"name", html_context->doc_cp);
if (al) form->name = al;
al = get_attr_val(a, (unsigned char *)"action", html_context->doc_cp);
/* The HTML specification at
* http://www.w3.org/TR/REC-html40/interact/forms.html#h-17.3 states
* that the behavior of an empty action attribute should be undefined.
* Mozilla handles action="" as action="<current-URI>" which seems
* reasonable. (bug 615) */
if (al && *al) {
form->action = join_urls(html_context->base_href, trim_chars(al, ' ', NULL));
mem_free(al);
} else {
enum uri_component components = URI_ORIGINAL;
mem_free_if(al);
/* We have to do following for GET method, because we would end
* up with two '?' otherwise. */
if (form->method == FORM_METHOD_GET)
components = URI_FORM_GET;
form->action = get_uri_string(html_context->base_href, components);
/* No action URI should contain post data */
assert(!form->action || !strchr((char *)form->action, POST_CHAR));
/* GET method URIs should not have '?'. */
assert(!form->action
|| form->method != FORM_METHOD_GET
|| !strchr((char *)form->action, '?'));
}
al = get_target(html_context->options, a);
form->target = al ? al : stracpy(html_context->base_target);
html_context->special_f(html_context, SP_FORM, form);
}
static void
html_input_format(struct html_context *html_context, unsigned char *a,
struct form_control *fc)
{
put_chrs(html_context, (unsigned char *)" ", 1);
html_stack_dup(html_context, ELEMENT_KILLABLE);
html_focusable(html_context, a);
format.form = fc;
if (format.title) mem_free(format.title);
format.title = get_attr_val(a, (unsigned char *)"title", html_context->doc_cp);
switch (fc->type) {
case FC_TEXT:
case FC_PASSWORD:
case FC_FILE:
{
int i;
format.style.attr |= AT_BOLD;
for (i = 0; i < fc->size; i++)
put_chrs(html_context, (unsigned char *)"_", 1);
break;
}
case FC_CHECKBOX:
format.style.attr |= AT_BOLD;
put_chrs(html_context, (unsigned char *)"[ ]", 8);
break;
case FC_RADIO:
format.style.attr |= AT_BOLD;
put_chrs(html_context, (unsigned char *)"( )", 8);
break;
case FC_IMAGE:
{
unsigned char *al;
mem_free_set(&format.image, NULL);
al = get_url_val(a, (unsigned char *)"src", html_context->doc_cp);
if (!al)
al = get_url_val(a, (unsigned char *)"dynsrc",
html_context->doc_cp);
if (al) {
format.image = join_urls(html_context->base_href, al);
mem_free(al);
}
format.style.attr |= AT_BOLD;
put_chrs(html_context, (unsigned char *)"[ ", 7);
format.style.attr |= AT_NO_ENTITIES;
if (fc->alt)
put_chrs(html_context, fc->alt, strlen((const char *)fc->alt));
else if (fc->name)
put_chrs(html_context, fc->name, strlen((const char *)fc->name));
else
put_chrs(html_context, (unsigned char *)"Submit", 6);
format.style.attr &= ~AT_NO_ENTITIES;
put_chrs(html_context, (unsigned char *)" ]", 7);
break;
}
case FC_SUBMIT:
case FC_RESET:
case FC_BUTTON:
format.style.attr |= AT_BOLD;
put_chrs(html_context, (unsigned char *)"[ ", 7);
if (fc->default_value) {
format.style.attr |= AT_NO_ENTITIES;
put_chrs(html_context, fc->default_value, strlen((const char *)fc->default_value));
format.style.attr &= ~AT_NO_ENTITIES;
}
put_chrs(html_context, (unsigned char *)" ]", 7);
break;
case FC_TEXTAREA:
case FC_SELECT:
case FC_HIDDEN:
INTERNAL("bad control type");
}
pop_html_element(html_context);
put_chrs(html_context, (unsigned char *)" ", 1);
}
void html_a( struct html_context *html_context, unsigned char *a, unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5 )
{
int eax;
int ecx;
int edx;
unsigned char *href;
if ( get_attr_value( a, "href", html_context->doc_cp, HTML_ATTR_EAT_NL ) )
{
unsigned char *target;
if ( html_context->stack.list_head_elinks )
{
mem_free( &html_context->stack.list_head_elinks );
}
strlen( &href[0] );
if ( href[0] == ' ' )
{
do
{
href[0] = href[1];
*(int*)(strlen( &href[0] ))--;
}
while ( href[0] == ' ' );
}
if ( eax )
{
if ( href[ eax - 1 ] == ' ' )
{
do
{
*(char*)(edx - 1) = 0;
if ( eax == 1 )
href[0] = href[0];
else
}
while ( *(char*)(edx - 2) != ' ' );
href[0] = href[0];
}
eax += 0;
memmove( &href[0], &href[0], 1 );
&html_context->stack.list_head_elinks = ((void)( struct css_stylesheet *, struct uri *, unsigned char *, int )*)join_urls( html_context->base_href, &href[0] );
mem_free( &href[0] );
if ( get_target( html_context->options, a ) )
{
if ( html_context->stack.list_head_elinks )
{
mem_free( &html_context->stack.list_head_elinks );
html_context->stack.next = html_context->stack.next;
}
&html_context->stack.list_head_elinks = &target[0];
}
else
{
if ( html_context->stack.list_head_elinks )
{
mem_free( &html_context->stack.list_head_elinks );
}
&html_context->stack.list_head_elinks = ((void)( struct css_stylesheet *, struct uri *, unsigned char *, int )*)stracpy( html_context->base_target );
}
if ( get_global_history_item( &html_context->stack.list_head_elinks ) )
{
&html_context->stack.list_head_elinks = &html_context->stack.list_head_elinks;
&html_context->stack.list_head_elinks &= -2;
&html_context->stack.list_head_elinks |= 2;
}
else
{
&html_context->stack.list_head_elinks = html_context->stack.next ? &html_context->stack.list_head_elinks : &html_context->stack.list_head_elinks;
&html_context->stack.list_head_elinks &= -3;
&html_context->stack.list_head_elinks |= 1;
}
if ( html_context->stack.list_head_elinks )
{
mem_free( &html_context->stack.list_head_elinks );
}
&html_context->stack.list_head_elinks = ((void)( struct css_stylesheet *, struct uri *, unsigned char *, int )*)get_attr_value( a, "title", html_context->doc_cp, HTML_ATTR_NONE );
html_focusable( &html_context[0], a );
xxx3 = "name";
html_context = &html_context[0];
a = a;
}
memmove( &href[0], &href[0], 1 );
&html_context->stack.list_head_elinks = ((void)( struct css_stylesheet *, struct uri *, unsigned char *, int )*)join_urls( html_context->base_href, &href[0] );
mem_free( &href[0] );
}
else
void
html_script(struct html_context *html_context, unsigned char *a,
unsigned char *html, unsigned char *eof, unsigned char **end)
{
#ifdef CONFIG_ECMASCRIPT
/* TODO: <noscript> processing. Well, same considerations apply as to
* CSS property display: none processing. */
/* TODO: Charsets for external scripts. */
unsigned char *type, *language, *src;
int in_comment = 0;
#endif
html_skip(html_context, a);
#ifdef CONFIG_ECMASCRIPT
/* We try to process nested <script> if we didn't process the parent
* one. That's why's all the fuzz. */
/* Ref:
* http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
*/
type = get_attr_val(a, (unsigned char *)"type", html_context->doc_cp);
if (type) {
unsigned char *pos = type;
if (!c_strncasecmp((const char *)type, "text/", 5)) {
pos += 5;
} else if (!c_strncasecmp((const char *)type, "application/", 12)) {
pos += 12;
} else {
mem_free(type);
not_processed:
/* Permit nested scripts and retreat. */
html_top->invisible++;
return;
}
if (!c_strncasecmp((const char *)pos, "javascript", 10)) {
int len = strlen((const char *)pos);
if (len > 10 && !isdigit(pos[10])) {
mem_free(type);
goto not_processed;
}
} else if (c_strcasecmp((const char *)pos, "ecmascript")
&& c_strcasecmp((const char *)pos, "jscript")
&& c_strcasecmp((const char *)pos, "livescript")
&& c_strcasecmp((const char *)pos, "x-javascript")
&& c_strcasecmp((const char *)pos, "x-ecmascript")) {
mem_free(type);
goto not_processed;
}
mem_free(type);
}
/* Check that the script content is ecmascript. The value of the
* language attribute can be JavaScript with optional version digits
* postfixed (like: ``JavaScript1.1'').
* That attribute is deprecated in favor of type by HTML 4.01 */
language = get_attr_val(a, (unsigned char *)"language", html_context->doc_cp);
if (language) {
int languagelen = strlen((const char *)language);
if (languagelen < 10
|| (languagelen > 10 && !isdigit(language[10]))
|| c_strncasecmp((const char *)language, "javascript", 10)) {
mem_free(language);
html_top->invisible++;
return;
//goto not_processed;
}
mem_free(language);
}
if (html_context->part->document
&& (src = get_attr_val(a, (unsigned char *)"src", html_context->doc_cp))) {
/* External reference. */
unsigned char *import_url;
struct uri *uri;
if (!get_opt_bool((const unsigned char *)"ecmascript.enable", NULL)) {
mem_free(src);
html_top->invisible++;
return;
//goto not_processed;
}
/* HTML <head> urls should already be fine but we can.t detect them. */
import_url = join_urls(html_context->base_href, src);
mem_free(src);
if (!import_url) goto imported;
uri = get_uri(import_url, URI_BASE);
if (!uri) goto imported;
/* Request the imported script as part of the document ... */
html_context->special_f(html_context, SP_SCRIPT, uri);
done_uri(uri);
/* Create URL reference onload snippet. */
insert_in_string(&import_url, 0, (const unsigned char *)"^", 1);
add_to_string_list(&html_context->part->document->onload_snippets,
import_url, -1);
imported:
/* Retreat. Do not permit nested scripts, tho'. */
if (import_url) mem_free(import_url);
return;
}
/* Positive, grab the rest and interpret it. */
/* First position to the real script start. */
while (html < eof && *html <= ' ') html++;
if (eof - html > 4 && !strncmp((const char *)html, "<!--", 4)) {
in_comment = 1;
/* We either skip to the end of line or to -->. */
for (; *html != '\n' && *html != '\r' && eof - html >= 3; html++) {
if (!strncmp((const char *)html, "-->", 3)) {
/* This means the document is probably broken.
* We will now try to process the rest of
* <script> contents, which is however likely
* to be empty. Should we try to process the
* comment too? Currently it seems safer but
* less tolerant to broken pages, if there are
* any like this. */
html += 3;
in_comment = 0;
break;
}
}
}
*end = html;
/* Now look ahead for the script end. The <script> contents is raw
* CDATA, so we just look for the ending tag and need not care for
* any quote marks counting etc - YET, we are more tolerant and permit
* </script> stuff inside of the script if the whole <script> element
* contents is wrapped in a comment. See i.e. Mozilla bug 26857 for fun
* reading regarding this. */
for (; *end < eof; (*end)++) {
unsigned char *name;
int namelen;
if (in_comment) {
/* TODO: If we ever get some standards-quirk mode
* distinction, this should be disabled in the
* standards mode (and we should just look for CDATA
* end, which is "</"). --pasky */
if (eof - *end >= 3 && !strncmp((const char *)*end, "-->", 3)) {
/* Next iteration will jump passed the ending '>' */
(*end) += 2;
in_comment = 0;
}
continue;
/* XXX: Scan for another comment? That's admittelly
* already stretching things a little bit to an
* extreme ;-). */
}
if (**end != '<')
continue;
/* We want to land before the closing element, that's why we
* don't pass @end also as the appropriate parse_element()
* argument. */
if (parse_element(*end, eof, &name, &namelen, NULL, NULL))
continue;
if (c_strlcasecmp(name, namelen, (const unsigned char *)"/script", 7))
continue;
/* We have won! */
break;
}
if (*end >= eof) {
/* Either the document is not completely loaded yet or it's
* broken. At any rate, run away screaming. */
*end = eof; /* Just for sanity. */
return;
}
if (html_context->part->document && *html != '^') {
add_to_string_list(&html_context->part->document->onload_snippets,
html, *end - html);
}
#endif
}