/**
* Callback to determine if a node is a linking element.
*
* \param pw HTML document
* \param n DOM node
* \param match Pointer to location to receive result
* \return CSS_OK.
*
* \post \a match will contain true if the node matches and false otherwise.
*/
css_error node_is_link(void *pw, void *n, bool *match)
{
dom_node *node = n;
dom_exception exc;
dom_string *node_name = NULL;
exc = dom_node_get_node_name(node, &node_name);
if ((exc != DOM_NO_ERR) || (node_name == NULL)) {
return CSS_NOMEM;
}
if (dom_string_caseless_lwc_isequal(node_name, corestring_lwc_a)) {
bool has_href;
exc = dom_element_has_attribute(node, corestring_dom_href,
&has_href);
if ((exc == DOM_NO_ERR) && (has_href)) {
*match = true;
} else {
*match = false;
}
} else {
*match = false;
}
dom_string_unref(node_name);
return CSS_OK;
}
/**
* Callback to determine if a node has an attribute with the given name.
*
* \param pw HTML document
* \param node DOM node
* \param qname Name to match
* \param match Pointer to location to receive result
* \return CSS_OK on success,
* CSS_NOMEM on memory exhaustion.
*
* \post \a match will contain true if the node matches and false otherwise.
*/
css_error node_has_attribute(void *pw, void *node,
const css_qname *qname, bool *match)
{
dom_node *n = node;
dom_string *name;
dom_exception err;
err = dom_string_create_interned(
(const uint8_t *) lwc_string_data(qname->name),
lwc_string_length(qname->name), &name);
if (err != DOM_NO_ERR)
return CSS_NOMEM;
err = dom_element_has_attribute(n, name, match);
if (err != DOM_NO_ERR) {
dom_string_unref(name);
return CSS_OK;
}
dom_string_unref(name);
return CSS_OK;
}
/**
* process a script with a src tag
*/
static dom_hubbub_error
exec_src_script(html_content *c,
dom_node *node,
dom_string *mimetype,
dom_string *src)
{
nserror ns_error;
nsurl *joined;
hlcache_child_context child;
struct html_script *nscript;
union content_msg_data msg_data;
bool async;
bool defer;
enum html_script_type script_type;
hlcache_handle_callback script_cb;
dom_hubbub_error ret = DOM_HUBBUB_OK;
dom_exception exc; /* returned by libdom functions */
/* src url */
ns_error = nsurl_join(c->base_url, dom_string_data(src), &joined);
if (ns_error != NSERROR_OK) {
msg_data.error = messages_get("NoMemory");
content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
return DOM_HUBBUB_NOMEM;
}
LOG(("script %i '%s'", c->scripts_count, nsurl_access(joined)));
/* there are three ways to process the script tag at this point:
*
* Syncronously pause the parent parse and continue after
* the script has downloaded and executed. (default)
* Async Start the script downloading and execute it when it
* becomes available.
* Defered Start the script downloading and execute it when
* the page has completed parsing, may be set along
* with async where it is ignored.
*/
/* we interpret the presence of the async and defer attribute
* as true and ignore its value, technically only the empty
* value or the attribute name itself are valid. However
* various browsers interpret this in various ways the most
* compatible approach is to be liberal and accept any
* value. Note setting the values to "false" still makes them true!
*/
exc = dom_element_has_attribute(node, corestring_dom_async, &async);
if (exc != DOM_NO_ERR) {
return DOM_HUBBUB_OK; /* dom error */
}
if (async) {
/* asyncronous script */
script_type = HTML_SCRIPT_ASYNC;
script_cb = convert_script_async_cb;
} else {
exc = dom_element_has_attribute(node,
corestring_dom_defer, &defer);
if (exc != DOM_NO_ERR) {
return DOM_HUBBUB_OK; /* dom error */
}
if (defer) {
/* defered script */
script_type = HTML_SCRIPT_DEFER;
script_cb = convert_script_defer_cb;
} else {
/* syncronous script */
script_type = HTML_SCRIPT_SYNC;
script_cb = convert_script_sync_cb;
}
}
nscript = html_process_new_script(c, mimetype, script_type);
if (nscript == NULL) {
nsurl_unref(joined);
msg_data.error = messages_get("NoMemory");
content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
return DOM_HUBBUB_NOMEM;
}
/* set up child fetch encoding and quirks */
child.charset = c->encoding;
child.quirks = c->base.quirks;
ns_error = hlcache_handle_retrieve(joined,
0,
content_get_url(&c->base),
NULL,
script_cb,
c,
&child,
CONTENT_SCRIPT,
&nscript->data.handle);
nsurl_unref(joined);
if (ns_error != NSERROR_OK) {
/* @todo Deal with fetch error better. currently assume
* fetch never became active
*/
/* mark duff script fetch as already started */
nscript->already_started = true;
LOG(("Fetch failed with error %d",ns_error));
} else {
/* update base content active fetch count */
c->base.active++;
LOG(("%d fetches active", c->base.active));
switch (script_type) {
case HTML_SCRIPT_SYNC:
ret = DOM_HUBBUB_HUBBUB_ERR | HUBBUB_PAUSED;
case HTML_SCRIPT_ASYNC:
break;
case HTML_SCRIPT_DEFER:
break;
default:
assert(0);
}
}
return ret;
}
static bool save_complete_handle_element(save_complete_ctx *ctx,
dom_node *node, save_complete_event_type event_type)
{
dom_string *name;
dom_namednodemap *attrs;
const char *name_data;
size_t name_len;
bool process = true;
dom_exception error;
ctx->iter_state = STATE_NORMAL;
error = dom_node_get_node_name(node, &name);
if (error != DOM_NO_ERR)
return false;
if (name == NULL)
return true;
name_data = dom_string_data(name);
name_len = dom_string_byte_length(name);
if (name_len == SLEN("base") &&
strncasecmp(name_data, "base", name_len) == 0) {
/* Elide BASE elements from the output */
process = false;
} else if (name_len == SLEN("meta") &&
strncasecmp(name_data, "meta", name_len) == 0) {
/* Don't emit close tags for META elements */
if (event_type == EVENT_LEAVE) {
process = false;
} else {
/* Elide meta charsets */
dom_string *value;
error = dom_element_get_attribute(node,
corestring_dom_http_equiv, &value);
if (error != DOM_NO_ERR) {
dom_string_unref(name);
return false;
}
if (value != NULL) {
if (dom_string_length(value) ==
SLEN("Content-Type") &&
strncasecmp(dom_string_data(value),
"Content-Type",
SLEN("Content-Type")) == 0)
process = false;
dom_string_unref(value);
} else {
bool yes;
error = dom_element_has_attribute(node,
corestring_dom_charset, &yes);
if (error != DOM_NO_ERR) {
dom_string_unref(name);
return false;
}
if (yes)
process = false;
}
}
} else if (event_type == EVENT_LEAVE &&
((name_len == SLEN("link") &&
strncasecmp(name_data, "link", name_len) == 0))) {
/* Don't emit close tags for void elements */
process = false;
}
if (process == false) {
dom_string_unref(name);
return true;
}
fputc('<', ctx->fp);
if (event_type == EVENT_LEAVE)
fputc('/', ctx->fp);
fwrite(name_data, sizeof(*name_data), name_len, ctx->fp);
if (event_type == EVENT_ENTER) {
error = dom_node_get_attributes(node, &attrs);
if (error != DOM_NO_ERR) {
dom_string_unref(name);
return false;
}
if (save_complete_handle_attrs(ctx, name, attrs) == false) {
dom_namednodemap_unref(attrs);
dom_string_unref(name);
return false;
}
dom_namednodemap_unref(attrs);
}
fputc('>', ctx->fp);
/* Rewrite contents of style elements */
if (event_type == EVENT_ENTER && name_len == SLEN("style") &&
strncasecmp(name_data, "style", name_len) == 0) {
dom_string *content;
error = dom_node_get_text_content(node, &content);
if (error != DOM_NO_ERR) {
dom_string_unref(name);
return false;
}
if (content != NULL) {
char *rewritten;
unsigned long len;
/* Rewrite @import rules */
rewritten = save_complete_rewrite_stylesheet_urls(
ctx,
dom_string_data(content),
dom_string_byte_length(content),
ctx->base,
&len);
if (rewritten == NULL) {
dom_string_unref(content);
dom_string_unref(name);
return false;
}
dom_string_unref(content);
fwrite(rewritten, sizeof(*rewritten), len, ctx->fp);
free(rewritten);
}
ctx->iter_state = STATE_IN_STYLE;
} else if (event_type == EVENT_ENTER && name_len == SLEN("head") &&
strncasecmp(name_data, "head", name_len) == 0) {
/* If this is a HEAD element, insert a meta charset */
fputs("<META http-equiv=\"Content-Type\" "
"content=\"text/html; charset=utf-8\">",
ctx->fp);
}
dom_string_unref(name);
return true;
}
/**
* Adds an imagemap entry to the list
*
* \param c The html content that the imagemap belongs to
* \param n The xmlNode representing the entry to add
* \param base_url Base URL for resolving relative URLs
* \param entry Pointer to list of entries
* \param tagtype The type of tag
* \return false on memory exhaustion, true otherwise
*/
static bool
imagemap_addtolist(const struct html_content *c, dom_node *n, nsurl *base_url,
struct mapentry **entry, dom_string *tagtype)
{
dom_exception exc;
dom_string *href = NULL, *target = NULL, *shape = NULL;
dom_string *coords = NULL;
struct mapentry *new_map, *temp;
bool ret = true;
if (dom_string_caseless_isequal(tagtype, corestring_dom_area)) {
bool nohref = false;
exc = dom_element_has_attribute(n,
corestring_dom_nohref, &nohref);
if ((exc != DOM_NO_ERR) || nohref)
/* Skip <area nohref="anything" /> */
goto ok_out;
}
exc = dom_element_get_attribute(n, corestring_dom_href, &href);
if (exc != DOM_NO_ERR || href == NULL) {
/* No href="" attribute, skip this element */
goto ok_out;
}
exc = dom_element_get_attribute(n, corestring_dom_target, &target);
if (exc != DOM_NO_ERR) {
goto ok_out;
}
exc = dom_element_get_attribute(n, corestring_dom_shape, &shape);
if (exc != DOM_NO_ERR) {
goto ok_out;
}
/* If there's no shape, we default to rectangles */
if (shape == NULL)
shape = dom_string_ref(corestring_dom_rect);
if (!dom_string_caseless_lwc_isequal(shape, corestring_lwc_default)) {
/* If not 'default' and there's no 'coords' give up */
exc = dom_element_get_attribute(n, corestring_dom_coords,
&coords);
if (exc != DOM_NO_ERR || coords == NULL) {
goto ok_out;
}
}
new_map = calloc(1, sizeof(*new_map));
if (new_map == NULL) {
goto bad_out;
}
if (dom_string_caseless_lwc_isequal(shape, corestring_lwc_rect) ||
dom_string_caseless_lwc_isequal(shape, corestring_lwc_rectangle))
new_map->type = IMAGEMAP_RECT;
else if (dom_string_caseless_lwc_isequal(shape, corestring_lwc_circle))
new_map->type = IMAGEMAP_CIRCLE;
else if (dom_string_caseless_lwc_isequal(shape, corestring_lwc_poly) ||
dom_string_caseless_lwc_isequal(shape, corestring_lwc_polygon))
new_map->type = IMAGEMAP_POLY;
else if (dom_string_caseless_lwc_isequal(shape, corestring_lwc_default))
new_map->type = IMAGEMAP_DEFAULT;
else
goto bad_out;
if (box_extract_link(c, href, base_url, &new_map->url) == false)
goto bad_out;
if (new_map->url == NULL) {
/* non-fatal error -> ignore this */
goto ok_free_map_out;
}
if (target != NULL) {
/* Copy target into the map */
new_map->target = malloc(dom_string_byte_length(target) + 1);
if (new_map->target == NULL)
goto bad_out;
/* Safe, but relies on dom_strings being NULL terminated */
/* \todo Do this better */
strcpy(new_map->target, dom_string_data(target));
}
if (new_map->type != IMAGEMAP_DEFAULT) {
int x, y;
float *xcoords, *ycoords;
/* coordinates are a comma-separated list of values */
char *val = strtok((char *)dom_string_data(coords), ",");
int num = 1;
switch (new_map->type) {
case IMAGEMAP_RECT:
/* (left, top, right, bottom) */
while (val != NULL && num <= 4) {
switch (num) {
case 1:
new_map->bounds.rect.x0 = atoi(val);
break;
case 2:
new_map->bounds.rect.y0 = atoi(val);
break;
case 3:
new_map->bounds.rect.x1 = atoi(val);
break;
case 4:
new_map->bounds.rect.y1 = atoi(val);
break;
}
num++;
val = strtok(NULL, ",");
}
break;
case IMAGEMAP_CIRCLE:
/* (x, y, radius ) */
while (val != NULL && num <= 3) {
switch (num) {
case 1:
new_map->bounds.circle.x = atoi(val);
break;
case 2:
new_map->bounds.circle.y = atoi(val);
break;
case 3:
new_map->bounds.circle.r = atoi(val);
break;
}
num++;
val = strtok(NULL, ",");
}
break;
case IMAGEMAP_POLY:
new_map->bounds.poly.xcoords = NULL;
new_map->bounds.poly.ycoords = NULL;
while (val != NULL) {
x = atoi(val);
val = strtok(NULL, ",");
if (val == NULL)
break;
y = atoi(val);
xcoords = realloc(new_map->bounds.poly.xcoords,
num * sizeof(float));
if (xcoords == NULL) {
goto bad_out;
}
new_map->bounds.poly.xcoords = xcoords;
ycoords = realloc(new_map->bounds.poly.ycoords,
num * sizeof(float));
if (ycoords == NULL) {
goto bad_out;
}
new_map->bounds.poly.ycoords = ycoords;
new_map->bounds.poly.xcoords[num - 1] = x;
new_map->bounds.poly.ycoords[num - 1] = y;
num++;
val = strtok(NULL, ",");
}
new_map->bounds.poly.num = num - 1;
break;
default:
break;
}
}
new_map->next = NULL;
if (*entry) {
/* add to END of list */
for (temp = (*entry); temp->next != NULL; temp = temp->next)
;
temp->next = new_map;
} else {
(*entry) = new_map;
}
/* All good, linked in, let's clean up */
goto ok_out;
bad_out:
ret = false;
ok_free_map_out:
if (new_map != NULL) {
if (new_map->url != NULL)
nsurl_unref(new_map->url);
if (new_map->type == IMAGEMAP_POLY &&
new_map->bounds.poly.ycoords != NULL)
free(new_map->bounds.poly.ycoords);
if (new_map->type == IMAGEMAP_POLY &&
new_map->bounds.poly.xcoords != NULL)
free(new_map->bounds.poly.xcoords);
if (new_map->target != NULL)
free(new_map->target);
free(new_map);
}
ok_out:
if (href != NULL)
dom_string_unref(href);
if (target != NULL)
dom_string_unref(target);
if (shape != NULL)
dom_string_unref(shape);
if (coords != NULL)
dom_string_unref(coords);
return ret;
}
