/* Initialise the document */
dom_exception _dom_document_initialise(dom_document *doc,
dom_events_default_action_fetcher daf)
{
dom_exception err;
dom_string *name;
err = dom_string_create((const uint8_t *) "#document",
SLEN("#document"), &name);
if (err != DOM_NO_ERR)
return err;
doc->nodelists = NULL;
err = _dom_node_initialise(&doc->base, doc, DOM_DOCUMENT_NODE,
name, NULL, NULL, NULL);
dom_string_unref(name);
if (err != DOM_NO_ERR)
return err;
list_init(&doc->pending_nodes);
doc->id_name = NULL;
doc->quirks = DOM_DOCUMENT_QUIRKS_MODE_NONE;
err = dom_string_create_interned((const uint8_t *) "class",
SLEN("class"), &doc->class_string);
if (err != DOM_NO_ERR)
return err;
/* We should not pass a NULL when all things hook up */
return _dom_document_event_internal_initialise(doc, &doc->dei, daf);
}
static hubbub_error add_attributes(void *parser, void *node,
const hubbub_attribute *attributes, uint32_t n_attributes)
{
dom_hubbub_parser *dom_parser = (dom_hubbub_parser *) parser;
dom_exception err;
uint32_t i;
for (i = 0; i < n_attributes; i++) {
dom_string *name, *value;
err = dom_string_create_interned(attributes[i].name.ptr,
attributes[i].name.len, &name);
if (err != DOM_NO_ERR) {
dom_parser->msg(DOM_MSG_CRITICAL, dom_parser->mctx,
"Can't create attribute name");
goto fail;
}
err = dom_string_create(attributes[i].value.ptr,
attributes[i].value.len, &value);
if (err != DOM_NO_ERR) {
dom_parser->msg(DOM_MSG_CRITICAL, dom_parser->mctx,
"Can't create attribute value");
dom_string_unref(name);
goto fail;
}
if (attributes[i].ns == HUBBUB_NS_NULL) {
err = dom_element_set_attribute(
(struct dom_element *) node, name,
value);
dom_string_unref(name);
dom_string_unref(value);
if (err != DOM_NO_ERR) {
dom_parser->msg(DOM_MSG_CRITICAL,
dom_parser->mctx,
"Can't add attribute");
}
} else {
err = dom_element_set_attribute_ns(
(struct dom_element *) node,
dom_namespaces[attributes[i].ns], name,
value);
dom_string_unref(name);
dom_string_unref(value);
if (err != DOM_NO_ERR) {
dom_parser->msg(DOM_MSG_CRITICAL,
dom_parser->mctx,
"Can't add attribute ns");
}
}
}
return HUBBUB_OK;
fail:
return HUBBUB_UNKNOWN;
}
/**
* Callback to determine if a node has an attribute with the given name whose
* value contains the substring given.
*
* \param pw HTML document
* \param node DOM node
* \param qname Name to match
* \param value Value 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_substring(void *pw, void *node,
const css_qname *qname, lwc_string *value,
bool *match)
{
dom_node *n = node;
dom_string *name;
dom_string *atr_val;
dom_exception err;
size_t vlen = lwc_string_length(value);
if (vlen == 0) {
*match = false;
return CSS_OK;
}
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_get_attribute(n, name, &atr_val);
if ((err != DOM_NO_ERR) || (atr_val == NULL)) {
dom_string_unref(name);
*match = false;
return CSS_OK;
}
dom_string_unref(name);
/* check for exact match */
*match = dom_string_caseless_lwc_isequal(atr_val, value);
/* check for prefix match */
if (*match == false) {
const char *vdata = lwc_string_data(value);
const char *start = (const char *) dom_string_data(atr_val);
size_t len = dom_string_byte_length(atr_val);
const char *last_start = start + len - vlen;
if (len >= vlen) {
while (start <= last_start) {
if (strncasecmp(start, vdata,
vlen) == 0) {
*match = true;
break;
}
start++;
}
}
}
dom_string_unref(atr_val);
return CSS_OK;
}
/**
* Callback to determine if a node has an attribute with the given name whose
* value includes that given.
*
* \param pw HTML document
* \param node DOM node
* \param qname Name to match
* \param value Value 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_includes(void *pw, void *node,
const css_qname *qname, lwc_string *value,
bool *match)
{
dom_node *n = node;
dom_string *name;
dom_string *atr_val;
dom_exception err;
size_t vlen = lwc_string_length(value);
const char *p;
const char *start;
const char *end;
*match = false;
if (vlen == 0) {
return CSS_OK;
}
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_get_attribute(n, name, &atr_val);
if ((err != DOM_NO_ERR) || (atr_val == NULL)) {
dom_string_unref(name);
*match = false;
return CSS_OK;
}
dom_string_unref(name);
/* check for match */
start = (const char *) dom_string_data(atr_val);
end = start + dom_string_byte_length(atr_val);
for (p = start; p <= end; p++) {
if (*p == ' ' || *p == '\0') {
if ((size_t) (p - start) == vlen &&
strncasecmp(start,
lwc_string_data(value),
vlen) == 0) {
*match = true;
break;
}
start = p + 1;
}
}
dom_string_unref(atr_val);
return CSS_OK;
}
static hubbub_error create_element(void *parser, const hubbub_tag *tag,
void **result)
{
dom_hubbub_parser *dom_parser = (dom_hubbub_parser *) parser;
dom_exception err;
dom_string *name;
struct dom_element *element = NULL;
hubbub_error herr;
*result = NULL;
err = dom_string_create_interned(tag->name.ptr, tag->name.len, &name);
if (err != DOM_NO_ERR) {
dom_parser->msg(DOM_MSG_CRITICAL, dom_parser->mctx,
"Can't create element name");
goto fail;
}
if (tag->ns == HUBBUB_NS_NULL) {
err = dom_document_create_element(dom_parser->doc, name,
&element);
if (err != DOM_NO_ERR) {
dom_parser->msg(DOM_MSG_CRITICAL, dom_parser->mctx,
"Can't create the DOM element");
goto clean1;
}
} else {
err = dom_document_create_element_ns(dom_parser->doc,
dom_namespaces[tag->ns], name, &element);
if (err != DOM_NO_ERR) {
dom_parser->msg(DOM_MSG_CRITICAL, dom_parser->mctx,
"Can't create the DOM element");
goto clean1;
}
}
if (element != NULL && tag->n_attributes > 0) {
herr = add_attributes(parser, element, tag->attributes,
tag->n_attributes);
if (herr != HUBBUB_OK)
goto clean1;
}
*result = element;
clean1:
dom_string_unref(name);
fail:
if (*result == NULL)
return HUBBUB_UNKNOWN;
else
return HUBBUB_OK;
}
dom_exception dom_element_get_attribute(dom_node *node, dom_string *string, dom_string **outAttribute) {
/* xmlAttrPtr attrPtr = xmlHasProp(node->node, (const xmlChar *)string->s); */
const char *attr = ezxml_attr(node->node, string->s);
if (NULL == attr) {
if (strchr(string->s, ':')) {
fprintf(stderr, "TODO:dom_element_get_attribute - namespace %s\n", string->s);
}
} else {
return dom_string_create_interned((const uint8_t *)attr, strlen(attr), outAttribute);
}
*outAttribute = NULL;
return DOM_NO_ERR;
}
/**
* Callback to determine if a node has an attribute with the given name whose
* value dashmatches that given.
*
* \param pw HTML document
* \param node DOM node
* \param qname Name to match
* \param value Value 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_dashmatch(void *pw, void *node,
const css_qname *qname, lwc_string *value,
bool *match)
{
dom_node *n = node;
dom_string *name;
dom_string *atr_val;
dom_exception err;
size_t vlen = lwc_string_length(value);
if (vlen == 0) {
*match = false;
return CSS_OK;
}
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_get_attribute(n, name, &atr_val);
if ((err != DOM_NO_ERR) || (atr_val == NULL)) {
dom_string_unref(name);
*match = false;
return CSS_OK;
}
dom_string_unref(name);
/* check for exact match */
*match = dom_string_caseless_lwc_isequal(atr_val, value);
/* check for dashmatch */
if (*match == false) {
const char *vdata = lwc_string_data(value);
const char *data = (const char *) dom_string_data(atr_val);
size_t len = dom_string_byte_length(atr_val);
if (len > vlen && data[vlen] == '-' &&
strncasecmp(data, vdata, vlen) == 0) {
*match = true;
}
}
dom_string_unref(atr_val);
return CSS_OK;
}
/** Convert the given string to lowercase
*
* \param source
* \param ascii_only Whether to only convert [a-z] to [A-Z]
* \param lower Result pointer for lowercase string. Caller owns ref
*
* \return DOM_NO_ERR on success.
*
* \note Right now, will return DOM_NOT_SUPPORTED_ERR if ascii_only is false.
*/
dom_exception
dom_string_tolower(dom_string *source, bool ascii_only, dom_string **lower)
{
const uint8_t *orig_s = (const uint8_t *) dom_string_data(source);
const size_t nbytes = dom_string_byte_length(source);
uint8_t *copy_s;
size_t index = 0, clen;
parserutils_error err;
dom_exception exc;
if (ascii_only == false)
return DOM_NOT_SUPPORTED_ERR;
copy_s = malloc(nbytes);
if (copy_s == NULL)
return DOM_NO_MEM_ERR;
memcpy(copy_s, orig_s, nbytes);
while (index < nbytes) {
err = parserutils_charset_utf8_char_byte_length(orig_s + index,
&clen);
if (err != PARSERUTILS_OK) {
free(copy_s);
/** \todo Find a better exception */
return DOM_NO_MEM_ERR;
}
if (clen == 1) {
if (orig_s[index] >= 'A' &&
orig_s[index] <= 'Z')
copy_s[index] += 'a' - 'A';
}
index += clen;
}
if (((dom_string_internal*)source)->type == DOM_STRING_CDATA) {
exc = dom_string_create(copy_s, nbytes, lower);
} else {
exc = dom_string_create_interned(copy_s, nbytes, lower);
}
free(copy_s);
return exc;
}
/**
* 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;
}
/**
* Callback to determine if a node has an attribute with given name and value.
*
* \param pw HTML document
* \param node DOM node
* \param qname Name to match
* \param value Value 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_equal(void *pw, void *node,
const css_qname *qname, lwc_string *value,
bool *match)
{
dom_node *n = node;
dom_string *name;
dom_string *atr_val;
dom_exception err;
size_t vlen = lwc_string_length(value);
if (vlen == 0) {
*match = false;
return CSS_OK;
}
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_get_attribute(n, name, &atr_val);
if ((err != DOM_NO_ERR) || (atr_val == NULL)) {
dom_string_unref(name);
*match = false;
return CSS_OK;
}
dom_string_unref(name);
*match = dom_string_caseless_lwc_isequal(atr_val, value);
dom_string_unref(atr_val);
return CSS_OK;
}
lwc_error lwc_intern_string(const char *data, size_t len, lwc_string **outString) {
return (lwc_error)dom_string_create_interned((const uint8_t *)data, len, outString);
}
dom_exception dom_node_get_node_name(dom_node *node, dom_string **outString) {
ezxml_t n = node->node;
dom_exception errCode = DOM_NO_ERR;
errCode = dom_string_create_interned(n->name, strlen((const char *)n->name), outString);
return errCode;
}
/* exported function documented in string.h */
dom_exception dom_string_whitespace_op(dom_string *s,
enum dom_whitespace_op op, dom_string **ret)
{
const uint8_t *src_text = (const uint8_t *) dom_string_data(s);
size_t len = dom_string_byte_length(s);
const uint8_t *src_pos;
const uint8_t *src_end;
dom_exception exc;
uint8_t *temp_pos;
uint8_t *temp;
if (len == 0) {
*ret = dom_string_ref(s);
}
temp = malloc(len);
if (temp == NULL) {
return DOM_NO_MEM_ERR;
}
src_pos = src_text;
src_end = src_text + len;
temp_pos = temp;
if (op & DOM_WHITESPACE_STRIP_LEADING) {
while (src_pos < src_end) {
if (*src_pos == ' ' || *src_pos == '\t' ||
*src_pos == '\n' || *src_pos == '\r' ||
*src_pos == '\f')
src_pos++;
else
break;
}
}
while (src_pos < src_end) {
if ((op & DOM_WHITESPACE_COLLAPSE) &&
(*src_pos == ' ' || *src_pos == '\t' ||
*src_pos == '\n' || *src_pos == '\r' ||
*src_pos == '\f')) {
/* Got a whitespace character */
do {
/* Skip all adjacent whitespace */
src_pos++;
} while (src_pos < src_end &&
(*src_pos == ' ' || *src_pos == '\t' ||
*src_pos == '\n' || *src_pos == '\r' ||
*src_pos == '\f'));
/* Gets replaced with single space in output */
*temp_pos++ = ' ';
} else {
/* Otherwise, copy to output */
*temp_pos++ = *src_pos++;
}
}
if (op & DOM_WHITESPACE_STRIP_TRAILING) {
while (temp_pos > temp) {
temp_pos--;
if (*temp_pos != ' ' && *temp_pos != '\t' &&
*temp_pos != '\n' && *temp_pos != '\r' &&
*temp_pos != '\f') {
temp_pos++;
break;
}
}
}
/* New length */
len = temp_pos - temp;
/* Make new string */
if (((dom_string_internal *) s)->type == DOM_STRING_CDATA) {
exc = dom_string_create(temp, len, ret);
} else {
exc = dom_string_create_interned(temp, len, ret);
}
free(temp);
return exc;
}
/**
* Create a Hubbub parser instance
*
* \param params The binding creation parameters
* \param parser Pointer to location to recive instance.
* \param document Pointer to location to receive document.
* \return Error code
*/
dom_hubbub_error
dom_hubbub_parser_create(dom_hubbub_parser_params *params,
dom_hubbub_parser **parser,
dom_document **document)
{
dom_hubbub_parser *binding;
hubbub_parser_optparams optparams;
hubbub_error error;
dom_exception err;
dom_string *idname = NULL;
/* check result parameters */
if (document == NULL) {
return DOM_HUBBUB_BADPARM;
}
if (parser == NULL) {
return DOM_HUBBUB_BADPARM;
}
/* setup binding parser context */
binding = malloc(sizeof(dom_hubbub_parser));
if (binding == NULL) {
return DOM_HUBBUB_NOMEM;
}
binding->parser = NULL;
binding->doc = NULL;
binding->encoding = params->enc;
if (params->enc != NULL) {
binding->encoding_source = DOM_HUBBUB_ENCODING_SOURCE_HEADER;
} else {
binding->encoding_source = DOM_HUBBUB_ENCODING_SOURCE_DETECTED;
}
binding->complete = false;
if (params->msg == NULL) {
binding->msg = dom_hubbub_parser_default_msg;
} else {
binding->msg = params->msg;
}
binding->mctx = params->ctx;
/* ensure script function is valid or use the default */
if (params->script == NULL) {
binding->script = dom_hubbub_parser_default_script;
} else {
binding->script = params->script;
}
/* create hubbub parser */
error = hubbub_parser_create(binding->encoding,
params->fix_enc,
dom_hubbub_alloc,
NULL,
&binding->parser);
if (error != HUBBUB_OK) {
free(binding);
return (DOM_HUBBUB_HUBBUB_ERR | error);
}
/* create DOM document */
err = dom_implementation_create_document(DOM_IMPLEMENTATION_HTML,
NULL,
NULL,
NULL,
params->daf,
params->ctx,
&binding->doc);
if (err != DOM_NO_ERR) {
hubbub_parser_destroy(binding->parser);
free(binding);
return DOM_HUBBUB_DOM;
}
binding->tree_handler = tree_handler;
binding->tree_handler.ctx = (void *)binding;
/* set tree handler on parser */
optparams.tree_handler = &binding->tree_handler;
hubbub_parser_setopt(binding->parser,
HUBBUB_PARSER_TREE_HANDLER,
&optparams);
/* set document node*/
optparams.document_node = dom_node_ref((struct dom_node *)binding->doc);
hubbub_parser_setopt(binding->parser,
HUBBUB_PARSER_DOCUMENT_NODE,
&optparams);
/* set scripting state */
optparams.enable_scripting = params->enable_script;
hubbub_parser_setopt(binding->parser,
HUBBUB_PARSER_ENABLE_SCRIPTING,
&optparams);
/* set the document id parameter before the parse so searches
* based on id succeed.
*/
err = dom_string_create_interned((const uint8_t *) "id",
SLEN("id"),
&idname);
if (err != DOM_NO_ERR) {
binding->msg(DOM_MSG_ERROR, binding->mctx, "Can't set DOM document id name");
hubbub_parser_destroy(binding->parser);
free(binding);
return DOM_HUBBUB_DOM;
}
_dom_document_set_id_name(binding->doc, idname);
dom_string_unref(idname);
/* set return parameters */
*document = (dom_document *)dom_node_ref(binding->doc);
*parser = binding;
return DOM_HUBBUB_OK;
}
