/* convert the whole document to EBCDIC */
APU_DECLARE(apr_status_t) apr_xml_parser_convert_doc(apr_pool_t *pool,
apr_xml_doc *pdoc,
apr_xlate_t *convset)
{
apr_status_t status;
/* Don't convert the namespaces: they are constant! */
if (pdoc->namespaces != NULL) {
int i;
apr_array_header_t *namespaces;
namespaces = apr_array_make(pool, pdoc->namespaces->nelts, sizeof(const char *));
if (namespaces == NULL)
return APR_ENOMEM;
for (i = 0; i < pdoc->namespaces->nelts; i++) {
apr_size_t inbytes_left, outbytes_left;
char *ptr = (char *) APR_XML_GET_URI_ITEM(pdoc->namespaces, i);
ptr = apr_pstrdup(pool, ptr);
if ( ptr == NULL)
return APR_ENOMEM;
inbytes_left = outbytes_left = strlen(ptr);
status = apr_xlate_conv_buffer(convset, ptr, &inbytes_left, ptr, &outbytes_left);
if (status) {
return status;
}
apr_xml_insert_uri(namespaces, ptr);
}
pdoc->namespaces = namespaces;
}
return apr_xml_parser_convert_elem(pdoc->root, convset);
}
APR_DECLARE(apr_xml_parser *) apr_xml_parser_create_ex(apr_pool_t *pool,
void *start_func, void *end_func, void *cdata_func)
{
apr_xml_parser *parser = apr_pcalloc(pool, sizeof(*parser));
parser->impl = apr_xml_get_parser_impl();
parser->p = pool;
parser->doc = apr_pcalloc(pool, sizeof(*parser->doc));
parser->doc->namespaces = apr_array_make(pool, 5, sizeof(const char *));
/* ### is there a way to avoid hard-coding this? */
apr_xml_insert_uri(parser->doc->namespaces, APR_KW_DAV);
parser->xp = XML_ParserCreate(NULL);
if (parser->xp == NULL) {
(*apr_pool_abort_get(pool))(APR_ENOMEM);
return NULL;
}
apr_pool_cleanup_register(pool, parser, cleanup_parser,
apr_pool_cleanup_null);
XML_SetUserData(parser->xp, parser);
XML_SetElementHandler(parser->xp, start_func, end_func);
XML_SetCharacterDataHandler(parser->xp, cdata_func);
/* Prevent the "billion laughs" attack against expat by disabling
* internal entity expansion. With 2.x, forcibly stop the parser
* if an entity is declared - this is safer and a more obvious
* failure mode. With older versions, installing a noop
* DefaultHandler means that internal entities will be expanded as
* the empty string, which is also sufficient to prevent the
* attack. */
#if XML_MAJOR_VERSION > 1
XML_SetEntityDeclHandler(parser->xp, entity_declaration);
#else
XML_SetDefaultHandler(parser->xp, default_handler);
#endif
return parser;
}
static void start_handler(void *userdata, const char *name, const char **attrs)
{
apr_xml_parser *parser = userdata;
apr_xml_elem *elem;
apr_xml_attr *attr;
apr_xml_attr *prev;
char *colon;
const char *quoted;
char *elem_name;
/* punt once we find an error */
if (parser->error)
return;
elem = apr_pcalloc(parser->p, sizeof(*elem));
/* prep the element */
elem->name = elem_name = apr_pstrdup(parser->p, name);
/* fill in the attributes (note: ends up in reverse order) */
while (*attrs) {
attr = apr_palloc(parser->p, sizeof(*attr));
attr->name = apr_pstrdup(parser->p, *attrs++);
attr->value = apr_pstrdup(parser->p, *attrs++);
attr->next = elem->attr;
elem->attr = attr;
}
/* hook the element into the tree */
if (parser->cur_elem == NULL) {
/* no current element; this also becomes the root */
parser->cur_elem = parser->doc->root = elem;
}
else {
/* this element appeared within the current elem */
elem->parent = parser->cur_elem;
/* set up the child/sibling links */
if (elem->parent->last_child == NULL) {
/* no first child either */
elem->parent->first_child = elem->parent->last_child = elem;
}
else {
/* hook onto the end of the parent's children */
elem->parent->last_child->next = elem;
elem->parent->last_child = elem;
}
/* this element is now the current element */
parser->cur_elem = elem;
}
/* scan the attributes for namespace declarations */
for (prev = NULL, attr = elem->attr;
attr;
attr = attr->next) {
if (strncmp(attr->name, APR_KW_xmlns, 5) == 0) {
const char *prefix = &attr->name[5];
apr_xml_ns_scope *ns_scope;
/* test for xmlns:foo= form and xmlns= form */
if (*prefix == 0x3A) {
/* a namespace prefix declaration must have a
non-empty value. */
if (attr->value[0] == '\0') {
parser->error = APR_XML_NS_ERROR_INVALID_DECL;
return;
}
++prefix;
}
else if (*prefix != '\0') {
/* advance "prev" since "attr" is still present */
prev = attr;
continue;
}
/* quote the URI before we ever start working with it */
quoted = apr_xml_quote_string(parser->p, attr->value, 1);
/* build and insert the new scope */
ns_scope = apr_pcalloc(parser->p, sizeof(*ns_scope));
ns_scope->prefix = prefix;
ns_scope->ns = apr_xml_insert_uri(parser->doc->namespaces, quoted);
ns_scope->emptyURI = *quoted == '\0';
ns_scope->next = elem->ns_scope;
elem->ns_scope = ns_scope;
/* remove this attribute from the element */
if (prev == NULL)
elem->attr = attr->next;
else
prev->next = attr->next;
/* Note: prev will not be advanced since we just removed "attr" */
}
else if (strcmp(attr->name, APR_KW_xmlns_lang) == 0) {
/* save away the language (in quoted form) */
elem->lang = apr_xml_quote_string(parser->p, attr->value, 1);
/* remove this attribute from the element */
if (prev == NULL)
elem->attr = attr->next;
else
prev->next = attr->next;
/* Note: prev will not be advanced since we just removed "attr" */
}
else {
/* advance "prev" since "attr" is still present */
prev = attr;
}
}
/*
** If an xml:lang attribute didn't exist (lang==NULL), then copy the
** language from the parent element (if present).
**
** NOTE: elem_size() *depends* upon this pointer equality.
*/
if (elem->lang == NULL && elem->parent != NULL)
elem->lang = elem->parent->lang;
/* adjust the element's namespace */
colon = strchr(elem_name, 0x3A);
if (colon == NULL) {
/*
* The element is using the default namespace, which will always
* be found. Either it will be "no namespace", or a default
* namespace URI has been specified at some point.
*/
elem->ns = find_prefix(parser, "");
}
else if (APR_XML_NS_IS_RESERVED(elem->name)) {
elem->ns = APR_XML_NS_NONE;
}
else {
*colon = '\0';
elem->ns = find_prefix(parser, elem->name);
elem->name = colon + 1;
if (APR_XML_NS_IS_ERROR(elem->ns)) {
parser->error = elem->ns;
return;
}
}
/* adjust all remaining attributes' namespaces */
for (attr = elem->attr; attr; attr = attr->next) {
/*
* apr_xml_attr defines this as "const" but we dup'd it, so we
* know that we can change it. a bit hacky, but the existing
* structure def is best.
*/
char *attr_name = (char *)attr->name;
colon = strchr(attr_name, 0x3A);
if (colon == NULL) {
/*
* Attributes do NOT use the default namespace. Therefore,
* we place them into the "no namespace" category.
*/
attr->ns = APR_XML_NS_NONE;
}
else if (APR_XML_NS_IS_RESERVED(attr->name)) {
attr->ns = APR_XML_NS_NONE;
}
else {
*colon = '\0';
attr->ns = find_prefix(parser, attr->name);
attr->name = colon + 1;
if (APR_XML_NS_IS_ERROR(attr->ns)) {
parser->error = attr->ns;
return;
}
}
}
}
