static inline void
add_sgml_attribute(struct dom_stack *stack,
struct dom_scanner_token *token, struct dom_scanner_token *valtoken)
{
struct sgml_parser *parser = get_sgml_parser(stack);
struct dom_node *parent = get_dom_stack_top(stack)->node;
struct dom_string *value = valtoken ? &valtoken->string : NULL;
struct sgml_node_info *info;
struct dom_node *node;
node = add_dom_attribute(parent, &token->string, value);
info = get_sgml_node_info(parser->info->attributes, node);
node->data.attribute.type = info->type;
node->data.attribute.id = !!(info->flags & SGML_ATTRIBUTE_IDENTIFIER);
node->data.attribute.reference = !!(info->flags & SGML_ATTRIBUTE_REFERENCE);
if (valtoken && valtoken->type == SGML_TOKEN_STRING)
node->data.attribute.quoted = 1;
if (!node || !push_dom_node(stack, node))
return;
pop_dom_node(stack);
}
static inline void
add_sgml_node(struct dom_stack *stack, enum dom_node_type type, struct dom_scanner_token *token)
{
struct dom_node *parent = get_dom_stack_top(stack)->node;
struct dom_node *node = add_dom_node(parent, type, &token->string);
if (!node) return;
if (token->type == SGML_TOKEN_SPACE)
node->data.text.only_space = 1;
if (push_dom_node(stack, node))
pop_dom_node(stack);
}
static void
sgml_parsing_pop(struct dom_stack *stack, struct dom_node *node, void *data)
{
struct sgml_parser *parser = get_sgml_parser(stack);
struct sgml_parsing_state *parsing = data;
/* Pop the stack back to the state it was in. This includes cleaning
* away even immutable states left on the stack. */
while (parsing->depth < parser->stack.depth) {
get_dom_stack_top(&parser->stack)->immutable = 0;
pop_dom_node(&parser->stack);
}
assert(parsing->depth == parser->stack.depth);
}
struct dom_node *
parse_sgml(struct sgml_parser *parser, struct dom_string *buffer)
{
struct sgml_parsing_state *parsing;
if (!parser->root) {
parser->root = add_sgml_document(&parser->stack, &parser->uri);
if (!parser->root)
return NULL;
get_dom_stack_top(&parser->stack)->immutable = 1;
}
parsing = init_sgml_parsing_state(parser, buffer);
if (!parsing) return NULL;
/* FIXME: Make parse_sgml_plain() return something (error code or if
* can be guarenteed a root node). */
parse_sgml_plain(&parser->stack, &parsing->scanner);
pop_dom_node(&parser->parsing);
return parser->root;
}
void
pop_dom_state(struct dom_stack *stack, struct dom_stack_state *target)
{
struct dom_stack_state *state;
unsigned int pos;
assert(stack);
if (!target) return;
if (dom_stack_is_empty(stack)) return;
foreachback_dom_stack_state (stack, state, pos) {
/* Don't pop past states marked immutable. */
if (state->immutable)
break;
/* Pop until the target state is reached. */
pop_dom_node(stack);
if (state == target)
break;
}
}
/* Parse a CSS3 selector and add selector nodes to the @select struct. */
static enum dom_code
parse_dom_select(struct dom_select *select, struct dom_stack *stack,
struct dom_string *string)
{
struct dom_scanner scanner;
struct dom_select_node sel;
init_dom_scanner(&scanner, &dom_css_scanner_info, string, 0, 0, 1, 0, 0);
memset(&sel, 0, sizeof(sel));
while (dom_scanner_has_tokens(&scanner)) {
struct dom_scanner_token *token = get_dom_scanner_token(&scanner);
enum dom_code code;
struct dom_select_node *select_node;
assert(token);
if (token->type == '{'
|| token->type == '}'
|| token->type == ';'
|| token->type == ',')
break;
/* Examine the selector fragment */
switch (token->type) {
case CSS_TOKEN_IDENT:
sel.node.type = DOM_NODE_ELEMENT;
copy_dom_string(&sel.node.string, &token->string);
if (dom_scanner_token_contains(token, "*"))
sel.match.element |= DOM_SELECT_ELEMENT_UNIVERSAL;
break;
case CSS_TOKEN_HASH:
case CSS_TOKEN_HEX_COLOR:
/* ID fragment */
sel.node.type = DOM_NODE_ATTRIBUTE;
sel.match.attribute |= DOM_SELECT_ATTRIBUTE_ID;
/* Skip the leading '#'. */
skip_dom_scanner_token_char(token);
break;
case '[':
sel.node.type = DOM_NODE_ATTRIBUTE;
code = parse_dom_select_attribute(&sel, &scanner);
if (code != DOM_CODE_OK)
return code;
break;
case '.':
token = get_next_dom_scanner_token(&scanner);
if (!token || token->type != CSS_TOKEN_IDENT)
return DOM_CODE_SYNTAX_ERR;
sel.node.type = DOM_NODE_ATTRIBUTE;
sel.match.attribute |= DOM_SELECT_ATTRIBUTE_SPACE_LIST;
set_dom_string(&sel.node.string, "class", -1);
copy_dom_string(&sel.node.data.attribute.value, &token->string);
break;
case ':':
code = parse_dom_select_pseudo(select, &sel, &scanner);
if (code != DOM_CODE_OK)
return code;
break;
case '>':
if (get_element_relation(&sel) != DOM_SELECT_RELATION_DESCENDANT)
return DOM_CODE_SYNTAX_ERR;
sel.match.element |= DOM_SELECT_RELATION_DIRECT_CHILD;
break;
case '+':
if (get_element_relation(&sel) != DOM_SELECT_RELATION_DESCENDANT)
return DOM_CODE_SYNTAX_ERR;
sel.match.element |= DOM_SELECT_RELATION_DIRECT_ADJACENT;
break;
case '~':
if (get_element_relation(&sel) != DOM_SELECT_RELATION_DESCENDANT)
return DOM_CODE_SYNTAX_ERR;
sel.match.element |= DOM_SELECT_RELATION_INDIRECT_ADJACENT;
break;
default:
return DOM_CODE_SYNTAX_ERR;
}
skip_dom_scanner_token(&scanner);
if (sel.node.type == DOM_NODE_UNKNOWN)
continue;
select_node = mem_calloc(1, sizeof(*select_node));
copy_struct(select_node, &sel);
if (!dom_stack_is_empty(stack)) {
struct dom_node *node = &select_node->node;
struct dom_node *parent = get_dom_stack_top(stack)->node;
struct dom_node_list **list = get_dom_node_list(parent, node);
int sort = (node->type == DOM_NODE_ATTRIBUTE);
int index;
assertm(list != NULL, "Adding node to bad parent [%d -> %d]",
node->type, parent->type);
index = *list && (*list)->size > 0 && sort
? get_dom_node_map_index(*list, node) : -1;
if (!add_to_dom_node_list(list, node, index)) {
done_dom_node(node);
return DOM_CODE_ALLOC_ERR;
}
node->parent = parent;
} else {
assert(!select->selector);
select->selector = select_node;
}
code = push_dom_node(stack, &select_node->node);
if (code != DOM_CODE_OK)
return code;
if (select_node->node.type != DOM_NODE_ELEMENT)
pop_dom_node(stack);
memset(&sel, 0, sizeof(sel));
}
if (select->selector)
return DOM_CODE_OK;
return DOM_CODE_ERR;
}
static void
parse_sgml_plain(struct dom_stack *stack, struct dom_scanner *scanner)
{
struct dom_scanner_token target;
while (dom_scanner_has_tokens(scanner)) {
struct dom_scanner_token *token = get_dom_scanner_token(scanner);
switch (token->type) {
case SGML_TOKEN_ELEMENT:
case SGML_TOKEN_ELEMENT_BEGIN:
if (!add_sgml_element(stack, token)) {
if (token->type == SGML_TOKEN_ELEMENT) {
skip_dom_scanner_token(scanner);
break;
}
skip_sgml_tokens(scanner, SGML_TOKEN_TAG_END);
break;
}
if (token->type == SGML_TOKEN_ELEMENT_BEGIN) {
parse_sgml_attributes(stack, scanner);
} else {
skip_dom_scanner_token(scanner);
}
break;
case SGML_TOKEN_ELEMENT_EMPTY_END:
pop_dom_node(stack);
skip_dom_scanner_token(scanner);
break;
case SGML_TOKEN_ELEMENT_END:
if (!token->string.length) {
pop_dom_node(stack);
} else {
struct dom_string string;
struct dom_stack_state *state;
set_dom_string(&string, token->string.string, token->string.length);
state = search_dom_stack(stack, DOM_NODE_ELEMENT,
&string);
if (state) {
struct sgml_parser_state *pstate;
pstate = get_sgml_parser_state(stack, state);
copy_struct(&pstate->end_token, token);
pop_dom_state(stack, state);
}
}
skip_dom_scanner_token(scanner);
break;
case SGML_TOKEN_NOTATION_COMMENT:
add_sgml_node(stack, DOM_NODE_COMMENT, token);
skip_dom_scanner_token(scanner);
break;
case SGML_TOKEN_NOTATION_ATTLIST:
case SGML_TOKEN_NOTATION_DOCTYPE:
case SGML_TOKEN_NOTATION_ELEMENT:
case SGML_TOKEN_NOTATION_ENTITY:
case SGML_TOKEN_NOTATION:
skip_dom_scanner_token(scanner);
break;
case SGML_TOKEN_CDATA_SECTION:
add_sgml_node(stack, DOM_NODE_CDATA_SECTION, token);
skip_dom_scanner_token(scanner);
break;
case SGML_TOKEN_PROCESS_XML_STYLESHEET:
case SGML_TOKEN_PROCESS_XML:
case SGML_TOKEN_PROCESS:
copy_struct(&target, token);
/* Skip the target token */
token = get_next_dom_scanner_token(scanner);
if (!token) break;
assert(token->type == SGML_TOKEN_PROCESS_DATA);
if (add_sgml_proc_instruction(stack, &target, token)
&& (target.type == SGML_TOKEN_PROCESS_XML
|| target.type == SGML_TOKEN_PROCESS_XML_STYLESHEET)
&& token->string.length > 0) {
/* Parse the <?xml data="attributes"?>. */
struct dom_scanner attr_scanner;
init_dom_scanner(&attr_scanner, &sgml_scanner_info,
&token->string, SGML_STATE_ELEMENT,
scanner->count_lines);
if (dom_scanner_has_tokens(&attr_scanner))
parse_sgml_attributes(stack, &attr_scanner);
}
pop_dom_node(stack);
skip_dom_scanner_token(scanner);
break;
case SGML_TOKEN_ENTITY:
add_sgml_node(stack, DOM_NODE_ENTITY_REFERENCE, token);
skip_dom_scanner_token(scanner);
break;
case SGML_TOKEN_SPACE:
case SGML_TOKEN_TEXT:
default:
add_sgml_node(stack, DOM_NODE_TEXT, token);
skip_dom_scanner_token(scanner);
}
}
}
int
main(int argc, char *argv[])
{
struct sgml_parser *parser;
enum sgml_document_type doctype = SGML_DOCTYPE_HTML;
enum sgml_parser_flag flags = 0;
enum sgml_parser_type type = SGML_PARSER_STREAM;
enum dom_code code = 0;
enum dom_config_flag normalize_flags = 0;
struct dom_config config;
int normalize = 0;
int dump = 0;
int complete = 1;
size_t read_stdin = 0;
struct dom_string uri = STATIC_DOM_STRING("dom://test");
struct dom_string source = STATIC_DOM_STRING("(no source)");
int i;
for (i = 1; i < argc; i++) {
char *arg = argv[i];
if (strncmp(arg, "--", 2))
break;
arg += 2;
if (get_test_opt(&arg, "uri", &i, argc, argv, "a URI")) {
set_dom_string(&uri, arg, strlen((const char *)arg));
} else if (get_test_opt(&arg, "src", &i, argc, argv, "a string")) {
set_dom_string(&source, arg, strlen((const char *)arg));
} else if (get_test_opt(&arg, "stdin", &i, argc, argv, "a number")) {
read_stdin = atoi(arg);
flags |= SGML_PARSER_INCREMENTAL;
} else if (get_test_opt(&arg, "normalize", &i, argc, argv, "a string")) {
normalize = 1;
normalize_flags = parse_dom_config(arg, ',');
type = SGML_PARSER_TREE;
} else if (!strcmp(arg, "print-lines")) {
flags |= SGML_PARSER_COUNT_LINES;
} else if (!strcmp(arg, "incomplete")) {
flags |= SGML_PARSER_INCREMENTAL;
complete = 0;
} else if (!strcmp(arg, "dump")) {
type = SGML_PARSER_TREE;
dump = 1;
} else if (!strcmp(arg, "error")) {
flags |= SGML_PARSER_DETECT_ERRORS;
} else if (!strcmp(arg, "help")) {
die(NULL);
} else {
die("Unknown argument '%s'", arg - 2);
}
}
parser = init_sgml_parser(type, doctype, &uri, flags);
if (!parser) return 1;
parser->error_func = sgml_error_function;
if (normalize)
add_dom_config_normalizer(&parser->stack, &config, normalize_flags);
else if (!dump)
add_dom_stack_context(&parser->stack, NULL, &sgml_parser_test_context_info);
if (read_stdin > 0) {
unsigned char *buffer;
buffer = mem_alloc(read_stdin);
if (!buffer)
die("Cannot allocate buffer");
complete = 0;
while (!complete) {
size_t size = fread(buffer, 1, read_stdin, stdin);
if (ferror(stdin))
die("error reading from stdin");
complete = feof(stdin);
code = parse_sgml(parser, buffer, size, complete);
switch (code) {
case DOM_CODE_OK:
break;
case DOM_CODE_INCOMPLETE:
if (!complete) break;
/* Error */
default:
complete = 1;
}
}
mem_free(buffer);
} else {
code = parse_sgml(parser, source.string, source.length, complete);
}
if (parser->root) {
assert(!complete || parser->stack.depth > 0);
while (!dom_stack_is_empty(&parser->stack)) {
get_dom_stack_top(&parser->stack)->immutable = 0;
pop_dom_node(&parser->stack);
}
if (normalize || dump) {
struct dom_stack stack;
/* Note, that we cannot free nodes when walking the DOM
* tree since walk_dom_node() uses an index to traverse
* the tree. */
init_dom_stack(&stack, DOM_STACK_FLAG_NONE);
/* XXX: This context needs to be added first because it
* assumes the parser can be accessed via
* stack->contexts[0].data. */
if (normalize)
add_dom_stack_context(&stack, parser, &sgml_parser_test_context_info);
else if (dump)
add_sgml_file_dumper(&stack, stdout);
walk_dom_nodes(&stack, parser->root);
done_dom_stack(&stack);
done_dom_node(parser->root);
}
}
done_sgml_parser(parser);
#ifdef DEBUG_MEMLEAK
check_memory_leaks();
#endif
return code != DOM_CODE_OK ? 1 : 0;
}
/* FIXME: Instead of walking all nodes in the tree only visit those which are
* of actual interest to the contexts on the stack. */
void
walk_dom_nodes(struct dom_stack *stack, struct dom_node *root)
{
struct dom_stack_context *context;
assert(root && stack);
context = add_dom_stack_context(stack, NULL, &dom_stack_walk_context_info);
if (!context)
return;
if (push_dom_node(stack, root) != DOM_CODE_OK)
return;
while (!dom_stack_is_empty(stack)) {
struct dom_stack_state *state = get_dom_stack_top(stack);
struct dom_stack_walk_state *wstate = get_dom_stack_state_data(context, state);
struct dom_node_list *list = wstate->list;
struct dom_node *node = state->node;
switch (node->type) {
case DOM_NODE_DOCUMENT:
if (!list) list = node->data.document.children;
break;
case DOM_NODE_ELEMENT:
if (!list) list = node->data.element.map;
if (list == node->data.element.children) break;
if (is_dom_node_list_member(list, wstate->index)
&& list == node->data.element.map)
break;
list = node->data.element.children;
break;
case DOM_NODE_PROCESSING_INSTRUCTION:
if (!list) list = node->data.proc_instruction.map;
break;
case DOM_NODE_DOCUMENT_TYPE:
if (!list) list = node->data.document_type.entities;
if (list == node->data.document_type.notations) break;
if (is_dom_node_list_member(list, wstate->index)
&& list == node->data.document_type.entities)
break;
list = node->data.document_type.notations;
break;
case DOM_NODE_ATTRIBUTE:
case DOM_NODE_TEXT:
case DOM_NODE_CDATA_SECTION:
case DOM_NODE_COMMENT:
case DOM_NODE_NOTATION:
case DOM_NODE_DOCUMENT_FRAGMENT:
case DOM_NODE_ENTITY_REFERENCE:
case DOM_NODE_ENTITY:
default:
break;
}
/* Reset list state if it is a new list */
if (list != wstate->list) {
wstate->list = list;
wstate->index = 0;
}
/* If we have next child node */
if (is_dom_node_list_member(list, wstate->index)) {
struct dom_node *child = list->entries[wstate->index++];
if (push_dom_node(stack, child) == DOM_CODE_OK)
continue;
}
pop_dom_node(stack);
}
done_dom_stack_context(stack, context);
}
