#include "treebuilder/treebuilder.h"
#include <stdio.h>

Classes
struct	element_context
	Item on the element stack. More...

struct	formatting_list_entry
	Entry in a formatting list. More...

struct	hubbub_treebuilder_context
	Context for a tree builder. More...

struct	hubbub_treebuilder
	Treebuilder object. More...

Macros
#define	ELEMENT_STACK_CHUNK 128

Typedefs
typedef struct element_context	element_context
	Item on the element stack.

typedef struct formatting_list_entry	formatting_list_entry
	Entry in a formatting list.

typedef struct hubbub_treebuilder_context	hubbub_treebuilder_context
	Context for a tree builder.

Enumerations
enum	element_type { ADDRESS , AREA , ARTICLE , ASIDE , BASE , BASEFONT , BGSOUND , BLOCKQUOTE , BODY , BR , CENTER , COL , COLGROUP , COMMAND , DATAGRID , DD , DETAILS , DIALOG , DIR , DIV , DL , DT , EMBED , FIELDSET , FIGCAPTION , FIGURE , FOOTER , FORM , FRAME , FRAMESET , H1 , H2 , H3 , H4 , H5 , H6 , HEAD , HEADER , HR , IFRAME , IMAGE , IMG , INPUT , ISINDEX , LI , LINK , LISTING , MAIN , MENU , META , NAV , NOEMBED , NOFRAMES , NOSCRIPT , OL , OPTGROUP , OPTION , P , PARAM , PLAINTEXT , PRE , SCRIPT , SECTION , SELECT , SPACER , STYLE , SUMMARY , TBODY , TEXTAREA , TFOOT , THEAD , TITLE , TR , UL , WBR , APPLET , BUTTON , CAPTION , HTML , MARQUEE , OBJECT , TABLE , TD , TH , A , B , BIG , CODE , EM , FONT , I , NOBR , S , SMALL , STRIKE , STRONG , TT , U , LABEL , OUTPUT , RP , RT , RUBY , SPAN , SUB , SUP , VAR , XMP , MATH , MGLYPH , MALIGNMARK , MI , MO , MN , MS , MTEXT , ANNOTATION_XML , SVG , FOREIGNOBJECT , DESC , UNKNOWN }

Functions
hubbub_error	hubbub_treebuilder_token_handler (const hubbub_token token, void pw)
	Handle tokeniser emitting a token.

hubbub_error	process_characters_expect_whitespace (hubbub_treebuilder treebuilder, const hubbub_token token, bool insert_into_current_node)
	Process a character token in cases where we expect only whitespace.

hubbub_error	process_comment_append (hubbub_treebuilder treebuilder, const hubbub_token token, void *parent)
	Process a comment token, appending it to the given parent.

hubbub_error	parse_generic_rcdata (hubbub_treebuilder treebuilder, const hubbub_token token, bool rcdata)
	Trigger parsing of generic (R)CDATA.

uint32_t	element_in_scope (hubbub_treebuilder *treebuilder, element_type type, bool in_table)
	Determine if an element is in (table) scope.

hubbub_error	reconstruct_active_formatting_list (hubbub_treebuilder *treebuilder)
	Reconstruct the list of active formatting elements.

void	clear_active_formatting_list_to_marker (hubbub_treebuilder *treebuilder)
	Clear the list of active formatting elements up to the last marker.

hubbub_error	remove_node_from_dom (hubbub_treebuilder treebuilder, void node)
	Remove a node from the DOM.

hubbub_error	insert_element (hubbub_treebuilder treebuilder, const hubbub_tag tag_name, bool push)
	Create element and insert it into the DOM, potentially pushing it on the stack.

void	close_implied_end_tags (hubbub_treebuilder *treebuilder, element_type except)
	Close implied end tags.

void	reset_insertion_mode (hubbub_treebuilder *treebuilder)
	Reset the insertion mode.

hubbub_error	append_text (hubbub_treebuilder treebuilder, const hubbub_string string)
	Append text to the current node, inserting into the last child of the current node, iff it's a Text node.

hubbub_error	complete_script (hubbub_treebuilder *treebuilder)
	Script processing and execution.

element_type	element_type_from_name (hubbub_treebuilder treebuilder, const hubbub_string tag_name)
	Convert an element name into an element type.

bool	is_special_element (element_type type)
	Determine if a node is a special element.

bool	is_scoping_element (element_type type)
	Determine if a node is a scoping element.

bool	is_formatting_element (element_type type)
	Determine if a node is a formatting element.

bool	is_phrasing_element (element_type type)
	Determine if a node is a phrasing element.

hubbub_error	element_stack_push (hubbub_treebuilder treebuilder, hubbub_ns ns, element_type type, void node)
	Push an element onto the stack of open elements.

hubbub_error	element_stack_pop (hubbub_treebuilder treebuilder, hubbub_ns ns, element_type type, void *node)
	Pop an element off the stack of open elements.

hubbub_error	element_stack_pop_until (hubbub_treebuilder *treebuilder, element_type type)
	Pop elements until an element of type "element" has been popped.

hubbub_error	element_stack_remove (hubbub_treebuilder treebuilder, uint32_t index, hubbub_ns ns, element_type type, void *removed)
	Remove a node from the stack of open elements.

uint32_t	current_table (hubbub_treebuilder *treebuilder)
	Find the stack index of the current table.

element_type	current_node (hubbub_treebuilder *treebuilder)
	Peek at the top element of the element stack.

element_type	prev_node (hubbub_treebuilder *treebuilder)
	Peek at the element below the top of the element stack.

hubbub_error	formatting_list_append (hubbub_treebuilder treebuilder, hubbub_ns ns, element_type type, void node, uint32_t stack_index)
	Append an element to the end of the list of active formatting elements.

hubbub_error	formatting_list_insert (hubbub_treebuilder treebuilder, formatting_list_entry prev, formatting_list_entry next, hubbub_ns ns, element_type type, void node, uint32_t stack_index)
	Insert an element into the list of active formatting elements.

hubbub_error	formatting_list_remove (hubbub_treebuilder treebuilder, formatting_list_entry entry, hubbub_ns ns, element_type type, void *node, uint32_t stack_index)
	Remove an element from the list of active formatting elements.

hubbub_error	formatting_list_replace (hubbub_treebuilder treebuilder, formatting_list_entry entry, hubbub_ns ns, element_type type, void node, uint32_t stack_index, hubbub_ns ons, element_type otype, void onode, uint32_t ostack_index)
	Remove an element from the list of active formatting elements.

void	adjust_mathml_attributes (hubbub_treebuilder treebuilder, hubbub_tag tag)
	Adjust MathML attributes.

void	adjust_svg_attributes (hubbub_treebuilder treebuilder, hubbub_tag tag)
	Adjust SVG attributes.

void	adjust_svg_tagname (hubbub_treebuilder treebuilder, hubbub_tag tag)
	Adjust SVG tagnmes.

void	adjust_foreign_attributes (hubbub_treebuilder treebuilder, hubbub_tag tag)
	Adjust foreign attributes.

hubbub_error	aa_insert_into_foster_parent (hubbub_treebuilder treebuilder, void node, void **inserted)
	Adoption agency: locate foster parent and insert node into it.

void	element_stack_dump (hubbub_treebuilder treebuilder, FILE fp)
	Dump an element stack to the given file pointer.

void	formatting_list_dump (hubbub_treebuilder treebuilder, FILE fp)
	Dump a formatting list to the given file pointer.

const char *	element_type_to_name (element_type type)
	Convert an element type to a name.

Macro Definition Documentation

◆ ELEMENT_STACK_CHUNK

#define ELEMENT_STACK_CHUNK 128

Typedef Documentation

◆ element_context

typedef struct element_context element_context

Item on the element stack.

◆ formatting_list_entry

typedef struct formatting_list_entry formatting_list_entry

Entry in a formatting list.

◆ hubbub_treebuilder_context

typedef struct hubbub_treebuilder_context hubbub_treebuilder_context

Context for a tree builder.

Enumeration Type Documentation

◆ element_type

enum element_type

Enumerator
ADDRESS
AREA
ARTICLE
ASIDE
BASE
BASEFONT
BGSOUND
BLOCKQUOTE
BODY
BR
CENTER
COL
COLGROUP
COMMAND
DATAGRID
DD
DETAILS
DIALOG
DIR
DIV
DL
DT
EMBED
FIELDSET
FIGCAPTION
FIGURE
FOOTER
FORM
FRAME
FRAMESET
H1
H2
H3
H4
H5
H6
HEAD
HEADER
HR
IFRAME
IMAGE
IMG
INPUT
ISINDEX
LI
LINK
LISTING
MAIN
MENU
META
NAV
NOEMBED
NOFRAMES
NOSCRIPT
OL
OPTGROUP
OPTION
P
PARAM
PLAINTEXT
PRE
SCRIPT
SECTION
SELECT
SPACER
STYLE
SUMMARY
TBODY
TEXTAREA
TFOOT
THEAD
TITLE
TR
UL
WBR
APPLET
BUTTON
CAPTION
HTML
MARQUEE
OBJECT
TABLE
TD
TH
A
B
BIG
CODE
EM
FONT
I
NOBR
S
SMALL
STRIKE
STRONG
TT
U
LABEL	Todo: Enumerate phrasing elements
OUTPUT
RP
RT
RUBY
SPAN
SUB
SUP
VAR
XMP
MATH
MGLYPH
MALIGNMARK
MI
MO
MN
MS
MTEXT
ANNOTATION_XML
SVG
FOREIGNOBJECT
DESC
UNKNOWN

Function Documentation

◆ aa_insert_into_foster_parent()

hubbub_error aa_insert_into_foster_parent	(	hubbub_treebuilder *	treebuilder,
		void *	node,
		void **	inserted
	)

Adoption agency: locate foster parent and insert node into it.

Parameters

treebuilder	The treebuilder instance
node	The node to insert
inserted	Pointer to location to receive inserted node

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ adjust_foreign_attributes()

void adjust_foreign_attributes	(	hubbub_treebuilder *	treebuilder,
		hubbub_tag *	tag
	)

Adjust foreign attributes.

Parameters

treebuilder	Treebuilder instance
tag	Tag to adjust the attributes of

◆ adjust_mathml_attributes()

void adjust_mathml_attributes	(	hubbub_treebuilder *	treebuilder,
		hubbub_tag *	tag
	)

Adjust MathML attributes.

Parameters

treebuilder	Treebuilder instance
tag	Tag to adjust the attributes of

◆ adjust_svg_attributes()

void adjust_svg_attributes	(	hubbub_treebuilder *	treebuilder,
		hubbub_tag *	tag
	)

Adjust SVG attributes.

Parameters

treebuilder	Treebuilder instance
tag	Tag to adjust the attributes of

◆ adjust_svg_tagname()

void adjust_svg_tagname	(	hubbub_treebuilder *	treebuilder,
		hubbub_tag *	tag
	)

Adjust SVG tagnmes.

Parameters

treebuilder	Treebuilder instance
tag	Tag to adjust the name of

◆ append_text()

hubbub_error append_text	(	hubbub_treebuilder *	treebuilder,
		const hubbub_string *	string
	)

Append text to the current node, inserting into the last child of the current node, iff it's a Text node.

Parameters

treebuilder	The treebuilder instance
string	The string to append

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ clear_active_formatting_list_to_marker()

void clear_active_formatting_list_to_marker ( hubbub_treebuilder * treebuilder )

Clear the list of active formatting elements up to the last marker.

Parameters

treebuilder The treebuilder instance containing the list

◆ close_implied_end_tags()

void close_implied_end_tags	(	hubbub_treebuilder *	treebuilder,
		element_type	except
	)

Close implied end tags.

Parameters

treebuilder	The treebuilder instance
except	Tag type to exclude from processing [DD,DT,LI,OPTION, OPTGROUP,P,RP,RT], UNKNOWN to exclude nothing

◆ complete_script()

hubbub_error complete_script ( hubbub_treebuilder * treebuilder )

Script processing and execution.

Parameters

treebuilder The treebuilder instance

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ current_node()

element_type current_node ( hubbub_treebuilder * treebuilder )

Peek at the top element of the element stack.

Parameters

treebuilder Treebuilder instance

Returns: Element type on the top of the stack

◆ current_table()

uint32_t current_table ( hubbub_treebuilder * treebuilder )

Find the stack index of the current table.

◆ element_in_scope()

uint32_t element_in_scope	(	hubbub_treebuilder *	treebuilder,
		element_type	type,
		bool	in_table
	)

Determine if an element is in (table) scope.

Parameters

treebuilder	Treebuilder to look in
type	Element type to find
in_table	Whether we're looking in table scope

Returns: Element stack index, or 0 if not in scope

◆ element_stack_dump()

void element_stack_dump	(	hubbub_treebuilder *	treebuilder,
		FILE *	fp
	)

Dump an element stack to the given file pointer.

Parameters

treebuilder	The treebuilder instance
fp	The file to dump to

◆ element_stack_pop()

hubbub_error element_stack_pop	(	hubbub_treebuilder *	treebuilder,
		hubbub_ns *	ns,
		element_type *	type,
		void **	node
	)

Pop an element off the stack of open elements.

Parameters

treebuilder	The treebuilder instance containing the stack
ns	Pointer to location to receive element namespace
type	Pointer to location to receive element type
node	Pointer to location to receive node

Returns: HUBBUB_OK on success, appropriate error otherwise.

Todo:: Can we optimise this? (i.e. by not traversing the entire list)

Todo:: reduce allocated stack size once there's enough free

◆ element_stack_pop_until()

hubbub_error element_stack_pop_until	(	hubbub_treebuilder *	treebuilder,
		element_type	type
	)

Pop elements until an element of type "element" has been popped.

Returns: HUBBUB_OK on success, appropriate error otherwise.

◆ element_stack_push()

hubbub_error element_stack_push	(	hubbub_treebuilder *	treebuilder,
		hubbub_ns	ns,
		element_type	type,
		void *	node
	)

Push an element onto the stack of open elements.

Parameters

treebuilder	The treebuilder instance containing the stack
ns	The namespace of element being pushed
type	The type of element being pushed
node	The node to push

Returns: HUBBUB_OK on success, appropriate error otherwise.

◆ element_stack_remove()

hubbub_error element_stack_remove	(	hubbub_treebuilder *	treebuilder,
		uint32_t	index,
		hubbub_ns *	ns,
		element_type *	type,
		void **	removed
	)

Remove a node from the stack of open elements.

Parameters

treebuilder	The treebuilder instance
index	The index of the node to remove
ns	Pointer to location to receive namespace
type	Pointer to location to receive type
removed	Pointer to location to receive removed node

Returns: HUBBUB_OK on success, appropriate error otherwise.

◆ element_type_from_name()

element_type element_type_from_name	(	hubbub_treebuilder *	treebuilder,
		const hubbub_string *	tag_name
	)

Convert an element name into an element type.

Parameters

treebuilder	The treebuilder instance
tag_name	The tag name to consider

Returns: The corresponding element type

Todo:: optimise this

◆ element_type_to_name()

const char * element_type_to_name ( element_type type )

Convert an element type to a name.

Parameters

type	The element type

Returns: Pointer to name

◆ formatting_list_append()

hubbub_error formatting_list_append	(	hubbub_treebuilder *	treebuilder,
		hubbub_ns	ns,
		element_type	type,
		void *	node,
		uint32_t	stack_index
	)

Append an element to the end of the list of active formatting elements.

Parameters

treebuilder	Treebuilder instance containing list
ns	Namespace of node being inserted
type	Type of node being inserted
node	Node being inserted
stack_index	Index into stack of open elements

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ formatting_list_dump()

void formatting_list_dump	(	hubbub_treebuilder *	treebuilder,
		FILE *	fp
	)

Dump a formatting list to the given file pointer.

Parameters

treebuilder	The treebuilder instance
fp	The file to dump to

◆ formatting_list_insert()

hubbub_error formatting_list_insert	(	hubbub_treebuilder *	treebuilder,
		formatting_list_entry *	prev,
		formatting_list_entry *	next,
		hubbub_ns	ns,
		element_type	type,
		void *	node,
		uint32_t	stack_index
	)

Insert an element into the list of active formatting elements.

Parameters

treebuilder	Treebuilder instance containing list
prev	Previous entry
next	Next entry
ns	Namespace of node being inserted
type	Type of node being inserted
node	Node being inserted
stack_index	Index into stack of open elements

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ formatting_list_remove()

hubbub_error formatting_list_remove	(	hubbub_treebuilder *	treebuilder,
		formatting_list_entry *	entry,
		hubbub_ns *	ns,
		element_type *	type,
		void **	node,
		uint32_t *	stack_index
	)

Remove an element from the list of active formatting elements.

Parameters

treebuilder	Treebuilder instance containing list
entry	The item to remove
ns	Pointer to location to receive namespace of node
type	Pointer to location to receive type of node
node	Pointer to location to receive node
stack_index	Pointer to location to receive stack index

Returns: HUBBUB_OK on success, appropriate error otherwise.

◆ formatting_list_replace()

hubbub_error formatting_list_replace	(	hubbub_treebuilder *	treebuilder,
		formatting_list_entry *	entry,
		hubbub_ns	ns,
		element_type	type,
		void *	node,
		uint32_t	stack_index,
		hubbub_ns *	ons,
		element_type *	otype,
		void **	onode,
		uint32_t *	ostack_index
	)

Remove an element from the list of active formatting elements.

Parameters

treebuilder	Treebuilder instance containing list
entry	The item to replace
ns	Replacement node namespace
type	Replacement node type
node	Replacement node
stack_index	Replacement stack index
ons	Pointer to location to receive old namespace
otype	Pointer to location to receive old type
onode	Pointer to location to receive old node
ostack_index	Pointer to location to receive old stack index

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ hubbub_treebuilder_token_handler()

hubbub_error hubbub_treebuilder_token_handler	(	const hubbub_token *	token,
		void *	pw
	)

Handle tokeniser emitting a token.

Parameters

token	The emitted token
pw	Pointer to treebuilder instance

◆ insert_element()

hubbub_error insert_element	(	hubbub_treebuilder *	treebuilder,
		const hubbub_tag *	tag,
		bool	push
	)

Create element and insert it into the DOM, potentially pushing it on the stack.

Parameters

treebuilder	The treebuilder instance
tag	The element to insert
push	Whether to push the element onto the stack

Returns: HUBBUB_OK on success, appropriate error otherwise.

◆ is_formatting_element()

bool is_formatting_element ( element_type type )

Determine if a node is a formatting element.

Parameters

type	Node type to consider

Returns: True iff node is a formatting element

◆ is_phrasing_element()

bool is_phrasing_element ( element_type type )

Determine if a node is a phrasing element.

Parameters

type	Node type to consider

Returns: True iff node is a phrasing element

◆ is_scoping_element()

bool is_scoping_element ( element_type type )

Determine if a node is a scoping element.

Parameters

type	Node type to consider

Returns: True iff node is a scoping element

◆ is_special_element()

bool is_special_element ( element_type type )

Determine if a node is a special element.

Parameters

type	Node type to consider

Returns: True iff node is a special element

◆ parse_generic_rcdata()

hubbub_error parse_generic_rcdata	(	hubbub_treebuilder *	treebuilder,
		const hubbub_token *	token,
		bool	rcdata
	)

Trigger parsing of generic (R)CDATA.

Parameters

treebuilder	The treebuilder instance
token	The current token
rcdata	True for RCDATA, false for CDATA

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ prev_node()

element_type prev_node ( hubbub_treebuilder * treebuilder )

Peek at the element below the top of the element stack.

Parameters

treebuilder Treebuilder instance

Returns: Element type of the element one below the top of the stack

◆ process_characters_expect_whitespace()

hubbub_error process_characters_expect_whitespace	(	hubbub_treebuilder *	treebuilder,
		const hubbub_token *	token,
		bool	insert_into_current_node
	)

Process a character token in cases where we expect only whitespace.

Parameters

treebuilder	The treebuilder instance
token	The character token
insert_into_current_node	Whether to insert whitespace into current node

Returns: HUBBUB_REPROCESS if the token needs reprocessing (token data updated to skip any leading whitespace), HUBBUB_OK if it contained only whitespace, appropriate error otherwise

◆ process_comment_append()

hubbub_error process_comment_append	(	hubbub_treebuilder *	treebuilder,
		const hubbub_token *	token,
		void *	parent
	)

Process a comment token, appending it to the given parent.

Parameters

treebuilder	The treebuilder instance
token	The comment token
parent	The node to append to

Returns: HUBBUB_OK on success, appropriate error otherwise

◆ reconstruct_active_formatting_list()

hubbub_error reconstruct_active_formatting_list ( hubbub_treebuilder * treebuilder )

Reconstruct the list of active formatting elements.

Parameters

treebuilder Treebuilder instance containing list

Returns: HUBBUB_OK on success, appropriate error otherwise.

◆ remove_node_from_dom()

hubbub_error remove_node_from_dom	(	hubbub_treebuilder *	treebuilder,
		void *	node
	)

Remove a node from the DOM.

Parameters

treebuilder	Treebuilder instance
node	Node to remove

Returns: HUBBUB_OK on success, appropriate error otherwise.

◆ reset_insertion_mode()

void reset_insertion_mode ( hubbub_treebuilder * treebuilder )

Reset the insertion mode.

Parameters

treebuilder The treebuilder to reset

Todo:: fragment parsing algorithm

Classes

Macros

Typedefs

Enumerations

Functions

Macro Definition Documentation

◆ ELEMENT_STACK_CHUNK

Typedef Documentation

◆ element_context

◆ formatting_list_entry

◆ hubbub_treebuilder_context

Enumeration Type Documentation

◆ element_type

Function Documentation

◆ aa_insert_into_foster_parent()

◆ adjust_foreign_attributes()

◆ adjust_mathml_attributes()

◆ adjust_svg_attributes()

◆ adjust_svg_tagname()

◆ append_text()

◆ clear_active_formatting_list_to_marker()

◆ close_implied_end_tags()

◆ complete_script()

◆ current_node()

◆ current_table()

◆ element_in_scope()

◆ element_stack_dump()

◆ element_stack_pop()

◆ element_stack_pop_until()

◆ element_stack_push()

◆ element_stack_remove()

◆ element_type_from_name()

◆ element_type_to_name()

◆ formatting_list_append()

◆ formatting_list_dump()

◆ formatting_list_insert()

◆ formatting_list_remove()

◆ formatting_list_replace()

◆ hubbub_treebuilder_token_handler()

◆ insert_element()

◆ is_formatting_element()

◆ is_phrasing_element()

◆ is_scoping_element()

◆ is_special_element()

◆ parse_generic_rcdata()

◆ prev_node()

◆ process_characters_expect_whitespace()

◆ process_comment_append()

◆ reconstruct_active_formatting_list()

◆ remove_node_from_dom()

◆ reset_insertion_mode()