Go to most recent revision | Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 3584 | sourcerer | 1 | /* |
| 2 | * This file is part of Hubbub. |
||
| 3 | * Licensed under the MIT License, |
||
| 4 | * http://www.opensource.org/licenses/mit-license.php |
||
| 5 | * Copyright 2008 John-Mark Bell |
||
| 6 | */ |
||
| 7 | |||
| 8 | #ifndef hubbub_treebuilder_internal_h_ |
||
| 9 | #define hubbub_treebuilder_internal_h_ |
||
| 10 | |||
| 11 | #include "treebuilder/treebuilder.h" |
||
| 12 | |||
| 13 | typedef enum |
||
| 14 | { |
||
| 15 | /* Special */ |
||
| 16 | ADDRESS, AREA, ARTICLE, ASIDE, BASE, BASEFONT, BGSOUND, BLOCKQUOTE, |
||
| 17 | BODY, BR, CENTER, COL, COLGROUP, COMMAND, DATAGRID, DD, DETAILS, |
||
| 18 | DIALOG, DIR, DIV, DL, DT, EMBED, FIELDSET, FIGURE, FOOTER, FORM, FRAME, |
||
| 19 | FRAMESET, H1, H2, H3, H4, H5, H6, HEAD, HEADER, HR, IFRAME, IMAGE, IMG, |
||
| 20 | INPUT, ISINDEX, LI, LINK, LISTING, MENU, META, NAV, NOEMBED, NOFRAMES, |
||
| 21 | NOSCRIPT, OL, OPTGROUP, OPTION, P, PARAM, PLAINTEXT, PRE, SCRIPT, |
||
| 22 | SECTION, SELECT, SPACER, STYLE, TBODY, TEXTAREA, TFOOT, THEAD, TITLE, |
||
| 23 | TR, UL, WBR, |
||
| 24 | /* Scoping */ |
||
| 25 | APPLET, BUTTON, CAPTION, HTML, MARQUEE, OBJECT, TABLE, TD, TH, |
||
| 26 | /* Formatting */ |
||
| 27 | A, B, BIG, CODE, EM, FONT, I, NOBR, S, SMALL, STRIKE, STRONG, TT, U, |
||
| 28 | /* Phrasing */ |
||
| 29 | /**< \todo Enumerate phrasing elements */ |
||
| 30 | LABEL, OUTPUT, RP, RT, RUBY, SPAN, SUB, SUP, VAR, XMP, |
||
| 31 | /* MathML */ |
||
| 32 | MATH, MGLYPH, MALIGNMARK, MI, MO, MN, MS, MTEXT, ANNOTATION_XML, |
||
| 33 | /* SVG */ |
||
| 34 | SVG, FOREIGNOBJECT, /* foreignobject is scoping, but only in SVG ns */ |
||
| 35 | DESC, |
||
| 36 | UNKNOWN |
||
| 37 | } element_type; |
||
| 38 | |||
| 39 | /** |
||
| 40 | * Item on the element stack |
||
| 41 | */ |
||
| 42 | typedef struct element_context |
||
| 43 | { |
||
| 44 | hubbub_ns ns; /**< Element namespace */ |
||
| 45 | element_type type; /**< Element type */ |
||
| 46 | uint8_t *name; /**< Element name (interned) */ |
||
| 47 | |||
| 48 | bool tainted; /**< Only for tables. "Once the |
||
| 49 | * current table has been tainted, |
||
| 50 | * whitespace characters are inserted |
||
| 51 | * into the foster parent element |
||
| 52 | * instead of the current node." */ |
||
| 53 | |||
| 54 | void *node; /**< Node pointer */ |
||
| 55 | } element_context; |
||
| 56 | |||
| 57 | /** |
||
| 58 | * Entry in a formatting list |
||
| 59 | */ |
||
| 60 | typedef struct formatting_list_entry |
||
| 61 | { |
||
| 62 | element_context details; /**< Entry details */ |
||
| 63 | |||
| 64 | uint32_t stack_index; /**< Index into element stack */ |
||
| 65 | |||
| 66 | struct formatting_list_entry *prev; /**< Previous in list */ |
||
| 67 | struct formatting_list_entry *next; /**< Next in list */ |
||
| 68 | } formatting_list_entry; |
||
| 69 | |||
| 70 | /** |
||
| 71 | * Context for a tree builder |
||
| 72 | */ |
||
| 73 | typedef struct hubbub_treebuilder_context |
||
| 74 | { |
||
| 75 | insertion_mode mode; /**< The current insertion mode */ |
||
| 76 | insertion_mode second_mode; /**< The secondary insertion mode */ |
||
| 77 | |||
| 78 | #define ELEMENT_STACK_CHUNK 128 |
||
| 79 | element_context *element_stack; /**< Stack of open elements */ |
||
| 80 | uint32_t stack_alloc; /**< Number of stack slots allocated */ |
||
| 81 | uint32_t current_node; /**< Index of current node in stack */ |
||
| 82 | |||
| 83 | formatting_list_entry *formatting_list; /**< List of active formatting |
||
| 84 | * elements */ |
||
| 85 | formatting_list_entry *formatting_list_end; /**< End of active |
||
| 86 | * formatting list */ |
||
| 87 | |||
| 88 | void *head_element; /**< Pointer to HEAD element */ |
||
| 89 | |||
| 90 | void *form_element; /**< Pointer to most recently |
||
| 91 | * opened FORM element */ |
||
| 92 | |||
| 93 | void *document; /**< Pointer to the document node */ |
||
| 94 | |||
| 95 | bool enable_scripting; /**< Whether scripting is enabled */ |
||
| 96 | |||
| 97 | struct { |
||
| 98 | insertion_mode mode; /**< Insertion mode to return to */ |
||
| 99 | element_type type; /**< Type of node */ |
||
| 100 | } collect; /**< Context for character collecting */ |
||
| 101 | |||
| 102 | bool strip_leading_lr; /**< Whether to strip a LR from the |
||
| 103 | * start of the next character sequence |
||
| 104 | * received */ |
||
| 105 | |||
| 106 | bool in_table_foster; /**< Whether nodes that would be |
||
| 107 | * inserted into the current node should |
||
| 108 | * be foster parented */ |
||
| 109 | |||
| 110 | bool frameset_ok; /**< Whether to process a frameset */ |
||
| 111 | } hubbub_treebuilder_context; |
||
| 112 | |||
| 113 | /** |
||
| 114 | * Treebuilder object |
||
| 115 | */ |
||
| 116 | struct hubbub_treebuilder |
||
| 117 | { |
||
| 118 | hubbub_tokeniser *tokeniser; /**< Underlying tokeniser */ |
||
| 119 | |||
| 120 | hubbub_treebuilder_context context; /**< Our context */ |
||
| 121 | |||
| 122 | hubbub_tree_handler *tree_handler; /**< Callback table */ |
||
| 123 | |||
| 124 | hubbub_error_handler error_handler; /**< Error handler */ |
||
| 125 | void *error_pw; /**< Error handler data */ |
||
| 126 | |||
| 127 | hubbub_allocator_fn alloc; /**< Memory (de)allocation function */ |
||
| 128 | void *alloc_pw; /**< Client private data */ |
||
| 129 | }; |
||
| 130 | |||
| 131 | hubbub_error hubbub_treebuilder_token_handler( |
||
| 132 | const hubbub_token *token, void *pw); |
||
| 133 | |||
| 134 | hubbub_error process_characters_expect_whitespace( |
||
| 135 | hubbub_treebuilder *treebuilder, const hubbub_token *token, |
||
| 136 | bool insert_into_current_node); |
||
| 137 | hubbub_error process_comment_append(hubbub_treebuilder *treebuilder, |
||
| 138 | const hubbub_token *token, void *parent); |
||
| 139 | hubbub_error parse_generic_rcdata(hubbub_treebuilder *treebuilder, |
||
| 140 | const hubbub_token *token, bool rcdata); |
||
| 141 | |||
| 142 | uint32_t element_in_scope(hubbub_treebuilder *treebuilder, |
||
| 143 | element_type type, bool in_table); |
||
| 144 | hubbub_error reconstruct_active_formatting_list( |
||
| 145 | hubbub_treebuilder *treebuilder); |
||
| 146 | void clear_active_formatting_list_to_marker( |
||
| 147 | hubbub_treebuilder *treebuilder); |
||
| 148 | hubbub_error remove_node_from_dom(hubbub_treebuilder *treebuilder, |
||
| 149 | void *node); |
||
| 150 | hubbub_error insert_element(hubbub_treebuilder *treebuilder, |
||
| 151 | const hubbub_tag *tag_name, bool push); |
||
| 152 | void close_implied_end_tags(hubbub_treebuilder *treebuilder, |
||
| 153 | element_type except); |
||
| 154 | void reset_insertion_mode(hubbub_treebuilder *treebuilder); |
||
| 155 | hubbub_error append_text(hubbub_treebuilder *treebuilder, |
||
| 156 | const hubbub_string *string); |
||
| 157 | hubbub_error complete_script(hubbub_treebuilder *treebuilder); |
||
| 158 | |||
| 159 | element_type element_type_from_name(hubbub_treebuilder *treebuilder, |
||
| 160 | const hubbub_string *tag_name); |
||
| 161 | |||
| 162 | bool is_special_element(element_type type); |
||
| 163 | bool is_scoping_element(element_type type); |
||
| 164 | bool is_formatting_element(element_type type); |
||
| 165 | bool is_phrasing_element(element_type type); |
||
| 166 | |||
| 167 | hubbub_error element_stack_push(hubbub_treebuilder *treebuilder, |
||
| 168 | hubbub_ns ns, element_type type, void *node); |
||
| 169 | hubbub_error element_stack_pop(hubbub_treebuilder *treebuilder, |
||
| 170 | hubbub_ns *ns, element_type *type, void **node); |
||
| 171 | hubbub_error element_stack_pop_until(hubbub_treebuilder *treebuilder, |
||
| 172 | element_type type); |
||
| 173 | hubbub_error element_stack_remove(hubbub_treebuilder *treebuilder, |
||
| 174 | uint32_t index, hubbub_ns *ns, element_type *type, |
||
| 175 | void **removed); |
||
| 176 | uint32_t current_table(hubbub_treebuilder *treebuilder); |
||
| 177 | element_type current_node(hubbub_treebuilder *treebuilder); |
||
| 178 | element_type prev_node(hubbub_treebuilder *treebuilder); |
||
| 179 | |||
| 180 | hubbub_error formatting_list_append(hubbub_treebuilder *treebuilder, |
||
| 181 | hubbub_ns ns, element_type type, void *node, |
||
| 182 | uint32_t stack_index); |
||
| 183 | hubbub_error formatting_list_insert(hubbub_treebuilder *treebuilder, |
||
| 184 | formatting_list_entry *prev, formatting_list_entry *next, |
||
| 185 | hubbub_ns ns, element_type type, void *node, |
||
| 186 | uint32_t stack_index); |
||
| 187 | hubbub_error formatting_list_remove(hubbub_treebuilder *treebuilder, |
||
| 188 | formatting_list_entry *entry, |
||
| 189 | hubbub_ns *ns, element_type *type, void **node, |
||
| 190 | uint32_t *stack_index); |
||
| 191 | hubbub_error formatting_list_replace(hubbub_treebuilder *treebuilder, |
||
| 192 | formatting_list_entry *entry, |
||
| 193 | hubbub_ns ns, element_type type, void *node, |
||
| 194 | uint32_t stack_index, |
||
| 195 | hubbub_ns *ons, element_type *otype, void **onode, |
||
| 196 | uint32_t *ostack_index); |
||
| 197 | |||
| 198 | /* in_foreign_content.c */ |
||
| 199 | void adjust_mathml_attributes(hubbub_treebuilder *treebuilder, hubbub_tag *tag); |
||
| 200 | void adjust_svg_attributes(hubbub_treebuilder *treebuilder, |
||
| 201 | hubbub_tag *tag); |
||
| 202 | void adjust_svg_tagname(hubbub_treebuilder *treebuilder, |
||
| 203 | hubbub_tag *tag); |
||
| 204 | void adjust_foreign_attributes(hubbub_treebuilder *treebuilder, |
||
| 205 | hubbub_tag *tag); |
||
| 206 | |||
| 207 | /* in_body.c */ |
||
| 208 | hubbub_error aa_insert_into_foster_parent(hubbub_treebuilder *treebuilder, |
||
| 209 | void *node, void **inserted); |
||
| 210 | |||
| 211 | #ifndef NDEBUG |
||
| 212 | #include |
||
| 213 | |||
| 214 | void element_stack_dump(hubbub_treebuilder *treebuilder, FILE *fp); |
||
| 215 | void formatting_list_dump(hubbub_treebuilder *treebuilder, FILE *fp); |
||
| 216 | |||
| 217 | const char *element_type_to_name(element_type type); |
||
| 218 | |||
| 219 | #endif |
||
| 220 | |||
| 221 | #endif |
||
| 222 |