src/document/dom/node.h

   1
   2 #ifndef EL__DOCUMENT_DOM_NODE_H
   3 #define EL__DOCUMENT_DOM_NODE_H
   4
   5 #include "document/dom/string.h"
   6 #include "util/hash.h"
   7
   8 struct dom_node_list;
   9
  10 enum dom_node_type {
  11         DOM_NODE_UNKNOWN                =  0, /* for internal purpose only */
  12
  13         DOM_NODE_ELEMENT                =  1,
  14         DOM_NODE_ATTRIBUTE              =  2,
  15         DOM_NODE_TEXT                   =  3,
  16         DOM_NODE_CDATA_SECTION          =  4,
  17         DOM_NODE_ENTITY_REFERENCE       =  5,
  18         DOM_NODE_ENTITY                 =  6,
  19         DOM_NODE_PROCESSING_INSTRUCTION =  7,
  20         DOM_NODE_COMMENT                =  8,
  21         DOM_NODE_DOCUMENT               =  9,
  22         DOM_NODE_DOCUMENT_TYPE          = 10,
  23         DOM_NODE_DOCUMENT_FRAGMENT      = 11,
  24         DOM_NODE_NOTATION               = 12,
  25
  26         DOM_NODES
  27 };
  28
  29 /* Following is the node specific datastructures. They may contain no more
  30  * than 3 pointers or something equivalent. */
  31
  32 struct dom_node_id_item {
  33         /* The attibute node containing the id value */
  34         struct dom_node *id_attribute;
  35
  36         /* The node with the @id attribute */
  37         struct dom_node *node;
  38 };
  39
  40 struct dom_document_node {
  41         /* The document URI is stored in the string / length members. */
  42         /* An id to node hash for fast lookup. */
  43         struct hash *element_ids; /* -> {struct dom_node_id_item} */
  44
  45         /* Any meta data the root node carries such as document type nodes,
  46          * entity and notation map nodes and maybe some internal CSS stylesheet
  47          * node. */
  48         struct dom_node_list *meta_nodes;
  49
  50         /* The child nodes. May be NULL. Ordered like they where inserted. */
  51         struct dom_node_list *children;
  52 };
  53
  54 struct dom_id {
  55         struct dom_string public_id;
  56         struct dom_string system_id;
  57 };
  58
  59 struct dom_doctype_subset_info {
  60         struct dom_string internal;
  61         struct dom_id external;
  62 };
  63
  64 struct dom_document_type_node {
  65         /* These are really maps and should be sorted alphabetically. */
  66         struct dom_node_list *entities;
  67         struct dom_node_list *notations;
  68
  69         /* The string/length members of dom_node hold the name of the document
  70          * type "<!DOCTYPE {name} ...>". This holds the ids for the external
  71          * subset and the string of the internal subset. */
  72         struct dom_doctype_subset_infot *subset;
  73 };
  74
  75 /* Element nodes are indexed nodes stored in node lists of either
  76  * other child nodes or the root node. */
  77 struct dom_element_node {
  78         /* The child nodes. May be NULL. Ordered like they where inserted. */
  79         struct dom_node_list *children;
  80
  81         /* Only element nodes can have attributes and element nodes can only be
  82          * child nodes so the map is put here.
  83          *
  84          * The @map may be NULL if there are none. The @map nodes are sorted
  85          * alphabetically according to the attributes name so it has fast
  86          * lookup. */
  87         struct dom_node_list *map;
  88
  89         /* For <xsl:stylesheet ...> elements this holds the offset of
  90          * 'stylesheet' */
  91         uint16_t namespace_offset;
  92
  93         /* Special implementation dependent type specifier for example
  94          * containing an enum value representing the element to reduce string
  95          * comparing and only do one fast find mapping. */
  96         uint16_t type;
  97 };
  98
  99 /* Attribute nodes are named nodes stored in a node map of an element node. */
 100 struct dom_attribute_node {
 101         /* The string that hold the attribute value. The @string / @length
 102          * members of {struct dom_node} holds the name that identifies the node
 103          * in the map. */
 104         struct dom_string value;
 105
 106         /* For xml:lang="en" attributes this holds the offset of 'lang' */
 107         uint16_t namespace_offset;
 108
 109         /* Special implementation dependent type specifier. For HTML it (will)
 110          * contain an enum value representing the attribute HTML_CLASS, HTML_ID etc.
 111          * to reduce string comparing and only do one fast find mapping. */
 112         uint16_t type;
 113
 114         /* Was the attribute specified in the DTD as a default attribute or was
 115          * it added from the document source. */
 116         unsigned int specified:1;
 117
 118         /* Was the node->string allocated */
 119         unsigned int allocated:1;
 120
 121         /* Has the node->string been converted to internal charset. */
 122         unsigned int converted:1;
 123
 124         /* Is the attribute a unique identifier meaning the owner (element)
 125          * should be added to the document nodes @element_id hash. */
 126         unsigned int id:1;
 127
 128         /* The attribute value references some other resource */
 129         unsigned int reference:1;
 130
 131         /* The attribute value is delimited by quotes */
 132         unsigned int quoted:1;
 133 };
 134
 135 struct dom_text_node {
 136         /* The number of newlines the text string contains */
 137         unsigned int newlines;
 138
 139         /* We will need to add text nodes even if they contain only whitespace.
 140          * In order to quickly identify such nodes this member is used. */
 141         unsigned int only_space:1;
 142
 143         /* Was the node->string allocated */
 144         unsigned int allocated:1;
 145
 146         /* Has the node->string been converted to internal charset. */
 147         unsigned int converted:1;
 148 };
 149
 150 enum dom_proc_instruction_type {
 151         DOM_PROC_INSTRUCTION,
 152
 153         /* Keep this group sorted */
 154         DOM_PROC_INSTRUCTION_DBHTML,    /* DocBook toolchain instruction */
 155         DOM_PROC_INSTRUCTION_ELINKS,    /* Internal instruction hook */
 156         DOM_PROC_INSTRUCTION_XML,       /* XML instructions */
 157
 158         DOM_PROC_INSTRUCTION_TYPES
 159 };
 160
 161 struct dom_proc_instruction_node {
 162         /* The target of the processing instruction (xml for '<?xml ...  ?>')
 163          * is in the @string / @length members. */
 164         /* This holds the value to be processed */
 165         struct dom_string instruction;
 166
 167         /* For fast checking of the target type */
 168         uint16_t type; /* enum dom_proc_instruction_type */
 169
 170         /* For some processing instructions like xml the instructions contain
 171          * attributes and those attribute can be collected in this @map. */
 172         struct dom_node_list *map;
 173 };
 174
 175 union dom_node_data {
 176         struct dom_document_node         document;
 177         struct dom_document_type_node    document_type;
 178         struct dom_element_node          element;
 179         struct dom_attribute_node        attribute;
 180         struct dom_text_node             text;
 181         struct dom_id                    notation;
 182         /* For entities string/length hold the notation name */
 183         struct dom_id                    entity;
 184         struct dom_proc_instruction_node proc_instruction;
 185
 186         /* Node types without a union member yet
 187          *
 188          * DOM_NODE_CDATA_SECTION,
 189          * DOM_NODE_COMMENT,
 190          * DOM_NODE_DOCUMENT_FRAGMENT,
 191          * DOM_NODE_ENTITY_REFERENCE,
 192          */
 193 };
 194
 195 /* This structure is size critical so keep ordering to make it easier to pack
 196  * and avoid unneeded members. */
 197 struct dom_node {
 198         /* The type of the node */
 199         uint16_t type; /* -> enum dom_node_type */
 200
 201         /* Can contain either stuff like element name or for attributes the
 202          * attribute name. */
 203         struct dom_string string;
 204
 205         struct dom_node *parent;
 206
 207         /* Various info depending on the type of the node. */
 208         union dom_node_data data;
 209 };
 210
 211 /* A node list can be used for storing indexed nodes */
 212 struct dom_node_list {
 213         size_t size;
 214         struct dom_node *entries[1];
 215 };
 216
 217 #define foreach_dom_node(i, node, list)                 \
 218         for ((i) = 0; (i) < (list)->size; (i)++)        \
 219                 if (((node) = (list)->entries[(i)]))
 220
 221 #define foreachback_dom_node(i, node, list)             \
 222         for ((i) = (list)->size - 1; (i) > 0; (i)--)    \
 223                 if (((node) = (list)->entries[(i)]))
 224
 225 #define is_dom_node_list_member(list, member)           \
 226         ((list) && 0 <= (member) && (member) < (list)->size)
 227
 228 /* Adds @node to the list pointed to by @list_ptr at the given @position. If
 229  * @position is -1 the node is added at the end. */
 230 struct dom_node_list *
 231 add_to_dom_node_list(struct dom_node_list **list_ptr,
 232                      struct dom_node *node, int position);
 233
 234 void done_dom_node_list(struct dom_node_list *list);
 235
 236 /* Returns the position or index where the @node should be inserted into the
 237  * node @list in order to the list to be alphabetically sorted.  Assumes that
 238  * @list is already sorted properly. */
 239 int get_dom_node_map_index(struct dom_node_list *list, struct dom_node *node);
 240
 241 /* Looks up the @node_map for a node matching the requested type and name.
 242  * The @subtype maybe be 0 indication unknown subtype and only name should be
 243  * tested else it will indicate either the element or attribute private
 244  * subtype. */
 245 struct dom_node *
 246 get_dom_node_map_entry(struct dom_node_list *node_map,
 247                        enum dom_node_type type, uint16_t subtype,
 248                        struct dom_string *name);
 249
 250 struct dom_node *
 251 init_dom_node_(unsigned char *file, int line,
 252                 struct dom_node *parent, enum dom_node_type type,
 253                 unsigned char *string, size_t length);
 254 #define init_dom_node(type, string, length) init_dom_node_(__FILE__, __LINE__, NULL, type, string, length)
 255 #define add_dom_node(parent, type, string, length) init_dom_node_(__FILE__, __LINE__, parent, type, string, length)
 256
 257 #define add_dom_element(parent, string, length) \
 258         add_dom_node(parent, DOM_NODE_ELEMENT, string, length)
 259
 260 static inline struct dom_node *
 261 add_dom_attribute(struct dom_node *parent, unsigned char *string, int length,
 262                   unsigned char *value, size_t valuelen)
 263 {
 264         struct dom_node *node = add_dom_node(parent, DOM_NODE_ATTRIBUTE, string, length);
 265
 266         if (node && value) {
 267                 set_dom_string(&node->data.attribute.value, value, valuelen);
 268         }
 269
 270         return node;
 271 }
 272
 273 static inline struct dom_node *
 274 add_dom_proc_instruction(struct dom_node *parent, unsigned char *string, int length,
 275                          unsigned char *instruction, size_t instructionlen)
 276 {
 277         struct dom_node *node = add_dom_node(parent, DOM_NODE_PROCESSING_INSTRUCTION, string, length);
 278
 279         if (node && instruction) {
 280                 set_dom_string(&node->data.proc_instruction.instruction, instruction, instructionlen);
 281         }
 282
 283         return node;
 284 }
 285
 286 /* Removes the node and all its children and free()s itself */
 287 void done_dom_node(struct dom_node *node);
 288
 289 /* Returns the name of the node in an allocated string. */
 290 struct dom_string *get_dom_node_name(struct dom_node *node);
 291
 292 /* Returns the value of the node or NULL if no value is defined for the node
 293  * type. */
 294 struct dom_string *get_dom_node_value(struct dom_node *node);
 295
 296 /* Returns the name used for identifying the node type. */
 297 struct dom_string *get_dom_node_type_name(enum dom_node_type type);
 298
 299 /* Returns a pointer to a node list containing attributes. */
 300 #define get_dom_node_attributes(node)                                   \
 301         ((node)->type == DOM_NODE_ELEMENT ? &(node)->data.element.map   \
 302                                           : NULL)
 303
 304 static inline struct dom_node_list **
 305 get_dom_node_list(struct dom_node *parent, struct dom_node *node)
 306 {
 307         switch (parent->type) {
 308         case DOM_NODE_DOCUMENT:
 309                 return &parent->data.document.children;
 310
 311         case DOM_NODE_ELEMENT:
 312                 switch (node->type) {
 313                 case DOM_NODE_ATTRIBUTE:
 314                         return &parent->data.element.map;
 315
 316                 default:
 317                         return &parent->data.element.children;
 318                 }
 319
 320         case DOM_NODE_DOCUMENT_TYPE:
 321                 switch (node->type) {
 322                 case DOM_NODE_ENTITY:
 323                         return &parent->data.document_type.entities;
 324
 325                 case DOM_NODE_NOTATION:
 326                         return &parent->data.document_type.notations;
 327
 328                 default:
 329                         return NULL;
 330                 }
 331
 332         case DOM_NODE_PROCESSING_INSTRUCTION:
 333                 switch (node->type) {
 334                 case DOM_NODE_ATTRIBUTE:
 335                         return &parent->data.proc_instruction.map;
 336
 337                 default:
 338                         return NULL;
 339                 }
 340
 341         default:
 342                 return NULL;
 343         }
 344 }
 345
 346 #endif