src/dom/sgml/sgml.h

   1
   2 #ifndef EL_DOM_SGML_SGML_H
   3 #define EL_DOM_SGML_SGML_H
   4
   5 #include <stdlib.h>
   6
   7 #include "dom/node.h"
   8 #include "dom/string.h"
   9
  10 /* The flags stored in the attribute sgml node info data */
  11 /* TODO: Other potential flags (there can be only 16)
  12  *
  13  * - interaction info for forms (diabled, readonly, maxlength) maybe tabindex,
  14  * - table layout,
  15  * - generic layout style attributes maybe giving color values an additional flag,
  16  * - meta information (rel, rev, title, alt, summary, caption, standby, lang),
  17  * - scripting hooks (onblur, ...)
  18  * - information about the referenced content (hreflang, codetype, media, type)
  19  *
  20  * Anyway the flags should of course optimally have a purpose to speed things up
  21  * by quickly making it possible to identify certain attribute groups. --jonas */
  22 enum sgml_attribute_flags {
  23         /* The value uniquely identifies the owner */
  24         SGML_ATTRIBUTE_IDENTIFIER = 1,
  25         /* The value contains an URI of some sort */
  26         SGML_ATTRIBUTE_REFERENCE = 2,
  27 };
  28
  29 /* TODO: We also need an element flag to signal to the parser that all the
  30  * content should be skipped; possible with some ugly hacks to not use the
  31  * scanner since it could get confused. The purpose of this flag is to group
  32  * all element content in one text node. Kind of like a ``verbatim'' thing
  33  * where not parsing should be done. For HTML the <script> and <style> tags
  34  * should use it. */
  35 enum sgml_element_flags {
  36         /* The start and end tags are optional */
  37         SGML_ELEMENT_OPTIONAL = 1,
  38
  39         /* The element is empty and end tags are forbidden */
  40         SGML_ELEMENT_EMPTY = 2,
  41
  42         /* The end tag is obtional */
  43         SGML_ELEMENT_END_OPTIONAL = 4,
  44 };
  45
  46 struct sgml_node_info {
  47         struct dom_string string;
  48         uint16_t type;
  49         uint16_t flags;
  50 };
  51
  52 /* The header node is special. It is used for storing the number of nodes and
  53  * for returning the default 'unknown' node. */
  54 #define SGML_NODE_HEAD(doctype, nodetype) \
  55         { INIT_DOM_STRING(NULL, doctype##_##nodetype##S - 1), doctype##_##nodetype##_UNKNOWN }
  56
  57 #define SGML_NODE_INFO(doctype, nodetype, name, data) \
  58         { INIT_DOM_STRING(#name, sizeof(#name) - 1), doctype##_##nodetype##_##name, data }
  59
  60 #define SGML_NODE_INF2(doctype, nodetype, name, ident, data) \
  61         { INIT_DOM_STRING(ident, sizeof(ident) - 1), doctype##_##nodetype##_##name, data }
  62
  63 #define SGML_NODE_INFO_TYPE(doctype, nodetype, name) doctype##_##nodetype##_##name
  64
  65 int sgml_info_strcmp(const void *key, const void *node);
  66
  67 static inline struct sgml_node_info *
  68 get_sgml_node_info(struct sgml_node_info list[], struct dom_node *node)
  69 {
  70         struct sgml_node_info *map = &list[1];
  71         size_t map_size = list->string.length;
  72         size_t obj_size = sizeof(struct sgml_node_info);
  73         void *match = bsearch(node, map, map_size, obj_size, sgml_info_strcmp);
  74
  75         return match ? match : list;
  76 }
  77
  78 enum sgml_document_type {
  79         SGML_DOCTYPE_HTML,
  80         SGML_DOCTYPE_RSS,
  81         SGML_DOCTYPE_XBEL,
  82
  83         SGML_DOCTYPES,
  84 };
  85
  86 struct sgml_info {
  87         enum sgml_document_type doctype;
  88         struct sgml_node_info *attributes;
  89         struct sgml_node_info *elements;
  90 };
  91
  92 struct sgml_info *get_sgml_info(enum sgml_document_type doctype);
  93
  94 #endif