src/dom/sgml/sgml.h

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