src/dom/select.h

   1 #ifndef EL_DOM_SELECT_H
   2 #define EL_DOM_SELECT_H
   3
   4 #include "dom/node.h"
   5
   6
   7 /* FIXME: Namespaces; *|E */
   8
   9 enum dom_select_element_match {
  10         /* Gives info about the relation required between two element nodes for
  11          * them to match. This is also referred to as combinators. */
  12         /* The following are mutually exclusive and at least one must be set.
  13          * DOM_SELECT_RELATION_DESCENDANT is the default. */
  14
  15         /* Matches any F descendant of E:               E   F */
  16         /* Bogus flag; it is an easy way to have a default. */
  17         DOM_SELECT_RELATION_DESCENDANT = 0,
  18         /* Matches F being a direct child of E:         E > F */
  19         DOM_SELECT_RELATION_DIRECT_CHILD = 1,
  20         /* Matches F immediate preceded by E:           E + F */
  21         DOM_SELECT_RELATION_DIRECT_ADJACENT = 2,
  22         /* Matches F preceded by E:                     E ~ F */
  23         DOM_SELECT_RELATION_INDIRECT_ADJACENT = 4,
  24
  25         DOM_SELECT_RELATION_FLAGS = DOM_SELECT_RELATION_DESCENDANT
  26                                   | DOM_SELECT_RELATION_DIRECT_CHILD
  27                                   | DOM_SELECT_RELATION_DIRECT_ADJACENT
  28                                   | DOM_SELECT_RELATION_INDIRECT_ADJACENT,
  29
  30         /* None of the following are mutual exclusive. They can co-exist
  31          * although combining them might not make a lot of sense. */
  32
  33         /* Matches any element:                         * */
  34         DOM_SELECT_ELEMENT_UNIVERSAL = 8,
  35         /* Matches the root node of the document:       :root or // */
  36         DOM_SELECT_ELEMENT_ROOT = 16,
  37         /* Matches the empty element (not even text):   :empty */
  38         DOM_SELECT_ELEMENT_EMPTY = 32,
  39
  40         /* Matches the some n-th child of its parent:   :nth-child(n), etc. */
  41         DOM_SELECT_ELEMENT_NTH_CHILD = 64,
  42
  43         /* Matches the some n-th sibling of its type:   :nth-of-type(n), etc. */
  44         DOM_SELECT_ELEMENT_NTH_TYPE = 128,
  45 };
  46
  47 /* The special CSS .bar class attribute syntax is represented as
  48  * E[class="bar"]. The ID flag will match against any attribute with it's
  49  * boolean id member set. XXX: These flags are ATM mutually exclusive. */
  50 enum dom_select_attribute_match {
  51         /* Matches any set value:                       E[foo] */
  52         DOM_SELECT_ATTRIBUTE_ANY = 1,
  53         /* Matches exact value "bar":                   E[foo="bar"] */
  54         DOM_SELECT_ATTRIBUTE_EXACT = 2,
  55         /* Matches space seprated list "z bar bee":     E[foo~="bar"] */
  56         DOM_SELECT_ATTRIBUTE_SPACE_LIST = 4,
  57         /* Matches hyphen separated list "z-bar-bee":   E[foo|="bar"] */
  58         DOM_SELECT_ATTRIBUTE_HYPHEN_LIST = 8,
  59         /* Matches value begining; "bar-z-bee":         E[foo^="bar"]*/
  60         DOM_SELECT_ATTRIBUTE_BEGIN = 16,
  61         /* Matches value ending; "z-bee-bar":           E[foo$="bar"] */
  62         DOM_SELECT_ATTRIBUTE_END = 32,
  63         /* Matches value containing; "m33p/bar\++":     E[foo*="bar"] */
  64         DOM_SELECT_ATTRIBUTE_CONTAINS = 64,
  65         /* Matches exact ID attribute value "bar":      #bar */
  66         DOM_SELECT_ATTRIBUTE_ID = 128,
  67 };
  68
  69 /* Info about text matching is stored in a DOM text node. */
  70 enum dom_select_text_match {
  71         /* Matches E containing substring "foo":        E:contains("foo") */
  72         DOM_SELECT_TEXT_CONTAINS = 1,
  73 };
  74
  75 /* Info about what nth child or type to match. The basic syntax is:
  76  *
  77  *      <step>n<index>
  78  *
  79  * with a little syntactic sugar.
  80  *
  81  * Examples:
  82  *
  83  *    0n+1 / 1          is first child (same as :first-child)
  84  *    2n+0 / 2n / even  is all even children
  85  *    2n+1 / odd        is all odd children
  86  *   -0n+2              is the last two children
  87  *   -0n+1 / -1         is last child (same as :last-child)
  88  *    1n+0 / n+0 / n    is all elements of type
  89  *    0n+0              is only element of type (a special internal syntax
  90  *                      used when storing nth-info)
  91  *
  92  * That is, a zero step (0n) means exact indexing, and non-zero step
  93  * means stepwise indexing.
  94  */
  95 struct dom_select_nth_match {
  96         size_t step;
  97         size_t index;
  98 };
  99
 100 #define set_dom_select_nth_match(nth, nthstep, nthindex) \
 101         do { (nth)->step = (nthstep); (nth)->index = (nthindex); } while(0)
 102
 103 /* This is supposed to be a simple selector. However, this struct is also used
 104  * for holding data for attribute matching and element text matching. */
 105 struct dom_select_node {
 106         /* This holds the DOM node which has data about the node being matched.
 107          * It can be either an element, attribute, or a text node. */
 108         /* XXX: Keep at the top. This is used for translating dom_node
 109          * reference to dom_select_node. */
 110         struct dom_node node;
 111
 112         /* Only meaningful for element nodes. */
 113         /* FIXME: Don't waste memory for non-element nodes. */
 114         struct dom_select_nth_match nth_child;
 115         struct dom_select_nth_match nth_type;
 116
 117         /* Flags, specifying how the matching should be done. */
 118         union {
 119                 enum dom_select_element_match element;
 120                 enum dom_select_attribute_match attribute;
 121                 enum dom_select_text_match text;
 122         } match;
 123 };
 124
 125
 126 enum dom_select_pseudo {
 127         DOM_SELECT_PSEUDO_UNKNOWN = 0,
 128
 129         /* Pseudo-elements: */
 130
 131         /* Matches first formatted line:                ::first-line */
 132         DOM_SELECT_PSEUDO_FIRST_LINE = 1,
 133         /* Matches first formatted letter:              ::first-letter */
 134         DOM_SELECT_PSEUDO_FIRST_LETTER = 2,
 135         /* Matches text selected by user:               ::selection */
 136         DOM_SELECT_PSEUDO_SELECTION = 4,
 137         /* Matches generated context after an element:  ::after */
 138         DOM_SELECT_PSEUDO_AFTER = 8,
 139         /* Matches generated content before an element: ::before */
 140         DOM_SELECT_PSEUDO_BEFORE = 16,
 141
 142         /* Pseudo-attributes: */
 143
 144         /* Link pseudo-classes: */
 145         DOM_SELECT_PSEUDO_LINK = 32,                    /* :link */
 146         DOM_SELECT_PSEUDO_VISITED = 64,                 /* :visited */
 147
 148         /* User action pseudo-classes: */
 149         DOM_SELECT_PSEUDO_ACTIVE = 128,                 /* :active */
 150         DOM_SELECT_PSEUDO_HOVER = 256,                  /* :hover */
 151         DOM_SELECT_PSEUDO_FOCUS = 512,                  /* :focus */
 152
 153         /* Target pseudo-class: */
 154         DOM_SELECT_PSEUDO_TARGET = 1024,                /* :target */
 155
 156         /* UI element states pseudo-classes: */
 157         DOM_SELECT_PSEUDO_ENABLED = 2048,               /* :enabled */
 158         DOM_SELECT_PSEUDO_DISABLED = 4096,              /* :disabled */
 159         DOM_SELECT_PSEUDO_CHECKED = 8192,               /* :checked */
 160         DOM_SELECT_PSEUDO_INDETERMINATE = 16384,        /* :indeterminate */
 161
 162         /* XXX: The following pseudo-classes are not kept in the pseudo member
 163          * of the dom_select struct so they should not be bitfields. They are
 164          * mostly for parsing purposes. */
 165
 166         DOM_SELECT_PSEUDO_CONTAINS = 10000,
 167
 168         DOM_SELECT_PSEUDO_NTH_CHILD,
 169         DOM_SELECT_PSEUDO_NTH_LAST_CHILD,
 170         DOM_SELECT_PSEUDO_FIRST_CHILD,
 171         DOM_SELECT_PSEUDO_LAST_CHILD,
 172         DOM_SELECT_PSEUDO_ONLY_CHILD,
 173
 174         DOM_SELECT_PSEUDO_NTH_TYPE,
 175         DOM_SELECT_PSEUDO_NTH_LAST_TYPE,
 176         DOM_SELECT_PSEUDO_FIRST_TYPE,
 177         DOM_SELECT_PSEUDO_LAST_TYPE,
 178         DOM_SELECT_PSEUDO_ONLY_TYPE,
 179
 180         DOM_SELECT_PSEUDO_ROOT,
 181         DOM_SELECT_PSEUDO_EMPTY,
 182 };
 183
 184 struct dom_select {
 185         struct dom_select_node *selector;
 186         unsigned long specificity;
 187         enum dom_select_pseudo pseudo;
 188 };
 189
 190 enum dom_select_syntax {
 191         DOM_SELECT_SYNTAX_CSS,  /* Example: 'p a[id=node] a:hover */
 192         DOM_SELECT_SYNTAX_PATH, /* Example: '//rss/channel/item' */
 193 };
 194
 195 struct dom_select *init_dom_select(enum dom_select_syntax syntax,
 196                                    struct dom_string *string);
 197
 198 void done_dom_select(struct dom_select *select);
 199
 200 struct dom_node_list *
 201 select_dom_nodes(struct dom_select *select, struct dom_node *root);
 202
 203 /*
 204  * +------------------------------------------------------------------------------------+
 205  * | Pattern               | Meaning                      | Type              | Version |
 206  * |-----------------------+------------------------------+-------------------+---------|
 207  * | *                     | any element                  | Universal         | 2       |
 208  * |                       |                              | selector          |         |
 209  * |-----------------------+------------------------------+-------------------+---------|
 210  * | E                     | an element of type E         | Type selector     | 1       |
 211  * |-----------------------+------------------------------+-------------------+---------|
 212  * | E F                   | an F element descendant of   | Descendant        | 1       |
 213  * |                       | an E element                 | combinator        |         |
 214  * |-----------------------+------------------------------+-------------------+---------|
 215  * | E > F                 | an F element child of an E   | Child combinator  | 2       |
 216  * |                       | element                      |                   |         |
 217  * |-----------------------+------------------------------+-------------------+---------|
 218  * | E + F                 | an F element immediately     | Direct adjacent   | 2       |
 219  * |                       | preceded by an E element     | combinator        |         |
 220  * |-----------------------+------------------------------+-------------------+---------|
 221  * | E ~ F                 | an F element preceded by an  | Indirect adjacent | 3       |
 222  * |                       | E element                    | combinator        |         |
 223  * |-----------------------+------------------------------+-------------------+---------|
 224  * | E:root                | an E element, root of the    | Structural        | 3       |
 225  * |                       | document                     | pseudo-classes    |         |
 226  * |-----------------------+------------------------------+-------------------+---------|
 227  * |                       | an E element that has no     | Structural        |         |
 228  * | E:empty               | children (including text     | pseudo-classes    | 3       |
 229  * |                       | nodes)                       |                   |         |
 230  * |-----------------------+------------------------------+-------------------+---------|
 231  * | E:first-child         | an E element, first child of | Structural        | 2       |
 232  * |                       | its parent                   | pseudo-classes    |         |
 233  * |-----------------------+------------------------------+-------------------+---------|
 234  * | E:last-child          | an E element, last child of  | Structural        | 3       |
 235  * |                       | its parent                   | pseudo-classes    |         |
 236  * |-----------------------+------------------------------+-------------------+---------|
 237  * | E:nth-child(n)        | an E element, the n-th child | Structural        | 3       |
 238  * |                       | of its parent                | pseudo-classes    |         |
 239  * |-----------------------+------------------------------+-------------------+---------|
 240  * |                       | an E element, the n-th child | Structural        |         |
 241  * | E:nth-last-child(n)   | of its parent, counting from | pseudo-classes    | 3       |
 242  * |                       | the last one                 |                   |         |
 243  * |-----------------------+------------------------------+-------------------+---------|
 244  * | E:first-of-type       | an E element, first sibling  | Structural        | 3       |
 245  * |                       | of its type                  | pseudo-classes    |         |
 246  * |-----------------------+------------------------------+-------------------+---------|
 247  * | E:last-of-type        | an E element, last sibling   | Structural        | 3       |
 248  * |                       | of its type                  | pseudo-classes    |         |
 249  * |-----------------------+------------------------------+-------------------+---------|
 250  * | E:nth-of-type(n)      | an E element, the n-th       | Structural        | 3       |
 251  * |                       | sibling of its type          | pseudo-classes    |         |
 252  * |-----------------------+------------------------------+-------------------+---------|
 253  * |                       | an E element, the n-th       | Structural        |         |
 254  * | E:nth-last-of-type(n) | sibling of its type,         | pseudo-classes    | 3       |
 255  * |                       | counting from the last one   |                   |         |
 256  * |-----------------------+------------------------------+-------------------+---------|
 257  * | E:only-child          | an E element, only child of  | Structural        | 3       |
 258  * |                       | its parent                   | pseudo-classes    |         |
 259  * |-----------------------+------------------------------+-------------------+---------|
 260  * | E:only-of-type        | an E element, only sibling   | Structural        | 3       |
 261  * |                       | of its type                  | pseudo-classes    |         |
 262  * |-----------------------+------------------------------+-------------------+---------|
 263  * |                       | an E element being the       |                   |         |
 264  * | E:link                | source anchor of a hyperlink | The link          |         |
 265  * | E:visited             | of which the target is not   | pseudo-classes    | 1       |
 266  * |                       | yet visited (:link) or       |                   |         |
 267  * |                       | already visited (:visited)   |                   |         |
 268  * |-----------------------+------------------------------+-------------------+---------|
 269  * | E:active              | an E element during certain  | The user action   |         |
 270  * | E:hover               | user actions                 | pseudo-classes    | 1 and 2 |
 271  * | E:focus               |                              |                   |         |
 272  * |-----------------------+------------------------------+-------------------+---------|
 273  * | E:target              | an E element being the       | The target        | 3       |
 274  * |                       | target of the referring URI  | pseudo-class      |         |
 275  * |-----------------------+------------------------------+-------------------+---------|
 276  * |                       | an element of type E in      |                   |         |
 277  * | E:lang(fr)            | language "fr" (the document  | The :lang()       | 2       |
 278  * | FIXME                 | language specifies how       | pseudo-class      |         |
 279  * |                       | language is determined)      |                   |         |
 280  * |-----------------------+------------------------------+-------------------+---------|
 281  * | E:enabled             | a user interface element E   | The UI element    |         |
 282  * | E:disabled            | which is enabled or disabled | states            | 3       |
 283  * |                       |                              | pseudo-classes    |         |
 284  * |-----------------------+------------------------------+-------------------+---------|
 285  * |                       | a user interface element E   |                   |         |
 286  * | E:checked             | which is checked or in an    | The UI element    |         |
 287  * | E:indeterminate       | indeterminate state (for     | states            | 3       |
 288  * |                       | instance a radio-button or   | pseudo-classes    |         |
 289  * |                       | checkbox)                    |                   |         |
 290  * |-----------------------+------------------------------+-------------------+---------|
 291  * |                       | an E element containing the  | Content           |         |
 292  * | E:contains("foo")     | substring "foo" in its       | pseudo-class      | 3       |
 293  * |                       | textual contents             |                   |         |
 294  * |-----------------------+------------------------------+-------------------+---------|
 295  * | E::first-line         | the first formatted line of  | The :first-line   | 1       |
 296  * |                       | an E element                 | pseudo-element    |         |
 297  * |-----------------------+------------------------------+-------------------+---------|
 298  * | E::first-letter       | the first formatted letter   | The :first-letter | 1       |
 299  * |                       | of an E element              | pseudo-element    |         |
 300  * |-----------------------+------------------------------+-------------------+---------|
 301  * |                       | the portion of an E element  | The UI element    |         |
 302  * | E::selection          | that is currently            | fragments         | 3       |
 303  * |                       | selected/highlighted by the  | pseudo-elements   |         |
 304  * |                       | user                         |                   |         |
 305  * |-----------------------+------------------------------+-------------------+---------|
 306  * | E::before             | generated content before an  | The :before       | 2       |
 307  * |                       | E element                    | pseudo-element    |         |
 308  * |-----------------------+------------------------------+-------------------+---------|
 309  * | E::after              | generated content after an E | The :after        | 2       |
 310  * |                       | element                      | pseudo-element    |         |
 311  * |-----------------------+------------------------------+-------------------+---------|
 312  * |                       | an E element whose class is  |                   |         |
 313  * | E.warning             | "warning" (the document      | Class selectors   | 1       |
 314  * |                       | language specifies how class |                   |         |
 315  * |                       | is determined).              |                   |         |
 316  * |-----------------------+------------------------------+-------------------+---------|
 317  * | E#myid                | an E element with ID equal   | ID selectors      | 1       |
 318  * |                       | to "myid".                   |                   |         |
 319  * |-----------------------+------------------------------+-------------------+---------|
 320  * | E[foo]                | an E element with a "foo"    | Attribute         | 2       |
 321  * |                       | attribute                    | selectors         |         |
 322  * |-----------------------+------------------------------+-------------------+---------|
 323  * |                       | an E element whose "foo"     | Attribute         |         |
 324  * | E[foo="bar"]          | attribute value is exactly   | selectors         | 2       |
 325  * |                       | equal to "bar"               |                   |         |
 326  * |-----------------------+------------------------------+-------------------+---------|
 327  * |                       | an E element whose "foo"     |                   |         |
 328  * |                       | attribute value is a list of | Attribute         |         |
 329  * | E[foo~="bar"]         | space-separated values, one  | selectors         | 2       |
 330  * |                       | of which is exactly equal to |                   |         |
 331  * |                       | "bar"                        |                   |         |
 332  * |-----------------------+------------------------------+-------------------+---------|
 333  * |                       | an E element whose "foo"     |                   |         |
 334  * | E[foo^="bar"]         | attribute value begins       | Attribute         | 3       |
 335  * |                       | exactly with the string      | selectors         |         |
 336  * |                       | "bar"                        |                   |         |
 337  * |-----------------------+------------------------------+-------------------+---------|
 338  * |                       | an E element whose "foo"     | Attribute         |         |
 339  * | E[foo$="bar"]         | attribute value ends exactly | selectors         | 3       |
 340  * |                       | with the string "bar"        |                   |         |
 341  * |-----------------------+------------------------------+-------------------+---------|
 342  * |                       | an E element whose "foo"     | Attribute         |         |
 343  * | E[foo*="bar"]         | attribute value contains the | selectors         | 3       |
 344  * |                       | substring "bar"              |                   |         |
 345  * |-----------------------+------------------------------+-------------------+---------|
 346  * |                       | an E element whose           |                   |         |
 347  * |                       | "hreflang" attribute has a   | Attribute         |         |
 348  * | E[hreflang|="en"]     | hyphen-separated list of     | selectors         | 2       |
 349  * |                       | values beginning (from the   |                   |         |
 350  * |                       | left) with "en"              |                   |         |
 351  * |-----------------------+------------------------------+-------------------+---------|
 352  * | E:not(s)              | an E element that does not   | Negation          | 3       |
 353  * | FIXME                 | match simple selector s      | pseudo-class      |         |
 354  * +------------------------------------------------------------------------------------+
 355  */
 356
 357 #endif