html.h 7.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289
  1. #ifndef MUPDF_HTML_H
  2. #define MUPDF_HTML_H
  3. #include "mupdf/fitz.h"
  4. typedef struct fz_html_font_face_s fz_html_font_face;
  5. typedef struct fz_html_font_set_s fz_html_font_set;
  6. typedef struct fz_html_s fz_html;
  7. typedef struct fz_html_box_s fz_html_box;
  8. typedef struct fz_html_flow_s fz_html_flow;
  9. typedef struct fz_css_s fz_css;
  10. typedef struct fz_css_rule_s fz_css_rule;
  11. typedef struct fz_css_match_prop_s fz_css_match_prop;
  12. typedef struct fz_css_match_s fz_css_match;
  13. typedef struct fz_css_style_s fz_css_style;
  14. typedef struct fz_css_selector_s fz_css_selector;
  15. typedef struct fz_css_condition_s fz_css_condition;
  16. typedef struct fz_css_property_s fz_css_property;
  17. typedef struct fz_css_value_s fz_css_value;
  18. typedef struct fz_css_number_s fz_css_number;
  19. typedef struct fz_css_color_s fz_css_color;
  20. struct fz_html_font_face_s
  21. {
  22. char *family;
  23. int is_bold;
  24. int is_italic;
  25. fz_font *font;
  26. char *src;
  27. fz_html_font_face *next;
  28. };
  29. struct fz_html_font_set_s
  30. {
  31. fz_font *fonts[12]; /* Times, Helvetica, Courier in R,I,B,BI */
  32. fz_html_font_face *custom;
  33. };
  34. enum
  35. {
  36. CSS_KEYWORD = 256,
  37. CSS_HASH,
  38. CSS_STRING,
  39. CSS_NUMBER,
  40. CSS_LENGTH,
  41. CSS_PERCENT,
  42. CSS_URI,
  43. };
  44. struct fz_css_s
  45. {
  46. fz_pool *pool;
  47. fz_css_rule *rule;
  48. };
  49. struct fz_css_rule_s
  50. {
  51. fz_css_selector *selector;
  52. fz_css_property *declaration;
  53. fz_css_rule *next;
  54. };
  55. struct fz_css_selector_s
  56. {
  57. char *name;
  58. int combine;
  59. fz_css_condition *cond;
  60. fz_css_selector *left;
  61. fz_css_selector *right;
  62. fz_css_selector *next;
  63. };
  64. struct fz_css_condition_s
  65. {
  66. int type;
  67. char *key;
  68. char *val;
  69. fz_css_condition *next;
  70. };
  71. struct fz_css_property_s
  72. {
  73. char *name;
  74. fz_css_value *value;
  75. short spec;
  76. short important;
  77. fz_css_property *next;
  78. };
  79. struct fz_css_value_s
  80. {
  81. int type;
  82. char *data;
  83. fz_css_value *args; /* function arguments */
  84. fz_css_value *next;
  85. };
  86. struct fz_css_match_prop_s
  87. {
  88. const char *name; /* not owned */
  89. fz_css_value *value; /* not owned */
  90. int spec;
  91. };
  92. struct fz_css_match_s
  93. {
  94. fz_css_match *up;
  95. int count;
  96. fz_css_match_prop prop[64];
  97. };
  98. enum { DIS_NONE, DIS_BLOCK, DIS_INLINE, DIS_LIST_ITEM, DIS_INLINE_BLOCK };
  99. enum { POS_STATIC, POS_RELATIVE, POS_ABSOLUTE, POS_FIXED };
  100. enum { TA_LEFT, TA_RIGHT, TA_CENTER, TA_JUSTIFY };
  101. enum { VA_BASELINE, VA_SUB, VA_SUPER, VA_TOP, VA_BOTTOM, VA_TEXT_TOP, VA_TEXT_BOTTOM };
  102. enum { BS_NONE, BS_SOLID };
  103. enum { V_VISIBLE, V_HIDDEN, V_COLLAPSE };
  104. enum { PB_AUTO, PB_ALWAYS, PB_AVOID, PB_LEFT, PB_RIGHT };
  105. enum {
  106. WS_COLLAPSE = 1,
  107. WS_ALLOW_BREAK_SPACE = 2,
  108. WS_FORCE_BREAK_NEWLINE = 4,
  109. WS_NORMAL = WS_COLLAPSE | WS_ALLOW_BREAK_SPACE,
  110. WS_PRE = WS_FORCE_BREAK_NEWLINE,
  111. WS_NOWRAP = WS_COLLAPSE,
  112. WS_PRE_WRAP = WS_ALLOW_BREAK_SPACE | WS_FORCE_BREAK_NEWLINE,
  113. WS_PRE_LINE = WS_COLLAPSE | WS_ALLOW_BREAK_SPACE | WS_FORCE_BREAK_NEWLINE
  114. };
  115. enum {
  116. LST_NONE,
  117. LST_DISC, LST_CIRCLE, LST_SQUARE,
  118. LST_DECIMAL, LST_DECIMAL_ZERO,
  119. LST_LC_ROMAN, LST_UC_ROMAN,
  120. LST_LC_GREEK, LST_UC_GREEK,
  121. LST_LC_LATIN, LST_UC_LATIN,
  122. LST_LC_ALPHA, LST_UC_ALPHA,
  123. LST_ARMENIAN, LST_GEORGIAN,
  124. };
  125. enum { N_NUMBER='u', N_LENGTH='p', N_SCALE='m', N_PERCENT='%', N_AUTO='a' };
  126. struct fz_css_number_s
  127. {
  128. float value;
  129. int unit;
  130. };
  131. struct fz_css_color_s
  132. {
  133. unsigned char r, g, b, a;
  134. };
  135. struct fz_css_style_s
  136. {
  137. fz_css_number font_size;
  138. fz_css_number width, height;
  139. fz_css_number margin[4];
  140. fz_css_number padding[4];
  141. fz_css_number border_width[4];
  142. fz_css_number text_indent;
  143. unsigned int visibility : 2;
  144. unsigned int white_space : 3;
  145. unsigned int text_align : 2;
  146. unsigned int vertical_align : 3;
  147. unsigned int list_style_type : 4;
  148. unsigned int page_break_before : 3;
  149. unsigned int page_break_after : 3;
  150. unsigned int border_style_0 : 1;
  151. unsigned int border_style_1 : 1;
  152. unsigned int border_style_2 : 1;
  153. unsigned int border_style_3 : 1;
  154. fz_css_number line_height;
  155. fz_css_color background_color;
  156. fz_css_color border_color[4];
  157. fz_css_color color;
  158. fz_font *font;
  159. };
  160. enum
  161. {
  162. BOX_BLOCK, /* block-level: contains block, break, and flow boxes */
  163. BOX_BREAK, /* block-level: empty <br> tag boxes */
  164. BOX_FLOW, /* block-level: contains only inline boxes */
  165. BOX_INLINE, /* inline-level: contains only inline boxes */
  166. };
  167. struct fz_html_s
  168. {
  169. fz_pool *pool; /* pool allocator for this html tree */
  170. float page_w, page_h;
  171. float page_margin[4];
  172. fz_html_box *root;
  173. };
  174. struct fz_html_box_s
  175. {
  176. unsigned int type : 2;
  177. unsigned int is_first_flow : 1; /* for text-indent */
  178. unsigned int markup_dir : 2;
  179. unsigned int list_item : 27;
  180. float x, y, w, h; /* content */
  181. float padding[4];
  182. float margin[4];
  183. float border[4];
  184. float em;
  185. fz_html_box *up, *down, *last, *next;
  186. fz_html_flow *flow_head, **flow_tail;
  187. char *id, *href;
  188. fz_css_style style;
  189. };
  190. enum
  191. {
  192. FLOW_WORD = 0,
  193. FLOW_SPACE = 1,
  194. FLOW_BREAK = 2,
  195. FLOW_IMAGE = 3,
  196. FLOW_SBREAK = 4,
  197. FLOW_SHYPHEN = 5,
  198. FLOW_ANCHOR = 6
  199. };
  200. struct fz_html_flow_s
  201. {
  202. /* What type of node */
  203. unsigned int type : 3;
  204. /* Whether this should expand during justification */
  205. unsigned int expand : 1;
  206. /* Whether this node is currently taken as a line break */
  207. unsigned int breaks_line : 1;
  208. /* Direction setting for text - UAX#9 says 125 is the max */
  209. unsigned int bidi_level : 7;
  210. /* The script detected by the bidi code. */
  211. unsigned int script : 8;
  212. /* Whether the markup specifies a given language. */
  213. unsigned int markup_lang : 15;
  214. float x, y, w, h;
  215. fz_html_box *box; /* for style and em */
  216. union {
  217. char *text;
  218. fz_image *image;
  219. } content;
  220. fz_html_flow *next;
  221. };
  222. fz_css *fz_new_css(fz_context *ctx);
  223. void fz_parse_css(fz_context *ctx, fz_css *css, const char *source, const char *file);
  224. fz_css_property *fz_parse_css_properties(fz_context *ctx, fz_pool *pool, const char *source);
  225. void fz_drop_css(fz_context *ctx, fz_css *css);
  226. void fz_match_css(fz_context *ctx, fz_css_match *match, fz_css *css, fz_xml *node);
  227. void fz_match_css_at_page(fz_context *ctx, fz_css_match *match, fz_css *css);
  228. int fz_get_css_match_display(fz_css_match *node);
  229. void fz_default_css_style(fz_context *ctx, fz_css_style *style);
  230. void fz_apply_css_style(fz_context *ctx, fz_html_font_set *set, fz_css_style *style, fz_css_match *match);
  231. float fz_from_css_number(fz_css_number, float em, float width);
  232. float fz_from_css_number_scale(fz_css_number number, float scale, float em, float width);
  233. fz_html_font_set *fz_new_html_font_set(fz_context *ctx);
  234. void fz_add_html_font_face(fz_context *ctx, fz_html_font_set *set,
  235. const char *family, int is_bold, int is_italic, const char *src, fz_font *font);
  236. fz_font *fz_load_html_font(fz_context *ctx, fz_html_font_set *set, const char *family, int is_bold, int is_italic);
  237. void fz_drop_html_font_set(fz_context *ctx, fz_html_font_set *htx);
  238. void fz_add_css_font_faces(fz_context *ctx, fz_html_font_set *set, fz_archive *zip, const char *base_uri, fz_css *css);
  239. fz_html *fz_parse_html(fz_context *ctx, fz_html_font_set *htx, fz_archive *zip, const char *base_uri, fz_buffer *buf, const char *user_css);
  240. void fz_layout_html(fz_context *ctx, fz_html *html, float w, float h, float em);
  241. void fz_draw_html(fz_context *ctx, fz_device *dev, const fz_matrix *ctm, fz_html *html, int page);
  242. float fz_find_html_target(fz_context *ctx, fz_html *html, const char *id);
  243. fz_link *fz_load_html_links(fz_context *ctx, fz_html *html, int page, const char *base_uri, void *doc);
  244. void fz_drop_html(fz_context *ctx, fz_html *html);
  245. fz_bookmark fz_make_html_bookmark(fz_context *ctx, fz_html *html, int page);
  246. int fz_lookup_html_bookmark(fz_context *ctx, fz_html *html, fz_bookmark mark);
  247. #endif