khtml/rendering/break_lines.h

   1 #ifndef BREAK_LINES_H
   2 #define BREAK_LINES_H
   3
   4 #include <QtCore/QString>
   5
   6 namespace khtml {
   7
   8     /*
   9       array of unicode codes where breaking shouldn't occur.
  10       (in sorted order because of using with binary search)
  11       these are currently for Japanese, though simply adding
  12       Korean, Chinese ones should work as well
  13     */
  14     /*
  15       dontbreakbefore[] contains characters not covered by QChar::Punctuation_Close that shouldn't be broken before.
  16       chars included in QChar::Punctuation_Close are listed below.(look at UAX #14)
  17          - 3001 ideographic comma
  18          - 3002 ideographic full stop
  19          - FE50 small comma
  20          - FF52 small full stop
  21          - FF0C fullwidth comma
  22          - FF0E fullwidth full stop
  23          - FF61 halfwidth ideographic full stop
  24          - FF64 halfwidth ideographic comma
  25       these character is commented out.
  26     */
  27     const ushort dontbreakbefore[] = {
  28         //0x3001,   //ideographic comma
  29         //0x3002,   //ideographic full stop
  30         0x3005, //ideographic iteration mark
  31         0x3009, //right angle bracket
  32         0x300b, //right double angle bracket
  33         0x300d, //right corner bracket
  34         0x300f, //right white corner bracket
  35         0x3011, //right black lenticular bracket
  36         0x3015, //right tortoise shell bracket
  37         0x3041, //small a hiragana
  38         0x3043, //small i hiragana
  39         0x3045, //small u hiragana
  40         0x3047, //small e hiragana
  41         0x3049, //small o hiragana
  42         0x3063, //small tsu hiragana
  43         0x3083, //small ya hiragana
  44         0x3085, //small yu hiragana
  45         0x3087, //small yo hiragana
  46         0x308E, //small wa hiragana
  47         0x309B, //jap voiced sound mark
  48         0x309C, //jap semi-voiced sound mark
  49         0x309D, //jap iteration mark hiragana
  50         0x309E, //jap voiced iteration mark hiragana
  51         0x30A1, //small a katakana
  52         0x30A3, //small i katakana
  53         0x30A5, //small u katakana
  54         0x30A7, //small e katakana
  55         0x30A9, //small o katakana
  56         0x30C3, //small tsu katakana
  57         0x30E3, //small ya katakana
  58         0x30E5, //small yu katakana
  59         0x30E7, //small yo katakana
  60         0x30EE, //small wa katakana
  61         0x30F5, //small ka katakana
  62         0x30F6, //small ke katakana
  63         0x30FC, //jap prolonged sound mark
  64         0x30FD, //jap iteration mark katakana
  65         0x30FE, //jap voiced iteration mark katakana
  66         //0xFE50,   //small comma
  67         //0xFF52,   //small full stop
  68         0xFF01, //fullwidth exclamation mark
  69         0xFF09, //fullwidth right parenthesis
  70         //0xFF0C,   //fullwidth comma
  71         0xFF0D, //fullwidth hypen-minus
  72         //0xFF0E,   //fullwidth full stop
  73         0xFF1F, //fullwidth question mark
  74         0xFF3D, //fullwidth right square bracket
  75         0xFF5D, //fullwidth right curly bracket
  76         //0xFF61,   //halfwidth ideographic full stop
  77         0xFF63, //halfwidth right corner bracket
  78         //0xFF64,   //halfwidth ideographic comma
  79         0xFF67, //halfwidth katakana letter small a
  80         0xFF68, //halfwidth katakana letter small i
  81         0xFF69, //halfwidth katakana letter small u
  82         0xFF6a, //halfwidth katakana letter small e
  83         0xFF6b, //halfwidth katakana letter small o
  84         0xFF6c, //halfwidth katakana letter small ya
  85         0xFF6d, //halfwidth katakana letter small yu
  86         0xFF6e, //halfwidth katakana letter small yo
  87         0xFF6f, //halfwidth katakana letter small tu
  88         0xFF70  //halfwidth katakana-hiragana prolonged sound mark
  89     };
  90
  91     // characters that aren't covered by QChar::Punctuation_Open
  92     const ushort dontbreakafter[] = {
  93         0x3012, //postal mark
  94         0xFF03, //full width pound mark
  95         0xFF04, //full width dollar sign
  96         0xFF20, //full width @
  97         0xFFE1, //full width british pound sign
  98         0xFFE5  //full width yen sign
  99     };
 100
 101     inline bool break_bsearch( const ushort* arr, const ushort val ) {
 102         int left = 0;
 103         int right = (sizeof(arr) / sizeof(ushort)) - 1;
 104
 105         while (1) {
 106             if (left == right)
 107                 return val != arr[left];
 108
 109             int i = (left + right) >> 1;
 110             if ( val == arr[i] )
 111                 return false;
 112             if ( val < arr[i] )
 113                 right = i;
 114             else
 115                 left = i + 1;
 116         }
 117     }
 118
 119     bool isBreakableThai( const QChar *string, const int pos, const int len);
 120     void cleanup_thaibreaks();
 121
 122     inline bool isBreakable( const QChar *str, const int pos, int len )
 123     {
 124         const QChar *c = str+pos;
 125         unsigned short ch = c->unicode();
 126         if ( ch > 0xff ) {
 127             // not latin1, need to do more sophisticated checks for asian fonts
 128             unsigned char row = c->row();
 129             if ( row == 0x0e ) {
 130                 // 0e00 - 0e7f == Thai
 131                 if ( c->cell() < 0x80 ) {
 132                     // consult libthai
 133                     return isBreakableThai(str, pos, len);
 134                 } else
 135                     return false;
 136             }
 137             if ( row > 0x2d && row < 0xfb || row == 0x11 ) {
 138                 /* asian line breaking. */
 139                 if ( pos == 0 )
 140                     return false; // never break before first character
 141
 142                 // check for simple punctuation cases
 143                 QChar::Category cat = c->category();
 144                 if ( cat == QChar::Punctuation_Close ||
 145                      cat == QChar::Punctuation_Other ||
 146                      (str+(pos-1))->category() == QChar::Punctuation_Open )
 147                     return false;
 148
 149                 // do binary search in dontbreak[]
 150                 return break_bsearch(dontbreakbefore, c->unicode()) &&
 151                        break_bsearch(dontbreakafter, (str+(pos-1))->unicode());
 152             } else // no asian font
 153                 return c->isSpace();
 154         } else {
 155             if ( ch == ' ' || ch == '\n' )
 156                 return true;
 157         }
 158         return false;
 159     }
 160
 161 }
 162
 163 #endif