vendor/bstr/scripts/regex/grapheme.sh

   1 #!/bin/sh
   2
   3 # vim: indentexpr= nosmartindent autoindent
   4 # vim: tabstop=2 shiftwidth=2 softtabstop=2
   5
   6 # This regex was manually written, derived from the rules in UAX #29.
   7 # Particularly, from Table 1c, which lays out a regex for grapheme clusters.
   8
   9 CR="\p{gcb=CR}"
  10 LF="\p{gcb=LF}"
  11 Control="\p{gcb=Control}"
  12 Prepend="\p{gcb=Prepend}"
  13 L="\p{gcb=L}"
  14 V="\p{gcb=V}"
  15 LV="\p{gcb=LV}"
  16 LVT="\p{gcb=LVT}"
  17 T="\p{gcb=T}"
  18 RI="\p{gcb=RI}"
  19 Extend="\p{gcb=Extend}"
  20 ZWJ="\p{gcb=ZWJ}"
  21 SpacingMark="\p{gcb=SpacingMark}"
  22
  23 Any="\p{any}"
  24 ExtendPict="\p{Extended_Pictographic}"
  25
  26 echo "(?x)
  27 $CR $LF
  28 |
  29 $Control
  30 |
  31 $Prepend*
  32 (
  33   (
  34     ($L* ($V+ | $LV $V* | $LVT) $T*)
  35     |
  36     $L+
  37     |
  38     $T+
  39   )
  40   |
  41   $RI $RI
  42   |
  43   $ExtendPict ($Extend* $ZWJ $ExtendPict)*
  44   |
  45   [^$Control $CR $LF]
  46 )
  47 [$Extend $ZWJ $SpacingMark]*
  48 |
  49 $Any
  50 "