home *** CD-ROM | disk | FTP | other *** search
- # $Id: style.lex,v 1.3 1996/09/12 17:25:24 steveb Exp $
- # Copyright (C) 1987-1995 Verity, Inc.
- # This is the one which is hit --hsc
- # Document Dataset Descriptor
- #
- $control: 1
- $charmap: 8859
- lex:
- {
- define: WHT "[ \t]"
- define: NL "{WHT}*\n"
-
- #token: WORD "[ΩεΣ╩╬─└δßα]+"
- token: WORD "[A-Za-z0-9/&'±╤└┴┬├─┼╞╟╚╔╩╦╠═╬╧╨╤╥╙╘╒╓╫╪┘┌█▄▌▐▀αßΓπΣσµτΦΘΩδφ∞ε∩≡±≥≤⌠⌡÷≈°∙·√ⁿ²■ ]+" #word
- token: WORD "[a-z0-9/&'±╤]+" #word
-
- token: WORD "[0-9]+\\.[0-9]+" #word
- token: EOS "[.?!]" #end of sentence
- token: NEWLINE "{NL}" #single end-of-line
- token: PARA "{NL}{NL}" #end of paragraph
- token: WHITE "{WHT}" #whitespace
- token: PUNCT "." #all other text
- }
- $$