source/factorcss/Tokeniser.x

--
-- This file is part of FactorCSS
-- Licensed under the MIT License,
--                http://www.opensource.org/licenses/mit-license
-- Copyright 2004 James Bursa <james@semichrome.net>
--

-- Tokeniser for CSS 2.1
-- See CSS 2.1 G.2

{
module Tokeniser where
}

%wrapper "basic"

$h              = [0-9a-f]
$nonascii       = [\x80-\xff]
@unicode        = \\$h{1,6}(\r\n|[\ \t\r\n\f])?
@escape         = @unicode|\\[\x20-\x7e\x80-\xff]
@nmstart        = [_a-z]|$nonascii|@escape
@nmchar         = [\-_a-zA-Z0-9]|$nonascii|@escape
@nl             = \n|\r\n|\r|\f
@string1        = \"([\t\ \!\#\x24\x25&\x28-\x7e]|\\@nl|\'|$nonascii|@escape|\.)*\"
@string2        = \'([\t\ \!\#\x24\x25&\x28-\x7e]|\\@nl|\"|$nonascii|@escape|\.)*\'
@ident          = @nmstart@nmchar*
@name           = @nmchar+
@num            = [\-\053]?[0-9]+|[0-9]*"."[0-9]+
@string         = @string1|@string2
@url            = ([\!\#\x24\x25&\x2a-\x7e]|$nonascii|@escape)*
$s              = [\ \t\r\n\f]
@w              = $s*
@range          = \?{1,6}|$h(\?{0,5}|$h(\?{0,4}|$h(\?{0,3}|$h(\?{0,2}|$h(\??|$h)))))
$all            = \x00-\xff
@comment        = \/\*$all#\x2a*\*+($all#[\/\x2a]$all#\x2a*\*+)*\/

tokens :-

$s+                     ;
@comment                ;

"<!--"                  ;
"-->"                   ;
"~="                    { \s -> INCLUDES }
"|="                    { \s -> DASHMATCH }

@w"{"                   { \s -> LBRACE }
@w"+"                   { \s -> PLUS }
@w">"                   { \s -> GREATER }
@w","                   { \s -> COMMA }

@string                 { \s -> STRING s }

@ident                  { \s -> IDENT s }

"#"@name                { \s -> HASH s }

"@import"               { \s -> IMPORT_SYM }
"@page"                 { \s -> PAGE_SYM }
"@media"                { \s -> MEDIA_SYM }
"@charset"              { \s -> CHARSET_SYM }

"!"@w"important"        { \s -> IMPORTANT_SYM }

@num em                 { \s -> EMS (read (take (length s - 2) s)) }
@num ex                 { \s -> EXS (read (take (length s - 2) s)) }
@num px                 { \s -> LENGTH s }
@num cm                 { \s -> LENGTH s }
@num mm                 { \s -> LENGTH s }
@num in                 { \s -> LENGTH s }
@num pt                 { \s -> LENGTH s }
@num pc                 { \s -> LENGTH s }
@num deg                { \s -> ANGLE s }
@num rad                { \s -> ANGLE s }
@num grad               { \s -> ANGLE s }
@num ms                 { \s -> TIME s }
@num s                  { \s -> TIME s }
@num Hz                 { \s -> FREQ s }
@num kHz                { \s -> FREQ s }
@num @ident             { \s -> DIMEN s }
@num "%"                { \s -> PERCENTAGE (read (take (length s - 1) s)) }
@num                    { \s -> NUMBER (read s) }

"url("@w@string@w")"    { \s -> URI s }
"url("@w@url@w")"       { \s -> URI s }
@ident"\050"            { \s -> FUNCTION s }

";"                     { \s -> SEMI }
"}"                     { \s -> RBRACE }
":"                     { \s -> COLON }
"/"                     { \s -> SLASH }
"-"                     { \s -> MINUS }
"."                     { \s -> DOT }
"*"                     { \s -> ASTERISK }
"["                     { \s -> LBRAC }
"]"                     { \s -> RBRAC }
"="                     { \s -> EQUALS }
")"                     { \s -> RPAREN }
.                       { \s -> DELIM (head s) }


{
data Token =
        INCLUDES        |
        DASHMATCH       |
        LBRACE          |
        PLUS            |
        GREATER         |
        COMMA           |
        STRING String   |
        IDENT String    |
        HASH String     |
        IMPORT_SYM      |
        PAGE_SYM        |
        MEDIA_SYM       |
        CHARSET_SYM     |
        IMPORTANT_SYM   |
        EMS Float       |
        EXS Float       |
        LENGTH String   |
        ANGLE String    |
        TIME String     |
        FREQ String     |
        DIMEN String    |
        PERCENTAGE Float |
        NUMBER Float    |
        URI String      |
        FUNCTION String |
        SEMI            |
        RBRACE          |
        COLON           |
        SLASH           |
        MINUS           |
        DOT             |
        ASTERISK        |
        LBRAC           |
        RBRAC           |
        EQUALS          |
        RPAREN          |
        DELIM Char
        deriving (Eq, Show)
}
1	james	63	--
2			-- This file is part of FactorCSS
3			-- Licensed under the MIT License,
4			-- http://www.opensource.org/licenses/mit-license
5			-- Copyright 2004 James Bursa <james@semichrome.net>
6			--
7
8	james	61	-- Tokeniser for CSS 2.1
9			-- See CSS 2.1 G.2
10
11			{
12			module Tokeniser where
13			}
14
15			%wrapper "basic"
16
17			$h = [0-9a-f]
18			$nonascii = [\x80-\xff]
19			@unicode = \\$h{1,6}(\r\n\|[\ \t\r\n\f])?
20			@escape = @unicode\|\\[\x20-\x7e\x80-\xff]
21			@nmstart = [_a-z]\|$nonascii\|@escape
22			@nmchar = [\-_a-zA-Z0-9]\|$nonascii\|@escape
23			@nl = \n\|\r\n\|\r\|\f
24	james	64	@string1 = \"([\t\ \!\#\x24\x25&\x28-\x7e]\|\\@nl\|\'\|$nonascii\|@escape\|\.)*\"
25			@string2 = \'([\t\ \!\#\x24\x25&\x28-\x7e]\|\\@nl\|\"\|$nonascii\|@escape\|\.)*\'
26	james	61	@ident = @nmstart@nmchar*
27			@name = @nmchar+
28			@num = [\-\053]?[0-9]+\|[0-9]*"."[0-9]+
29			@string = @string1\|@string2
30	james	64	@url = ([\!\#\x24\x25&\x2a-\x7e]\|$nonascii\|@escape)*
31	james	61	$s = [\ \t\r\n\f]
32			@w = $s*
33			@range = \?{1,6}\|$h(\?{0,5}\|$h(\?{0,4}\|$h(\?{0,3}\|$h(\?{0,2}\|$h(\??\|$h)))))
34	james	64	$all = \x00-\xff
35			@comment = \/\$all#\x2a\+($all#[\/\x2a]$all#\x2a\+)\/
36	james	61
37			tokens :-
38
39			$s+ ;
40			@comment ;
41
42			"<!--" ;
43			"-->" ;
44			"~=" { \s -> INCLUDES }
45			"\|=" { \s -> DASHMATCH }
46
47			@w"{" { \s -> LBRACE }
48			@w"+" { \s -> PLUS }
49			@w">" { \s -> GREATER }
50			@w"," { \s -> COMMA }
51
52			@string { \s -> STRING s }
53
54			@ident { \s -> IDENT s }
55
56			"#"@name { \s -> HASH s }
57
58			"@import" { \s -> IMPORT_SYM }
59			"@page" { \s -> PAGE_SYM }
60			"@media" { \s -> MEDIA_SYM }
61			"@charset" { \s -> CHARSET_SYM }
62
63			"!"@w"important" { \s -> IMPORTANT_SYM }
64
65			@num em { \s -> EMS (read (take (length s - 2) s)) }
66			@num ex { \s -> EXS (read (take (length s - 2) s)) }
67			@num px { \s -> LENGTH s }
68			@num cm { \s -> LENGTH s }
69			@num mm { \s -> LENGTH s }
70			@num in { \s -> LENGTH s }
71			@num pt { \s -> LENGTH s }
72			@num pc { \s -> LENGTH s }
73			@num deg { \s -> ANGLE s }
74			@num rad { \s -> ANGLE s }
75			@num grad { \s -> ANGLE s }
76			@num ms { \s -> TIME s }
77			@num s { \s -> TIME s }
78			@num Hz { \s -> FREQ s }
79			@num kHz { \s -> FREQ s }
80			@num @ident { \s -> DIMEN s }
81			@num "%" { \s -> PERCENTAGE (read (take (length s - 1) s)) }
82			@num { \s -> NUMBER (read s) }
83
84			"url("@w@string@w")" { \s -> URI s }
85			"url("@w@url@w")" { \s -> URI s }
86			@ident"\050" { \s -> FUNCTION s }
87
88			";" { \s -> SEMI }
89			"}" { \s -> RBRACE }
90			":" { \s -> COLON }
91			"/" { \s -> SLASH }
92			"-" { \s -> MINUS }
93			"." { \s -> DOT }
94			"*" { \s -> ASTERISK }
95			"[" { \s -> LBRAC }
96			"]" { \s -> RBRAC }
97			"=" { \s -> EQUALS }
98			")" { \s -> RPAREN }
99			. { \s -> DELIM (head s) }
100
101
102			{
103			data Token =
104			INCLUDES \|
105			DASHMATCH \|
106			LBRACE \|
107			PLUS \|
108			GREATER \|
109			COMMA \|
110			STRING String \|
111			IDENT String \|
112			HASH String \|
113			IMPORT_SYM \|
114			PAGE_SYM \|
115			MEDIA_SYM \|
116			CHARSET_SYM \|
117			IMPORTANT_SYM \|
118			EMS Float \|
119			EXS Float \|
120			LENGTH String \|
121			ANGLE String \|
122			TIME String \|
123			FREQ String \|
124			DIMEN String \|
125			PERCENTAGE Float \|
126			NUMBER Float \|
127			URI String \|
128			FUNCTION String \|
129			SEMI \|
130			RBRACE \|
131			COLON \|
132			SLASH \|
133			MINUS \|
134			DOT \|
135			ASTERISK \|
136			LBRAC \|
137			RBRAC \|
138			EQUALS \|
139			RPAREN \|
140			DELIM Char
141			deriving (Eq, Show)
142			}