source/factorcss/Tokeniser.x

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