bzr branch
http://gegoxaren.bato24.eu/bzr/lenasys/0.1
|
4.1.1
by c11emian
Added <head> and <title> tags |
1 |
// Token class and storage definition |
2 |
function token (kind,val,fromchar,tochar,row) { |
|
3 |
this.kind = kind; |
|
4 |
this.val = val; |
|
5 |
this.from = fromchar; |
|
6 |
this.to = tochar; |
|
7 |
this.row = row; |
|
8 |
} |
|
9 |
|
|
10 |
var tokens = []; // Array to hold the tokens. |
|
11 |
||
12 |
//---------------------------------------------------------- |
|
13 |
// Store token in tokens array |
|
14 |
// Creates a new token object using the constructor |
|
15 |
//---------------------------------------------------------- |
|
16 |
||
17 |
function maketoken(kind,val,from,to,rowno) |
|
18 |
{ |
|
19 |
newtoken=new token(kind,val,from,to,rowno); |
|
20 |
tokens.push(newtoken); |
|
21 |
} |
|
22 |
||
23 |
//---------------------------------------------------------- |
|
24 |
// Writes error from tokenizer |
|
25 |
//---------------------------------------------------------- |
|
26 |
|
|
27 |
function error(str,val,row) |
|
28 |
{ |
|
29 |
alert("Tokenizer Error: "+str+val+" at row "+row); |
|
30 |
} |
|
31 |
||
32 |
//---------------------------------------------------------- |
|
33 |
// Tokenize function |
|
34 |
// Tokenizer partly based on ideas from the very clever tokenizer written by Douglas Cockford |
|
35 |
// The tokenizer is passed a string, and a string of prefix and suffix terminators |
|
36 |
//---------------------------------------------------------- |
|
37 |
|
|
38 |
function tokenize(instring,inprefix,insuffix){ |
|
39 |
|
|
40 |
var from; // index of the start of the token. |
|
41 |
var i = 0; // index of the current character. |
|
42 |
var length=instring.length; // length of the string |
|
43 |
||
44 |
var c; // current character. |
|
45 |
var n; // current numerical value |
|
46 |
var q; // current quote character |
|
47 |
var str; // current string value. |
|
48 |
var row=1; // current row value |
|
49 |
|
|
50 |
||
51 |
c = instring.charAt(i); |
|
52 |
while (c) { |
|
53 |
|
|
54 |
from = i; |
|
55 |
||
56 |
if (c <= ' '){ // White space |
|
57 |
if((c=='\n')||(c=='\r')||(c == '')) row++; // Add row if this white space is a row terminator |
|
58 |
i++; |
|
59 |
c = instring.charAt(i); |
|
60 |
}else if ((c >='a'&&c<='z')||(c>='A'&&c<='Z')) { // Names i.e. Text |
|
61 |
str = c; |
|
62 |
i++; |
|
63 |
for (;;) { |
|
64 |
c = instring.charAt(i); |
|
65 |
if ((c >='a'&&c<='z')||(c>='A'&&c<='Z')||(c>='0'&&c<='9')||c=='_'){ |
|
66 |
str += c; |
|
67 |
i++; |
|
68 |
}else{ |
|
69 |
break; |
|
70 |
} |
|
71 |
} |
|
72 |
maketoken('name',str,from,i,row); |
|
73 |
} else if (c >= '0' && c <= '9') { // Number token |
|
74 |
str = c; |
|
75 |
i++; |
|
76 |
for (;;) { |
|
77 |
c = instring.charAt(i); |
|
78 |
if (c < '0' || c > '9') break; |
|
79 |
i++; |
|
80 |
str+=c; |
|
81 |
} |
|
82 |
if(c=='.'){ |
|
83 |
i++; |
|
84 |
str+=c; |
|
85 |
for(;;){ |
|
86 |
c=instring.charAt(i); |
|
87 |
if (c < '0' || c > '9') break; |
|
88 |
i++; |
|
89 |
str+=c; |
|
90 |
} |
|
91 |
} |
|
92 |
if (c=='e'||c=='E') { |
|
93 |
i++; |
|
94 |
str+=c; |
|
95 |
c=instring.charAt(i); |
|
96 |
if(c=='-'||c=='+') { |
|
97 |
i+=1; |
|
98 |
str+=c; |
|
99 |
c=instring.charAt(i); |
|
100 |
} |
|
101 |
if (c < '0' || c > '9') error('Bad Exponent in Number: ',str,row); |
|
102 |
do { |
|
103 |
i++; |
|
104 |
str+=c; |
|
105 |
c=instring.charAt(i); |
|
106 |
}while(c>='0'&&c<='9'); |
|
107 |
} |
|
108 |
if (c>='a'&&c<='z'){ |
|
109 |
str += c; |
|
110 |
i += 1; |
|
111 |
error('Bad Number: ',str,row); |
|
112 |
} |
|
113 |
n=+str; |
|
114 |
if (isFinite(n)) { |
|
115 |
maketoken('number',n,from,i,row); |
|
116 |
} else { |
|
117 |
error('Bad Number: ',str,row); |
|
118 |
} |
|
119 |
||
120 |
} else if(c=='\''||c=='"'){ // String .. handles c style breaking codes |
|
121 |
str=''; |
|
122 |
q=c; |
|
123 |
i++; |
|
124 |
for (;;) { |
|
125 |
c=instring.charAt(i); |
|
126 |
||
127 |
if (c<' '){ |
|
128 |
if((c=='\n')||(c=='\r')||(c == '')) row++; // Add row if this white space is a row terminator |
|
129 |
error('Unterminated String: ',str,row); |
|
130 |
} |
|
131 |
||
132 |
if (c==q) break; |
|
133 |
||
134 |
if (c=='\\'){ |
|
135 |
i += 1; |
|
136 |
if (i >= length) { |
|
137 |
error('Unterminated String: ',str,row); |
|
138 |
} |
|
139 |
c=instring.charAt(i); |
|
140 |
|
|
141 |
if(c=='b'){ c='\b'; break; } |
|
142 |
if(c=='f'){ c='\f'; break; } |
|
143 |
if(c=='n'){ c='\n'; break; } |
|
144 |
if(c=='r'){ c='\r'; break; } |
|
145 |
if(c=='t'){ c='\t'; break; } |
|
146 |
if(c=='u'){ |
|
147 |
if (i >= length) { |
|
148 |
error('Unterminated String: ',str,row); |
|
149 |
} |
|
150 |
c = parseInt(this.substr(i + 1, 4), 16); |
|
151 |
if (!isFinite(c) || c < 0) { |
|
152 |
error('Unterminated String: ',str,row); |
|
153 |
} |
|
154 |
c = String.fromCharCode(c); |
|
155 |
i+=4; |
|
156 |
break; |
|
157 |
} |
|
158 |
} |
|
159 |
str += c; |
|
160 |
i++; |
|
161 |
} |
|
162 |
i++; |
|
163 |
maketoken('string',str,from,i,row); |
|
164 |
c=instring.charAt(i); |
|
165 |
||
166 |
} else if (c=='/'&&instring.charAt(i+1)=='/'){ // Comment of // type ... does not cover block comments |
|
167 |
i++; |
|
168 |
for (;;){ |
|
169 |
c=instring.charAt(i); |
|
170 |
if (c=='\n'||c=='\r'||c=='') { |
|
171 |
row++; |
|
172 |
break; |
|
173 |
} |
|
174 |
i++; |
|
175 |
} |
|
176 |
} else if (c=='/'&&instring.charAt(i+1)=='*'){ // Block comment of /* type |
|
177 |
i++; |
|
178 |
for (;;){ |
|
179 |
c=instring.charAt(i); |
|
180 |
if ((c=='*'&&instring.charAt(i+1)=='/')||(i==length)) { |
|
181 |
i+=2; |
|
182 |
c=instring.charAt(i); |
|
183 |
break; |
|
184 |
} |
|
185 |
if (c=='\n'||c=='\r'||c=='') { |
|
186 |
row++; |
|
187 |
} |
|
188 |
i++; |
|
189 |
} |
|
190 |
}else if(inprefix.indexOf(c) >= 0) { // Multi-character Operators |
|
191 |
str = c; |
|
192 |
i++; |
|
193 |
while (true){ |
|
194 |
c=instring.charAt(i); |
|
195 |
if (i >= length || insuffix.indexOf(c) < 0) { |
|
196 |
break; |
|
197 |
} |
|
198 |
str += c; |
|
199 |
i++; |
|
200 |
} |
|
201 |
maketoken('operator',str,from,i,row); |
|
202 |
} else { // Single-character Operators |
|
203 |
i++; |
|
204 |
maketoken('operator',c,from,i,row); |
|
205 |
c = instring.charAt(i); |
|
206 |
} |
|
207 |
|
|
208 |
} |
|
209 |
} |
|
210 |
|
|
211 |
function newbutton() |
|
212 |
{ |
|
213 |
var funclist; |
|
214 |
var oplist; |
|
215 |
|
|
216 |
oplist=document.getElementById('infobox'); |
|
217 |
|
|
218 |
string_tokenize="foo // Lederhosen\nsin'foo'(200.0*r)+(-21.4/1.51e-6)+/*Plodder*/(Feeeb+400)*ln(3000);\ncos(200.1-atan2(41.0,51.0));\nHello+85-4*2.6-1.51e-6;"; |
|
219 |
||
220 |
tokenize(string_tokenize,"<>+-&","=>&:"); |
|
221 |
||
222 |
|
|
223 |
// Iterate over token objects and print kind of each token and token type in window |
|
224 |
printout=document.getElementById('infobox'); |
|
225 |
for(i=0;i<tokens.length;i++){ |
|
226 |
printout.innerHTML+=tokens[i].kind+" "+tokens[i].val+" ("+tokens[i].from+"-"+tokens[i].to+") at: "+tokens[i].row+"<br>"; |
|
227 |
} |
|
228 |
|
|
229 |
} |