Mercurial > repos > rhope

comparison string.rhope @ 51:7d6a6906b648

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added integer type conversions and started on the implementation of String
author Mike Pavone <pavone@retrodev.com>
date Thu, 22 Apr 2010 02:18:26 -0400
parents
children 079200bc3e75
comparison
equal deleted inserted replaced
50:689fb73e7612 51:7d6a6906b648
1
2 Blueprint Null String
3 {
4 }
5
6 Null String[:out(Null String)]
7 {
8 out <- Build[Null String()]
9 }
10
11 Depth@Null String[in:out(Int32)]
12 {
13 out <- 0i32
14 }
15
16 Blueprint Base String
17 {
18 Buffer
19 Length(Int32,Naked)
20 }
21
22 UTF8 Expect[num,arr,index,count,consumed:out]
23 {
24 byte <- [arr]Index[index]
25 {
26 If[[128u8]>[byte]]
27 {
28 //Error: ASCII byte when we were expecting part of a mutlibyte sequence
29 //treat each byte as a separate character
30 ncount <- [1i32]+[[count]+[consumed]]
31 }{
32 If[[192u8]>[byte]]
33 {
34 If[[num]=[1]]
35 {
36 //Sequence is complete count as single character
37 ncount <- [1i32]+[count]
38 }{
39 out <- UTF8 Expect[[num]-[1], arr, [index]+[1], count, [1i32]+[consumed]]
40 }
41 }{
42 //Error: too high to be a continuation byte
43 ncount <- [1i32]+[[count]+[consumed]]
44 }
45 }
46 }{
47 //Error: string ended in the middle of a multi-byte sequence
48 out <- [count]+[consumed]
49 }
50 Val[ncount]
51 {
52 [arr]Next[index]
53 {
54 out <- Count UTF8[arr, ~, ncount]
55 }{
56 out <- Val[ncount]
57 }
58 }
59 }
60
61 Count UTF8[arr,index,count:out]
62 {
63 byte <- [arr]Index[index]
64 If[[128u8]>[byte]]
65 { ncount <- [1i32]+[count] }
66 {
67 If[[192u8]>[byte]]
68 {
69 //Error: Encoding for 2nd,3rd or 4th byte of sequence
70 //treat as a single character
71 ncount <- [1i32]+[count]
72 }{
73 If[[224u8]>[byte]]
74 {
75 out <- UTF8 Expect[1, arr, [index]+[1], count, 1]
76 }{
77 If[[240u8]>[byte]]
78 {
79 out <- UTF8 Expect[2, arr, [index]+[1], count, 1]
80 }{
81 If[[245u8]>[byte]]
82 {
83 out <- UTF8 Expect[3, arr, [index]+[1], count, 1]
84 }{
85 //Error: Out of range of Unicode standard
86 //treat as a single character
87 ncount <- [1i32]+[count]
88 }
89 }
90 }
91 }
92 }
93 [arr]Next[index]
94 {
95 out <- Count UTF8[arr, ~, ncount]
96 }{
97 out <- Val[ncount]
98 }
99 }
100
101 Blueprint String
102 {
103 Left
104 Right
105 L Offset(Int32,Naked)
106 L Length(Int32,Naked)
107 Depth(Int32,Naked)
108 Length(Int32,Naked)
109 }
110
111 String[in(Array):out(Base String)]
112 {
113 out <- [[Build[Base String()]]Buffer <<[in]]Length <<[Count UTF8[in, 0, 0]]
114 }
115
116 Main[]
117 {
118 text <- [[[[[[[[[[[[[Array[1]
119 ]Append[36u8]
120 ]Append[194u8]
121 ]Append[162u8]
122 ]Append[236u8]
123 ]Append[130u8]
124 ]Append[172u8]
125 ]Append[240u8]
126 ]Append[164u8]
127 ]Append[173u8]
128 ]Append[162u8]
129 ]Append[194u8]
130 ]Append[36u8]
131 ]Append[162u8]
132 Print[Count UTF8[text, 0, 0]]
133 }
134