Line | Count | Source |
1 | | |
2 | | #include "wc.h" |
3 | | #include "iso2022.h" |
4 | | #include "hz.h" |
5 | | #include "wtf.h" |
6 | | #ifdef USE_UNICODE |
7 | | #include "ucs.h" |
8 | | #endif |
9 | | |
10 | | Str |
11 | | wc_conv_from_hz(Str is, wc_ces ces) |
12 | 276 | { |
13 | 276 | Str os; |
14 | 276 | wc_uchar *sp = (wc_uchar *)is->ptr; |
15 | 276 | wc_uchar *ep = sp + is->length; |
16 | 276 | wc_uchar *p; |
17 | 276 | int state = WC_HZ_NOSTATE; |
18 | | |
19 | 583 | for (p = sp; p < ep && *p < 0x80 && *p != WC_C_HZ_TILDA; p++) |
20 | 307 | ; |
21 | 276 | if (p == ep) |
22 | 28 | return is; |
23 | 248 | os = Strnew_size(is->length); |
24 | 248 | if (p > sp) |
25 | 27 | Strcat_charp_n(os, is->ptr, (int)(p - sp)); |
26 | | |
27 | 1.03M | for (; p < ep; p++) { |
28 | 1.03M | switch (state) { |
29 | 704k | case WC_HZ_NOSTATE: |
30 | 704k | if (*p == WC_C_HZ_TILDA) |
31 | 2.74k | state = WC_HZ_TILDA; |
32 | 701k | else if (WC_ISO_MAP[*p] == WC_ISO_MAP_GR) |
33 | 324k | state = WC_HZ_MBYTE1_GR; /* GB 2312 ? */ |
34 | 376k | else if (*p & 0x80) |
35 | 208k | wtf_push_unknown(os, p, 1); |
36 | 168k | else |
37 | 168k | Strcat_char(os, (char)*p); |
38 | 704k | break; |
39 | 2.92k | case WC_HZ_TILDA: |
40 | 2.92k | if (*p == WC_C_HZ_SI) { |
41 | 691 | state = WC_HZ_MBYTE; |
42 | 691 | break; |
43 | 2.23k | } else if (*p == WC_C_HZ_TILDA) |
44 | 513 | Strcat_char(os, (char)*p); |
45 | 1.72k | else if (*p == '\n') |
46 | 195 | break; |
47 | 1.52k | else |
48 | 1.52k | wtf_push_unknown(os, p-1, 2); |
49 | 2.04k | state = WC_HZ_NOSTATE; |
50 | 2.04k | break; |
51 | 1.19k | case WC_HZ_TILDA_MB: |
52 | 1.19k | if (*p == WC_C_HZ_SO || *p == '\n') { |
53 | 605 | state = WC_HZ_NOSTATE; |
54 | 605 | break; |
55 | 605 | } |
56 | 587 | else if (WC_ISO_MAP[*p & 0x7f] == WC_ISO_MAP_GL) |
57 | 362 | wtf_push(os, WC_CCS_GB_2312, ((wc_uint32)*(p-1) << 8) | *p); |
58 | 225 | else |
59 | 225 | wtf_push_unknown(os, p-1, 2); |
60 | 587 | state = WC_HZ_MBYTE; |
61 | 587 | break; |
62 | 3.67k | case WC_HZ_MBYTE: |
63 | 3.67k | if (*p == WC_C_HZ_TILDA) |
64 | 1.19k | state = WC_HZ_TILDA_MB; |
65 | 2.47k | else if (WC_ISO_MAP[*p & 0x7f] == WC_ISO_MAP_GL) |
66 | 1.45k | state = WC_HZ_MBYTE1; |
67 | 1.02k | else |
68 | 1.02k | wtf_push_unknown(os, p, 1); |
69 | 3.67k | break; |
70 | 1.43k | case WC_HZ_MBYTE1: |
71 | 1.43k | if (WC_ISO_MAP[*p & 0x7f] == WC_ISO_MAP_GL) |
72 | 964 | wtf_push(os, WC_CCS_GB_2312, ((wc_uint32)*(p-1) << 8) | *p); |
73 | 475 | else |
74 | 475 | wtf_push_unknown(os, p-1, 2); |
75 | 1.43k | state = WC_HZ_MBYTE; |
76 | 1.43k | break; |
77 | 324k | case WC_HZ_MBYTE1_GR: |
78 | 324k | if (WC_ISO_MAP[*p] == WC_ISO_MAP_GR) |
79 | 304k | wtf_push(os, WC_CCS_GB_2312, ((wc_uint32)*(p-1) << 8) | *p); |
80 | 19.5k | else |
81 | 19.5k | wtf_push_unknown(os, p-1, 2); |
82 | 324k | state = WC_HZ_NOSTATE; |
83 | 324k | break; |
84 | 1.03M | } |
85 | 1.03M | } |
86 | 248 | switch (state) { |
87 | 11 | case WC_HZ_TILDA: |
88 | 17 | case WC_HZ_TILDA_MB: |
89 | 32 | case WC_HZ_MBYTE1: |
90 | 52 | case WC_HZ_MBYTE1_GR: |
91 | 52 | wtf_push_unknown(os, p-1, 1); |
92 | 52 | break; |
93 | 248 | } |
94 | 248 | return os; |
95 | 248 | } |
96 | | |
97 | | void |
98 | | wc_push_to_hz(Str os, wc_wchar_t cc, wc_status *st) |
99 | 24.9M | { |
100 | 46.2M | while (1) { |
101 | 46.2M | switch (cc.ccs) { |
102 | 3.23M | case WC_CCS_US_ASCII: |
103 | 3.23M | if (st->gl) { |
104 | 2.98M | Strcat_char(os, WC_C_HZ_TILDA); |
105 | 2.98M | Strcat_char(os, WC_C_HZ_SO); |
106 | 2.98M | st->gl = 0; |
107 | 2.98M | } |
108 | 3.23M | if ((char)cc.code == WC_C_HZ_TILDA) |
109 | 2.93M | Strcat_char(os, WC_C_HZ_TILDA); |
110 | 3.23M | Strcat_char(os, (char)cc.code); |
111 | 3.23M | return; |
112 | 12.0M | case WC_CCS_GB_2312: |
113 | 12.0M | if (! st->gl) { |
114 | 11.6M | Strcat_char(os, WC_C_HZ_TILDA); |
115 | 11.6M | Strcat_char(os, WC_C_HZ_SI); |
116 | 11.6M | st->gl = 1; |
117 | 11.6M | } |
118 | 12.0M | Strcat_char(os, (char)((cc.code >> 8) & 0x7f)); |
119 | 12.0M | Strcat_char(os, (char)(cc.code & 0x7f)); |
120 | 12.0M | return; |
121 | 8.74M | case WC_CCS_UNKNOWN_W: |
122 | 8.74M | if (WcOption.no_replace) |
123 | 0 | return; |
124 | 8.74M | if (st->gl) { |
125 | 8.63M | Strcat_char(os, WC_C_HZ_TILDA); |
126 | 8.63M | Strcat_char(os, WC_C_HZ_SO); |
127 | 8.63M | st->gl = 0; |
128 | 8.63M | } |
129 | 8.74M | Strcat_charp(os, WC_REPLACE_W); |
130 | 8.74M | return; |
131 | 912k | case WC_CCS_UNKNOWN: |
132 | 912k | if (WcOption.no_replace) |
133 | 0 | return; |
134 | 912k | if (st->gl) { |
135 | 23.1k | Strcat_char(os, WC_C_HZ_TILDA); |
136 | 23.1k | Strcat_char(os, WC_C_HZ_SO); |
137 | 23.1k | st->gl = 0; |
138 | 23.1k | } |
139 | 912k | Strcat_charp(os, WC_REPLACE); |
140 | 912k | return; |
141 | 21.3M | default: |
142 | 21.3M | #ifdef USE_UNICODE |
143 | 21.3M | if (WcOption.ucs_conv) |
144 | 21.3M | cc = wc_any_to_any_ces(cc, st); |
145 | 0 | else |
146 | 0 | #endif |
147 | 0 | cc.ccs = WC_CCS_IS_WIDE(cc.ccs) ? WC_CCS_UNKNOWN_W : WC_CCS_UNKNOWN; |
148 | 21.3M | continue; |
149 | 46.2M | } |
150 | 46.2M | } |
151 | 24.9M | } |
152 | | |
153 | | void |
154 | | wc_push_to_hz_end(Str os, wc_status *st) |
155 | 459 | { |
156 | 459 | if (st->gl) { |
157 | 223 | Strcat_char(os, WC_C_HZ_TILDA); |
158 | | Strcat_char(os, WC_C_HZ_SO); |
159 | 223 | st->gl = 0; |
160 | 223 | } |
161 | 459 | } |