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 | 274 | { |
13 | 274 | Str os; |
14 | 274 | wc_uchar *sp = (wc_uchar *)is->ptr; |
15 | 274 | wc_uchar *ep = sp + is->length; |
16 | 274 | wc_uchar *p; |
17 | 274 | int state = WC_HZ_NOSTATE; |
18 | | |
19 | 792 | for (p = sp; p < ep && *p < 0x80 && *p != WC_C_HZ_TILDA; p++) |
20 | 518 | ; |
21 | 274 | if (p == ep) |
22 | 35 | return is; |
23 | 239 | os = Strnew_size(is->length); |
24 | 239 | if (p > sp) |
25 | 24 | Strcat_charp_n(os, is->ptr, (int)(p - sp)); |
26 | | |
27 | 969k | for (; p < ep; p++) { |
28 | 969k | switch (state) { |
29 | 755k | case WC_HZ_NOSTATE: |
30 | 755k | if (*p == WC_C_HZ_TILDA) |
31 | 2.80k | state = WC_HZ_TILDA; |
32 | 752k | else if (WC_ISO_MAP[*p] == WC_ISO_MAP_GR) |
33 | 157k | state = WC_HZ_MBYTE1_GR; /* GB 2312 ? */ |
34 | 594k | else if (*p & 0x80) |
35 | 477k | wtf_push_unknown(os, p, 1); |
36 | 117k | else |
37 | 117k | Strcat_char(os, (char)*p); |
38 | 755k | break; |
39 | 3.03k | case WC_HZ_TILDA: |
40 | 3.03k | if (*p == WC_C_HZ_SI) { |
41 | 643 | state = WC_HZ_MBYTE; |
42 | 643 | break; |
43 | 2.39k | } else if (*p == WC_C_HZ_TILDA) |
44 | 507 | Strcat_char(os, (char)*p); |
45 | 1.88k | else if (*p == '\n') |
46 | 239 | break; |
47 | 1.64k | else |
48 | 1.64k | wtf_push_unknown(os, p-1, 2); |
49 | 2.15k | state = WC_HZ_NOSTATE; |
50 | 2.15k | break; |
51 | 1.50k | case WC_HZ_TILDA_MB: |
52 | 1.50k | if (*p == WC_C_HZ_SO || *p == '\n') { |
53 | 563 | state = WC_HZ_NOSTATE; |
54 | 563 | break; |
55 | 563 | } |
56 | 939 | else if (WC_ISO_MAP[*p & 0x7f] == WC_ISO_MAP_GL) |
57 | 692 | wtf_push(os, WC_CCS_GB_2312, ((wc_uint32)*(p-1) << 8) | *p); |
58 | 247 | else |
59 | 247 | wtf_push_unknown(os, p-1, 2); |
60 | 939 | state = WC_HZ_MBYTE; |
61 | 939 | break; |
62 | 34.9k | case WC_HZ_MBYTE: |
63 | 34.9k | if (*p == WC_C_HZ_TILDA) |
64 | 1.50k | state = WC_HZ_TILDA_MB; |
65 | 33.4k | else if (WC_ISO_MAP[*p & 0x7f] == WC_ISO_MAP_GL) |
66 | 16.8k | state = WC_HZ_MBYTE1; |
67 | 16.5k | else |
68 | 16.5k | wtf_push_unknown(os, p, 1); |
69 | 34.9k | break; |
70 | 16.8k | case WC_HZ_MBYTE1: |
71 | 16.8k | if (WC_ISO_MAP[*p & 0x7f] == WC_ISO_MAP_GL) |
72 | 14.9k | wtf_push(os, WC_CCS_GB_2312, ((wc_uint32)*(p-1) << 8) | *p); |
73 | 1.93k | else |
74 | 1.93k | wtf_push_unknown(os, p-1, 2); |
75 | 16.8k | state = WC_HZ_MBYTE; |
76 | 16.8k | break; |
77 | 157k | case WC_HZ_MBYTE1_GR: |
78 | 157k | if (WC_ISO_MAP[*p] == WC_ISO_MAP_GR) |
79 | 56.9k | wtf_push(os, WC_CCS_GB_2312, ((wc_uint32)*(p-1) << 8) | *p); |
80 | 100k | else |
81 | 100k | wtf_push_unknown(os, p-1, 2); |
82 | 157k | state = WC_HZ_NOSTATE; |
83 | 157k | break; |
84 | 969k | } |
85 | 969k | } |
86 | 239 | switch (state) { |
87 | 9 | case WC_HZ_TILDA: |
88 | 14 | case WC_HZ_TILDA_MB: |
89 | 26 | case WC_HZ_MBYTE1: |
90 | 44 | case WC_HZ_MBYTE1_GR: |
91 | 44 | wtf_push_unknown(os, p-1, 1); |
92 | 44 | break; |
93 | 239 | } |
94 | 239 | return os; |
95 | 239 | } |
96 | | |
97 | | void |
98 | | wc_push_to_hz(Str os, wc_wchar_t cc, wc_status *st) |
99 | 24.7M | { |
100 | 45.3M | while (1) { |
101 | 45.3M | switch (cc.ccs) { |
102 | 3.49M | case WC_CCS_US_ASCII: |
103 | 3.49M | if (st->gl) { |
104 | 2.58M | Strcat_char(os, WC_C_HZ_TILDA); |
105 | 2.58M | Strcat_char(os, WC_C_HZ_SO); |
106 | 2.58M | st->gl = 0; |
107 | 2.58M | } |
108 | 3.49M | if ((char)cc.code == WC_C_HZ_TILDA) |
109 | 2.48M | Strcat_char(os, WC_C_HZ_TILDA); |
110 | 3.49M | Strcat_char(os, (char)cc.code); |
111 | 3.49M | return; |
112 | 12.3M | case WC_CCS_GB_2312: |
113 | 12.3M | if (! st->gl) { |
114 | 10.7M | Strcat_char(os, WC_C_HZ_TILDA); |
115 | 10.7M | Strcat_char(os, WC_C_HZ_SI); |
116 | 10.7M | st->gl = 1; |
117 | 10.7M | } |
118 | 12.3M | Strcat_char(os, (char)((cc.code >> 8) & 0x7f)); |
119 | 12.3M | Strcat_char(os, (char)(cc.code & 0x7f)); |
120 | 12.3M | return; |
121 | 8.24M | case WC_CCS_UNKNOWN_W: |
122 | 8.24M | if (WcOption.no_replace) |
123 | 0 | return; |
124 | 8.24M | if (st->gl) { |
125 | 8.09M | Strcat_char(os, WC_C_HZ_TILDA); |
126 | 8.09M | Strcat_char(os, WC_C_HZ_SO); |
127 | 8.09M | st->gl = 0; |
128 | 8.09M | } |
129 | 8.24M | Strcat_charp(os, WC_REPLACE_W); |
130 | 8.24M | return; |
131 | 679k | case WC_CCS_UNKNOWN: |
132 | 679k | if (WcOption.no_replace) |
133 | 0 | return; |
134 | 679k | if (st->gl) { |
135 | 38.5k | Strcat_char(os, WC_C_HZ_TILDA); |
136 | 38.5k | Strcat_char(os, WC_C_HZ_SO); |
137 | 38.5k | st->gl = 0; |
138 | 38.5k | } |
139 | 679k | Strcat_charp(os, WC_REPLACE); |
140 | 679k | return; |
141 | 20.6M | default: |
142 | 20.6M | #ifdef USE_UNICODE |
143 | 20.6M | if (WcOption.ucs_conv) |
144 | 20.6M | 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 | 20.6M | continue; |
149 | 45.3M | } |
150 | 45.3M | } |
151 | 24.7M | } |
152 | | |
153 | | void |
154 | | wc_push_to_hz_end(Str os, wc_status *st) |
155 | 457 | { |
156 | 457 | if (st->gl) { |
157 | 230 | Strcat_char(os, WC_C_HZ_TILDA); |
158 | | Strcat_char(os, WC_C_HZ_SO); |
159 | 230 | st->gl = 0; |
160 | 230 | } |
161 | 457 | } |