Commit 6817ca585a44eec2c50961ba7c714b419774da36
1 parent
69848546
Bidirectional transcoding for win, mac, pdf, utf8, utf16
Showing
3 changed files
with
490 additions
and
257 deletions
libqpdf/QUtil.cc
| ... | ... | @@ -66,6 +66,171 @@ static unsigned short pdf_doc_to_unicode[] = { |
| 66 | 66 | 0xfffd, // 0x9f UNDEFINED |
| 67 | 67 | 0x20ac, // 0xa0 EURO SIGN |
| 68 | 68 | }; |
| 69 | +static unsigned short win_ansi_to_unicode[] = { | |
| 70 | + 0x20ac, // 0x80 | |
| 71 | + 0xfffd, // 0x81 | |
| 72 | + 0x201a, // 0x82 | |
| 73 | + 0x0192, // 0x83 | |
| 74 | + 0x201e, // 0x84 | |
| 75 | + 0x2026, // 0x85 | |
| 76 | + 0x2020, // 0x86 | |
| 77 | + 0x2021, // 0x87 | |
| 78 | + 0x02c6, // 0x88 | |
| 79 | + 0x2030, // 0x89 | |
| 80 | + 0x0160, // 0x8a | |
| 81 | + 0x2039, // 0x8b | |
| 82 | + 0x0152, // 0x8c | |
| 83 | + 0xfffd, // 0x8d | |
| 84 | + 0x017d, // 0x8e | |
| 85 | + 0xfffd, // 0x8f | |
| 86 | + 0xfffd, // 0x90 | |
| 87 | + 0x2018, // 0x91 | |
| 88 | + 0x2019, // 0x92 | |
| 89 | + 0x201c, // 0x93 | |
| 90 | + 0x201d, // 0x94 | |
| 91 | + 0x2022, // 0x95 | |
| 92 | + 0x2013, // 0x96 | |
| 93 | + 0x2014, // 0x97 | |
| 94 | + 0x0303, // 0x98 | |
| 95 | + 0x2122, // 0x99 | |
| 96 | + 0x0161, // 0x9a | |
| 97 | + 0x203a, // 0x9b | |
| 98 | + 0x0153, // 0x9c | |
| 99 | + 0xfffd, // 0x9d | |
| 100 | + 0x017e, // 0x9e | |
| 101 | + 0x0178, // 0x9f | |
| 102 | + 0x00a0, // 0xa0 | |
| 103 | +}; | |
| 104 | +static unsigned short mac_roman_to_unicode[] = { | |
| 105 | + 0x00c4, // 0x80 | |
| 106 | + 0x00c5, // 0x81 | |
| 107 | + 0x00c7, // 0x82 | |
| 108 | + 0x00c9, // 0x83 | |
| 109 | + 0x00d1, // 0x84 | |
| 110 | + 0x00d6, // 0x85 | |
| 111 | + 0x00dc, // 0x86 | |
| 112 | + 0x00e1, // 0x87 | |
| 113 | + 0x00e0, // 0x88 | |
| 114 | + 0x00e2, // 0x89 | |
| 115 | + 0x00e4, // 0x8a | |
| 116 | + 0x00e3, // 0x8b | |
| 117 | + 0x00e5, // 0x8c | |
| 118 | + 0x00e7, // 0x8d | |
| 119 | + 0x00e9, // 0x8e | |
| 120 | + 0x00e8, // 0x8f | |
| 121 | + 0x00ea, // 0x90 | |
| 122 | + 0x00eb, // 0x91 | |
| 123 | + 0x00ed, // 0x92 | |
| 124 | + 0x00ec, // 0x93 | |
| 125 | + 0x00ee, // 0x94 | |
| 126 | + 0x00ef, // 0x95 | |
| 127 | + 0x00f1, // 0x96 | |
| 128 | + 0x00f3, // 0x97 | |
| 129 | + 0x00f2, // 0x98 | |
| 130 | + 0x00f4, // 0x99 | |
| 131 | + 0x00f6, // 0x9a | |
| 132 | + 0x00f5, // 0x9b | |
| 133 | + 0x00fa, // 0x9c | |
| 134 | + 0x00f9, // 0x9d | |
| 135 | + 0x00fb, // 0x9e | |
| 136 | + 0x00fc, // 0x9f | |
| 137 | + 0x2020, // 0xa0 | |
| 138 | + 0x00b0, // 0xa1 | |
| 139 | + 0x00a2, // 0xa2 | |
| 140 | + 0x00a3, // 0xa3 | |
| 141 | + 0x00a7, // 0xa4 | |
| 142 | + 0x2022, // 0xa5 | |
| 143 | + 0x00b6, // 0xa6 | |
| 144 | + 0x00df, // 0xa7 | |
| 145 | + 0x00ae, // 0xa8 | |
| 146 | + 0x00a9, // 0xa9 | |
| 147 | + 0x2122, // 0xaa | |
| 148 | + 0x0301, // 0xab | |
| 149 | + 0x0308, // 0xac | |
| 150 | + 0xfffd, // 0xad | |
| 151 | + 0x00c6, // 0xae | |
| 152 | + 0x00d8, // 0xaf | |
| 153 | + 0xfffd, // 0xb0 | |
| 154 | + 0x00b1, // 0xb1 | |
| 155 | + 0xfffd, // 0xb2 | |
| 156 | + 0xfffd, // 0xb3 | |
| 157 | + 0x00a5, // 0xb4 | |
| 158 | + 0x03bc, // 0xb5 | |
| 159 | + 0xfffd, // 0xb6 | |
| 160 | + 0xfffd, // 0xb7 | |
| 161 | + 0xfffd, // 0xb8 | |
| 162 | + 0xfffd, // 0xb9 | |
| 163 | + 0xfffd, // 0xba | |
| 164 | + 0x1d43, // 0xbb | |
| 165 | + 0x1d52, // 0xbc | |
| 166 | + 0xfffd, // 0xbd | |
| 167 | + 0x00e6, // 0xbe | |
| 168 | + 0x00f8, // 0xbf | |
| 169 | + 0x00bf, // 0xc0 | |
| 170 | + 0x00a1, // 0xc1 | |
| 171 | + 0x00ac, // 0xc2 | |
| 172 | + 0xfffd, // 0xc3 | |
| 173 | + 0x0192, // 0xc4 | |
| 174 | + 0xfffd, // 0xc5 | |
| 175 | + 0xfffd, // 0xc6 | |
| 176 | + 0x00ab, // 0xc7 | |
| 177 | + 0x00bb, // 0xc8 | |
| 178 | + 0x2026, // 0xc9 | |
| 179 | + 0xfffd, // 0xca | |
| 180 | + 0x00c0, // 0xcb | |
| 181 | + 0x00c3, // 0xcc | |
| 182 | + 0x00d5, // 0xcd | |
| 183 | + 0x0152, // 0xce | |
| 184 | + 0x0153, // 0xcf | |
| 185 | + 0x2013, // 0xd0 | |
| 186 | + 0x2014, // 0xd1 | |
| 187 | + 0x201c, // 0xd2 | |
| 188 | + 0x201d, // 0xd3 | |
| 189 | + 0x2018, // 0xd4 | |
| 190 | + 0x2019, // 0xd5 | |
| 191 | + 0x00f7, // 0xd6 | |
| 192 | + 0xfffd, // 0xd7 | |
| 193 | + 0x00ff, // 0xd8 | |
| 194 | + 0x0178, // 0xd9 | |
| 195 | + 0x2044, // 0xda | |
| 196 | + 0x00a4, // 0xdb | |
| 197 | + 0x2039, // 0xdc | |
| 198 | + 0x203a, // 0xdd | |
| 199 | + 0xfb01, // 0xde | |
| 200 | + 0xfb02, // 0xdf | |
| 201 | + 0x2021, // 0xe0 | |
| 202 | + 0x00b7, // 0xe1 | |
| 203 | + 0x201a, // 0xe2 | |
| 204 | + 0x201e, // 0xe3 | |
| 205 | + 0x2030, // 0xe4 | |
| 206 | + 0x00c2, // 0xe5 | |
| 207 | + 0x00ca, // 0xe6 | |
| 208 | + 0x00c1, // 0xe7 | |
| 209 | + 0x00cb, // 0xe8 | |
| 210 | + 0x00c8, // 0xe9 | |
| 211 | + 0x00cd, // 0xea | |
| 212 | + 0x00ce, // 0xeb | |
| 213 | + 0x00cf, // 0xec | |
| 214 | + 0x00cc, // 0xed | |
| 215 | + 0x00d3, // 0xee | |
| 216 | + 0x00d4, // 0xef | |
| 217 | + 0xfffd, // 0xf0 | |
| 218 | + 0x00d2, // 0xf1 | |
| 219 | + 0x00da, // 0xf2 | |
| 220 | + 0x00db, // 0xf3 | |
| 221 | + 0x00d9, // 0xf4 | |
| 222 | + 0x0131, // 0xf5 | |
| 223 | + 0x02c6, // 0xf6 | |
| 224 | + 0x0303, // 0xf7 | |
| 225 | + 0x0304, // 0xf8 | |
| 226 | + 0x0306, // 0xf9 | |
| 227 | + 0x0307, // 0xfa | |
| 228 | + 0x030a, // 0xfb | |
| 229 | + 0x0327, // 0xfc | |
| 230 | + 0x030b, // 0xfd | |
| 231 | + 0x0328, // 0xfe | |
| 232 | + 0x02c7, // 0xff | |
| 233 | +}; | |
| 69 | 234 | |
| 70 | 235 | std::string |
| 71 | 236 | QUtil::int_to_string(long long num, int length) |
| ... | ... | @@ -946,23 +1111,17 @@ encode_winansi(unsigned long codepoint) |
| 946 | 1111 | case 0x20ac: |
| 947 | 1112 | ch = 0x80; |
| 948 | 1113 | break; |
| 949 | - case 0x152: | |
| 950 | - ch = 0x8c; | |
| 951 | - break; | |
| 952 | - case 0x160: | |
| 953 | - ch = 0x8a; | |
| 954 | - break; | |
| 955 | - case 0x178: | |
| 956 | - ch = 0x9f; | |
| 1114 | + case 0x201a: | |
| 1115 | + ch = 0x82; | |
| 957 | 1116 | break; |
| 958 | - case 0x17d: | |
| 959 | - ch = 0x8e; | |
| 1117 | + case 0x192: | |
| 1118 | + ch = 0x83; | |
| 960 | 1119 | break; |
| 961 | - case 0x2022: | |
| 962 | - ch = 0x95; | |
| 1120 | + case 0x201e: | |
| 1121 | + ch = 0x84; | |
| 963 | 1122 | break; |
| 964 | - case 0x2c6: | |
| 965 | - ch = 0x88; | |
| 1123 | + case 0x2026: | |
| 1124 | + ch = 0x85; | |
| 966 | 1125 | break; |
| 967 | 1126 | case 0x2020: |
| 968 | 1127 | ch = 0x86; |
| ... | ... | @@ -970,32 +1129,29 @@ encode_winansi(unsigned long codepoint) |
| 970 | 1129 | case 0x2021: |
| 971 | 1130 | ch = 0x87; |
| 972 | 1131 | break; |
| 973 | - case 0x2026: | |
| 974 | - ch = 0x85; | |
| 975 | - break; | |
| 976 | - case 0x2014: | |
| 977 | - ch = 0x97; | |
| 1132 | + case 0x2c6: | |
| 1133 | + ch = 0x88; | |
| 978 | 1134 | break; |
| 979 | - case 0x2013: | |
| 980 | - ch = 0x96; | |
| 1135 | + case 0x2030: | |
| 1136 | + ch = 0x89; | |
| 981 | 1137 | break; |
| 982 | - case 0x192: | |
| 983 | - ch = 0x83; | |
| 1138 | + case 0x160: | |
| 1139 | + ch = 0x8a; | |
| 984 | 1140 | break; |
| 985 | 1141 | case 0x2039: |
| 986 | 1142 | ch = 0x8b; |
| 987 | 1143 | break; |
| 988 | - case 0x203a: | |
| 989 | - ch = 0x9b; | |
| 1144 | + case 0x152: | |
| 1145 | + ch = 0x8c; | |
| 990 | 1146 | break; |
| 991 | - case 0x153: | |
| 992 | - ch = 0x9c; | |
| 1147 | + case 0x17d: | |
| 1148 | + ch = 0x8e; | |
| 993 | 1149 | break; |
| 994 | - case 0x2030: | |
| 995 | - ch = 0x89; | |
| 1150 | + case 0x2018: | |
| 1151 | + ch = 0x91; | |
| 996 | 1152 | break; |
| 997 | - case 0x201e: | |
| 998 | - ch = 0x84; | |
| 1153 | + case 0x2019: | |
| 1154 | + ch = 0x92; | |
| 999 | 1155 | break; |
| 1000 | 1156 | case 0x201c: |
| 1001 | 1157 | ch = 0x93; |
| ... | ... | @@ -1003,17 +1159,14 @@ encode_winansi(unsigned long codepoint) |
| 1003 | 1159 | case 0x201d: |
| 1004 | 1160 | ch = 0x94; |
| 1005 | 1161 | break; |
| 1006 | - case 0x2018: | |
| 1007 | - ch = 0x91; | |
| 1008 | - break; | |
| 1009 | - case 0x2019: | |
| 1010 | - ch = 0x92; | |
| 1162 | + case 0x2022: | |
| 1163 | + ch = 0x95; | |
| 1011 | 1164 | break; |
| 1012 | - case 0x201a: | |
| 1013 | - ch = 0x82; | |
| 1165 | + case 0x2013: | |
| 1166 | + ch = 0x96; | |
| 1014 | 1167 | break; |
| 1015 | - case 0x161: | |
| 1016 | - ch = 0x9a; | |
| 1168 | + case 0x2014: | |
| 1169 | + ch = 0x97; | |
| 1017 | 1170 | break; |
| 1018 | 1171 | case 0x303: |
| 1019 | 1172 | ch = 0x98; |
| ... | ... | @@ -1021,9 +1174,24 @@ encode_winansi(unsigned long codepoint) |
| 1021 | 1174 | case 0x2122: |
| 1022 | 1175 | ch = 0x99; |
| 1023 | 1176 | break; |
| 1177 | + case 0x161: | |
| 1178 | + ch = 0x9a; | |
| 1179 | + break; | |
| 1180 | + case 0x203a: | |
| 1181 | + ch = 0x9b; | |
| 1182 | + break; | |
| 1183 | + case 0x153: | |
| 1184 | + ch = 0x9c; | |
| 1185 | + break; | |
| 1024 | 1186 | case 0x17e: |
| 1025 | 1187 | ch = 0x9e; |
| 1026 | 1188 | break; |
| 1189 | + case 0x178: | |
| 1190 | + ch = 0x9f; | |
| 1191 | + break; | |
| 1192 | + case 0xa0: | |
| 1193 | + ch = 0xa0; | |
| 1194 | + break; | |
| 1027 | 1195 | default: |
| 1028 | 1196 | break; |
| 1029 | 1197 | } |
| ... | ... | @@ -1038,290 +1206,209 @@ encode_macroman(unsigned long codepoint) |
| 1038 | 1206 | unsigned char ch = '\0'; |
| 1039 | 1207 | switch (codepoint) |
| 1040 | 1208 | { |
| 1041 | - case 0xc6: | |
| 1042 | - ch = 0xae; | |
| 1043 | - break; | |
| 1044 | - case 0xc1: | |
| 1045 | - ch = 0xe7; | |
| 1046 | - break; | |
| 1047 | - case 0xc2: | |
| 1048 | - ch = 0xe5; | |
| 1049 | - break; | |
| 1050 | 1209 | case 0xc4: |
| 1051 | 1210 | ch = 0x80; |
| 1052 | 1211 | break; |
| 1053 | - case 0xc0: | |
| 1054 | - ch = 0xcb; | |
| 1055 | - break; | |
| 1056 | 1212 | case 0xc5: |
| 1057 | 1213 | ch = 0x81; |
| 1058 | 1214 | break; |
| 1059 | - case 0xc3: | |
| 1060 | - ch = 0xcc; | |
| 1061 | - break; | |
| 1062 | 1215 | case 0xc7: |
| 1063 | 1216 | ch = 0x82; |
| 1064 | 1217 | break; |
| 1065 | 1218 | case 0xc9: |
| 1066 | 1219 | ch = 0x83; |
| 1067 | 1220 | break; |
| 1068 | - case 0xca: | |
| 1069 | - ch = 0xe6; | |
| 1070 | - break; | |
| 1071 | - case 0xcb: | |
| 1072 | - ch = 0xe8; | |
| 1073 | - break; | |
| 1074 | - case 0xc8: | |
| 1075 | - ch = 0xe9; | |
| 1076 | - break; | |
| 1077 | - case 0xcd: | |
| 1078 | - ch = 0xea; | |
| 1079 | - break; | |
| 1080 | - case 0xce: | |
| 1081 | - ch = 0xeb; | |
| 1082 | - break; | |
| 1083 | - case 0xcf: | |
| 1084 | - ch = 0xec; | |
| 1085 | - break; | |
| 1086 | - case 0xcc: | |
| 1087 | - ch = 0xed; | |
| 1088 | - break; | |
| 1089 | 1221 | case 0xd1: |
| 1090 | 1222 | ch = 0x84; |
| 1091 | 1223 | break; |
| 1092 | - case 0x152: | |
| 1093 | - ch = 0xce; | |
| 1094 | - break; | |
| 1095 | - case 0xd3: | |
| 1096 | - ch = 0xee; | |
| 1097 | - break; | |
| 1098 | - case 0xd4: | |
| 1099 | - ch = 0xef; | |
| 1100 | - break; | |
| 1101 | 1224 | case 0xd6: |
| 1102 | 1225 | ch = 0x85; |
| 1103 | 1226 | break; |
| 1104 | - case 0xd2: | |
| 1105 | - ch = 0xf1; | |
| 1106 | - break; | |
| 1107 | - case 0xd8: | |
| 1108 | - ch = 0xaf; | |
| 1109 | - break; | |
| 1110 | - case 0xd5: | |
| 1111 | - ch = 0xcd; | |
| 1112 | - break; | |
| 1113 | - case 0xda: | |
| 1114 | - ch = 0xf2; | |
| 1115 | - break; | |
| 1116 | - case 0xdb: | |
| 1117 | - ch = 0xf3; | |
| 1118 | - break; | |
| 1119 | 1227 | case 0xdc: |
| 1120 | 1228 | ch = 0x86; |
| 1121 | 1229 | break; |
| 1122 | - case 0xd9: | |
| 1123 | - ch = 0xf4; | |
| 1124 | - break; | |
| 1125 | - case 0x178: | |
| 1126 | - ch = 0xd9; | |
| 1127 | - break; | |
| 1128 | 1230 | case 0xe1: |
| 1129 | 1231 | ch = 0x87; |
| 1130 | 1232 | break; |
| 1233 | + case 0xe0: | |
| 1234 | + ch = 0x88; | |
| 1235 | + break; | |
| 1131 | 1236 | case 0xe2: |
| 1132 | 1237 | ch = 0x89; |
| 1133 | 1238 | break; |
| 1134 | - case 0x301: | |
| 1135 | - ch = 0xab; | |
| 1136 | - break; | |
| 1137 | 1239 | case 0xe4: |
| 1138 | 1240 | ch = 0x8a; |
| 1139 | 1241 | break; |
| 1140 | - case 0xe6: | |
| 1141 | - ch = 0xbe; | |
| 1142 | - break; | |
| 1143 | - case 0xe0: | |
| 1144 | - ch = 0x88; | |
| 1145 | - break; | |
| 1146 | - case 0xe5: | |
| 1147 | - ch = 0x8c; | |
| 1148 | - break; | |
| 1149 | 1242 | case 0xe3: |
| 1150 | 1243 | ch = 0x8b; |
| 1151 | 1244 | break; |
| 1152 | - case 0x306: | |
| 1153 | - ch = 0xf9; | |
| 1154 | - break; | |
| 1155 | - case 0x2022: | |
| 1156 | - ch = 0xa5; | |
| 1157 | - break; | |
| 1158 | - case 0x2c7: | |
| 1159 | - ch = 0xff; | |
| 1245 | + case 0xe5: | |
| 1246 | + ch = 0x8c; | |
| 1160 | 1247 | break; |
| 1161 | 1248 | case 0xe7: |
| 1162 | 1249 | ch = 0x8d; |
| 1163 | 1250 | break; |
| 1164 | - case 0x327: | |
| 1165 | - ch = 0xfc; | |
| 1251 | + case 0xe9: | |
| 1252 | + ch = 0x8e; | |
| 1166 | 1253 | break; |
| 1167 | - case 0xa2: | |
| 1168 | - ch = 0xa2; | |
| 1254 | + case 0xe8: | |
| 1255 | + ch = 0x8f; | |
| 1169 | 1256 | break; |
| 1170 | - case 0x2c6: | |
| 1171 | - ch = 0xf6; | |
| 1257 | + case 0xea: | |
| 1258 | + ch = 0x90; | |
| 1172 | 1259 | break; |
| 1173 | - case 0xa9: | |
| 1174 | - ch = 0xa9; | |
| 1260 | + case 0xeb: | |
| 1261 | + ch = 0x91; | |
| 1175 | 1262 | break; |
| 1176 | - case 0xa4: | |
| 1177 | - ch = 0xdb; | |
| 1263 | + case 0xed: | |
| 1264 | + ch = 0x92; | |
| 1178 | 1265 | break; |
| 1179 | - case 0x2020: | |
| 1180 | - ch = 0xa0; | |
| 1266 | + case 0xec: | |
| 1267 | + ch = 0x93; | |
| 1181 | 1268 | break; |
| 1182 | - case 0x2021: | |
| 1183 | - ch = 0xe0; | |
| 1269 | + case 0xee: | |
| 1270 | + ch = 0x94; | |
| 1184 | 1271 | break; |
| 1185 | - case 0xb0: | |
| 1186 | - ch = 0xa1; | |
| 1272 | + case 0xef: | |
| 1273 | + ch = 0x95; | |
| 1187 | 1274 | break; |
| 1188 | - case 0x308: | |
| 1189 | - ch = 0xac; | |
| 1275 | + case 0xf1: | |
| 1276 | + ch = 0x96; | |
| 1190 | 1277 | break; |
| 1191 | - case 0xf7: | |
| 1192 | - ch = 0xd6; | |
| 1278 | + case 0xf3: | |
| 1279 | + ch = 0x97; | |
| 1193 | 1280 | break; |
| 1194 | - case 0x307: | |
| 1195 | - ch = 0xfa; | |
| 1281 | + case 0xf2: | |
| 1282 | + ch = 0x98; | |
| 1196 | 1283 | break; |
| 1197 | - case 0x131: | |
| 1198 | - ch = 0xf5; | |
| 1284 | + case 0xf4: | |
| 1285 | + ch = 0x99; | |
| 1199 | 1286 | break; |
| 1200 | - case 0xe9: | |
| 1201 | - ch = 0x8e; | |
| 1287 | + case 0xf6: | |
| 1288 | + ch = 0x9a; | |
| 1202 | 1289 | break; |
| 1203 | - case 0xea: | |
| 1204 | - ch = 0x90; | |
| 1290 | + case 0xf5: | |
| 1291 | + ch = 0x9b; | |
| 1205 | 1292 | break; |
| 1206 | - case 0xeb: | |
| 1207 | - ch = 0x91; | |
| 1293 | + case 0xfa: | |
| 1294 | + ch = 0x9c; | |
| 1208 | 1295 | break; |
| 1209 | - case 0xe8: | |
| 1210 | - ch = 0x8f; | |
| 1296 | + case 0xf9: | |
| 1297 | + ch = 0x9d; | |
| 1211 | 1298 | break; |
| 1212 | - case 0x2026: | |
| 1213 | - ch = 0xc9; | |
| 1299 | + case 0xfb: | |
| 1300 | + ch = 0x9e; | |
| 1214 | 1301 | break; |
| 1215 | - case 0x2014: | |
| 1216 | - ch = 0xd1; | |
| 1302 | + case 0xfc: | |
| 1303 | + ch = 0x9f; | |
| 1217 | 1304 | break; |
| 1218 | - case 0x2013: | |
| 1219 | - ch = 0xd0; | |
| 1305 | + case 0x2020: | |
| 1306 | + ch = 0xa0; | |
| 1220 | 1307 | break; |
| 1221 | - case 0xa1: | |
| 1222 | - ch = 0xc1; | |
| 1308 | + case 0xb0: | |
| 1309 | + ch = 0xa1; | |
| 1223 | 1310 | break; |
| 1224 | - case 0xfb01: | |
| 1225 | - ch = 0xde; | |
| 1311 | + case 0xa2: | |
| 1312 | + ch = 0xa2; | |
| 1226 | 1313 | break; |
| 1227 | - case 0xfb02: | |
| 1228 | - ch = 0xdf; | |
| 1314 | + case 0xa3: | |
| 1315 | + ch = 0xa3; | |
| 1229 | 1316 | break; |
| 1230 | - case 0x192: | |
| 1231 | - ch = 0xc4; | |
| 1317 | + case 0xa7: | |
| 1318 | + ch = 0xa4; | |
| 1232 | 1319 | break; |
| 1233 | - case 0x2044: | |
| 1234 | - ch = 0xda; | |
| 1320 | + case 0x2022: | |
| 1321 | + ch = 0xa5; | |
| 1322 | + break; | |
| 1323 | + case 0xb6: | |
| 1324 | + ch = 0xa6; | |
| 1235 | 1325 | break; |
| 1236 | 1326 | case 0xdf: |
| 1237 | 1327 | ch = 0xa7; |
| 1238 | 1328 | break; |
| 1239 | - case 0xab: | |
| 1240 | - ch = 0xc7; | |
| 1241 | - break; | |
| 1242 | - case 0xbb: | |
| 1243 | - ch = 0xc8; | |
| 1244 | - break; | |
| 1245 | - case 0x2039: | |
| 1246 | - ch = 0xdc; | |
| 1329 | + case 0xae: | |
| 1330 | + ch = 0xa8; | |
| 1247 | 1331 | break; |
| 1248 | - case 0x203a: | |
| 1249 | - ch = 0xdd; | |
| 1332 | + case 0xa9: | |
| 1333 | + ch = 0xa9; | |
| 1250 | 1334 | break; |
| 1251 | - case 0x30b: | |
| 1252 | - ch = 0xfd; | |
| 1335 | + case 0x2122: | |
| 1336 | + ch = 0xaa; | |
| 1253 | 1337 | break; |
| 1254 | - case 0xed: | |
| 1255 | - ch = 0x92; | |
| 1338 | + case 0x301: | |
| 1339 | + ch = 0xab; | |
| 1256 | 1340 | break; |
| 1257 | - case 0xee: | |
| 1258 | - ch = 0x94; | |
| 1341 | + case 0x308: | |
| 1342 | + ch = 0xac; | |
| 1259 | 1343 | break; |
| 1260 | - case 0xef: | |
| 1261 | - ch = 0x95; | |
| 1344 | + case 0xc6: | |
| 1345 | + ch = 0xae; | |
| 1262 | 1346 | break; |
| 1263 | - case 0xec: | |
| 1264 | - ch = 0x93; | |
| 1347 | + case 0xd8: | |
| 1348 | + ch = 0xaf; | |
| 1265 | 1349 | break; |
| 1266 | - case 0xac: | |
| 1267 | - ch = 0xc2; | |
| 1350 | + case 0xb1: | |
| 1351 | + ch = 0xb1; | |
| 1268 | 1352 | break; |
| 1269 | - case 0x304: | |
| 1270 | - ch = 0xf8; | |
| 1353 | + case 0xa5: | |
| 1354 | + ch = 0xb4; | |
| 1271 | 1355 | break; |
| 1272 | 1356 | case 0x3bc: |
| 1273 | 1357 | ch = 0xb5; |
| 1274 | 1358 | break; |
| 1275 | - case 0xf1: | |
| 1276 | - ch = 0x96; | |
| 1359 | + case 0x1d43: | |
| 1360 | + ch = 0xbb; | |
| 1277 | 1361 | break; |
| 1278 | - case 0xf3: | |
| 1279 | - ch = 0x97; | |
| 1362 | + case 0x1d52: | |
| 1363 | + ch = 0xbc; | |
| 1280 | 1364 | break; |
| 1281 | - case 0xf4: | |
| 1282 | - ch = 0x99; | |
| 1365 | + case 0xe6: | |
| 1366 | + ch = 0xbe; | |
| 1283 | 1367 | break; |
| 1284 | - case 0xf6: | |
| 1285 | - ch = 0x9a; | |
| 1368 | + case 0xf8: | |
| 1369 | + ch = 0xbf; | |
| 1286 | 1370 | break; |
| 1287 | - case 0x153: | |
| 1288 | - ch = 0xcf; | |
| 1371 | + case 0xbf: | |
| 1372 | + ch = 0xc0; | |
| 1289 | 1373 | break; |
| 1290 | - case 0x328: | |
| 1291 | - ch = 0xfe; | |
| 1374 | + case 0xa1: | |
| 1375 | + ch = 0xc1; | |
| 1292 | 1376 | break; |
| 1293 | - case 0xf2: | |
| 1294 | - ch = 0x98; | |
| 1377 | + case 0xac: | |
| 1378 | + ch = 0xc2; | |
| 1295 | 1379 | break; |
| 1296 | - case 0x1d43: | |
| 1297 | - ch = 0xbb; | |
| 1380 | + case 0x192: | |
| 1381 | + ch = 0xc4; | |
| 1298 | 1382 | break; |
| 1299 | - case 0x1d52: | |
| 1300 | - ch = 0xbc; | |
| 1383 | + case 0xab: | |
| 1384 | + ch = 0xc7; | |
| 1301 | 1385 | break; |
| 1302 | - case 0xf8: | |
| 1303 | - ch = 0xbf; | |
| 1386 | + case 0xbb: | |
| 1387 | + ch = 0xc8; | |
| 1304 | 1388 | break; |
| 1305 | - case 0xf5: | |
| 1306 | - ch = 0x9b; | |
| 1389 | + case 0x2026: | |
| 1390 | + ch = 0xc9; | |
| 1307 | 1391 | break; |
| 1308 | - case 0xb6: | |
| 1309 | - ch = 0xa6; | |
| 1392 | + case 0xc0: | |
| 1393 | + ch = 0xcb; | |
| 1310 | 1394 | break; |
| 1311 | - case 0xb7: | |
| 1312 | - ch = 0xe1; | |
| 1395 | + case 0xc3: | |
| 1396 | + ch = 0xcc; | |
| 1313 | 1397 | break; |
| 1314 | - case 0x2030: | |
| 1315 | - ch = 0xe4; | |
| 1398 | + case 0xd5: | |
| 1399 | + ch = 0xcd; | |
| 1316 | 1400 | break; |
| 1317 | - case 0xb1: | |
| 1318 | - ch = 0xb1; | |
| 1401 | + case 0x152: | |
| 1402 | + ch = 0xce; | |
| 1319 | 1403 | break; |
| 1320 | - case 0xbf: | |
| 1321 | - ch = 0xc0; | |
| 1404 | + case 0x153: | |
| 1405 | + ch = 0xcf; | |
| 1322 | 1406 | break; |
| 1323 | - case 0x201e: | |
| 1324 | - ch = 0xe3; | |
| 1407 | + case 0x2013: | |
| 1408 | + ch = 0xd0; | |
| 1409 | + break; | |
| 1410 | + case 0x2014: | |
| 1411 | + ch = 0xd1; | |
| 1325 | 1412 | break; |
| 1326 | 1413 | case 0x201c: |
| 1327 | 1414 | ch = 0xd2; |
| ... | ... | @@ -1335,44 +1422,125 @@ encode_macroman(unsigned long codepoint) |
| 1335 | 1422 | case 0x2019: |
| 1336 | 1423 | ch = 0xd5; |
| 1337 | 1424 | break; |
| 1425 | + case 0xf7: | |
| 1426 | + ch = 0xd6; | |
| 1427 | + break; | |
| 1428 | + case 0xff: | |
| 1429 | + ch = 0xd8; | |
| 1430 | + break; | |
| 1431 | + case 0x178: | |
| 1432 | + ch = 0xd9; | |
| 1433 | + break; | |
| 1434 | + case 0x2044: | |
| 1435 | + ch = 0xda; | |
| 1436 | + break; | |
| 1437 | + case 0xa4: | |
| 1438 | + ch = 0xdb; | |
| 1439 | + break; | |
| 1440 | + case 0x2039: | |
| 1441 | + ch = 0xdc; | |
| 1442 | + break; | |
| 1443 | + case 0x203a: | |
| 1444 | + ch = 0xdd; | |
| 1445 | + break; | |
| 1446 | + case 0xfb01: | |
| 1447 | + ch = 0xde; | |
| 1448 | + break; | |
| 1449 | + case 0xfb02: | |
| 1450 | + ch = 0xdf; | |
| 1451 | + break; | |
| 1452 | + case 0x2021: | |
| 1453 | + ch = 0xe0; | |
| 1454 | + break; | |
| 1455 | + case 0xb7: | |
| 1456 | + ch = 0xe1; | |
| 1457 | + break; | |
| 1338 | 1458 | case 0x201a: |
| 1339 | 1459 | ch = 0xe2; |
| 1340 | 1460 | break; |
| 1341 | - case 0xae: | |
| 1342 | - ch = 0xa8; | |
| 1461 | + case 0x201e: | |
| 1462 | + ch = 0xe3; | |
| 1343 | 1463 | break; |
| 1344 | - case 0x30a: | |
| 1345 | - ch = 0xfb; | |
| 1464 | + case 0x2030: | |
| 1465 | + ch = 0xe4; | |
| 1346 | 1466 | break; |
| 1347 | - case 0xa7: | |
| 1348 | - ch = 0xa4; | |
| 1467 | + case 0xc2: | |
| 1468 | + ch = 0xe5; | |
| 1349 | 1469 | break; |
| 1350 | - case 0xa3: | |
| 1351 | - ch = 0xa3; | |
| 1470 | + case 0xca: | |
| 1471 | + ch = 0xe6; | |
| 1472 | + break; | |
| 1473 | + case 0xc1: | |
| 1474 | + ch = 0xe7; | |
| 1475 | + break; | |
| 1476 | + case 0xcb: | |
| 1477 | + ch = 0xe8; | |
| 1478 | + break; | |
| 1479 | + case 0xc8: | |
| 1480 | + ch = 0xe9; | |
| 1481 | + break; | |
| 1482 | + case 0xcd: | |
| 1483 | + ch = 0xea; | |
| 1484 | + break; | |
| 1485 | + case 0xce: | |
| 1486 | + ch = 0xeb; | |
| 1487 | + break; | |
| 1488 | + case 0xcf: | |
| 1489 | + ch = 0xec; | |
| 1490 | + break; | |
| 1491 | + case 0xcc: | |
| 1492 | + ch = 0xed; | |
| 1493 | + break; | |
| 1494 | + case 0xd3: | |
| 1495 | + ch = 0xee; | |
| 1496 | + break; | |
| 1497 | + case 0xd4: | |
| 1498 | + ch = 0xef; | |
| 1499 | + break; | |
| 1500 | + case 0xd2: | |
| 1501 | + ch = 0xf1; | |
| 1502 | + break; | |
| 1503 | + case 0xda: | |
| 1504 | + ch = 0xf2; | |
| 1505 | + break; | |
| 1506 | + case 0xdb: | |
| 1507 | + ch = 0xf3; | |
| 1508 | + break; | |
| 1509 | + case 0xd9: | |
| 1510 | + ch = 0xf4; | |
| 1511 | + break; | |
| 1512 | + case 0x131: | |
| 1513 | + ch = 0xf5; | |
| 1514 | + break; | |
| 1515 | + case 0x2c6: | |
| 1516 | + ch = 0xf6; | |
| 1352 | 1517 | break; |
| 1353 | 1518 | case 0x303: |
| 1354 | 1519 | ch = 0xf7; |
| 1355 | 1520 | break; |
| 1356 | - case 0x2122: | |
| 1357 | - ch = 0xaa; | |
| 1521 | + case 0x304: | |
| 1522 | + ch = 0xf8; | |
| 1358 | 1523 | break; |
| 1359 | - case 0xfa: | |
| 1360 | - ch = 0x9c; | |
| 1524 | + case 0x306: | |
| 1525 | + ch = 0xf9; | |
| 1361 | 1526 | break; |
| 1362 | - case 0xfb: | |
| 1363 | - ch = 0x9e; | |
| 1527 | + case 0x307: | |
| 1528 | + ch = 0xfa; | |
| 1364 | 1529 | break; |
| 1365 | - case 0xfc: | |
| 1366 | - ch = 0x9f; | |
| 1530 | + case 0x30a: | |
| 1531 | + ch = 0xfb; | |
| 1367 | 1532 | break; |
| 1368 | - case 0xf9: | |
| 1369 | - ch = 0x9d; | |
| 1533 | + case 0x327: | |
| 1534 | + ch = 0xfc; | |
| 1370 | 1535 | break; |
| 1371 | - case 0xff: | |
| 1372 | - ch = 0xd8; | |
| 1536 | + case 0x30b: | |
| 1537 | + ch = 0xfd; | |
| 1373 | 1538 | break; |
| 1374 | - case 0xa5: | |
| 1375 | - ch = 0xb4; | |
| 1539 | + case 0x328: | |
| 1540 | + ch = 0xfe; | |
| 1541 | + break; | |
| 1542 | + case 0x2c7: | |
| 1543 | + ch = 0xff; | |
| 1376 | 1544 | break; |
| 1377 | 1545 | default: |
| 1378 | 1546 | break; |
| ... | ... | @@ -1561,7 +1729,7 @@ transcode_utf8(std::string const& utf8_val, encoding_e encoding, |
| 1561 | 1729 | { |
| 1562 | 1730 | result += QUtil::toUTF16(codepoint); |
| 1563 | 1731 | } |
| 1564 | - else if ((codepoint >= 160) && (codepoint < 256) && | |
| 1732 | + else if ((codepoint > 160) && (codepoint < 256) && | |
| 1565 | 1733 | ((encoding == e_winansi) || (encoding == e_pdfdoc))) |
| 1566 | 1734 | { |
| 1567 | 1735 | ch = static_cast<unsigned char>(codepoint & 0xff); |
| ... | ... | @@ -1686,13 +1854,37 @@ QUtil::utf16_to_utf8(std::string const& val) |
| 1686 | 1854 | std::string |
| 1687 | 1855 | QUtil::win_ansi_to_utf8(std::string const& val) |
| 1688 | 1856 | { |
| 1689 | - return "QXXXQ"; | |
| 1857 | + std::string result; | |
| 1858 | + size_t len = val.length(); | |
| 1859 | + for (unsigned int i = 0; i < len; ++i) | |
| 1860 | + { | |
| 1861 | + unsigned char ch = static_cast<unsigned char>(val.at(i)); | |
| 1862 | + unsigned short val = ch; | |
| 1863 | + if ((ch >= 128) && (ch <= 160)) | |
| 1864 | + { | |
| 1865 | + val = win_ansi_to_unicode[ch - 128]; | |
| 1866 | + } | |
| 1867 | + result += QUtil::toUTF8(val); | |
| 1868 | + } | |
| 1869 | + return result; | |
| 1690 | 1870 | } |
| 1691 | 1871 | |
| 1692 | 1872 | std::string |
| 1693 | 1873 | QUtil::mac_roman_to_utf8(std::string const& val) |
| 1694 | 1874 | { |
| 1695 | - return "QXXXQ"; | |
| 1875 | + std::string result; | |
| 1876 | + size_t len = val.length(); | |
| 1877 | + for (unsigned int i = 0; i < len; ++i) | |
| 1878 | + { | |
| 1879 | + unsigned char ch = static_cast<unsigned char>(val.at(i)); | |
| 1880 | + unsigned short val = ch; | |
| 1881 | + if (ch >= 128) | |
| 1882 | + { | |
| 1883 | + val = mac_roman_to_unicode[ch - 128]; | |
| 1884 | + } | |
| 1885 | + result += QUtil::toUTF8(val); | |
| 1886 | + } | |
| 1887 | + return result; | |
| 1696 | 1888 | } |
| 1697 | 1889 | |
| 1698 | 1890 | std::string | ... | ... |
libtests/qtest/qutil/qutil.out
| ... | ... | @@ -53,6 +53,10 @@ HAGOOGAMAGOOGLE: 0 |
| 53 | 53 | *Does * have fingers? |
| 54 | 54 | <bf>Does * have fingers? |
| 55 | 55 | <c0>Does * have fingers? |
| 56 | +---- transcoding | |
| 57 | +bidirectional pdf doc done | |
| 58 | +bidirectional win ansi done | |
| 59 | +bidirectional mac roman done | |
| 56 | 60 | ---- whoami |
| 57 | 61 | quack1 |
| 58 | 62 | quack2 | ... | ... |
libtests/qutil.cc
| ... | ... | @@ -240,6 +240,41 @@ void utf8_to_ascii_test() |
| 240 | 240 | << ">" << b.substr(1) << std::endl; |
| 241 | 241 | } |
| 242 | 242 | |
| 243 | +void transcoding_test(std::string (*to_utf8)(std::string const&), | |
| 244 | + std::string (*from_utf8)(std::string const&, char), | |
| 245 | + int last, std::string unknown) | |
| 246 | +{ | |
| 247 | + std::string in(" "); | |
| 248 | + std::string out; | |
| 249 | + std::string back; | |
| 250 | + for (int i = 128; i <= last; ++i) | |
| 251 | + { | |
| 252 | + in.at(0) = static_cast<unsigned char>(i); | |
| 253 | + out = (*to_utf8)(in); | |
| 254 | + std::string wanted = (out == "\xef\xbf\xbd") ? unknown : in; | |
| 255 | + back = (*from_utf8)(out, '?'); | |
| 256 | + if (back != wanted) | |
| 257 | + { | |
| 258 | + std::cout << i << ": " << in << " -> " << out | |
| 259 | + << " -> " << back << " (wanted " << wanted << ")" | |
| 260 | + << std::endl; | |
| 261 | + } | |
| 262 | + } | |
| 263 | +} | |
| 264 | + | |
| 265 | +void transcoding_test() | |
| 266 | +{ | |
| 267 | + transcoding_test(&QUtil::pdf_doc_to_utf8, | |
| 268 | + &QUtil::utf8_to_pdf_doc, 160, "\x9f"); | |
| 269 | + std::cout << "bidirectional pdf doc done" << std::endl; | |
| 270 | + transcoding_test(&QUtil::win_ansi_to_utf8, | |
| 271 | + &QUtil::utf8_to_win_ansi, 160, "?"); | |
| 272 | + std::cout << "bidirectional win ansi done" << std::endl; | |
| 273 | + transcoding_test(&QUtil::mac_roman_to_utf8, | |
| 274 | + &QUtil::utf8_to_mac_roman, 255, "?"); | |
| 275 | + std::cout << "bidirectional mac roman done" << std::endl; | |
| 276 | +} | |
| 277 | + | |
| 243 | 278 | void print_whoami(char const* str) |
| 244 | 279 | { |
| 245 | 280 | PointerHolder<char> dup(true, QUtil::copy_string(str)); |
| ... | ... | @@ -350,6 +385,8 @@ int main(int argc, char* argv[]) |
| 350 | 385 | to_utf16_test(); |
| 351 | 386 | std::cout << "---- utf8_to_ascii" << std::endl; |
| 352 | 387 | utf8_to_ascii_test(); |
| 388 | + std::cout << "---- transcoding" << std::endl; | |
| 389 | + transcoding_test(); | |
| 353 | 390 | std::cout << "---- whoami" << std::endl; |
| 354 | 391 | get_whoami_test(); |
| 355 | 392 | std::cout << "---- file" << std::endl; | ... | ... |