* xmlParseCharEncoding: * @name: the encoding name as parsed, in UTF-8 format (ASCII actually) * * Compare the string to the encoding schemes already known. Note * that the comparison is case insensitive accordingly to the section * [XML] 4.3.3 Character Encoding in Entities. * * Returns one of the XML_CHAR_ENCODING_... values or XML_CHAR_ENCODING_NONE * if not recognized. */
| 1094 | * if not recognized. |
| 1095 | */ |
| 1096 | xmlCharEncoding |
| 1097 | xmlParseCharEncoding(const char* name) |
| 1098 | { |
| 1099 | const char *alias; |
| 1100 | char upper[500]; |
| 1101 | int i; |
| 1102 | |
| 1103 | if (name == NULL) |
| 1104 | return(XML_CHAR_ENCODING_NONE); |
| 1105 | |
| 1106 | /* |
| 1107 | * Do the alias resolution |
| 1108 | */ |
| 1109 | alias = xmlGetEncodingAlias(name); |
| 1110 | if (alias != NULL) |
| 1111 | name = alias; |
| 1112 | |
| 1113 | for (i = 0;i < 499;i++) { |
| 1114 | upper[i] = (char) toupper((unsigned char) name[i]); |
| 1115 | if (upper[i] == 0) break; |
| 1116 | } |
| 1117 | upper[i] = 0; |
| 1118 | |
| 1119 | if (!strcmp(upper, "")) return(XML_CHAR_ENCODING_NONE); |
| 1120 | if (!strcmp(upper, "UTF-8")) return(XML_CHAR_ENCODING_UTF8); |
| 1121 | if (!strcmp(upper, "UTF8")) return(XML_CHAR_ENCODING_UTF8); |
| 1122 | |
| 1123 | /* |
| 1124 | * NOTE: if we were able to parse this, the endianness of UTF16 is |
| 1125 | * already found and in use |
| 1126 | */ |
| 1127 | if (!strcmp(upper, "UTF-16")) return(XML_CHAR_ENCODING_UTF16LE); |
| 1128 | if (!strcmp(upper, "UTF16")) return(XML_CHAR_ENCODING_UTF16LE); |
| 1129 | |
| 1130 | if (!strcmp(upper, "ISO-10646-UCS-2")) return(XML_CHAR_ENCODING_UCS2); |
| 1131 | if (!strcmp(upper, "UCS-2")) return(XML_CHAR_ENCODING_UCS2); |
| 1132 | if (!strcmp(upper, "UCS2")) return(XML_CHAR_ENCODING_UCS2); |
| 1133 | |
| 1134 | /* |
| 1135 | * NOTE: if we were able to parse this, the endianness of UCS4 is |
| 1136 | * already found and in use |
| 1137 | */ |
| 1138 | if (!strcmp(upper, "ISO-10646-UCS-4")) return(XML_CHAR_ENCODING_UCS4LE); |
| 1139 | if (!strcmp(upper, "UCS-4")) return(XML_CHAR_ENCODING_UCS4LE); |
| 1140 | if (!strcmp(upper, "UCS4")) return(XML_CHAR_ENCODING_UCS4LE); |
| 1141 | |
| 1142 | |
| 1143 | if (!strcmp(upper, "ISO-8859-1")) return(XML_CHAR_ENCODING_8859_1); |
| 1144 | if (!strcmp(upper, "ISO-LATIN-1")) return(XML_CHAR_ENCODING_8859_1); |
| 1145 | if (!strcmp(upper, "ISO LATIN 1")) return(XML_CHAR_ENCODING_8859_1); |
| 1146 | |
| 1147 | if (!strcmp(upper, "ISO-8859-2")) return(XML_CHAR_ENCODING_8859_2); |
| 1148 | if (!strcmp(upper, "ISO-LATIN-2")) return(XML_CHAR_ENCODING_8859_2); |
| 1149 | if (!strcmp(upper, "ISO LATIN 2")) return(XML_CHAR_ENCODING_8859_2); |
| 1150 | |
| 1151 | if (!strcmp(upper, "ISO-8859-3")) return(XML_CHAR_ENCODING_8859_3); |
| 1152 | if (!strcmp(upper, "ISO-8859-4")) return(XML_CHAR_ENCODING_8859_4); |
| 1153 | if (!strcmp(upper, "ISO-8859-5")) return(XML_CHAR_ENCODING_8859_5); |
no test coverage detected