Go to most recent revision | Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 3584 | sourcerer | 1 | /* |
| 2 | * This file is part of libdom. |
||
| 3 | * Licensed under the MIT License, |
||
| 4 | * http://www.opensource.org/licenses/mit-license.php |
||
| 5 | * Copyright 2009 Bo Yang |
||
| 6 | */ |
||
| 7 | |||
| 8 | #include |
||
| 9 | #include |
||
| 10 | |||
| 11 | #include "utils/validate.h" |
||
| 12 | |||
| 13 | #include |
||
| 14 | |||
| 15 | #include "utils/character_valid.h" |
||
| 16 | #include "utils/namespace.h" |
||
| 17 | #include "utils/utils.h" |
||
| 18 | |||
| 19 | #include |
||
| 20 | |||
| 21 | /* An combination of various tests */ |
||
| 22 | static bool is_first_char(uint32_t ch); |
||
| 23 | static bool is_name_char(uint32_t ch); |
||
| 24 | |||
| 25 | /* Test whether the character can be the first character of |
||
| 26 | * a NCName. */ |
||
| 27 | static bool is_first_char(uint32_t ch) |
||
| 28 | { |
||
| 29 | /* Refer http://www.w3.org/TR/REC-xml/ for detail */ |
||
| 30 | if (((ch >= 'a') && (ch <= 'z')) || |
||
| 31 | ((ch >= 'A') && (ch <= 'Z')) || |
||
| 32 | (ch == '_') || (ch == ':') || |
||
| 33 | ((ch >= 0xC0) && (ch <= 0xD6)) || |
||
| 34 | ((ch >= 0xD8) && (ch <= 0xF6)) || |
||
| 35 | ((ch >= 0xF8) && (ch <= 0x2FF)) || |
||
| 36 | ((ch >= 0x370) && (ch <= 0x37D)) || |
||
| 37 | ((ch >= 0x37F) && (ch <= 0x1FFF)) || |
||
| 38 | ((ch >= 0x200C) && (ch <= 0x200D)) || |
||
| 39 | ((ch >= 0x2070) && (ch <= 0x218F)) || |
||
| 40 | ((ch >= 0x2C00) && (ch <= 0x2FEF)) || |
||
| 41 | ((ch >= 0x3001) && (ch <= 0xD7FF)) || |
||
| 42 | ((ch >= 0xF900) && (ch <= 0xFDCF)) || |
||
| 43 | ((ch >= 0xFDF0) && (ch <= 0xFFFD)) || |
||
| 44 | ((ch >= 0x10000) && (ch <= 0xEFFFF))) |
||
| 45 | return true; |
||
| 46 | |||
| 47 | if (is_letter(ch) || ch == (uint32_t) '_' || ch == (uint32_t) ':') { |
||
| 48 | return true; |
||
| 49 | } |
||
| 50 | |||
| 51 | return false; |
||
| 52 | } |
||
| 53 | |||
| 54 | /* Test whether the character can be a part of a NCName */ |
||
| 55 | static bool is_name_char(uint32_t ch) |
||
| 56 | { |
||
| 57 | /* Refer http://www.w3.org/TR/REC-xml/ for detail */ |
||
| 58 | if (((ch >= 'a') && (ch <= 'z')) || |
||
| 59 | ((ch >= 'A') && (ch <= 'Z')) || |
||
| 60 | ((ch >= '0') && (ch <= '9')) || /* !start */ |
||
| 61 | (ch == '_') || (ch == ':') || |
||
| 62 | (ch == '-') || (ch == '.') || (ch == 0xB7) || /* !start */ |
||
| 63 | ((ch >= 0xC0) && (ch <= 0xD6)) || |
||
| 64 | ((ch >= 0xD8) && (ch <= 0xF6)) || |
||
| 65 | ((ch >= 0xF8) && (ch <= 0x2FF)) || |
||
| 66 | ((ch >= 0x300) && (ch <= 0x36F)) || /* !start */ |
||
| 67 | ((ch >= 0x370) && (ch <= 0x37D)) || |
||
| 68 | ((ch >= 0x37F) && (ch <= 0x1FFF)) || |
||
| 69 | ((ch >= 0x200C) && (ch <= 0x200D)) || |
||
| 70 | ((ch >= 0x203F) && (ch <= 0x2040)) || /* !start */ |
||
| 71 | ((ch >= 0x2070) && (ch <= 0x218F)) || |
||
| 72 | ((ch >= 0x2C00) && (ch <= 0x2FEF)) || |
||
| 73 | ((ch >= 0x3001) && (ch <= 0xD7FF)) || |
||
| 74 | ((ch >= 0xF900) && (ch <= 0xFDCF)) || |
||
| 75 | ((ch >= 0xFDF0) && (ch <= 0xFFFD)) || |
||
| 76 | ((ch >= 0x10000) && (ch <= 0xEFFFF))) |
||
| 77 | return true; |
||
| 78 | |||
| 79 | if (is_letter(ch) == true) |
||
| 80 | return true; |
||
| 81 | if (is_digit(ch) == true) |
||
| 82 | return true; |
||
| 83 | if (is_combining_char(ch) == true) |
||
| 84 | return true; |
||
| 85 | if (is_extender(ch) == true) |
||
| 86 | return true; |
||
| 87 | |||
| 88 | if (ch == (uint32_t) '.' || ch == (uint32_t) '-' || |
||
| 89 | ch == (uint32_t) '_' || ch == (uint32_t) ':') |
||
| 90 | return true; |
||
| 91 | |||
| 92 | return false; |
||
| 93 | } |
||
| 94 | |||
| 95 | /** |
||
| 96 | * Test whether the name is a valid one according XML 1.0 standard. |
||
| 97 | * For the standard please refer: |
||
| 98 | * |
||
| 99 | * http://www.w3.org/TR/2004/REC-xml-20040204/ |
||
| 100 | * |
||
| 101 | * \param name The name need to be tested |
||
| 102 | * \return true if ::name is valid, false otherwise. |
||
| 103 | */ |
||
| 104 | bool _dom_validate_name(dom_string *name) |
||
| 105 | { |
||
| 106 | uint32_t ch; |
||
| 107 | size_t clen, slen; |
||
| 108 | parserutils_error err; |
||
| 109 | const uint8_t *s; |
||
| 110 | |||
| 111 | if (name == NULL) |
||
| 112 | return false; |
||
| 113 | |||
| 114 | slen = dom_string_length(name); |
||
| 115 | if (slen == 0) |
||
| 116 | return false; |
||
| 117 | |||
| 118 | s = (const uint8_t *) dom_string_data(name); |
||
| 119 | slen = dom_string_byte_length(name); |
||
| 120 | |||
| 121 | err = parserutils_charset_utf8_to_ucs4(s, slen, &ch, &clen); |
||
| 122 | if (err != PARSERUTILS_OK) { |
||
| 123 | return false; |
||
| 124 | } |
||
| 125 | |||
| 126 | if (is_first_char(ch) == false) |
||
| 127 | return false; |
||
| 128 | |||
| 129 | s += clen; |
||
| 130 | slen -= clen; |
||
| 131 | |||
| 132 | while (slen > 0) { |
||
| 133 | err = parserutils_charset_utf8_to_ucs4(s, slen, &ch, &clen); |
||
| 134 | if (err != PARSERUTILS_OK) { |
||
| 135 | return false; |
||
| 136 | } |
||
| 137 | |||
| 138 | if (is_name_char(ch) == false) |
||
| 139 | return false; |
||
| 140 | |||
| 141 | s += clen; |
||
| 142 | slen -= clen; |
||
| 143 | } |
||
| 144 | |||
| 145 | return true; |
||
| 146 | } |
||
| 147 | |||
| 148 | /** |
||
| 149 | * Validate whether the string is a legal NCName. |
||
| 150 | * Refer http://www.w3.org/TR/REC-xml-names/ for detail. |
||
| 151 | * |
||
| 152 | * \param str The name to validate |
||
| 153 | * \return true if ::name is valid, false otherwise. |
||
| 154 | */ |
||
| 155 | bool _dom_validate_ncname(dom_string *name) |
||
| 156 | { |
||
| 157 | uint32_t ch; |
||
| 158 | size_t clen, slen; |
||
| 159 | parserutils_error err; |
||
| 160 | const uint8_t *s; |
||
| 161 | |||
| 162 | if (name == NULL) |
||
| 163 | return false; |
||
| 164 | |||
| 165 | slen = dom_string_length(name); |
||
| 166 | if (slen == 0) |
||
| 167 | return false; |
||
| 168 | |||
| 169 | s = (const uint8_t *) dom_string_data(name); |
||
| 170 | slen = dom_string_byte_length(name); |
||
| 171 | |||
| 172 | err = parserutils_charset_utf8_to_ucs4(s, slen, &ch, &clen); |
||
| 173 | if (err != PARSERUTILS_OK) { |
||
| 174 | return false; |
||
| 175 | } |
||
| 176 | |||
| 177 | if (is_letter(ch) == false && ch != (uint32_t) '_') |
||
| 178 | return false; |
||
| 179 | |||
| 180 | s += clen; |
||
| 181 | slen -= clen; |
||
| 182 | |||
| 183 | while (slen > 0) { |
||
| 184 | err = parserutils_charset_utf8_to_ucs4(s, slen, &ch, &clen); |
||
| 185 | if (err != PARSERUTILS_OK) { |
||
| 186 | return false; |
||
| 187 | } |
||
| 188 | |||
| 189 | if (is_name_char(ch) == false) |
||
| 190 | return false; |
||
| 191 | |||
| 192 | if (ch == (uint32_t) ':') |
||
| 193 | return false; |
||
| 194 | |||
| 195 | s += clen; |
||
| 196 | slen -= clen; |
||
| 197 | } |
||
| 198 | |||
| 199 | return true; |
||
| 200 | } |
||
| 201 |