1 /* 2 __ __ _ 3 ___\ \/ /_ __ __ _| |_ 4 / _ \\ /| '_ \ / _` | __| 5 | __// \| |_) | (_| | |_ 6 \___/_/\_\ .__/ \__,_|\__| 7 |_| XML parser 8 9 Copyright (c) 1997-2000 Thai Open Source Software Center Ltd 10 Copyright (c) 2000 Clark Cooper <coopercc@users.sourceforge.net> 11 Copyright (c) 2002 Fred L. Drake, Jr. <fdrake@users.sourceforge.net> 12 Copyright (c) 2002-2005 Karl Waclawek <karl@waclawek.net> 13 Copyright (c) 2016-2026 Sebastian Pipping <sebastian@pipping.org> 14 Copyright (c) 2017 Rhodri James <rhodri@wildebeest.org.uk> 15 Licensed under the MIT license: 16 17 Permission is hereby granted, free of charge, to any person obtaining 18 a copy of this software and associated documentation files (the 19 "Software"), to deal in the Software without restriction, including 20 without limitation the rights to use, copy, modify, merge, publish, 21 distribute, sublicense, and/or sell copies of the Software, and to permit 22 persons to whom the Software is furnished to do so, subject to the 23 following conditions: 24 25 The above copyright notice and this permission notice shall be included 26 in all copies or substantial portions of the Software. 27 28 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 29 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 30 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN 31 NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, 32 DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 33 OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 34 USE OR OTHER DEALINGS IN THE SOFTWARE. 35 36 SPDX-License-Identifier: MIT 37 */ 38 39 #ifndef XmlTok_INCLUDED 40 # define XmlTok_INCLUDED 1 41 42 # include <stdint.h> // uint64_t 43 44 # ifdef __cplusplus 45 extern "C" { 46 # endif 47 48 /* The following token may be returned by XmlContentTok */ 49 # define XML_TOK_TRAILING_RSQB \ 50 -5 /* ] or ]] at the end of the scan; might be \ 51 start of illegal ]]> sequence */ 52 /* The following tokens may be returned by both XmlPrologTok and 53 XmlContentTok. 54 */ 55 # define XML_TOK_NONE -4 /* The string to be scanned is empty */ 56 # define XML_TOK_TRAILING_CR \ 57 -3 /* A CR at the end of the scan; \ 58 might be part of CRLF sequence */ 59 # define XML_TOK_PARTIAL_CHAR -2 /* only part of a multibyte sequence */ 60 # define XML_TOK_PARTIAL -1 /* only part of a token */ 61 # define XML_TOK_INVALID 0 62 63 /* The following tokens are returned by XmlContentTok; some are also 64 returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok. 65 */ 66 # define XML_TOK_START_TAG_WITH_ATTS 1 67 # define XML_TOK_START_TAG_NO_ATTS 2 68 # define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */ 69 # define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4 70 # define XML_TOK_END_TAG 5 71 # define XML_TOK_DATA_CHARS 6 72 # define XML_TOK_DATA_NEWLINE 7 73 # define XML_TOK_CDATA_SECT_OPEN 8 74 # define XML_TOK_ENTITY_REF 9 75 # define XML_TOK_CHAR_REF 10 /* numeric character reference */ 76 77 /* The following tokens may be returned by both XmlPrologTok and 78 XmlContentTok. 79 */ 80 # define XML_TOK_PI 11 /* processing instruction */ 81 # define XML_TOK_XML_DECL 12 /* XML decl or text decl */ 82 # define XML_TOK_COMMENT 13 83 # define XML_TOK_BOM 14 /* Byte order mark */ 84 85 /* The following tokens are returned only by XmlPrologTok */ 86 # define XML_TOK_PROLOG_S 15 87 # define XML_TOK_DECL_OPEN 16 /* <!foo */ 88 # define XML_TOK_DECL_CLOSE 17 /* > */ 89 # define XML_TOK_NAME 18 90 # define XML_TOK_NMTOKEN 19 91 # define XML_TOK_POUND_NAME 20 /* #name */ 92 # define XML_TOK_OR 21 /* | */ 93 # define XML_TOK_PERCENT 22 94 # define XML_TOK_OPEN_PAREN 23 95 # define XML_TOK_CLOSE_PAREN 24 96 # define XML_TOK_OPEN_BRACKET 25 97 # define XML_TOK_CLOSE_BRACKET 26 98 # define XML_TOK_LITERAL 27 99 # define XML_TOK_PARAM_ENTITY_REF 28 100 # define XML_TOK_INSTANCE_START 29 101 102 /* The following occur only in element type declarations */ 103 # define XML_TOK_NAME_QUESTION 30 /* name? */ 104 # define XML_TOK_NAME_ASTERISK 31 /* name* */ 105 # define XML_TOK_NAME_PLUS 32 /* name+ */ 106 # define XML_TOK_COND_SECT_OPEN 33 /* <![ */ 107 # define XML_TOK_COND_SECT_CLOSE 34 /* ]]> */ 108 # define XML_TOK_CLOSE_PAREN_QUESTION 35 /* )? */ 109 # define XML_TOK_CLOSE_PAREN_ASTERISK 36 /* )* */ 110 # define XML_TOK_CLOSE_PAREN_PLUS 37 /* )+ */ 111 # define XML_TOK_COMMA 38 112 113 /* The following token is returned only by XmlAttributeValueTok */ 114 # define XML_TOK_ATTRIBUTE_VALUE_S 39 115 116 /* The following token is returned only by XmlCdataSectionTok */ 117 # define XML_TOK_CDATA_SECT_CLOSE 40 118 119 /* With namespace processing this is returned by XmlPrologTok for a 120 name with a colon. 121 */ 122 # define XML_TOK_PREFIXED_NAME 41 123 124 # ifdef XML_DTD 125 # define XML_TOK_IGNORE_SECT 42 126 # endif /* XML_DTD */ 127 128 # ifdef XML_DTD 129 # define XML_N_STATES 4 130 # else /* not XML_DTD */ 131 # define XML_N_STATES 3 132 # endif /* not XML_DTD */ 133 134 # define XML_PROLOG_STATE 0 135 # define XML_CONTENT_STATE 1 136 # define XML_CDATA_SECTION_STATE 2 137 # ifdef XML_DTD 138 # define XML_IGNORE_SECTION_STATE 3 139 # endif /* XML_DTD */ 140 141 # define XML_N_LITERAL_TYPES 2 142 # define XML_ATTRIBUTE_VALUE_LITERAL 0 143 # define XML_ENTITY_VALUE_LITERAL 1 144 145 /* The size of the buffer passed to XmlUtf8Encode must be at least this. */ 146 # define XML_UTF8_ENCODE_MAX 4 147 /* The size of the buffer passed to XmlUtf16Encode must be at least this. */ 148 # define XML_UTF16_ENCODE_MAX 2 149 150 typedef struct position { 151 /* first line and first column are 0 not 1 */ 152 uint64_t lineNumber; 153 uint64_t columnNumber; 154 } POSITION; 155 156 typedef struct { 157 const char *name; 158 const char *valuePtr; 159 const char *valueEnd; 160 char normalized; 161 } ATTRIBUTE; 162 163 struct encoding; 164 typedef struct encoding ENCODING; 165 166 typedef int(PTRCALL *SCANNER)(const ENCODING *, const char *, const char *, 167 const char **); 168 169 enum XML_Convert_Result { 170 XML_CONVERT_COMPLETED = 0, 171 XML_CONVERT_INPUT_INCOMPLETE = 1, 172 XML_CONVERT_OUTPUT_EXHAUSTED = 2 /* and therefore potentially input remaining 173 as well */ 174 }; 175 176 struct encoding { 177 SCANNER scanners[XML_N_STATES]; 178 SCANNER literalScanners[XML_N_LITERAL_TYPES]; 179 int(PTRCALL *nameMatchesAscii)(const ENCODING *, const char *, const char *, 180 const char *); 181 int(PTRFASTCALL *nameLength)(const ENCODING *, const char *); 182 const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *); 183 int(PTRCALL *getAtts)(const ENCODING *enc, const char *ptr, int attsMax, 184 ATTRIBUTE *atts); 185 int(PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr); 186 int(PTRCALL *predefinedEntityName)(const ENCODING *, const char *, 187 const char *); 188 void(PTRCALL *updatePosition)(const ENCODING *, const char *ptr, 189 const char *end, POSITION *); 190 int(PTRCALL *isPublicId)(const ENCODING *enc, const char *ptr, 191 const char *end, const char **badPtr); 192 enum XML_Convert_Result(PTRCALL *utf8Convert)(const ENCODING *enc, 193 const char **fromP, 194 const char *fromLim, char **toP, 195 const char *toLim); 196 enum XML_Convert_Result(PTRCALL *utf16Convert)(const ENCODING *enc, 197 const char **fromP, 198 const char *fromLim, 199 unsigned short **toP, 200 const unsigned short *toLim); 201 int minBytesPerChar; 202 char isUtf8; 203 char isUtf16; 204 }; 205 206 /* Scan the string starting at ptr until the end of the next complete 207 token, but do not scan past eptr. Return an integer giving the 208 type of token. 209 210 Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set. 211 212 Return XML_TOK_PARTIAL when the string does not contain a complete 213 token; nextTokPtr will not be set. 214 215 Return XML_TOK_INVALID when the string does not start a valid 216 token; nextTokPtr will be set to point to the character which made 217 the token invalid. 218 219 Otherwise the string starts with a valid token; nextTokPtr will be 220 set to point to the character following the end of that token. 221 222 Each data character counts as a single token, but adjacent data 223 characters may be returned together. Similarly for characters in 224 the prolog outside literals, comments and processing instructions. 225 */ 226 227 # define XmlTok(enc, state, ptr, end, nextTokPtr) \ 228 (((enc)->scanners[state])(enc, ptr, end, nextTokPtr)) 229 230 # define XmlPrologTok(enc, ptr, end, nextTokPtr) \ 231 XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr) 232 233 # define XmlContentTok(enc, ptr, end, nextTokPtr) \ 234 XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr) 235 236 # define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \ 237 XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr) 238 239 # ifdef XML_DTD 240 241 # define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \ 242 XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr) 243 244 # endif /* XML_DTD */ 245 246 /* This is used for performing a 2nd-level tokenization on the content 247 of a literal that has already been returned by XmlTok. 248 */ 249 # define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \ 250 (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr)) 251 252 # define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \ 253 XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr) 254 255 # define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \ 256 XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr) 257 258 # define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \ 259 (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2)) 260 261 # define XmlNameLength(enc, ptr) (((enc)->nameLength)(enc, ptr)) 262 263 # define XmlSkipS(enc, ptr) (((enc)->skipS)(enc, ptr)) 264 265 # define XmlGetAttributes(enc, ptr, attsMax, atts) \ 266 (((enc)->getAtts)(enc, ptr, attsMax, atts)) 267 268 # define XmlCharRefNumber(enc, ptr) (((enc)->charRefNumber)(enc, ptr)) 269 270 # define XmlPredefinedEntityName(enc, ptr, end) \ 271 (((enc)->predefinedEntityName)(enc, ptr, end)) 272 273 # define XmlUpdatePosition(enc, ptr, end, pos) \ 274 (((enc)->updatePosition)(enc, ptr, end, pos)) 275 276 # define XmlIsPublicId(enc, ptr, end, badPtr) \ 277 (((enc)->isPublicId)(enc, ptr, end, badPtr)) 278 279 # define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \ 280 (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim)) 281 282 # define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \ 283 (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim)) 284 285 typedef struct { 286 ENCODING initEnc; 287 const ENCODING **encPtr; 288 } INIT_ENCODING; 289 290 int XmlParseXmlDecl(int isGeneralTextEntity, const ENCODING *enc, 291 const char *ptr, const char *end, const char **badPtr, 292 const char **versionPtr, const char **versionEndPtr, 293 const char **encodingNamePtr, 294 const ENCODING **namedEncodingPtr, int *standalonePtr); 295 296 int XmlInitEncoding(INIT_ENCODING *p, const ENCODING **encPtr, 297 const char *name); 298 const ENCODING *XmlGetUtf8InternalEncoding(void); 299 const ENCODING *XmlGetUtf16InternalEncoding(void); 300 int FASTCALL XmlUtf8Encode(int charNumber, char *buf); 301 int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf); 302 int XmlSizeOfUnknownEncoding(void); 303 304 typedef int(XMLCALL *CONVERTER)(void *userData, const char *p); 305 306 ENCODING *XmlInitUnknownEncoding(void *mem, const int *table, CONVERTER convert, 307 void *userData); 308 309 int XmlParseXmlDeclNS(int isGeneralTextEntity, const ENCODING *enc, 310 const char *ptr, const char *end, const char **badPtr, 311 const char **versionPtr, const char **versionEndPtr, 312 const char **encodingNamePtr, 313 const ENCODING **namedEncodingPtr, int *standalonePtr); 314 315 int XmlInitEncodingNS(INIT_ENCODING *p, const ENCODING **encPtr, 316 const char *name); 317 const ENCODING *XmlGetUtf8InternalEncodingNS(void); 318 const ENCODING *XmlGetUtf16InternalEncodingNS(void); 319 ENCODING *XmlInitUnknownEncodingNS(void *mem, const int *table, 320 CONVERTER convert, void *userData); 321 # ifdef __cplusplus 322 } 323 # endif 324 325 #endif /* not XmlTok_INCLUDED */ 326