Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 1 | /** |
| 2 | * @file xml.c |
| 3 | * @author Radek Krejci <rkrejci@cesnet.cz> |
| 4 | * @brief Generic XML parser implementation for libyang |
| 5 | * |
| 6 | * Copyright (c) 2015 - 2018 CESNET, z.s.p.o. |
| 7 | * |
| 8 | * This source code is licensed under BSD 3-Clause License (the "License"). |
| 9 | * You may not use this file except in compliance with the License. |
| 10 | * You may obtain a copy of the License at |
| 11 | * |
| 12 | * https://opensource.org/licenses/BSD-3-Clause |
| 13 | */ |
| 14 | |
Radek Krejci | c1c03d6 | 2018-11-27 10:52:43 +0100 | [diff] [blame] | 15 | #include "common.h" |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 16 | |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 17 | #include <assert.h> |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 18 | #include <ctype.h> |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 19 | #include <stdbool.h> |
| 20 | #include <stdint.h> |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 21 | #include <string.h> |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 22 | |
| 23 | #include "libyang.h" |
| 24 | #include "xml.h" |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 25 | |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 26 | /* Move input p by s characters, if EOF log with lyxml_context c */ |
| 27 | #define move_input(c,p,s) p += s; LY_CHECK_ERR_RET(!p[0], LOGVAL(c->ctx, LY_VLOG_LINE, &c->line, LY_VCODE_EOF), LY_EVALID) |
| 28 | |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 29 | /* Ignore whitespaces in the input string p */ |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 30 | #define ign_xmlws(c,p) while (is_xmlws(*(p))) {if (*(p) == '\n') {++c->line;} ++p;} |
| 31 | |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 32 | /** |
| 33 | * @brief Ignore any characters until the delim of the size delim_len is read |
| 34 | * |
| 35 | * Detects number of read new lines. |
| 36 | * Returns the pointer to the beginning of the detected delim, or NULL in case the delim not found in |
| 37 | * NULL-terminated input string. |
| 38 | * */ |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 39 | static const char * |
| 40 | ign_todelim(register const char *input, const char *delim, size_t delim_len, size_t *newlines) |
| 41 | { |
| 42 | size_t i; |
| 43 | register const char *a, *b; |
| 44 | |
| 45 | (*newlines) = 0; |
| 46 | for ( ; *input; ++input) { |
| 47 | if (*input != *delim) { |
| 48 | if (*input == '\n') { |
| 49 | ++(*newlines); |
| 50 | } |
| 51 | continue; |
| 52 | } |
| 53 | a = input; |
| 54 | b = delim; |
| 55 | for (i = 0; i < delim_len; ++i) { |
| 56 | if (*a++ != *b++) { |
| 57 | break; |
| 58 | } |
| 59 | } |
| 60 | if (i == delim_len) { |
| 61 | return input; |
| 62 | } |
| 63 | } |
| 64 | return NULL; |
| 65 | } |
| 66 | |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 67 | /** |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 68 | * Store UTF-8 character specified as 4byte integer into the dst buffer. |
| 69 | * Returns number of written bytes (4 max), expects that dst has enough space. |
| 70 | * |
| 71 | * UTF-8 mapping: |
| 72 | * 00000000 -- 0000007F: 0xxxxxxx |
| 73 | * 00000080 -- 000007FF: 110xxxxx 10xxxxxx |
| 74 | * 00000800 -- 0000FFFF: 1110xxxx 10xxxxxx 10xxxxxx |
| 75 | * 00010000 -- 001FFFFF: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx |
| 76 | * |
| 77 | * Includes checking for valid characters (following RFC 7950, sec 9.4) |
| 78 | */ |
| 79 | static LY_ERR |
Radek Krejci | 117d208 | 2018-09-26 10:05:14 +0200 | [diff] [blame] | 80 | lyxml_pututf8(char *dst, uint32_t value, size_t *bytes_written) |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 81 | { |
| 82 | if (value < 0x80) { |
| 83 | /* one byte character */ |
| 84 | if (value < 0x20 && |
| 85 | value != 0x09 && |
| 86 | value != 0x0a && |
| 87 | value != 0x0d) { |
| 88 | return LY_EINVAL; |
| 89 | } |
| 90 | |
| 91 | dst[0] = value; |
| 92 | (*bytes_written) = 1; |
| 93 | } else if (value < 0x800) { |
| 94 | /* two bytes character */ |
| 95 | dst[0] = 0xc0 | (value >> 6); |
| 96 | dst[1] = 0x80 | (value & 0x3f); |
| 97 | (*bytes_written) = 2; |
| 98 | } else if (value < 0xfffe) { |
| 99 | /* three bytes character */ |
| 100 | if (((value & 0xf800) == 0xd800) || |
| 101 | (value >= 0xfdd0 && value <= 0xfdef)) { |
| 102 | /* exclude surrogate blocks %xD800-DFFF */ |
| 103 | /* exclude noncharacters %xFDD0-FDEF */ |
| 104 | return LY_EINVAL; |
| 105 | } |
| 106 | |
| 107 | dst[0] = 0xe0 | (value >> 12); |
| 108 | dst[1] = 0x80 | ((value >> 6) & 0x3f); |
| 109 | dst[2] = 0x80 | (value & 0x3f); |
| 110 | |
| 111 | (*bytes_written) = 3; |
| 112 | } else if (value < 0x10fffe) { |
| 113 | if ((value & 0xffe) == 0xffe) { |
| 114 | /* exclude noncharacters %xFFFE-FFFF, %x1FFFE-1FFFF, %x2FFFE-2FFFF, %x3FFFE-3FFFF, %x4FFFE-4FFFF, |
| 115 | * %x5FFFE-5FFFF, %x6FFFE-6FFFF, %x7FFFE-7FFFF, %x8FFFE-8FFFF, %x9FFFE-9FFFF, %xAFFFE-AFFFF, |
| 116 | * %xBFFFE-BFFFF, %xCFFFE-CFFFF, %xDFFFE-DFFFF, %xEFFFE-EFFFF, %xFFFFE-FFFFF, %x10FFFE-10FFFF */ |
| 117 | return LY_EINVAL; |
| 118 | } |
| 119 | /* four bytes character */ |
| 120 | dst[0] = 0xf0 | (value >> 18); |
| 121 | dst[1] = 0x80 | ((value >> 12) & 0x3f); |
| 122 | dst[2] = 0x80 | ((value >> 6) & 0x3f); |
| 123 | dst[3] = 0x80 | (value & 0x3f); |
| 124 | |
| 125 | (*bytes_written) = 4; |
| 126 | } |
| 127 | return LY_SUCCESS; |
| 128 | } |
| 129 | |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 130 | /** |
| 131 | * @brief Check/Get an XML qualified name from the input string. |
| 132 | * |
| 133 | * The identifier must have at least one valid character complying the name start character constraints. |
| 134 | * The identifier is terminated by the first character, which does not comply to the name character constraints. |
| 135 | * |
| 136 | * See https://www.w3.org/TR/xml-names/#NT-NCName |
| 137 | * |
| 138 | * @param[in] context XML context to track lines or store errors into libyang context. |
| 139 | * @param[in,out] input Input string to process, updated according to the processed/read data. |
| 140 | * Note that the term_char is also read, so input points after the term_char at the end. |
| 141 | * @param[out] term_char The first character in the input string which does not compy to the name constraints. |
| 142 | * @param[out] term_char_len Number of bytes used to encode UTF8 term_char. Serves to be able to go back in input string. |
| 143 | * @return LY_ERR value. |
| 144 | */ |
| 145 | static LY_ERR |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 146 | lyxml_check_qname(struct lyxml_context *context, const char **input, unsigned int *term_char, size_t *term_char_len) |
| 147 | { |
| 148 | unsigned int c; |
| 149 | const char *id = (*input); |
| 150 | LY_ERR rc; |
| 151 | |
| 152 | /* check NameStartChar (minus colon) */ |
Radek Krejci | b416be6 | 2018-10-01 14:51:45 +0200 | [diff] [blame] | 153 | LY_CHECK_ERR_RET(ly_getutf8(input, &c, NULL) != LY_SUCCESS, |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 154 | LOGVAL(context->ctx, LY_VLOG_LINE, &context->line, LY_VCODE_INCHAR, (*input)[0]), LY_EVALID); |
| 155 | LY_CHECK_ERR_RET(!is_xmlqnamestartchar(c), |
| 156 | LOGVAL(context->ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, |
| 157 | "Identifier \"%s\" starts with invalid character.", id), |
| 158 | LY_EVALID); |
| 159 | |
| 160 | /* check rest of the identifier */ |
Radek Krejci | b416be6 | 2018-10-01 14:51:45 +0200 | [diff] [blame] | 161 | for (rc = ly_getutf8(input, &c, term_char_len); |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 162 | rc == LY_SUCCESS && is_xmlqnamechar(c); |
Radek Krejci | b416be6 | 2018-10-01 14:51:45 +0200 | [diff] [blame] | 163 | rc = ly_getutf8(input, &c, term_char_len)); |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 164 | LY_CHECK_ERR_RET(rc != LY_SUCCESS, LOGVAL(context->ctx, LY_VLOG_LINE, &context->line, LY_VCODE_INCHAR, (*input)[0]), LY_EVALID); |
| 165 | |
| 166 | (*term_char) = c; |
| 167 | return LY_SUCCESS; |
| 168 | } |
| 169 | |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 170 | LY_ERR |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 171 | lyxml_get_string(struct lyxml_context *context, const char **input, char **buffer, size_t *buffer_size, char **output, size_t *length, int *dynamic) |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 172 | { |
| 173 | #define BUFSIZE 4096 |
| 174 | #define BUFSIZE_STEP 4096 |
| 175 | #define BUFSIZE_CHECK(CTX, BUF, SIZE, CURR, NEED) \ |
| 176 | if (CURR+NEED >= SIZE) { \ |
| 177 | BUF = ly_realloc(BUF, SIZE + BUFSIZE_STEP); \ |
| 178 | LY_CHECK_ERR_RET(!BUF, LOGMEM(CTX), LY_EMEM); \ |
| 179 | SIZE += BUFSIZE_STEP; \ |
| 180 | } |
| 181 | |
| 182 | struct ly_ctx *ctx = context->ctx; /* shortcut */ |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 183 | const char *in = (*input), *start; |
| 184 | char *buf = NULL, delim; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 185 | size_t offset; /* read offset in input buffer */ |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 186 | size_t len; /* length of the output string (write offset in output buffer) */ |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 187 | size_t size; /* size of the output buffer */ |
| 188 | void *p; |
Radek Krejci | 117d208 | 2018-09-26 10:05:14 +0200 | [diff] [blame] | 189 | uint32_t n; |
| 190 | size_t u, newlines; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 191 | bool empty_content = false; |
| 192 | LY_ERR rc; |
| 193 | |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 194 | assert(context); |
| 195 | assert(context->status == LYXML_ELEM_CONTENT || context->status == LYXML_ATTR_CONTENT); |
| 196 | |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 197 | if (in[0] == '\'') { |
| 198 | delim = '\''; |
| 199 | ++in; |
| 200 | } else if (in[0] == '"') { |
| 201 | delim = '"'; |
| 202 | ++in; |
| 203 | } else { |
| 204 | delim = '<'; |
| 205 | empty_content = true; |
| 206 | } |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 207 | start = in; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 208 | |
| 209 | if (empty_content) { |
| 210 | /* only when processing element's content - try to ignore whitespaces used to format XML data |
| 211 | * before element's child or closing tag */ |
Radek Krejci | 117d208 | 2018-09-26 10:05:14 +0200 | [diff] [blame] | 212 | for (offset = newlines = 0; in[offset] && is_xmlws(in[offset]); ++offset) { |
| 213 | if (in[offset] == '\n') { |
| 214 | ++newlines; |
| 215 | } |
| 216 | } |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 217 | LY_CHECK_ERR_RET(!in[offset], LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_EOF), LY_EVALID); |
Radek Krejci | 117d208 | 2018-09-26 10:05:14 +0200 | [diff] [blame] | 218 | context->line += newlines; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 219 | if (in[offset] == '<') { |
Radek Krejci | ed6c6ad | 2018-09-26 09:10:18 +0200 | [diff] [blame] | 220 | (*input) = in + offset; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 221 | return LY_EINVAL; |
| 222 | } |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 223 | } |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 224 | /* init */ |
| 225 | offset = len = 0; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 226 | |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 227 | if (0) { |
| 228 | getbuffer: |
| 229 | /* prepare output buffer */ |
| 230 | if (*buffer) { |
| 231 | buf = *buffer; |
| 232 | size = *buffer_size; |
| 233 | } else { |
| 234 | buf = malloc(BUFSIZE); |
| 235 | size = BUFSIZE; |
| 236 | LY_CHECK_ERR_RET(!buf, LOGMEM(ctx), LY_EMEM); |
| 237 | } |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 238 | } |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 239 | |
| 240 | /* parse */ |
| 241 | while (in[offset]) { |
| 242 | if (in[offset] == '&') { |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 243 | if (!buf) { |
| 244 | /* it is necessary to modify the input, so we will need a dynamically allocated buffer */ |
| 245 | goto getbuffer; |
| 246 | } |
| 247 | |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 248 | if (offset) { |
| 249 | /* store what we have so far */ |
| 250 | BUFSIZE_CHECK(ctx, buf, size, len, offset); |
| 251 | memcpy(&buf[len], in, offset); |
| 252 | len += offset; |
| 253 | in += offset; |
| 254 | offset = 0; |
| 255 | } |
| 256 | /* process reference */ |
| 257 | /* we will need 4 bytes at most since we support only the predefined |
| 258 | * (one-char) entities and character references */ |
| 259 | BUFSIZE_CHECK(ctx, buf, size, len, 4); |
| 260 | ++offset; |
| 261 | if (in[offset] != '#') { |
| 262 | /* entity reference - only predefined references are supported */ |
| 263 | if (!strncmp(&in[offset], "lt;", 3)) { |
| 264 | buf[len++] = '<'; |
| 265 | in += 4; /* < */ |
| 266 | } else if (!strncmp(&in[offset], "gt;", 3)) { |
| 267 | buf[len++] = '>'; |
| 268 | in += 4; /* > */ |
| 269 | } else if (!strncmp(&in[offset], "amp;", 4)) { |
| 270 | buf[len++] = '&'; |
| 271 | in += 5; /* & */ |
| 272 | } else if (!strncmp(&in[offset], "apos;", 5)) { |
| 273 | buf[len++] = '\''; |
| 274 | in += 6; /* ' */ |
| 275 | } else if (!strncmp(&in[offset], "quot;", 5)) { |
| 276 | buf[len++] = '\"'; |
| 277 | in += 6; /* " */ |
| 278 | } else { |
Radek Krejci | ed6c6ad | 2018-09-26 09:10:18 +0200 | [diff] [blame] | 279 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, |
| 280 | "Entity reference \"%.*s\" not supported, only predefined references allowed.", 10, &in[offset-1]); |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 281 | goto error; |
| 282 | } |
| 283 | offset = 0; |
| 284 | } else { |
| 285 | p = (void*)&in[offset - 1]; |
| 286 | /* character reference */ |
| 287 | ++offset; |
| 288 | if (isdigit(in[offset])) { |
| 289 | for (n = 0; isdigit(in[offset]); offset++) { |
| 290 | n = (10 * n) + (in[offset] - '0'); |
| 291 | } |
| 292 | } else if (in[offset] == 'x' && isxdigit(in[offset + 1])) { |
| 293 | for (n = 0, ++offset; isxdigit(in[offset]); offset++) { |
| 294 | if (isdigit(in[offset])) { |
| 295 | u = (in[offset] - '0'); |
| 296 | } else if (in[offset] > 'F') { |
| 297 | u = 10 + (in[offset] - 'a'); |
| 298 | } else { |
| 299 | u = 10 + (in[offset] - 'A'); |
| 300 | } |
| 301 | n = (16 * n) + u; |
| 302 | } |
| 303 | } else { |
Radek Krejci | ed6c6ad | 2018-09-26 09:10:18 +0200 | [diff] [blame] | 304 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, "Invalid character reference \"%.*s\".", 12, p); |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 305 | goto error; |
| 306 | |
| 307 | } |
| 308 | LY_CHECK_ERR_GOTO(in[offset] != ';', |
| 309 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_INSTREXP, |
| 310 | LY_VCODE_INSTREXP_len(&in[offset]), &in[offset], ";"), |
| 311 | error); |
| 312 | ++offset; |
| 313 | rc = lyxml_pututf8(&buf[len], n, &u); |
| 314 | LY_CHECK_ERR_GOTO(rc, LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, |
Radek Krejci | 117d208 | 2018-09-26 10:05:14 +0200 | [diff] [blame] | 315 | "Invalid character reference \"%.*s\" (0x%08x).", 12, p, n), |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 316 | error); |
| 317 | len += u; |
| 318 | in += offset; |
| 319 | offset = 0; |
| 320 | } |
| 321 | } else if (in[offset] == delim) { |
| 322 | /* end of string */ |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 323 | if (buf) { |
| 324 | if (len + offset >= size) { |
| 325 | buf = ly_realloc(buf, len + offset + 1); |
| 326 | LY_CHECK_ERR_RET(!buf, LOGMEM(ctx), LY_EMEM); |
| 327 | size = len + offset + 1; |
| 328 | } |
| 329 | memcpy(&buf[len], in, offset); |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 330 | } |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 331 | len += offset; |
| 332 | /* in case of element content, keep the leading <, |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 333 | * for attribute's value move after the terminating quotation mark */ |
| 334 | if (context->status == LYXML_ELEM_CONTENT) { |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 335 | in += offset; |
| 336 | } else { |
| 337 | in += offset + 1; |
| 338 | } |
| 339 | goto success; |
| 340 | } else { |
| 341 | /* log lines */ |
| 342 | if (in[offset] == '\n') { |
| 343 | ++context->line; |
| 344 | } |
| 345 | |
| 346 | /* continue */ |
| 347 | ++offset; |
| 348 | } |
| 349 | } |
| 350 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_EOF); |
| 351 | error: |
| 352 | if (!(*buffer)) { |
Radek Krejci | bb9b198 | 2019-04-08 14:24:59 +0200 | [diff] [blame] | 353 | /* buffer not provided, buf is local */ |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 354 | free(buf); |
Radek Krejci | bb9b198 | 2019-04-08 14:24:59 +0200 | [diff] [blame] | 355 | } else if (buf) { |
| 356 | /* buf is shared with caller via buffer, but buf could be reallocated, so update the provided buffer */ |
| 357 | (*buffer) = buf; |
| 358 | (*buffer_size) = size; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 359 | } |
| 360 | return LY_EVALID; |
| 361 | |
| 362 | success: |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 363 | if (buf) { |
| 364 | if (!(*buffer) && size != len + 1) { |
| 365 | /* not using provided buffer, so fit the allocated buffer to what we really have inside */ |
| 366 | p = realloc(buf, len + 1); |
| 367 | /* ignore realloc fail because we are reducing the buffer, |
| 368 | * so just return bigger buffer than needed */ |
| 369 | if (p) { |
| 370 | size = len + 1; |
| 371 | buf = p; |
| 372 | } |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 373 | } |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 374 | /* set terminating NULL byte */ |
| 375 | buf[len] = '\0'; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 376 | } |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 377 | |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 378 | context->status -= 1; |
Radek Krejci | d70d107 | 2018-10-09 14:20:47 +0200 | [diff] [blame] | 379 | if (buf) { |
| 380 | (*buffer) = buf; |
| 381 | (*buffer_size) = size; |
| 382 | (*output) = buf; |
| 383 | (*dynamic) = 1; |
| 384 | } else { |
| 385 | (*output) = (char*)start; |
| 386 | (*dynamic) = 0; |
| 387 | } |
| 388 | (*length) = len; |
| 389 | |
Radek Krejci | 28e8cb5 | 2019-03-08 11:31:31 +0100 | [diff] [blame] | 390 | if (context->status == LYXML_ATTRIBUTE) { |
| 391 | if (in[0] == '>') { |
| 392 | /* element terminated by > - termination of the opening tag */ |
| 393 | context->status = LYXML_ELEM_CONTENT; |
| 394 | ++in; |
| 395 | } else if (in[0] == '/' && in[1] == '>') { |
| 396 | /* element terminated by /> - termination of an empty element */ |
| 397 | context->status = LYXML_ELEMENT; |
| 398 | in += 2; |
| 399 | |
| 400 | /* remove the closed element record from the tags list */ |
| 401 | free(context->elements.objs[context->elements.count - 1]); |
| 402 | --context->elements.count; |
| 403 | } |
| 404 | } |
| 405 | |
| 406 | (*input) = in; |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 407 | return LY_SUCCESS; |
| 408 | |
| 409 | #undef BUFSIZE |
| 410 | #undef BUFSIZE_STEP |
| 411 | #undef BUFSIZE_CHECK |
| 412 | } |
| 413 | |
Radek Krejci | d972c25 | 2018-09-25 13:23:39 +0200 | [diff] [blame] | 414 | LY_ERR |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 415 | lyxml_get_attribute(struct lyxml_context *context, const char **input, |
Radek Krejci | d972c25 | 2018-09-25 13:23:39 +0200 | [diff] [blame] | 416 | const char **prefix, size_t *prefix_len, const char **name, size_t *name_len) |
| 417 | { |
| 418 | struct ly_ctx *ctx = context->ctx; /* shortcut */ |
| 419 | const char *in = (*input); |
| 420 | const char *id; |
| 421 | const char *endtag; |
| 422 | LY_ERR rc; |
| 423 | unsigned int c; |
| 424 | size_t endtag_len; |
| 425 | |
| 426 | /* initialize output variables */ |
| 427 | (*prefix) = (*name) = NULL; |
| 428 | (*prefix_len) = (*name_len) = 0; |
| 429 | |
| 430 | /* skip initial whitespaces */ |
| 431 | ign_xmlws(context, in); |
| 432 | |
| 433 | if (in[0] == '\0') { |
| 434 | /* EOF - not expected at this place */ |
| 435 | return LY_EINVAL; |
Radek Krejci | d972c25 | 2018-09-25 13:23:39 +0200 | [diff] [blame] | 436 | } |
| 437 | |
| 438 | /* remember the identifier start before checking its format */ |
| 439 | id = in; |
| 440 | rc = lyxml_check_qname(context, &in, &c, &endtag_len); |
| 441 | LY_CHECK_RET(rc); |
| 442 | if (c == ':') { |
| 443 | /* we have prefixed identifier */ |
| 444 | endtag = in - endtag_len; |
| 445 | |
| 446 | rc = lyxml_check_qname(context, &in, &c, &endtag_len); |
| 447 | LY_CHECK_RET(rc); |
| 448 | |
| 449 | (*prefix) = id; |
| 450 | (*prefix_len) = endtag - id; |
| 451 | id = endtag + 1; |
| 452 | } |
| 453 | if (!is_xmlws(c) && c != '=') { |
| 454 | in = in - endtag_len; |
| 455 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_INSTREXP, LY_VCODE_INSTREXP_len(in), in, "whitespace or '='"); |
| 456 | return LY_EVALID; |
| 457 | } |
| 458 | in = in - endtag_len; |
| 459 | (*name) = id; |
| 460 | (*name_len) = in - id; |
| 461 | |
| 462 | /* eat '=' and stop at the value beginning */ |
| 463 | ign_xmlws(context, in); |
| 464 | if (in[0] != '=') { |
| 465 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_INSTREXP, LY_VCODE_INSTREXP_len(in), in, "'='"); |
| 466 | return LY_EVALID; |
| 467 | } |
| 468 | ++in; |
| 469 | ign_xmlws(context, in); |
| 470 | if (in[0] != '\'' && in[0] != '"') { |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 471 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_INSTREXP, |
| 472 | LY_VCODE_INSTREXP_len(in), in, "either single or double quotation mark"); |
Radek Krejci | d972c25 | 2018-09-25 13:23:39 +0200 | [diff] [blame] | 473 | return LY_EVALID; |
| 474 | } |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 475 | context->status = LYXML_ATTR_CONTENT; |
Radek Krejci | d972c25 | 2018-09-25 13:23:39 +0200 | [diff] [blame] | 476 | |
Radek Krejci | d972c25 | 2018-09-25 13:23:39 +0200 | [diff] [blame] | 477 | /* move caller's input */ |
| 478 | (*input) = in; |
| 479 | return LY_SUCCESS; |
| 480 | } |
| 481 | |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 482 | LY_ERR |
Radek Krejci | 7a7fa90 | 2018-09-25 17:08:21 +0200 | [diff] [blame] | 483 | lyxml_get_element(struct lyxml_context *context, const char **input, |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 484 | const char **prefix, size_t *prefix_len, const char **name, size_t *name_len) |
| 485 | { |
| 486 | struct ly_ctx *ctx = context->ctx; /* shortcut */ |
| 487 | const char *in = (*input); |
| 488 | const char *endtag; |
| 489 | const char *sectname; |
| 490 | const char *id; |
| 491 | size_t endtag_len, newlines; |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 492 | bool loop = true, closing = false; |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 493 | unsigned int c; |
| 494 | LY_ERR rc; |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 495 | struct lyxml_elem *e; |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 496 | |
| 497 | /* initialize output variables */ |
| 498 | (*prefix) = (*name) = NULL; |
| 499 | (*prefix_len) = (*name_len) = 0; |
| 500 | |
| 501 | while (loop) { |
| 502 | ign_xmlws(context, in); |
| 503 | |
| 504 | if (in[0] == '\0') { |
| 505 | /* EOF */ |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 506 | context->status = LYXML_END; |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 507 | goto success; |
| 508 | } else if (in[0] != '<') { |
| 509 | return LY_EINVAL; |
| 510 | } |
| 511 | move_input(context, in, 1); |
| 512 | |
| 513 | if (in[0] == '!') { |
| 514 | move_input(context, in, 1); |
| 515 | /* sections to ignore */ |
| 516 | if (!strncmp(in, "--", 2)) { |
| 517 | /* comment */ |
| 518 | move_input(context, in, 2); |
| 519 | sectname = "Comment"; |
| 520 | endtag = "-->"; |
| 521 | endtag_len = 3; |
| 522 | } else if (!strncmp(in, "[CDATA[", 7)) { |
| 523 | /* CDATA section */ |
| 524 | move_input(context, in, 7); |
| 525 | sectname = "CData"; |
| 526 | endtag = "]]>"; |
| 527 | endtag_len = 3; |
| 528 | } else if (!strncmp(in, "DOCTYPE", 7)) { |
| 529 | /* Document type declaration - not supported */ |
| 530 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_NSUPP, "Document Type Declaration"); |
| 531 | return LY_EVALID; |
Radek Krejci | c5c31bb | 2019-04-08 14:40:52 +0200 | [diff] [blame] | 532 | } else { |
| 533 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, "Unknown XML section \"%.20s\".", &in[-2]); |
| 534 | return LY_EVALID; |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 535 | } |
| 536 | in = ign_todelim(in, endtag, endtag_len, &newlines); |
| 537 | LY_CHECK_ERR_RET(!in, LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_NTERM, sectname), LY_EVALID); |
| 538 | context->line += newlines; |
| 539 | in += endtag_len; |
| 540 | } else if (in[0] == '?') { |
| 541 | in = ign_todelim(in, "?>", 2, &newlines); |
| 542 | LY_CHECK_ERR_RET(!in, LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_NTERM, "Declaration"), LY_EVALID); |
| 543 | context->line += newlines; |
| 544 | in += 2; |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 545 | } else if (in[0] == '/') { |
| 546 | /* closing element */ |
| 547 | closing = true; |
| 548 | ++in; |
| 549 | goto element; |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 550 | } else { |
| 551 | /* element */ |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 552 | element: |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 553 | ign_xmlws(context, in); |
| 554 | LY_CHECK_ERR_RET(!in[0], LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_EOF), LY_EVALID); |
| 555 | |
| 556 | /* remember the identifier start before checking its format */ |
| 557 | id = in; |
| 558 | rc = lyxml_check_qname(context, &in, &c, &endtag_len); |
| 559 | LY_CHECK_RET(rc); |
| 560 | if (c == ':') { |
| 561 | /* we have prefixed identifier */ |
| 562 | endtag = in - endtag_len; |
| 563 | |
| 564 | rc = lyxml_check_qname(context, &in, &c, &endtag_len); |
| 565 | LY_CHECK_RET(rc); |
| 566 | |
| 567 | (*prefix) = id; |
| 568 | (*prefix_len) = endtag - id; |
| 569 | id = endtag + 1; |
| 570 | } |
| 571 | if (!is_xmlws(c) && c != '/' && c != '>') { |
| 572 | in = in - endtag_len; |
Radek Krejci | d972c25 | 2018-09-25 13:23:39 +0200 | [diff] [blame] | 573 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_INSTREXP, LY_VCODE_INSTREXP_len(in), in, |
| 574 | "whitespace or element tag termination ('>' or '/>'"); |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 575 | return LY_EVALID; |
| 576 | } |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 577 | (*name) = id; |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 578 | (*name_len) = in - endtag_len - id; |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 579 | |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 580 | if (is_xmlws(c)) { |
| 581 | /* go to the next meaningful input */ |
| 582 | ign_xmlws(context, in); |
| 583 | LY_CHECK_ERR_RET(!in[0], LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_EOF), LY_EVALID); |
| 584 | c = in[0]; |
| 585 | ++in; |
| 586 | endtag_len = 1; |
| 587 | } |
| 588 | |
| 589 | if (closing) { |
| 590 | /* match opening and closing element tags */ |
| 591 | LY_CHECK_ERR_RET( |
| 592 | !context->elements.count, |
Radek Krejci | 3fbc987 | 2019-04-16 16:50:01 +0200 | [diff] [blame] | 593 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, "Opening and closing elements tag missmatch (\"%.*s\").", *name_len, *name), |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 594 | LY_EVALID); |
| 595 | e = (struct lyxml_elem*)context->elements.objs[context->elements.count - 1]; |
| 596 | LY_CHECK_ERR_RET(e->prefix_len != *prefix_len || e->name_len != *name_len |
| 597 | || (*prefix_len && strncmp(*prefix, e->prefix, e->prefix_len)) || strncmp(*name, e->name, e->name_len), |
Radek Krejci | 3fbc987 | 2019-04-16 16:50:01 +0200 | [diff] [blame] | 598 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, "Opening and closing elements tag missmatch (\"%.*s\").", *name_len, *name), |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 599 | LY_EVALID); |
| 600 | /* opening and closing element tags matches, remove record from the opening tags list */ |
| 601 | free(e); |
| 602 | --context->elements.count; |
| 603 | /* do not return element information to announce closing element being currently processed */ |
| 604 | *name = *prefix = NULL; |
| 605 | *name_len = *prefix_len = 0; |
| 606 | |
| 607 | if (c == '>') { |
| 608 | /* end of closing element */ |
| 609 | context->status = LYXML_ELEMENT; |
| 610 | } else { |
| 611 | in -= endtag_len; |
| 612 | LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX, "Unexpected data \"%.*s\" in closing element tag.", |
| 613 | LY_VCODE_INSTREXP_len(in), in); |
| 614 | return LY_EVALID; |
| 615 | } |
| 616 | } else { |
| 617 | if (c == '>') { |
| 618 | /* end of opening element */ |
| 619 | context->status = LYXML_ELEM_CONTENT; |
| 620 | } else if (c == '/' && in[0] == '>') { |
| 621 | /* empty element closing */ |
| 622 | context->status = LYXML_ELEMENT; |
| 623 | ++in; |
| 624 | } else { |
| 625 | /* attribute */ |
| 626 | context->status = LYXML_ATTRIBUTE; |
| 627 | in -= endtag_len; |
| 628 | } |
| 629 | |
| 630 | if (context->status != LYXML_ELEMENT) { |
| 631 | /* store element opening tag information */ |
| 632 | e = malloc(sizeof *e); |
| 633 | LY_CHECK_ERR_RET(!e, LOGMEM(ctx), LY_EMEM); |
| 634 | e->name = *name; |
| 635 | e->prefix = *prefix; |
| 636 | e->name_len = *name_len; |
| 637 | e->prefix_len = *prefix_len; |
| 638 | ly_set_add(&context->elements, e, LY_SET_OPT_USEASLIST); |
| 639 | } |
| 640 | } |
Radek Krejci | d91dbaf | 2018-09-21 15:51:39 +0200 | [diff] [blame] | 641 | loop = false; |
| 642 | } |
| 643 | } |
| 644 | |
| 645 | success: |
| 646 | /* move caller's input */ |
| 647 | (*input) = in; |
| 648 | return LY_SUCCESS; |
| 649 | } |
| 650 | |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 651 | LY_ERR |
| 652 | lyxml_ns_add(struct lyxml_context *context, const char *element_name, const char *prefix, size_t prefix_len, char *uri) |
| 653 | { |
| 654 | struct lyxml_ns *ns; |
| 655 | |
| 656 | ns = malloc(sizeof *ns); |
| 657 | LY_CHECK_ERR_RET(!ns, LOGMEM(context->ctx), LY_EMEM); |
| 658 | |
Radek Krejci | e0734d2 | 2019-04-05 15:54:28 +0200 | [diff] [blame] | 659 | /* to distinguish 2 elements, we need not only the name, but also its depth in the XML tree. |
| 660 | * In case some dictionary is used to store elements' names (so name strings of 2 distinguish nodes |
| 661 | * actually points to the same memory), so the depth is necessary to distinguish parent/child nodes |
| 662 | * of the same name. Otherwise, the namespace defined in parent could be removed when leaving child node. */ |
| 663 | ns->element_depth = context->elements.count; |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 664 | ns->element = element_name; |
Radek Krejci | e0734d2 | 2019-04-05 15:54:28 +0200 | [diff] [blame] | 665 | |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 666 | ns->uri = uri; |
| 667 | if (prefix) { |
| 668 | ns->prefix = strndup(prefix, prefix_len); |
| 669 | LY_CHECK_ERR_RET(!ns->prefix, LOGMEM(context->ctx); free(ns), LY_EMEM); |
| 670 | } else { |
| 671 | ns->prefix = NULL; |
| 672 | } |
| 673 | |
| 674 | LY_CHECK_ERR_RET(ly_set_add(&context->ns, ns, LY_SET_OPT_USEASLIST) == -1, free(ns->prefix), LY_EMEM); |
| 675 | return LY_SUCCESS; |
| 676 | } |
| 677 | |
| 678 | const struct lyxml_ns * |
| 679 | lyxml_ns_get(struct lyxml_context *context, const char *prefix, size_t prefix_len) |
| 680 | { |
| 681 | unsigned int u; |
| 682 | struct lyxml_ns *ns; |
| 683 | |
| 684 | for (u = context->ns.count - 1; u + 1 > 0; --u) { |
| 685 | ns = (struct lyxml_ns *)context->ns.objs[u]; |
| 686 | if (prefix) { |
| 687 | if (!strncmp(prefix, ns->prefix, prefix_len) && ns->prefix[prefix_len] == '\0') { |
| 688 | return ns; |
| 689 | } |
| 690 | } else if (!ns->prefix) { |
| 691 | /* default namespace */ |
| 692 | return ns; |
| 693 | } |
| 694 | } |
| 695 | |
| 696 | return NULL; |
| 697 | } |
| 698 | |
| 699 | LY_ERR |
| 700 | lyxml_ns_rm(struct lyxml_context *context, const char *element_name) |
| 701 | { |
| 702 | unsigned int u; |
| 703 | |
| 704 | for (u = context->ns.count - 1; u + 1 > 0; --u) { |
Radek Krejci | e0734d2 | 2019-04-05 15:54:28 +0200 | [diff] [blame] | 705 | if (((struct lyxml_ns *)context->ns.objs[u])->element != element_name || |
| 706 | ((struct lyxml_ns *)context->ns.objs[u])->element_depth != context->elements.count + 1) { |
| 707 | /* we are done, the namespaces from a single element are supposed to be together; |
| 708 | * the second condition is there to distinguish parent/child elements with the same name |
| 709 | * (which are for some reason stored at the same memory chunk), so we need to distinguish |
| 710 | * level of the node */ |
Radek Krejci | 4b74d5e | 2018-09-26 14:30:55 +0200 | [diff] [blame] | 711 | break; |
| 712 | } |
| 713 | /* remove the ns structure */ |
| 714 | free(((struct lyxml_ns *)context->ns.objs[u])->prefix); |
| 715 | free(((struct lyxml_ns *)context->ns.objs[u])->uri); |
| 716 | free(context->ns.objs[u]); |
| 717 | --context->ns.count; |
| 718 | } |
| 719 | |
| 720 | if (!context->ns.count) { |
| 721 | /* cleanup the context's namespaces storage */ |
| 722 | ly_set_erase(&context->ns, NULL); |
| 723 | } |
| 724 | |
| 725 | return LY_SUCCESS; |
| 726 | } |
Radek Krejci | b189064 | 2018-10-03 14:05:40 +0200 | [diff] [blame] | 727 | |
| 728 | void |
| 729 | lyxml_context_clear(struct lyxml_context *context) |
| 730 | { |
| 731 | unsigned int u; |
| 732 | |
| 733 | ly_set_erase(&context->elements, free); |
| 734 | for (u = context->ns.count - 1; u + 1 > 0; --u) { |
| 735 | /* remove the ns structure */ |
| 736 | free(((struct lyxml_ns *)context->ns.objs[u])->prefix); |
| 737 | free(((struct lyxml_ns *)context->ns.objs[u])->uri); |
| 738 | free(context->ns.objs[u]); |
| 739 | } |
| 740 | ly_set_erase(&context->ns, NULL); |
| 741 | } |