Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 1 | /** |
| 2 | * @file xpath.c |
| 3 | * @author Michal Vasko <mvasko@cesnet.cz> |
| 4 | * @brief YANG XPath evaluation functions |
| 5 | * |
| 6 | * Copyright (c) 2015 - 2017 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 | |
| 15 | #include "common.h" |
| 16 | |
| 17 | #include <ctype.h> |
| 18 | #include <stdio.h> |
| 19 | #include <stdlib.h> |
| 20 | #include <stdint.h> |
| 21 | #include <string.h> |
| 22 | #include <assert.h> |
| 23 | #include <limits.h> |
| 24 | #include <errno.h> |
| 25 | #include <math.h> |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 26 | |
| 27 | #include "xpath.h" |
| 28 | #include "xml.h" |
| 29 | |
| 30 | /** |
| 31 | * @brief Parse NCName. |
| 32 | * |
| 33 | * @param[in] ncname Name to parse. |
| 34 | * |
| 35 | * @return Length of \p ncname valid bytes. |
| 36 | */ |
Radek Krejci | d427026 | 2019-01-07 15:07:25 +0100 | [diff] [blame] | 37 | static long int |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 38 | parse_ncname(const char *ncname) |
| 39 | { |
| 40 | unsigned int uc; |
Radek Krejci | d427026 | 2019-01-07 15:07:25 +0100 | [diff] [blame] | 41 | size_t size; |
| 42 | long int len = 0; |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 43 | |
| 44 | LY_CHECK_RET(ly_getutf8(&ncname, &uc, &size), 0); |
| 45 | if (!is_xmlqnamestartchar(uc) || (uc == ':')) { |
| 46 | return len; |
| 47 | } |
| 48 | |
| 49 | do { |
| 50 | len += size; |
Radek Krejci | 9a564c9 | 2019-01-07 14:53:57 +0100 | [diff] [blame] | 51 | if (!*ncname) { |
| 52 | break; |
| 53 | } |
Radek Krejci | d427026 | 2019-01-07 15:07:25 +0100 | [diff] [blame] | 54 | LY_CHECK_RET(ly_getutf8(&ncname, &uc, &size), -len); |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 55 | } while (is_xmlqnamechar(uc) && (uc != ':')); |
| 56 | |
| 57 | return len; |
| 58 | } |
| 59 | |
| 60 | /** |
| 61 | * @brief Add \p token into the expression \p exp. |
| 62 | * |
| 63 | * @param[in] ctx libyang context to log in. |
| 64 | * @param[in] exp Expression to use. |
| 65 | * @param[in] token Token to add. |
| 66 | * @param[in] expr_pos Token position in the XPath expression. |
| 67 | * @param[in] tok_len Token length in the XPath expression. |
| 68 | * @return LY_ERR value |
| 69 | */ |
| 70 | static LY_ERR |
| 71 | exp_add_token(struct ly_ctx *ctx, struct lyxp_expr *exp, enum lyxp_token token, uint16_t expr_pos, uint16_t tok_len) |
| 72 | { |
| 73 | uint32_t prev; |
| 74 | |
| 75 | if (exp->used == exp->size) { |
| 76 | prev = exp->size; |
| 77 | exp->size += LYXP_EXPR_SIZE_STEP; |
| 78 | if (prev > exp->size) { |
| 79 | LOGINT(ctx); |
| 80 | return LY_EINT; |
| 81 | } |
| 82 | |
| 83 | exp->tokens = ly_realloc(exp->tokens, exp->size * sizeof *exp->tokens); |
| 84 | LY_CHECK_ERR_RET(!exp->tokens, LOGMEM(ctx), LY_EMEM); |
| 85 | exp->tok_pos = ly_realloc(exp->tok_pos, exp->size * sizeof *exp->tok_pos); |
| 86 | LY_CHECK_ERR_RET(!exp->tok_pos, LOGMEM(ctx), LY_EMEM); |
| 87 | exp->tok_len = ly_realloc(exp->tok_len, exp->size * sizeof *exp->tok_len); |
| 88 | LY_CHECK_ERR_RET(!exp->tok_len, LOGMEM(ctx), LY_EMEM); |
| 89 | } |
| 90 | |
| 91 | exp->tokens[exp->used] = token; |
| 92 | exp->tok_pos[exp->used] = expr_pos; |
| 93 | exp->tok_len[exp->used] = tok_len; |
| 94 | ++exp->used; |
| 95 | return LY_SUCCESS; |
| 96 | } |
| 97 | |
| 98 | void |
| 99 | lyxp_expr_free(struct ly_ctx *ctx, struct lyxp_expr *expr) |
| 100 | { |
| 101 | uint16_t i; |
| 102 | |
| 103 | if (!expr) { |
| 104 | return; |
| 105 | } |
| 106 | |
| 107 | lydict_remove(ctx, expr->expr); |
| 108 | free(expr->tokens); |
| 109 | free(expr->tok_pos); |
| 110 | free(expr->tok_len); |
| 111 | if (expr->repeat) { |
| 112 | for (i = 0; i < expr->used; ++i) { |
| 113 | free(expr->repeat[i]); |
| 114 | } |
| 115 | } |
| 116 | free(expr->repeat); |
| 117 | free(expr); |
| 118 | } |
| 119 | |
| 120 | struct lyxp_expr * |
| 121 | lyxp_expr_parse(struct ly_ctx *ctx, const char *expr) |
| 122 | { |
| 123 | struct lyxp_expr *ret; |
Radek Krejci | d427026 | 2019-01-07 15:07:25 +0100 | [diff] [blame] | 124 | size_t parsed = 0, tok_len; |
| 125 | long int ncname_len; |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 126 | enum lyxp_token tok_type; |
| 127 | int prev_function_check = 0; |
| 128 | |
| 129 | if (strlen(expr) > UINT16_MAX) { |
| 130 | LOGERR(ctx, LY_EINVAL, "XPath expression cannot be longer than %ud characters.", UINT16_MAX); |
| 131 | return NULL; |
| 132 | } |
| 133 | |
| 134 | /* init lyxp_expr structure */ |
| 135 | ret = calloc(1, sizeof *ret); |
| 136 | LY_CHECK_ERR_GOTO(!ret, LOGMEM(ctx), error); |
| 137 | ret->expr = lydict_insert(ctx, expr, strlen(expr)); |
| 138 | LY_CHECK_ERR_GOTO(!ret->expr, LOGMEM(ctx), error); |
| 139 | ret->used = 0; |
| 140 | ret->size = LYXP_EXPR_SIZE_START; |
| 141 | ret->tokens = malloc(ret->size * sizeof *ret->tokens); |
| 142 | LY_CHECK_ERR_GOTO(!ret->tokens, LOGMEM(ctx), error); |
| 143 | |
| 144 | ret->tok_pos = malloc(ret->size * sizeof *ret->tok_pos); |
| 145 | LY_CHECK_ERR_GOTO(!ret->tok_pos, LOGMEM(ctx), error); |
| 146 | |
| 147 | ret->tok_len = malloc(ret->size * sizeof *ret->tok_len); |
| 148 | LY_CHECK_ERR_GOTO(!ret->tok_len, LOGMEM(ctx), error); |
| 149 | |
| 150 | while (is_xmlws(expr[parsed])) { |
| 151 | ++parsed; |
| 152 | } |
| 153 | |
| 154 | do { |
| 155 | if (expr[parsed] == '(') { |
| 156 | |
| 157 | /* '(' */ |
| 158 | tok_len = 1; |
| 159 | tok_type = LYXP_TOKEN_PAR1; |
| 160 | |
| 161 | if (prev_function_check && ret->used && (ret->tokens[ret->used - 1] == LYXP_TOKEN_NAMETEST)) { |
| 162 | /* it is a NodeType/FunctionName after all */ |
| 163 | if (((ret->tok_len[ret->used - 1] == 4) |
| 164 | && (!strncmp(&expr[ret->tok_pos[ret->used - 1]], "node", 4) |
| 165 | || !strncmp(&expr[ret->tok_pos[ret->used - 1]], "text", 4))) || |
| 166 | ((ret->tok_len[ret->used - 1] == 7) |
| 167 | && !strncmp(&expr[ret->tok_pos[ret->used - 1]], "comment", 7))) { |
| 168 | ret->tokens[ret->used - 1] = LYXP_TOKEN_NODETYPE; |
| 169 | } else { |
| 170 | ret->tokens[ret->used - 1] = LYXP_TOKEN_FUNCNAME; |
| 171 | } |
| 172 | prev_function_check = 0; |
| 173 | } |
| 174 | |
| 175 | } else if (expr[parsed] == ')') { |
| 176 | |
| 177 | /* ')' */ |
| 178 | tok_len = 1; |
| 179 | tok_type = LYXP_TOKEN_PAR2; |
| 180 | |
| 181 | } else if (expr[parsed] == '[') { |
| 182 | |
| 183 | /* '[' */ |
| 184 | tok_len = 1; |
| 185 | tok_type = LYXP_TOKEN_BRACK1; |
| 186 | |
| 187 | } else if (expr[parsed] == ']') { |
| 188 | |
| 189 | /* ']' */ |
| 190 | tok_len = 1; |
| 191 | tok_type = LYXP_TOKEN_BRACK2; |
| 192 | |
| 193 | } else if (!strncmp(&expr[parsed], "..", 2)) { |
| 194 | |
| 195 | /* '..' */ |
| 196 | tok_len = 2; |
| 197 | tok_type = LYXP_TOKEN_DDOT; |
| 198 | |
| 199 | } else if ((expr[parsed] == '.') && (!isdigit(expr[parsed + 1]))) { |
| 200 | |
| 201 | /* '.' */ |
| 202 | tok_len = 1; |
| 203 | tok_type = LYXP_TOKEN_DOT; |
| 204 | |
| 205 | } else if (expr[parsed] == '@') { |
| 206 | |
| 207 | /* '@' */ |
| 208 | tok_len = 1; |
| 209 | tok_type = LYXP_TOKEN_AT; |
| 210 | |
| 211 | } else if (expr[parsed] == ',') { |
| 212 | |
| 213 | /* ',' */ |
| 214 | tok_len = 1; |
| 215 | tok_type = LYXP_TOKEN_COMMA; |
| 216 | |
| 217 | } else if (expr[parsed] == '\'') { |
| 218 | |
| 219 | /* Literal with ' */ |
| 220 | for (tok_len = 1; (expr[parsed + tok_len] != '\0') && (expr[parsed + tok_len] != '\''); ++tok_len); |
| 221 | LY_CHECK_ERR_GOTO(expr[parsed + tok_len] == '\0', |
| 222 | LOGVAL(ctx, LY_VLOG_NONE, NULL, LY_VCODE_XP_EOE, expr[parsed], &expr[parsed]), error); |
| 223 | ++tok_len; |
| 224 | tok_type = LYXP_TOKEN_LITERAL; |
| 225 | |
| 226 | } else if (expr[parsed] == '\"') { |
| 227 | |
| 228 | /* Literal with " */ |
| 229 | for (tok_len = 1; (expr[parsed + tok_len] != '\0') && (expr[parsed + tok_len] != '\"'); ++tok_len); |
| 230 | LY_CHECK_ERR_GOTO(expr[parsed + tok_len] == '\0', |
| 231 | LOGVAL(ctx, LY_VLOG_NONE, NULL, LY_VCODE_XP_EOE, expr[parsed], &expr[parsed]), error); |
| 232 | ++tok_len; |
| 233 | tok_type = LYXP_TOKEN_LITERAL; |
| 234 | |
| 235 | } else if ((expr[parsed] == '.') || (isdigit(expr[parsed]))) { |
| 236 | |
| 237 | /* Number */ |
| 238 | for (tok_len = 0; isdigit(expr[parsed + tok_len]); ++tok_len); |
| 239 | if (expr[parsed + tok_len] == '.') { |
| 240 | ++tok_len; |
| 241 | for (; isdigit(expr[parsed + tok_len]); ++tok_len); |
| 242 | } |
| 243 | tok_type = LYXP_TOKEN_NUMBER; |
| 244 | |
| 245 | } else if (expr[parsed] == '/') { |
| 246 | |
| 247 | /* Operator '/', '//' */ |
| 248 | if (!strncmp(&expr[parsed], "//", 2)) { |
| 249 | tok_len = 2; |
| 250 | } else { |
| 251 | tok_len = 1; |
| 252 | } |
| 253 | tok_type = LYXP_TOKEN_OPERATOR_PATH; |
| 254 | |
| 255 | } else if (!strncmp(&expr[parsed], "!=", 2) || !strncmp(&expr[parsed], "<=", 2) |
| 256 | || !strncmp(&expr[parsed], ">=", 2)) { |
| 257 | |
| 258 | /* Operator '!=', '<=', '>=' */ |
| 259 | tok_len = 2; |
| 260 | tok_type = LYXP_TOKEN_OPERATOR_COMP; |
| 261 | |
| 262 | } else if (expr[parsed] == '|') { |
| 263 | |
| 264 | /* Operator '|' */ |
| 265 | tok_len = 1; |
| 266 | tok_type = LYXP_TOKEN_OPERATOR_UNI; |
| 267 | |
| 268 | } else if ((expr[parsed] == '+') || (expr[parsed] == '-')) { |
| 269 | |
| 270 | /* Operator '+', '-' */ |
| 271 | tok_len = 1; |
| 272 | tok_type = LYXP_TOKEN_OPERATOR_MATH; |
| 273 | |
| 274 | } else if ((expr[parsed] == '=') || (expr[parsed] == '<') || (expr[parsed] == '>')) { |
| 275 | |
| 276 | /* Operator '=', '<', '>' */ |
| 277 | tok_len = 1; |
| 278 | tok_type = LYXP_TOKEN_OPERATOR_COMP; |
| 279 | |
| 280 | } else if (ret->used && (ret->tokens[ret->used - 1] != LYXP_TOKEN_AT) |
| 281 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_PAR1) |
| 282 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_BRACK1) |
| 283 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_COMMA) |
| 284 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_OPERATOR_LOG) |
| 285 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_OPERATOR_COMP) |
| 286 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_OPERATOR_MATH) |
| 287 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_OPERATOR_UNI) |
| 288 | && (ret->tokens[ret->used - 1] != LYXP_TOKEN_OPERATOR_PATH)) { |
| 289 | |
| 290 | /* Operator '*', 'or', 'and', 'mod', or 'div' */ |
| 291 | if (expr[parsed] == '*') { |
| 292 | tok_len = 1; |
| 293 | tok_type = LYXP_TOKEN_OPERATOR_MATH; |
| 294 | |
| 295 | } else if (!strncmp(&expr[parsed], "or", 2)) { |
| 296 | tok_len = 2; |
| 297 | tok_type = LYXP_TOKEN_OPERATOR_LOG; |
| 298 | |
| 299 | } else if (!strncmp(&expr[parsed], "and", 3)) { |
| 300 | tok_len = 3; |
| 301 | tok_type = LYXP_TOKEN_OPERATOR_LOG; |
| 302 | |
| 303 | } else if (!strncmp(&expr[parsed], "mod", 3) || !strncmp(&expr[parsed], "div", 3)) { |
| 304 | tok_len = 3; |
| 305 | tok_type = LYXP_TOKEN_OPERATOR_MATH; |
| 306 | |
| 307 | } else if (prev_function_check) { |
| 308 | LOGVAL(ctx, LY_VLOG_NONE, NULL, LYVE_XPATH, "Invalid character 0x%x, perhaps \"%.*s\" is supposed to be a function call.", |
| 309 | expr[parsed], &expr[parsed], ret->tok_len[ret->used - 1], &ret->expr[ret->tok_pos[ret->used - 1]]); |
| 310 | goto error; |
| 311 | } else { |
Radek Krejci | d427026 | 2019-01-07 15:07:25 +0100 | [diff] [blame] | 312 | LOGVAL(ctx, LY_VLOG_NONE, NULL, LY_VCODE_XP_INEXPR, parsed + 1, expr); |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 313 | goto error; |
| 314 | } |
| 315 | } else if (expr[parsed] == '*') { |
| 316 | |
| 317 | /* NameTest '*' */ |
| 318 | tok_len = 1; |
| 319 | tok_type = LYXP_TOKEN_NAMETEST; |
| 320 | |
| 321 | } else { |
| 322 | |
| 323 | /* NameTest (NCName ':' '*' | QName) or NodeType/FunctionName */ |
| 324 | ncname_len = parse_ncname(&expr[parsed]); |
Radek Krejci | d427026 | 2019-01-07 15:07:25 +0100 | [diff] [blame] | 325 | LY_CHECK_ERR_GOTO(ncname_len < 0, LOGVAL(ctx, LY_VLOG_NONE, NULL, LY_VCODE_XP_INEXPR, parsed - ncname_len + 1, expr), error); |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 326 | tok_len = ncname_len; |
| 327 | |
| 328 | if (expr[parsed + tok_len] == ':') { |
| 329 | ++tok_len; |
| 330 | if (expr[parsed + tok_len] == '*') { |
| 331 | ++tok_len; |
| 332 | } else { |
| 333 | ncname_len = parse_ncname(&expr[parsed + tok_len]); |
Radek Krejci | d427026 | 2019-01-07 15:07:25 +0100 | [diff] [blame] | 334 | LY_CHECK_ERR_GOTO(ncname_len < 0, LOGVAL(ctx, LY_VLOG_NONE, NULL, LY_VCODE_XP_INEXPR, parsed - ncname_len + 1, expr), error); |
Radek Krejci | b1646a9 | 2018-11-02 16:08:26 +0100 | [diff] [blame] | 335 | tok_len += ncname_len; |
| 336 | } |
| 337 | /* remove old flag to prevent ambiguities */ |
| 338 | prev_function_check = 0; |
| 339 | tok_type = LYXP_TOKEN_NAMETEST; |
| 340 | } else { |
| 341 | /* there is no prefix so it can still be NodeType/FunctionName, we can't finally decide now */ |
| 342 | prev_function_check = 1; |
| 343 | tok_type = LYXP_TOKEN_NAMETEST; |
| 344 | } |
| 345 | } |
| 346 | |
| 347 | /* store the token, move on to the next one */ |
| 348 | LY_CHECK_GOTO(exp_add_token(ctx, ret, tok_type, parsed, tok_len), error); |
| 349 | parsed += tok_len; |
| 350 | while (is_xmlws(expr[parsed])) { |
| 351 | ++parsed; |
| 352 | } |
| 353 | |
| 354 | } while (expr[parsed]); |
| 355 | |
| 356 | /* prealloc repeat */ |
| 357 | ret->repeat = calloc(ret->size, sizeof *ret->repeat); |
| 358 | LY_CHECK_ERR_GOTO(!ret->repeat, LOGMEM(ctx), error); |
| 359 | |
| 360 | return ret; |
| 361 | |
| 362 | error: |
| 363 | lyxp_expr_free(ctx, ret); |
| 364 | return NULL; |
| 365 | } |
| 366 | |