| 1 |
%top{ |
| 2 |
/*------------------------------------------------------------------------- |
| 3 |
* |
| 4 |
* scan.l |
| 5 |
* lexical scanner for PostgreSQL |
| 6 |
* |
| 7 |
* NOTE NOTE NOTE: |
| 8 |
* |
| 9 |
* The rules in this file must be kept in sync with src/fe_utils/psqlscan.l |
| 10 |
* and src/interfaces/ecpg/preproc/pgc.l! |
| 11 |
* |
| 12 |
* The rules are designed so that the scanner never has to backtrack, |
| 13 |
* in the sense that there is always a rule that can match the input |
| 14 |
* consumed so far (the rule action may internally throw back some input |
| 15 |
* with yyless(), however). As explained in the flex manual, this makes |
| 16 |
* for a useful speed increase --- several percent faster when measuring |
| 17 |
* raw parsing (Flex + Bison). The extra complexity is mostly in the rules |
| 18 |
* for handling float numbers and continued string literals. If you change |
| 19 |
* the lexical rules, verify that you haven't broken the no-backtrack |
| 20 |
* property by running flex with the "-b" option and checking that the |
| 21 |
* resulting "lex.backup" file says that no backing up is needed. (As of |
| 22 |
* Postgres 9.2, this check is made automatically by the Makefile.) |
| 23 |
* |
| 24 |
* |
| 25 |
* Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group |
| 26 |
* Portions Copyright (c) 1994, Regents of the University of California |
| 27 |
* |
| 28 |
* IDENTIFICATION |
| 29 |
* src/backend/parser/scan.l |
| 30 |
* |
| 31 |
*------------------------------------------------------------------------- |
| 32 |
*/ |
| 33 |
#include "postgres.h" |
| 34 |
|
| 35 |
#include <ctype.h> |
| 36 |
#include <unistd.h> |
| 37 |
|
| 38 |
#include "common/string.h" |
| 39 |
#include "gramparse.h" |
| 40 |
#include "nodes/miscnodes.h" |
| 41 |
#include "parser/parser.h" /* only needed for GUC variables */ |
| 42 |
#include "parser/scansup.h" |
| 43 |
#include "port/pg_bitutils.h" |
| 44 |
#include "mb/pg_wchar.h" |
| 45 |
#include "utils/builtins.h" |
| 46 |
} |
| 47 |
|
| 48 |
%{ |
| 49 |
|
| 50 |
/* LCOV_EXCL_START */ |
| 51 |
|
| 52 |
/* Avoid exit() on fatal scanner errors (a bit ugly -- see yy_fatal_error) */ |
| 53 |
#undef fprintf |
| 54 |
#define fprintf(file, fmt, msg) fprintf_to_ereport(fmt, msg) |
| 55 |
|
| 56 |
static void |
| 57 |
fprintf_to_ereport(const char *fmt, const char *msg) |
| 58 |
{ |
| 59 |
ereport(ERROR, (errmsg_internal("%s", msg))); |
| 60 |
} |
| 61 |
|
| 62 |
/* |
| 63 |
* GUC variables. This is a DIRECT violation of the warning given at the |
| 64 |
* head of gram.y, ie flex/bison code must not depend on any GUC variables; |
| 65 |
* as such, changing their values can induce very unintuitive behavior. |
| 66 |
* But we shall have to live with it until we can remove these variables. |
| 67 |
*/ |
| 68 |
int backslash_quote = BACKSLASH_QUOTE_SAFE_ENCODING; |
| 69 |
bool escape_string_warning = true; |
| 70 |
bool standard_conforming_strings = true; |
| 71 |
|
| 72 |
/* |
| 73 |
* Constant data exported from this file. This array maps from the |
| 74 |
* zero-based keyword numbers returned by ScanKeywordLookup to the |
| 75 |
* Bison token numbers needed by gram.y. This is exported because |
| 76 |
* callers need to pass it to scanner_init, if they are using the |
| 77 |
* standard keyword list ScanKeywords. |
| 78 |
*/ |
| 79 |
#define PG_KEYWORD(kwname, value, category, collabel) value, |
| 80 |
|
| 81 |
const uint16 ScanKeywordTokens[] = { |
| 82 |
#include "parser/kwlist.h" |
| 83 |
}; |
| 84 |
|
| 85 |
#undef PG_KEYWORD |
| 86 |
|
| 87 |
/* |
| 88 |
* Set the type of YYSTYPE. |
| 89 |
*/ |
| 90 |
#define YYSTYPE core_YYSTYPE |
| 91 |
|
| 92 |
/* |
| 93 |
* Each call to yylex must set yylloc to the location of the found token |
| 94 |
* (expressed as a byte offset from the start of the input text). |
| 95 |
* When we parse a token that requires multiple lexer rules to process, |
| 96 |
* this should be done in the first such rule, else yylloc will point |
| 97 |
* into the middle of the token. |
| 98 |
*/ |
| 99 |
#define SET_YYLLOC() (*(yylloc) = yytext - yyextra->scanbuf) |
| 100 |
|
| 101 |
/* |
| 102 |
* Advance yylloc by the given number of bytes. |
| 103 |
*/ |
| 104 |
#define ADVANCE_YYLLOC(delta) ( *(yylloc) += (delta) ) |
| 105 |
|
| 106 |
/* |
| 107 |
* Sometimes, we do want yylloc to point into the middle of a token; this is |
| 108 |
* useful for instance to throw an error about an escape sequence within a |
| 109 |
* string literal. But if we find no error there, we want to revert yylloc |
| 110 |
* to the token start, so that that's the location reported to the parser. |
| 111 |
* Use PUSH_YYLLOC/POP_YYLLOC to save/restore yylloc around such code. |
| 112 |
* (Currently the implied "stack" is just one location, but someday we might |
| 113 |
* need to nest these.) |
| 114 |
*/ |
| 115 |
#define PUSH_YYLLOC() (yyextra->save_yylloc = *(yylloc)) |
| 116 |
#define POP_YYLLOC() (*(yylloc) = yyextra->save_yylloc) |
| 117 |
|
| 118 |
#define startlit() ( yyextra->literallen = 0 ) |
| 119 |
static void addlit(char *ytext, int yleng, core_yyscan_t yyscanner); |
| 120 |
static void addlitchar(unsigned char ychar, core_yyscan_t yyscanner); |
| 121 |
static char *litbufdup(core_yyscan_t yyscanner); |
| 122 |
static unsigned char unescape_single_char(unsigned char c, core_yyscan_t yyscanner); |
| 123 |
static int process_integer_literal(const char *token, YYSTYPE *lval, int base); |
| 124 |
static void addunicode(pg_wchar c, yyscan_t yyscanner); |
| 125 |
|
| 126 |
#define yyerror(msg) scanner_yyerror(msg, yyscanner) |
| 127 |
|
| 128 |
#define lexer_errposition() scanner_errposition(*(yylloc), yyscanner) |
| 129 |
|
| 130 |
static void check_string_escape_warning(unsigned char ychar, core_yyscan_t yyscanner); |
| 131 |
static void check_escape_warning(core_yyscan_t yyscanner); |
| 132 |
|
| 133 |
%} |
| 134 |
|
| 135 |
%option reentrant |
| 136 |
%option bison-bridge |
| 137 |
%option bison-locations |
| 138 |
%option 8bit |
| 139 |
%option never-interactive |
| 140 |
%option nodefault |
| 141 |
%option noinput |
| 142 |
%option nounput |
| 143 |
%option noyywrap |
| 144 |
%option noyyalloc |
| 145 |
%option noyyrealloc |
| 146 |
%option noyyfree |
| 147 |
%option warn |
| 148 |
%option prefix="core_yy" |
| 149 |
%option extra-type="core_yy_extra_type *" |
| 150 |
|
| 151 |
/* |
| 152 |
* OK, here is a short description of lex/flex rules behavior. |
| 153 |
* The longest pattern which matches an input string is always chosen. |
| 154 |
* For equal-length patterns, the first occurring in the rules list is chosen. |
| 155 |
* INITIAL is the starting state, to which all non-conditional rules apply. |
| 156 |
* Exclusive states change parsing rules while the state is active. When in |
| 157 |
* an exclusive state, only those rules defined for that state apply. |
| 158 |
* |
| 159 |
* We use exclusive states for quoted strings, extended comments, |
| 160 |
* and to eliminate parsing troubles for numeric strings. |
| 161 |
* Exclusive states: |
| 162 |
* <xb> bit string literal |
| 163 |
* <xc> extended C-style comments |
| 164 |
* <xd> delimited identifiers (double-quoted identifiers) |
| 165 |
* <xh> hexadecimal byte string |
| 166 |
* <xq> standard quoted strings |
| 167 |
* <xqs> quote stop (detect continued strings) |
| 168 |
* <xe> extended quoted strings (support backslash escape sequences) |
| 169 |
* <xdolq> $foo$ quoted strings |
| 170 |
* <xui> quoted identifier with Unicode escapes |
| 171 |
* <xus> quoted string with Unicode escapes |
| 172 |
* <xeu> Unicode surrogate pair in extended quoted string |
| 173 |
* |
| 174 |
* Remember to add an <<EOF>> case whenever you add a new exclusive state! |
| 175 |
* The default one is probably not the right thing. |
| 176 |
*/ |
| 177 |
|
| 178 |
%x xb |
| 179 |
%x xc |
| 180 |
%x xd |
| 181 |
%x xh |
| 182 |
%x xq |
| 183 |
%x xqs |
| 184 |
%x xe |
| 185 |
%x xdolq |
| 186 |
%x xui |
| 187 |
%x xus |
| 188 |
%x xeu |
| 189 |
|
| 190 |
/* |
| 191 |
* In order to make the world safe for Windows and Mac clients as well as |
| 192 |
* Unix ones, we accept either \n or \r as a newline. A DOS-style \r\n |
| 193 |
* sequence will be seen as two successive newlines, but that doesn't cause |
| 194 |
* any problems. Comments that start with -- and extend to the next |
| 195 |
* newline are treated as equivalent to a single whitespace character. |
| 196 |
* |
| 197 |
* NOTE a fine point: if there is no newline following --, we will absorb |
| 198 |
* everything to the end of the input as a comment. This is correct. Older |
| 199 |
* versions of Postgres failed to recognize -- as a comment if the input |
| 200 |
* did not end with a newline. |
| 201 |
* |
| 202 |
* non_newline_space tracks all the other space characters except newlines. |
| 203 |
* |
| 204 |
* XXX if you change the set of whitespace characters, fix scanner_isspace() |
| 205 |
* to agree. |
| 206 |
*/ |
| 207 |
|
| 208 |
space [ \t\n\r\f\v] |
| 209 |
non_newline_space [ \t\f\v] |
| 210 |
newline [\n\r] |
| 211 |
non_newline [^\n\r] |
| 212 |
|
| 213 |
comment ("--"{non_newline}*) |
| 214 |
|
| 215 |
whitespace ({space}+|{comment}) |
| 216 |
|
| 217 |
/* |
| 218 |
* SQL requires at least one newline in the whitespace separating |
| 219 |
* string literals that are to be concatenated. Silly, but who are we |
| 220 |
* to argue? Note that {whitespace_with_newline} should not have * after |
| 221 |
* it, whereas {whitespace} should generally have a * after it... |
| 222 |
*/ |
| 223 |
|
| 224 |
special_whitespace ({space}+|{comment}{newline}) |
| 225 |
non_newline_whitespace ({non_newline_space}|{comment}) |
| 226 |
whitespace_with_newline ({non_newline_whitespace}*{newline}{special_whitespace}*) |
| 227 |
|
| 228 |
quote ' |
| 229 |
/* If we see {quote} then {quotecontinue}, the quoted string continues */ |
| 230 |
quotecontinue {whitespace_with_newline}{quote} |
| 231 |
|
| 232 |
/* |
| 233 |
* {quotecontinuefail} is needed to avoid lexer backup when we fail to match |
| 234 |
* {quotecontinue}. It might seem that this could just be {whitespace}*, |
| 235 |
* but if there's a dash after {whitespace_with_newline}, it must be consumed |
| 236 |
* to see if there's another dash --- which would start a {comment} and thus |
| 237 |
* allow continuation of the {quotecontinue} token. |
| 238 |
*/ |
| 239 |
quotecontinuefail {whitespace}*"-"? |
| 240 |
|
| 241 |
/* Bit string |
| 242 |
* It is tempting to scan the string for only those characters |
| 243 |
* which are allowed. However, this leads to silently swallowed |
| 244 |
* characters if illegal characters are included in the string. |
| 245 |
* For example, if xbinside is [01] then B'ABCD' is interpreted |
| 246 |
* as a zero-length string, and the ABCD' is lost! |
| 247 |
* Better to pass the string forward and let the input routines |
| 248 |
* validate the contents. |
| 249 |
*/ |
| 250 |
xbstart [bB]{quote} |
| 251 |
xbinside [^']* |
| 252 |
|
| 253 |
/* Hexadecimal byte string */ |
| 254 |
xhstart [xX]{quote} |
| 255 |
xhinside [^']* |
| 256 |
|
| 257 |
/* National character */ |
| 258 |
xnstart [nN]{quote} |
| 259 |
|
| 260 |
/* Quoted string that allows backslash escapes */ |
| 261 |
xestart [eE]{quote} |
| 262 |
xeinside [^\\']+ |
| 263 |
xeescape [\\][^0-7] |
| 264 |
xeoctesc [\\][0-7]{1,3} |
| 265 |
xehexesc [\\]x[0-9A-Fa-f]{1,2} |
| 266 |
xeunicode [\\](u[0-9A-Fa-f]{4}|U[0-9A-Fa-f]{8}) |
| 267 |
xeunicodefail [\\](u[0-9A-Fa-f]{0,3}|U[0-9A-Fa-f]{0,7}) |
| 268 |
|
| 269 |
/* Extended quote |
| 270 |
* xqdouble implements embedded quote, '''' |
| 271 |
*/ |
| 272 |
xqstart {quote} |
| 273 |
xqdouble {quote}{quote} |
| 274 |
xqinside [^']+ |
| 275 |
|
| 276 |
/* $foo$ style quotes ("dollar quoting") |
| 277 |
* The quoted string starts with $foo$ where "foo" is an optional string |
| 278 |
* in the form of an identifier, except that it may not contain "$", |
| 279 |
* and extends to the first occurrence of an identical string. |
| 280 |
* There is *no* processing of the quoted text. |
| 281 |
* |
| 282 |
* {dolqfailed} is an error rule to avoid scanner backup when {dolqdelim} |
| 283 |
* fails to match its trailing "$". |
| 284 |
*/ |
| 285 |
dolq_start [A-Za-z\200-\377_] |
| 286 |
dolq_cont [A-Za-z\200-\377_0-9] |
| 287 |
dolqdelim \$({dolq_start}{dolq_cont}*)?\$ |
| 288 |
dolqfailed \${dolq_start}{dolq_cont}* |
| 289 |
dolqinside [^$]+ |
| 290 |
|
| 291 |
/* Double quote |
| 292 |
* Allows embedded spaces and other special characters into identifiers. |
| 293 |
*/ |
| 294 |
dquote \" |
| 295 |
xdstart {dquote} |
| 296 |
xdstop {dquote} |
| 297 |
xddouble {dquote}{dquote} |
| 298 |
xdinside [^"]+ |
| 299 |
|
| 300 |
/* Quoted identifier with Unicode escapes */ |
| 301 |
xuistart [uU]&{dquote} |
| 302 |
|
| 303 |
/* Quoted string with Unicode escapes */ |
| 304 |
xusstart [uU]&{quote} |
| 305 |
|
| 306 |
/* error rule to avoid backup */ |
| 307 |
xufailed [uU]& |
| 308 |
|
| 309 |
|
| 310 |
/* C-style comments |
| 311 |
* |
| 312 |
* The "extended comment" syntax closely resembles allowable operator syntax. |
| 313 |
* The tricky part here is to get lex to recognize a string starting with |
| 314 |
* slash-star as a comment, when interpreting it as an operator would produce |
| 315 |
* a longer match --- remember lex will prefer a longer match! Also, if we |
| 316 |
* have something like plus-slash-star, lex will think this is a 3-character |
| 317 |
* operator whereas we want to see it as a + operator and a comment start. |
| 318 |
* The solution is two-fold: |
| 319 |
* 1. append {op_chars}* to xcstart so that it matches as much text as |
| 320 |
* {operator} would. Then the tie-breaker (first matching rule of same |
| 321 |
* length) ensures xcstart wins. We put back the extra stuff with yyless() |
| 322 |
* in case it contains a star-slash that should terminate the comment. |
| 323 |
* 2. In the operator rule, check for slash-star within the operator, and |
| 324 |
* if found throw it back with yyless(). This handles the plus-slash-star |
| 325 |
* problem. |
| 326 |
* Dash-dash comments have similar interactions with the operator rule. |
| 327 |
*/ |
| 328 |
xcstart \/\*{op_chars}* |
| 329 |
xcstop \*+\/ |
| 330 |
xcinside [^*/]+ |
| 331 |
|
| 332 |
ident_start [A-Za-z\200-\377_] |
| 333 |
ident_cont [A-Za-z\200-\377_0-9\$] |
| 334 |
|
| 335 |
identifier {ident_start}{ident_cont}* |
| 336 |
|
| 337 |
/* Assorted special-case operators and operator-like tokens */ |
| 338 |
typecast "::" |
| 339 |
dot_dot \.\. |
| 340 |
colon_equals ":=" |
| 341 |
|
| 342 |
/* |
| 343 |
* These operator-like tokens (unlike the above ones) also match the {operator} |
| 344 |
* rule, which means that they might be overridden by a longer match if they |
| 345 |
* are followed by a comment start or a + or - character. Accordingly, if you |
| 346 |
* add to this list, you must also add corresponding code to the {operator} |
| 347 |
* block to return the correct token in such cases. (This is not needed in |
| 348 |
* psqlscan.l since the token value is ignored there.) |
| 349 |
*/ |
| 350 |
equals_greater "=>" |
| 351 |
less_equals "<=" |
| 352 |
greater_equals ">=" |
| 353 |
less_greater "<>" |
| 354 |
not_equals "!=" |
| 355 |
|
| 356 |
/* |
| 357 |
* "self" is the set of chars that should be returned as single-character |
| 358 |
* tokens. "op_chars" is the set of chars that can make up "Op" tokens, |
| 359 |
* which can be one or more characters long (but if a single-char token |
| 360 |
* appears in the "self" set, it is not to be returned as an Op). Note |
| 361 |
* that the sets overlap, but each has some chars that are not in the other. |
| 362 |
* |
| 363 |
* If you change either set, adjust the character lists appearing in the |
| 364 |
* rule for "operator"! |
| 365 |
*/ |
| 366 |
self [,()\[\].;\:\+\-\*\/\%\^\<\>\=] |
| 367 |
op_chars [\~\!\@\#\^\&\|\`\?\+\-\*\/\%\<\>\=] |
| 368 |
operator {op_chars}+ |
| 369 |
|
| 370 |
/* |
| 371 |
* Numbers |
| 372 |
* |
| 373 |
* Unary minus is not part of a number here. Instead we pass it separately to |
| 374 |
* the parser, and there it gets coerced via doNegate(). |
| 375 |
* |
| 376 |
* {numericfail} is used because we would like "1..10" to lex as 1, dot_dot, 10. |
| 377 |
* |
| 378 |
* {realfail} is added to prevent the need for scanner |
| 379 |
* backup when the {real} rule fails to match completely. |
| 380 |
*/ |
| 381 |
decdigit [0-9] |
| 382 |
hexdigit [0-9A-Fa-f] |
| 383 |
octdigit [0-7] |
| 384 |
bindigit [0-1] |
| 385 |
|
| 386 |
decinteger {decdigit}(_?{decdigit})* |
| 387 |
hexinteger 0[xX](_?{hexdigit})+ |
| 388 |
octinteger 0[oO](_?{octdigit})+ |
| 389 |
bininteger 0[bB](_?{bindigit})+ |
| 390 |
|
| 391 |
hexfail 0[xX]_? |
| 392 |
octfail 0[oO]_? |
| 393 |
binfail 0[bB]_? |
| 394 |
|
| 395 |
numeric (({decinteger}\.{decinteger}?)|(\.{decinteger})) |
| 396 |
numericfail {decinteger}\.\. |
| 397 |
|
| 398 |
real ({decinteger}|{numeric})[Ee][-+]?{decinteger} |
| 399 |
realfail ({decinteger}|{numeric})[Ee][-+] |
| 400 |
|
| 401 |
/* Positional parameters don't accept underscores. */ |
| 402 |
param \${decdigit}+ |
| 403 |
|
| 404 |
/* |
| 405 |
* An identifier immediately following an integer literal is disallowed because |
| 406 |
* in some cases it's ambiguous what is meant: for example, 0x1234 could be |
| 407 |
* either a hexinteger or a decinteger "0" and an identifier "x1234". We can |
| 408 |
* detect such problems by seeing if integer_junk matches a longer substring |
| 409 |
* than any of the XXXinteger patterns (decinteger, hexinteger, octinteger, |
| 410 |
* bininteger). One "junk" pattern is sufficient because |
| 411 |
* {decinteger}{identifier} will match all the same strings we'd match with |
| 412 |
* {hexinteger}{identifier} etc. |
| 413 |
* |
| 414 |
* Note that the rule for integer_junk must appear after the ones for |
| 415 |
* XXXinteger to make this work correctly: 0x1234 will match both hexinteger |
| 416 |
* and integer_junk, and we need hexinteger to be chosen in that case. |
| 417 |
* |
| 418 |
* Also disallow strings matched by numeric_junk, real_junk and param_junk |
| 419 |
* for consistency. |
| 420 |
*/ |
| 421 |
integer_junk {decinteger}{identifier} |
| 422 |
numeric_junk {numeric}{identifier} |
| 423 |
real_junk {real}{identifier} |
| 424 |
param_junk \${decdigit}+{identifier} |
| 425 |
|
| 426 |
other . |
| 427 |
|
| 428 |
/* |
| 429 |
* Dollar quoted strings are totally opaque, and no escaping is done on them. |
| 430 |
* Other quoted strings must allow some special characters such as single-quote |
| 431 |
* and newline. |
| 432 |
* Embedded single-quotes are implemented both in the SQL standard |
| 433 |
* style of two adjacent single quotes "''" and in the Postgres/Java style |
| 434 |
* of escaped-quote "\'". |
| 435 |
* Other embedded escaped characters are matched explicitly and the leading |
| 436 |
* backslash is dropped from the string. |
| 437 |
* Note that xcstart must appear before operator, as explained above! |
| 438 |
* Also whitespace (comment) must appear before operator. |
| 439 |
*/ |
| 440 |
|
| 441 |
%% |
| 442 |
|
| 443 |
{whitespace} { |
| 444 |
/* ignore */ |
| 445 |
} |
| 446 |
|
| 447 |
{xcstart} { |
| 448 |
/* Set location in case of syntax error in comment */ |
| 449 |
SET_YYLLOC(); |
| 450 |
yyextra->xcdepth = 0; |
| 451 |
BEGIN(xc); |
| 452 |
/* Put back any characters past slash-star; see above */ |
| 453 |
yyless(2); |
| 454 |
} |
| 455 |
|
| 456 |
<xc>{ |
| 457 |
{xcstart} { |
| 458 |
(yyextra->xcdepth)++; |
| 459 |
/* Put back any characters past slash-star; see above */ |
| 460 |
yyless(2); |
| 461 |
} |
| 462 |
|
| 463 |
{xcstop} { |
| 464 |
if (yyextra->xcdepth <= 0) |
| 465 |
BEGIN(INITIAL); |
| 466 |
else |
| 467 |
(yyextra->xcdepth)--; |
| 468 |
} |
| 469 |
|
| 470 |
{xcinside} { |
| 471 |
/* ignore */ |
| 472 |
} |
| 473 |
|
| 474 |
{op_chars} { |
| 475 |
/* ignore */ |
| 476 |
} |
| 477 |
|
| 478 |
\*+ { |
| 479 |
/* ignore */ |
| 480 |
} |
| 481 |
|
| 482 |
<<EOF>> { |
| 483 |
yyerror("unterminated /* comment"); |
| 484 |
} |
| 485 |
} /* <xc> */ |
| 486 |
|
| 487 |
{xbstart} { |
| 488 |
/* Binary bit type. |
| 489 |
* At some point we should simply pass the string |
| 490 |
* forward to the parser and label it there. |
| 491 |
* In the meantime, place a leading "b" on the string |
| 492 |
* to mark it for the input routine as a binary string. |
| 493 |
*/ |
| 494 |
SET_YYLLOC(); |
| 495 |
BEGIN(xb); |
| 496 |
startlit(); |
| 497 |
addlitchar('b', yyscanner); |
| 498 |
} |
| 499 |
<xh>{xhinside} | |
| 500 |
<xb>{xbinside} { |
| 501 |
addlit(yytext, yyleng, yyscanner); |
| 502 |
} |
| 503 |
<xb><<EOF>> { yyerror("unterminated bit string literal"); } |
| 504 |
|
| 505 |
{xhstart} { |
| 506 |
/* Hexadecimal bit type. |
| 507 |
* At some point we should simply pass the string |
| 508 |
* forward to the parser and label it there. |
| 509 |
* In the meantime, place a leading "x" on the string |
| 510 |
* to mark it for the input routine as a hex string. |
| 511 |
*/ |
| 512 |
SET_YYLLOC(); |
| 513 |
BEGIN(xh); |
| 514 |
startlit(); |
| 515 |
addlitchar('x', yyscanner); |
| 516 |
} |
| 517 |
<xh><<EOF>> { yyerror("unterminated hexadecimal string literal"); } |
| 518 |
|
| 519 |
{xnstart} { |
| 520 |
/* National character. |
| 521 |
* We will pass this along as a normal character string, |
| 522 |
* but preceded with an internally-generated "NCHAR". |
| 523 |
*/ |
| 524 |
int kwnum; |
| 525 |
|
| 526 |
SET_YYLLOC(); |
| 527 |
yyless(1); /* eat only 'n' this time */ |
| 528 |
|
| 529 |
kwnum = ScanKeywordLookup("nchar", |
| 530 |
yyextra->keywordlist); |
| 531 |
if (kwnum >= 0) |
| 532 |
{ |
| 533 |
yylval->keyword = GetScanKeyword(kwnum, |
| 534 |
yyextra->keywordlist); |
| 535 |
return yyextra->keyword_tokens[kwnum]; |
| 536 |
} |
| 537 |
else |
| 538 |
{ |
| 539 |
/* If NCHAR isn't a keyword, just return "n" */ |
| 540 |
yylval->str = pstrdup("n"); |
| 541 |
return IDENT; |
| 542 |
} |
| 543 |
} |
| 544 |
|
| 545 |
{xqstart} { |
| 546 |
yyextra->warn_on_first_escape = true; |
| 547 |
yyextra->saw_non_ascii = false; |
| 548 |
SET_YYLLOC(); |
| 549 |
if (yyextra->standard_conforming_strings) |
| 550 |
BEGIN(xq); |
| 551 |
else |
| 552 |
BEGIN(xe); |
| 553 |
startlit(); |
| 554 |
} |
| 555 |
{xestart} { |
| 556 |
yyextra->warn_on_first_escape = false; |
| 557 |
yyextra->saw_non_ascii = false; |
| 558 |
SET_YYLLOC(); |
| 559 |
BEGIN(xe); |
| 560 |
startlit(); |
| 561 |
} |
| 562 |
{xusstart} { |
| 563 |
SET_YYLLOC(); |
| 564 |
if (!yyextra->standard_conforming_strings) |
| 565 |
ereport(ERROR, |
| 566 |
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED), |
| 567 |
errmsg("unsafe use of string constant with Unicode escapes"), |
| 568 |
errdetail("String constants with Unicode escapes cannot be used when \"standard_conforming_strings\" is off."), |
| 569 |
lexer_errposition())); |
| 570 |
BEGIN(xus); |
| 571 |
startlit(); |
| 572 |
} |
| 573 |
|
| 574 |
<xb,xh,xq,xe,xus>{quote} { |
| 575 |
/* |
| 576 |
* When we are scanning a quoted string and see an end |
| 577 |
* quote, we must look ahead for a possible continuation. |
| 578 |
* If we don't see one, we know the end quote was in fact |
| 579 |
* the end of the string. To reduce the lexer table size, |
| 580 |
* we use a single "xqs" state to do the lookahead for all |
| 581 |
* types of strings. |
| 582 |
*/ |
| 583 |
yyextra->state_before_str_stop = YYSTATE; |
| 584 |
BEGIN(xqs); |
| 585 |
} |
| 586 |
<xqs>{quotecontinue} { |
| 587 |
/* |
| 588 |
* Found a quote continuation, so return to the in-quote |
| 589 |
* state and continue scanning the literal. Nothing is |
| 590 |
* added to the literal's contents. |
| 591 |
*/ |
| 592 |
BEGIN(yyextra->state_before_str_stop); |
| 593 |
} |
| 594 |
<xqs>{quotecontinuefail} | |
| 595 |
<xqs>{other} | |
| 596 |
<xqs><<EOF>> { |
| 597 |
/* |
| 598 |
* Failed to see a quote continuation. Throw back |
| 599 |
* everything after the end quote, and handle the string |
| 600 |
* according to the state we were in previously. |
| 601 |
*/ |
| 602 |
yyless(0); |
| 603 |
BEGIN(INITIAL); |
| 604 |
|
| 605 |
switch (yyextra->state_before_str_stop) |
| 606 |
{ |
| 607 |
case xb: |
| 608 |
yylval->str = litbufdup(yyscanner); |
| 609 |
return BCONST; |
| 610 |
case xh: |
| 611 |
yylval->str = litbufdup(yyscanner); |
| 612 |
return XCONST; |
| 613 |
case xq: |
| 614 |
case xe: |
| 615 |
/* |
| 616 |
* Check that the data remains valid, if it might |
| 617 |
* have been made invalid by unescaping any chars. |
| 618 |
*/ |
| 619 |
if (yyextra->saw_non_ascii) |
| 620 |
pg_verifymbstr(yyextra->literalbuf, |
| 621 |
yyextra->literallen, |
| 622 |
false); |
| 623 |
yylval->str = litbufdup(yyscanner); |
| 624 |
return SCONST; |
| 625 |
case xus: |
| 626 |
yylval->str = litbufdup(yyscanner); |
| 627 |
return USCONST; |
| 628 |
default: |
| 629 |
yyerror("unhandled previous state in xqs"); |
| 630 |
} |
| 631 |
} |
| 632 |
|
| 633 |
<xq,xe,xus>{xqdouble} { |
| 634 |
addlitchar('\'', yyscanner); |
| 635 |
} |
| 636 |
<xq,xus>{xqinside} { |
| 637 |
addlit(yytext, yyleng, yyscanner); |
| 638 |
} |
| 639 |
<xe>{xeinside} { |
| 640 |
addlit(yytext, yyleng, yyscanner); |
| 641 |
} |
| 642 |
<xe>{xeunicode} { |
| 643 |
pg_wchar c = strtoul(yytext + 2, NULL, 16); |
| 644 |
|
| 645 |
/* |
| 646 |
* For consistency with other productions, issue any |
| 647 |
* escape warning with cursor pointing to start of string. |
| 648 |
* We might want to change that, someday. |
| 649 |
*/ |
| 650 |
check_escape_warning(yyscanner); |
| 651 |
|
| 652 |
/* Remember start of overall string token ... */ |
| 653 |
PUSH_YYLLOC(); |
| 654 |
/* ... and set the error cursor to point at this esc seq */ |
| 655 |
SET_YYLLOC(); |
| 656 |
|
| 657 |
if (is_utf16_surrogate_first(c)) |
| 658 |
{ |
| 659 |
yyextra->utf16_first_part = c; |
| 660 |
BEGIN(xeu); |
| 661 |
} |
| 662 |
else if (is_utf16_surrogate_second(c)) |
| 663 |
yyerror("invalid Unicode surrogate pair"); |
| 664 |
else |
| 665 |
addunicode(c, yyscanner); |
| 666 |
|
| 667 |
/* Restore yylloc to be start of string token */ |
| 668 |
POP_YYLLOC(); |
| 669 |
} |
| 670 |
<xeu>{xeunicode} { |
| 671 |
pg_wchar c = strtoul(yytext + 2, NULL, 16); |
| 672 |
|
| 673 |
/* Remember start of overall string token ... */ |
| 674 |
PUSH_YYLLOC(); |
| 675 |
/* ... and set the error cursor to point at this esc seq */ |
| 676 |
SET_YYLLOC(); |
| 677 |
|
| 678 |
if (!is_utf16_surrogate_second(c)) |
| 679 |
yyerror("invalid Unicode surrogate pair"); |
| 680 |
|
| 681 |
c = surrogate_pair_to_codepoint(yyextra->utf16_first_part, c); |
| 682 |
|
| 683 |
addunicode(c, yyscanner); |
| 684 |
|
| 685 |
/* Restore yylloc to be start of string token */ |
| 686 |
POP_YYLLOC(); |
| 687 |
|
| 688 |
BEGIN(xe); |
| 689 |
} |
| 690 |
<xeu>. | |
| 691 |
<xeu>\n | |
| 692 |
<xeu><<EOF>> { |
| 693 |
/* Set the error cursor to point at missing esc seq */ |
| 694 |
SET_YYLLOC(); |
| 695 |
yyerror("invalid Unicode surrogate pair"); |
| 696 |
} |
| 697 |
<xe,xeu>{xeunicodefail} { |
| 698 |
/* Set the error cursor to point at malformed esc seq */ |
| 699 |
SET_YYLLOC(); |
| 700 |
ereport(ERROR, |
| 701 |
(errcode(ERRCODE_INVALID_ESCAPE_SEQUENCE), |
| 702 |
errmsg("invalid Unicode escape"), |
| 703 |
errhint("Unicode escapes must be \\uXXXX or \\UXXXXXXXX."), |
| 704 |
lexer_errposition())); |
| 705 |
} |
| 706 |
<xe>{xeescape} { |
| 707 |
if (yytext[1] == '\'') |
| 708 |
{ |
| 709 |
if (yyextra->backslash_quote == BACKSLASH_QUOTE_OFF || |
| 710 |
(yyextra->backslash_quote == BACKSLASH_QUOTE_SAFE_ENCODING && |
| 711 |
PG_ENCODING_IS_CLIENT_ONLY(pg_get_client_encoding()))) |
| 712 |
ereport(ERROR, |
| 713 |
(errcode(ERRCODE_NONSTANDARD_USE_OF_ESCAPE_CHARACTER), |
| 714 |
errmsg("unsafe use of \\' in a string literal"), |
| 715 |
errhint("Use '' to write quotes in strings. \\' is insecure in client-only encodings."), |
| 716 |
lexer_errposition())); |
| 717 |
} |
| 718 |
check_string_escape_warning(yytext[1], yyscanner); |
| 719 |
addlitchar(unescape_single_char(yytext[1], yyscanner), |
| 720 |
yyscanner); |
| 721 |
} |
| 722 |
<xe>{xeoctesc} { |
| 723 |
unsigned char c = strtoul(yytext + 1, NULL, 8); |
| 724 |
|
| 725 |
check_escape_warning(yyscanner); |
| 726 |
addlitchar(c, yyscanner); |
| 727 |
if (c == '\0' || IS_HIGHBIT_SET(c)) |
| 728 |
yyextra->saw_non_ascii = true; |
| 729 |
} |
| 730 |
<xe>{xehexesc} { |
| 731 |
unsigned char c = strtoul(yytext + 2, NULL, 16); |
| 732 |
|
| 733 |
check_escape_warning(yyscanner); |
| 734 |
addlitchar(c, yyscanner); |
| 735 |
if (c == '\0' || IS_HIGHBIT_SET(c)) |
| 736 |
yyextra->saw_non_ascii = true; |
| 737 |
} |
| 738 |
<xe>. { |
| 739 |
/* This is only needed for \ just before EOF */ |
| 740 |
addlitchar(yytext[0], yyscanner); |
| 741 |
} |
| 742 |
<xq,xe,xus><<EOF>> { yyerror("unterminated quoted string"); } |
| 743 |
|
| 744 |
{dolqdelim} { |
| 745 |
SET_YYLLOC(); |
| 746 |
yyextra->dolqstart = pstrdup(yytext); |
| 747 |
BEGIN(xdolq); |
| 748 |
startlit(); |
| 749 |
} |
| 750 |
{dolqfailed} { |
| 751 |
SET_YYLLOC(); |
| 752 |
/* throw back all but the initial "$" */ |
| 753 |
yyless(1); |
| 754 |
/* and treat it as {other} */ |
| 755 |
return yytext[0]; |
| 756 |
} |
| 757 |
<xdolq>{dolqdelim} { |
| 758 |
if (strcmp(yytext, yyextra->dolqstart) == 0) |
| 759 |
{ |
| 760 |
pfree(yyextra->dolqstart); |
| 761 |
yyextra->dolqstart = NULL; |
| 762 |
BEGIN(INITIAL); |
| 763 |
yylval->str = litbufdup(yyscanner); |
| 764 |
return SCONST; |
| 765 |
} |
| 766 |
else |
| 767 |
{ |
| 768 |
/* |
| 769 |
* When we fail to match $...$ to dolqstart, transfer |
| 770 |
* the $... part to the output, but put back the final |
| 771 |
* $ for rescanning. Consider $delim$...$junk$delim$ |
| 772 |
*/ |
| 773 |
addlit(yytext, yyleng - 1, yyscanner); |
| 774 |
yyless(yyleng - 1); |
| 775 |
} |
| 776 |
} |
| 777 |
<xdolq>{dolqinside} { |
| 778 |
addlit(yytext, yyleng, yyscanner); |
| 779 |
} |
| 780 |
<xdolq>{dolqfailed} { |
| 781 |
addlit(yytext, yyleng, yyscanner); |
| 782 |
} |
| 783 |
<xdolq>. { |
| 784 |
/* This is only needed for $ inside the quoted text */ |
| 785 |
addlitchar(yytext[0], yyscanner); |
| 786 |
} |
| 787 |
<xdolq><<EOF>> { yyerror("unterminated dollar-quoted string"); } |
| 788 |
|
| 789 |
{xdstart} { |
| 790 |
SET_YYLLOC(); |
| 791 |
BEGIN(xd); |
| 792 |
startlit(); |
| 793 |
} |
| 794 |
{xuistart} { |
| 795 |
SET_YYLLOC(); |
| 796 |
BEGIN(xui); |
| 797 |
startlit(); |
| 798 |
} |
| 799 |
<xd>{xdstop} { |
| 800 |
char *ident; |
| 801 |
|
| 802 |
BEGIN(INITIAL); |
| 803 |
if (yyextra->literallen == 0) |
| 804 |
yyerror("zero-length delimited identifier"); |
| 805 |
ident = litbufdup(yyscanner); |
| 806 |
if (yyextra->literallen >= NAMEDATALEN) |
| 807 |
truncate_identifier(ident, yyextra->literallen, true); |
| 808 |
yylval->str = ident; |
| 809 |
return IDENT; |
| 810 |
} |
| 811 |
<xui>{dquote} { |
| 812 |
BEGIN(INITIAL); |
| 813 |
if (yyextra->literallen == 0) |
| 814 |
yyerror("zero-length delimited identifier"); |
| 815 |
/* can't truncate till after we de-escape the ident */ |
| 816 |
yylval->str = litbufdup(yyscanner); |
| 817 |
return UIDENT; |
| 818 |
} |
| 819 |
<xd,xui>{xddouble} { |
| 820 |
addlitchar('"', yyscanner); |
| 821 |
} |
| 822 |
<xd,xui>{xdinside} { |
| 823 |
addlit(yytext, yyleng, yyscanner); |
| 824 |
} |
| 825 |
<xd,xui><<EOF>> { yyerror("unterminated quoted identifier"); } |
| 826 |
|
| 827 |
{xufailed} { |
| 828 |
char *ident; |
| 829 |
|
| 830 |
SET_YYLLOC(); |
| 831 |
/* throw back all but the initial u/U */ |
| 832 |
yyless(1); |
| 833 |
/* and treat it as {identifier} */ |
| 834 |
ident = downcase_truncate_identifier(yytext, yyleng, true); |
| 835 |
yylval->str = ident; |
| 836 |
return IDENT; |
| 837 |
} |
| 838 |
|
| 839 |
{typecast} { |
| 840 |
SET_YYLLOC(); |
| 841 |
return TYPECAST; |
| 842 |
} |
| 843 |
|
| 844 |
{dot_dot} { |
| 845 |
SET_YYLLOC(); |
| 846 |
return DOT_DOT; |
| 847 |
} |
| 848 |
|
| 849 |
{colon_equals} { |
| 850 |
SET_YYLLOC(); |
| 851 |
return COLON_EQUALS; |
| 852 |
} |
| 853 |
|
| 854 |
{equals_greater} { |
| 855 |
SET_YYLLOC(); |
| 856 |
return EQUALS_GREATER; |
| 857 |
} |
| 858 |
|
| 859 |
{less_equals} { |
| 860 |
SET_YYLLOC(); |
| 861 |
return LESS_EQUALS; |
| 862 |
} |
| 863 |
|
| 864 |
{greater_equals} { |
| 865 |
SET_YYLLOC(); |
| 866 |
return GREATER_EQUALS; |
| 867 |
} |
| 868 |
|
| 869 |
{less_greater} { |
| 870 |
/* We accept both "<>" and "!=" as meaning NOT_EQUALS */ |
| 871 |
SET_YYLLOC(); |
| 872 |
return NOT_EQUALS; |
| 873 |
} |
| 874 |
|
| 875 |
{not_equals} { |
| 876 |
/* We accept both "<>" and "!=" as meaning NOT_EQUALS */ |
| 877 |
SET_YYLLOC(); |
| 878 |
return NOT_EQUALS; |
| 879 |
} |
| 880 |
|
| 881 |
{self} { |
| 882 |
SET_YYLLOC(); |
| 883 |
return yytext[0]; |
| 884 |
} |
| 885 |
|
| 886 |
{operator} { |
| 887 |
/* |
| 888 |
* Check for embedded slash-star or dash-dash; those |
| 889 |
* are comment starts, so operator must stop there. |
| 890 |
* Note that slash-star or dash-dash at the first |
| 891 |
* character will match a prior rule, not this one. |
| 892 |
*/ |
| 893 |
int nchars = yyleng; |
| 894 |
char *slashstar = strstr(yytext, "/*"); |
| 895 |
char *dashdash = strstr(yytext, "--"); |
| 896 |
|
| 897 |
if (slashstar && dashdash) |
| 898 |
{ |
| 899 |
/* if both appear, take the first one */ |
| 900 |
if (slashstar > dashdash) |
| 901 |
slashstar = dashdash; |
| 902 |
} |
| 903 |
else if (!slashstar) |
| 904 |
slashstar = dashdash; |
| 905 |
if (slashstar) |
| 906 |
nchars = slashstar - yytext; |
| 907 |
|
| 908 |
/* |
| 909 |
* For SQL compatibility, '+' and '-' cannot be the |
| 910 |
* last char of a multi-char operator unless the operator |
| 911 |
* contains chars that are not in SQL operators. |
| 912 |
* The idea is to lex '=-' as two operators, but not |
| 913 |
* to forbid operator names like '?-' that could not be |
| 914 |
* sequences of SQL operators. |
| 915 |
*/ |
| 916 |
if (nchars > 1 && |
| 917 |
(yytext[nchars - 1] == '+' || |
| 918 |
yytext[nchars - 1] == '-')) |
| 919 |
{ |
| 920 |
int ic; |
| 921 |
|
| 922 |
for (ic = nchars - 2; ic >= 0; ic--) |
| 923 |
{ |
| 924 |
char c = yytext[ic]; |
| 925 |
if (c == '~' || c == '!' || c == '@' || |
| 926 |
c == '#' || c == '^' || c == '&' || |
| 927 |
c == '|' || c == '`' || c == '?' || |
| 928 |
c == '%') |
| 929 |
break; |
| 930 |
} |
| 931 |
if (ic < 0) |
| 932 |
{ |
| 933 |
/* |
| 934 |
* didn't find a qualifying character, so remove |
| 935 |
* all trailing [+-] |
| 936 |
*/ |
| 937 |
do { |
| 938 |
nchars--; |
| 939 |
} while (nchars > 1 && |
| 940 |
(yytext[nchars - 1] == '+' || |
| 941 |
yytext[nchars - 1] == '-')); |
| 942 |
} |
| 943 |
} |
| 944 |
|
| 945 |
SET_YYLLOC(); |
| 946 |
|
| 947 |
if (nchars < yyleng) |
| 948 |
{ |
| 949 |
/* Strip the unwanted chars from the token */ |
| 950 |
yyless(nchars); |
| 951 |
/* |
| 952 |
* If what we have left is only one char, and it's |
| 953 |
* one of the characters matching "self", then |
| 954 |
* return it as a character token the same way |
| 955 |
* that the "self" rule would have. |
| 956 |
*/ |
| 957 |
if (nchars == 1 && |
| 958 |
strchr(",()[].;:+-*/%^<>=", yytext[0])) |
| 959 |
return yytext[0]; |
| 960 |
/* |
| 961 |
* Likewise, if what we have left is two chars, and |
| 962 |
* those match the tokens ">=", "<=", "=>", "<>" or |
| 963 |
* "!=", then we must return the appropriate token |
| 964 |
* rather than the generic Op. |
| 965 |
*/ |
| 966 |
if (nchars == 2) |
| 967 |
{ |
| 968 |
if (yytext[0] == '=' && yytext[1] == '>') |
| 969 |
return EQUALS_GREATER; |
| 970 |
if (yytext[0] == '>' && yytext[1] == '=') |
| 971 |
return GREATER_EQUALS; |
| 972 |
if (yytext[0] == '<' && yytext[1] == '=') |
| 973 |
return LESS_EQUALS; |
| 974 |
if (yytext[0] == '<' && yytext[1] == '>') |
| 975 |
return NOT_EQUALS; |
| 976 |
if (yytext[0] == '!' && yytext[1] == '=') |
| 977 |
return NOT_EQUALS; |
| 978 |
} |
| 979 |
} |
| 980 |
|
| 981 |
/* |
| 982 |
* Complain if operator is too long. Unlike the case |
| 983 |
* for identifiers, we make this an error not a notice- |
| 984 |
* and-truncate, because the odds are we are looking at |
| 985 |
* a syntactic mistake anyway. |
| 986 |
*/ |
| 987 |
if (nchars >= NAMEDATALEN) |
| 988 |
yyerror("operator too long"); |
| 989 |
|
| 990 |
yylval->str = pstrdup(yytext); |
| 991 |
return Op; |
| 992 |
} |
| 993 |
|
| 994 |
{param} { |
| 995 |
ErrorSaveContext escontext = {T_ErrorSaveContext}; |
| 996 |
int32 val; |
| 997 |
|
| 998 |
SET_YYLLOC(); |
| 999 |
val = pg_strtoint32_safe(yytext + 1, (Node *) &escontext); |
| 1000 |
if (escontext.error_occurred) |
| 1001 |
yyerror("parameter number too large"); |
| 1002 |
yylval->ival = val; |
| 1003 |
return PARAM; |
| 1004 |
} |
| 1005 |
{param_junk} { |
| 1006 |
SET_YYLLOC(); |
| 1007 |
yyerror("trailing junk after parameter"); |
| 1008 |
} |
| 1009 |
|
| 1010 |
{decinteger} { |
| 1011 |
SET_YYLLOC(); |
| 1012 |
return process_integer_literal(yytext, yylval, 10); |
| 1013 |
} |
| 1014 |
{hexinteger} { |
| 1015 |
SET_YYLLOC(); |
| 1016 |
return process_integer_literal(yytext, yylval, 16); |
| 1017 |
} |
| 1018 |
{octinteger} { |
| 1019 |
SET_YYLLOC(); |
| 1020 |
return process_integer_literal(yytext, yylval, 8); |
| 1021 |
} |
| 1022 |
{bininteger} { |
| 1023 |
SET_YYLLOC(); |
| 1024 |
return process_integer_literal(yytext, yylval, 2); |
| 1025 |
} |
| 1026 |
{hexfail} { |
| 1027 |
SET_YYLLOC(); |
| 1028 |
yyerror("invalid hexadecimal integer"); |
| 1029 |
} |
| 1030 |
{octfail} { |
| 1031 |
SET_YYLLOC(); |
| 1032 |
yyerror("invalid octal integer"); |
| 1033 |
} |
| 1034 |
{binfail} { |
| 1035 |
SET_YYLLOC(); |
| 1036 |
yyerror("invalid binary integer"); |
| 1037 |
} |
| 1038 |
{numeric} { |
| 1039 |
SET_YYLLOC(); |
| 1040 |
yylval->str = pstrdup(yytext); |
| 1041 |
return FCONST; |
| 1042 |
} |
| 1043 |
{numericfail} { |
| 1044 |
/* throw back the .., and treat as integer */ |
| 1045 |
yyless(yyleng - 2); |
| 1046 |
SET_YYLLOC(); |
| 1047 |
return process_integer_literal(yytext, yylval, 10); |
| 1048 |
} |
| 1049 |
{real} { |
| 1050 |
SET_YYLLOC(); |
| 1051 |
yylval->str = pstrdup(yytext); |
| 1052 |
return FCONST; |
| 1053 |
} |
| 1054 |
{realfail} { |
| 1055 |
SET_YYLLOC(); |
| 1056 |
yyerror("trailing junk after numeric literal"); |
| 1057 |
} |
| 1058 |
{integer_junk} { |
| 1059 |
SET_YYLLOC(); |
| 1060 |
yyerror("trailing junk after numeric literal"); |
| 1061 |
} |
| 1062 |
{numeric_junk} { |
| 1063 |
SET_YYLLOC(); |
| 1064 |
yyerror("trailing junk after numeric literal"); |
| 1065 |
} |
| 1066 |
{real_junk} { |
| 1067 |
SET_YYLLOC(); |
| 1068 |
yyerror("trailing junk after numeric literal"); |
| 1069 |
} |
| 1070 |
|
| 1071 |
|
| 1072 |
{identifier} { |
| 1073 |
int kwnum; |
| 1074 |
char *ident; |
| 1075 |
|
| 1076 |
SET_YYLLOC(); |
| 1077 |
|
| 1078 |
/* Is it a keyword? */ |
| 1079 |
kwnum = ScanKeywordLookup(yytext, |
| 1080 |
yyextra->keywordlist); |
| 1081 |
if (kwnum >= 0) |
| 1082 |
{ |
| 1083 |
yylval->keyword = GetScanKeyword(kwnum, |
| 1084 |
yyextra->keywordlist); |
| 1085 |
return yyextra->keyword_tokens[kwnum]; |
| 1086 |
} |
| 1087 |
|
| 1088 |
/* |
| 1089 |
* No. Convert the identifier to lower case, and truncate |
| 1090 |
* if necessary. |
| 1091 |
*/ |
| 1092 |
ident = downcase_truncate_identifier(yytext, yyleng, true); |
| 1093 |
yylval->str = ident; |
| 1094 |
return IDENT; |
| 1095 |
} |
| 1096 |
|
| 1097 |
{other} { |
| 1098 |
SET_YYLLOC(); |
| 1099 |
return yytext[0]; |
| 1100 |
} |
| 1101 |
|
| 1102 |
<<EOF>> { |
| 1103 |
SET_YYLLOC(); |
| 1104 |
yyterminate(); |
| 1105 |
} |
| 1106 |
|
| 1107 |
%% |
| 1108 |
|
| 1109 |
/* LCOV_EXCL_STOP */ |
| 1110 |
|
| 1111 |
/* |
| 1112 |
* Arrange access to yyextra for subroutines of the main yylex() function. |
| 1113 |
* We expect each subroutine to have a yyscanner parameter. Rather than |
| 1114 |
* use the yyget_xxx functions, which might or might not get inlined by the |
| 1115 |
* compiler, we cheat just a bit and cast yyscanner to the right type. |
| 1116 |
*/ |
| 1117 |
#undef yyextra |
| 1118 |
#define yyextra (((struct yyguts_t *) yyscanner)->yyextra_r) |
| 1119 |
|
| 1120 |
/* Likewise for a couple of other things we need. */ |
| 1121 |
#undef yylloc |
| 1122 |
#define yylloc (((struct yyguts_t *) yyscanner)->yylloc_r) |
| 1123 |
#undef yyleng |
| 1124 |
#define yyleng (((struct yyguts_t *) yyscanner)->yyleng_r) |
| 1125 |
|
| 1126 |
|
| 1127 |
/* |
| 1128 |
* scanner_errposition |
| 1129 |
* Report a lexer or grammar error cursor position, if possible. |
| 1130 |
* |
| 1131 |
* This is expected to be used within an ereport() call, or via an error |
| 1132 |
* callback such as setup_scanner_errposition_callback(). The return value |
| 1133 |
* is a dummy (always 0, in fact). |
| 1134 |
* |
| 1135 |
* Note that this can only be used for messages emitted during raw parsing |
| 1136 |
* (essentially, scan.l, parser.c, and gram.y), since it requires the |
| 1137 |
* yyscanner struct to still be available. |
| 1138 |
*/ |
| 1139 |
int |
| 1140 |
scanner_errposition(int location, core_yyscan_t yyscanner) |
| 1141 |
{ |
| 1142 |
int pos; |
| 1143 |
|
| 1144 |
if (location < 0) |
| 1145 |
return 0; /* no-op if location is unknown */ |
| 1146 |
|
| 1147 |
/* Convert byte offset to character number */ |
| 1148 |
pos = pg_mbstrlen_with_len(yyextra->scanbuf, location) + 1; |
| 1149 |
/* And pass it to the ereport mechanism */ |
| 1150 |
return errposition(pos); |
| 1151 |
} |
| 1152 |
|
| 1153 |
/* |
| 1154 |
* Error context callback for inserting scanner error location. |
| 1155 |
* |
| 1156 |
* Note that this will be called for *any* error occurring while the |
| 1157 |
* callback is installed. We avoid inserting an irrelevant error location |
| 1158 |
* if the error is a query cancel --- are there any other important cases? |
| 1159 |
*/ |
| 1160 |
static void |
| 1161 |
scb_error_callback(void *arg) |
| 1162 |
{ |
| 1163 |
ScannerCallbackState *scbstate = (ScannerCallbackState *) arg; |
| 1164 |
|
| 1165 |
if (geterrcode() != ERRCODE_QUERY_CANCELED) |
| 1166 |
(void) scanner_errposition(scbstate->location, scbstate->yyscanner); |
| 1167 |
} |
| 1168 |
|
| 1169 |
/* |
| 1170 |
* setup_scanner_errposition_callback |
| 1171 |
* Arrange for non-scanner errors to report an error position |
| 1172 |
* |
| 1173 |
* Sometimes the scanner calls functions that aren't part of the scanner |
| 1174 |
* subsystem and can't reasonably be passed the yyscanner pointer; yet |
| 1175 |
* we would like any errors thrown in those functions to be tagged with an |
| 1176 |
* error location. Use this function to set up an error context stack |
| 1177 |
* entry that will accomplish that. Usage pattern: |
| 1178 |
* |
| 1179 |
* declare a local variable "ScannerCallbackState scbstate" |
| 1180 |
* ... |
| 1181 |
* setup_scanner_errposition_callback(&scbstate, yyscanner, location); |
| 1182 |
* call function that might throw error; |
| 1183 |
* cancel_scanner_errposition_callback(&scbstate); |
| 1184 |
*/ |
| 1185 |
void |
| 1186 |
setup_scanner_errposition_callback(ScannerCallbackState *scbstate, |
| 1187 |
core_yyscan_t yyscanner, |
| 1188 |
int location) |
| 1189 |
{ |
| 1190 |
/* Setup error traceback support for ereport() */ |
| 1191 |
scbstate->yyscanner = yyscanner; |
| 1192 |
scbstate->location = location; |
| 1193 |
scbstate->errcallback.callback = scb_error_callback; |
| 1194 |
scbstate->errcallback.arg = scbstate; |
| 1195 |
scbstate->errcallback.previous = error_context_stack; |
| 1196 |
error_context_stack = &scbstate->errcallback; |
| 1197 |
} |
| 1198 |
|
| 1199 |
/* |
| 1200 |
* Cancel a previously-set-up errposition callback. |
| 1201 |
*/ |
| 1202 |
void |
| 1203 |
cancel_scanner_errposition_callback(ScannerCallbackState *scbstate) |
| 1204 |
{ |
| 1205 |
/* Pop the error context stack */ |
| 1206 |
error_context_stack = scbstate->errcallback.previous; |
| 1207 |
} |
| 1208 |
|
| 1209 |
/* |
| 1210 |
* scanner_yyerror |
| 1211 |
* Report a lexer or grammar error. |
| 1212 |
* |
| 1213 |
* The message's cursor position is whatever YYLLOC was last set to, |
| 1214 |
* ie, the start of the current token if called within yylex(), or the |
| 1215 |
* most recently lexed token if called from the grammar. |
| 1216 |
* This is OK for syntax error messages from the Bison parser, because Bison |
| 1217 |
* parsers report error as soon as the first unparsable token is reached. |
| 1218 |
* Beware of using yyerror for other purposes, as the cursor position might |
| 1219 |
* be misleading! |
| 1220 |
*/ |
| 1221 |
void |
| 1222 |
scanner_yyerror(const char *message, core_yyscan_t yyscanner) |
| 1223 |
{ |
| 1224 |
const char *loc = yyextra->scanbuf + *yylloc; |
| 1225 |
|
| 1226 |
if (*loc == YY_END_OF_BUFFER_CHAR) |
| 1227 |
{ |
| 1228 |
ereport(ERROR, |
| 1229 |
(errcode(ERRCODE_SYNTAX_ERROR), |
| 1230 |
/* translator: %s is typically the translation of "syntax error" */ |
| 1231 |
errmsg("%s at end of input", _(message)), |
| 1232 |
lexer_errposition())); |
| 1233 |
} |
| 1234 |
else |
| 1235 |
{ |
| 1236 |
ereport(ERROR, |
| 1237 |
(errcode(ERRCODE_SYNTAX_ERROR), |
| 1238 |
/* translator: first %s is typically the translation of "syntax error" */ |
| 1239 |
errmsg("%s at or near \"%s\"", _(message), loc), |
| 1240 |
lexer_errposition())); |
| 1241 |
} |
| 1242 |
} |
| 1243 |
|
| 1244 |
|
| 1245 |
/* |
| 1246 |
* Called before any actual parsing is done |
| 1247 |
*/ |
| 1248 |
core_yyscan_t |
| 1249 |
scanner_init(const char *str, |
| 1250 |
core_yy_extra_type *yyext, |
| 1251 |
const ScanKeywordList *keywordlist, |
| 1252 |
const uint16 *keyword_tokens) |
| 1253 |
{ |
| 1254 |
Size slen = strlen(str); |
| 1255 |
yyscan_t scanner; |
| 1256 |
|
| 1257 |
if (yylex_init(&scanner) != 0) |
| 1258 |
elog(ERROR, "yylex_init() failed: %m"); |
| 1259 |
|
| 1260 |
core_yyset_extra(yyext, scanner); |
| 1261 |
|
| 1262 |
yyext->keywordlist = keywordlist; |
| 1263 |
yyext->keyword_tokens = keyword_tokens; |
| 1264 |
|
| 1265 |
yyext->backslash_quote = backslash_quote; |
| 1266 |
yyext->escape_string_warning = escape_string_warning; |
| 1267 |
yyext->standard_conforming_strings = standard_conforming_strings; |
| 1268 |
|
| 1269 |
/* |
| 1270 |
* Make a scan buffer with special termination needed by flex. |
| 1271 |
*/ |
| 1272 |
yyext->scanbuf = (char *) palloc(slen + 2); |
| 1273 |
yyext->scanbuflen = slen; |
| 1274 |
memcpy(yyext->scanbuf, str, slen); |
| 1275 |
yyext->scanbuf[slen] = yyext->scanbuf[slen + 1] = YY_END_OF_BUFFER_CHAR; |
| 1276 |
yy_scan_buffer(yyext->scanbuf, slen + 2, scanner); |
| 1277 |
|
| 1278 |
/* initialize literal buffer to a reasonable but expansible size */ |
| 1279 |
yyext->literalalloc = 1024; |
| 1280 |
yyext->literalbuf = (char *) palloc(yyext->literalalloc); |
| 1281 |
yyext->literallen = 0; |
| 1282 |
|
| 1283 |
return scanner; |
| 1284 |
} |
| 1285 |
|
| 1286 |
|
| 1287 |
/* |
| 1288 |
* Called after parsing is done to clean up after scanner_init() |
| 1289 |
*/ |
| 1290 |
void |
| 1291 |
scanner_finish(core_yyscan_t yyscanner) |
| 1292 |
{ |
| 1293 |
/* |
| 1294 |
* We don't bother to call yylex_destroy(), because all it would do is |
| 1295 |
* pfree a small amount of control storage. It's cheaper to leak the |
| 1296 |
* storage until the parsing context is destroyed. The amount of space |
| 1297 |
* involved is usually negligible compared to the output parse tree |
| 1298 |
* anyway. |
| 1299 |
* |
| 1300 |
* We do bother to pfree the scanbuf and literal buffer, but only if they |
| 1301 |
* represent a nontrivial amount of space. The 8K cutoff is arbitrary. |
| 1302 |
*/ |
| 1303 |
if (yyextra->scanbuflen >= 8192) |
| 1304 |
pfree(yyextra->scanbuf); |
| 1305 |
if (yyextra->literalalloc >= 8192) |
| 1306 |
pfree(yyextra->literalbuf); |
| 1307 |
} |
| 1308 |
|
| 1309 |
|
| 1310 |
static void |
| 1311 |
addlit(char *ytext, int yleng, core_yyscan_t yyscanner) |
| 1312 |
{ |
| 1313 |
/* enlarge buffer if needed */ |
| 1314 |
if ((yyextra->literallen + yleng) >= yyextra->literalalloc) |
| 1315 |
{ |
| 1316 |
yyextra->literalalloc = pg_nextpower2_32(yyextra->literallen + yleng + 1); |
| 1317 |
yyextra->literalbuf = (char *) repalloc(yyextra->literalbuf, |
| 1318 |
yyextra->literalalloc); |
| 1319 |
} |
| 1320 |
/* append new data */ |
| 1321 |
memcpy(yyextra->literalbuf + yyextra->literallen, ytext, yleng); |
| 1322 |
yyextra->literallen += yleng; |
| 1323 |
} |
| 1324 |
|
| 1325 |
|
| 1326 |
static void |
| 1327 |
addlitchar(unsigned char ychar, core_yyscan_t yyscanner) |
| 1328 |
{ |
| 1329 |
/* enlarge buffer if needed */ |
| 1330 |
if ((yyextra->literallen + 1) >= yyextra->literalalloc) |
| 1331 |
{ |
| 1332 |
yyextra->literalalloc *= 2; |
| 1333 |
yyextra->literalbuf = (char *) repalloc(yyextra->literalbuf, |
| 1334 |
yyextra->literalalloc); |
| 1335 |
} |
| 1336 |
/* append new data */ |
| 1337 |
yyextra->literalbuf[yyextra->literallen] = ychar; |
| 1338 |
yyextra->literallen += 1; |
| 1339 |
} |
| 1340 |
|
| 1341 |
|
| 1342 |
/* |
| 1343 |
* Create a palloc'd copy of literalbuf, adding a trailing null. |
| 1344 |
*/ |
| 1345 |
static char * |
| 1346 |
litbufdup(core_yyscan_t yyscanner) |
| 1347 |
{ |
| 1348 |
int llen = yyextra->literallen; |
| 1349 |
char *new; |
| 1350 |
|
| 1351 |
new = palloc(llen + 1); |
| 1352 |
memcpy(new, yyextra->literalbuf, llen); |
| 1353 |
new[llen] = '\0'; |
| 1354 |
return new; |
| 1355 |
} |
| 1356 |
|
| 1357 |
/* |
| 1358 |
* Process {decinteger}, {hexinteger}, etc. Note this will also do the right |
| 1359 |
* thing with {numeric}, ie digits and a decimal point. |
| 1360 |
*/ |
| 1361 |
static int |
| 1362 |
process_integer_literal(const char *token, YYSTYPE *lval, int base) |
| 1363 |
{ |
| 1364 |
ErrorSaveContext escontext = {T_ErrorSaveContext}; |
| 1365 |
int32 val; |
| 1366 |
|
| 1367 |
val = pg_strtoint32_safe(token, (Node *) &escontext); |
| 1368 |
if (escontext.error_occurred) |
| 1369 |
{ |
| 1370 |
/* integer too large (or contains decimal pt), treat it as a float */ |
| 1371 |
lval->str = pstrdup(token); |
| 1372 |
return FCONST; |
| 1373 |
} |
| 1374 |
lval->ival = val; |
| 1375 |
return ICONST; |
| 1376 |
} |
| 1377 |
|
| 1378 |
static void |
| 1379 |
addunicode(pg_wchar c, core_yyscan_t yyscanner) |
| 1380 |
{ |
| 1381 |
ScannerCallbackState scbstate; |
| 1382 |
char buf[MAX_UNICODE_EQUIVALENT_STRING + 1]; |
| 1383 |
|
| 1384 |
if (!is_valid_unicode_codepoint(c)) |
| 1385 |
yyerror("invalid Unicode escape value"); |
| 1386 |
|
| 1387 |
/* |
| 1388 |
* We expect that pg_unicode_to_server() will complain about any |
| 1389 |
* unconvertible code point, so we don't have to set saw_non_ascii. |
| 1390 |
*/ |
| 1391 |
setup_scanner_errposition_callback(&scbstate, yyscanner, *(yylloc)); |
| 1392 |
pg_unicode_to_server(c, (unsigned char *) buf); |
| 1393 |
cancel_scanner_errposition_callback(&scbstate); |
| 1394 |
addlit(buf, strlen(buf), yyscanner); |
| 1395 |
} |
| 1396 |
|
| 1397 |
static unsigned char |
| 1398 |
unescape_single_char(unsigned char c, core_yyscan_t yyscanner) |
| 1399 |
{ |
| 1400 |
switch (c) |
| 1401 |
{ |
| 1402 |
case 'b': |
| 1403 |
return '\b'; |
| 1404 |
case 'f': |
| 1405 |
return '\f'; |
| 1406 |
case 'n': |
| 1407 |
return '\n'; |
| 1408 |
case 'r': |
| 1409 |
return '\r'; |
| 1410 |
case 't': |
| 1411 |
return '\t'; |
| 1412 |
case 'v': |
| 1413 |
return '\v'; |
| 1414 |
default: |
| 1415 |
/* check for backslash followed by non-7-bit-ASCII */ |
| 1416 |
if (c == '\0' || IS_HIGHBIT_SET(c)) |
| 1417 |
yyextra->saw_non_ascii = true; |
| 1418 |
|
| 1419 |
return c; |
| 1420 |
} |
| 1421 |
} |
| 1422 |
|
| 1423 |
static void |
| 1424 |
check_string_escape_warning(unsigned char ychar, core_yyscan_t yyscanner) |
| 1425 |
{ |
| 1426 |
if (ychar == '\'') |
| 1427 |
{ |
| 1428 |
if (yyextra->warn_on_first_escape && yyextra->escape_string_warning) |
| 1429 |
ereport(WARNING, |
| 1430 |
(errcode(ERRCODE_NONSTANDARD_USE_OF_ESCAPE_CHARACTER), |
| 1431 |
errmsg("nonstandard use of \\' in a string literal"), |
| 1432 |
errhint("Use '' to write quotes in strings, or use the escape string syntax (E'...')."), |
| 1433 |
lexer_errposition())); |
| 1434 |
yyextra->warn_on_first_escape = false; /* warn only once per string */ |
| 1435 |
} |
| 1436 |
else if (ychar == '\\') |
| 1437 |
{ |
| 1438 |
if (yyextra->warn_on_first_escape && yyextra->escape_string_warning) |
| 1439 |
ereport(WARNING, |
| 1440 |
(errcode(ERRCODE_NONSTANDARD_USE_OF_ESCAPE_CHARACTER), |
| 1441 |
errmsg("nonstandard use of \\\\ in a string literal"), |
| 1442 |
errhint("Use the escape string syntax for backslashes, e.g., E'\\\\'."), |
| 1443 |
lexer_errposition())); |
| 1444 |
yyextra->warn_on_first_escape = false; /* warn only once per string */ |
| 1445 |
} |
| 1446 |
else |
| 1447 |
check_escape_warning(yyscanner); |
| 1448 |
} |
| 1449 |
|
| 1450 |
static void |
| 1451 |
check_escape_warning(core_yyscan_t yyscanner) |
| 1452 |
{ |
| 1453 |
if (yyextra->warn_on_first_escape && yyextra->escape_string_warning) |
| 1454 |
ereport(WARNING, |
| 1455 |
(errcode(ERRCODE_NONSTANDARD_USE_OF_ESCAPE_CHARACTER), |
| 1456 |
errmsg("nonstandard use of escape in a string literal"), |
| 1457 |
errhint("Use the escape string syntax for escapes, e.g., E'\\r\\n'."), |
| 1458 |
lexer_errposition())); |
| 1459 |
yyextra->warn_on_first_escape = false; /* warn only once per string */ |
| 1460 |
} |
| 1461 |
|
| 1462 |
/* |
| 1463 |
* Interface functions to make flex use palloc() instead of malloc(). |
| 1464 |
* It'd be better to make these static, but flex insists otherwise. |
| 1465 |
*/ |
| 1466 |
|
| 1467 |
void * |
| 1468 |
core_yyalloc(yy_size_t bytes, core_yyscan_t yyscanner) |
| 1469 |
{ |
| 1470 |
return palloc(bytes); |
| 1471 |
} |
| 1472 |
|
| 1473 |
void * |
| 1474 |
core_yyrealloc(void *ptr, yy_size_t bytes, core_yyscan_t yyscanner) |
| 1475 |
{ |
| 1476 |
if (ptr) |
| 1477 |
return repalloc(ptr, bytes); |
| 1478 |
else |
| 1479 |
return palloc(bytes); |
| 1480 |
} |
| 1481 |
|
| 1482 |
void |
| 1483 |
core_yyfree(void *ptr, core_yyscan_t yyscanner) |
| 1484 |
{ |
| 1485 |
if (ptr) |
| 1486 |
pfree(ptr); |
| 1487 |
} |