granicus.if.org Git - postgresql/blob - src/interfaces/ecpg/preproc/keywords.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  * keywords.c
   4  *        lexical token lookup for reserved words in PostgreSQL
   5  *
   6  * Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
   7  * Portions Copyright (c) 1994, Regents of the University of California
   8  *
   9  *
  10  * IDENTIFICATION
  11  *        $Header: /cvsroot/pgsql/src/interfaces/ecpg/preproc/keywords.c,v 1.51 2002/06/30 09:34:13 meskes Exp $
  12  *
  13  *-------------------------------------------------------------------------
  14  */
  15 #include "postgres_fe.h"
  16
  17 #include <ctype.h>
  18
  19 #include "extern.h"
  20 #include "preproc.h"
  21
  22
  23 /*
  24  * List of (keyword-name, keyword-token-value) pairs.
  25  *
  26  * !!WARNING!!: This list must be sorted, because binary
  27  *               search is used to locate entries.
  28  */
  29 static ScanKeyword ScanKeywords[] = {
  30         /* name, value */
  31         {"abort", ABORT_TRANS},
  32         {"absolute", ABSOLUTE},
  33         {"access", ACCESS},
  34         {"action", ACTION},
  35         {"add", ADD},
  36         {"after", AFTER},
  37         {"aggregate", AGGREGATE},
  38         {"all", ALL},
  39         {"alter", ALTER},
  40         {"analyse", ANALYSE},           /* British spelling */
  41         {"analyze", ANALYZE},
  42         {"and", AND},
  43         {"any", ANY},
  44         {"as", AS},
  45         {"asc", ASC},
  46         {"assertion", ASSERTION},
  47         {"assignment", ASSIGNMENT},
  48         {"at", AT},
  49         {"authorization", AUTHORIZATION},
  50         {"backward", BACKWARD},
  51         {"before", BEFORE},
  52         {"begin", BEGIN_TRANS},
  53         {"between", BETWEEN},
  54         {"bigint", BIGINT},
  55         {"binary", BINARY},
  56         {"bit", BIT},
  57         {"boolean", BOOLEAN},
  58         {"both", BOTH},
  59         {"by", BY},
  60         {"cache", CACHE},
  61         {"called", CALLED},
  62         {"cascade", CASCADE},
  63         {"case", CASE},
  64         {"cast", CAST},
  65         {"chain", CHAIN},
  66         {"char", CHAR_P},
  67         {"character", CHARACTER},
  68         {"characteristics", CHARACTERISTICS},
  69         {"check", CHECK},
  70         {"checkpoint", CHECKPOINT},
  71         {"close", CLOSE},
  72         {"cluster", CLUSTER},
  73         {"coalesce", COALESCE},
  74         {"collate", COLLATE},
  75         {"column", COLUMN},
  76         {"comment", COMMENT},
  77         {"commit", COMMIT},
  78         {"committed", COMMITTED},
  79         {"constraint", CONSTRAINT},
  80         {"constraints", CONSTRAINTS},
  81         {"copy", COPY},
  82         {"create", CREATE},
  83         {"createdb", CREATEDB},
  84         {"createuser", CREATEUSER},
  85         {"cross", CROSS},
  86         {"current_date", CURRENT_DATE},
  87         {"current_time", CURRENT_TIME},
  88         {"current_timestamp", CURRENT_TIMESTAMP},
  89         {"current_user", CURRENT_USER},
  90         {"cursor", CURSOR},
  91         {"cycle", CYCLE},
  92         {"database", DATABASE},
  93         {"day", DAY_P},
  94         {"dec", DEC},
  95         {"decimal", DECIMAL},
  96         {"declare", DECLARE},
  97         {"default", DEFAULT},
  98         {"deferrable", DEFERRABLE},
  99         {"deferred", DEFERRED},
 100         {"definer", DEFINER},
 101         {"delete", DELETE_P},
 102         {"delimiter", DELIMITER},
 103         {"delimiters", DELIMITERS},
 104         {"desc", DESC},
 105         {"distinct", DISTINCT},
 106         {"do", DO},
 107         {"domain", DOMAIN_P},
 108         {"double", DOUBLE},
 109         {"drop", DROP},
 110         {"each", EACH},
 111         {"else", ELSE},
 112         {"encoding", ENCODING},
 113         {"encrypted", ENCRYPTED},
 114         {"end", END_TRANS},
 115         {"escape", ESCAPE},
 116         {"except", EXCEPT},
 117         {"exclusive", EXCLUSIVE},
 118         {"execute", EXECUTE},
 119         {"exists", EXISTS},
 120         {"explain", EXPLAIN},
 121         {"external", EXTERNAL},
 122         {"extract", EXTRACT},
 123         {"false", FALSE_P},
 124         {"fetch", FETCH},
 125         {"float", FLOAT_P},
 126         {"for", FOR},
 127         {"force", FORCE},
 128         {"foreign", FOREIGN},
 129         {"forward", FORWARD},
 130         {"freeze", FREEZE},
 131         {"from", FROM},
 132         {"full", FULL},
 133         {"function", FUNCTION},
 134         {"global", GLOBAL},
 135         {"grant", GRANT},
 136         {"group", GROUP_P},
 137         {"handler", HANDLER},
 138         {"having", HAVING},
 139         {"hour", HOUR_P},
 140         {"ilike", ILIKE},
 141         {"immediate", IMMEDIATE},
 142         {"immutable", IMMUTABLE},
 143         {"implicit", IMPLICIT},
 144         {"in", IN_P},
 145         {"increment", INCREMENT},
 146         {"index", INDEX},
 147         {"inherits", INHERITS},
 148         {"initially", INITIALLY},
 149         {"inner", INNER_P},
 150         {"inout", INOUT},
 151         {"input", INPUT},
 152         {"insensitive", INSENSITIVE},
 153         {"insert", INSERT},
 154         {"instead", INSTEAD},
 155         {"int", INT},
 156         {"integer", INTEGER},
 157         {"intersect", INTERSECT},
 158         {"interval", INTERVAL},
 159         {"into", INTO},
 160         {"invoker", INVOKER},
 161         {"is", IS},
 162         {"isnull", ISNULL},
 163         {"isolation", ISOLATION},
 164         {"join", JOIN},
 165         {"key", KEY},
 166         {"lancompiler", LANCOMPILER},
 167         {"language", LANGUAGE},
 168         {"leading", LEADING},
 169         {"left", LEFT},
 170         {"level", LEVEL},
 171         {"like", LIKE},
 172         {"limit", LIMIT},
 173         {"listen", LISTEN},
 174         {"load", LOAD},
 175         {"local", LOCAL},
 176         {"localtime", LOCALTIME},
 177         {"localtimestamp", LOCALTIMESTAMP},
 178         {"location", LOCATION},
 179         {"lock", LOCK_P},
 180         {"match", MATCH},
 181         {"maxvalue", MAXVALUE},
 182         {"minute", MINUTE_P},
 183         {"minvalue", MINVALUE},
 184         {"mode", MODE},
 185         {"month", MONTH_P},
 186         {"move", MOVE},
 187         {"names", NAMES},
 188         {"national", NATIONAL},
 189         {"natural", NATURAL},
 190         {"nchar", NCHAR},
 191         {"new", NEW},
 192         {"next", NEXT},
 193         {"no", NO},
 194         {"nocreatedb", NOCREATEDB},
 195         {"nocreateuser", NOCREATEUSER},
 196         {"none", NONE},
 197         {"not", NOT},
 198         {"nothing", NOTHING},
 199         {"notify", NOTIFY},
 200         {"notnull", NOTNULL},
 201         {"null", NULL_P},
 202         {"nullif", NULLIF},
 203         {"numeric", NUMERIC},
 204         {"of", OF},
 205         {"off", OFF},
 206         {"offset", OFFSET},
 207         {"oids", OIDS},
 208         {"old", OLD},
 209         {"on", ON},
 210         {"only", ONLY},
 211         {"operator", OPERATOR},
 212         {"option", OPTION},
 213         {"or", OR},
 214         {"order", ORDER},
 215         {"out", OUT_P},
 216         {"outer", OUTER_P},
 217         {"overlaps", OVERLAPS},
 218         {"overlay", OVERLAY},
 219         {"owner", OWNER},
 220         {"partial", PARTIAL},
 221         {"password", PASSWORD},
 222         {"path", PATH_P},
 223         {"pendant", PENDANT},
 224         {"placing", PLACING},
 225         {"position", POSITION},
 226         {"precision", PRECISION},
 227         {"primary", PRIMARY},
 228         {"prior", PRIOR},
 229         {"privileges", PRIVILEGES},
 230         {"procedural", PROCEDURAL},
 231         {"procedure", PROCEDURE},
 232         {"read", READ},
 233         {"real", REAL},
 234         {"references", REFERENCES},
 235         {"reindex", REINDEX},
 236         {"relative", RELATIVE},
 237         {"rename", RENAME},
 238         {"replace", REPLACE},
 239         {"reset", RESET},
 240         {"restrict", RESTRICT},
 241         {"returns", RETURNS},
 242         {"revoke", REVOKE},
 243         {"right", RIGHT},
 244         {"rollback", ROLLBACK},
 245         {"row", ROW},
 246         {"rule", RULE},
 247         {"schema", SCHEMA},
 248         {"scroll", SCROLL},
 249         {"second", SECOND_P},
 250         {"security", SECURITY},
 251         {"select", SELECT},
 252         {"sequence", SEQUENCE},
 253         {"serializable", SERIALIZABLE},
 254         {"session", SESSION},
 255         {"session_user", SESSION_USER},
 256         {"set", SET},
 257         {"setof", SETOF},
 258         {"share", SHARE},
 259         {"show", SHOW},
 260         {"similar", SIMILAR},
 261         {"simple", SIMPLE},
 262         {"smallint", SMALLINT},
 263         {"some", SOME},
 264         {"stable", STABLE},
 265         {"start", START},
 266         {"statement", STATEMENT},
 267         {"statistics", STATISTICS},
 268         {"stdin", STDIN},
 269         {"stdout", STDOUT},
 270         {"storage", STORAGE},
 271         {"strict", STRICT},
 272         {"substring", SUBSTRING},
 273         {"sysid", SYSID},
 274         {"table", TABLE},
 275         {"temp", TEMP},
 276         {"template", TEMPLATE},
 277         {"temporary", TEMPORARY},
 278         {"then", THEN},
 279         {"time", TIME},
 280         {"timestamp", TIMESTAMP},
 281         {"to", TO},
 282         {"toast", TOAST},
 283         {"trailing", TRAILING},
 284         {"transaction", TRANSACTION},
 285         {"trigger", TRIGGER},
 286         {"trim", TRIM},
 287         {"true", TRUE_P},
 288         {"truncate", TRUNCATE},
 289         {"trusted", TRUSTED},
 290         {"type", TYPE_P},
 291         {"unencrypted", UNENCRYPTED},
 292         {"union", UNION},
 293         {"unique", UNIQUE},
 294         {"unknown", UNKNOWN},
 295         {"unlisten", UNLISTEN},
 296         {"until", UNTIL},
 297         {"update", UPDATE},
 298         {"usage", USAGE},
 299         {"user", USER},
 300         {"using", USING},
 301         {"vacuum", VACUUM},
 302         {"valid", VALID},
 303         {"validator", VALIDATOR},
 304         {"values", VALUES},
 305         {"varchar", VARCHAR},
 306         {"varying", VARYING},
 307         {"verbose", VERBOSE},
 308         {"version", VERSION},
 309         {"view", VIEW},
 310         {"volatile", VOLATILE},
 311         {"when", WHEN},
 312         {"where", WHERE},
 313         {"with", WITH},
 314         {"without", WITHOUT},
 315         {"work", WORK},
 316         {"write", WRITE},
 317         {"year", YEAR_P},
 318         {"zone", ZONE},
 319 };
 320
 321 /*
 322  * ScanKeywordLookup - see if a given word is a keyword
 323  *
 324  * Returns a pointer to the ScanKeyword table entry, or NULL if no match.
 325  *
 326  * The match is done case-insensitively.  Note that we deliberately use a
 327  * dumbed-down case conversion that will only translate 'A'-'Z' into 'a'-'z',
 328  * even if we are in a locale where tolower() would produce more or different
 329  * translations.  This is to conform to the SQL99 spec, which says that
 330  * keywords are to be matched in this way even though non-keyword identifiers
 331  * receive a different case-normalization mapping.
 332  */
 333 ScanKeyword *
 334 ScanKeywordLookup(char *text)
 335 {
 336         int                     len,
 337                                 i;
 338         char            word[NAMEDATALEN];
 339         ScanKeyword *low;
 340         ScanKeyword *high;
 341
 342         len = strlen(text);
 343         /* We assume all keywords are shorter than NAMEDATALEN. */
 344         if (len >= NAMEDATALEN)
 345                 return NULL;
 346
 347         /*
 348          * Apply an ASCII-only downcasing.      We must not use tolower() since it
 349          * may produce the wrong translation in some locales (eg, Turkish),
 350          * and we don't trust isupper() very much either.  In an ASCII-based
 351          * encoding the tests against A and Z are sufficient, but we also
 352          * check isupper() so that we will work correctly under EBCDIC.  The
 353          * actual case conversion step should work for either ASCII or EBCDIC.
 354          */
 355         for (i = 0; i < len; i++)
 356         {
 357                 char            ch = text[i];
 358
 359                 if (ch >= 'A' && ch <= 'Z' && isupper((unsigned char) ch))
 360                         ch += 'a' - 'A';
 361                 word[i] = ch;
 362         }
 363         word[len] = '\0';
 364
 365         /*
 366          * Now do a binary search using plain strcmp() comparison.
 367          */
 368         low = &ScanKeywords[0];
 369         high = endof(ScanKeywords) - 1;
 370         while (low <= high)
 371         {
 372                 ScanKeyword *middle;
 373                 int                     difference;
 374
 375                 middle = low + (high - low) / 2;
 376                 difference = strcmp(middle->name, word);
 377                 if (difference == 0)
 378                         return middle;
 379                 else if (difference < 0)
 380                         low = middle + 1;
 381                 else
 382                         high = middle - 1;
 383         }
 384
 385         return NULL;
 386 }