2 ** $Id: llex.c,v 2.20.1.2 2009/11/23 14:58:22 roberto Exp $
4 ** See Copyright Notice in lua.h
28 #define next(ls) (ls->current = zgetc(ls->z))
33 #define currIsNewline(ls) (ls->current == '\n' || ls->current == '\r')
37 const char *const luaX_tokens [] = {
38 "and", "break", "do", "else", "elseif",
39 "end", "false", "for", "function", "if",
40 "in", "local", "nil", "not", "or", "repeat",
41 "return", "then", "true", "until", "while",
42 "..", "...", "==", ">=", "<=", "~=",
43 "<number>", "<name>", "<string>", "<eof>",
48 #define save_and_next(ls) (save(ls, ls->current), next(ls))
51 static void save (LexState *ls, int c) {
52 Mbuffer *b = ls->buff;
53 if (b->n + 1 > b->buffsize) {
55 if (b->buffsize >= MAX_SIZET/2)
56 luaX_lexerror(ls, "lexical element too long", 0);
57 newsize = b->buffsize * 2;
58 luaZ_resizebuffer(ls->L, b, newsize);
60 b->buffer[b->n++] = cast(char, c);
64 void luaX_init (lua_State *L) {
66 for (i=0; i<NUM_RESERVED; i++) {
67 TString *ts = luaS_new(L, luaX_tokens[i]);
68 luaS_fix(ts); /* reserved words are never collected */
69 lua_assert(strlen(luaX_tokens[i])+1 <= TOKEN_LEN);
70 ts->tsv.reserved = cast_byte(i+1); /* reserved word */
78 const char *luaX_token2str (LexState *ls, int token) {
79 if (token < FIRST_RESERVED) {
80 lua_assert(token == cast(unsigned char, token));
81 return (iscntrl(token)) ? luaO_pushfstring(ls->L, "char(%d)", token) :
82 luaO_pushfstring(ls->L, "%c", token);
85 return luaX_tokens[token-FIRST_RESERVED];
89 static const char *txtToken (LexState *ls, int token) {
95 return luaZ_buffer(ls->buff);
97 return luaX_token2str(ls, token);
102 void luaX_lexerror (LexState *ls, const char *msg, int token) {
104 luaO_chunkid(buff, getstr(ls->source), MAXSRC);
105 msg = luaO_pushfstring(ls->L, "%s:%d: %s", buff, ls->linenumber, msg);
107 luaO_pushfstring(ls->L, "%s near " LUA_QS, msg, txtToken(ls, token));
108 luaD_throw(ls->L, LUA_ERRSYNTAX);
112 void luaX_syntaxerror (LexState *ls, const char *msg) {
113 luaX_lexerror(ls, msg, ls->t.token);
117 TString *luaX_newstring (LexState *ls, const char *str, size_t l) {
118 lua_State *L = ls->L;
119 TString *ts = luaS_newlstr(L, str, l);
120 TValue *o = luaH_setstr(L, ls->fs->h, ts); /* entry for `str' */
122 setbvalue(o, 1); /* make sure `str' will not be collected */
129 static void inclinenumber (LexState *ls) {
130 int old = ls->current;
131 lua_assert(currIsNewline(ls));
132 next(ls); /* skip `\n' or `\r' */
133 if (currIsNewline(ls) && ls->current != old)
134 next(ls); /* skip `\n\r' or `\r\n' */
135 if (++ls->linenumber >= MAX_INT)
136 luaX_syntaxerror(ls, "chunk has too many lines");
140 void luaX_setinput (lua_State *L, LexState *ls, ZIO *z, TString *source) {
143 ls->lookahead.token = TK_EOS; /* no look-ahead token */
149 luaZ_resizebuffer(ls->L, ls->buff, LUA_MINBUFFER); /* initialize buffer */
150 next(ls); /* read first char */
156 ** =======================================================
158 ** =======================================================
163 static int check_next (LexState *ls, const char *set) {
164 if (!strchr(set, ls->current))
171 static void buffreplace (LexState *ls, char from, char to) {
172 size_t n = luaZ_bufflen(ls->buff);
173 char *p = luaZ_buffer(ls->buff);
175 if (p[n] == from) p[n] = to;
179 static void trydecpoint (LexState *ls, SemInfo *seminfo) {
180 /* format error: try to update decimal point separator */
181 struct lconv *cv = localeconv();
182 char old = ls->decpoint;
183 ls->decpoint = (cv ? cv->decimal_point[0] : '.');
184 buffreplace(ls, old, ls->decpoint); /* try updated decimal separator */
185 if (!luaO_str2d(luaZ_buffer(ls->buff), &seminfo->r)) {
186 /* format error with correct decimal point: no more options */
187 buffreplace(ls, ls->decpoint, '.'); /* undo change (for error message) */
188 luaX_lexerror(ls, "malformed number", TK_NUMBER);
194 static void read_numeral (LexState *ls, SemInfo *seminfo) {
195 lua_assert(isdigit(ls->current));
198 } while (isdigit(ls->current) || ls->current == '.');
199 if (check_next(ls, "Ee")) /* `E'? */
200 check_next(ls, "+-"); /* optional exponent sign */
201 while (isalnum(ls->current) || ls->current == '_')
204 buffreplace(ls, '.', ls->decpoint); /* follow locale for decimal point */
205 if (!luaO_str2d(luaZ_buffer(ls->buff), &seminfo->r)) /* format error? */
206 trydecpoint(ls, seminfo); /* try to update decimal point separator */
210 static int skip_sep (LexState *ls) {
213 lua_assert(s == '[' || s == ']');
215 while (ls->current == '=') {
219 return (ls->current == s) ? count : (-count) - 1;
223 static void read_long_string (LexState *ls, SemInfo *seminfo, int sep) {
225 (void)(cont); /* avoid warnings when `cont' is not used */
226 save_and_next(ls); /* skip 2nd `[' */
227 if (currIsNewline(ls)) /* string starts with a newline? */
228 inclinenumber(ls); /* skip it */
230 switch (ls->current) {
232 luaX_lexerror(ls, (seminfo) ? "unfinished long string" :
233 "unfinished long comment", TK_EOS);
234 break; /* to avoid warnings */
235 #if defined(LUA_COMPAT_LSTR)
237 if (skip_sep(ls) == sep) {
238 save_and_next(ls); /* skip 2nd `[' */
240 #if LUA_COMPAT_LSTR == 1
242 luaX_lexerror(ls, "nesting of [[...]] is deprecated", '[');
249 if (skip_sep(ls) == sep) {
250 save_and_next(ls); /* skip 2nd `]' */
251 #if defined(LUA_COMPAT_LSTR) && LUA_COMPAT_LSTR == 2
253 if (sep == 0 && cont >= 0) break;
263 if (!seminfo) luaZ_resetbuffer(ls->buff); /* avoid wasting space */
267 if (seminfo) save_and_next(ls);
273 seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + (2 + sep),
274 luaZ_bufflen(ls->buff) - 2*(2 + sep));
278 static void read_string (LexState *ls, int del, SemInfo *seminfo) {
280 while (ls->current != del) {
281 switch (ls->current) {
283 luaX_lexerror(ls, "unfinished string", TK_EOS);
284 continue; /* to avoid warnings */
287 luaX_lexerror(ls, "unfinished string", TK_STRING);
288 continue; /* to avoid warnings */
291 next(ls); /* do not save the `\' */
292 switch (ls->current) {
293 case 'a': c = '\a'; break;
294 case 'b': c = '\b'; break;
295 case 'f': c = '\f'; break;
296 case 'n': c = '\n'; break;
297 case 'r': c = '\r'; break;
298 case 't': c = '\t'; break;
299 case 'v': c = '\v'; break;
300 case '\n': /* go through */
301 case '\r': save(ls, '\n'); inclinenumber(ls); continue;
302 case EOZ: continue; /* will raise an error next loop */
304 if (!isdigit(ls->current))
305 save_and_next(ls); /* handles \\, \", \', and \? */
310 c = 10*c + (ls->current-'0');
312 } while (++i<3 && isdigit(ls->current));
314 luaX_lexerror(ls, "escape sequence too large", TK_STRING);
328 save_and_next(ls); /* skip delimiter */
329 seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + 1,
330 luaZ_bufflen(ls->buff) - 2);
334 static int llex (LexState *ls, SemInfo *seminfo) {
335 luaZ_resetbuffer(ls->buff);
337 switch (ls->current) {
345 if (ls->current != '-') return '-';
346 /* else is a comment */
348 if (ls->current == '[') {
349 int sep = skip_sep(ls);
350 luaZ_resetbuffer(ls->buff); /* `skip_sep' may dirty the buffer */
352 read_long_string(ls, NULL, sep); /* long comment */
353 luaZ_resetbuffer(ls->buff);
357 /* else short comment */
358 while (!currIsNewline(ls) && ls->current != EOZ)
363 int sep = skip_sep(ls);
365 read_long_string(ls, seminfo, sep);
368 else if (sep == -1) return '[';
369 else luaX_lexerror(ls, "invalid long string delimiter", TK_STRING);
373 if (ls->current != '=') return '=';
374 else { next(ls); return TK_EQ; }
378 if (ls->current != '=') return '<';
379 else { next(ls); return TK_LE; }
383 if (ls->current != '=') return '>';
384 else { next(ls); return TK_GE; }
388 if (ls->current != '=') return '~';
389 else { next(ls); return TK_NE; }
393 read_string(ls, ls->current, seminfo);
398 if (check_next(ls, ".")) {
399 if (check_next(ls, "."))
400 return TK_DOTS; /* ... */
401 else return TK_CONCAT; /* .. */
403 else if (!isdigit(ls->current)) return '.';
405 read_numeral(ls, seminfo);
413 if (isspace(ls->current)) {
414 lua_assert(!currIsNewline(ls));
418 else if (isdigit(ls->current)) {
419 read_numeral(ls, seminfo);
422 else if (isalpha(ls->current) || ls->current == '_') {
423 /* identifier or reserved word */
427 } while (isalnum(ls->current) || ls->current == '_');
428 ts = luaX_newstring(ls, luaZ_buffer(ls->buff),
429 luaZ_bufflen(ls->buff));
430 if (ts->tsv.reserved > 0) /* reserved word? */
431 return ts->tsv.reserved - 1 + FIRST_RESERVED;
440 return c; /* single-char tokens (+ - / ...) */
448 void luaX_next (LexState *ls) {
449 ls->lastline = ls->linenumber;
450 if (ls->lookahead.token != TK_EOS) { /* is there a look-ahead token? */
451 ls->t = ls->lookahead; /* use this one */
452 ls->lookahead.token = TK_EOS; /* and discharge it */
455 ls->t.token = llex(ls, &ls->t.seminfo); /* read next token */
459 void luaX_lookahead (LexState *ls) {
460 lua_assert(ls->lookahead.token == TK_EOS);
461 ls->lookahead.token = llex(ls, &ls->lookahead.seminfo);