2
** $Id: llex.c,v 2.20.1.1 2007/12/27 13:02:25 roberto Exp $
4
** See Copyright Notice in lua.h
29
#define next(ls) (ls->current = zgetc(ls->z))
34
#define currIsNewline(ls) (ls->current == '\n' || ls->current == '\r')
38
const char *const luaX_tokens [] = {
39
"and", "break", "do", "else", "elseif",
40
"end", "false", "for", "function", "if",
41
"in", "local", "nil", "not", "or", "repeat",
42
"return", "then", "true", "until", "while",
43
"..", "...", "==", ">=", "<=", "~=",
44
"<number>", "<name>", "<string>", "<eof>",
49
#define save_and_next(ls) (save(ls, ls->current), next(ls))
52
static void save (LexState *ls, int c) {
53
Mbuffer *b = ls->buff;
54
if (b->n + 1 > b->buffsize) {
56
if (b->buffsize >= MAX_SIZET/2)
57
luaX_lexerror(ls, "lexical element too long", 0);
58
newsize = b->buffsize * 2;
59
luaZ_resizebuffer(ls->L, b, newsize);
61
b->buffer[b->n++] = cast(char, c);
65
void luaX_init (lua_State *L) {
67
for (i=0; i<NUM_RESERVED; i++) {
68
TString *ts = luaS_new(L, luaX_tokens[i]);
69
luaS_fix(ts); /* reserved words are never collected */
70
lua_assert(strlen(luaX_tokens[i])+1 <= TOKEN_LEN);
71
ts->tsv.reserved = cast_byte(i+1); /* reserved word */
79
const char *luaX_token2str (LexState *ls, int token) {
80
if (token < FIRST_RESERVED) {
81
lua_assert(token == cast(unsigned char, token));
82
return (iscntrl(token)) ? luaO_pushfstring(ls->L, "char(%d)", token) :
83
luaO_pushfstring(ls->L, "%c", token);
86
return luaX_tokens[token-FIRST_RESERVED];
90
static const char *txtToken (LexState *ls, int token) {
96
return luaZ_buffer(ls->buff);
98
return luaX_token2str(ls, token);
103
void luaX_lexerror (LexState *ls, const char *msg, int token) {
105
luaO_chunkid(buff, getstr(ls->source), MAXSRC);
106
msg = luaO_pushfstring(ls->L, "%s:%d: %s", buff, ls->linenumber, msg);
108
luaO_pushfstring(ls->L, "%s near " LUA_QS, msg, txtToken(ls, token));
109
luaD_throw(ls->L, LUA_ERRSYNTAX);
113
void luaX_syntaxerror (LexState *ls, const char *msg) {
114
luaX_lexerror(ls, msg, ls->t.token);
118
TString *luaX_newstring (LexState *ls, const char *str, size_t l) {
119
lua_State *L = ls->L;
120
TString *ts = luaS_newlstr(L, str, l);
121
TValue *o = luaH_setstr(L, ls->fs->h, ts); /* entry for `str' */
123
setbvalue(o, 1); /* make sure `str' will not be collected */
128
static void inclinenumber (LexState *ls) {
129
int old = ls->current;
130
lua_assert(currIsNewline(ls));
131
next(ls); /* skip `\n' or `\r' */
132
if (currIsNewline(ls) && ls->current != old)
133
next(ls); /* skip `\n\r' or `\r\n' */
134
if (++ls->linenumber >= MAX_INT)
135
luaX_syntaxerror(ls, "chunk has too many lines");
139
void luaX_setinput (lua_State *L, LexState *ls, ZIO *z, TString *source) {
142
ls->lookahead.token = TK_EOS; /* no look-ahead token */
148
luaZ_resizebuffer(ls->L, ls->buff, LUA_MINBUFFER); /* initialize buffer */
149
next(ls); /* read first char */
155
** =======================================================
157
** =======================================================
162
static int check_next (LexState *ls, const char *set) {
163
if (!strchr(set, ls->current))
170
static void buffreplace (LexState *ls, char from, char to) {
171
size_t n = luaZ_bufflen(ls->buff);
172
char *p = luaZ_buffer(ls->buff);
174
if (p[n] == from) p[n] = to;
178
static void trydecpoint (LexState *ls, SemInfo *seminfo) {
179
/* 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] : '.');
185
char old = ls->decpoint;
188
buffreplace(ls, old, ls->decpoint); /* try updated decimal separator */
189
if (!luaO_str2d(luaZ_buffer(ls->buff), &seminfo->r)) {
190
/* format error with correct decimal point: no more options */
191
buffreplace(ls, ls->decpoint, '.'); /* undo change (for error message) */
192
luaX_lexerror(ls, "malformed number", TK_NUMBER);
198
static void read_numeral (LexState *ls, SemInfo *seminfo) {
199
lua_assert(isdigit(ls->current));
202
} while (isdigit(ls->current) || ls->current == '.');
203
if (check_next(ls, "Ee")) /* `E'? */
204
check_next(ls, "+-"); /* optional exponent sign */
205
while (isalnum(ls->current) || ls->current == '_')
208
buffreplace(ls, '.', ls->decpoint); /* follow locale for decimal point */
209
if (!luaO_str2d(luaZ_buffer(ls->buff), &seminfo->r)) /* format error? */
210
trydecpoint(ls, seminfo); /* try to update decimal point separator */
214
static int skip_sep (LexState *ls) {
217
lua_assert(s == '[' || s == ']');
219
while (ls->current == '=') {
223
return (ls->current == s) ? count : (-count) - 1;
227
static void read_long_string (LexState *ls, SemInfo *seminfo, int sep) {
229
(void)(cont); /* avoid warnings when `cont' is not used */
230
save_and_next(ls); /* skip 2nd `[' */
231
if (currIsNewline(ls)) /* string starts with a newline? */
232
inclinenumber(ls); /* skip it */
234
switch (ls->current) {
236
luaX_lexerror(ls, (seminfo) ? "unfinished long string" :
237
"unfinished long comment", TK_EOS);
238
break; /* to avoid warnings */
239
#if defined(LUA_COMPAT_LSTR)
241
if (skip_sep(ls) == sep) {
242
save_and_next(ls); /* skip 2nd `[' */
244
#if LUA_COMPAT_LSTR == 1
246
luaX_lexerror(ls, "nesting of [[...]] is deprecated", '[');
253
if (skip_sep(ls) == sep) {
254
save_and_next(ls); /* skip 2nd `]' */
255
#if defined(LUA_COMPAT_LSTR) && LUA_COMPAT_LSTR == 2
257
if (sep == 0 && cont >= 0) break;
267
if (!seminfo) luaZ_resetbuffer(ls->buff); /* avoid wasting space */
271
if (seminfo) save_and_next(ls);
277
seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + (2 + sep),
278
luaZ_bufflen(ls->buff) - 2*(2 + sep));
282
static void read_string (LexState *ls, int del, SemInfo *seminfo) {
284
while (ls->current != del) {
285
switch (ls->current) {
287
luaX_lexerror(ls, "unfinished string", TK_EOS);
288
continue; /* to avoid warnings */
291
luaX_lexerror(ls, "unfinished string", TK_STRING);
292
continue; /* to avoid warnings */
295
next(ls); /* do not save the `\' */
296
switch (ls->current) {
297
case 'a': c = '\a'; break;
298
case 'b': c = '\b'; break;
299
case 'f': c = '\f'; break;
300
case 'n': c = '\n'; break;
301
case 'r': c = '\r'; break;
302
case 't': c = '\t'; break;
303
case 'v': c = '\v'; break;
304
case '\n': /* go through */
305
case '\r': save(ls, '\n'); inclinenumber(ls); continue;
306
case EOZ: continue; /* will raise an error next loop */
308
if (!isdigit(ls->current))
309
save_and_next(ls); /* handles \\, \", \', and \? */
314
c = 10*c + (ls->current-'0');
316
} while (++i<3 && isdigit(ls->current));
318
luaX_lexerror(ls, "escape sequence too large", TK_STRING);
332
save_and_next(ls); /* skip delimiter */
333
seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + 1,
334
luaZ_bufflen(ls->buff) - 2);
338
static int llex (LexState *ls, SemInfo *seminfo) {
339
luaZ_resetbuffer(ls->buff);
341
switch (ls->current) {
349
if (ls->current != '-') return '-';
350
/* else is a comment */
352
if (ls->current == '[') {
353
int sep = skip_sep(ls);
354
luaZ_resetbuffer(ls->buff); /* `skip_sep' may dirty the buffer */
356
read_long_string(ls, NULL, sep); /* long comment */
357
luaZ_resetbuffer(ls->buff);
361
/* else short comment */
362
while (!currIsNewline(ls) && ls->current != EOZ)
367
int sep = skip_sep(ls);
369
read_long_string(ls, seminfo, sep);
372
else if (sep == -1) return '[';
373
else luaX_lexerror(ls, "invalid long string delimiter", TK_STRING);
377
if (ls->current != '=') return '=';
378
else { next(ls); return TK_EQ; }
382
if (ls->current != '=') return '<';
383
else { next(ls); return TK_LE; }
387
if (ls->current != '=') return '>';
388
else { next(ls); return TK_GE; }
392
if (ls->current != '=') return '~';
393
else { next(ls); return TK_NE; }
397
read_string(ls, ls->current, seminfo);
402
if (check_next(ls, ".")) {
403
if (check_next(ls, "."))
404
return TK_DOTS; /* ... */
405
else return TK_CONCAT; /* .. */
407
else if (!isdigit(ls->current)) return '.';
409
read_numeral(ls, seminfo);
417
if (isspace(ls->current)) {
418
lua_assert(!currIsNewline(ls));
422
else if (isdigit(ls->current)) {
423
read_numeral(ls, seminfo);
426
else if (isalpha(ls->current) || ls->current == '_') {
427
/* identifier or reserved word */
431
} while (isalnum(ls->current) || ls->current == '_');
432
ts = luaX_newstring(ls, luaZ_buffer(ls->buff),
433
luaZ_bufflen(ls->buff));
434
if (ts->tsv.reserved > 0) /* reserved word? */
435
return ts->tsv.reserved - 1 + FIRST_RESERVED;
444
return c; /* single-char tokens (+ - / ...) */
452
void luaX_next (LexState *ls) {
453
ls->lastline = ls->linenumber;
454
if (ls->lookahead.token != TK_EOS) { /* is there a look-ahead token? */
455
ls->t = ls->lookahead; /* use this one */
456
ls->lookahead.token = TK_EOS; /* and discharge it */
459
ls->t.token = llex(ls, &ls->t.seminfo); /* read next token */
463
void luaX_lookahead (LexState *ls) {
464
lua_assert(ls->lookahead.token == TK_EOS);
465
ls->lookahead.token = llex(ls, &ls->lookahead.seminfo);