1/*
2** $Id: llex.c,v 2.63.1.3 2015/02/09 17:56:34 roberto Exp $
3** Lexical Analyzer
4** See Copyright Notice in lua.h
5*/
6
7#include <sys/zfs_context.h>
8
9#define llex_c
10#define LUA_CORE
11
12#include "lua.h"
13
14#include "lctype.h"
15#include "ldo.h"
16#include "llex.h"
17#include "lobject.h"
18#include "lparser.h"
19#include "lstate.h"
20#include "lstring.h"
21#include "ltable.h"
22#include "lzio.h"
23
24
25
26#define next(ls) (ls->current = zgetc(ls->z))
27
28
29
30#define currIsNewline(ls)	(ls->current == '\n' || ls->current == '\r')
31
32
33/* ORDER RESERVED */
34static const char *const luaX_tokens [] = {
35    "and", "break", "do", "else", "elseif",
36    "end", "false", "for", "function", "goto", "if",
37    "in", "local", "nil", "not", "or", "repeat",
38    "return", "then", "true", "until", "while",
39    "..", "...", "==", ">=", "<=", "~=", "::", "<eof>",
40    "<number>", "<name>", "<string>"
41};
42
43
44#define save_and_next(ls) (save(ls, ls->current), next(ls))
45
46
47static l_noret lexerror (LexState *ls, const char *msg, int token);
48
49
50static void save (LexState *ls, int c) {
51  Mbuffer *b = ls->buff;
52  if (luaZ_bufflen(b) + 1 > luaZ_sizebuffer(b)) {
53    size_t newsize;
54    if (luaZ_sizebuffer(b) >= MAX_SIZET/2)
55      lexerror(ls, "lexical element too long", 0);
56    newsize = luaZ_sizebuffer(b) * 2;
57    luaZ_resizebuffer(ls->L, b, newsize);
58  }
59  b->buffer[luaZ_bufflen(b)++] = cast(char, c);
60}
61
62
63void luaX_init (lua_State *L) {
64  int i;
65  for (i=0; i<NUM_RESERVED; i++) {
66    TString *ts = luaS_new(L, luaX_tokens[i]);
67    luaS_fix(ts);  /* reserved words are never collected */
68    ts->tsv.extra = cast_byte(i+1);  /* reserved word */
69  }
70}
71
72
73const char *luaX_token2str (LexState *ls, int token) {
74  if (token < FIRST_RESERVED) {  /* single-byte symbols? */
75    lua_assert(token == cast(unsigned char, token));
76    return (lisprint(token)) ? luaO_pushfstring(ls->L, LUA_QL("%c"), token) :
77                              luaO_pushfstring(ls->L, "char(%d)", token);
78  }
79  else {
80    const char *s = luaX_tokens[token - FIRST_RESERVED];
81    if (token < TK_EOS)  /* fixed format (symbols and reserved words)? */
82      return luaO_pushfstring(ls->L, LUA_QS, s);
83    else  /* names, strings, and numerals */
84      return s;
85  }
86}
87
88
89static const char *txtToken (LexState *ls, int token) {
90  switch (token) {
91    case TK_NAME:
92    case TK_STRING:
93    case TK_NUMBER:
94      save(ls, '\0');
95      return luaO_pushfstring(ls->L, LUA_QS, luaZ_buffer(ls->buff));
96    default:
97      return luaX_token2str(ls, token);
98  }
99}
100
101
102static l_noret lexerror (LexState *ls, const char *msg, int token) {
103  char buff[LUA_IDSIZE];
104  luaO_chunkid(buff, getstr(ls->source), LUA_IDSIZE);
105  msg = luaO_pushfstring(ls->L, "%s:%d: %s", buff, ls->linenumber, msg);
106  if (token)
107    luaO_pushfstring(ls->L, "%s near %s", msg, txtToken(ls, token));
108  luaD_throw(ls->L, LUA_ERRSYNTAX);
109}
110
111
112l_noret luaX_syntaxerror (LexState *ls, const char *msg) {
113  lexerror(ls, msg, ls->t.token);
114}
115
116
117/*
118** creates a new string and anchors it in function's table so that
119** it will not be collected until the end of the function's compilation
120** (by that time it should be anchored in function's prototype)
121*/
122TString *luaX_newstring (LexState *ls, const char *str, size_t l) {
123  lua_State *L = ls->L;
124  TValue *o;  /* entry for `str' */
125  TString *ts = luaS_newlstr(L, str, l);  /* create new string */
126  setsvalue2s(L, L->top++, ts);  /* temporarily anchor it in stack */
127  o = luaH_set(L, ls->fs->h, L->top - 1);
128  if (ttisnil(o)) {  /* not in use yet? (see 'addK') */
129    /* boolean value does not need GC barrier;
130       table has no metatable, so it does not need to invalidate cache */
131    setbvalue(o, 1);  /* t[string] = true */
132    luaC_checkGC(L);
133  }
134  else {  /* string already present */
135    ts = rawtsvalue(keyfromval(o));  /* re-use value previously stored */
136  }
137  L->top--;  /* remove string from stack */
138  return ts;
139}
140
141
142/*
143** increment line number and skips newline sequence (any of
144** \n, \r, \n\r, or \r\n)
145*/
146static void inclinenumber (LexState *ls) {
147  int old = ls->current;
148  lua_assert(currIsNewline(ls));
149  next(ls);  /* skip `\n' or `\r' */
150  if (currIsNewline(ls) && ls->current != old)
151    next(ls);  /* skip `\n\r' or `\r\n' */
152  if (++ls->linenumber >= MAX_INT)
153    lexerror(ls, "chunk has too many lines", 0);
154}
155
156
157void luaX_setinput (lua_State *L, LexState *ls, ZIO *z, TString *source,
158                    int firstchar) {
159  ls->decpoint = '.';
160  ls->L = L;
161  ls->current = firstchar;
162  ls->lookahead.token = TK_EOS;  /* no look-ahead token */
163  ls->z = z;
164  ls->fs = NULL;
165  ls->linenumber = 1;
166  ls->lastline = 1;
167  ls->source = source;
168  ls->envn = luaS_new(L, LUA_ENV);  /* create env name */
169  luaS_fix(ls->envn);  /* never collect this name */
170  luaZ_resizebuffer(ls->L, ls->buff, LUA_MINBUFFER);  /* initialize buffer */
171}
172
173
174
175/*
176** =======================================================
177** LEXICAL ANALYZER
178** =======================================================
179*/
180
181
182
183static int check_next (LexState *ls, const char *set) {
184  if (ls->current == '\0' || !strchr(set, ls->current))
185    return 0;
186  save_and_next(ls);
187  return 1;
188}
189
190
191/*
192** change all characters 'from' in buffer to 'to'
193*/
194static void buffreplace (LexState *ls, char from, char to) {
195  size_t n = luaZ_bufflen(ls->buff);
196  char *p = luaZ_buffer(ls->buff);
197  while (n--)
198    if (p[n] == from) p[n] = to;
199}
200
201
202#if !defined(getlocaledecpoint)
203#define getlocaledecpoint()	(localeconv()->decimal_point[0])
204#endif
205
206
207#define buff2d(b,e)	luaO_str2d(luaZ_buffer(b), luaZ_bufflen(b) - 1, e)
208
209/*
210** in case of format error, try to change decimal point separator to
211** the one defined in the current locale and check again
212*/
213static void trydecpoint (LexState *ls, SemInfo *seminfo) {
214  char old = ls->decpoint;
215  ls->decpoint = getlocaledecpoint();
216  buffreplace(ls, old, ls->decpoint);  /* try new decimal separator */
217  if (!buff2d(ls->buff, &seminfo->r)) {
218    /* format error with correct decimal point: no more options */
219    buffreplace(ls, ls->decpoint, '.');  /* undo change (for error message) */
220    lexerror(ls, "malformed number", TK_NUMBER);
221  }
222}
223
224
225/* LUA_NUMBER */
226/*
227** this function is quite liberal in what it accepts, as 'luaO_str2d'
228** will reject ill-formed numerals.
229*/
230static void read_numeral (LexState *ls, SemInfo *seminfo) {
231  const char *expo = "Ee";
232  int first = ls->current;
233  lua_assert(lisdigit(ls->current));
234  save_and_next(ls);
235  if (first == '0' && check_next(ls, "Xx"))  /* hexadecimal? */
236    expo = "Pp";
237  for (;;) {
238    if (check_next(ls, expo))  /* exponent part? */
239      check_next(ls, "+-");  /* optional exponent sign */
240    if (lisxdigit(ls->current) || ls->current == '.')
241      save_and_next(ls);
242    else  break;
243  }
244  save(ls, '\0');
245  buffreplace(ls, '.', ls->decpoint);  /* follow locale for decimal point */
246  if (!buff2d(ls->buff, &seminfo->r))  /* format error? */
247    trydecpoint(ls, seminfo); /* try to update decimal point separator */
248}
249
250
251/*
252** skip a sequence '[=*[' or ']=*]' and return its number of '='s or
253** -1 if sequence is malformed
254*/
255static int skip_sep (LexState *ls) {
256  int count = 0;
257  int s = ls->current;
258  lua_assert(s == '[' || s == ']');
259  save_and_next(ls);
260  while (ls->current == '=') {
261    save_and_next(ls);
262    count++;
263  }
264  return (ls->current == s) ? count : (-count) - 1;
265}
266
267
268static void read_long_string (LexState *ls, SemInfo *seminfo, int sep) {
269  save_and_next(ls);  /* skip 2nd `[' */
270  if (currIsNewline(ls))  /* string starts with a newline? */
271    inclinenumber(ls);  /* skip it */
272  for (;;) {
273    switch (ls->current) {
274      case EOZ:
275        lexerror(ls, (seminfo) ? "unfinished long string" :
276                                 "unfinished long comment", TK_EOS);
277        break;  /* to avoid warnings */
278      case ']': {
279        if (skip_sep(ls) == sep) {
280          save_and_next(ls);  /* skip 2nd `]' */
281          goto endloop;
282        }
283        break;
284      }
285      case '\n': case '\r': {
286        save(ls, '\n');
287        inclinenumber(ls);
288        if (!seminfo) luaZ_resetbuffer(ls->buff);  /* avoid wasting space */
289        break;
290      }
291      default: {
292        if (seminfo) save_and_next(ls);
293        else next(ls);
294      }
295    }
296  } endloop:
297  if (seminfo)
298    seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + (2 + sep),
299                                     luaZ_bufflen(ls->buff) - 2*(2 + sep));
300}
301
302
303static void escerror (LexState *ls, int *c, int n, const char *msg) {
304  int i;
305  luaZ_resetbuffer(ls->buff);  /* prepare error message */
306  save(ls, '\\');
307  for (i = 0; i < n && c[i] != EOZ; i++)
308    save(ls, c[i]);
309  lexerror(ls, msg, TK_STRING);
310}
311
312
313static int readhexaesc (LexState *ls) {
314  int c[3], i;  /* keep input for error message */
315  int r = 0;  /* result accumulator */
316  c[0] = 'x';  /* for error message */
317  for (i = 1; i < 3; i++) {  /* read two hexadecimal digits */
318    c[i] = next(ls);
319    if (!lisxdigit(c[i]))
320      escerror(ls, c, i + 1, "hexadecimal digit expected");
321    r = (r << 4) + luaO_hexavalue(c[i]);
322  }
323  return r;
324}
325
326
327static int readdecesc (LexState *ls) {
328  int c[3], i;
329  int r = 0;  /* result accumulator */
330  for (i = 0; i < 3 && lisdigit(ls->current); i++) {  /* read up to 3 digits */
331    c[i] = ls->current;
332    r = 10*r + c[i] - '0';
333    next(ls);
334  }
335  if (r > UCHAR_MAX)
336    escerror(ls, c, i, "decimal escape too large");
337  return r;
338}
339
340
341static void read_string (LexState *ls, int del, SemInfo *seminfo) {
342  save_and_next(ls);  /* keep delimiter (for error messages) */
343  while (ls->current != del) {
344    switch (ls->current) {
345      case EOZ:
346        lexerror(ls, "unfinished string", TK_EOS);
347        break;  /* to avoid warnings */
348      case '\n':
349      case '\r':
350        lexerror(ls, "unfinished string", TK_STRING);
351        break;  /* to avoid warnings */
352      case '\\': {  /* escape sequences */
353        int c;  /* final character to be saved */
354        next(ls);  /* do not save the `\' */
355        switch (ls->current) {
356          case 'a': c = '\a'; goto read_save;
357          case 'b': c = '\b'; goto read_save;
358          case 'f': c = '\f'; goto read_save;
359          case 'n': c = '\n'; goto read_save;
360          case 'r': c = '\r'; goto read_save;
361          case 't': c = '\t'; goto read_save;
362          case 'v': c = '\v'; goto read_save;
363          case 'x': c = readhexaesc(ls); goto read_save;
364          case '\n': case '\r':
365            inclinenumber(ls); c = '\n'; goto only_save;
366          case '\\': case '\"': case '\'':
367            c = ls->current; goto read_save;
368          case EOZ: goto no_save;  /* will raise an error next loop */
369          case 'z': {  /* zap following span of spaces */
370            next(ls);  /* skip the 'z' */
371            while (lisspace(ls->current)) {
372              if (currIsNewline(ls)) inclinenumber(ls);
373              else next(ls);
374            }
375            goto no_save;
376          }
377          default: {
378            if (!lisdigit(ls->current))
379              escerror(ls, &ls->current, 1, "invalid escape sequence");
380            /* digital escape \ddd */
381            c = readdecesc(ls);
382            goto only_save;
383          }
384        }
385       read_save: next(ls);  /* read next character */
386       only_save: save(ls, c);  /* save 'c' */
387       no_save: break;
388      }
389      default:
390        save_and_next(ls);
391    }
392  }
393  save_and_next(ls);  /* skip delimiter */
394  seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + 1,
395                                   luaZ_bufflen(ls->buff) - 2);
396}
397
398
399static int llex (LexState *ls, SemInfo *seminfo) {
400  luaZ_resetbuffer(ls->buff);
401  for (;;) {
402    switch (ls->current) {
403      case '\n': case '\r': {  /* line breaks */
404        inclinenumber(ls);
405        break;
406      }
407      case ' ': case '\f': case '\t': case '\v': {  /* spaces */
408        next(ls);
409        break;
410      }
411      case '-': {  /* '-' or '--' (comment) */
412        next(ls);
413        if (ls->current != '-') return '-';
414        /* else is a comment */
415        next(ls);
416        if (ls->current == '[') {  /* long comment? */
417          int sep = skip_sep(ls);
418          luaZ_resetbuffer(ls->buff);  /* `skip_sep' may dirty the buffer */
419          if (sep >= 0) {
420            read_long_string(ls, NULL, sep);  /* skip long comment */
421            luaZ_resetbuffer(ls->buff);  /* previous call may dirty the buff. */
422            break;
423          }
424        }
425        /* else short comment */
426        while (!currIsNewline(ls) && ls->current != EOZ)
427          next(ls);  /* skip until end of line (or end of file) */
428        break;
429      }
430      case '[': {  /* long string or simply '[' */
431        int sep = skip_sep(ls);
432        if (sep >= 0) {
433          read_long_string(ls, seminfo, sep);
434          return TK_STRING;
435        }
436        else if (sep == -1) return '[';
437        else lexerror(ls, "invalid long string delimiter", TK_STRING);
438      }
439      case '=': {
440        next(ls);
441        if (ls->current != '=') return '=';
442        else { next(ls); return TK_EQ; }
443      }
444      case '<': {
445        next(ls);
446        if (ls->current != '=') return '<';
447        else { next(ls); return TK_LE; }
448      }
449      case '>': {
450        next(ls);
451        if (ls->current != '=') return '>';
452        else { next(ls); return TK_GE; }
453      }
454      case '~': {
455        next(ls);
456        if (ls->current != '=') return '~';
457        else { next(ls); return TK_NE; }
458      }
459      case ':': {
460        next(ls);
461        if (ls->current != ':') return ':';
462        else { next(ls); return TK_DBCOLON; }
463      }
464      case '"': case '\'': {  /* short literal strings */
465        read_string(ls, ls->current, seminfo);
466        return TK_STRING;
467      }
468      case '.': {  /* '.', '..', '...', or number */
469        save_and_next(ls);
470        if (check_next(ls, ".")) {
471          if (check_next(ls, "."))
472            return TK_DOTS;   /* '...' */
473          else return TK_CONCAT;   /* '..' */
474        }
475        else if (!lisdigit(ls->current)) return '.';
476        /* else go through */
477      }
478      /* FALLTHROUGH */
479      case '0': case '1': case '2': case '3': case '4':
480      case '5': case '6': case '7': case '8': case '9': {
481        read_numeral(ls, seminfo);
482        return TK_NUMBER;
483      }
484      case EOZ: {
485        return TK_EOS;
486      }
487      default: {
488        if (lislalpha(ls->current)) {  /* identifier or reserved word? */
489          TString *ts;
490          do {
491            save_and_next(ls);
492          } while (lislalnum(ls->current));
493          ts = luaX_newstring(ls, luaZ_buffer(ls->buff),
494                                  luaZ_bufflen(ls->buff));
495          seminfo->ts = ts;
496          if (isreserved(ts))  /* reserved word? */
497            return ts->tsv.extra - 1 + FIRST_RESERVED;
498          else {
499            return TK_NAME;
500          }
501        }
502        else {  /* single-char tokens (+ - / ...) */
503          int c = ls->current;
504          next(ls);
505          return c;
506        }
507      }
508    }
509  }
510}
511
512
513void luaX_next (LexState *ls) {
514  ls->lastline = ls->linenumber;
515  if (ls->lookahead.token != TK_EOS) {  /* is there a look-ahead token? */
516    ls->t = ls->lookahead;  /* use this one */
517    ls->lookahead.token = TK_EOS;  /* and discharge it */
518  }
519  else
520    ls->t.token = llex(ls, &ls->t.seminfo);  /* read next token */
521}
522
523
524int luaX_lookahead (LexState *ls) {
525  lua_assert(ls->lookahead.token == TK_EOS);
526  ls->lookahead.token = llex(ls, &ls->lookahead.seminfo);
527  return ls->lookahead.token;
528}
529
530