1*22ce4affSfengbojiang /* BEGIN CSTYLED */
2*22ce4affSfengbojiang /*
3*22ce4affSfengbojiang ** $Id: lstrlib.c,v 1.178.1.1 2013/04/12 18:48:47 roberto Exp $
4*22ce4affSfengbojiang ** Standard library for string operations and pattern-matching
5*22ce4affSfengbojiang ** See Copyright Notice in lua.h
6*22ce4affSfengbojiang */
7*22ce4affSfengbojiang
8*22ce4affSfengbojiang
9*22ce4affSfengbojiang #define lstrlib_c
10*22ce4affSfengbojiang #define LUA_LIB
11*22ce4affSfengbojiang
12*22ce4affSfengbojiang #include <sys/lua/lua.h>
13*22ce4affSfengbojiang
14*22ce4affSfengbojiang #include <sys/lua/lauxlib.h>
15*22ce4affSfengbojiang #include <sys/lua/lualib.h>
16*22ce4affSfengbojiang
17*22ce4affSfengbojiang
18*22ce4affSfengbojiang /*
19*22ce4affSfengbojiang ** maximum number of captures that a pattern can do during
20*22ce4affSfengbojiang ** pattern-matching. This limit is arbitrary.
21*22ce4affSfengbojiang */
22*22ce4affSfengbojiang #if !defined(LUA_MAXCAPTURES)
23*22ce4affSfengbojiang #define LUA_MAXCAPTURES 16
24*22ce4affSfengbojiang #endif
25*22ce4affSfengbojiang
26*22ce4affSfengbojiang
27*22ce4affSfengbojiang /* macro to `unsign' a character */
28*22ce4affSfengbojiang #define uchar(c) ((unsigned char)(c))
29*22ce4affSfengbojiang
30*22ce4affSfengbojiang /*
31*22ce4affSfengbojiang * The provided version of sprintf returns a char *, but str_format expects
32*22ce4affSfengbojiang * it to return the number of characters printed. This version has the expected
33*22ce4affSfengbojiang * behavior.
34*22ce4affSfengbojiang */
str_sprintf(char * buf,const char * fmt,...)35*22ce4affSfengbojiang static size_t str_sprintf(char *buf, const char *fmt, ...) {
36*22ce4affSfengbojiang va_list args;
37*22ce4affSfengbojiang size_t len;
38*22ce4affSfengbojiang
39*22ce4affSfengbojiang va_start(args, fmt);
40*22ce4affSfengbojiang len = vsnprintf(buf, INT_MAX, fmt, args);
41*22ce4affSfengbojiang va_end(args);
42*22ce4affSfengbojiang
43*22ce4affSfengbojiang return len;
44*22ce4affSfengbojiang }
45*22ce4affSfengbojiang
46*22ce4affSfengbojiang
str_len(lua_State * L)47*22ce4affSfengbojiang static int str_len (lua_State *L) {
48*22ce4affSfengbojiang size_t l;
49*22ce4affSfengbojiang luaL_checklstring(L, 1, &l);
50*22ce4affSfengbojiang lua_pushinteger(L, (lua_Integer)l);
51*22ce4affSfengbojiang return 1;
52*22ce4affSfengbojiang }
53*22ce4affSfengbojiang
54*22ce4affSfengbojiang
55*22ce4affSfengbojiang /* translate a relative string position: negative means back from end */
posrelat(ptrdiff_t pos,size_t len)56*22ce4affSfengbojiang static size_t posrelat (ptrdiff_t pos, size_t len) {
57*22ce4affSfengbojiang if (pos >= 0) return (size_t)pos;
58*22ce4affSfengbojiang else if (0u - (size_t)pos > len) return 0;
59*22ce4affSfengbojiang else return len - ((size_t)-pos) + 1;
60*22ce4affSfengbojiang }
61*22ce4affSfengbojiang
62*22ce4affSfengbojiang
str_sub(lua_State * L)63*22ce4affSfengbojiang static int str_sub (lua_State *L) {
64*22ce4affSfengbojiang size_t l;
65*22ce4affSfengbojiang const char *s = luaL_checklstring(L, 1, &l);
66*22ce4affSfengbojiang size_t start = posrelat(luaL_checkinteger(L, 2), l);
67*22ce4affSfengbojiang size_t end = posrelat(luaL_optinteger(L, 3, -1), l);
68*22ce4affSfengbojiang if (start < 1) start = 1;
69*22ce4affSfengbojiang if (end > l) end = l;
70*22ce4affSfengbojiang if (start <= end)
71*22ce4affSfengbojiang lua_pushlstring(L, s + start - 1, end - start + 1);
72*22ce4affSfengbojiang else lua_pushliteral(L, "");
73*22ce4affSfengbojiang return 1;
74*22ce4affSfengbojiang }
75*22ce4affSfengbojiang
76*22ce4affSfengbojiang
str_reverse(lua_State * L)77*22ce4affSfengbojiang static int str_reverse (lua_State *L) {
78*22ce4affSfengbojiang size_t l, i;
79*22ce4affSfengbojiang luaL_Buffer b;
80*22ce4affSfengbojiang const char *s = luaL_checklstring(L, 1, &l);
81*22ce4affSfengbojiang char *p = luaL_buffinitsize(L, &b, l);
82*22ce4affSfengbojiang for (i = 0; i < l; i++)
83*22ce4affSfengbojiang p[i] = s[l - i - 1];
84*22ce4affSfengbojiang luaL_pushresultsize(&b, l);
85*22ce4affSfengbojiang return 1;
86*22ce4affSfengbojiang }
87*22ce4affSfengbojiang
88*22ce4affSfengbojiang
str_lower(lua_State * L)89*22ce4affSfengbojiang static int str_lower (lua_State *L) {
90*22ce4affSfengbojiang size_t l;
91*22ce4affSfengbojiang size_t i;
92*22ce4affSfengbojiang luaL_Buffer b;
93*22ce4affSfengbojiang const char *s = luaL_checklstring(L, 1, &l);
94*22ce4affSfengbojiang char *p = luaL_buffinitsize(L, &b, l);
95*22ce4affSfengbojiang for (i=0; i<l; i++)
96*22ce4affSfengbojiang p[i] = tolower(uchar(s[i]));
97*22ce4affSfengbojiang luaL_pushresultsize(&b, l);
98*22ce4affSfengbojiang return 1;
99*22ce4affSfengbojiang }
100*22ce4affSfengbojiang
101*22ce4affSfengbojiang
str_upper(lua_State * L)102*22ce4affSfengbojiang static int str_upper (lua_State *L) {
103*22ce4affSfengbojiang size_t l;
104*22ce4affSfengbojiang size_t i;
105*22ce4affSfengbojiang luaL_Buffer b;
106*22ce4affSfengbojiang const char *s = luaL_checklstring(L, 1, &l);
107*22ce4affSfengbojiang char *p = luaL_buffinitsize(L, &b, l);
108*22ce4affSfengbojiang for (i=0; i<l; i++)
109*22ce4affSfengbojiang p[i] = toupper(uchar(s[i]));
110*22ce4affSfengbojiang luaL_pushresultsize(&b, l);
111*22ce4affSfengbojiang return 1;
112*22ce4affSfengbojiang }
113*22ce4affSfengbojiang
114*22ce4affSfengbojiang
115*22ce4affSfengbojiang /* reasonable limit to avoid arithmetic overflow */
116*22ce4affSfengbojiang #define MAXSIZE ((~(size_t)0) >> 1)
117*22ce4affSfengbojiang
str_rep(lua_State * L)118*22ce4affSfengbojiang static int str_rep (lua_State *L) {
119*22ce4affSfengbojiang size_t l, lsep;
120*22ce4affSfengbojiang const char *s = luaL_checklstring(L, 1, &l);
121*22ce4affSfengbojiang int n = luaL_checkint(L, 2);
122*22ce4affSfengbojiang const char *sep = luaL_optlstring(L, 3, "", &lsep);
123*22ce4affSfengbojiang if (n <= 0) lua_pushliteral(L, "");
124*22ce4affSfengbojiang else if (l + lsep < l || l + lsep >= MAXSIZE / n) /* may overflow? */
125*22ce4affSfengbojiang return luaL_error(L, "resulting string too large");
126*22ce4affSfengbojiang else {
127*22ce4affSfengbojiang size_t totallen = n * l + (n - 1) * lsep;
128*22ce4affSfengbojiang luaL_Buffer b;
129*22ce4affSfengbojiang char *p = luaL_buffinitsize(L, &b, totallen);
130*22ce4affSfengbojiang while (n-- > 1) { /* first n-1 copies (followed by separator) */
131*22ce4affSfengbojiang memcpy(p, s, l * sizeof(char)); p += l;
132*22ce4affSfengbojiang if (lsep > 0) { /* avoid empty 'memcpy' (may be expensive) */
133*22ce4affSfengbojiang memcpy(p, sep, lsep * sizeof(char)); p += lsep;
134*22ce4affSfengbojiang }
135*22ce4affSfengbojiang }
136*22ce4affSfengbojiang memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */
137*22ce4affSfengbojiang luaL_pushresultsize(&b, totallen);
138*22ce4affSfengbojiang }
139*22ce4affSfengbojiang return 1;
140*22ce4affSfengbojiang }
141*22ce4affSfengbojiang
142*22ce4affSfengbojiang
str_byte(lua_State * L)143*22ce4affSfengbojiang static int str_byte (lua_State *L) {
144*22ce4affSfengbojiang size_t l;
145*22ce4affSfengbojiang const char *s = luaL_checklstring(L, 1, &l);
146*22ce4affSfengbojiang size_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
147*22ce4affSfengbojiang size_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
148*22ce4affSfengbojiang int n, i;
149*22ce4affSfengbojiang if (posi < 1) posi = 1;
150*22ce4affSfengbojiang if (pose > l) pose = l;
151*22ce4affSfengbojiang if (posi > pose) return 0; /* empty interval; return no values */
152*22ce4affSfengbojiang n = (int)(pose - posi + 1);
153*22ce4affSfengbojiang if (posi + n <= pose) /* (size_t -> int) overflow? */
154*22ce4affSfengbojiang return luaL_error(L, "string slice too long");
155*22ce4affSfengbojiang luaL_checkstack(L, n, "string slice too long");
156*22ce4affSfengbojiang for (i=0; i<n; i++)
157*22ce4affSfengbojiang lua_pushinteger(L, uchar(s[posi+i-1]));
158*22ce4affSfengbojiang return n;
159*22ce4affSfengbojiang }
160*22ce4affSfengbojiang
161*22ce4affSfengbojiang
str_char(lua_State * L)162*22ce4affSfengbojiang static int str_char (lua_State *L) {
163*22ce4affSfengbojiang int n = lua_gettop(L); /* number of arguments */
164*22ce4affSfengbojiang int i;
165*22ce4affSfengbojiang luaL_Buffer b;
166*22ce4affSfengbojiang char *p = luaL_buffinitsize(L, &b, n);
167*22ce4affSfengbojiang for (i=1; i<=n; i++) {
168*22ce4affSfengbojiang int c = luaL_checkint(L, i);
169*22ce4affSfengbojiang luaL_argcheck(L, uchar(c) == c, i, "value out of range");
170*22ce4affSfengbojiang p[i - 1] = uchar(c);
171*22ce4affSfengbojiang }
172*22ce4affSfengbojiang luaL_pushresultsize(&b, n);
173*22ce4affSfengbojiang return 1;
174*22ce4affSfengbojiang }
175*22ce4affSfengbojiang
176*22ce4affSfengbojiang
177*22ce4affSfengbojiang #if defined(LUA_USE_DUMP)
writer(lua_State * L,const void * b,size_t size,void * B)178*22ce4affSfengbojiang static int writer (lua_State *L, const void* b, size_t size, void* B) {
179*22ce4affSfengbojiang (void)L;
180*22ce4affSfengbojiang luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);
181*22ce4affSfengbojiang return 0;
182*22ce4affSfengbojiang }
183*22ce4affSfengbojiang
184*22ce4affSfengbojiang
str_dump(lua_State * L)185*22ce4affSfengbojiang static int str_dump (lua_State *L) {
186*22ce4affSfengbojiang luaL_Buffer b;
187*22ce4affSfengbojiang luaL_checktype(L, 1, LUA_TFUNCTION);
188*22ce4affSfengbojiang lua_settop(L, 1);
189*22ce4affSfengbojiang luaL_buffinit(L,&b);
190*22ce4affSfengbojiang if (lua_dump(L, writer, &b) != 0)
191*22ce4affSfengbojiang return luaL_error(L, "unable to dump given function");
192*22ce4affSfengbojiang luaL_pushresult(&b);
193*22ce4affSfengbojiang return 1;
194*22ce4affSfengbojiang }
195*22ce4affSfengbojiang #endif
196*22ce4affSfengbojiang
197*22ce4affSfengbojiang
198*22ce4affSfengbojiang /*
199*22ce4affSfengbojiang ** {======================================================
200*22ce4affSfengbojiang ** PATTERN MATCHING
201*22ce4affSfengbojiang ** =======================================================
202*22ce4affSfengbojiang */
203*22ce4affSfengbojiang
204*22ce4affSfengbojiang
205*22ce4affSfengbojiang #define CAP_UNFINISHED (-1)
206*22ce4affSfengbojiang #define CAP_POSITION (-2)
207*22ce4affSfengbojiang
208*22ce4affSfengbojiang
209*22ce4affSfengbojiang typedef struct MatchState {
210*22ce4affSfengbojiang int matchdepth; /* control for recursive depth (to avoid C stack overflow) */
211*22ce4affSfengbojiang const char *src_init; /* init of source string */
212*22ce4affSfengbojiang const char *src_end; /* end ('\0') of source string */
213*22ce4affSfengbojiang const char *p_end; /* end ('\0') of pattern */
214*22ce4affSfengbojiang lua_State *L;
215*22ce4affSfengbojiang int level; /* total number of captures (finished or unfinished) */
216*22ce4affSfengbojiang struct {
217*22ce4affSfengbojiang const char *init;
218*22ce4affSfengbojiang ptrdiff_t len;
219*22ce4affSfengbojiang } capture[LUA_MAXCAPTURES];
220*22ce4affSfengbojiang } MatchState;
221*22ce4affSfengbojiang
222*22ce4affSfengbojiang
223*22ce4affSfengbojiang /* recursive function */
224*22ce4affSfengbojiang static const char *match (MatchState *ms, const char *s, const char *p);
225*22ce4affSfengbojiang
226*22ce4affSfengbojiang
227*22ce4affSfengbojiang /* maximum recursion depth for 'match' */
228*22ce4affSfengbojiang #if !defined(MAXCCALLS)
229*22ce4affSfengbojiang #define MAXCCALLS 200
230*22ce4affSfengbojiang #endif
231*22ce4affSfengbojiang
232*22ce4affSfengbojiang
233*22ce4affSfengbojiang #define L_ESC '%'
234*22ce4affSfengbojiang #define SPECIALS "^$*+?.([%-"
235*22ce4affSfengbojiang
236*22ce4affSfengbojiang
check_capture(MatchState * ms,int l)237*22ce4affSfengbojiang static int check_capture (MatchState *ms, int l) {
238*22ce4affSfengbojiang l -= '1';
239*22ce4affSfengbojiang if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
240*22ce4affSfengbojiang return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
241*22ce4affSfengbojiang return l;
242*22ce4affSfengbojiang }
243*22ce4affSfengbojiang
244*22ce4affSfengbojiang
capture_to_close(MatchState * ms)245*22ce4affSfengbojiang static int capture_to_close (MatchState *ms) {
246*22ce4affSfengbojiang int level = ms->level;
247*22ce4affSfengbojiang for (level--; level>=0; level--)
248*22ce4affSfengbojiang if (ms->capture[level].len == CAP_UNFINISHED) return level;
249*22ce4affSfengbojiang return luaL_error(ms->L, "invalid pattern capture");
250*22ce4affSfengbojiang }
251*22ce4affSfengbojiang
252*22ce4affSfengbojiang
classend(MatchState * ms,const char * p)253*22ce4affSfengbojiang static const char *classend (MatchState *ms, const char *p) {
254*22ce4affSfengbojiang switch (*p++) {
255*22ce4affSfengbojiang case L_ESC: {
256*22ce4affSfengbojiang if (p == ms->p_end)
257*22ce4affSfengbojiang luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
258*22ce4affSfengbojiang return p+1;
259*22ce4affSfengbojiang }
260*22ce4affSfengbojiang case '[': {
261*22ce4affSfengbojiang if (*p == '^') p++;
262*22ce4affSfengbojiang do { /* look for a `]' */
263*22ce4affSfengbojiang if (p == ms->p_end)
264*22ce4affSfengbojiang luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
265*22ce4affSfengbojiang if (*(p++) == L_ESC && p < ms->p_end)
266*22ce4affSfengbojiang p++; /* skip escapes (e.g. `%]') */
267*22ce4affSfengbojiang } while (*p != ']');
268*22ce4affSfengbojiang return p+1;
269*22ce4affSfengbojiang }
270*22ce4affSfengbojiang default: {
271*22ce4affSfengbojiang return p;
272*22ce4affSfengbojiang }
273*22ce4affSfengbojiang }
274*22ce4affSfengbojiang }
275*22ce4affSfengbojiang
276*22ce4affSfengbojiang
match_class(int c,int cl)277*22ce4affSfengbojiang static int match_class (int c, int cl) {
278*22ce4affSfengbojiang int res;
279*22ce4affSfengbojiang switch (tolower(cl)) {
280*22ce4affSfengbojiang case 'a' : res = isalpha(c); break;
281*22ce4affSfengbojiang case 'c' : res = iscntrl(c); break;
282*22ce4affSfengbojiang case 'd' : res = isdigit(c); break;
283*22ce4affSfengbojiang case 'g' : res = isgraph(c); break;
284*22ce4affSfengbojiang case 'l' : res = islower(c); break;
285*22ce4affSfengbojiang case 'p' : res = ispunct(c); break;
286*22ce4affSfengbojiang case 's' : res = isspace(c); break;
287*22ce4affSfengbojiang case 'u' : res = isupper(c); break;
288*22ce4affSfengbojiang case 'w' : res = isalnum(c); break;
289*22ce4affSfengbojiang case 'x' : res = isxdigit(c); break;
290*22ce4affSfengbojiang case 'z' : res = (c == 0); break; /* deprecated option */
291*22ce4affSfengbojiang default: return (cl == c);
292*22ce4affSfengbojiang }
293*22ce4affSfengbojiang return (islower(cl) ? res : !res);
294*22ce4affSfengbojiang }
295*22ce4affSfengbojiang
296*22ce4affSfengbojiang
matchbracketclass(int c,const char * p,const char * ec)297*22ce4affSfengbojiang static int matchbracketclass (int c, const char *p, const char *ec) {
298*22ce4affSfengbojiang int sig = 1;
299*22ce4affSfengbojiang if (*(p+1) == '^') {
300*22ce4affSfengbojiang sig = 0;
301*22ce4affSfengbojiang p++; /* skip the `^' */
302*22ce4affSfengbojiang }
303*22ce4affSfengbojiang while (++p < ec) {
304*22ce4affSfengbojiang if (*p == L_ESC) {
305*22ce4affSfengbojiang p++;
306*22ce4affSfengbojiang if (match_class(c, uchar(*p)))
307*22ce4affSfengbojiang return sig;
308*22ce4affSfengbojiang }
309*22ce4affSfengbojiang else if ((*(p+1) == '-') && (p+2 < ec)) {
310*22ce4affSfengbojiang p+=2;
311*22ce4affSfengbojiang if (uchar(*(p-2)) <= c && c <= uchar(*p))
312*22ce4affSfengbojiang return sig;
313*22ce4affSfengbojiang }
314*22ce4affSfengbojiang else if (uchar(*p) == c) return sig;
315*22ce4affSfengbojiang }
316*22ce4affSfengbojiang return !sig;
317*22ce4affSfengbojiang }
318*22ce4affSfengbojiang
319*22ce4affSfengbojiang
singlematch(MatchState * ms,const char * s,const char * p,const char * ep)320*22ce4affSfengbojiang static int singlematch (MatchState *ms, const char *s, const char *p,
321*22ce4affSfengbojiang const char *ep) {
322*22ce4affSfengbojiang if (s >= ms->src_end)
323*22ce4affSfengbojiang return 0;
324*22ce4affSfengbojiang else {
325*22ce4affSfengbojiang int c = uchar(*s);
326*22ce4affSfengbojiang switch (*p) {
327*22ce4affSfengbojiang case '.': return 1; /* matches any char */
328*22ce4affSfengbojiang case L_ESC: return match_class(c, uchar(*(p+1)));
329*22ce4affSfengbojiang case '[': return matchbracketclass(c, p, ep-1);
330*22ce4affSfengbojiang default: return (uchar(*p) == c);
331*22ce4affSfengbojiang }
332*22ce4affSfengbojiang }
333*22ce4affSfengbojiang }
334*22ce4affSfengbojiang
335*22ce4affSfengbojiang
matchbalance(MatchState * ms,const char * s,const char * p)336*22ce4affSfengbojiang static const char *matchbalance (MatchState *ms, const char *s,
337*22ce4affSfengbojiang const char *p) {
338*22ce4affSfengbojiang if (p >= ms->p_end - 1)
339*22ce4affSfengbojiang luaL_error(ms->L, "malformed pattern "
340*22ce4affSfengbojiang "(missing arguments to " LUA_QL("%%b") ")");
341*22ce4affSfengbojiang if (*s != *p) return NULL;
342*22ce4affSfengbojiang else {
343*22ce4affSfengbojiang int b = *p;
344*22ce4affSfengbojiang int e = *(p+1);
345*22ce4affSfengbojiang int cont = 1;
346*22ce4affSfengbojiang while (++s < ms->src_end) {
347*22ce4affSfengbojiang if (*s == e) {
348*22ce4affSfengbojiang if (--cont == 0) return s+1;
349*22ce4affSfengbojiang }
350*22ce4affSfengbojiang else if (*s == b) cont++;
351*22ce4affSfengbojiang }
352*22ce4affSfengbojiang }
353*22ce4affSfengbojiang return NULL; /* string ends out of balance */
354*22ce4affSfengbojiang }
355*22ce4affSfengbojiang
356*22ce4affSfengbojiang
max_expand(MatchState * ms,const char * s,const char * p,const char * ep)357*22ce4affSfengbojiang static const char *max_expand (MatchState *ms, const char *s,
358*22ce4affSfengbojiang const char *p, const char *ep) {
359*22ce4affSfengbojiang ptrdiff_t i = 0; /* counts maximum expand for item */
360*22ce4affSfengbojiang while (singlematch(ms, s + i, p, ep))
361*22ce4affSfengbojiang i++;
362*22ce4affSfengbojiang /* keeps trying to match with the maximum repetitions */
363*22ce4affSfengbojiang while (i>=0) {
364*22ce4affSfengbojiang const char *res = match(ms, (s+i), ep+1);
365*22ce4affSfengbojiang if (res) return res;
366*22ce4affSfengbojiang i--; /* else didn't match; reduce 1 repetition to try again */
367*22ce4affSfengbojiang }
368*22ce4affSfengbojiang return NULL;
369*22ce4affSfengbojiang }
370*22ce4affSfengbojiang
371*22ce4affSfengbojiang
min_expand(MatchState * ms,const char * s,const char * p,const char * ep)372*22ce4affSfengbojiang static const char *min_expand (MatchState *ms, const char *s,
373*22ce4affSfengbojiang const char *p, const char *ep) {
374*22ce4affSfengbojiang for (;;) {
375*22ce4affSfengbojiang const char *res = match(ms, s, ep+1);
376*22ce4affSfengbojiang if (res != NULL)
377*22ce4affSfengbojiang return res;
378*22ce4affSfengbojiang else if (singlematch(ms, s, p, ep))
379*22ce4affSfengbojiang s++; /* try with one more repetition */
380*22ce4affSfengbojiang else return NULL;
381*22ce4affSfengbojiang }
382*22ce4affSfengbojiang }
383*22ce4affSfengbojiang
384*22ce4affSfengbojiang
start_capture(MatchState * ms,const char * s,const char * p,int what)385*22ce4affSfengbojiang static const char *start_capture (MatchState *ms, const char *s,
386*22ce4affSfengbojiang const char *p, int what) {
387*22ce4affSfengbojiang const char *res;
388*22ce4affSfengbojiang int level = ms->level;
389*22ce4affSfengbojiang if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
390*22ce4affSfengbojiang ms->capture[level].init = s;
391*22ce4affSfengbojiang ms->capture[level].len = what;
392*22ce4affSfengbojiang ms->level = level+1;
393*22ce4affSfengbojiang if ((res=match(ms, s, p)) == NULL) /* match failed? */
394*22ce4affSfengbojiang ms->level--; /* undo capture */
395*22ce4affSfengbojiang return res;
396*22ce4affSfengbojiang }
397*22ce4affSfengbojiang
398*22ce4affSfengbojiang
end_capture(MatchState * ms,const char * s,const char * p)399*22ce4affSfengbojiang static const char *end_capture (MatchState *ms, const char *s,
400*22ce4affSfengbojiang const char *p) {
401*22ce4affSfengbojiang int l = capture_to_close(ms);
402*22ce4affSfengbojiang const char *res;
403*22ce4affSfengbojiang ms->capture[l].len = s - ms->capture[l].init; /* close capture */
404*22ce4affSfengbojiang if ((res = match(ms, s, p)) == NULL) /* match failed? */
405*22ce4affSfengbojiang ms->capture[l].len = CAP_UNFINISHED; /* undo capture */
406*22ce4affSfengbojiang return res;
407*22ce4affSfengbojiang }
408*22ce4affSfengbojiang
409*22ce4affSfengbojiang
match_capture(MatchState * ms,const char * s,int l)410*22ce4affSfengbojiang static const char *match_capture (MatchState *ms, const char *s, int l) {
411*22ce4affSfengbojiang size_t len;
412*22ce4affSfengbojiang l = check_capture(ms, l);
413*22ce4affSfengbojiang len = ms->capture[l].len;
414*22ce4affSfengbojiang if ((size_t)(ms->src_end-s) >= len &&
415*22ce4affSfengbojiang memcmp(ms->capture[l].init, s, len) == 0)
416*22ce4affSfengbojiang return s+len;
417*22ce4affSfengbojiang else return NULL;
418*22ce4affSfengbojiang }
419*22ce4affSfengbojiang
420*22ce4affSfengbojiang
match(MatchState * ms,const char * s,const char * p)421*22ce4affSfengbojiang static const char *match (MatchState *ms, const char *s, const char *p) {
422*22ce4affSfengbojiang if (ms->matchdepth-- == 0)
423*22ce4affSfengbojiang luaL_error(ms->L, "pattern too complex");
424*22ce4affSfengbojiang init: /* using goto's to optimize tail recursion */
425*22ce4affSfengbojiang if (p != ms->p_end) { /* end of pattern? */
426*22ce4affSfengbojiang switch (*p) {
427*22ce4affSfengbojiang case '(': { /* start capture */
428*22ce4affSfengbojiang if (*(p + 1) == ')') /* position capture? */
429*22ce4affSfengbojiang s = start_capture(ms, s, p + 2, CAP_POSITION);
430*22ce4affSfengbojiang else
431*22ce4affSfengbojiang s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
432*22ce4affSfengbojiang break;
433*22ce4affSfengbojiang }
434*22ce4affSfengbojiang case ')': { /* end capture */
435*22ce4affSfengbojiang s = end_capture(ms, s, p + 1);
436*22ce4affSfengbojiang break;
437*22ce4affSfengbojiang }
438*22ce4affSfengbojiang case '$': {
439*22ce4affSfengbojiang if ((p + 1) != ms->p_end) /* is the `$' the last char in pattern? */
440*22ce4affSfengbojiang goto dflt; /* no; go to default */
441*22ce4affSfengbojiang s = (s == ms->src_end) ? s : NULL; /* check end of string */
442*22ce4affSfengbojiang break;
443*22ce4affSfengbojiang }
444*22ce4affSfengbojiang case L_ESC: { /* escaped sequences not in the format class[*+?-]? */
445*22ce4affSfengbojiang switch (*(p + 1)) {
446*22ce4affSfengbojiang case 'b': { /* balanced string? */
447*22ce4affSfengbojiang s = matchbalance(ms, s, p + 2);
448*22ce4affSfengbojiang if (s != NULL) {
449*22ce4affSfengbojiang p += 4; goto init; /* return match(ms, s, p + 4); */
450*22ce4affSfengbojiang } /* else fail (s == NULL) */
451*22ce4affSfengbojiang break;
452*22ce4affSfengbojiang }
453*22ce4affSfengbojiang case 'f': { /* frontier? */
454*22ce4affSfengbojiang const char *ep; char previous;
455*22ce4affSfengbojiang p += 2;
456*22ce4affSfengbojiang if (*p != '[')
457*22ce4affSfengbojiang luaL_error(ms->L, "missing " LUA_QL("[") " after "
458*22ce4affSfengbojiang LUA_QL("%%f") " in pattern");
459*22ce4affSfengbojiang ep = classend(ms, p); /* points to what is next */
460*22ce4affSfengbojiang previous = (s == ms->src_init) ? '\0' : *(s - 1);
461*22ce4affSfengbojiang if (!matchbracketclass(uchar(previous), p, ep - 1) &&
462*22ce4affSfengbojiang matchbracketclass(uchar(*s), p, ep - 1)) {
463*22ce4affSfengbojiang p = ep; goto init; /* return match(ms, s, ep); */
464*22ce4affSfengbojiang }
465*22ce4affSfengbojiang s = NULL; /* match failed */
466*22ce4affSfengbojiang break;
467*22ce4affSfengbojiang }
468*22ce4affSfengbojiang case '0': case '1': case '2': case '3':
469*22ce4affSfengbojiang case '4': case '5': case '6': case '7':
470*22ce4affSfengbojiang case '8': case '9': { /* capture results (%0-%9)? */
471*22ce4affSfengbojiang s = match_capture(ms, s, uchar(*(p + 1)));
472*22ce4affSfengbojiang if (s != NULL) {
473*22ce4affSfengbojiang p += 2; goto init; /* return match(ms, s, p + 2) */
474*22ce4affSfengbojiang }
475*22ce4affSfengbojiang break;
476*22ce4affSfengbojiang }
477*22ce4affSfengbojiang default: goto dflt;
478*22ce4affSfengbojiang }
479*22ce4affSfengbojiang break;
480*22ce4affSfengbojiang }
481*22ce4affSfengbojiang default: dflt: { /* pattern class plus optional suffix */
482*22ce4affSfengbojiang const char *ep = classend(ms, p); /* points to optional suffix */
483*22ce4affSfengbojiang /* does not match at least once? */
484*22ce4affSfengbojiang if (!singlematch(ms, s, p, ep)) {
485*22ce4affSfengbojiang if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */
486*22ce4affSfengbojiang p = ep + 1; goto init; /* return match(ms, s, ep + 1); */
487*22ce4affSfengbojiang }
488*22ce4affSfengbojiang else /* '+' or no suffix */
489*22ce4affSfengbojiang s = NULL; /* fail */
490*22ce4affSfengbojiang }
491*22ce4affSfengbojiang else { /* matched once */
492*22ce4affSfengbojiang switch (*ep) { /* handle optional suffix */
493*22ce4affSfengbojiang case '?': { /* optional */
494*22ce4affSfengbojiang const char *res;
495*22ce4affSfengbojiang if ((res = match(ms, s + 1, ep + 1)) != NULL)
496*22ce4affSfengbojiang s = res;
497*22ce4affSfengbojiang else {
498*22ce4affSfengbojiang p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */
499*22ce4affSfengbojiang }
500*22ce4affSfengbojiang break;
501*22ce4affSfengbojiang }
502*22ce4affSfengbojiang case '+': /* 1 or more repetitions */
503*22ce4affSfengbojiang s++; /* 1 match already done */
504*22ce4affSfengbojiang /* FALLTHROUGH */
505*22ce4affSfengbojiang case '*': /* 0 or more repetitions */
506*22ce4affSfengbojiang s = max_expand(ms, s, p, ep);
507*22ce4affSfengbojiang break;
508*22ce4affSfengbojiang case '-': /* 0 or more repetitions (minimum) */
509*22ce4affSfengbojiang s = min_expand(ms, s, p, ep);
510*22ce4affSfengbojiang break;
511*22ce4affSfengbojiang default: /* no suffix */
512*22ce4affSfengbojiang s++; p = ep; goto init; /* return match(ms, s + 1, ep); */
513*22ce4affSfengbojiang }
514*22ce4affSfengbojiang }
515*22ce4affSfengbojiang break;
516*22ce4affSfengbojiang }
517*22ce4affSfengbojiang }
518*22ce4affSfengbojiang }
519*22ce4affSfengbojiang ms->matchdepth++;
520*22ce4affSfengbojiang return s;
521*22ce4affSfengbojiang }
522*22ce4affSfengbojiang
523*22ce4affSfengbojiang
524*22ce4affSfengbojiang
lmemfind(const char * s1,size_t l1,const char * s2,size_t l2)525*22ce4affSfengbojiang static const char *lmemfind (const char *s1, size_t l1,
526*22ce4affSfengbojiang const char *s2, size_t l2) {
527*22ce4affSfengbojiang if (l2 == 0) return s1; /* empty strings are everywhere */
528*22ce4affSfengbojiang else if (l2 > l1) return NULL; /* avoids a negative `l1' */
529*22ce4affSfengbojiang else {
530*22ce4affSfengbojiang const char *init; /* to search for a `*s2' inside `s1' */
531*22ce4affSfengbojiang l2--; /* 1st char will be checked by `memchr' */
532*22ce4affSfengbojiang l1 = l1-l2; /* `s2' cannot be found after that */
533*22ce4affSfengbojiang while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
534*22ce4affSfengbojiang init++; /* 1st char is already checked */
535*22ce4affSfengbojiang if (memcmp(init, s2+1, l2) == 0)
536*22ce4affSfengbojiang return init-1;
537*22ce4affSfengbojiang else { /* correct `l1' and `s1' to try again */
538*22ce4affSfengbojiang l1 -= init-s1;
539*22ce4affSfengbojiang s1 = init;
540*22ce4affSfengbojiang }
541*22ce4affSfengbojiang }
542*22ce4affSfengbojiang return NULL; /* not found */
543*22ce4affSfengbojiang }
544*22ce4affSfengbojiang }
545*22ce4affSfengbojiang
546*22ce4affSfengbojiang
push_onecapture(MatchState * ms,int i,const char * s,const char * e)547*22ce4affSfengbojiang static void push_onecapture (MatchState *ms, int i, const char *s,
548*22ce4affSfengbojiang const char *e) {
549*22ce4affSfengbojiang if (i >= ms->level) {
550*22ce4affSfengbojiang if (i == 0) /* ms->level == 0, too */
551*22ce4affSfengbojiang lua_pushlstring(ms->L, s, e - s); /* add whole match */
552*22ce4affSfengbojiang else
553*22ce4affSfengbojiang luaL_error(ms->L, "invalid capture index");
554*22ce4affSfengbojiang }
555*22ce4affSfengbojiang else {
556*22ce4affSfengbojiang ptrdiff_t l = ms->capture[i].len;
557*22ce4affSfengbojiang if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
558*22ce4affSfengbojiang if (l == CAP_POSITION)
559*22ce4affSfengbojiang lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
560*22ce4affSfengbojiang else
561*22ce4affSfengbojiang lua_pushlstring(ms->L, ms->capture[i].init, l);
562*22ce4affSfengbojiang }
563*22ce4affSfengbojiang }
564*22ce4affSfengbojiang
565*22ce4affSfengbojiang
push_captures(MatchState * ms,const char * s,const char * e)566*22ce4affSfengbojiang static int push_captures (MatchState *ms, const char *s, const char *e) {
567*22ce4affSfengbojiang int i;
568*22ce4affSfengbojiang int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
569*22ce4affSfengbojiang luaL_checkstack(ms->L, nlevels, "too many captures");
570*22ce4affSfengbojiang for (i = 0; i < nlevels; i++)
571*22ce4affSfengbojiang push_onecapture(ms, i, s, e);
572*22ce4affSfengbojiang return nlevels; /* number of strings pushed */
573*22ce4affSfengbojiang }
574*22ce4affSfengbojiang
575*22ce4affSfengbojiang
576*22ce4affSfengbojiang /* check whether pattern has no special characters */
nospecials(const char * p,size_t l)577*22ce4affSfengbojiang static int nospecials (const char *p, size_t l) {
578*22ce4affSfengbojiang size_t upto = 0;
579*22ce4affSfengbojiang do {
580*22ce4affSfengbojiang if (strpbrk(p + upto, SPECIALS))
581*22ce4affSfengbojiang return 0; /* pattern has a special character */
582*22ce4affSfengbojiang upto += strlen(p + upto) + 1; /* may have more after \0 */
583*22ce4affSfengbojiang } while (upto <= l);
584*22ce4affSfengbojiang return 1; /* no special chars found */
585*22ce4affSfengbojiang }
586*22ce4affSfengbojiang
587*22ce4affSfengbojiang
str_find_aux(lua_State * L,int find)588*22ce4affSfengbojiang static int str_find_aux (lua_State *L, int find) {
589*22ce4affSfengbojiang size_t ls, lp;
590*22ce4affSfengbojiang const char *s = luaL_checklstring(L, 1, &ls);
591*22ce4affSfengbojiang const char *p = luaL_checklstring(L, 2, &lp);
592*22ce4affSfengbojiang size_t init = posrelat(luaL_optinteger(L, 3, 1), ls);
593*22ce4affSfengbojiang if (init < 1) init = 1;
594*22ce4affSfengbojiang else if (init > ls + 1) { /* start after string's end? */
595*22ce4affSfengbojiang lua_pushnil(L); /* cannot find anything */
596*22ce4affSfengbojiang return 1;
597*22ce4affSfengbojiang }
598*22ce4affSfengbojiang /* explicit request or no special characters? */
599*22ce4affSfengbojiang if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
600*22ce4affSfengbojiang /* do a plain search */
601*22ce4affSfengbojiang const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp);
602*22ce4affSfengbojiang if (s2) {
603*22ce4affSfengbojiang lua_pushinteger(L, s2 - s + 1);
604*22ce4affSfengbojiang lua_pushinteger(L, s2 - s + lp);
605*22ce4affSfengbojiang return 2;
606*22ce4affSfengbojiang }
607*22ce4affSfengbojiang }
608*22ce4affSfengbojiang else {
609*22ce4affSfengbojiang MatchState ms;
610*22ce4affSfengbojiang const char *s1 = s + init - 1;
611*22ce4affSfengbojiang int anchor = (*p == '^');
612*22ce4affSfengbojiang if (anchor) {
613*22ce4affSfengbojiang p++; lp--; /* skip anchor character */
614*22ce4affSfengbojiang }
615*22ce4affSfengbojiang ms.L = L;
616*22ce4affSfengbojiang ms.matchdepth = MAXCCALLS;
617*22ce4affSfengbojiang ms.src_init = s;
618*22ce4affSfengbojiang ms.src_end = s + ls;
619*22ce4affSfengbojiang ms.p_end = p + lp;
620*22ce4affSfengbojiang do {
621*22ce4affSfengbojiang const char *res;
622*22ce4affSfengbojiang ms.level = 0;
623*22ce4affSfengbojiang lua_assert(ms.matchdepth == MAXCCALLS);
624*22ce4affSfengbojiang if ((res=match(&ms, s1, p)) != NULL) {
625*22ce4affSfengbojiang if (find) {
626*22ce4affSfengbojiang lua_pushinteger(L, s1 - s + 1); /* start */
627*22ce4affSfengbojiang lua_pushinteger(L, res - s); /* end */
628*22ce4affSfengbojiang return push_captures(&ms, NULL, 0) + 2;
629*22ce4affSfengbojiang }
630*22ce4affSfengbojiang else
631*22ce4affSfengbojiang return push_captures(&ms, s1, res);
632*22ce4affSfengbojiang }
633*22ce4affSfengbojiang } while (s1++ < ms.src_end && !anchor);
634*22ce4affSfengbojiang }
635*22ce4affSfengbojiang lua_pushnil(L); /* not found */
636*22ce4affSfengbojiang return 1;
637*22ce4affSfengbojiang }
638*22ce4affSfengbojiang
639*22ce4affSfengbojiang
str_find(lua_State * L)640*22ce4affSfengbojiang static int str_find (lua_State *L) {
641*22ce4affSfengbojiang return str_find_aux(L, 1);
642*22ce4affSfengbojiang }
643*22ce4affSfengbojiang
644*22ce4affSfengbojiang
str_match(lua_State * L)645*22ce4affSfengbojiang static int str_match (lua_State *L) {
646*22ce4affSfengbojiang return str_find_aux(L, 0);
647*22ce4affSfengbojiang }
648*22ce4affSfengbojiang
649*22ce4affSfengbojiang
gmatch_aux(lua_State * L)650*22ce4affSfengbojiang static int gmatch_aux (lua_State *L) {
651*22ce4affSfengbojiang MatchState ms;
652*22ce4affSfengbojiang size_t ls, lp;
653*22ce4affSfengbojiang const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
654*22ce4affSfengbojiang const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);
655*22ce4affSfengbojiang const char *src;
656*22ce4affSfengbojiang ms.L = L;
657*22ce4affSfengbojiang ms.matchdepth = MAXCCALLS;
658*22ce4affSfengbojiang ms.src_init = s;
659*22ce4affSfengbojiang ms.src_end = s+ls;
660*22ce4affSfengbojiang ms.p_end = p + lp;
661*22ce4affSfengbojiang for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
662*22ce4affSfengbojiang src <= ms.src_end;
663*22ce4affSfengbojiang src++) {
664*22ce4affSfengbojiang const char *e;
665*22ce4affSfengbojiang ms.level = 0;
666*22ce4affSfengbojiang lua_assert(ms.matchdepth == MAXCCALLS);
667*22ce4affSfengbojiang if ((e = match(&ms, src, p)) != NULL) {
668*22ce4affSfengbojiang lua_Integer newstart = e-s;
669*22ce4affSfengbojiang if (e == src) newstart++; /* empty match? go at least one position */
670*22ce4affSfengbojiang lua_pushinteger(L, newstart);
671*22ce4affSfengbojiang lua_replace(L, lua_upvalueindex(3));
672*22ce4affSfengbojiang return push_captures(&ms, src, e);
673*22ce4affSfengbojiang }
674*22ce4affSfengbojiang }
675*22ce4affSfengbojiang return 0; /* not found */
676*22ce4affSfengbojiang }
677*22ce4affSfengbojiang
678*22ce4affSfengbojiang
str_gmatch(lua_State * L)679*22ce4affSfengbojiang static int str_gmatch (lua_State *L) {
680*22ce4affSfengbojiang luaL_checkstring(L, 1);
681*22ce4affSfengbojiang luaL_checkstring(L, 2);
682*22ce4affSfengbojiang lua_settop(L, 2);
683*22ce4affSfengbojiang lua_pushinteger(L, 0);
684*22ce4affSfengbojiang lua_pushcclosure(L, gmatch_aux, 3);
685*22ce4affSfengbojiang return 1;
686*22ce4affSfengbojiang }
687*22ce4affSfengbojiang
688*22ce4affSfengbojiang
add_s(MatchState * ms,luaL_Buffer * b,const char * s,const char * e)689*22ce4affSfengbojiang static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
690*22ce4affSfengbojiang const char *e) {
691*22ce4affSfengbojiang size_t l, i;
692*22ce4affSfengbojiang const char *news = lua_tolstring(ms->L, 3, &l);
693*22ce4affSfengbojiang for (i = 0; i < l; i++) {
694*22ce4affSfengbojiang if (news[i] != L_ESC)
695*22ce4affSfengbojiang luaL_addchar(b, news[i]);
696*22ce4affSfengbojiang else {
697*22ce4affSfengbojiang i++; /* skip ESC */
698*22ce4affSfengbojiang if (!isdigit(uchar(news[i]))) {
699*22ce4affSfengbojiang if (news[i] != L_ESC)
700*22ce4affSfengbojiang luaL_error(ms->L, "invalid use of " LUA_QL("%c")
701*22ce4affSfengbojiang " in replacement string", L_ESC);
702*22ce4affSfengbojiang luaL_addchar(b, news[i]);
703*22ce4affSfengbojiang }
704*22ce4affSfengbojiang else if (news[i] == '0')
705*22ce4affSfengbojiang luaL_addlstring(b, s, e - s);
706*22ce4affSfengbojiang else {
707*22ce4affSfengbojiang push_onecapture(ms, news[i] - '1', s, e);
708*22ce4affSfengbojiang luaL_addvalue(b); /* add capture to accumulated result */
709*22ce4affSfengbojiang }
710*22ce4affSfengbojiang }
711*22ce4affSfengbojiang }
712*22ce4affSfengbojiang }
713*22ce4affSfengbojiang
714*22ce4affSfengbojiang
add_value(MatchState * ms,luaL_Buffer * b,const char * s,const char * e,int tr)715*22ce4affSfengbojiang static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
716*22ce4affSfengbojiang const char *e, int tr) {
717*22ce4affSfengbojiang lua_State *L = ms->L;
718*22ce4affSfengbojiang switch (tr) {
719*22ce4affSfengbojiang case LUA_TFUNCTION: {
720*22ce4affSfengbojiang int n;
721*22ce4affSfengbojiang lua_pushvalue(L, 3);
722*22ce4affSfengbojiang n = push_captures(ms, s, e);
723*22ce4affSfengbojiang lua_call(L, n, 1);
724*22ce4affSfengbojiang break;
725*22ce4affSfengbojiang }
726*22ce4affSfengbojiang case LUA_TTABLE: {
727*22ce4affSfengbojiang push_onecapture(ms, 0, s, e);
728*22ce4affSfengbojiang lua_gettable(L, 3);
729*22ce4affSfengbojiang break;
730*22ce4affSfengbojiang }
731*22ce4affSfengbojiang default: { /* LUA_TNUMBER or LUA_TSTRING */
732*22ce4affSfengbojiang add_s(ms, b, s, e);
733*22ce4affSfengbojiang return;
734*22ce4affSfengbojiang }
735*22ce4affSfengbojiang }
736*22ce4affSfengbojiang if (!lua_toboolean(L, -1)) { /* nil or false? */
737*22ce4affSfengbojiang lua_pop(L, 1);
738*22ce4affSfengbojiang lua_pushlstring(L, s, e - s); /* keep original text */
739*22ce4affSfengbojiang }
740*22ce4affSfengbojiang else if (!lua_isstring(L, -1))
741*22ce4affSfengbojiang luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
742*22ce4affSfengbojiang luaL_addvalue(b); /* add result to accumulator */
743*22ce4affSfengbojiang }
744*22ce4affSfengbojiang
745*22ce4affSfengbojiang
str_gsub(lua_State * L)746*22ce4affSfengbojiang static int str_gsub (lua_State *L) {
747*22ce4affSfengbojiang size_t srcl, lp;
748*22ce4affSfengbojiang const char *src = luaL_checklstring(L, 1, &srcl);
749*22ce4affSfengbojiang const char *p = luaL_checklstring(L, 2, &lp);
750*22ce4affSfengbojiang int tr = lua_type(L, 3);
751*22ce4affSfengbojiang size_t max_s = luaL_optinteger(L, 4, srcl+1);
752*22ce4affSfengbojiang int anchor = (*p == '^');
753*22ce4affSfengbojiang size_t n = 0;
754*22ce4affSfengbojiang MatchState ms;
755*22ce4affSfengbojiang luaL_Buffer b;
756*22ce4affSfengbojiang luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
757*22ce4affSfengbojiang tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
758*22ce4affSfengbojiang "string/function/table expected");
759*22ce4affSfengbojiang luaL_buffinit(L, &b);
760*22ce4affSfengbojiang if (anchor) {
761*22ce4affSfengbojiang p++; lp--; /* skip anchor character */
762*22ce4affSfengbojiang }
763*22ce4affSfengbojiang ms.L = L;
764*22ce4affSfengbojiang ms.matchdepth = MAXCCALLS;
765*22ce4affSfengbojiang ms.src_init = src;
766*22ce4affSfengbojiang ms.src_end = src+srcl;
767*22ce4affSfengbojiang ms.p_end = p + lp;
768*22ce4affSfengbojiang while (n < max_s) {
769*22ce4affSfengbojiang const char *e;
770*22ce4affSfengbojiang ms.level = 0;
771*22ce4affSfengbojiang lua_assert(ms.matchdepth == MAXCCALLS);
772*22ce4affSfengbojiang e = match(&ms, src, p);
773*22ce4affSfengbojiang if (e) {
774*22ce4affSfengbojiang n++;
775*22ce4affSfengbojiang add_value(&ms, &b, src, e, tr);
776*22ce4affSfengbojiang }
777*22ce4affSfengbojiang if (e && e>src) /* non empty match? */
778*22ce4affSfengbojiang src = e; /* skip it */
779*22ce4affSfengbojiang else if (src < ms.src_end)
780*22ce4affSfengbojiang luaL_addchar(&b, *src++);
781*22ce4affSfengbojiang else break;
782*22ce4affSfengbojiang if (anchor) break;
783*22ce4affSfengbojiang }
784*22ce4affSfengbojiang luaL_addlstring(&b, src, ms.src_end-src);
785*22ce4affSfengbojiang luaL_pushresult(&b);
786*22ce4affSfengbojiang lua_pushinteger(L, n); /* number of substitutions */
787*22ce4affSfengbojiang return 2;
788*22ce4affSfengbojiang }
789*22ce4affSfengbojiang
790*22ce4affSfengbojiang /* }====================================================== */
791*22ce4affSfengbojiang
792*22ce4affSfengbojiang
793*22ce4affSfengbojiang
794*22ce4affSfengbojiang /*
795*22ce4affSfengbojiang ** {======================================================
796*22ce4affSfengbojiang ** STRING FORMAT
797*22ce4affSfengbojiang ** =======================================================
798*22ce4affSfengbojiang */
799*22ce4affSfengbojiang
800*22ce4affSfengbojiang /*
801*22ce4affSfengbojiang ** LUA_INTFRMLEN is the length modifier for integer conversions in
802*22ce4affSfengbojiang ** 'string.format'; LUA_INTFRM_T is the integer type corresponding to
803*22ce4affSfengbojiang ** the previous length
804*22ce4affSfengbojiang */
805*22ce4affSfengbojiang #if !defined(LUA_INTFRMLEN) /* { */
806*22ce4affSfengbojiang #if defined(LUA_USE_LONGLONG)
807*22ce4affSfengbojiang
808*22ce4affSfengbojiang #define LUA_INTFRMLEN "ll"
809*22ce4affSfengbojiang #define LUA_INTFRM_T long long
810*22ce4affSfengbojiang
811*22ce4affSfengbojiang #else
812*22ce4affSfengbojiang
813*22ce4affSfengbojiang #define LUA_INTFRMLEN "l"
814*22ce4affSfengbojiang #define LUA_INTFRM_T long
815*22ce4affSfengbojiang
816*22ce4affSfengbojiang #endif
817*22ce4affSfengbojiang #endif /* } */
818*22ce4affSfengbojiang
819*22ce4affSfengbojiang
820*22ce4affSfengbojiang /*
821*22ce4affSfengbojiang ** LUA_FLTFRMLEN is the length modifier for float conversions in
822*22ce4affSfengbojiang ** 'string.format'; LUA_FLTFRM_T is the float type corresponding to
823*22ce4affSfengbojiang ** the previous length
824*22ce4affSfengbojiang */
825*22ce4affSfengbojiang #if !defined(LUA_FLTFRMLEN)
826*22ce4affSfengbojiang
827*22ce4affSfengbojiang #define LUA_FLTFRMLEN ""
828*22ce4affSfengbojiang #define LUA_FLTFRM_T double
829*22ce4affSfengbojiang
830*22ce4affSfengbojiang #endif
831*22ce4affSfengbojiang
832*22ce4affSfengbojiang
833*22ce4affSfengbojiang /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
834*22ce4affSfengbojiang #define MAX_ITEM 512
835*22ce4affSfengbojiang /* valid flags in a format specification */
836*22ce4affSfengbojiang #define FLAGS "-+ #0"
837*22ce4affSfengbojiang /*
838*22ce4affSfengbojiang ** maximum size of each format specification (such as '%-099.99d')
839*22ce4affSfengbojiang ** (+10 accounts for %99.99x plus margin of error)
840*22ce4affSfengbojiang */
841*22ce4affSfengbojiang #define MAX_FORMAT (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
842*22ce4affSfengbojiang
843*22ce4affSfengbojiang
addquoted(lua_State * L,luaL_Buffer * b,int arg)844*22ce4affSfengbojiang static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
845*22ce4affSfengbojiang size_t l;
846*22ce4affSfengbojiang const char *s = luaL_checklstring(L, arg, &l);
847*22ce4affSfengbojiang luaL_addchar(b, '"');
848*22ce4affSfengbojiang while (l--) {
849*22ce4affSfengbojiang if (*s == '"' || *s == '\\' || *s == '\n') {
850*22ce4affSfengbojiang luaL_addchar(b, '\\');
851*22ce4affSfengbojiang luaL_addchar(b, *s);
852*22ce4affSfengbojiang }
853*22ce4affSfengbojiang else if (*s == '\0' || iscntrl(uchar(*s))) {
854*22ce4affSfengbojiang char buff[10];
855*22ce4affSfengbojiang if (!isdigit(uchar(*(s+1))))
856*22ce4affSfengbojiang snprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s));
857*22ce4affSfengbojiang else
858*22ce4affSfengbojiang snprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s));
859*22ce4affSfengbojiang luaL_addstring(b, buff);
860*22ce4affSfengbojiang }
861*22ce4affSfengbojiang else
862*22ce4affSfengbojiang luaL_addchar(b, *s);
863*22ce4affSfengbojiang s++;
864*22ce4affSfengbojiang }
865*22ce4affSfengbojiang luaL_addchar(b, '"');
866*22ce4affSfengbojiang }
867*22ce4affSfengbojiang
scanformat(lua_State * L,const char * strfrmt,char * form)868*22ce4affSfengbojiang static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
869*22ce4affSfengbojiang const char *p = strfrmt;
870*22ce4affSfengbojiang while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */
871*22ce4affSfengbojiang if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
872*22ce4affSfengbojiang luaL_error(L, "invalid format (repeated flags)");
873*22ce4affSfengbojiang if (isdigit(uchar(*p))) p++; /* skip width */
874*22ce4affSfengbojiang if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
875*22ce4affSfengbojiang if (*p == '.') {
876*22ce4affSfengbojiang p++;
877*22ce4affSfengbojiang if (isdigit(uchar(*p))) p++; /* skip precision */
878*22ce4affSfengbojiang if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
879*22ce4affSfengbojiang }
880*22ce4affSfengbojiang if (isdigit(uchar(*p)))
881*22ce4affSfengbojiang luaL_error(L, "invalid format (width or precision too long)");
882*22ce4affSfengbojiang *(form++) = '%';
883*22ce4affSfengbojiang memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char));
884*22ce4affSfengbojiang form += p - strfrmt + 1;
885*22ce4affSfengbojiang *form = '\0';
886*22ce4affSfengbojiang return p;
887*22ce4affSfengbojiang }
888*22ce4affSfengbojiang
889*22ce4affSfengbojiang
890*22ce4affSfengbojiang /*
891*22ce4affSfengbojiang ** add length modifier into formats
892*22ce4affSfengbojiang */
addlenmod(char * form,const char * lenmod,size_t size)893*22ce4affSfengbojiang static void addlenmod (char *form, const char *lenmod, size_t size) {
894*22ce4affSfengbojiang size_t l = strlen(form);
895*22ce4affSfengbojiang size_t lm = strlen(lenmod);
896*22ce4affSfengbojiang char spec = form[l - 1];
897*22ce4affSfengbojiang strlcpy(form + l - 1, lenmod, size - (l - 1));
898*22ce4affSfengbojiang form[l + lm - 1] = spec;
899*22ce4affSfengbojiang form[l + lm] = '\0';
900*22ce4affSfengbojiang }
901*22ce4affSfengbojiang
902*22ce4affSfengbojiang
str_format(lua_State * L)903*22ce4affSfengbojiang static int str_format (lua_State *L) {
904*22ce4affSfengbojiang int top = lua_gettop(L);
905*22ce4affSfengbojiang int arg = 1;
906*22ce4affSfengbojiang size_t sfl;
907*22ce4affSfengbojiang const char *strfrmt = luaL_checklstring(L, arg, &sfl);
908*22ce4affSfengbojiang const char *strfrmt_end = strfrmt+sfl;
909*22ce4affSfengbojiang luaL_Buffer b;
910*22ce4affSfengbojiang luaL_buffinit(L, &b);
911*22ce4affSfengbojiang while (strfrmt < strfrmt_end) {
912*22ce4affSfengbojiang if (*strfrmt != L_ESC)
913*22ce4affSfengbojiang luaL_addchar(&b, *strfrmt++);
914*22ce4affSfengbojiang else if (*++strfrmt == L_ESC)
915*22ce4affSfengbojiang luaL_addchar(&b, *strfrmt++); /* %% */
916*22ce4affSfengbojiang else { /* format item */
917*22ce4affSfengbojiang char form[MAX_FORMAT]; /* to store the format (`%...') */
918*22ce4affSfengbojiang char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */
919*22ce4affSfengbojiang int nb = 0; /* number of bytes in added item */
920*22ce4affSfengbojiang if (++arg > top)
921*22ce4affSfengbojiang luaL_argerror(L, arg, "no value");
922*22ce4affSfengbojiang strfrmt = scanformat(L, strfrmt, form);
923*22ce4affSfengbojiang switch (*strfrmt++) {
924*22ce4affSfengbojiang case 'c': {
925*22ce4affSfengbojiang nb = str_sprintf(buff, form, luaL_checkint(L, arg));
926*22ce4affSfengbojiang break;
927*22ce4affSfengbojiang }
928*22ce4affSfengbojiang case 'd': case 'i': {
929*22ce4affSfengbojiang lua_Number n = luaL_checknumber(L, arg);
930*22ce4affSfengbojiang LUA_INTFRM_T ni = (LUA_INTFRM_T)n;
931*22ce4affSfengbojiang lua_Number diff = n - (lua_Number)ni;
932*22ce4affSfengbojiang luaL_argcheck(L, -1 < diff && diff < 1, arg,
933*22ce4affSfengbojiang "not a number in proper range");
934*22ce4affSfengbojiang addlenmod(form, LUA_INTFRMLEN, MAX_FORMAT);
935*22ce4affSfengbojiang nb = str_sprintf(buff, form, ni);
936*22ce4affSfengbojiang break;
937*22ce4affSfengbojiang }
938*22ce4affSfengbojiang case 'o': case 'u': case 'x': case 'X': {
939*22ce4affSfengbojiang lua_Number n = luaL_checknumber(L, arg);
940*22ce4affSfengbojiang unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n;
941*22ce4affSfengbojiang lua_Number diff = n - (lua_Number)ni;
942*22ce4affSfengbojiang luaL_argcheck(L, -1 < diff && diff < 1, arg,
943*22ce4affSfengbojiang "not a non-negative number in proper range");
944*22ce4affSfengbojiang addlenmod(form, LUA_INTFRMLEN, MAX_FORMAT);
945*22ce4affSfengbojiang nb = str_sprintf(buff, form, ni);
946*22ce4affSfengbojiang break;
947*22ce4affSfengbojiang }
948*22ce4affSfengbojiang #if defined(LUA_USE_FLOAT_FORMATS)
949*22ce4affSfengbojiang case 'e': case 'E': case 'f':
950*22ce4affSfengbojiang #if defined(LUA_USE_AFORMAT)
951*22ce4affSfengbojiang case 'a': case 'A':
952*22ce4affSfengbojiang #endif
953*22ce4affSfengbojiang case 'g': case 'G': {
954*22ce4affSfengbojiang addlenmod(form, LUA_FLTFRMLEN, MAX_FORMAT);
955*22ce4affSfengbojiang nb = str_sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg));
956*22ce4affSfengbojiang break;
957*22ce4affSfengbojiang }
958*22ce4affSfengbojiang #endif
959*22ce4affSfengbojiang case 'q': {
960*22ce4affSfengbojiang addquoted(L, &b, arg);
961*22ce4affSfengbojiang break;
962*22ce4affSfengbojiang }
963*22ce4affSfengbojiang case 's': {
964*22ce4affSfengbojiang size_t l;
965*22ce4affSfengbojiang const char *s = luaL_tolstring(L, arg, &l);
966*22ce4affSfengbojiang if (!strchr(form, '.') && l >= 100) {
967*22ce4affSfengbojiang /* no precision and string is too long to be formatted;
968*22ce4affSfengbojiang keep original string */
969*22ce4affSfengbojiang luaL_addvalue(&b);
970*22ce4affSfengbojiang break;
971*22ce4affSfengbojiang }
972*22ce4affSfengbojiang else {
973*22ce4affSfengbojiang nb = str_sprintf(buff, form, s);
974*22ce4affSfengbojiang lua_pop(L, 1); /* remove result from 'luaL_tolstring' */
975*22ce4affSfengbojiang break;
976*22ce4affSfengbojiang }
977*22ce4affSfengbojiang }
978*22ce4affSfengbojiang default: { /* also treat cases `pnLlh' */
979*22ce4affSfengbojiang return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
980*22ce4affSfengbojiang LUA_QL("format"), *(strfrmt - 1));
981*22ce4affSfengbojiang }
982*22ce4affSfengbojiang }
983*22ce4affSfengbojiang luaL_addsize(&b, nb);
984*22ce4affSfengbojiang }
985*22ce4affSfengbojiang }
986*22ce4affSfengbojiang luaL_pushresult(&b);
987*22ce4affSfengbojiang return 1;
988*22ce4affSfengbojiang }
989*22ce4affSfengbojiang
990*22ce4affSfengbojiang /* }====================================================== */
991*22ce4affSfengbojiang
992*22ce4affSfengbojiang
993*22ce4affSfengbojiang static const luaL_Reg strlib[] = {
994*22ce4affSfengbojiang {"byte", str_byte},
995*22ce4affSfengbojiang {"char", str_char},
996*22ce4affSfengbojiang #if defined(LUA_USE_DUMP)
997*22ce4affSfengbojiang {"dump", str_dump},
998*22ce4affSfengbojiang #endif
999*22ce4affSfengbojiang {"find", str_find},
1000*22ce4affSfengbojiang {"format", str_format},
1001*22ce4affSfengbojiang {"gmatch", str_gmatch},
1002*22ce4affSfengbojiang {"gsub", str_gsub},
1003*22ce4affSfengbojiang {"len", str_len},
1004*22ce4affSfengbojiang {"lower", str_lower},
1005*22ce4affSfengbojiang {"match", str_match},
1006*22ce4affSfengbojiang {"rep", str_rep},
1007*22ce4affSfengbojiang {"reverse", str_reverse},
1008*22ce4affSfengbojiang {"sub", str_sub},
1009*22ce4affSfengbojiang {"upper", str_upper},
1010*22ce4affSfengbojiang {NULL, NULL}
1011*22ce4affSfengbojiang };
1012*22ce4affSfengbojiang
1013*22ce4affSfengbojiang
createmetatable(lua_State * L)1014*22ce4affSfengbojiang static void createmetatable (lua_State *L) {
1015*22ce4affSfengbojiang lua_createtable(L, 0, 1); /* table to be metatable for strings */
1016*22ce4affSfengbojiang lua_pushliteral(L, ""); /* dummy string */
1017*22ce4affSfengbojiang lua_pushvalue(L, -2); /* copy table */
1018*22ce4affSfengbojiang lua_setmetatable(L, -2); /* set table as metatable for strings */
1019*22ce4affSfengbojiang lua_pop(L, 1); /* pop dummy string */
1020*22ce4affSfengbojiang lua_pushvalue(L, -2); /* get string library */
1021*22ce4affSfengbojiang lua_setfield(L, -2, "__index"); /* metatable.__index = string */
1022*22ce4affSfengbojiang lua_pop(L, 1); /* pop metatable */
1023*22ce4affSfengbojiang }
1024*22ce4affSfengbojiang
1025*22ce4affSfengbojiang
1026*22ce4affSfengbojiang /*
1027*22ce4affSfengbojiang ** Open string library
1028*22ce4affSfengbojiang */
luaopen_string(lua_State * L)1029*22ce4affSfengbojiang LUAMOD_API int luaopen_string (lua_State *L) {
1030*22ce4affSfengbojiang luaL_newlib(L, strlib);
1031*22ce4affSfengbojiang createmetatable(L);
1032*22ce4affSfengbojiang return 1;
1033*22ce4affSfengbojiang }
1034*22ce4affSfengbojiang
1035*22ce4affSfengbojiang #if defined(_KERNEL)
1036*22ce4affSfengbojiang
1037*22ce4affSfengbojiang EXPORT_SYMBOL(luaopen_string);
1038*22ce4affSfengbojiang
1039*22ce4affSfengbojiang #endif
1040*22ce4affSfengbojiang /* END CSTYLED */
1041