diff options
author | Philipp Janda <siffiejoe@gmx.net> | 2015-01-21 14:37:41 +0100 |
---|---|---|
committer | Philipp Janda <siffiejoe@gmx.net> | 2015-01-21 14:37:41 +0100 |
commit | eec3f9e44bee334542348bdd1d428417e22ed5e5 (patch) | |
tree | 82fe996fb72ee5a86b9f872fcf57df8fb888a280 | |
parent | c57750c925553f627701f2c4897ee6bd6488c7c1 (diff) | |
download | lua-compat-5.3-eec3f9e44bee334542348bdd1d428417e22ed5e5.tar.gz lua-compat-5.3-eec3f9e44bee334542348bdd1d428417e22ed5e5.tar.bz2 lua-compat-5.3-eec3f9e44bee334542348bdd1d428417e22ed5e5.zip |
use Lua 5.3's source for string packing
-rw-r--r-- | compat53.lua | 27 | ||||
-rw-r--r-- | lprefix.h | 42 | ||||
-rw-r--r-- | lstrlib.c | 1430 | ||||
-rw-r--r-- | rockspecs/compat53-scm-0.rockspec | 1 |
4 files changed, 1491 insertions, 9 deletions
diff --git a/compat53.lua b/compat53.lua index 67a8c8f..4a78260 100644 --- a/compat53.lua +++ b/compat53.lua | |||
@@ -46,15 +46,24 @@ if lua_version < "5.3" then | |||
46 | end | 46 | end |
47 | 47 | ||
48 | 48 | ||
49 | -- use Roberto's struct module for string packing/unpacking for now | 49 | -- load string packing functions |
50 | -- maybe we'll later extract the functions from the 5.3 string | 50 | local str_ok, strlib = pcall(require, "compat53.string") |
51 | -- library for greater compatiblity, but it uses the 5.3 buffer API | 51 | if str_ok then |
52 | -- which cannot easily be backported to Lua 5.1. | 52 | for k,v in pairs(strlib) do |
53 | local struct_ok, struct = pcall(require, "struct") | 53 | string[k] = v |
54 | if struct_ok then | 54 | end |
55 | string.pack = struct.pack | 55 | end |
56 | string.packsize = struct.size | 56 | |
57 | string.unpack = struct.unpack | 57 | |
58 | -- try Roberto's struct module for string packing/unpacking if | ||
59 | -- compat53.string is unavailable | ||
60 | if not str_ok then | ||
61 | local struct_ok, struct = pcall(require, "struct") | ||
62 | if struct_ok then | ||
63 | string.pack = struct.pack | ||
64 | string.packsize = struct.size | ||
65 | string.unpack = struct.unpack | ||
66 | end | ||
58 | end | 67 | end |
59 | 68 | ||
60 | 69 | ||
@@ -54,6 +54,9 @@ | |||
54 | 54 | ||
55 | #ifdef ltablib_c | 55 | #ifdef ltablib_c |
56 | # define luaopen_table luaopen_compat53_table | 56 | # define luaopen_table luaopen_compat53_table |
57 | /* lua_rawgeti in compat53.h is implemented as a macro, so the | ||
58 | * function signature doesn't match when you use a function pointer | ||
59 | */ | ||
57 | static int compat53_rawgeti (lua_State *L, int i, lua_Integer n) { | 60 | static int compat53_rawgeti (lua_State *L, int i, lua_Integer n) { |
58 | return lua_rawgeti(L, i, n); | 61 | return lua_rawgeti(L, i, n); |
59 | } | 62 | } |
@@ -64,11 +67,50 @@ static void compat53_rawseti (lua_State *L, int i, lua_Integer n) { | |||
64 | } | 67 | } |
65 | # undef lua_rawseti | 68 | # undef lua_rawseti |
66 | # define lua_rawseti compat53_rawseti | 69 | # define lua_rawseti compat53_rawseti |
70 | /* we have lua_compare emulation for Lua 5.1, but it involves calling | ||
71 | * Lua code, and the only use in the table library is for '<', so ... | ||
72 | */ | ||
67 | # if LUA_VERSION_NUM == 501 | 73 | # if LUA_VERSION_NUM == 501 |
68 | # undef lua_compare | 74 | # undef lua_compare |
69 | # define lua_compare(L, a, b, op) lua_lessthan(L, a, b) | 75 | # define lua_compare(L, a, b, op) lua_lessthan(L, a, b) |
70 | # endif | 76 | # endif |
71 | #endif /* ltablib_c */ | 77 | #endif /* ltablib_c */ |
72 | 78 | ||
79 | #ifdef lstrlib_c | ||
80 | /* move the string library open function out of the way (we only take | ||
81 | * the string packing functions)! | ||
82 | */ | ||
83 | # define luaopen_string luaopen_string_XXX | ||
84 | /* used in string.format implementation, which we don't use: */ | ||
85 | # ifndef LUA_INTEGER_FRMLEN | ||
86 | # define LUA_INTEGER_FRMLEN "" | ||
87 | # define LUA_NUMBER_FRMLEN "" | ||
88 | # endif | ||
89 | # if LUA_VERSION_NUM < 503 | ||
90 | /* lstrlib assumes that lua_Integer and lua_Unsigned have the same | ||
91 | * size, so we use the unsigned equivalent of ptrdiff_t! */ | ||
92 | # define lua_Unsigned size_t | ||
93 | # endif | ||
94 | static int str_pack (lua_State *L); | ||
95 | static int str_packsize (lua_State *L); | ||
96 | static int str_unpack (lua_State *L); | ||
97 | LUAMOD_API int luaopen_compat53_string (lua_State *L) { | ||
98 | luaL_Reg const funcs[] = { | ||
99 | { "pack", str_pack }, | ||
100 | { "packsize", str_packsize }, | ||
101 | { "unpack", str_unpack }, | ||
102 | { NULL, NULL } | ||
103 | }; | ||
104 | luaL_newlib(L, funcs); | ||
105 | return 1; | ||
106 | } | ||
107 | /* make luaopen_string(_XXX) static, so it (and all other referenced | ||
108 | * string functions) won't be included in the resulting dll | ||
109 | * (hopefully). | ||
110 | */ | ||
111 | # undef LUAMOD_API | ||
112 | # define LUAMOD_API static | ||
113 | #endif /* lstrlib.c */ | ||
114 | |||
73 | #endif | 115 | #endif |
74 | 116 | ||
diff --git a/lstrlib.c b/lstrlib.c new file mode 100644 index 0000000..a650b76 --- /dev/null +++ b/lstrlib.c | |||
@@ -0,0 +1,1430 @@ | |||
1 | /* | ||
2 | ** $Id: lstrlib.c,v 1.221 2014/12/11 14:03:07 roberto Exp $ | ||
3 | ** Standard library for string operations and pattern-matching | ||
4 | ** See Copyright Notice in lua.h | ||
5 | */ | ||
6 | |||
7 | #define lstrlib_c | ||
8 | #define LUA_LIB | ||
9 | |||
10 | #include "lprefix.h" | ||
11 | |||
12 | |||
13 | #include <ctype.h> | ||
14 | #include <limits.h> | ||
15 | #include <stddef.h> | ||
16 | #include <stdio.h> | ||
17 | #include <stdlib.h> | ||
18 | #include <string.h> | ||
19 | |||
20 | #include "lua.h" | ||
21 | |||
22 | #include "lauxlib.h" | ||
23 | #include "lualib.h" | ||
24 | |||
25 | |||
26 | /* | ||
27 | ** maximum number of captures that a pattern can do during | ||
28 | ** pattern-matching. This limit is arbitrary. | ||
29 | */ | ||
30 | #if !defined(LUA_MAXCAPTURES) | ||
31 | #define LUA_MAXCAPTURES 32 | ||
32 | #endif | ||
33 | |||
34 | |||
35 | /* macro to 'unsign' a character */ | ||
36 | #define uchar(c) ((unsigned char)(c)) | ||
37 | |||
38 | |||
39 | /* | ||
40 | ** Some sizes are better limited to fit in 'int', but must also fit in | ||
41 | ** 'size_t'. (We assume that 'lua_Integer' cannot be smaller than 'int'.) | ||
42 | */ | ||
43 | #define MAXSIZE \ | ||
44 | (sizeof(size_t) < sizeof(int) ? (~(size_t)0) : (size_t)(INT_MAX)) | ||
45 | |||
46 | |||
47 | |||
48 | |||
49 | static int str_len (lua_State *L) { | ||
50 | size_t l; | ||
51 | luaL_checklstring(L, 1, &l); | ||
52 | lua_pushinteger(L, (lua_Integer)l); | ||
53 | return 1; | ||
54 | } | ||
55 | |||
56 | |||
57 | /* translate a relative string position: negative means back from end */ | ||
58 | static lua_Integer posrelat (lua_Integer pos, size_t len) { | ||
59 | if (pos >= 0) return pos; | ||
60 | else if (0u - (size_t)pos > len) return 0; | ||
61 | else return (lua_Integer)len + pos + 1; | ||
62 | } | ||
63 | |||
64 | |||
65 | static int str_sub (lua_State *L) { | ||
66 | size_t l; | ||
67 | const char *s = luaL_checklstring(L, 1, &l); | ||
68 | lua_Integer start = posrelat(luaL_checkinteger(L, 2), l); | ||
69 | lua_Integer end = posrelat(luaL_optinteger(L, 3, -1), l); | ||
70 | if (start < 1) start = 1; | ||
71 | if (end > (lua_Integer)l) end = l; | ||
72 | if (start <= end) | ||
73 | lua_pushlstring(L, s + start - 1, (size_t)(end - start + 1)); | ||
74 | else lua_pushliteral(L, ""); | ||
75 | return 1; | ||
76 | } | ||
77 | |||
78 | |||
79 | static int str_reverse (lua_State *L) { | ||
80 | size_t l, i; | ||
81 | luaL_Buffer b; | ||
82 | const char *s = luaL_checklstring(L, 1, &l); | ||
83 | char *p = luaL_buffinitsize(L, &b, l); | ||
84 | for (i = 0; i < l; i++) | ||
85 | p[i] = s[l - i - 1]; | ||
86 | luaL_pushresultsize(&b, l); | ||
87 | return 1; | ||
88 | } | ||
89 | |||
90 | |||
91 | static int str_lower (lua_State *L) { | ||
92 | size_t l; | ||
93 | size_t i; | ||
94 | luaL_Buffer b; | ||
95 | const char *s = luaL_checklstring(L, 1, &l); | ||
96 | char *p = luaL_buffinitsize(L, &b, l); | ||
97 | for (i=0; i<l; i++) | ||
98 | p[i] = tolower(uchar(s[i])); | ||
99 | luaL_pushresultsize(&b, l); | ||
100 | return 1; | ||
101 | } | ||
102 | |||
103 | |||
104 | static int str_upper (lua_State *L) { | ||
105 | size_t l; | ||
106 | size_t i; | ||
107 | luaL_Buffer b; | ||
108 | const char *s = luaL_checklstring(L, 1, &l); | ||
109 | char *p = luaL_buffinitsize(L, &b, l); | ||
110 | for (i=0; i<l; i++) | ||
111 | p[i] = toupper(uchar(s[i])); | ||
112 | luaL_pushresultsize(&b, l); | ||
113 | return 1; | ||
114 | } | ||
115 | |||
116 | |||
117 | static int str_rep (lua_State *L) { | ||
118 | size_t l, lsep; | ||
119 | const char *s = luaL_checklstring(L, 1, &l); | ||
120 | lua_Integer n = luaL_checkinteger(L, 2); | ||
121 | const char *sep = luaL_optlstring(L, 3, "", &lsep); | ||
122 | if (n <= 0) lua_pushliteral(L, ""); | ||
123 | else if (l + lsep < l || l + lsep > MAXSIZE / n) /* may overflow? */ | ||
124 | return luaL_error(L, "resulting string too large"); | ||
125 | else { | ||
126 | size_t totallen = (size_t)n * l + (size_t)(n - 1) * lsep; | ||
127 | luaL_Buffer b; | ||
128 | char *p = luaL_buffinitsize(L, &b, totallen); | ||
129 | while (n-- > 1) { /* first n-1 copies (followed by separator) */ | ||
130 | memcpy(p, s, l * sizeof(char)); p += l; | ||
131 | if (lsep > 0) { /* empty 'memcpy' is not that cheap */ | ||
132 | memcpy(p, sep, lsep * sizeof(char)); | ||
133 | p += lsep; | ||
134 | } | ||
135 | } | ||
136 | memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */ | ||
137 | luaL_pushresultsize(&b, totallen); | ||
138 | } | ||
139 | return 1; | ||
140 | } | ||
141 | |||
142 | |||
143 | static int str_byte (lua_State *L) { | ||
144 | size_t l; | ||
145 | const char *s = luaL_checklstring(L, 1, &l); | ||
146 | lua_Integer posi = posrelat(luaL_optinteger(L, 2, 1), l); | ||
147 | lua_Integer pose = posrelat(luaL_optinteger(L, 3, posi), l); | ||
148 | int n, i; | ||
149 | if (posi < 1) posi = 1; | ||
150 | if (pose > (lua_Integer)l) pose = l; | ||
151 | if (posi > pose) return 0; /* empty interval; return no values */ | ||
152 | n = (int)(pose - posi + 1); | ||
153 | if (posi + n <= pose) /* arithmetic overflow? */ | ||
154 | return luaL_error(L, "string slice too long"); | ||
155 | luaL_checkstack(L, n, "string slice too long"); | ||
156 | for (i=0; i<n; i++) | ||
157 | lua_pushinteger(L, uchar(s[posi+i-1])); | ||
158 | return n; | ||
159 | } | ||
160 | |||
161 | |||
162 | static int str_char (lua_State *L) { | ||
163 | int n = lua_gettop(L); /* number of arguments */ | ||
164 | int i; | ||
165 | luaL_Buffer b; | ||
166 | char *p = luaL_buffinitsize(L, &b, n); | ||
167 | for (i=1; i<=n; i++) { | ||
168 | lua_Integer c = luaL_checkinteger(L, i); | ||
169 | luaL_argcheck(L, uchar(c) == c, i, "value out of range"); | ||
170 | p[i - 1] = uchar(c); | ||
171 | } | ||
172 | luaL_pushresultsize(&b, n); | ||
173 | return 1; | ||
174 | } | ||
175 | |||
176 | |||
177 | static int writer (lua_State *L, const void *b, size_t size, void *B) { | ||
178 | (void)L; | ||
179 | luaL_addlstring((luaL_Buffer *) B, (const char *)b, size); | ||
180 | return 0; | ||
181 | } | ||
182 | |||
183 | |||
184 | static int str_dump (lua_State *L) { | ||
185 | luaL_Buffer b; | ||
186 | int strip = lua_toboolean(L, 2); | ||
187 | luaL_checktype(L, 1, LUA_TFUNCTION); | ||
188 | lua_settop(L, 1); | ||
189 | luaL_buffinit(L,&b); | ||
190 | if (lua_dump(L, writer, &b, strip) != 0) | ||
191 | return luaL_error(L, "unable to dump given function"); | ||
192 | luaL_pushresult(&b); | ||
193 | return 1; | ||
194 | } | ||
195 | |||
196 | |||
197 | |||
198 | /* | ||
199 | ** {====================================================== | ||
200 | ** PATTERN MATCHING | ||
201 | ** ======================================================= | ||
202 | */ | ||
203 | |||
204 | |||
205 | #define CAP_UNFINISHED (-1) | ||
206 | #define CAP_POSITION (-2) | ||
207 | |||
208 | |||
209 | typedef struct MatchState { | ||
210 | int matchdepth; /* control for recursive depth (to avoid C stack overflow) */ | ||
211 | const char *src_init; /* init of source string */ | ||
212 | const char *src_end; /* end ('\0') of source string */ | ||
213 | const char *p_end; /* end ('\0') of pattern */ | ||
214 | lua_State *L; | ||
215 | int level; /* total number of captures (finished or unfinished) */ | ||
216 | struct { | ||
217 | const char *init; | ||
218 | ptrdiff_t len; | ||
219 | } capture[LUA_MAXCAPTURES]; | ||
220 | } MatchState; | ||
221 | |||
222 | |||
223 | /* recursive function */ | ||
224 | static const char *match (MatchState *ms, const char *s, const char *p); | ||
225 | |||
226 | |||
227 | /* maximum recursion depth for 'match' */ | ||
228 | #if !defined(MAXCCALLS) | ||
229 | #define MAXCCALLS 200 | ||
230 | #endif | ||
231 | |||
232 | |||
233 | #define L_ESC '%' | ||
234 | #define SPECIALS "^$*+?.([%-" | ||
235 | |||
236 | |||
237 | static int check_capture (MatchState *ms, int l) { | ||
238 | l -= '1'; | ||
239 | if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED) | ||
240 | return luaL_error(ms->L, "invalid capture index %%%d", l + 1); | ||
241 | return l; | ||
242 | } | ||
243 | |||
244 | |||
245 | static int capture_to_close (MatchState *ms) { | ||
246 | int level = ms->level; | ||
247 | for (level--; level>=0; level--) | ||
248 | if (ms->capture[level].len == CAP_UNFINISHED) return level; | ||
249 | return luaL_error(ms->L, "invalid pattern capture"); | ||
250 | } | ||
251 | |||
252 | |||
253 | static const char *classend (MatchState *ms, const char *p) { | ||
254 | switch (*p++) { | ||
255 | case L_ESC: { | ||
256 | if (p == ms->p_end) | ||
257 | luaL_error(ms->L, "malformed pattern (ends with '%%')"); | ||
258 | return p+1; | ||
259 | } | ||
260 | case '[': { | ||
261 | if (*p == '^') p++; | ||
262 | do { /* look for a ']' */ | ||
263 | if (p == ms->p_end) | ||
264 | luaL_error(ms->L, "malformed pattern (missing ']')"); | ||
265 | if (*(p++) == L_ESC && p < ms->p_end) | ||
266 | p++; /* skip escapes (e.g. '%]') */ | ||
267 | } while (*p != ']'); | ||
268 | return p+1; | ||
269 | } | ||
270 | default: { | ||
271 | return p; | ||
272 | } | ||
273 | } | ||
274 | } | ||
275 | |||
276 | |||
277 | static int match_class (int c, int cl) { | ||
278 | int res; | ||
279 | switch (tolower(cl)) { | ||
280 | case 'a' : res = isalpha(c); break; | ||
281 | case 'c' : res = iscntrl(c); break; | ||
282 | case 'd' : res = isdigit(c); break; | ||
283 | case 'g' : res = isgraph(c); break; | ||
284 | case 'l' : res = islower(c); break; | ||
285 | case 'p' : res = ispunct(c); break; | ||
286 | case 's' : res = isspace(c); break; | ||
287 | case 'u' : res = isupper(c); break; | ||
288 | case 'w' : res = isalnum(c); break; | ||
289 | case 'x' : res = isxdigit(c); break; | ||
290 | case 'z' : res = (c == 0); break; /* deprecated option */ | ||
291 | default: return (cl == c); | ||
292 | } | ||
293 | return (islower(cl) ? res : !res); | ||
294 | } | ||
295 | |||
296 | |||
297 | static int matchbracketclass (int c, const char *p, const char *ec) { | ||
298 | int sig = 1; | ||
299 | if (*(p+1) == '^') { | ||
300 | sig = 0; | ||
301 | p++; /* skip the '^' */ | ||
302 | } | ||
303 | while (++p < ec) { | ||
304 | if (*p == L_ESC) { | ||
305 | p++; | ||
306 | if (match_class(c, uchar(*p))) | ||
307 | return sig; | ||
308 | } | ||
309 | else if ((*(p+1) == '-') && (p+2 < ec)) { | ||
310 | p+=2; | ||
311 | if (uchar(*(p-2)) <= c && c <= uchar(*p)) | ||
312 | return sig; | ||
313 | } | ||
314 | else if (uchar(*p) == c) return sig; | ||
315 | } | ||
316 | return !sig; | ||
317 | } | ||
318 | |||
319 | |||
320 | static int singlematch (MatchState *ms, const char *s, const char *p, | ||
321 | const char *ep) { | ||
322 | if (s >= ms->src_end) | ||
323 | return 0; | ||
324 | else { | ||
325 | int c = uchar(*s); | ||
326 | switch (*p) { | ||
327 | case '.': return 1; /* matches any char */ | ||
328 | case L_ESC: return match_class(c, uchar(*(p+1))); | ||
329 | case '[': return matchbracketclass(c, p, ep-1); | ||
330 | default: return (uchar(*p) == c); | ||
331 | } | ||
332 | } | ||
333 | } | ||
334 | |||
335 | |||
336 | static const char *matchbalance (MatchState *ms, const char *s, | ||
337 | const char *p) { | ||
338 | if (p >= ms->p_end - 1) | ||
339 | luaL_error(ms->L, "malformed pattern (missing arguments to '%%b')"); | ||
340 | if (*s != *p) return NULL; | ||
341 | else { | ||
342 | int b = *p; | ||
343 | int e = *(p+1); | ||
344 | int cont = 1; | ||
345 | while (++s < ms->src_end) { | ||
346 | if (*s == e) { | ||
347 | if (--cont == 0) return s+1; | ||
348 | } | ||
349 | else if (*s == b) cont++; | ||
350 | } | ||
351 | } | ||
352 | return NULL; /* string ends out of balance */ | ||
353 | } | ||
354 | |||
355 | |||
356 | static const char *max_expand (MatchState *ms, const char *s, | ||
357 | const char *p, const char *ep) { | ||
358 | ptrdiff_t i = 0; /* counts maximum expand for item */ | ||
359 | while (singlematch(ms, s + i, p, ep)) | ||
360 | i++; | ||
361 | /* keeps trying to match with the maximum repetitions */ | ||
362 | while (i>=0) { | ||
363 | const char *res = match(ms, (s+i), ep+1); | ||
364 | if (res) return res; | ||
365 | i--; /* else didn't match; reduce 1 repetition to try again */ | ||
366 | } | ||
367 | return NULL; | ||
368 | } | ||
369 | |||
370 | |||
371 | static const char *min_expand (MatchState *ms, const char *s, | ||
372 | const char *p, const char *ep) { | ||
373 | for (;;) { | ||
374 | const char *res = match(ms, s, ep+1); | ||
375 | if (res != NULL) | ||
376 | return res; | ||
377 | else if (singlematch(ms, s, p, ep)) | ||
378 | s++; /* try with one more repetition */ | ||
379 | else return NULL; | ||
380 | } | ||
381 | } | ||
382 | |||
383 | |||
384 | static const char *start_capture (MatchState *ms, const char *s, | ||
385 | const char *p, int what) { | ||
386 | const char *res; | ||
387 | int level = ms->level; | ||
388 | if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures"); | ||
389 | ms->capture[level].init = s; | ||
390 | ms->capture[level].len = what; | ||
391 | ms->level = level+1; | ||
392 | if ((res=match(ms, s, p)) == NULL) /* match failed? */ | ||
393 | ms->level--; /* undo capture */ | ||
394 | return res; | ||
395 | } | ||
396 | |||
397 | |||
398 | static const char *end_capture (MatchState *ms, const char *s, | ||
399 | const char *p) { | ||
400 | int l = capture_to_close(ms); | ||
401 | const char *res; | ||
402 | ms->capture[l].len = s - ms->capture[l].init; /* close capture */ | ||
403 | if ((res = match(ms, s, p)) == NULL) /* match failed? */ | ||
404 | ms->capture[l].len = CAP_UNFINISHED; /* undo capture */ | ||
405 | return res; | ||
406 | } | ||
407 | |||
408 | |||
409 | static const char *match_capture (MatchState *ms, const char *s, int l) { | ||
410 | size_t len; | ||
411 | l = check_capture(ms, l); | ||
412 | len = ms->capture[l].len; | ||
413 | if ((size_t)(ms->src_end-s) >= len && | ||
414 | memcmp(ms->capture[l].init, s, len) == 0) | ||
415 | return s+len; | ||
416 | else return NULL; | ||
417 | } | ||
418 | |||
419 | |||
420 | static const char *match (MatchState *ms, const char *s, const char *p) { | ||
421 | if (ms->matchdepth-- == 0) | ||
422 | luaL_error(ms->L, "pattern too complex"); | ||
423 | init: /* using goto's to optimize tail recursion */ | ||
424 | if (p != ms->p_end) { /* end of pattern? */ | ||
425 | switch (*p) { | ||
426 | case '(': { /* start capture */ | ||
427 | if (*(p + 1) == ')') /* position capture? */ | ||
428 | s = start_capture(ms, s, p + 2, CAP_POSITION); | ||
429 | else | ||
430 | s = start_capture(ms, s, p + 1, CAP_UNFINISHED); | ||
431 | break; | ||
432 | } | ||
433 | case ')': { /* end capture */ | ||
434 | s = end_capture(ms, s, p + 1); | ||
435 | break; | ||
436 | } | ||
437 | case '$': { | ||
438 | if ((p + 1) != ms->p_end) /* is the '$' the last char in pattern? */ | ||
439 | goto dflt; /* no; go to default */ | ||
440 | s = (s == ms->src_end) ? s : NULL; /* check end of string */ | ||
441 | break; | ||
442 | } | ||
443 | case L_ESC: { /* escaped sequences not in the format class[*+?-]? */ | ||
444 | switch (*(p + 1)) { | ||
445 | case 'b': { /* balanced string? */ | ||
446 | s = matchbalance(ms, s, p + 2); | ||
447 | if (s != NULL) { | ||
448 | p += 4; goto init; /* return match(ms, s, p + 4); */ | ||
449 | } /* else fail (s == NULL) */ | ||
450 | break; | ||
451 | } | ||
452 | case 'f': { /* frontier? */ | ||
453 | const char *ep; char previous; | ||
454 | p += 2; | ||
455 | if (*p != '[') | ||
456 | luaL_error(ms->L, "missing '[' after '%%f' in pattern"); | ||
457 | ep = classend(ms, p); /* points to what is next */ | ||
458 | previous = (s == ms->src_init) ? '\0' : *(s - 1); | ||
459 | if (!matchbracketclass(uchar(previous), p, ep - 1) && | ||
460 | matchbracketclass(uchar(*s), p, ep - 1)) { | ||
461 | p = ep; goto init; /* return match(ms, s, ep); */ | ||
462 | } | ||
463 | s = NULL; /* match failed */ | ||
464 | break; | ||
465 | } | ||
466 | case '0': case '1': case '2': case '3': | ||
467 | case '4': case '5': case '6': case '7': | ||
468 | case '8': case '9': { /* capture results (%0-%9)? */ | ||
469 | s = match_capture(ms, s, uchar(*(p + 1))); | ||
470 | if (s != NULL) { | ||
471 | p += 2; goto init; /* return match(ms, s, p + 2) */ | ||
472 | } | ||
473 | break; | ||
474 | } | ||
475 | default: goto dflt; | ||
476 | } | ||
477 | break; | ||
478 | } | ||
479 | default: dflt: { /* pattern class plus optional suffix */ | ||
480 | const char *ep = classend(ms, p); /* points to optional suffix */ | ||
481 | /* does not match at least once? */ | ||
482 | if (!singlematch(ms, s, p, ep)) { | ||
483 | if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */ | ||
484 | p = ep + 1; goto init; /* return match(ms, s, ep + 1); */ | ||
485 | } | ||
486 | else /* '+' or no suffix */ | ||
487 | s = NULL; /* fail */ | ||
488 | } | ||
489 | else { /* matched once */ | ||
490 | switch (*ep) { /* handle optional suffix */ | ||
491 | case '?': { /* optional */ | ||
492 | const char *res; | ||
493 | if ((res = match(ms, s + 1, ep + 1)) != NULL) | ||
494 | s = res; | ||
495 | else { | ||
496 | p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */ | ||
497 | } | ||
498 | break; | ||
499 | } | ||
500 | case '+': /* 1 or more repetitions */ | ||
501 | s++; /* 1 match already done */ | ||
502 | /* go through */ | ||
503 | case '*': /* 0 or more repetitions */ | ||
504 | s = max_expand(ms, s, p, ep); | ||
505 | break; | ||
506 | case '-': /* 0 or more repetitions (minimum) */ | ||
507 | s = min_expand(ms, s, p, ep); | ||
508 | break; | ||
509 | default: /* no suffix */ | ||
510 | s++; p = ep; goto init; /* return match(ms, s + 1, ep); */ | ||
511 | } | ||
512 | } | ||
513 | break; | ||
514 | } | ||
515 | } | ||
516 | } | ||
517 | ms->matchdepth++; | ||
518 | return s; | ||
519 | } | ||
520 | |||
521 | |||
522 | |||
523 | static const char *lmemfind (const char *s1, size_t l1, | ||
524 | const char *s2, size_t l2) { | ||
525 | if (l2 == 0) return s1; /* empty strings are everywhere */ | ||
526 | else if (l2 > l1) return NULL; /* avoids a negative 'l1' */ | ||
527 | else { | ||
528 | const char *init; /* to search for a '*s2' inside 's1' */ | ||
529 | l2--; /* 1st char will be checked by 'memchr' */ | ||
530 | l1 = l1-l2; /* 's2' cannot be found after that */ | ||
531 | while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) { | ||
532 | init++; /* 1st char is already checked */ | ||
533 | if (memcmp(init, s2+1, l2) == 0) | ||
534 | return init-1; | ||
535 | else { /* correct 'l1' and 's1' to try again */ | ||
536 | l1 -= init-s1; | ||
537 | s1 = init; | ||
538 | } | ||
539 | } | ||
540 | return NULL; /* not found */ | ||
541 | } | ||
542 | } | ||
543 | |||
544 | |||
545 | static void push_onecapture (MatchState *ms, int i, const char *s, | ||
546 | const char *e) { | ||
547 | if (i >= ms->level) { | ||
548 | if (i == 0) /* ms->level == 0, too */ | ||
549 | lua_pushlstring(ms->L, s, e - s); /* add whole match */ | ||
550 | else | ||
551 | luaL_error(ms->L, "invalid capture index %%%d", i + 1); | ||
552 | } | ||
553 | else { | ||
554 | ptrdiff_t l = ms->capture[i].len; | ||
555 | if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture"); | ||
556 | if (l == CAP_POSITION) | ||
557 | lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1); | ||
558 | else | ||
559 | lua_pushlstring(ms->L, ms->capture[i].init, l); | ||
560 | } | ||
561 | } | ||
562 | |||
563 | |||
564 | static int push_captures (MatchState *ms, const char *s, const char *e) { | ||
565 | int i; | ||
566 | int nlevels = (ms->level == 0 && s) ? 1 : ms->level; | ||
567 | luaL_checkstack(ms->L, nlevels, "too many captures"); | ||
568 | for (i = 0; i < nlevels; i++) | ||
569 | push_onecapture(ms, i, s, e); | ||
570 | return nlevels; /* number of strings pushed */ | ||
571 | } | ||
572 | |||
573 | |||
574 | /* check whether pattern has no special characters */ | ||
575 | static int nospecials (const char *p, size_t l) { | ||
576 | size_t upto = 0; | ||
577 | do { | ||
578 | if (strpbrk(p + upto, SPECIALS)) | ||
579 | return 0; /* pattern has a special character */ | ||
580 | upto += strlen(p + upto) + 1; /* may have more after \0 */ | ||
581 | } while (upto <= l); | ||
582 | return 1; /* no special chars found */ | ||
583 | } | ||
584 | |||
585 | |||
586 | static int str_find_aux (lua_State *L, int find) { | ||
587 | size_t ls, lp; | ||
588 | const char *s = luaL_checklstring(L, 1, &ls); | ||
589 | const char *p = luaL_checklstring(L, 2, &lp); | ||
590 | lua_Integer init = posrelat(luaL_optinteger(L, 3, 1), ls); | ||
591 | if (init < 1) init = 1; | ||
592 | else if (init > (lua_Integer)ls + 1) { /* start after string's end? */ | ||
593 | lua_pushnil(L); /* cannot find anything */ | ||
594 | return 1; | ||
595 | } | ||
596 | /* explicit request or no special characters? */ | ||
597 | if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) { | ||
598 | /* do a plain search */ | ||
599 | const char *s2 = lmemfind(s + init - 1, ls - (size_t)init + 1, p, lp); | ||
600 | if (s2) { | ||
601 | lua_pushinteger(L, s2 - s + 1); | ||
602 | lua_pushinteger(L, s2 - s + lp); | ||
603 | return 2; | ||
604 | } | ||
605 | } | ||
606 | else { | ||
607 | MatchState ms; | ||
608 | const char *s1 = s + init - 1; | ||
609 | int anchor = (*p == '^'); | ||
610 | if (anchor) { | ||
611 | p++; lp--; /* skip anchor character */ | ||
612 | } | ||
613 | ms.L = L; | ||
614 | ms.matchdepth = MAXCCALLS; | ||
615 | ms.src_init = s; | ||
616 | ms.src_end = s + ls; | ||
617 | ms.p_end = p + lp; | ||
618 | do { | ||
619 | const char *res; | ||
620 | ms.level = 0; | ||
621 | lua_assert(ms.matchdepth == MAXCCALLS); | ||
622 | if ((res=match(&ms, s1, p)) != NULL) { | ||
623 | if (find) { | ||
624 | lua_pushinteger(L, s1 - s + 1); /* start */ | ||
625 | lua_pushinteger(L, res - s); /* end */ | ||
626 | return push_captures(&ms, NULL, 0) + 2; | ||
627 | } | ||
628 | else | ||
629 | return push_captures(&ms, s1, res); | ||
630 | } | ||
631 | } while (s1++ < ms.src_end && !anchor); | ||
632 | } | ||
633 | lua_pushnil(L); /* not found */ | ||
634 | return 1; | ||
635 | } | ||
636 | |||
637 | |||
638 | static int str_find (lua_State *L) { | ||
639 | return str_find_aux(L, 1); | ||
640 | } | ||
641 | |||
642 | |||
643 | static int str_match (lua_State *L) { | ||
644 | return str_find_aux(L, 0); | ||
645 | } | ||
646 | |||
647 | |||
648 | static int gmatch_aux (lua_State *L) { | ||
649 | MatchState ms; | ||
650 | size_t ls, lp; | ||
651 | const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls); | ||
652 | const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp); | ||
653 | const char *src; | ||
654 | ms.L = L; | ||
655 | ms.matchdepth = MAXCCALLS; | ||
656 | ms.src_init = s; | ||
657 | ms.src_end = s+ls; | ||
658 | ms.p_end = p + lp; | ||
659 | for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3)); | ||
660 | src <= ms.src_end; | ||
661 | src++) { | ||
662 | const char *e; | ||
663 | ms.level = 0; | ||
664 | lua_assert(ms.matchdepth == MAXCCALLS); | ||
665 | if ((e = match(&ms, src, p)) != NULL) { | ||
666 | lua_Integer newstart = e-s; | ||
667 | if (e == src) newstart++; /* empty match? go at least one position */ | ||
668 | lua_pushinteger(L, newstart); | ||
669 | lua_replace(L, lua_upvalueindex(3)); | ||
670 | return push_captures(&ms, src, e); | ||
671 | } | ||
672 | } | ||
673 | return 0; /* not found */ | ||
674 | } | ||
675 | |||
676 | |||
677 | static int gmatch (lua_State *L) { | ||
678 | luaL_checkstring(L, 1); | ||
679 | luaL_checkstring(L, 2); | ||
680 | lua_settop(L, 2); | ||
681 | lua_pushinteger(L, 0); | ||
682 | lua_pushcclosure(L, gmatch_aux, 3); | ||
683 | return 1; | ||
684 | } | ||
685 | |||
686 | |||
687 | static void add_s (MatchState *ms, luaL_Buffer *b, const char *s, | ||
688 | const char *e) { | ||
689 | size_t l, i; | ||
690 | lua_State *L = ms->L; | ||
691 | const char *news = lua_tolstring(L, 3, &l); | ||
692 | for (i = 0; i < l; i++) { | ||
693 | if (news[i] != L_ESC) | ||
694 | luaL_addchar(b, news[i]); | ||
695 | else { | ||
696 | i++; /* skip ESC */ | ||
697 | if (!isdigit(uchar(news[i]))) { | ||
698 | if (news[i] != L_ESC) | ||
699 | luaL_error(L, "invalid use of '%c' in replacement string", L_ESC); | ||
700 | luaL_addchar(b, news[i]); | ||
701 | } | ||
702 | else if (news[i] == '0') | ||
703 | luaL_addlstring(b, s, e - s); | ||
704 | else { | ||
705 | push_onecapture(ms, news[i] - '1', s, e); | ||
706 | luaL_tolstring(L, -1, NULL); /* if number, convert it to string */ | ||
707 | lua_remove(L, -2); /* remove original value */ | ||
708 | luaL_addvalue(b); /* add capture to accumulated result */ | ||
709 | } | ||
710 | } | ||
711 | } | ||
712 | } | ||
713 | |||
714 | |||
715 | static void add_value (MatchState *ms, luaL_Buffer *b, const char *s, | ||
716 | const char *e, int tr) { | ||
717 | lua_State *L = ms->L; | ||
718 | switch (tr) { | ||
719 | case LUA_TFUNCTION: { | ||
720 | int n; | ||
721 | lua_pushvalue(L, 3); | ||
722 | n = push_captures(ms, s, e); | ||
723 | lua_call(L, n, 1); | ||
724 | break; | ||
725 | } | ||
726 | case LUA_TTABLE: { | ||
727 | push_onecapture(ms, 0, s, e); | ||
728 | lua_gettable(L, 3); | ||
729 | break; | ||
730 | } | ||
731 | default: { /* LUA_TNUMBER or LUA_TSTRING */ | ||
732 | add_s(ms, b, s, e); | ||
733 | return; | ||
734 | } | ||
735 | } | ||
736 | if (!lua_toboolean(L, -1)) { /* nil or false? */ | ||
737 | lua_pop(L, 1); | ||
738 | lua_pushlstring(L, s, e - s); /* keep original text */ | ||
739 | } | ||
740 | else if (!lua_isstring(L, -1)) | ||
741 | luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1)); | ||
742 | luaL_addvalue(b); /* add result to accumulator */ | ||
743 | } | ||
744 | |||
745 | |||
746 | static int str_gsub (lua_State *L) { | ||
747 | size_t srcl, lp; | ||
748 | const char *src = luaL_checklstring(L, 1, &srcl); | ||
749 | const char *p = luaL_checklstring(L, 2, &lp); | ||
750 | int tr = lua_type(L, 3); | ||
751 | lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1); | ||
752 | int anchor = (*p == '^'); | ||
753 | lua_Integer n = 0; | ||
754 | MatchState ms; | ||
755 | luaL_Buffer b; | ||
756 | luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING || | ||
757 | tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3, | ||
758 | "string/function/table expected"); | ||
759 | luaL_buffinit(L, &b); | ||
760 | if (anchor) { | ||
761 | p++; lp--; /* skip anchor character */ | ||
762 | } | ||
763 | ms.L = L; | ||
764 | ms.matchdepth = MAXCCALLS; | ||
765 | ms.src_init = src; | ||
766 | ms.src_end = src+srcl; | ||
767 | ms.p_end = p + lp; | ||
768 | while (n < max_s) { | ||
769 | const char *e; | ||
770 | ms.level = 0; | ||
771 | lua_assert(ms.matchdepth == MAXCCALLS); | ||
772 | e = match(&ms, src, p); | ||
773 | if (e) { | ||
774 | n++; | ||
775 | add_value(&ms, &b, src, e, tr); | ||
776 | } | ||
777 | if (e && e>src) /* non empty match? */ | ||
778 | src = e; /* skip it */ | ||
779 | else if (src < ms.src_end) | ||
780 | luaL_addchar(&b, *src++); | ||
781 | else break; | ||
782 | if (anchor) break; | ||
783 | } | ||
784 | luaL_addlstring(&b, src, ms.src_end-src); | ||
785 | luaL_pushresult(&b); | ||
786 | lua_pushinteger(L, n); /* number of substitutions */ | ||
787 | return 2; | ||
788 | } | ||
789 | |||
790 | /* }====================================================== */ | ||
791 | |||
792 | |||
793 | |||
794 | /* | ||
795 | ** {====================================================== | ||
796 | ** STRING FORMAT | ||
797 | ** ======================================================= | ||
798 | */ | ||
799 | |||
800 | /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */ | ||
801 | #define MAX_ITEM 512 | ||
802 | |||
803 | /* valid flags in a format specification */ | ||
804 | #define FLAGS "-+ #0" | ||
805 | |||
806 | /* | ||
807 | ** maximum size of each format specification (such as "%-099.99d") | ||
808 | ** (+2 for length modifiers; +10 accounts for %99.99x plus margin of error) | ||
809 | */ | ||
810 | #define MAX_FORMAT (sizeof(FLAGS) + 2 + 10) | ||
811 | |||
812 | |||
813 | static void addquoted (lua_State *L, luaL_Buffer *b, int arg) { | ||
814 | size_t l; | ||
815 | const char *s = luaL_checklstring(L, arg, &l); | ||
816 | luaL_addchar(b, '"'); | ||
817 | while (l--) { | ||
818 | if (*s == '"' || *s == '\\' || *s == '\n') { | ||
819 | luaL_addchar(b, '\\'); | ||
820 | luaL_addchar(b, *s); | ||
821 | } | ||
822 | else if (*s == '\0' || iscntrl(uchar(*s))) { | ||
823 | char buff[10]; | ||
824 | if (!isdigit(uchar(*(s+1)))) | ||
825 | sprintf(buff, "\\%d", (int)uchar(*s)); | ||
826 | else | ||
827 | sprintf(buff, "\\%03d", (int)uchar(*s)); | ||
828 | luaL_addstring(b, buff); | ||
829 | } | ||
830 | else | ||
831 | luaL_addchar(b, *s); | ||
832 | s++; | ||
833 | } | ||
834 | luaL_addchar(b, '"'); | ||
835 | } | ||
836 | |||
837 | static const char *scanformat (lua_State *L, const char *strfrmt, char *form) { | ||
838 | const char *p = strfrmt; | ||
839 | while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */ | ||
840 | if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char)) | ||
841 | luaL_error(L, "invalid format (repeated flags)"); | ||
842 | if (isdigit(uchar(*p))) p++; /* skip width */ | ||
843 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ | ||
844 | if (*p == '.') { | ||
845 | p++; | ||
846 | if (isdigit(uchar(*p))) p++; /* skip precision */ | ||
847 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ | ||
848 | } | ||
849 | if (isdigit(uchar(*p))) | ||
850 | luaL_error(L, "invalid format (width or precision too long)"); | ||
851 | *(form++) = '%'; | ||
852 | memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char)); | ||
853 | form += p - strfrmt + 1; | ||
854 | *form = '\0'; | ||
855 | return p; | ||
856 | } | ||
857 | |||
858 | |||
859 | /* | ||
860 | ** add length modifier into formats | ||
861 | */ | ||
862 | static void addlenmod (char *form, const char *lenmod) { | ||
863 | size_t l = strlen(form); | ||
864 | size_t lm = strlen(lenmod); | ||
865 | char spec = form[l - 1]; | ||
866 | strcpy(form + l - 1, lenmod); | ||
867 | form[l + lm - 1] = spec; | ||
868 | form[l + lm] = '\0'; | ||
869 | } | ||
870 | |||
871 | |||
872 | static int str_format (lua_State *L) { | ||
873 | int top = lua_gettop(L); | ||
874 | int arg = 1; | ||
875 | size_t sfl; | ||
876 | const char *strfrmt = luaL_checklstring(L, arg, &sfl); | ||
877 | const char *strfrmt_end = strfrmt+sfl; | ||
878 | luaL_Buffer b; | ||
879 | luaL_buffinit(L, &b); | ||
880 | while (strfrmt < strfrmt_end) { | ||
881 | if (*strfrmt != L_ESC) | ||
882 | luaL_addchar(&b, *strfrmt++); | ||
883 | else if (*++strfrmt == L_ESC) | ||
884 | luaL_addchar(&b, *strfrmt++); /* %% */ | ||
885 | else { /* format item */ | ||
886 | char form[MAX_FORMAT]; /* to store the format ('%...') */ | ||
887 | char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */ | ||
888 | int nb = 0; /* number of bytes in added item */ | ||
889 | if (++arg > top) | ||
890 | luaL_argerror(L, arg, "no value"); | ||
891 | strfrmt = scanformat(L, strfrmt, form); | ||
892 | switch (*strfrmt++) { | ||
893 | case 'c': { | ||
894 | nb = sprintf(buff, form, (int)luaL_checkinteger(L, arg)); | ||
895 | break; | ||
896 | } | ||
897 | case 'd': case 'i': | ||
898 | case 'o': case 'u': case 'x': case 'X': { | ||
899 | lua_Integer n = luaL_checkinteger(L, arg); | ||
900 | addlenmod(form, LUA_INTEGER_FRMLEN); | ||
901 | nb = sprintf(buff, form, n); | ||
902 | break; | ||
903 | } | ||
904 | #if defined(LUA_USE_AFORMAT) | ||
905 | case 'a': case 'A': | ||
906 | #endif | ||
907 | case 'e': case 'E': case 'f': | ||
908 | case 'g': case 'G': { | ||
909 | addlenmod(form, LUA_NUMBER_FRMLEN); | ||
910 | nb = sprintf(buff, form, luaL_checknumber(L, arg)); | ||
911 | break; | ||
912 | } | ||
913 | case 'q': { | ||
914 | addquoted(L, &b, arg); | ||
915 | break; | ||
916 | } | ||
917 | case 's': { | ||
918 | size_t l; | ||
919 | const char *s = luaL_tolstring(L, arg, &l); | ||
920 | if (!strchr(form, '.') && l >= 100) { | ||
921 | /* no precision and string is too long to be formatted; | ||
922 | keep original string */ | ||
923 | luaL_addvalue(&b); | ||
924 | break; | ||
925 | } | ||
926 | else { | ||
927 | nb = sprintf(buff, form, s); | ||
928 | lua_pop(L, 1); /* remove result from 'luaL_tolstring' */ | ||
929 | break; | ||
930 | } | ||
931 | } | ||
932 | default: { /* also treat cases 'pnLlh' */ | ||
933 | return luaL_error(L, "invalid option '%%%c' to 'format'", | ||
934 | *(strfrmt - 1)); | ||
935 | } | ||
936 | } | ||
937 | luaL_addsize(&b, nb); | ||
938 | } | ||
939 | } | ||
940 | luaL_pushresult(&b); | ||
941 | return 1; | ||
942 | } | ||
943 | |||
944 | /* }====================================================== */ | ||
945 | |||
946 | |||
947 | /* | ||
948 | ** {====================================================== | ||
949 | ** PACK/UNPACK | ||
950 | ** ======================================================= | ||
951 | */ | ||
952 | |||
953 | |||
954 | /* value used for padding */ | ||
955 | #if !defined(LUA_PACKPADBYTE) | ||
956 | #define LUA_PACKPADBYTE 0x00 | ||
957 | #endif | ||
958 | |||
959 | /* maximum size for the binary representation of an integer */ | ||
960 | #define MAXINTSIZE 16 | ||
961 | |||
962 | /* number of bits in a character */ | ||
963 | #define NB CHAR_BIT | ||
964 | |||
965 | /* mask for one character (NB 1's) */ | ||
966 | #define MC ((1 << NB) - 1) | ||
967 | |||
968 | /* size of a lua_Integer */ | ||
969 | #define SZINT ((int)sizeof(lua_Integer)) | ||
970 | |||
971 | |||
972 | /* dummy union to get native endianness */ | ||
973 | static const union { | ||
974 | int dummy; | ||
975 | char little; /* true iff machine is little endian */ | ||
976 | } nativeendian = {1}; | ||
977 | |||
978 | |||
979 | /* dummy structure to get native alignment requirements */ | ||
980 | struct cD { | ||
981 | char c; | ||
982 | union { double d; void *p; lua_Integer i; lua_Number n; } u; | ||
983 | }; | ||
984 | |||
985 | #define MAXALIGN (offsetof(struct cD, u)) | ||
986 | |||
987 | |||
988 | /* | ||
989 | ** Union for serializing floats | ||
990 | */ | ||
991 | typedef union Ftypes { | ||
992 | float f; | ||
993 | double d; | ||
994 | lua_Number n; | ||
995 | char buff[5 * sizeof(lua_Number)]; /* enough for any float type */ | ||
996 | } Ftypes; | ||
997 | |||
998 | |||
999 | /* | ||
1000 | ** information to pack/unpack stuff | ||
1001 | */ | ||
1002 | typedef struct Header { | ||
1003 | lua_State *L; | ||
1004 | int islittle; | ||
1005 | int maxalign; | ||
1006 | } Header; | ||
1007 | |||
1008 | |||
1009 | /* | ||
1010 | ** options for pack/unpack | ||
1011 | */ | ||
1012 | typedef enum KOption { | ||
1013 | Kint, /* signed integers */ | ||
1014 | Kuint, /* unsigned integers */ | ||
1015 | Kfloat, /* floating-point numbers */ | ||
1016 | Kchar, /* fixed-length strings */ | ||
1017 | Kstring, /* strings with prefixed length */ | ||
1018 | Kzstr, /* zero-terminated strings */ | ||
1019 | Kpadding, /* padding */ | ||
1020 | Kpaddalign, /* padding for alignment */ | ||
1021 | Knop /* no-op (configuration or spaces) */ | ||
1022 | } KOption; | ||
1023 | |||
1024 | |||
1025 | /* | ||
1026 | ** Read an integer numeral from string 'fmt' or return 'df' if | ||
1027 | ** there is no numeral | ||
1028 | */ | ||
1029 | static int digit (int c) { return '0' <= c && c <= '9'; } | ||
1030 | |||
1031 | static int getnum (const char **fmt, int df) { | ||
1032 | if (!digit(**fmt)) /* no number? */ | ||
1033 | return df; /* return default value */ | ||
1034 | else { | ||
1035 | int a = 0; | ||
1036 | do { | ||
1037 | a = a*10 + (*((*fmt)++) - '0'); | ||
1038 | } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10); | ||
1039 | return a; | ||
1040 | } | ||
1041 | } | ||
1042 | |||
1043 | |||
1044 | /* | ||
1045 | ** Read an integer numeral and raises an error if it is larger | ||
1046 | ** than the maximum size for integers. | ||
1047 | */ | ||
1048 | static int getnumlimit (Header *h, const char **fmt, int df) { | ||
1049 | int sz = getnum(fmt, df); | ||
1050 | if (sz > MAXINTSIZE || sz <= 0) | ||
1051 | luaL_error(h->L, "integral size (%d) out of limits [1,%d]", | ||
1052 | sz, MAXINTSIZE); | ||
1053 | return sz; | ||
1054 | } | ||
1055 | |||
1056 | |||
1057 | /* | ||
1058 | ** Initialize Header | ||
1059 | */ | ||
1060 | static void initheader (lua_State *L, Header *h) { | ||
1061 | h->L = L; | ||
1062 | h->islittle = nativeendian.little; | ||
1063 | h->maxalign = 1; | ||
1064 | } | ||
1065 | |||
1066 | |||
1067 | /* | ||
1068 | ** Read and classify next option. 'size' is filled with option's size. | ||
1069 | */ | ||
1070 | static KOption getoption (Header *h, const char **fmt, int *size) { | ||
1071 | int opt = *((*fmt)++); | ||
1072 | *size = 0; /* default */ | ||
1073 | switch (opt) { | ||
1074 | case 'b': *size = sizeof(char); return Kint; | ||
1075 | case 'B': *size = sizeof(char); return Kuint; | ||
1076 | case 'h': *size = sizeof(short); return Kint; | ||
1077 | case 'H': *size = sizeof(short); return Kuint; | ||
1078 | case 'l': *size = sizeof(long); return Kint; | ||
1079 | case 'L': *size = sizeof(long); return Kuint; | ||
1080 | case 'j': *size = sizeof(lua_Integer); return Kint; | ||
1081 | case 'J': *size = sizeof(lua_Integer); return Kuint; | ||
1082 | case 'T': *size = sizeof(size_t); return Kuint; | ||
1083 | case 'f': *size = sizeof(float); return Kfloat; | ||
1084 | case 'd': *size = sizeof(double); return Kfloat; | ||
1085 | case 'n': *size = sizeof(lua_Number); return Kfloat; | ||
1086 | case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint; | ||
1087 | case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint; | ||
1088 | case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring; | ||
1089 | case 'c': | ||
1090 | *size = getnum(fmt, -1); | ||
1091 | if (*size == -1) | ||
1092 | luaL_error(h->L, "missing size for format option 'c'"); | ||
1093 | return Kchar; | ||
1094 | case 'z': return Kzstr; | ||
1095 | case 'x': *size = 1; return Kpadding; | ||
1096 | case 'X': return Kpaddalign; | ||
1097 | case ' ': break; | ||
1098 | case '<': h->islittle = 1; break; | ||
1099 | case '>': h->islittle = 0; break; | ||
1100 | case '=': h->islittle = nativeendian.little; break; | ||
1101 | case '!': h->maxalign = getnumlimit(h, fmt, MAXALIGN); break; | ||
1102 | default: luaL_error(h->L, "invalid format option '%c'", opt); | ||
1103 | } | ||
1104 | return Knop; | ||
1105 | } | ||
1106 | |||
1107 | |||
1108 | /* | ||
1109 | ** Read, classify, and fill other details about the next option. | ||
1110 | ** 'psize' is filled with option's size, 'notoalign' with its | ||
1111 | ** alignment requirements. | ||
1112 | ** Local variable 'size' gets the size to be aligned. (Kpadal option | ||
1113 | ** always gets its full alignment, other options are limited by | ||
1114 | ** the maximum alignment ('maxalign'). Kchar option needs no alignment | ||
1115 | ** despite its size. | ||
1116 | */ | ||
1117 | static KOption getdetails (Header *h, size_t totalsize, | ||
1118 | const char **fmt, int *psize, int *ntoalign) { | ||
1119 | KOption opt = getoption(h, fmt, psize); | ||
1120 | int align = *psize; /* usually, alignment follows size */ | ||
1121 | if (opt == Kpaddalign) { /* 'X' gets alignment from following option */ | ||
1122 | if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0) | ||
1123 | luaL_argerror(h->L, 1, "invalid next option for option 'X'"); | ||
1124 | } | ||
1125 | if (align <= 1 || opt == Kchar) /* need no alignment? */ | ||
1126 | *ntoalign = 0; | ||
1127 | else { | ||
1128 | if (align > h->maxalign) /* enforce maximum alignment */ | ||
1129 | align = h->maxalign; | ||
1130 | if ((align & (align - 1)) != 0) /* is 'align' not a power of 2? */ | ||
1131 | luaL_argerror(h->L, 1, "format asks for alignment not power of 2"); | ||
1132 | *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1); | ||
1133 | } | ||
1134 | return opt; | ||
1135 | } | ||
1136 | |||
1137 | |||
1138 | /* | ||
1139 | ** Pack integer 'n' with 'size' bytes and 'islittle' endianness. | ||
1140 | ** The final 'if' handles the case when 'size' is larger than | ||
1141 | ** the size of a Lua integer, correcting the extra sign-extension | ||
1142 | ** bytes if necessary (by default they would be zeros). | ||
1143 | */ | ||
1144 | static void packint (luaL_Buffer *b, lua_Unsigned n, | ||
1145 | int islittle, int size, int neg) { | ||
1146 | char *buff = luaL_prepbuffsize(b, size); | ||
1147 | int i; | ||
1148 | buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */ | ||
1149 | for (i = 1; i < size; i++) { | ||
1150 | n >>= NB; | ||
1151 | buff[islittle ? i : size - 1 - i] = (char)(n & MC); | ||
1152 | } | ||
1153 | if (neg && size > SZINT) { /* negative number need sign extension? */ | ||
1154 | for (i = SZINT; i < size; i++) /* correct extra bytes */ | ||
1155 | buff[islittle ? i : size - 1 - i] = (char)MC; | ||
1156 | } | ||
1157 | luaL_addsize(b, size); /* add result to buffer */ | ||
1158 | } | ||
1159 | |||
1160 | |||
1161 | /* | ||
1162 | ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if | ||
1163 | ** given 'islittle' is different from native endianness. | ||
1164 | */ | ||
1165 | static void copywithendian (volatile char *dest, volatile const char *src, | ||
1166 | int size, int islittle) { | ||
1167 | if (islittle == nativeendian.little) { | ||
1168 | while (size-- != 0) | ||
1169 | *(dest++) = *(src++); | ||
1170 | } | ||
1171 | else { | ||
1172 | dest += size - 1; | ||
1173 | while (size-- != 0) | ||
1174 | *(dest--) = *(src++); | ||
1175 | } | ||
1176 | } | ||
1177 | |||
1178 | |||
1179 | static int str_pack (lua_State *L) { | ||
1180 | luaL_Buffer b; | ||
1181 | Header h; | ||
1182 | const char *fmt = luaL_checkstring(L, 1); /* format string */ | ||
1183 | int arg = 1; /* current argument to pack */ | ||
1184 | size_t totalsize = 0; /* accumulate total size of result */ | ||
1185 | initheader(L, &h); | ||
1186 | lua_pushnil(L); /* mark to separate arguments from string buffer */ | ||
1187 | luaL_buffinit(L, &b); | ||
1188 | while (*fmt != '\0') { | ||
1189 | int size, ntoalign; | ||
1190 | KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); | ||
1191 | totalsize += ntoalign + size; | ||
1192 | while (ntoalign-- > 0) | ||
1193 | luaL_addchar(&b, LUA_PACKPADBYTE); /* fill alignment */ | ||
1194 | arg++; | ||
1195 | switch (opt) { | ||
1196 | case Kint: { /* signed integers */ | ||
1197 | lua_Integer n = luaL_checkinteger(L, arg); | ||
1198 | if (size < SZINT) { /* need overflow check? */ | ||
1199 | lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1); | ||
1200 | luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow"); | ||
1201 | } | ||
1202 | packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0)); | ||
1203 | break; | ||
1204 | } | ||
1205 | case Kuint: { /* unsigned integers */ | ||
1206 | lua_Integer n = luaL_checkinteger(L, arg); | ||
1207 | if (size < SZINT) /* need overflow check? */ | ||
1208 | luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)), | ||
1209 | arg, "unsigned overflow"); | ||
1210 | packint(&b, (lua_Unsigned)n, h.islittle, size, 0); | ||
1211 | break; | ||
1212 | } | ||
1213 | case Kfloat: { /* floating-point options */ | ||
1214 | volatile Ftypes u; | ||
1215 | char *buff = luaL_prepbuffsize(&b, size); | ||
1216 | lua_Number n = luaL_checknumber(L, arg); /* get argument */ | ||
1217 | if (size == sizeof(u.f)) u.f = (float)n; /* copy it into 'u' */ | ||
1218 | else if (size == sizeof(u.d)) u.d = (double)n; | ||
1219 | else u.n = n; | ||
1220 | /* move 'u' to final result, correcting endianness if needed */ | ||
1221 | copywithendian(buff, u.buff, size, h.islittle); | ||
1222 | luaL_addsize(&b, size); | ||
1223 | break; | ||
1224 | } | ||
1225 | case Kchar: { /* fixed-size string */ | ||
1226 | size_t len; | ||
1227 | const char *s = luaL_checklstring(L, arg, &len); | ||
1228 | luaL_argcheck(L, len == (size_t)size, arg, "wrong length"); | ||
1229 | luaL_addlstring(&b, s, size); | ||
1230 | break; | ||
1231 | } | ||
1232 | case Kstring: { /* strings with length count */ | ||
1233 | size_t len; | ||
1234 | const char *s = luaL_checklstring(L, arg, &len); | ||
1235 | luaL_argcheck(L, size >= (int)sizeof(size_t) || | ||
1236 | len < ((size_t)1 << (size * NB)), | ||
1237 | arg, "string length does not fit in given size"); | ||
1238 | packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */ | ||
1239 | luaL_addlstring(&b, s, len); | ||
1240 | totalsize += len; | ||
1241 | break; | ||
1242 | } | ||
1243 | case Kzstr: { /* zero-terminated string */ | ||
1244 | size_t len; | ||
1245 | const char *s = luaL_checklstring(L, arg, &len); | ||
1246 | luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros"); | ||
1247 | luaL_addlstring(&b, s, len); | ||
1248 | luaL_addchar(&b, '\0'); /* add zero at the end */ | ||
1249 | totalsize += len + 1; | ||
1250 | break; | ||
1251 | } | ||
1252 | case Kpadding: luaL_addchar(&b, LUA_PACKPADBYTE); /* go through */ | ||
1253 | case Kpaddalign: case Knop: | ||
1254 | arg--; /* undo increment */ | ||
1255 | break; | ||
1256 | } | ||
1257 | } | ||
1258 | luaL_pushresult(&b); | ||
1259 | return 1; | ||
1260 | } | ||
1261 | |||
1262 | |||
1263 | static int str_packsize (lua_State *L) { | ||
1264 | Header h; | ||
1265 | const char *fmt = luaL_checkstring(L, 1); /* format string */ | ||
1266 | size_t totalsize = 0; /* accumulate total size of result */ | ||
1267 | initheader(L, &h); | ||
1268 | while (*fmt != '\0') { | ||
1269 | int size, ntoalign; | ||
1270 | KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); | ||
1271 | size += ntoalign; /* total space used by option */ | ||
1272 | luaL_argcheck(L, totalsize <= MAXSIZE - size, 1, | ||
1273 | "format result too large"); | ||
1274 | totalsize += size; | ||
1275 | switch (opt) { | ||
1276 | case Kstring: /* strings with length count */ | ||
1277 | case Kzstr: /* zero-terminated string */ | ||
1278 | luaL_argerror(L, 1, "variable-length format"); | ||
1279 | break; | ||
1280 | default: break; | ||
1281 | } | ||
1282 | } | ||
1283 | lua_pushinteger(L, (lua_Integer)totalsize); | ||
1284 | return 1; | ||
1285 | } | ||
1286 | |||
1287 | |||
1288 | /* | ||
1289 | ** Unpack an integer with 'size' bytes and 'islittle' endianness. | ||
1290 | ** If size is smaller than the size of a Lua integer and integer | ||
1291 | ** is signed, must do sign extension (propagating the sign to the | ||
1292 | ** higher bits); if size is larger than the size of a Lua integer, | ||
1293 | ** it must check the unread bytes to see whether they do not cause an | ||
1294 | ** overflow. | ||
1295 | */ | ||
1296 | static lua_Integer unpackint (lua_State *L, const char *str, | ||
1297 | int islittle, int size, int issigned) { | ||
1298 | lua_Unsigned res = 0; | ||
1299 | int i; | ||
1300 | int limit = (size <= SZINT) ? size : SZINT; | ||
1301 | for (i = limit - 1; i >= 0; i--) { | ||
1302 | res <<= NB; | ||
1303 | res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i]; | ||
1304 | } | ||
1305 | if (size < SZINT) { /* real size smaller than lua_Integer? */ | ||
1306 | if (issigned) { /* needs sign extension? */ | ||
1307 | lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1); | ||
1308 | res = ((res ^ mask) - mask); /* do sign extension */ | ||
1309 | } | ||
1310 | } | ||
1311 | else if (size > SZINT) { /* must check unread bytes */ | ||
1312 | int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC; | ||
1313 | for (i = limit; i < size; i++) { | ||
1314 | if ((unsigned char)str[islittle ? i : size - 1 - i] != mask) | ||
1315 | luaL_error(L, "%d-byte integer does not fit into Lua Integer", size); | ||
1316 | } | ||
1317 | } | ||
1318 | return (lua_Integer)res; | ||
1319 | } | ||
1320 | |||
1321 | |||
1322 | static int str_unpack (lua_State *L) { | ||
1323 | Header h; | ||
1324 | const char *fmt = luaL_checkstring(L, 1); | ||
1325 | size_t ld; | ||
1326 | const char *data = luaL_checklstring(L, 2, &ld); | ||
1327 | size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1; | ||
1328 | int n = 0; /* number of results */ | ||
1329 | luaL_argcheck(L, pos <= ld, 3, "initial position out of string"); | ||
1330 | initheader(L, &h); | ||
1331 | while (*fmt != '\0') { | ||
1332 | int size, ntoalign; | ||
1333 | KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign); | ||
1334 | if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld) | ||
1335 | luaL_argerror(L, 2, "data string too short"); | ||
1336 | pos += ntoalign; /* skip alignment */ | ||
1337 | /* stack space for item + next position */ | ||
1338 | luaL_checkstack(L, 2, "too many results"); | ||
1339 | n++; | ||
1340 | switch (opt) { | ||
1341 | case Kint: | ||
1342 | case Kuint: { | ||
1343 | lua_Integer res = unpackint(L, data + pos, h.islittle, size, | ||
1344 | (opt == Kint)); | ||
1345 | lua_pushinteger(L, res); | ||
1346 | break; | ||
1347 | } | ||
1348 | case Kfloat: { | ||
1349 | volatile Ftypes u; | ||
1350 | lua_Number num; | ||
1351 | copywithendian(u.buff, data + pos, size, h.islittle); | ||
1352 | if (size == sizeof(u.f)) num = (lua_Number)u.f; | ||
1353 | else if (size == sizeof(u.d)) num = (lua_Number)u.d; | ||
1354 | else num = u.n; | ||
1355 | lua_pushnumber(L, num); | ||
1356 | break; | ||
1357 | } | ||
1358 | case Kchar: { | ||
1359 | lua_pushlstring(L, data + pos, size); | ||
1360 | break; | ||
1361 | } | ||
1362 | case Kstring: { | ||
1363 | size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0); | ||
1364 | luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short"); | ||
1365 | lua_pushlstring(L, data + pos + size, len); | ||
1366 | pos += len; /* skip string */ | ||
1367 | break; | ||
1368 | } | ||
1369 | case Kzstr: { | ||
1370 | size_t len = (int)strlen(data + pos); | ||
1371 | lua_pushlstring(L, data + pos, len); | ||
1372 | pos += len + 1; /* skip string plus final '\0' */ | ||
1373 | break; | ||
1374 | } | ||
1375 | case Kpaddalign: case Kpadding: case Knop: | ||
1376 | n--; /* undo increment */ | ||
1377 | break; | ||
1378 | } | ||
1379 | pos += size; | ||
1380 | } | ||
1381 | lua_pushinteger(L, pos + 1); /* next position */ | ||
1382 | return n + 1; | ||
1383 | } | ||
1384 | |||
1385 | /* }====================================================== */ | ||
1386 | |||
1387 | |||
1388 | static const luaL_Reg strlib[] = { | ||
1389 | {"byte", str_byte}, | ||
1390 | {"char", str_char}, | ||
1391 | {"dump", str_dump}, | ||
1392 | {"find", str_find}, | ||
1393 | {"format", str_format}, | ||
1394 | {"gmatch", gmatch}, | ||
1395 | {"gsub", str_gsub}, | ||
1396 | {"len", str_len}, | ||
1397 | {"lower", str_lower}, | ||
1398 | {"match", str_match}, | ||
1399 | {"rep", str_rep}, | ||
1400 | {"reverse", str_reverse}, | ||
1401 | {"sub", str_sub}, | ||
1402 | {"upper", str_upper}, | ||
1403 | {"pack", str_pack}, | ||
1404 | {"packsize", str_packsize}, | ||
1405 | {"unpack", str_unpack}, | ||
1406 | {NULL, NULL} | ||
1407 | }; | ||
1408 | |||
1409 | |||
1410 | static void createmetatable (lua_State *L) { | ||
1411 | lua_createtable(L, 0, 1); /* table to be metatable for strings */ | ||
1412 | lua_pushliteral(L, ""); /* dummy string */ | ||
1413 | lua_pushvalue(L, -2); /* copy table */ | ||
1414 | lua_setmetatable(L, -2); /* set table as metatable for strings */ | ||
1415 | lua_pop(L, 1); /* pop dummy string */ | ||
1416 | lua_pushvalue(L, -2); /* get string library */ | ||
1417 | lua_setfield(L, -2, "__index"); /* metatable.__index = string */ | ||
1418 | lua_pop(L, 1); /* pop metatable */ | ||
1419 | } | ||
1420 | |||
1421 | |||
1422 | /* | ||
1423 | ** Open string library | ||
1424 | */ | ||
1425 | LUAMOD_API int luaopen_string (lua_State *L) { | ||
1426 | luaL_newlib(L, strlib); | ||
1427 | createmetatable(L); | ||
1428 | return 1; | ||
1429 | } | ||
1430 | |||
diff --git a/rockspecs/compat53-scm-0.rockspec b/rockspecs/compat53-scm-0.rockspec index ec56e51..2dd8cba 100644 --- a/rockspecs/compat53-scm-0.rockspec +++ b/rockspecs/compat53-scm-0.rockspec | |||
@@ -25,6 +25,7 @@ build = { | |||
25 | ["compat53"] = "compat53.lua", | 25 | ["compat53"] = "compat53.lua", |
26 | ["compat53.utf8"] = "lutf8lib.c", | 26 | ["compat53.utf8"] = "lutf8lib.c", |
27 | ["compat53.table"] = "ltablib.c", | 27 | ["compat53.table"] = "ltablib.c", |
28 | ["compat53.string"] = "lstrlib.c", | ||
28 | } | 29 | } |
29 | } | 30 | } |
30 | 31 | ||