Contents of /trunk/mkinitrd-magellan/busybox/editors/awk.c
Parent Directory | Revision Log
Revision 984 -
(show annotations)
(download)
Sun May 30 11:32:42 2010 UTC (14 years ago) by niro
File MIME type: text/plain
File size: 66134 byte(s)
Sun May 30 11:32:42 2010 UTC (14 years ago) by niro
File MIME type: text/plain
File size: 66134 byte(s)
-updated to busybox-1.16.1 and enabled blkid/uuid support in default config
1 | /* vi: set sw=4 ts=4: */ |
2 | /* |
3 | * awk implementation for busybox |
4 | * |
5 | * Copyright (C) 2002 by Dmitry Zakharov <dmit@crp.bank.gov.ua> |
6 | * |
7 | * Licensed under the GPL v2 or later, see the file LICENSE in this tarball. |
8 | */ |
9 | |
10 | #include "libbb.h" |
11 | #include "xregex.h" |
12 | #include <math.h> |
13 | |
14 | /* This is a NOEXEC applet. Be very careful! */ |
15 | |
16 | |
17 | #define MAXVARFMT 240 |
18 | #define MINNVBLOCK 64 |
19 | |
20 | /* variable flags */ |
21 | #define VF_NUMBER 0x0001 /* 1 = primary type is number */ |
22 | #define VF_ARRAY 0x0002 /* 1 = it's an array */ |
23 | |
24 | #define VF_CACHED 0x0100 /* 1 = num/str value has cached str/num eq */ |
25 | #define VF_USER 0x0200 /* 1 = user input (may be numeric string) */ |
26 | #define VF_SPECIAL 0x0400 /* 1 = requires extra handling when changed */ |
27 | #define VF_WALK 0x0800 /* 1 = variable has alloc'd x.walker list */ |
28 | #define VF_FSTR 0x1000 /* 1 = var::string points to fstring buffer */ |
29 | #define VF_CHILD 0x2000 /* 1 = function arg; x.parent points to source */ |
30 | #define VF_DIRTY 0x4000 /* 1 = variable was set explicitly */ |
31 | |
32 | /* these flags are static, don't change them when value is changed */ |
33 | #define VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY) |
34 | |
35 | /* Variable */ |
36 | typedef struct var_s { |
37 | unsigned type; /* flags */ |
38 | double number; |
39 | char *string; |
40 | union { |
41 | int aidx; /* func arg idx (for compilation stage) */ |
42 | struct xhash_s *array; /* array ptr */ |
43 | struct var_s *parent; /* for func args, ptr to actual parameter */ |
44 | char **walker; /* list of array elements (for..in) */ |
45 | } x; |
46 | } var; |
47 | |
48 | /* Node chain (pattern-action chain, BEGIN, END, function bodies) */ |
49 | typedef struct chain_s { |
50 | struct node_s *first; |
51 | struct node_s *last; |
52 | const char *programname; |
53 | } chain; |
54 | |
55 | /* Function */ |
56 | typedef struct func_s { |
57 | unsigned nargs; |
58 | struct chain_s body; |
59 | } func; |
60 | |
61 | /* I/O stream */ |
62 | typedef struct rstream_s { |
63 | FILE *F; |
64 | char *buffer; |
65 | int adv; |
66 | int size; |
67 | int pos; |
68 | smallint is_pipe; |
69 | } rstream; |
70 | |
71 | typedef struct hash_item_s { |
72 | union { |
73 | struct var_s v; /* variable/array hash */ |
74 | struct rstream_s rs; /* redirect streams hash */ |
75 | struct func_s f; /* functions hash */ |
76 | } data; |
77 | struct hash_item_s *next; /* next in chain */ |
78 | char name[1]; /* really it's longer */ |
79 | } hash_item; |
80 | |
81 | typedef struct xhash_s { |
82 | unsigned nel; /* num of elements */ |
83 | unsigned csize; /* current hash size */ |
84 | unsigned nprime; /* next hash size in PRIMES[] */ |
85 | unsigned glen; /* summary length of item names */ |
86 | struct hash_item_s **items; |
87 | } xhash; |
88 | |
89 | /* Tree node */ |
90 | typedef struct node_s { |
91 | uint32_t info; |
92 | unsigned lineno; |
93 | union { |
94 | struct node_s *n; |
95 | var *v; |
96 | int i; |
97 | char *s; |
98 | regex_t *re; |
99 | } l; |
100 | union { |
101 | struct node_s *n; |
102 | regex_t *ire; |
103 | func *f; |
104 | int argno; |
105 | } r; |
106 | union { |
107 | struct node_s *n; |
108 | } a; |
109 | } node; |
110 | |
111 | /* Block of temporary variables */ |
112 | typedef struct nvblock_s { |
113 | int size; |
114 | var *pos; |
115 | struct nvblock_s *prev; |
116 | struct nvblock_s *next; |
117 | var nv[]; |
118 | } nvblock; |
119 | |
120 | typedef struct tsplitter_s { |
121 | node n; |
122 | regex_t re[2]; |
123 | } tsplitter; |
124 | |
125 | /* simple token classes */ |
126 | /* Order and hex values are very important!!! See next_token() */ |
127 | #define TC_SEQSTART 1 /* ( */ |
128 | #define TC_SEQTERM (1 << 1) /* ) */ |
129 | #define TC_REGEXP (1 << 2) /* /.../ */ |
130 | #define TC_OUTRDR (1 << 3) /* | > >> */ |
131 | #define TC_UOPPOST (1 << 4) /* unary postfix operator */ |
132 | #define TC_UOPPRE1 (1 << 5) /* unary prefix operator */ |
133 | #define TC_BINOPX (1 << 6) /* two-opnd operator */ |
134 | #define TC_IN (1 << 7) |
135 | #define TC_COMMA (1 << 8) |
136 | #define TC_PIPE (1 << 9) /* input redirection pipe */ |
137 | #define TC_UOPPRE2 (1 << 10) /* unary prefix operator */ |
138 | #define TC_ARRTERM (1 << 11) /* ] */ |
139 | #define TC_GRPSTART (1 << 12) /* { */ |
140 | #define TC_GRPTERM (1 << 13) /* } */ |
141 | #define TC_SEMICOL (1 << 14) |
142 | #define TC_NEWLINE (1 << 15) |
143 | #define TC_STATX (1 << 16) /* ctl statement (for, next...) */ |
144 | #define TC_WHILE (1 << 17) |
145 | #define TC_ELSE (1 << 18) |
146 | #define TC_BUILTIN (1 << 19) |
147 | #define TC_GETLINE (1 << 20) |
148 | #define TC_FUNCDECL (1 << 21) /* `function' `func' */ |
149 | #define TC_BEGIN (1 << 22) |
150 | #define TC_END (1 << 23) |
151 | #define TC_EOF (1 << 24) |
152 | #define TC_VARIABLE (1 << 25) |
153 | #define TC_ARRAY (1 << 26) |
154 | #define TC_FUNCTION (1 << 27) |
155 | #define TC_STRING (1 << 28) |
156 | #define TC_NUMBER (1 << 29) |
157 | |
158 | #define TC_UOPPRE (TC_UOPPRE1 | TC_UOPPRE2) |
159 | |
160 | /* combined token classes */ |
161 | #define TC_BINOP (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN) |
162 | #define TC_UNARYOP (TC_UOPPRE | TC_UOPPOST) |
163 | #define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION \ |
164 | | TC_BUILTIN | TC_GETLINE | TC_SEQSTART | TC_STRING | TC_NUMBER) |
165 | |
166 | #define TC_STATEMNT (TC_STATX | TC_WHILE) |
167 | #define TC_OPTERM (TC_SEMICOL | TC_NEWLINE) |
168 | |
169 | /* word tokens, cannot mean something else if not expected */ |
170 | #define TC_WORD (TC_IN | TC_STATEMNT | TC_ELSE | TC_BUILTIN \ |
171 | | TC_GETLINE | TC_FUNCDECL | TC_BEGIN | TC_END) |
172 | |
173 | /* discard newlines after these */ |
174 | #define TC_NOTERM (TC_COMMA | TC_GRPSTART | TC_GRPTERM \ |
175 | | TC_BINOP | TC_OPTERM) |
176 | |
177 | /* what can expression begin with */ |
178 | #define TC_OPSEQ (TC_OPERAND | TC_UOPPRE | TC_REGEXP) |
179 | /* what can group begin with */ |
180 | #define TC_GRPSEQ (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART) |
181 | |
182 | /* if previous token class is CONCAT1 and next is CONCAT2, concatenation */ |
183 | /* operator is inserted between them */ |
184 | #define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM \ |
185 | | TC_STRING | TC_NUMBER | TC_UOPPOST) |
186 | #define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE) |
187 | |
188 | #define OF_RES1 0x010000 |
189 | #define OF_RES2 0x020000 |
190 | #define OF_STR1 0x040000 |
191 | #define OF_STR2 0x080000 |
192 | #define OF_NUM1 0x100000 |
193 | #define OF_CHECKED 0x200000 |
194 | |
195 | /* combined operator flags */ |
196 | #define xx 0 |
197 | #define xV OF_RES2 |
198 | #define xS (OF_RES2 | OF_STR2) |
199 | #define Vx OF_RES1 |
200 | #define VV (OF_RES1 | OF_RES2) |
201 | #define Nx (OF_RES1 | OF_NUM1) |
202 | #define NV (OF_RES1 | OF_NUM1 | OF_RES2) |
203 | #define Sx (OF_RES1 | OF_STR1) |
204 | #define SV (OF_RES1 | OF_STR1 | OF_RES2) |
205 | #define SS (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2) |
206 | |
207 | #define OPCLSMASK 0xFF00 |
208 | #define OPNMASK 0x007F |
209 | |
210 | /* operator priority is a highest byte (even: r->l, odd: l->r grouping) |
211 | * For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1, |
212 | * n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string |
213 | */ |
214 | #define P(x) (x << 24) |
215 | #define PRIMASK 0x7F000000 |
216 | #define PRIMASK2 0x7E000000 |
217 | |
218 | /* Operation classes */ |
219 | |
220 | #define SHIFT_TIL_THIS 0x0600 |
221 | #define RECUR_FROM_THIS 0x1000 |
222 | |
223 | enum { |
224 | OC_DELETE = 0x0100, OC_EXEC = 0x0200, OC_NEWSOURCE = 0x0300, |
225 | OC_PRINT = 0x0400, OC_PRINTF = 0x0500, OC_WALKINIT = 0x0600, |
226 | |
227 | OC_BR = 0x0700, OC_BREAK = 0x0800, OC_CONTINUE = 0x0900, |
228 | OC_EXIT = 0x0a00, OC_NEXT = 0x0b00, OC_NEXTFILE = 0x0c00, |
229 | OC_TEST = 0x0d00, OC_WALKNEXT = 0x0e00, |
230 | |
231 | OC_BINARY = 0x1000, OC_BUILTIN = 0x1100, OC_COLON = 0x1200, |
232 | OC_COMMA = 0x1300, OC_COMPARE = 0x1400, OC_CONCAT = 0x1500, |
233 | OC_FBLTIN = 0x1600, OC_FIELD = 0x1700, OC_FNARG = 0x1800, |
234 | OC_FUNC = 0x1900, OC_GETLINE = 0x1a00, OC_IN = 0x1b00, |
235 | OC_LAND = 0x1c00, OC_LOR = 0x1d00, OC_MATCH = 0x1e00, |
236 | OC_MOVE = 0x1f00, OC_PGETLINE = 0x2000, OC_REGEXP = 0x2100, |
237 | OC_REPLACE = 0x2200, OC_RETURN = 0x2300, OC_SPRINTF = 0x2400, |
238 | OC_TERNARY = 0x2500, OC_UNARY = 0x2600, OC_VAR = 0x2700, |
239 | OC_DONE = 0x2800, |
240 | |
241 | ST_IF = 0x3000, ST_DO = 0x3100, ST_FOR = 0x3200, |
242 | ST_WHILE = 0x3300 |
243 | }; |
244 | |
245 | /* simple builtins */ |
246 | enum { |
247 | F_in, F_rn, F_co, F_ex, F_lg, F_si, F_sq, F_sr, |
248 | F_ti, F_le, F_sy, F_ff, F_cl |
249 | }; |
250 | |
251 | /* builtins */ |
252 | enum { |
253 | B_a2, B_ix, B_ma, B_sp, B_ss, B_ti, B_mt, B_lo, B_up, |
254 | B_ge, B_gs, B_su, |
255 | B_an, B_co, B_ls, B_or, B_rs, B_xo, |
256 | }; |
257 | |
258 | /* tokens and their corresponding info values */ |
259 | |
260 | #define NTC "\377" /* switch to next token class (tc<<1) */ |
261 | #define NTCC '\377' |
262 | |
263 | #define OC_B OC_BUILTIN |
264 | |
265 | static const char tokenlist[] ALIGN1 = |
266 | "\1(" NTC |
267 | "\1)" NTC |
268 | "\1/" NTC /* REGEXP */ |
269 | "\2>>" "\1>" "\1|" NTC /* OUTRDR */ |
270 | "\2++" "\2--" NTC /* UOPPOST */ |
271 | "\2++" "\2--" "\1$" NTC /* UOPPRE1 */ |
272 | "\2==" "\1=" "\2+=" "\2-=" /* BINOPX */ |
273 | "\2*=" "\2/=" "\2%=" "\2^=" |
274 | "\1+" "\1-" "\3**=" "\2**" |
275 | "\1/" "\1%" "\1^" "\1*" |
276 | "\2!=" "\2>=" "\2<=" "\1>" |
277 | "\1<" "\2!~" "\1~" "\2&&" |
278 | "\2||" "\1?" "\1:" NTC |
279 | "\2in" NTC |
280 | "\1," NTC |
281 | "\1|" NTC |
282 | "\1+" "\1-" "\1!" NTC /* UOPPRE2 */ |
283 | "\1]" NTC |
284 | "\1{" NTC |
285 | "\1}" NTC |
286 | "\1;" NTC |
287 | "\1\n" NTC |
288 | "\2if" "\2do" "\3for" "\5break" /* STATX */ |
289 | "\10continue" "\6delete" "\5print" |
290 | "\6printf" "\4next" "\10nextfile" |
291 | "\6return" "\4exit" NTC |
292 | "\5while" NTC |
293 | "\4else" NTC |
294 | |
295 | "\3and" "\5compl" "\6lshift" "\2or" |
296 | "\6rshift" "\3xor" |
297 | "\5close" "\6system" "\6fflush" "\5atan2" /* BUILTIN */ |
298 | "\3cos" "\3exp" "\3int" "\3log" |
299 | "\4rand" "\3sin" "\4sqrt" "\5srand" |
300 | "\6gensub" "\4gsub" "\5index" "\6length" |
301 | "\5match" "\5split" "\7sprintf" "\3sub" |
302 | "\6substr" "\7systime" "\10strftime" "\6mktime" |
303 | "\7tolower" "\7toupper" NTC |
304 | "\7getline" NTC |
305 | "\4func" "\10function" NTC |
306 | "\5BEGIN" NTC |
307 | "\3END" "\0" |
308 | ; |
309 | |
310 | static const uint32_t tokeninfo[] = { |
311 | 0, |
312 | 0, |
313 | OC_REGEXP, |
314 | xS|'a', xS|'w', xS|'|', |
315 | OC_UNARY|xV|P(9)|'p', OC_UNARY|xV|P(9)|'m', |
316 | OC_UNARY|xV|P(9)|'P', OC_UNARY|xV|P(9)|'M', |
317 | OC_FIELD|xV|P(5), |
318 | OC_COMPARE|VV|P(39)|5, OC_MOVE|VV|P(74), |
319 | OC_REPLACE|NV|P(74)|'+', OC_REPLACE|NV|P(74)|'-', |
320 | OC_REPLACE|NV|P(74)|'*', OC_REPLACE|NV|P(74)|'/', |
321 | OC_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&', |
322 | OC_BINARY|NV|P(29)|'+', OC_BINARY|NV|P(29)|'-', |
323 | OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&', |
324 | OC_BINARY|NV|P(25)|'/', OC_BINARY|NV|P(25)|'%', |
325 | OC_BINARY|NV|P(15)|'&', OC_BINARY|NV|P(25)|'*', |
326 | OC_COMPARE|VV|P(39)|4, OC_COMPARE|VV|P(39)|3, |
327 | OC_COMPARE|VV|P(39)|0, OC_COMPARE|VV|P(39)|1, |
328 | OC_COMPARE|VV|P(39)|2, OC_MATCH|Sx|P(45)|'!', |
329 | OC_MATCH|Sx|P(45)|'~', OC_LAND|Vx|P(55), |
330 | OC_LOR|Vx|P(59), OC_TERNARY|Vx|P(64)|'?', |
331 | OC_COLON|xx|P(67)|':', |
332 | OC_IN|SV|P(49), |
333 | OC_COMMA|SS|P(80), |
334 | OC_PGETLINE|SV|P(37), |
335 | OC_UNARY|xV|P(19)|'+', OC_UNARY|xV|P(19)|'-', |
336 | OC_UNARY|xV|P(19)|'!', |
337 | 0, |
338 | 0, |
339 | 0, |
340 | 0, |
341 | 0, |
342 | ST_IF, ST_DO, ST_FOR, OC_BREAK, |
343 | OC_CONTINUE, OC_DELETE|Vx, OC_PRINT, |
344 | OC_PRINTF, OC_NEXT, OC_NEXTFILE, |
345 | OC_RETURN|Vx, OC_EXIT|Nx, |
346 | ST_WHILE, |
347 | 0, |
348 | |
349 | OC_B|B_an|P(0x83), OC_B|B_co|P(0x41), OC_B|B_ls|P(0x83), OC_B|B_or|P(0x83), |
350 | OC_B|B_rs|P(0x83), OC_B|B_xo|P(0x83), |
351 | OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83), |
352 | OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg, |
353 | OC_FBLTIN|F_rn, OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr, |
354 | OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), OC_FBLTIN|Sx|F_le, |
355 | OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF, OC_B|B_su|P(0xb6), |
356 | OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b), OC_B|B_mt|P(0x0b), |
357 | OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49), |
358 | OC_GETLINE|SV|P(0), |
359 | 0, 0, |
360 | 0, |
361 | 0 |
362 | }; |
363 | |
364 | /* internal variable names and their initial values */ |
365 | /* asterisk marks SPECIAL vars; $ is just no-named Field0 */ |
366 | enum { |
367 | CONVFMT, OFMT, FS, OFS, |
368 | ORS, RS, RT, FILENAME, |
369 | SUBSEP, F0, ARGIND, ARGC, |
370 | ARGV, ERRNO, FNR, NR, |
371 | NF, IGNORECASE, ENVIRON, NUM_INTERNAL_VARS |
372 | }; |
373 | |
374 | static const char vNames[] ALIGN1 = |
375 | "CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0" |
376 | "ORS\0" "RS\0*" "RT\0" "FILENAME\0" |
377 | "SUBSEP\0" "$\0*" "ARGIND\0" "ARGC\0" |
378 | "ARGV\0" "ERRNO\0" "FNR\0" "NR\0" |
379 | "NF\0*" "IGNORECASE\0*" "ENVIRON\0" "\0"; |
380 | |
381 | static const char vValues[] ALIGN1 = |
382 | "%.6g\0" "%.6g\0" " \0" " \0" |
383 | "\n\0" "\n\0" "\0" "\0" |
384 | "\034\0" "\0" "\377"; |
385 | |
386 | /* hash size may grow to these values */ |
387 | #define FIRST_PRIME 61 |
388 | static const uint16_t PRIMES[] ALIGN2 = { 251, 1021, 4093, 16381, 65521 }; |
389 | |
390 | |
391 | /* Globals. Split in two parts so that first one is addressed |
392 | * with (mostly short) negative offsets. |
393 | * NB: it's unsafe to put members of type "double" |
394 | * into globals2 (gcc may fail to align them). |
395 | */ |
396 | struct globals { |
397 | double t_double; |
398 | chain beginseq, mainseq, endseq; |
399 | chain *seq; |
400 | node *break_ptr, *continue_ptr; |
401 | rstream *iF; |
402 | xhash *vhash, *ahash, *fdhash, *fnhash; |
403 | const char *g_progname; |
404 | int g_lineno; |
405 | int nfields; |
406 | int maxfields; /* used in fsrealloc() only */ |
407 | var *Fields; |
408 | nvblock *g_cb; |
409 | char *g_pos; |
410 | char *g_buf; |
411 | smallint icase; |
412 | smallint exiting; |
413 | smallint nextrec; |
414 | smallint nextfile; |
415 | smallint is_f0_split; |
416 | }; |
417 | struct globals2 { |
418 | uint32_t t_info; /* often used */ |
419 | uint32_t t_tclass; |
420 | char *t_string; |
421 | int t_lineno; |
422 | int t_rollback; |
423 | |
424 | var *intvar[NUM_INTERNAL_VARS]; /* often used */ |
425 | |
426 | /* former statics from various functions */ |
427 | char *split_f0__fstrings; |
428 | |
429 | uint32_t next_token__save_tclass; |
430 | uint32_t next_token__save_info; |
431 | uint32_t next_token__ltclass; |
432 | smallint next_token__concat_inserted; |
433 | |
434 | smallint next_input_file__files_happen; |
435 | rstream next_input_file__rsm; |
436 | |
437 | var *evaluate__fnargs; |
438 | unsigned evaluate__seed; |
439 | regex_t evaluate__sreg; |
440 | |
441 | var ptest__v; |
442 | |
443 | tsplitter exec_builtin__tspl; |
444 | |
445 | /* biggest and least used members go last */ |
446 | tsplitter fsplitter, rsplitter; |
447 | }; |
448 | #define G1 (ptr_to_globals[-1]) |
449 | #define G (*(struct globals2 *)ptr_to_globals) |
450 | /* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */ |
451 | /*char G1size[sizeof(G1)]; - 0x74 */ |
452 | /*char Gsize[sizeof(G)]; - 0x1c4 */ |
453 | /* Trying to keep most of members accessible with short offsets: */ |
454 | /*char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */ |
455 | #define t_double (G1.t_double ) |
456 | #define beginseq (G1.beginseq ) |
457 | #define mainseq (G1.mainseq ) |
458 | #define endseq (G1.endseq ) |
459 | #define seq (G1.seq ) |
460 | #define break_ptr (G1.break_ptr ) |
461 | #define continue_ptr (G1.continue_ptr) |
462 | #define iF (G1.iF ) |
463 | #define vhash (G1.vhash ) |
464 | #define ahash (G1.ahash ) |
465 | #define fdhash (G1.fdhash ) |
466 | #define fnhash (G1.fnhash ) |
467 | #define g_progname (G1.g_progname ) |
468 | #define g_lineno (G1.g_lineno ) |
469 | #define nfields (G1.nfields ) |
470 | #define maxfields (G1.maxfields ) |
471 | #define Fields (G1.Fields ) |
472 | #define g_cb (G1.g_cb ) |
473 | #define g_pos (G1.g_pos ) |
474 | #define g_buf (G1.g_buf ) |
475 | #define icase (G1.icase ) |
476 | #define exiting (G1.exiting ) |
477 | #define nextrec (G1.nextrec ) |
478 | #define nextfile (G1.nextfile ) |
479 | #define is_f0_split (G1.is_f0_split ) |
480 | #define t_info (G.t_info ) |
481 | #define t_tclass (G.t_tclass ) |
482 | #define t_string (G.t_string ) |
483 | #define t_lineno (G.t_lineno ) |
484 | #define t_rollback (G.t_rollback ) |
485 | #define intvar (G.intvar ) |
486 | #define fsplitter (G.fsplitter ) |
487 | #define rsplitter (G.rsplitter ) |
488 | #define INIT_G() do { \ |
489 | SET_PTR_TO_GLOBALS((char*)xzalloc(sizeof(G1)+sizeof(G)) + sizeof(G1)); \ |
490 | G.next_token__ltclass = TC_OPTERM; \ |
491 | G.evaluate__seed = 1; \ |
492 | } while (0) |
493 | |
494 | |
495 | /* function prototypes */ |
496 | static void handle_special(var *); |
497 | static node *parse_expr(uint32_t); |
498 | static void chain_group(void); |
499 | static var *evaluate(node *, var *); |
500 | static rstream *next_input_file(void); |
501 | static int fmt_num(char *, int, const char *, double, int); |
502 | static int awk_exit(int) NORETURN; |
503 | |
504 | /* ---- error handling ---- */ |
505 | |
506 | static const char EMSG_INTERNAL_ERROR[] ALIGN1 = "Internal error"; |
507 | static const char EMSG_UNEXP_EOS[] ALIGN1 = "Unexpected end of string"; |
508 | static const char EMSG_UNEXP_TOKEN[] ALIGN1 = "Unexpected token"; |
509 | static const char EMSG_DIV_BY_ZERO[] ALIGN1 = "Division by zero"; |
510 | static const char EMSG_INV_FMT[] ALIGN1 = "Invalid format specifier"; |
511 | static const char EMSG_TOO_FEW_ARGS[] ALIGN1 = "Too few arguments for builtin"; |
512 | static const char EMSG_NOT_ARRAY[] ALIGN1 = "Not an array"; |
513 | static const char EMSG_POSSIBLE_ERROR[] ALIGN1 = "Possible syntax error"; |
514 | static const char EMSG_UNDEF_FUNC[] ALIGN1 = "Call to undefined function"; |
515 | #if !ENABLE_FEATURE_AWK_LIBM |
516 | static const char EMSG_NO_MATH[] ALIGN1 = "Math support is not compiled in"; |
517 | #endif |
518 | |
519 | static void zero_out_var(var *vp) |
520 | { |
521 | memset(vp, 0, sizeof(*vp)); |
522 | } |
523 | |
524 | static void syntax_error(const char *message) NORETURN; |
525 | static void syntax_error(const char *message) |
526 | { |
527 | bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message); |
528 | } |
529 | |
530 | /* ---- hash stuff ---- */ |
531 | |
532 | static unsigned hashidx(const char *name) |
533 | { |
534 | unsigned idx = 0; |
535 | |
536 | while (*name) |
537 | idx = *name++ + (idx << 6) - idx; |
538 | return idx; |
539 | } |
540 | |
541 | /* create new hash */ |
542 | static xhash *hash_init(void) |
543 | { |
544 | xhash *newhash; |
545 | |
546 | newhash = xzalloc(sizeof(*newhash)); |
547 | newhash->csize = FIRST_PRIME; |
548 | newhash->items = xzalloc(FIRST_PRIME * sizeof(newhash->items[0])); |
549 | |
550 | return newhash; |
551 | } |
552 | |
553 | /* find item in hash, return ptr to data, NULL if not found */ |
554 | static void *hash_search(xhash *hash, const char *name) |
555 | { |
556 | hash_item *hi; |
557 | |
558 | hi = hash->items[hashidx(name) % hash->csize]; |
559 | while (hi) { |
560 | if (strcmp(hi->name, name) == 0) |
561 | return &(hi->data); |
562 | hi = hi->next; |
563 | } |
564 | return NULL; |
565 | } |
566 | |
567 | /* grow hash if it becomes too big */ |
568 | static void hash_rebuild(xhash *hash) |
569 | { |
570 | unsigned newsize, i, idx; |
571 | hash_item **newitems, *hi, *thi; |
572 | |
573 | if (hash->nprime == ARRAY_SIZE(PRIMES)) |
574 | return; |
575 | |
576 | newsize = PRIMES[hash->nprime++]; |
577 | newitems = xzalloc(newsize * sizeof(newitems[0])); |
578 | |
579 | for (i = 0; i < hash->csize; i++) { |
580 | hi = hash->items[i]; |
581 | while (hi) { |
582 | thi = hi; |
583 | hi = thi->next; |
584 | idx = hashidx(thi->name) % newsize; |
585 | thi->next = newitems[idx]; |
586 | newitems[idx] = thi; |
587 | } |
588 | } |
589 | |
590 | free(hash->items); |
591 | hash->csize = newsize; |
592 | hash->items = newitems; |
593 | } |
594 | |
595 | /* find item in hash, add it if necessary. Return ptr to data */ |
596 | static void *hash_find(xhash *hash, const char *name) |
597 | { |
598 | hash_item *hi; |
599 | unsigned idx; |
600 | int l; |
601 | |
602 | hi = hash_search(hash, name); |
603 | if (!hi) { |
604 | if (++hash->nel / hash->csize > 10) |
605 | hash_rebuild(hash); |
606 | |
607 | l = strlen(name) + 1; |
608 | hi = xzalloc(sizeof(*hi) + l); |
609 | strcpy(hi->name, name); |
610 | |
611 | idx = hashidx(name) % hash->csize; |
612 | hi->next = hash->items[idx]; |
613 | hash->items[idx] = hi; |
614 | hash->glen += l; |
615 | } |
616 | return &(hi->data); |
617 | } |
618 | |
619 | #define findvar(hash, name) ((var*) hash_find((hash), (name))) |
620 | #define newvar(name) ((var*) hash_find(vhash, (name))) |
621 | #define newfile(name) ((rstream*)hash_find(fdhash, (name))) |
622 | #define newfunc(name) ((func*) hash_find(fnhash, (name))) |
623 | |
624 | static void hash_remove(xhash *hash, const char *name) |
625 | { |
626 | hash_item *hi, **phi; |
627 | |
628 | phi = &(hash->items[hashidx(name) % hash->csize]); |
629 | while (*phi) { |
630 | hi = *phi; |
631 | if (strcmp(hi->name, name) == 0) { |
632 | hash->glen -= (strlen(name) + 1); |
633 | hash->nel--; |
634 | *phi = hi->next; |
635 | free(hi); |
636 | break; |
637 | } |
638 | phi = &(hi->next); |
639 | } |
640 | } |
641 | |
642 | /* ------ some useful functions ------ */ |
643 | |
644 | static void skip_spaces(char **s) |
645 | { |
646 | char *p = *s; |
647 | |
648 | while (1) { |
649 | if (*p == '\\' && p[1] == '\n') { |
650 | p++; |
651 | t_lineno++; |
652 | } else if (*p != ' ' && *p != '\t') { |
653 | break; |
654 | } |
655 | p++; |
656 | } |
657 | *s = p; |
658 | } |
659 | |
660 | static char *nextword(char **s) |
661 | { |
662 | char *p = *s; |
663 | while (*(*s)++) |
664 | continue; |
665 | return p; |
666 | } |
667 | |
668 | static char nextchar(char **s) |
669 | { |
670 | char c, *pps; |
671 | |
672 | c = *((*s)++); |
673 | pps = *s; |
674 | if (c == '\\') |
675 | c = bb_process_escape_sequence((const char**)s); |
676 | if (c == '\\' && *s == pps) |
677 | c = *((*s)++); |
678 | return c; |
679 | } |
680 | |
681 | static ALWAYS_INLINE int isalnum_(int c) |
682 | { |
683 | return (isalnum(c) || c == '_'); |
684 | } |
685 | |
686 | static double my_strtod(char **pp) |
687 | { |
688 | #if ENABLE_DESKTOP |
689 | if ((*pp)[0] == '0' |
690 | && ((((*pp)[1] | 0x20) == 'x') || isdigit((*pp)[1])) |
691 | ) { |
692 | return strtoull(*pp, pp, 0); |
693 | } |
694 | #endif |
695 | return strtod(*pp, pp); |
696 | } |
697 | |
698 | /* -------- working with variables (set/get/copy/etc) -------- */ |
699 | |
700 | static xhash *iamarray(var *v) |
701 | { |
702 | var *a = v; |
703 | |
704 | while (a->type & VF_CHILD) |
705 | a = a->x.parent; |
706 | |
707 | if (!(a->type & VF_ARRAY)) { |
708 | a->type |= VF_ARRAY; |
709 | a->x.array = hash_init(); |
710 | } |
711 | return a->x.array; |
712 | } |
713 | |
714 | static void clear_array(xhash *array) |
715 | { |
716 | unsigned i; |
717 | hash_item *hi, *thi; |
718 | |
719 | for (i = 0; i < array->csize; i++) { |
720 | hi = array->items[i]; |
721 | while (hi) { |
722 | thi = hi; |
723 | hi = hi->next; |
724 | free(thi->data.v.string); |
725 | free(thi); |
726 | } |
727 | array->items[i] = NULL; |
728 | } |
729 | array->glen = array->nel = 0; |
730 | } |
731 | |
732 | /* clear a variable */ |
733 | static var *clrvar(var *v) |
734 | { |
735 | if (!(v->type & VF_FSTR)) |
736 | free(v->string); |
737 | |
738 | v->type &= VF_DONTTOUCH; |
739 | v->type |= VF_DIRTY; |
740 | v->string = NULL; |
741 | return v; |
742 | } |
743 | |
744 | /* assign string value to variable */ |
745 | static var *setvar_p(var *v, char *value) |
746 | { |
747 | clrvar(v); |
748 | v->string = value; |
749 | handle_special(v); |
750 | return v; |
751 | } |
752 | |
753 | /* same as setvar_p but make a copy of string */ |
754 | static var *setvar_s(var *v, const char *value) |
755 | { |
756 | return setvar_p(v, (value && *value) ? xstrdup(value) : NULL); |
757 | } |
758 | |
759 | /* same as setvar_s but sets USER flag */ |
760 | static var *setvar_u(var *v, const char *value) |
761 | { |
762 | v = setvar_s(v, value); |
763 | v->type |= VF_USER; |
764 | return v; |
765 | } |
766 | |
767 | /* set array element to user string */ |
768 | static void setari_u(var *a, int idx, const char *s) |
769 | { |
770 | var *v; |
771 | |
772 | v = findvar(iamarray(a), itoa(idx)); |
773 | setvar_u(v, s); |
774 | } |
775 | |
776 | /* assign numeric value to variable */ |
777 | static var *setvar_i(var *v, double value) |
778 | { |
779 | clrvar(v); |
780 | v->type |= VF_NUMBER; |
781 | v->number = value; |
782 | handle_special(v); |
783 | return v; |
784 | } |
785 | |
786 | static const char *getvar_s(var *v) |
787 | { |
788 | /* if v is numeric and has no cached string, convert it to string */ |
789 | if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) { |
790 | fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE); |
791 | v->string = xstrdup(g_buf); |
792 | v->type |= VF_CACHED; |
793 | } |
794 | return (v->string == NULL) ? "" : v->string; |
795 | } |
796 | |
797 | static double getvar_i(var *v) |
798 | { |
799 | char *s; |
800 | |
801 | if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) { |
802 | v->number = 0; |
803 | s = v->string; |
804 | if (s && *s) { |
805 | v->number = my_strtod(&s); |
806 | if (v->type & VF_USER) { |
807 | skip_spaces(&s); |
808 | if (*s != '\0') |
809 | v->type &= ~VF_USER; |
810 | } |
811 | } else { |
812 | v->type &= ~VF_USER; |
813 | } |
814 | v->type |= VF_CACHED; |
815 | } |
816 | return v->number; |
817 | } |
818 | |
819 | /* Used for operands of bitwise ops */ |
820 | static unsigned long getvar_i_int(var *v) |
821 | { |
822 | double d = getvar_i(v); |
823 | |
824 | /* Casting doubles to longs is undefined for values outside |
825 | * of target type range. Try to widen it as much as possible */ |
826 | if (d >= 0) |
827 | return (unsigned long)d; |
828 | /* Why? Think about d == -4294967295.0 (assuming 32bit longs) */ |
829 | return - (long) (unsigned long) (-d); |
830 | } |
831 | |
832 | static var *copyvar(var *dest, const var *src) |
833 | { |
834 | if (dest != src) { |
835 | clrvar(dest); |
836 | dest->type |= (src->type & ~(VF_DONTTOUCH | VF_FSTR)); |
837 | dest->number = src->number; |
838 | if (src->string) |
839 | dest->string = xstrdup(src->string); |
840 | } |
841 | handle_special(dest); |
842 | return dest; |
843 | } |
844 | |
845 | static var *incvar(var *v) |
846 | { |
847 | return setvar_i(v, getvar_i(v) + 1.0); |
848 | } |
849 | |
850 | /* return true if v is number or numeric string */ |
851 | static int is_numeric(var *v) |
852 | { |
853 | getvar_i(v); |
854 | return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY)); |
855 | } |
856 | |
857 | /* return 1 when value of v corresponds to true, 0 otherwise */ |
858 | static int istrue(var *v) |
859 | { |
860 | if (is_numeric(v)) |
861 | return (v->number != 0); |
862 | return (v->string && v->string[0]); |
863 | } |
864 | |
865 | /* temporary variables allocator. Last allocated should be first freed */ |
866 | static var *nvalloc(int n) |
867 | { |
868 | nvblock *pb = NULL; |
869 | var *v, *r; |
870 | int size; |
871 | |
872 | while (g_cb) { |
873 | pb = g_cb; |
874 | if ((g_cb->pos - g_cb->nv) + n <= g_cb->size) |
875 | break; |
876 | g_cb = g_cb->next; |
877 | } |
878 | |
879 | if (!g_cb) { |
880 | size = (n <= MINNVBLOCK) ? MINNVBLOCK : n; |
881 | g_cb = xzalloc(sizeof(nvblock) + size * sizeof(var)); |
882 | g_cb->size = size; |
883 | g_cb->pos = g_cb->nv; |
884 | g_cb->prev = pb; |
885 | /*g_cb->next = NULL; - xzalloc did it */ |
886 | if (pb) |
887 | pb->next = g_cb; |
888 | } |
889 | |
890 | v = r = g_cb->pos; |
891 | g_cb->pos += n; |
892 | |
893 | while (v < g_cb->pos) { |
894 | v->type = 0; |
895 | v->string = NULL; |
896 | v++; |
897 | } |
898 | |
899 | return r; |
900 | } |
901 | |
902 | static void nvfree(var *v) |
903 | { |
904 | var *p; |
905 | |
906 | if (v < g_cb->nv || v >= g_cb->pos) |
907 | syntax_error(EMSG_INTERNAL_ERROR); |
908 | |
909 | for (p = v; p < g_cb->pos; p++) { |
910 | if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) { |
911 | clear_array(iamarray(p)); |
912 | free(p->x.array->items); |
913 | free(p->x.array); |
914 | } |
915 | if (p->type & VF_WALK) |
916 | free(p->x.walker); |
917 | |
918 | clrvar(p); |
919 | } |
920 | |
921 | g_cb->pos = v; |
922 | while (g_cb->prev && g_cb->pos == g_cb->nv) { |
923 | g_cb = g_cb->prev; |
924 | } |
925 | } |
926 | |
927 | /* ------- awk program text parsing ------- */ |
928 | |
929 | /* Parse next token pointed by global pos, place results into global ttt. |
930 | * If token isn't expected, give away. Return token class |
931 | */ |
932 | static uint32_t next_token(uint32_t expected) |
933 | { |
934 | #define concat_inserted (G.next_token__concat_inserted) |
935 | #define save_tclass (G.next_token__save_tclass) |
936 | #define save_info (G.next_token__save_info) |
937 | /* Initialized to TC_OPTERM: */ |
938 | #define ltclass (G.next_token__ltclass) |
939 | |
940 | char *p, *pp, *s; |
941 | const char *tl; |
942 | uint32_t tc; |
943 | const uint32_t *ti; |
944 | int l; |
945 | |
946 | if (t_rollback) { |
947 | t_rollback = FALSE; |
948 | |
949 | } else if (concat_inserted) { |
950 | concat_inserted = FALSE; |
951 | t_tclass = save_tclass; |
952 | t_info = save_info; |
953 | |
954 | } else { |
955 | p = g_pos; |
956 | readnext: |
957 | skip_spaces(&p); |
958 | g_lineno = t_lineno; |
959 | if (*p == '#') |
960 | while (*p != '\n' && *p != '\0') |
961 | p++; |
962 | |
963 | if (*p == '\n') |
964 | t_lineno++; |
965 | |
966 | if (*p == '\0') { |
967 | tc = TC_EOF; |
968 | |
969 | } else if (*p == '\"') { |
970 | /* it's a string */ |
971 | t_string = s = ++p; |
972 | while (*p != '\"') { |
973 | if (*p == '\0' || *p == '\n') |
974 | syntax_error(EMSG_UNEXP_EOS); |
975 | *(s++) = nextchar(&p); |
976 | } |
977 | p++; |
978 | *s = '\0'; |
979 | tc = TC_STRING; |
980 | |
981 | } else if ((expected & TC_REGEXP) && *p == '/') { |
982 | /* it's regexp */ |
983 | t_string = s = ++p; |
984 | while (*p != '/') { |
985 | if (*p == '\0' || *p == '\n') |
986 | syntax_error(EMSG_UNEXP_EOS); |
987 | *s = *p++; |
988 | if (*s++ == '\\') { |
989 | pp = p; |
990 | *(s-1) = bb_process_escape_sequence((const char **)&p); |
991 | if (*pp == '\\') |
992 | *s++ = '\\'; |
993 | if (p == pp) |
994 | *s++ = *p++; |
995 | } |
996 | } |
997 | p++; |
998 | *s = '\0'; |
999 | tc = TC_REGEXP; |
1000 | |
1001 | } else if (*p == '.' || isdigit(*p)) { |
1002 | /* it's a number */ |
1003 | t_double = my_strtod(&p); |
1004 | if (*p == '.') |
1005 | syntax_error(EMSG_UNEXP_TOKEN); |
1006 | tc = TC_NUMBER; |
1007 | |
1008 | } else { |
1009 | /* search for something known */ |
1010 | tl = tokenlist; |
1011 | tc = 0x00000001; |
1012 | ti = tokeninfo; |
1013 | while (*tl) { |
1014 | l = *(tl++); |
1015 | if (l == NTCC) { |
1016 | tc <<= 1; |
1017 | continue; |
1018 | } |
1019 | /* if token class is expected, token |
1020 | * matches and it's not a longer word, |
1021 | * then this is what we are looking for |
1022 | */ |
1023 | if ((tc & (expected | TC_WORD | TC_NEWLINE)) |
1024 | && *tl == *p && strncmp(p, tl, l) == 0 |
1025 | && !((tc & TC_WORD) && isalnum_(p[l])) |
1026 | ) { |
1027 | t_info = *ti; |
1028 | p += l; |
1029 | break; |
1030 | } |
1031 | ti++; |
1032 | tl += l; |
1033 | } |
1034 | |
1035 | if (!*tl) { |
1036 | /* it's a name (var/array/function), |
1037 | * otherwise it's something wrong |
1038 | */ |
1039 | if (!isalnum_(*p)) |
1040 | syntax_error(EMSG_UNEXP_TOKEN); |
1041 | |
1042 | t_string = --p; |
1043 | while (isalnum_(*(++p))) { |
1044 | *(p-1) = *p; |
1045 | } |
1046 | *(p-1) = '\0'; |
1047 | tc = TC_VARIABLE; |
1048 | /* also consume whitespace between functionname and bracket */ |
1049 | if (!(expected & TC_VARIABLE)) |
1050 | skip_spaces(&p); |
1051 | if (*p == '(') { |
1052 | tc = TC_FUNCTION; |
1053 | } else { |
1054 | if (*p == '[') { |
1055 | p++; |
1056 | tc = TC_ARRAY; |
1057 | } |
1058 | } |
1059 | } |
1060 | } |
1061 | g_pos = p; |
1062 | |
1063 | /* skipping newlines in some cases */ |
1064 | if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE)) |
1065 | goto readnext; |
1066 | |
1067 | /* insert concatenation operator when needed */ |
1068 | if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) { |
1069 | concat_inserted = TRUE; |
1070 | save_tclass = tc; |
1071 | save_info = t_info; |
1072 | tc = TC_BINOP; |
1073 | t_info = OC_CONCAT | SS | P(35); |
1074 | } |
1075 | |
1076 | t_tclass = tc; |
1077 | } |
1078 | ltclass = t_tclass; |
1079 | |
1080 | /* Are we ready for this? */ |
1081 | if (!(ltclass & expected)) |
1082 | syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ? |
1083 | EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN); |
1084 | |
1085 | return ltclass; |
1086 | #undef concat_inserted |
1087 | #undef save_tclass |
1088 | #undef save_info |
1089 | #undef ltclass |
1090 | } |
1091 | |
1092 | static void rollback_token(void) |
1093 | { |
1094 | t_rollback = TRUE; |
1095 | } |
1096 | |
1097 | static node *new_node(uint32_t info) |
1098 | { |
1099 | node *n; |
1100 | |
1101 | n = xzalloc(sizeof(node)); |
1102 | n->info = info; |
1103 | n->lineno = g_lineno; |
1104 | return n; |
1105 | } |
1106 | |
1107 | static node *mk_re_node(const char *s, node *n, regex_t *re) |
1108 | { |
1109 | n->info = OC_REGEXP; |
1110 | n->l.re = re; |
1111 | n->r.ire = re + 1; |
1112 | xregcomp(re, s, REG_EXTENDED); |
1113 | xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE); |
1114 | |
1115 | return n; |
1116 | } |
1117 | |
1118 | static node *condition(void) |
1119 | { |
1120 | next_token(TC_SEQSTART); |
1121 | return parse_expr(TC_SEQTERM); |
1122 | } |
1123 | |
1124 | /* parse expression terminated by given argument, return ptr |
1125 | * to built subtree. Terminator is eaten by parse_expr */ |
1126 | static node *parse_expr(uint32_t iexp) |
1127 | { |
1128 | node sn; |
1129 | node *cn = &sn; |
1130 | node *vn, *glptr; |
1131 | uint32_t tc, xtc; |
1132 | var *v; |
1133 | |
1134 | sn.info = PRIMASK; |
1135 | sn.r.n = glptr = NULL; |
1136 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp; |
1137 | |
1138 | while (!((tc = next_token(xtc)) & iexp)) { |
1139 | if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) { |
1140 | /* input redirection (<) attached to glptr node */ |
1141 | cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37)); |
1142 | cn->a.n = glptr; |
1143 | xtc = TC_OPERAND | TC_UOPPRE; |
1144 | glptr = NULL; |
1145 | |
1146 | } else if (tc & (TC_BINOP | TC_UOPPOST)) { |
1147 | /* for binary and postfix-unary operators, jump back over |
1148 | * previous operators with higher priority */ |
1149 | vn = cn; |
1150 | while (((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2)) |
1151 | || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON)) |
1152 | ) { |
1153 | vn = vn->a.n; |
1154 | } |
1155 | if ((t_info & OPCLSMASK) == OC_TERNARY) |
1156 | t_info += P(6); |
1157 | cn = vn->a.n->r.n = new_node(t_info); |
1158 | cn->a.n = vn->a.n; |
1159 | if (tc & TC_BINOP) { |
1160 | cn->l.n = vn; |
1161 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
1162 | if ((t_info & OPCLSMASK) == OC_PGETLINE) { |
1163 | /* it's a pipe */ |
1164 | next_token(TC_GETLINE); |
1165 | /* give maximum priority to this pipe */ |
1166 | cn->info &= ~PRIMASK; |
1167 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1168 | } |
1169 | } else { |
1170 | cn->r.n = vn; |
1171 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1172 | } |
1173 | vn->a.n = cn; |
1174 | |
1175 | } else { |
1176 | /* for operands and prefix-unary operators, attach them |
1177 | * to last node */ |
1178 | vn = cn; |
1179 | cn = vn->r.n = new_node(t_info); |
1180 | cn->a.n = vn; |
1181 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
1182 | if (tc & (TC_OPERAND | TC_REGEXP)) { |
1183 | xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp; |
1184 | /* one should be very careful with switch on tclass - |
1185 | * only simple tclasses should be used! */ |
1186 | switch (tc) { |
1187 | case TC_VARIABLE: |
1188 | case TC_ARRAY: |
1189 | cn->info = OC_VAR; |
1190 | v = hash_search(ahash, t_string); |
1191 | if (v != NULL) { |
1192 | cn->info = OC_FNARG; |
1193 | cn->l.i = v->x.aidx; |
1194 | } else { |
1195 | cn->l.v = newvar(t_string); |
1196 | } |
1197 | if (tc & TC_ARRAY) { |
1198 | cn->info |= xS; |
1199 | cn->r.n = parse_expr(TC_ARRTERM); |
1200 | } |
1201 | break; |
1202 | |
1203 | case TC_NUMBER: |
1204 | case TC_STRING: |
1205 | cn->info = OC_VAR; |
1206 | v = cn->l.v = xzalloc(sizeof(var)); |
1207 | if (tc & TC_NUMBER) |
1208 | setvar_i(v, t_double); |
1209 | else |
1210 | setvar_s(v, t_string); |
1211 | break; |
1212 | |
1213 | case TC_REGEXP: |
1214 | mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2)); |
1215 | break; |
1216 | |
1217 | case TC_FUNCTION: |
1218 | cn->info = OC_FUNC; |
1219 | cn->r.f = newfunc(t_string); |
1220 | cn->l.n = condition(); |
1221 | break; |
1222 | |
1223 | case TC_SEQSTART: |
1224 | cn = vn->r.n = parse_expr(TC_SEQTERM); |
1225 | cn->a.n = vn; |
1226 | break; |
1227 | |
1228 | case TC_GETLINE: |
1229 | glptr = cn; |
1230 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1231 | break; |
1232 | |
1233 | case TC_BUILTIN: |
1234 | cn->l.n = condition(); |
1235 | break; |
1236 | } |
1237 | } |
1238 | } |
1239 | } |
1240 | return sn.r.n; |
1241 | } |
1242 | |
1243 | /* add node to chain. Return ptr to alloc'd node */ |
1244 | static node *chain_node(uint32_t info) |
1245 | { |
1246 | node *n; |
1247 | |
1248 | if (!seq->first) |
1249 | seq->first = seq->last = new_node(0); |
1250 | |
1251 | if (seq->programname != g_progname) { |
1252 | seq->programname = g_progname; |
1253 | n = chain_node(OC_NEWSOURCE); |
1254 | n->l.s = xstrdup(g_progname); |
1255 | } |
1256 | |
1257 | n = seq->last; |
1258 | n->info = info; |
1259 | seq->last = n->a.n = new_node(OC_DONE); |
1260 | |
1261 | return n; |
1262 | } |
1263 | |
1264 | static void chain_expr(uint32_t info) |
1265 | { |
1266 | node *n; |
1267 | |
1268 | n = chain_node(info); |
1269 | n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
1270 | if (t_tclass & TC_GRPTERM) |
1271 | rollback_token(); |
1272 | } |
1273 | |
1274 | static node *chain_loop(node *nn) |
1275 | { |
1276 | node *n, *n2, *save_brk, *save_cont; |
1277 | |
1278 | save_brk = break_ptr; |
1279 | save_cont = continue_ptr; |
1280 | |
1281 | n = chain_node(OC_BR | Vx); |
1282 | continue_ptr = new_node(OC_EXEC); |
1283 | break_ptr = new_node(OC_EXEC); |
1284 | chain_group(); |
1285 | n2 = chain_node(OC_EXEC | Vx); |
1286 | n2->l.n = nn; |
1287 | n2->a.n = n; |
1288 | continue_ptr->a.n = n2; |
1289 | break_ptr->a.n = n->r.n = seq->last; |
1290 | |
1291 | continue_ptr = save_cont; |
1292 | break_ptr = save_brk; |
1293 | |
1294 | return n; |
1295 | } |
1296 | |
1297 | /* parse group and attach it to chain */ |
1298 | static void chain_group(void) |
1299 | { |
1300 | uint32_t c; |
1301 | node *n, *n2, *n3; |
1302 | |
1303 | do { |
1304 | c = next_token(TC_GRPSEQ); |
1305 | } while (c & TC_NEWLINE); |
1306 | |
1307 | if (c & TC_GRPSTART) { |
1308 | while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) { |
1309 | if (t_tclass & TC_NEWLINE) continue; |
1310 | rollback_token(); |
1311 | chain_group(); |
1312 | } |
1313 | } else if (c & (TC_OPSEQ | TC_OPTERM)) { |
1314 | rollback_token(); |
1315 | chain_expr(OC_EXEC | Vx); |
1316 | } else { /* TC_STATEMNT */ |
1317 | switch (t_info & OPCLSMASK) { |
1318 | case ST_IF: |
1319 | n = chain_node(OC_BR | Vx); |
1320 | n->l.n = condition(); |
1321 | chain_group(); |
1322 | n2 = chain_node(OC_EXEC); |
1323 | n->r.n = seq->last; |
1324 | if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE) == TC_ELSE) { |
1325 | chain_group(); |
1326 | n2->a.n = seq->last; |
1327 | } else { |
1328 | rollback_token(); |
1329 | } |
1330 | break; |
1331 | |
1332 | case ST_WHILE: |
1333 | n2 = condition(); |
1334 | n = chain_loop(NULL); |
1335 | n->l.n = n2; |
1336 | break; |
1337 | |
1338 | case ST_DO: |
1339 | n2 = chain_node(OC_EXEC); |
1340 | n = chain_loop(NULL); |
1341 | n2->a.n = n->a.n; |
1342 | next_token(TC_WHILE); |
1343 | n->l.n = condition(); |
1344 | break; |
1345 | |
1346 | case ST_FOR: |
1347 | next_token(TC_SEQSTART); |
1348 | n2 = parse_expr(TC_SEMICOL | TC_SEQTERM); |
1349 | if (t_tclass & TC_SEQTERM) { /* for-in */ |
1350 | if ((n2->info & OPCLSMASK) != OC_IN) |
1351 | syntax_error(EMSG_UNEXP_TOKEN); |
1352 | n = chain_node(OC_WALKINIT | VV); |
1353 | n->l.n = n2->l.n; |
1354 | n->r.n = n2->r.n; |
1355 | n = chain_loop(NULL); |
1356 | n->info = OC_WALKNEXT | Vx; |
1357 | n->l.n = n2->l.n; |
1358 | } else { /* for (;;) */ |
1359 | n = chain_node(OC_EXEC | Vx); |
1360 | n->l.n = n2; |
1361 | n2 = parse_expr(TC_SEMICOL); |
1362 | n3 = parse_expr(TC_SEQTERM); |
1363 | n = chain_loop(n3); |
1364 | n->l.n = n2; |
1365 | if (!n2) |
1366 | n->info = OC_EXEC; |
1367 | } |
1368 | break; |
1369 | |
1370 | case OC_PRINT: |
1371 | case OC_PRINTF: |
1372 | n = chain_node(t_info); |
1373 | n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM); |
1374 | if (t_tclass & TC_OUTRDR) { |
1375 | n->info |= t_info; |
1376 | n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
1377 | } |
1378 | if (t_tclass & TC_GRPTERM) |
1379 | rollback_token(); |
1380 | break; |
1381 | |
1382 | case OC_BREAK: |
1383 | n = chain_node(OC_EXEC); |
1384 | n->a.n = break_ptr; |
1385 | break; |
1386 | |
1387 | case OC_CONTINUE: |
1388 | n = chain_node(OC_EXEC); |
1389 | n->a.n = continue_ptr; |
1390 | break; |
1391 | |
1392 | /* delete, next, nextfile, return, exit */ |
1393 | default: |
1394 | chain_expr(t_info); |
1395 | } |
1396 | } |
1397 | } |
1398 | |
1399 | static void parse_program(char *p) |
1400 | { |
1401 | uint32_t tclass; |
1402 | node *cn; |
1403 | func *f; |
1404 | var *v; |
1405 | |
1406 | g_pos = p; |
1407 | t_lineno = 1; |
1408 | while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART | |
1409 | TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) { |
1410 | |
1411 | if (tclass & TC_OPTERM) |
1412 | continue; |
1413 | |
1414 | seq = &mainseq; |
1415 | if (tclass & TC_BEGIN) { |
1416 | seq = &beginseq; |
1417 | chain_group(); |
1418 | |
1419 | } else if (tclass & TC_END) { |
1420 | seq = &endseq; |
1421 | chain_group(); |
1422 | |
1423 | } else if (tclass & TC_FUNCDECL) { |
1424 | next_token(TC_FUNCTION); |
1425 | g_pos++; |
1426 | f = newfunc(t_string); |
1427 | f->body.first = NULL; |
1428 | f->nargs = 0; |
1429 | while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) { |
1430 | v = findvar(ahash, t_string); |
1431 | v->x.aidx = (f->nargs)++; |
1432 | |
1433 | if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM) |
1434 | break; |
1435 | } |
1436 | seq = &(f->body); |
1437 | chain_group(); |
1438 | clear_array(ahash); |
1439 | |
1440 | } else if (tclass & TC_OPSEQ) { |
1441 | rollback_token(); |
1442 | cn = chain_node(OC_TEST); |
1443 | cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART); |
1444 | if (t_tclass & TC_GRPSTART) { |
1445 | rollback_token(); |
1446 | chain_group(); |
1447 | } else { |
1448 | chain_node(OC_PRINT); |
1449 | } |
1450 | cn->r.n = mainseq.last; |
1451 | |
1452 | } else /* if (tclass & TC_GRPSTART) */ { |
1453 | rollback_token(); |
1454 | chain_group(); |
1455 | } |
1456 | } |
1457 | } |
1458 | |
1459 | |
1460 | /* -------- program execution part -------- */ |
1461 | |
1462 | static node *mk_splitter(const char *s, tsplitter *spl) |
1463 | { |
1464 | regex_t *re, *ire; |
1465 | node *n; |
1466 | |
1467 | re = &spl->re[0]; |
1468 | ire = &spl->re[1]; |
1469 | n = &spl->n; |
1470 | if ((n->info & OPCLSMASK) == OC_REGEXP) { |
1471 | regfree(re); |
1472 | regfree(ire); // TODO: nuke ire, use re+1? |
1473 | } |
1474 | if (strlen(s) > 1) { |
1475 | mk_re_node(s, n, re); |
1476 | } else { |
1477 | n->info = (uint32_t) *s; |
1478 | } |
1479 | |
1480 | return n; |
1481 | } |
1482 | |
1483 | /* use node as a regular expression. Supplied with node ptr and regex_t |
1484 | * storage space. Return ptr to regex (if result points to preg, it should |
1485 | * be later regfree'd manually |
1486 | */ |
1487 | static regex_t *as_regex(node *op, regex_t *preg) |
1488 | { |
1489 | int cflags; |
1490 | var *v; |
1491 | const char *s; |
1492 | |
1493 | if ((op->info & OPCLSMASK) == OC_REGEXP) { |
1494 | return icase ? op->r.ire : op->l.re; |
1495 | } |
1496 | v = nvalloc(1); |
1497 | s = getvar_s(evaluate(op, v)); |
1498 | |
1499 | cflags = icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED; |
1500 | /* Testcase where REG_EXTENDED fails (unpaired '{'): |
1501 | * echo Hi | awk 'gsub("@(samp|code|file)\{","");' |
1502 | * gawk 3.1.5 eats this. We revert to ~REG_EXTENDED |
1503 | * (maybe gsub is not supposed to use REG_EXTENDED?). |
1504 | */ |
1505 | if (regcomp(preg, s, cflags)) { |
1506 | cflags &= ~REG_EXTENDED; |
1507 | xregcomp(preg, s, cflags); |
1508 | } |
1509 | nvfree(v); |
1510 | return preg; |
1511 | } |
1512 | |
1513 | /* gradually increasing buffer */ |
1514 | static void qrealloc(char **b, int n, int *size) |
1515 | { |
1516 | if (!*b || n >= *size) { |
1517 | *size = n + (n>>1) + 80; |
1518 | *b = xrealloc(*b, *size); |
1519 | } |
1520 | } |
1521 | |
1522 | /* resize field storage space */ |
1523 | static void fsrealloc(int size) |
1524 | { |
1525 | int i; |
1526 | |
1527 | if (size >= maxfields) { |
1528 | i = maxfields; |
1529 | maxfields = size + 16; |
1530 | Fields = xrealloc(Fields, maxfields * sizeof(var)); |
1531 | for (; i < maxfields; i++) { |
1532 | Fields[i].type = VF_SPECIAL; |
1533 | Fields[i].string = NULL; |
1534 | } |
1535 | } |
1536 | |
1537 | if (size < nfields) { |
1538 | for (i = size; i < nfields; i++) { |
1539 | clrvar(Fields + i); |
1540 | } |
1541 | } |
1542 | nfields = size; |
1543 | } |
1544 | |
1545 | static int awk_split(const char *s, node *spl, char **slist) |
1546 | { |
1547 | int l, n = 0; |
1548 | char c[4]; |
1549 | char *s1; |
1550 | regmatch_t pmatch[2]; // TODO: why [2]? [1] is enough... |
1551 | |
1552 | /* in worst case, each char would be a separate field */ |
1553 | *slist = s1 = xzalloc(strlen(s) * 2 + 3); |
1554 | strcpy(s1, s); |
1555 | |
1556 | c[0] = c[1] = (char)spl->info; |
1557 | c[2] = c[3] = '\0'; |
1558 | if (*getvar_s(intvar[RS]) == '\0') |
1559 | c[2] = '\n'; |
1560 | |
1561 | if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */ |
1562 | if (!*s) |
1563 | return n; /* "": zero fields */ |
1564 | n++; /* at least one field will be there */ |
1565 | do { |
1566 | l = strcspn(s, c+2); /* len till next NUL or \n */ |
1567 | if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0 |
1568 | && pmatch[0].rm_so <= l |
1569 | ) { |
1570 | l = pmatch[0].rm_so; |
1571 | if (pmatch[0].rm_eo == 0) { |
1572 | l++; |
1573 | pmatch[0].rm_eo++; |
1574 | } |
1575 | n++; /* we saw yet another delimiter */ |
1576 | } else { |
1577 | pmatch[0].rm_eo = l; |
1578 | if (s[l]) |
1579 | pmatch[0].rm_eo++; |
1580 | } |
1581 | memcpy(s1, s, l); |
1582 | /* make sure we remove *all* of the separator chars */ |
1583 | do { |
1584 | s1[l] = '\0'; |
1585 | } while (++l < pmatch[0].rm_eo); |
1586 | nextword(&s1); |
1587 | s += pmatch[0].rm_eo; |
1588 | } while (*s); |
1589 | return n; |
1590 | } |
1591 | if (c[0] == '\0') { /* null split */ |
1592 | while (*s) { |
1593 | *s1++ = *s++; |
1594 | *s1++ = '\0'; |
1595 | n++; |
1596 | } |
1597 | return n; |
1598 | } |
1599 | if (c[0] != ' ') { /* single-character split */ |
1600 | if (icase) { |
1601 | c[0] = toupper(c[0]); |
1602 | c[1] = tolower(c[1]); |
1603 | } |
1604 | if (*s1) n++; |
1605 | while ((s1 = strpbrk(s1, c))) { |
1606 | *s1++ = '\0'; |
1607 | n++; |
1608 | } |
1609 | return n; |
1610 | } |
1611 | /* space split */ |
1612 | while (*s) { |
1613 | s = skip_whitespace(s); |
1614 | if (!*s) break; |
1615 | n++; |
1616 | while (*s && !isspace(*s)) |
1617 | *s1++ = *s++; |
1618 | *s1++ = '\0'; |
1619 | } |
1620 | return n; |
1621 | } |
1622 | |
1623 | static void split_f0(void) |
1624 | { |
1625 | /* static char *fstrings; */ |
1626 | #define fstrings (G.split_f0__fstrings) |
1627 | |
1628 | int i, n; |
1629 | char *s; |
1630 | |
1631 | if (is_f0_split) |
1632 | return; |
1633 | |
1634 | is_f0_split = TRUE; |
1635 | free(fstrings); |
1636 | fsrealloc(0); |
1637 | n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings); |
1638 | fsrealloc(n); |
1639 | s = fstrings; |
1640 | for (i = 0; i < n; i++) { |
1641 | Fields[i].string = nextword(&s); |
1642 | Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY); |
1643 | } |
1644 | |
1645 | /* set NF manually to avoid side effects */ |
1646 | clrvar(intvar[NF]); |
1647 | intvar[NF]->type = VF_NUMBER | VF_SPECIAL; |
1648 | intvar[NF]->number = nfields; |
1649 | #undef fstrings |
1650 | } |
1651 | |
1652 | /* perform additional actions when some internal variables changed */ |
1653 | static void handle_special(var *v) |
1654 | { |
1655 | int n; |
1656 | char *b; |
1657 | const char *sep, *s; |
1658 | int sl, l, len, i, bsize; |
1659 | |
1660 | if (!(v->type & VF_SPECIAL)) |
1661 | return; |
1662 | |
1663 | if (v == intvar[NF]) { |
1664 | n = (int)getvar_i(v); |
1665 | fsrealloc(n); |
1666 | |
1667 | /* recalculate $0 */ |
1668 | sep = getvar_s(intvar[OFS]); |
1669 | sl = strlen(sep); |
1670 | b = NULL; |
1671 | len = 0; |
1672 | for (i = 0; i < n; i++) { |
1673 | s = getvar_s(&Fields[i]); |
1674 | l = strlen(s); |
1675 | if (b) { |
1676 | memcpy(b+len, sep, sl); |
1677 | len += sl; |
1678 | } |
1679 | qrealloc(&b, len+l+sl, &bsize); |
1680 | memcpy(b+len, s, l); |
1681 | len += l; |
1682 | } |
1683 | if (b) |
1684 | b[len] = '\0'; |
1685 | setvar_p(intvar[F0], b); |
1686 | is_f0_split = TRUE; |
1687 | |
1688 | } else if (v == intvar[F0]) { |
1689 | is_f0_split = FALSE; |
1690 | |
1691 | } else if (v == intvar[FS]) { |
1692 | mk_splitter(getvar_s(v), &fsplitter); |
1693 | |
1694 | } else if (v == intvar[RS]) { |
1695 | mk_splitter(getvar_s(v), &rsplitter); |
1696 | |
1697 | } else if (v == intvar[IGNORECASE]) { |
1698 | icase = istrue(v); |
1699 | |
1700 | } else { /* $n */ |
1701 | n = getvar_i(intvar[NF]); |
1702 | setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1); |
1703 | /* right here v is invalid. Just to note... */ |
1704 | } |
1705 | } |
1706 | |
1707 | /* step through func/builtin/etc arguments */ |
1708 | static node *nextarg(node **pn) |
1709 | { |
1710 | node *n; |
1711 | |
1712 | n = *pn; |
1713 | if (n && (n->info & OPCLSMASK) == OC_COMMA) { |
1714 | *pn = n->r.n; |
1715 | n = n->l.n; |
1716 | } else { |
1717 | *pn = NULL; |
1718 | } |
1719 | return n; |
1720 | } |
1721 | |
1722 | static void hashwalk_init(var *v, xhash *array) |
1723 | { |
1724 | char **w; |
1725 | hash_item *hi; |
1726 | unsigned i; |
1727 | |
1728 | if (v->type & VF_WALK) |
1729 | free(v->x.walker); |
1730 | |
1731 | v->type |= VF_WALK; |
1732 | w = v->x.walker = xzalloc(2 + 2*sizeof(char *) + array->glen); |
1733 | w[0] = w[1] = (char *)(w + 2); |
1734 | for (i = 0; i < array->csize; i++) { |
1735 | hi = array->items[i]; |
1736 | while (hi) { |
1737 | strcpy(*w, hi->name); |
1738 | nextword(w); |
1739 | hi = hi->next; |
1740 | } |
1741 | } |
1742 | } |
1743 | |
1744 | static int hashwalk_next(var *v) |
1745 | { |
1746 | char **w; |
1747 | |
1748 | w = v->x.walker; |
1749 | if (w[1] == w[0]) |
1750 | return FALSE; |
1751 | |
1752 | setvar_s(v, nextword(w+1)); |
1753 | return TRUE; |
1754 | } |
1755 | |
1756 | /* evaluate node, return 1 when result is true, 0 otherwise */ |
1757 | static int ptest(node *pattern) |
1758 | { |
1759 | /* ptest__v is "static": to save stack space? */ |
1760 | return istrue(evaluate(pattern, &G.ptest__v)); |
1761 | } |
1762 | |
1763 | /* read next record from stream rsm into a variable v */ |
1764 | static int awk_getline(rstream *rsm, var *v) |
1765 | { |
1766 | char *b; |
1767 | regmatch_t pmatch[2]; |
1768 | int a, p, pp=0, size; |
1769 | int fd, so, eo, r, rp; |
1770 | char c, *m, *s; |
1771 | |
1772 | /* we're using our own buffer since we need access to accumulating |
1773 | * characters |
1774 | */ |
1775 | fd = fileno(rsm->F); |
1776 | m = rsm->buffer; |
1777 | a = rsm->adv; |
1778 | p = rsm->pos; |
1779 | size = rsm->size; |
1780 | c = (char) rsplitter.n.info; |
1781 | rp = 0; |
1782 | |
1783 | if (!m) qrealloc(&m, 256, &size); |
1784 | do { |
1785 | b = m + a; |
1786 | so = eo = p; |
1787 | r = 1; |
1788 | if (p > 0) { |
1789 | if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) { |
1790 | if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re, |
1791 | b, 1, pmatch, 0) == 0) { |
1792 | so = pmatch[0].rm_so; |
1793 | eo = pmatch[0].rm_eo; |
1794 | if (b[eo] != '\0') |
1795 | break; |
1796 | } |
1797 | } else if (c != '\0') { |
1798 | s = strchr(b+pp, c); |
1799 | if (!s) s = memchr(b+pp, '\0', p - pp); |
1800 | if (s) { |
1801 | so = eo = s-b; |
1802 | eo++; |
1803 | break; |
1804 | } |
1805 | } else { |
1806 | while (b[rp] == '\n') |
1807 | rp++; |
1808 | s = strstr(b+rp, "\n\n"); |
1809 | if (s) { |
1810 | so = eo = s-b; |
1811 | while (b[eo] == '\n') eo++; |
1812 | if (b[eo] != '\0') |
1813 | break; |
1814 | } |
1815 | } |
1816 | } |
1817 | |
1818 | if (a > 0) { |
1819 | memmove(m, (const void *)(m+a), p+1); |
1820 | b = m; |
1821 | a = 0; |
1822 | } |
1823 | |
1824 | qrealloc(&m, a+p+128, &size); |
1825 | b = m + a; |
1826 | pp = p; |
1827 | p += safe_read(fd, b+p, size-p-1); |
1828 | if (p < pp) { |
1829 | p = 0; |
1830 | r = 0; |
1831 | setvar_i(intvar[ERRNO], errno); |
1832 | } |
1833 | b[p] = '\0'; |
1834 | |
1835 | } while (p > pp); |
1836 | |
1837 | if (p == 0) { |
1838 | r--; |
1839 | } else { |
1840 | c = b[so]; b[so] = '\0'; |
1841 | setvar_s(v, b+rp); |
1842 | v->type |= VF_USER; |
1843 | b[so] = c; |
1844 | c = b[eo]; b[eo] = '\0'; |
1845 | setvar_s(intvar[RT], b+so); |
1846 | b[eo] = c; |
1847 | } |
1848 | |
1849 | rsm->buffer = m; |
1850 | rsm->adv = a + eo; |
1851 | rsm->pos = p - eo; |
1852 | rsm->size = size; |
1853 | |
1854 | return r; |
1855 | } |
1856 | |
1857 | static int fmt_num(char *b, int size, const char *format, double n, int int_as_int) |
1858 | { |
1859 | int r = 0; |
1860 | char c; |
1861 | const char *s = format; |
1862 | |
1863 | if (int_as_int && n == (int)n) { |
1864 | r = snprintf(b, size, "%d", (int)n); |
1865 | } else { |
1866 | do { c = *s; } while (c && *++s); |
1867 | if (strchr("diouxX", c)) { |
1868 | r = snprintf(b, size, format, (int)n); |
1869 | } else if (strchr("eEfgG", c)) { |
1870 | r = snprintf(b, size, format, n); |
1871 | } else { |
1872 | syntax_error(EMSG_INV_FMT); |
1873 | } |
1874 | } |
1875 | return r; |
1876 | } |
1877 | |
1878 | /* formatted output into an allocated buffer, return ptr to buffer */ |
1879 | static char *awk_printf(node *n) |
1880 | { |
1881 | char *b = NULL; |
1882 | char *fmt, *s, *f; |
1883 | const char *s1; |
1884 | int i, j, incr, bsize; |
1885 | char c, c1; |
1886 | var *v, *arg; |
1887 | |
1888 | v = nvalloc(1); |
1889 | fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v))); |
1890 | |
1891 | i = 0; |
1892 | while (*f) { |
1893 | s = f; |
1894 | while (*f && (*f != '%' || *(++f) == '%')) |
1895 | f++; |
1896 | while (*f && !isalpha(*f)) { |
1897 | if (*f == '*') |
1898 | syntax_error("%*x formats are not supported"); |
1899 | f++; |
1900 | } |
1901 | |
1902 | incr = (f - s) + MAXVARFMT; |
1903 | qrealloc(&b, incr + i, &bsize); |
1904 | c = *f; |
1905 | if (c != '\0') f++; |
1906 | c1 = *f; |
1907 | *f = '\0'; |
1908 | arg = evaluate(nextarg(&n), v); |
1909 | |
1910 | j = i; |
1911 | if (c == 'c' || !c) { |
1912 | i += sprintf(b+i, s, is_numeric(arg) ? |
1913 | (char)getvar_i(arg) : *getvar_s(arg)); |
1914 | } else if (c == 's') { |
1915 | s1 = getvar_s(arg); |
1916 | qrealloc(&b, incr+i+strlen(s1), &bsize); |
1917 | i += sprintf(b+i, s, s1); |
1918 | } else { |
1919 | i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE); |
1920 | } |
1921 | *f = c1; |
1922 | |
1923 | /* if there was an error while sprintf, return value is negative */ |
1924 | if (i < j) i = j; |
1925 | } |
1926 | |
1927 | b = xrealloc(b, i + 1); |
1928 | free(fmt); |
1929 | nvfree(v); |
1930 | b[i] = '\0'; |
1931 | return b; |
1932 | } |
1933 | |
1934 | /* common substitution routine |
1935 | * replace (nm) substring of (src) that match (n) with (repl), store |
1936 | * result into (dest), return number of substitutions. If nm=0, replace |
1937 | * all matches. If src or dst is NULL, use $0. If ex=TRUE, enable |
1938 | * subexpression matching (\1-\9) |
1939 | */ |
1940 | static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int ex) |
1941 | { |
1942 | char *ds = NULL; |
1943 | const char *s; |
1944 | const char *sp; |
1945 | int c, i, j, di, rl, so, eo, nbs, n, dssize; |
1946 | regmatch_t pmatch[10]; |
1947 | regex_t sreg, *re; |
1948 | |
1949 | re = as_regex(rn, &sreg); |
1950 | if (!src) src = intvar[F0]; |
1951 | if (!dest) dest = intvar[F0]; |
1952 | |
1953 | i = di = 0; |
1954 | sp = getvar_s(src); |
1955 | rl = strlen(repl); |
1956 | while (regexec(re, sp, 10, pmatch, sp==getvar_s(src) ? 0 : REG_NOTBOL) == 0) { |
1957 | so = pmatch[0].rm_so; |
1958 | eo = pmatch[0].rm_eo; |
1959 | |
1960 | qrealloc(&ds, di + eo + rl, &dssize); |
1961 | memcpy(ds + di, sp, eo); |
1962 | di += eo; |
1963 | if (++i >= nm) { |
1964 | /* replace */ |
1965 | di -= (eo - so); |
1966 | nbs = 0; |
1967 | for (s = repl; *s; s++) { |
1968 | ds[di++] = c = *s; |
1969 | if (c == '\\') { |
1970 | nbs++; |
1971 | continue; |
1972 | } |
1973 | if (c == '&' || (ex && c >= '0' && c <= '9')) { |
1974 | di -= ((nbs + 3) >> 1); |
1975 | j = 0; |
1976 | if (c != '&') { |
1977 | j = c - '0'; |
1978 | nbs++; |
1979 | } |
1980 | if (nbs % 2) { |
1981 | ds[di++] = c; |
1982 | } else { |
1983 | n = pmatch[j].rm_eo - pmatch[j].rm_so; |
1984 | qrealloc(&ds, di + rl + n, &dssize); |
1985 | memcpy(ds + di, sp + pmatch[j].rm_so, n); |
1986 | di += n; |
1987 | } |
1988 | } |
1989 | nbs = 0; |
1990 | } |
1991 | } |
1992 | |
1993 | sp += eo; |
1994 | if (i == nm) |
1995 | break; |
1996 | if (eo == so) { |
1997 | ds[di] = *sp++; |
1998 | if (!ds[di++]) |
1999 | break; |
2000 | } |
2001 | } |
2002 | |
2003 | qrealloc(&ds, di + strlen(sp), &dssize); |
2004 | strcpy(ds + di, sp); |
2005 | setvar_p(dest, ds); |
2006 | if (re == &sreg) |
2007 | regfree(re); |
2008 | return i; |
2009 | } |
2010 | |
2011 | static NOINLINE int do_mktime(const char *ds) |
2012 | { |
2013 | struct tm then; |
2014 | int count; |
2015 | |
2016 | /*memset(&then, 0, sizeof(then)); - not needed */ |
2017 | then.tm_isdst = -1; /* default is unknown */ |
2018 | |
2019 | /* manpage of mktime says these fields are ints, |
2020 | * so we can sscanf stuff directly into them */ |
2021 | count = sscanf(ds, "%u %u %u %u %u %u %d", |
2022 | &then.tm_year, &then.tm_mon, &then.tm_mday, |
2023 | &then.tm_hour, &then.tm_min, &then.tm_sec, |
2024 | &then.tm_isdst); |
2025 | |
2026 | if (count < 6 |
2027 | || (unsigned)then.tm_mon < 1 |
2028 | || (unsigned)then.tm_year < 1900 |
2029 | ) { |
2030 | return -1; |
2031 | } |
2032 | |
2033 | then.tm_mon -= 1; |
2034 | then.tm_year -= 1900; |
2035 | |
2036 | return mktime(&then); |
2037 | } |
2038 | |
2039 | static NOINLINE var *exec_builtin(node *op, var *res) |
2040 | { |
2041 | #define tspl (G.exec_builtin__tspl) |
2042 | |
2043 | var *tv; |
2044 | node *an[4]; |
2045 | var *av[4]; |
2046 | const char *as[4]; |
2047 | regmatch_t pmatch[2]; |
2048 | regex_t sreg, *re; |
2049 | node *spl; |
2050 | uint32_t isr, info; |
2051 | int nargs; |
2052 | time_t tt; |
2053 | char *s, *s1; |
2054 | int i, l, ll, n; |
2055 | |
2056 | tv = nvalloc(4); |
2057 | isr = info = op->info; |
2058 | op = op->l.n; |
2059 | |
2060 | av[2] = av[3] = NULL; |
2061 | for (i = 0; i < 4 && op; i++) { |
2062 | an[i] = nextarg(&op); |
2063 | if (isr & 0x09000000) av[i] = evaluate(an[i], &tv[i]); |
2064 | if (isr & 0x08000000) as[i] = getvar_s(av[i]); |
2065 | isr >>= 1; |
2066 | } |
2067 | |
2068 | nargs = i; |
2069 | if ((uint32_t)nargs < (info >> 30)) |
2070 | syntax_error(EMSG_TOO_FEW_ARGS); |
2071 | |
2072 | info &= OPNMASK; |
2073 | switch (info) { |
2074 | |
2075 | case B_a2: |
2076 | #if ENABLE_FEATURE_AWK_LIBM |
2077 | setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1]))); |
2078 | #else |
2079 | syntax_error(EMSG_NO_MATH); |
2080 | #endif |
2081 | break; |
2082 | |
2083 | case B_sp: |
2084 | if (nargs > 2) { |
2085 | spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ? |
2086 | an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl); |
2087 | } else { |
2088 | spl = &fsplitter.n; |
2089 | } |
2090 | |
2091 | n = awk_split(as[0], spl, &s); |
2092 | s1 = s; |
2093 | clear_array(iamarray(av[1])); |
2094 | for (i = 1; i <= n; i++) |
2095 | setari_u(av[1], i, nextword(&s1)); |
2096 | free(s); |
2097 | setvar_i(res, n); |
2098 | break; |
2099 | |
2100 | case B_ss: |
2101 | l = strlen(as[0]); |
2102 | i = getvar_i(av[1]) - 1; |
2103 | if (i > l) i = l; |
2104 | if (i < 0) i = 0; |
2105 | n = (nargs > 2) ? getvar_i(av[2]) : l-i; |
2106 | if (n < 0) n = 0; |
2107 | s = xstrndup(as[0]+i, n); |
2108 | setvar_p(res, s); |
2109 | break; |
2110 | |
2111 | /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5: |
2112 | * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */ |
2113 | case B_an: |
2114 | setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1])); |
2115 | break; |
2116 | |
2117 | case B_co: |
2118 | setvar_i(res, ~getvar_i_int(av[0])); |
2119 | break; |
2120 | |
2121 | case B_ls: |
2122 | setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1])); |
2123 | break; |
2124 | |
2125 | case B_or: |
2126 | setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1])); |
2127 | break; |
2128 | |
2129 | case B_rs: |
2130 | setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1])); |
2131 | break; |
2132 | |
2133 | case B_xo: |
2134 | setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1])); |
2135 | break; |
2136 | |
2137 | case B_lo: |
2138 | case B_up: |
2139 | s1 = s = xstrdup(as[0]); |
2140 | while (*s1) { |
2141 | //*s1 = (info == B_up) ? toupper(*s1) : tolower(*s1); |
2142 | if ((unsigned char)((*s1 | 0x20) - 'a') <= ('z' - 'a')) |
2143 | *s1 = (info == B_up) ? (*s1 & 0xdf) : (*s1 | 0x20); |
2144 | s1++; |
2145 | } |
2146 | setvar_p(res, s); |
2147 | break; |
2148 | |
2149 | case B_ix: |
2150 | n = 0; |
2151 | ll = strlen(as[1]); |
2152 | l = strlen(as[0]) - ll; |
2153 | if (ll > 0 && l >= 0) { |
2154 | if (!icase) { |
2155 | s = strstr(as[0], as[1]); |
2156 | if (s) n = (s - as[0]) + 1; |
2157 | } else { |
2158 | /* this piece of code is terribly slow and |
2159 | * really should be rewritten |
2160 | */ |
2161 | for (i=0; i<=l; i++) { |
2162 | if (strncasecmp(as[0]+i, as[1], ll) == 0) { |
2163 | n = i+1; |
2164 | break; |
2165 | } |
2166 | } |
2167 | } |
2168 | } |
2169 | setvar_i(res, n); |
2170 | break; |
2171 | |
2172 | case B_ti: |
2173 | if (nargs > 1) |
2174 | tt = getvar_i(av[1]); |
2175 | else |
2176 | time(&tt); |
2177 | //s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"; |
2178 | i = strftime(g_buf, MAXVARFMT, |
2179 | ((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"), |
2180 | localtime(&tt)); |
2181 | g_buf[i] = '\0'; |
2182 | setvar_s(res, g_buf); |
2183 | break; |
2184 | |
2185 | case B_mt: |
2186 | setvar_i(res, do_mktime(as[0])); |
2187 | break; |
2188 | |
2189 | case B_ma: |
2190 | re = as_regex(an[1], &sreg); |
2191 | n = regexec(re, as[0], 1, pmatch, 0); |
2192 | if (n == 0) { |
2193 | pmatch[0].rm_so++; |
2194 | pmatch[0].rm_eo++; |
2195 | } else { |
2196 | pmatch[0].rm_so = 0; |
2197 | pmatch[0].rm_eo = -1; |
2198 | } |
2199 | setvar_i(newvar("RSTART"), pmatch[0].rm_so); |
2200 | setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so); |
2201 | setvar_i(res, pmatch[0].rm_so); |
2202 | if (re == &sreg) regfree(re); |
2203 | break; |
2204 | |
2205 | case B_ge: |
2206 | awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE); |
2207 | break; |
2208 | |
2209 | case B_gs: |
2210 | setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE)); |
2211 | break; |
2212 | |
2213 | case B_su: |
2214 | setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE)); |
2215 | break; |
2216 | } |
2217 | |
2218 | nvfree(tv); |
2219 | return res; |
2220 | #undef tspl |
2221 | } |
2222 | |
2223 | /* |
2224 | * Evaluate node - the heart of the program. Supplied with subtree |
2225 | * and place where to store result. returns ptr to result. |
2226 | */ |
2227 | #define XC(n) ((n) >> 8) |
2228 | |
2229 | static var *evaluate(node *op, var *res) |
2230 | { |
2231 | /* This procedure is recursive so we should count every byte */ |
2232 | #define fnargs (G.evaluate__fnargs) |
2233 | /* seed is initialized to 1 */ |
2234 | #define seed (G.evaluate__seed) |
2235 | #define sreg (G.evaluate__sreg) |
2236 | |
2237 | node *op1; |
2238 | var *v1; |
2239 | union { |
2240 | var *v; |
2241 | const char *s; |
2242 | double d; |
2243 | int i; |
2244 | } L, R; |
2245 | uint32_t opinfo; |
2246 | int opn; |
2247 | union { |
2248 | char *s; |
2249 | rstream *rsm; |
2250 | FILE *F; |
2251 | var *v; |
2252 | regex_t *re; |
2253 | uint32_t info; |
2254 | } X; |
2255 | |
2256 | if (!op) |
2257 | return setvar_s(res, NULL); |
2258 | |
2259 | v1 = nvalloc(2); |
2260 | |
2261 | while (op) { |
2262 | opinfo = op->info; |
2263 | opn = (opinfo & OPNMASK); |
2264 | g_lineno = op->lineno; |
2265 | |
2266 | /* execute inevitable things */ |
2267 | op1 = op->l.n; |
2268 | if (opinfo & OF_RES1) X.v = L.v = evaluate(op1, v1); |
2269 | if (opinfo & OF_RES2) R.v = evaluate(op->r.n, v1+1); |
2270 | if (opinfo & OF_STR1) L.s = getvar_s(L.v); |
2271 | if (opinfo & OF_STR2) R.s = getvar_s(R.v); |
2272 | if (opinfo & OF_NUM1) L.d = getvar_i(L.v); |
2273 | |
2274 | switch (XC(opinfo & OPCLSMASK)) { |
2275 | |
2276 | /* -- iterative node type -- */ |
2277 | |
2278 | /* test pattern */ |
2279 | case XC( OC_TEST ): |
2280 | if ((op1->info & OPCLSMASK) == OC_COMMA) { |
2281 | /* it's range pattern */ |
2282 | if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) { |
2283 | op->info |= OF_CHECKED; |
2284 | if (ptest(op1->r.n)) |
2285 | op->info &= ~OF_CHECKED; |
2286 | |
2287 | op = op->a.n; |
2288 | } else { |
2289 | op = op->r.n; |
2290 | } |
2291 | } else { |
2292 | op = (ptest(op1)) ? op->a.n : op->r.n; |
2293 | } |
2294 | break; |
2295 | |
2296 | /* just evaluate an expression, also used as unconditional jump */ |
2297 | case XC( OC_EXEC ): |
2298 | break; |
2299 | |
2300 | /* branch, used in if-else and various loops */ |
2301 | case XC( OC_BR ): |
2302 | op = istrue(L.v) ? op->a.n : op->r.n; |
2303 | break; |
2304 | |
2305 | /* initialize for-in loop */ |
2306 | case XC( OC_WALKINIT ): |
2307 | hashwalk_init(L.v, iamarray(R.v)); |
2308 | break; |
2309 | |
2310 | /* get next array item */ |
2311 | case XC( OC_WALKNEXT ): |
2312 | op = hashwalk_next(L.v) ? op->a.n : op->r.n; |
2313 | break; |
2314 | |
2315 | case XC( OC_PRINT ): |
2316 | case XC( OC_PRINTF ): |
2317 | X.F = stdout; |
2318 | if (op->r.n) { |
2319 | X.rsm = newfile(R.s); |
2320 | if (!X.rsm->F) { |
2321 | if (opn == '|') { |
2322 | X.rsm->F = popen(R.s, "w"); |
2323 | if (X.rsm->F == NULL) |
2324 | bb_perror_msg_and_die("popen"); |
2325 | X.rsm->is_pipe = 1; |
2326 | } else { |
2327 | X.rsm->F = xfopen(R.s, opn=='w' ? "w" : "a"); |
2328 | } |
2329 | } |
2330 | X.F = X.rsm->F; |
2331 | } |
2332 | |
2333 | if ((opinfo & OPCLSMASK) == OC_PRINT) { |
2334 | if (!op1) { |
2335 | fputs(getvar_s(intvar[F0]), X.F); |
2336 | } else { |
2337 | while (op1) { |
2338 | L.v = evaluate(nextarg(&op1), v1); |
2339 | if (L.v->type & VF_NUMBER) { |
2340 | fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]), |
2341 | getvar_i(L.v), TRUE); |
2342 | fputs(g_buf, X.F); |
2343 | } else { |
2344 | fputs(getvar_s(L.v), X.F); |
2345 | } |
2346 | |
2347 | if (op1) fputs(getvar_s(intvar[OFS]), X.F); |
2348 | } |
2349 | } |
2350 | fputs(getvar_s(intvar[ORS]), X.F); |
2351 | |
2352 | } else { /* OC_PRINTF */ |
2353 | L.s = awk_printf(op1); |
2354 | fputs(L.s, X.F); |
2355 | free((char*)L.s); |
2356 | } |
2357 | fflush(X.F); |
2358 | break; |
2359 | |
2360 | case XC( OC_DELETE ): |
2361 | X.info = op1->info & OPCLSMASK; |
2362 | if (X.info == OC_VAR) { |
2363 | R.v = op1->l.v; |
2364 | } else if (X.info == OC_FNARG) { |
2365 | R.v = &fnargs[op1->l.i]; |
2366 | } else { |
2367 | syntax_error(EMSG_NOT_ARRAY); |
2368 | } |
2369 | |
2370 | if (op1->r.n) { |
2371 | clrvar(L.v); |
2372 | L.s = getvar_s(evaluate(op1->r.n, v1)); |
2373 | hash_remove(iamarray(R.v), L.s); |
2374 | } else { |
2375 | clear_array(iamarray(R.v)); |
2376 | } |
2377 | break; |
2378 | |
2379 | case XC( OC_NEWSOURCE ): |
2380 | g_progname = op->l.s; |
2381 | break; |
2382 | |
2383 | case XC( OC_RETURN ): |
2384 | copyvar(res, L.v); |
2385 | break; |
2386 | |
2387 | case XC( OC_NEXTFILE ): |
2388 | nextfile = TRUE; |
2389 | case XC( OC_NEXT ): |
2390 | nextrec = TRUE; |
2391 | case XC( OC_DONE ): |
2392 | clrvar(res); |
2393 | break; |
2394 | |
2395 | case XC( OC_EXIT ): |
2396 | awk_exit(L.d); |
2397 | |
2398 | /* -- recursive node type -- */ |
2399 | |
2400 | case XC( OC_VAR ): |
2401 | L.v = op->l.v; |
2402 | if (L.v == intvar[NF]) |
2403 | split_f0(); |
2404 | goto v_cont; |
2405 | |
2406 | case XC( OC_FNARG ): |
2407 | L.v = &fnargs[op->l.i]; |
2408 | v_cont: |
2409 | res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v; |
2410 | break; |
2411 | |
2412 | case XC( OC_IN ): |
2413 | setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0); |
2414 | break; |
2415 | |
2416 | case XC( OC_REGEXP ): |
2417 | op1 = op; |
2418 | L.s = getvar_s(intvar[F0]); |
2419 | goto re_cont; |
2420 | |
2421 | case XC( OC_MATCH ): |
2422 | op1 = op->r.n; |
2423 | re_cont: |
2424 | X.re = as_regex(op1, &sreg); |
2425 | R.i = regexec(X.re, L.s, 0, NULL, 0); |
2426 | if (X.re == &sreg) regfree(X.re); |
2427 | setvar_i(res, (R.i == 0) ^ (opn == '!')); |
2428 | break; |
2429 | |
2430 | case XC( OC_MOVE ): |
2431 | /* if source is a temporary string, jusk relink it to dest */ |
2432 | //Disabled: if R.v is numeric but happens to have cached R.v->string, |
2433 | //then L.v ends up being a string, which is wrong |
2434 | // if (R.v == v1+1 && R.v->string) { |
2435 | // res = setvar_p(L.v, R.v->string); |
2436 | // R.v->string = NULL; |
2437 | // } else { |
2438 | res = copyvar(L.v, R.v); |
2439 | // } |
2440 | break; |
2441 | |
2442 | case XC( OC_TERNARY ): |
2443 | if ((op->r.n->info & OPCLSMASK) != OC_COLON) |
2444 | syntax_error(EMSG_POSSIBLE_ERROR); |
2445 | res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res); |
2446 | break; |
2447 | |
2448 | case XC( OC_FUNC ): |
2449 | if (!op->r.f->body.first) |
2450 | syntax_error(EMSG_UNDEF_FUNC); |
2451 | |
2452 | X.v = R.v = nvalloc(op->r.f->nargs + 1); |
2453 | while (op1) { |
2454 | L.v = evaluate(nextarg(&op1), v1); |
2455 | copyvar(R.v, L.v); |
2456 | R.v->type |= VF_CHILD; |
2457 | R.v->x.parent = L.v; |
2458 | if (++R.v - X.v >= op->r.f->nargs) |
2459 | break; |
2460 | } |
2461 | |
2462 | R.v = fnargs; |
2463 | fnargs = X.v; |
2464 | |
2465 | L.s = g_progname; |
2466 | res = evaluate(op->r.f->body.first, res); |
2467 | g_progname = L.s; |
2468 | |
2469 | nvfree(fnargs); |
2470 | fnargs = R.v; |
2471 | break; |
2472 | |
2473 | case XC( OC_GETLINE ): |
2474 | case XC( OC_PGETLINE ): |
2475 | if (op1) { |
2476 | X.rsm = newfile(L.s); |
2477 | if (!X.rsm->F) { |
2478 | if ((opinfo & OPCLSMASK) == OC_PGETLINE) { |
2479 | X.rsm->F = popen(L.s, "r"); |
2480 | X.rsm->is_pipe = TRUE; |
2481 | } else { |
2482 | X.rsm->F = fopen_for_read(L.s); /* not xfopen! */ |
2483 | } |
2484 | } |
2485 | } else { |
2486 | if (!iF) iF = next_input_file(); |
2487 | X.rsm = iF; |
2488 | } |
2489 | |
2490 | if (!X.rsm->F) { |
2491 | setvar_i(intvar[ERRNO], errno); |
2492 | setvar_i(res, -1); |
2493 | break; |
2494 | } |
2495 | |
2496 | if (!op->r.n) |
2497 | R.v = intvar[F0]; |
2498 | |
2499 | L.i = awk_getline(X.rsm, R.v); |
2500 | if (L.i > 0) { |
2501 | if (!op1) { |
2502 | incvar(intvar[FNR]); |
2503 | incvar(intvar[NR]); |
2504 | } |
2505 | } |
2506 | setvar_i(res, L.i); |
2507 | break; |
2508 | |
2509 | /* simple builtins */ |
2510 | case XC( OC_FBLTIN ): |
2511 | switch (opn) { |
2512 | |
2513 | case F_in: |
2514 | R.d = (int)L.d; |
2515 | break; |
2516 | |
2517 | case F_rn: |
2518 | R.d = (double)rand() / (double)RAND_MAX; |
2519 | break; |
2520 | #if ENABLE_FEATURE_AWK_LIBM |
2521 | case F_co: |
2522 | R.d = cos(L.d); |
2523 | break; |
2524 | |
2525 | case F_ex: |
2526 | R.d = exp(L.d); |
2527 | break; |
2528 | |
2529 | case F_lg: |
2530 | R.d = log(L.d); |
2531 | break; |
2532 | |
2533 | case F_si: |
2534 | R.d = sin(L.d); |
2535 | break; |
2536 | |
2537 | case F_sq: |
2538 | R.d = sqrt(L.d); |
2539 | break; |
2540 | #else |
2541 | case F_co: |
2542 | case F_ex: |
2543 | case F_lg: |
2544 | case F_si: |
2545 | case F_sq: |
2546 | syntax_error(EMSG_NO_MATH); |
2547 | break; |
2548 | #endif |
2549 | case F_sr: |
2550 | R.d = (double)seed; |
2551 | seed = op1 ? (unsigned)L.d : (unsigned)time(NULL); |
2552 | srand(seed); |
2553 | break; |
2554 | |
2555 | case F_ti: |
2556 | R.d = time(NULL); |
2557 | break; |
2558 | |
2559 | case F_le: |
2560 | if (!op1) |
2561 | L.s = getvar_s(intvar[F0]); |
2562 | R.d = strlen(L.s); |
2563 | break; |
2564 | |
2565 | case F_sy: |
2566 | fflush_all(); |
2567 | R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s) |
2568 | ? (system(L.s) >> 8) : 0; |
2569 | break; |
2570 | |
2571 | case F_ff: |
2572 | if (!op1) |
2573 | fflush(stdout); |
2574 | else { |
2575 | if (L.s && *L.s) { |
2576 | X.rsm = newfile(L.s); |
2577 | fflush(X.rsm->F); |
2578 | } else { |
2579 | fflush_all(); |
2580 | } |
2581 | } |
2582 | break; |
2583 | |
2584 | case F_cl: |
2585 | X.rsm = (rstream *)hash_search(fdhash, L.s); |
2586 | if (X.rsm) { |
2587 | R.i = X.rsm->is_pipe ? pclose(X.rsm->F) : fclose(X.rsm->F); |
2588 | free(X.rsm->buffer); |
2589 | hash_remove(fdhash, L.s); |
2590 | } |
2591 | if (R.i != 0) |
2592 | setvar_i(intvar[ERRNO], errno); |
2593 | R.d = (double)R.i; |
2594 | break; |
2595 | } |
2596 | setvar_i(res, R.d); |
2597 | break; |
2598 | |
2599 | case XC( OC_BUILTIN ): |
2600 | res = exec_builtin(op, res); |
2601 | break; |
2602 | |
2603 | case XC( OC_SPRINTF ): |
2604 | setvar_p(res, awk_printf(op1)); |
2605 | break; |
2606 | |
2607 | case XC( OC_UNARY ): |
2608 | X.v = R.v; |
2609 | L.d = R.d = getvar_i(R.v); |
2610 | switch (opn) { |
2611 | case 'P': |
2612 | L.d = ++R.d; |
2613 | goto r_op_change; |
2614 | case 'p': |
2615 | R.d++; |
2616 | goto r_op_change; |
2617 | case 'M': |
2618 | L.d = --R.d; |
2619 | goto r_op_change; |
2620 | case 'm': |
2621 | R.d--; |
2622 | goto r_op_change; |
2623 | case '!': |
2624 | L.d = !istrue(X.v); |
2625 | break; |
2626 | case '-': |
2627 | L.d = -R.d; |
2628 | break; |
2629 | r_op_change: |
2630 | setvar_i(X.v, R.d); |
2631 | } |
2632 | setvar_i(res, L.d); |
2633 | break; |
2634 | |
2635 | case XC( OC_FIELD ): |
2636 | R.i = (int)getvar_i(R.v); |
2637 | if (R.i == 0) { |
2638 | res = intvar[F0]; |
2639 | } else { |
2640 | split_f0(); |
2641 | if (R.i > nfields) |
2642 | fsrealloc(R.i); |
2643 | res = &Fields[R.i - 1]; |
2644 | } |
2645 | break; |
2646 | |
2647 | /* concatenation (" ") and index joining (",") */ |
2648 | case XC( OC_CONCAT ): |
2649 | case XC( OC_COMMA ): |
2650 | opn = strlen(L.s) + strlen(R.s) + 2; |
2651 | X.s = xmalloc(opn); |
2652 | strcpy(X.s, L.s); |
2653 | if ((opinfo & OPCLSMASK) == OC_COMMA) { |
2654 | L.s = getvar_s(intvar[SUBSEP]); |
2655 | X.s = xrealloc(X.s, opn + strlen(L.s)); |
2656 | strcat(X.s, L.s); |
2657 | } |
2658 | strcat(X.s, R.s); |
2659 | setvar_p(res, X.s); |
2660 | break; |
2661 | |
2662 | case XC( OC_LAND ): |
2663 | setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0); |
2664 | break; |
2665 | |
2666 | case XC( OC_LOR ): |
2667 | setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n)); |
2668 | break; |
2669 | |
2670 | case XC( OC_BINARY ): |
2671 | case XC( OC_REPLACE ): |
2672 | R.d = getvar_i(R.v); |
2673 | switch (opn) { |
2674 | case '+': |
2675 | L.d += R.d; |
2676 | break; |
2677 | case '-': |
2678 | L.d -= R.d; |
2679 | break; |
2680 | case '*': |
2681 | L.d *= R.d; |
2682 | break; |
2683 | case '/': |
2684 | if (R.d == 0) |
2685 | syntax_error(EMSG_DIV_BY_ZERO); |
2686 | L.d /= R.d; |
2687 | break; |
2688 | case '&': |
2689 | #if ENABLE_FEATURE_AWK_LIBM |
2690 | L.d = pow(L.d, R.d); |
2691 | #else |
2692 | syntax_error(EMSG_NO_MATH); |
2693 | #endif |
2694 | break; |
2695 | case '%': |
2696 | if (R.d == 0) |
2697 | syntax_error(EMSG_DIV_BY_ZERO); |
2698 | L.d -= (int)(L.d / R.d) * R.d; |
2699 | break; |
2700 | } |
2701 | res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : X.v, L.d); |
2702 | break; |
2703 | |
2704 | case XC( OC_COMPARE ): |
2705 | if (is_numeric(L.v) && is_numeric(R.v)) { |
2706 | L.d = getvar_i(L.v) - getvar_i(R.v); |
2707 | } else { |
2708 | L.s = getvar_s(L.v); |
2709 | R.s = getvar_s(R.v); |
2710 | L.d = icase ? strcasecmp(L.s, R.s) : strcmp(L.s, R.s); |
2711 | } |
2712 | switch (opn & 0xfe) { |
2713 | case 0: |
2714 | R.i = (L.d > 0); |
2715 | break; |
2716 | case 2: |
2717 | R.i = (L.d >= 0); |
2718 | break; |
2719 | case 4: |
2720 | R.i = (L.d == 0); |
2721 | break; |
2722 | } |
2723 | setvar_i(res, (opn & 1 ? R.i : !R.i) ? 1 : 0); |
2724 | break; |
2725 | |
2726 | default: |
2727 | syntax_error(EMSG_POSSIBLE_ERROR); |
2728 | } |
2729 | if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS) |
2730 | op = op->a.n; |
2731 | if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS) |
2732 | break; |
2733 | if (nextrec) |
2734 | break; |
2735 | } |
2736 | nvfree(v1); |
2737 | return res; |
2738 | #undef fnargs |
2739 | #undef seed |
2740 | #undef sreg |
2741 | } |
2742 | |
2743 | |
2744 | /* -------- main & co. -------- */ |
2745 | |
2746 | static int awk_exit(int r) |
2747 | { |
2748 | var tv; |
2749 | unsigned i; |
2750 | hash_item *hi; |
2751 | |
2752 | zero_out_var(&tv); |
2753 | |
2754 | if (!exiting) { |
2755 | exiting = TRUE; |
2756 | nextrec = FALSE; |
2757 | evaluate(endseq.first, &tv); |
2758 | } |
2759 | |
2760 | /* waiting for children */ |
2761 | for (i = 0; i < fdhash->csize; i++) { |
2762 | hi = fdhash->items[i]; |
2763 | while (hi) { |
2764 | if (hi->data.rs.F && hi->data.rs.is_pipe) |
2765 | pclose(hi->data.rs.F); |
2766 | hi = hi->next; |
2767 | } |
2768 | } |
2769 | |
2770 | exit(r); |
2771 | } |
2772 | |
2773 | /* if expr looks like "var=value", perform assignment and return 1, |
2774 | * otherwise return 0 */ |
2775 | static int is_assignment(const char *expr) |
2776 | { |
2777 | char *exprc, *s, *s0, *s1; |
2778 | |
2779 | exprc = xstrdup(expr); |
2780 | if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) { |
2781 | free(exprc); |
2782 | return FALSE; |
2783 | } |
2784 | |
2785 | *(s++) = '\0'; |
2786 | s0 = s1 = s; |
2787 | while (*s) |
2788 | *(s1++) = nextchar(&s); |
2789 | |
2790 | *s1 = '\0'; |
2791 | setvar_u(newvar(exprc), s0); |
2792 | free(exprc); |
2793 | return TRUE; |
2794 | } |
2795 | |
2796 | /* switch to next input file */ |
2797 | static rstream *next_input_file(void) |
2798 | { |
2799 | #define rsm (G.next_input_file__rsm) |
2800 | #define files_happen (G.next_input_file__files_happen) |
2801 | |
2802 | FILE *F = NULL; |
2803 | const char *fname, *ind; |
2804 | |
2805 | if (rsm.F) |
2806 | fclose(rsm.F); |
2807 | rsm.F = NULL; |
2808 | rsm.pos = rsm.adv = 0; |
2809 | |
2810 | do { |
2811 | if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) { |
2812 | if (files_happen) |
2813 | return NULL; |
2814 | fname = "-"; |
2815 | F = stdin; |
2816 | } else { |
2817 | ind = getvar_s(incvar(intvar[ARGIND])); |
2818 | fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind)); |
2819 | if (fname && *fname && !is_assignment(fname)) |
2820 | F = xfopen_stdin(fname); |
2821 | } |
2822 | } while (!F); |
2823 | |
2824 | files_happen = TRUE; |
2825 | setvar_s(intvar[FILENAME], fname); |
2826 | rsm.F = F; |
2827 | return &rsm; |
2828 | #undef rsm |
2829 | #undef files_happen |
2830 | } |
2831 | |
2832 | int awk_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
2833 | int awk_main(int argc, char **argv) |
2834 | { |
2835 | unsigned opt; |
2836 | char *opt_F, *opt_W; |
2837 | llist_t *list_v = NULL; |
2838 | llist_t *list_f = NULL; |
2839 | int i, j; |
2840 | var *v; |
2841 | var tv; |
2842 | char **envp; |
2843 | char *vnames = (char *)vNames; /* cheat */ |
2844 | char *vvalues = (char *)vValues; |
2845 | |
2846 | INIT_G(); |
2847 | |
2848 | /* Undo busybox.c, or else strtod may eat ','! This breaks parsing: |
2849 | * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */ |
2850 | if (ENABLE_LOCALE_SUPPORT) |
2851 | setlocale(LC_NUMERIC, "C"); |
2852 | |
2853 | zero_out_var(&tv); |
2854 | |
2855 | /* allocate global buffer */ |
2856 | g_buf = xmalloc(MAXVARFMT + 1); |
2857 | |
2858 | vhash = hash_init(); |
2859 | ahash = hash_init(); |
2860 | fdhash = hash_init(); |
2861 | fnhash = hash_init(); |
2862 | |
2863 | /* initialize variables */ |
2864 | for (i = 0; *vnames; i++) { |
2865 | intvar[i] = v = newvar(nextword(&vnames)); |
2866 | if (*vvalues != '\377') |
2867 | setvar_s(v, nextword(&vvalues)); |
2868 | else |
2869 | setvar_i(v, 0); |
2870 | |
2871 | if (*vnames == '*') { |
2872 | v->type |= VF_SPECIAL; |
2873 | vnames++; |
2874 | } |
2875 | } |
2876 | |
2877 | handle_special(intvar[FS]); |
2878 | handle_special(intvar[RS]); |
2879 | |
2880 | newfile("/dev/stdin")->F = stdin; |
2881 | newfile("/dev/stdout")->F = stdout; |
2882 | newfile("/dev/stderr")->F = stderr; |
2883 | |
2884 | /* Huh, people report that sometimes environ is NULL. Oh well. */ |
2885 | if (environ) for (envp = environ; *envp; envp++) { |
2886 | /* environ is writable, thus we don't strdup it needlessly */ |
2887 | char *s = *envp; |
2888 | char *s1 = strchr(s, '='); |
2889 | if (s1) { |
2890 | *s1 = '\0'; |
2891 | /* Both findvar and setvar_u take const char* |
2892 | * as 2nd arg -> environment is not trashed */ |
2893 | setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1); |
2894 | *s1 = '='; |
2895 | } |
2896 | } |
2897 | opt_complementary = "v::f::"; /* -v and -f can occur multiple times */ |
2898 | opt = getopt32(argv, "F:v:f:W:", &opt_F, &list_v, &list_f, &opt_W); |
2899 | argv += optind; |
2900 | argc -= optind; |
2901 | if (opt & 0x1) |
2902 | setvar_s(intvar[FS], opt_F); // -F |
2903 | while (list_v) { /* -v */ |
2904 | if (!is_assignment(llist_pop(&list_v))) |
2905 | bb_show_usage(); |
2906 | } |
2907 | if (list_f) { /* -f */ |
2908 | do { |
2909 | char *s = NULL; |
2910 | FILE *from_file; |
2911 | |
2912 | g_progname = llist_pop(&list_f); |
2913 | from_file = xfopen_stdin(g_progname); |
2914 | /* one byte is reserved for some trick in next_token */ |
2915 | for (i = j = 1; j > 0; i += j) { |
2916 | s = xrealloc(s, i + 4096); |
2917 | j = fread(s + i, 1, 4094, from_file); |
2918 | } |
2919 | s[i] = '\0'; |
2920 | fclose(from_file); |
2921 | parse_program(s + 1); |
2922 | free(s); |
2923 | } while (list_f); |
2924 | argc++; |
2925 | } else { // no -f: take program from 1st parameter |
2926 | if (!argc) |
2927 | bb_show_usage(); |
2928 | g_progname = "cmd. line"; |
2929 | parse_program(*argv++); |
2930 | } |
2931 | if (opt & 0x8) // -W |
2932 | bb_error_msg("warning: unrecognized option '-W %s' ignored", opt_W); |
2933 | |
2934 | /* fill in ARGV array */ |
2935 | setvar_i(intvar[ARGC], argc); |
2936 | setari_u(intvar[ARGV], 0, "awk"); |
2937 | i = 0; |
2938 | while (*argv) |
2939 | setari_u(intvar[ARGV], ++i, *argv++); |
2940 | |
2941 | evaluate(beginseq.first, &tv); |
2942 | if (!mainseq.first && !endseq.first) |
2943 | awk_exit(EXIT_SUCCESS); |
2944 | |
2945 | /* input file could already be opened in BEGIN block */ |
2946 | if (!iF) |
2947 | iF = next_input_file(); |
2948 | |
2949 | /* passing through input files */ |
2950 | while (iF) { |
2951 | nextfile = FALSE; |
2952 | setvar_i(intvar[FNR], 0); |
2953 | |
2954 | while ((i = awk_getline(iF, intvar[F0])) > 0) { |
2955 | nextrec = FALSE; |
2956 | incvar(intvar[NR]); |
2957 | incvar(intvar[FNR]); |
2958 | evaluate(mainseq.first, &tv); |
2959 | |
2960 | if (nextfile) |
2961 | break; |
2962 | } |
2963 | |
2964 | if (i < 0) |
2965 | syntax_error(strerror(errno)); |
2966 | |
2967 | iF = next_input_file(); |
2968 | } |
2969 | |
2970 | awk_exit(EXIT_SUCCESS); |
2971 | /*return 0;*/ |
2972 | } |