Contents of /trunk/mkinitrd-magellan/busybox/editors/awk.c
Parent Directory | Revision Log
Revision 1123 -
(show annotations)
(download)
Wed Aug 18 21:56:57 2010 UTC (14 years, 1 month ago) by niro
File MIME type: text/plain
File size: 68598 byte(s)
Wed Aug 18 21:56:57 2010 UTC (14 years, 1 month ago) by niro
File MIME type: text/plain
File size: 68598 byte(s)
-updated to busybox-1.17.1
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 | /* If you comment out one of these below, it will be #defined later |
18 | * to perform debug printfs to stderr: */ |
19 | #define debug_printf_walker(...) do {} while (0) |
20 | |
21 | #ifndef debug_printf_walker |
22 | # define debug_printf_walker(...) (fprintf(stderr, __VA_ARGS__)) |
23 | #endif |
24 | |
25 | |
26 | |
27 | #define MAXVARFMT 240 |
28 | #define MINNVBLOCK 64 |
29 | |
30 | /* variable flags */ |
31 | #define VF_NUMBER 0x0001 /* 1 = primary type is number */ |
32 | #define VF_ARRAY 0x0002 /* 1 = it's an array */ |
33 | |
34 | #define VF_CACHED 0x0100 /* 1 = num/str value has cached str/num eq */ |
35 | #define VF_USER 0x0200 /* 1 = user input (may be numeric string) */ |
36 | #define VF_SPECIAL 0x0400 /* 1 = requires extra handling when changed */ |
37 | #define VF_WALK 0x0800 /* 1 = variable has alloc'd x.walker list */ |
38 | #define VF_FSTR 0x1000 /* 1 = var::string points to fstring buffer */ |
39 | #define VF_CHILD 0x2000 /* 1 = function arg; x.parent points to source */ |
40 | #define VF_DIRTY 0x4000 /* 1 = variable was set explicitly */ |
41 | |
42 | /* these flags are static, don't change them when value is changed */ |
43 | #define VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY) |
44 | |
45 | typedef struct walker_list { |
46 | char *end; |
47 | char *cur; |
48 | struct walker_list *prev; |
49 | char wbuf[1]; |
50 | } walker_list; |
51 | |
52 | /* Variable */ |
53 | typedef struct var_s { |
54 | unsigned type; /* flags */ |
55 | double number; |
56 | char *string; |
57 | union { |
58 | int aidx; /* func arg idx (for compilation stage) */ |
59 | struct xhash_s *array; /* array ptr */ |
60 | struct var_s *parent; /* for func args, ptr to actual parameter */ |
61 | walker_list *walker; /* list of array elements (for..in) */ |
62 | } x; |
63 | } var; |
64 | |
65 | /* Node chain (pattern-action chain, BEGIN, END, function bodies) */ |
66 | typedef struct chain_s { |
67 | struct node_s *first; |
68 | struct node_s *last; |
69 | const char *programname; |
70 | } chain; |
71 | |
72 | /* Function */ |
73 | typedef struct func_s { |
74 | unsigned nargs; |
75 | struct chain_s body; |
76 | } func; |
77 | |
78 | /* I/O stream */ |
79 | typedef struct rstream_s { |
80 | FILE *F; |
81 | char *buffer; |
82 | int adv; |
83 | int size; |
84 | int pos; |
85 | smallint is_pipe; |
86 | } rstream; |
87 | |
88 | typedef struct hash_item_s { |
89 | union { |
90 | struct var_s v; /* variable/array hash */ |
91 | struct rstream_s rs; /* redirect streams hash */ |
92 | struct func_s f; /* functions hash */ |
93 | } data; |
94 | struct hash_item_s *next; /* next in chain */ |
95 | char name[1]; /* really it's longer */ |
96 | } hash_item; |
97 | |
98 | typedef struct xhash_s { |
99 | unsigned nel; /* num of elements */ |
100 | unsigned csize; /* current hash size */ |
101 | unsigned nprime; /* next hash size in PRIMES[] */ |
102 | unsigned glen; /* summary length of item names */ |
103 | struct hash_item_s **items; |
104 | } xhash; |
105 | |
106 | /* Tree node */ |
107 | typedef struct node_s { |
108 | uint32_t info; |
109 | unsigned lineno; |
110 | union { |
111 | struct node_s *n; |
112 | var *v; |
113 | int aidx; |
114 | char *new_progname; |
115 | regex_t *re; |
116 | } l; |
117 | union { |
118 | struct node_s *n; |
119 | regex_t *ire; |
120 | func *f; |
121 | } r; |
122 | union { |
123 | struct node_s *n; |
124 | } a; |
125 | } node; |
126 | |
127 | /* Block of temporary variables */ |
128 | typedef struct nvblock_s { |
129 | int size; |
130 | var *pos; |
131 | struct nvblock_s *prev; |
132 | struct nvblock_s *next; |
133 | var nv[]; |
134 | } nvblock; |
135 | |
136 | typedef struct tsplitter_s { |
137 | node n; |
138 | regex_t re[2]; |
139 | } tsplitter; |
140 | |
141 | /* simple token classes */ |
142 | /* Order and hex values are very important!!! See next_token() */ |
143 | #define TC_SEQSTART 1 /* ( */ |
144 | #define TC_SEQTERM (1 << 1) /* ) */ |
145 | #define TC_REGEXP (1 << 2) /* /.../ */ |
146 | #define TC_OUTRDR (1 << 3) /* | > >> */ |
147 | #define TC_UOPPOST (1 << 4) /* unary postfix operator */ |
148 | #define TC_UOPPRE1 (1 << 5) /* unary prefix operator */ |
149 | #define TC_BINOPX (1 << 6) /* two-opnd operator */ |
150 | #define TC_IN (1 << 7) |
151 | #define TC_COMMA (1 << 8) |
152 | #define TC_PIPE (1 << 9) /* input redirection pipe */ |
153 | #define TC_UOPPRE2 (1 << 10) /* unary prefix operator */ |
154 | #define TC_ARRTERM (1 << 11) /* ] */ |
155 | #define TC_GRPSTART (1 << 12) /* { */ |
156 | #define TC_GRPTERM (1 << 13) /* } */ |
157 | #define TC_SEMICOL (1 << 14) |
158 | #define TC_NEWLINE (1 << 15) |
159 | #define TC_STATX (1 << 16) /* ctl statement (for, next...) */ |
160 | #define TC_WHILE (1 << 17) |
161 | #define TC_ELSE (1 << 18) |
162 | #define TC_BUILTIN (1 << 19) |
163 | #define TC_GETLINE (1 << 20) |
164 | #define TC_FUNCDECL (1 << 21) /* `function' `func' */ |
165 | #define TC_BEGIN (1 << 22) |
166 | #define TC_END (1 << 23) |
167 | #define TC_EOF (1 << 24) |
168 | #define TC_VARIABLE (1 << 25) |
169 | #define TC_ARRAY (1 << 26) |
170 | #define TC_FUNCTION (1 << 27) |
171 | #define TC_STRING (1 << 28) |
172 | #define TC_NUMBER (1 << 29) |
173 | |
174 | #define TC_UOPPRE (TC_UOPPRE1 | TC_UOPPRE2) |
175 | |
176 | /* combined token classes */ |
177 | #define TC_BINOP (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN) |
178 | #define TC_UNARYOP (TC_UOPPRE | TC_UOPPOST) |
179 | #define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION \ |
180 | | TC_BUILTIN | TC_GETLINE | TC_SEQSTART | TC_STRING | TC_NUMBER) |
181 | |
182 | #define TC_STATEMNT (TC_STATX | TC_WHILE) |
183 | #define TC_OPTERM (TC_SEMICOL | TC_NEWLINE) |
184 | |
185 | /* word tokens, cannot mean something else if not expected */ |
186 | #define TC_WORD (TC_IN | TC_STATEMNT | TC_ELSE | TC_BUILTIN \ |
187 | | TC_GETLINE | TC_FUNCDECL | TC_BEGIN | TC_END) |
188 | |
189 | /* discard newlines after these */ |
190 | #define TC_NOTERM (TC_COMMA | TC_GRPSTART | TC_GRPTERM \ |
191 | | TC_BINOP | TC_OPTERM) |
192 | |
193 | /* what can expression begin with */ |
194 | #define TC_OPSEQ (TC_OPERAND | TC_UOPPRE | TC_REGEXP) |
195 | /* what can group begin with */ |
196 | #define TC_GRPSEQ (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART) |
197 | |
198 | /* if previous token class is CONCAT1 and next is CONCAT2, concatenation */ |
199 | /* operator is inserted between them */ |
200 | #define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM \ |
201 | | TC_STRING | TC_NUMBER | TC_UOPPOST) |
202 | #define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE) |
203 | |
204 | #define OF_RES1 0x010000 |
205 | #define OF_RES2 0x020000 |
206 | #define OF_STR1 0x040000 |
207 | #define OF_STR2 0x080000 |
208 | #define OF_NUM1 0x100000 |
209 | #define OF_CHECKED 0x200000 |
210 | |
211 | /* combined operator flags */ |
212 | #define xx 0 |
213 | #define xV OF_RES2 |
214 | #define xS (OF_RES2 | OF_STR2) |
215 | #define Vx OF_RES1 |
216 | #define VV (OF_RES1 | OF_RES2) |
217 | #define Nx (OF_RES1 | OF_NUM1) |
218 | #define NV (OF_RES1 | OF_NUM1 | OF_RES2) |
219 | #define Sx (OF_RES1 | OF_STR1) |
220 | #define SV (OF_RES1 | OF_STR1 | OF_RES2) |
221 | #define SS (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2) |
222 | |
223 | #define OPCLSMASK 0xFF00 |
224 | #define OPNMASK 0x007F |
225 | |
226 | /* operator priority is a highest byte (even: r->l, odd: l->r grouping) |
227 | * For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1, |
228 | * n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string |
229 | */ |
230 | #define P(x) (x << 24) |
231 | #define PRIMASK 0x7F000000 |
232 | #define PRIMASK2 0x7E000000 |
233 | |
234 | /* Operation classes */ |
235 | |
236 | #define SHIFT_TIL_THIS 0x0600 |
237 | #define RECUR_FROM_THIS 0x1000 |
238 | |
239 | enum { |
240 | OC_DELETE = 0x0100, OC_EXEC = 0x0200, OC_NEWSOURCE = 0x0300, |
241 | OC_PRINT = 0x0400, OC_PRINTF = 0x0500, OC_WALKINIT = 0x0600, |
242 | |
243 | OC_BR = 0x0700, OC_BREAK = 0x0800, OC_CONTINUE = 0x0900, |
244 | OC_EXIT = 0x0a00, OC_NEXT = 0x0b00, OC_NEXTFILE = 0x0c00, |
245 | OC_TEST = 0x0d00, OC_WALKNEXT = 0x0e00, |
246 | |
247 | OC_BINARY = 0x1000, OC_BUILTIN = 0x1100, OC_COLON = 0x1200, |
248 | OC_COMMA = 0x1300, OC_COMPARE = 0x1400, OC_CONCAT = 0x1500, |
249 | OC_FBLTIN = 0x1600, OC_FIELD = 0x1700, OC_FNARG = 0x1800, |
250 | OC_FUNC = 0x1900, OC_GETLINE = 0x1a00, OC_IN = 0x1b00, |
251 | OC_LAND = 0x1c00, OC_LOR = 0x1d00, OC_MATCH = 0x1e00, |
252 | OC_MOVE = 0x1f00, OC_PGETLINE = 0x2000, OC_REGEXP = 0x2100, |
253 | OC_REPLACE = 0x2200, OC_RETURN = 0x2300, OC_SPRINTF = 0x2400, |
254 | OC_TERNARY = 0x2500, OC_UNARY = 0x2600, OC_VAR = 0x2700, |
255 | OC_DONE = 0x2800, |
256 | |
257 | ST_IF = 0x3000, ST_DO = 0x3100, ST_FOR = 0x3200, |
258 | ST_WHILE = 0x3300 |
259 | }; |
260 | |
261 | /* simple builtins */ |
262 | enum { |
263 | F_in, F_rn, F_co, F_ex, F_lg, F_si, F_sq, F_sr, |
264 | F_ti, F_le, F_sy, F_ff, F_cl |
265 | }; |
266 | |
267 | /* builtins */ |
268 | enum { |
269 | B_a2, B_ix, B_ma, B_sp, B_ss, B_ti, B_mt, B_lo, B_up, |
270 | B_ge, B_gs, B_su, |
271 | B_an, B_co, B_ls, B_or, B_rs, B_xo, |
272 | }; |
273 | |
274 | /* tokens and their corresponding info values */ |
275 | |
276 | #define NTC "\377" /* switch to next token class (tc<<1) */ |
277 | #define NTCC '\377' |
278 | |
279 | #define OC_B OC_BUILTIN |
280 | |
281 | static const char tokenlist[] ALIGN1 = |
282 | "\1(" NTC |
283 | "\1)" NTC |
284 | "\1/" NTC /* REGEXP */ |
285 | "\2>>" "\1>" "\1|" NTC /* OUTRDR */ |
286 | "\2++" "\2--" NTC /* UOPPOST */ |
287 | "\2++" "\2--" "\1$" NTC /* UOPPRE1 */ |
288 | "\2==" "\1=" "\2+=" "\2-=" /* BINOPX */ |
289 | "\2*=" "\2/=" "\2%=" "\2^=" |
290 | "\1+" "\1-" "\3**=" "\2**" |
291 | "\1/" "\1%" "\1^" "\1*" |
292 | "\2!=" "\2>=" "\2<=" "\1>" |
293 | "\1<" "\2!~" "\1~" "\2&&" |
294 | "\2||" "\1?" "\1:" NTC |
295 | "\2in" NTC |
296 | "\1," NTC |
297 | "\1|" NTC |
298 | "\1+" "\1-" "\1!" NTC /* UOPPRE2 */ |
299 | "\1]" NTC |
300 | "\1{" NTC |
301 | "\1}" NTC |
302 | "\1;" NTC |
303 | "\1\n" NTC |
304 | "\2if" "\2do" "\3for" "\5break" /* STATX */ |
305 | "\10continue" "\6delete" "\5print" |
306 | "\6printf" "\4next" "\10nextfile" |
307 | "\6return" "\4exit" NTC |
308 | "\5while" NTC |
309 | "\4else" NTC |
310 | |
311 | "\3and" "\5compl" "\6lshift" "\2or" |
312 | "\6rshift" "\3xor" |
313 | "\5close" "\6system" "\6fflush" "\5atan2" /* BUILTIN */ |
314 | "\3cos" "\3exp" "\3int" "\3log" |
315 | "\4rand" "\3sin" "\4sqrt" "\5srand" |
316 | "\6gensub" "\4gsub" "\5index" "\6length" |
317 | "\5match" "\5split" "\7sprintf" "\3sub" |
318 | "\6substr" "\7systime" "\10strftime" "\6mktime" |
319 | "\7tolower" "\7toupper" NTC |
320 | "\7getline" NTC |
321 | "\4func" "\10function" NTC |
322 | "\5BEGIN" NTC |
323 | "\3END" "\0" |
324 | ; |
325 | |
326 | static const uint32_t tokeninfo[] = { |
327 | 0, |
328 | 0, |
329 | OC_REGEXP, |
330 | xS|'a', xS|'w', xS|'|', |
331 | OC_UNARY|xV|P(9)|'p', OC_UNARY|xV|P(9)|'m', |
332 | OC_UNARY|xV|P(9)|'P', OC_UNARY|xV|P(9)|'M', |
333 | OC_FIELD|xV|P(5), |
334 | OC_COMPARE|VV|P(39)|5, OC_MOVE|VV|P(74), |
335 | OC_REPLACE|NV|P(74)|'+', OC_REPLACE|NV|P(74)|'-', |
336 | OC_REPLACE|NV|P(74)|'*', OC_REPLACE|NV|P(74)|'/', |
337 | OC_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&', |
338 | OC_BINARY|NV|P(29)|'+', OC_BINARY|NV|P(29)|'-', |
339 | OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&', |
340 | OC_BINARY|NV|P(25)|'/', OC_BINARY|NV|P(25)|'%', |
341 | OC_BINARY|NV|P(15)|'&', OC_BINARY|NV|P(25)|'*', |
342 | OC_COMPARE|VV|P(39)|4, OC_COMPARE|VV|P(39)|3, |
343 | OC_COMPARE|VV|P(39)|0, OC_COMPARE|VV|P(39)|1, |
344 | OC_COMPARE|VV|P(39)|2, OC_MATCH|Sx|P(45)|'!', |
345 | OC_MATCH|Sx|P(45)|'~', OC_LAND|Vx|P(55), |
346 | OC_LOR|Vx|P(59), OC_TERNARY|Vx|P(64)|'?', |
347 | OC_COLON|xx|P(67)|':', |
348 | OC_IN|SV|P(49), |
349 | OC_COMMA|SS|P(80), |
350 | OC_PGETLINE|SV|P(37), |
351 | OC_UNARY|xV|P(19)|'+', OC_UNARY|xV|P(19)|'-', |
352 | OC_UNARY|xV|P(19)|'!', |
353 | 0, |
354 | 0, |
355 | 0, |
356 | 0, |
357 | 0, |
358 | ST_IF, ST_DO, ST_FOR, OC_BREAK, |
359 | OC_CONTINUE, OC_DELETE|Vx, OC_PRINT, |
360 | OC_PRINTF, OC_NEXT, OC_NEXTFILE, |
361 | OC_RETURN|Vx, OC_EXIT|Nx, |
362 | ST_WHILE, |
363 | 0, |
364 | |
365 | OC_B|B_an|P(0x83), OC_B|B_co|P(0x41), OC_B|B_ls|P(0x83), OC_B|B_or|P(0x83), |
366 | OC_B|B_rs|P(0x83), OC_B|B_xo|P(0x83), |
367 | OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83), |
368 | OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg, |
369 | OC_FBLTIN|F_rn, OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr, |
370 | OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), OC_FBLTIN|Sx|F_le, |
371 | OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF, OC_B|B_su|P(0xb6), |
372 | OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b), OC_B|B_mt|P(0x0b), |
373 | OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49), |
374 | OC_GETLINE|SV|P(0), |
375 | 0, 0, |
376 | 0, |
377 | 0 |
378 | }; |
379 | |
380 | /* internal variable names and their initial values */ |
381 | /* asterisk marks SPECIAL vars; $ is just no-named Field0 */ |
382 | enum { |
383 | CONVFMT, OFMT, FS, OFS, |
384 | ORS, RS, RT, FILENAME, |
385 | SUBSEP, F0, ARGIND, ARGC, |
386 | ARGV, ERRNO, FNR, NR, |
387 | NF, IGNORECASE, ENVIRON, NUM_INTERNAL_VARS |
388 | }; |
389 | |
390 | static const char vNames[] ALIGN1 = |
391 | "CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0" |
392 | "ORS\0" "RS\0*" "RT\0" "FILENAME\0" |
393 | "SUBSEP\0" "$\0*" "ARGIND\0" "ARGC\0" |
394 | "ARGV\0" "ERRNO\0" "FNR\0" "NR\0" |
395 | "NF\0*" "IGNORECASE\0*" "ENVIRON\0" "\0"; |
396 | |
397 | static const char vValues[] ALIGN1 = |
398 | "%.6g\0" "%.6g\0" " \0" " \0" |
399 | "\n\0" "\n\0" "\0" "\0" |
400 | "\034\0" "\0" "\377"; |
401 | |
402 | /* hash size may grow to these values */ |
403 | #define FIRST_PRIME 61 |
404 | static const uint16_t PRIMES[] ALIGN2 = { 251, 1021, 4093, 16381, 65521 }; |
405 | |
406 | |
407 | /* Globals. Split in two parts so that first one is addressed |
408 | * with (mostly short) negative offsets. |
409 | * NB: it's unsafe to put members of type "double" |
410 | * into globals2 (gcc may fail to align them). |
411 | */ |
412 | struct globals { |
413 | double t_double; |
414 | chain beginseq, mainseq, endseq; |
415 | chain *seq; |
416 | node *break_ptr, *continue_ptr; |
417 | rstream *iF; |
418 | xhash *vhash, *ahash, *fdhash, *fnhash; |
419 | const char *g_progname; |
420 | int g_lineno; |
421 | int nfields; |
422 | int maxfields; /* used in fsrealloc() only */ |
423 | var *Fields; |
424 | nvblock *g_cb; |
425 | char *g_pos; |
426 | char *g_buf; |
427 | smallint icase; |
428 | smallint exiting; |
429 | smallint nextrec; |
430 | smallint nextfile; |
431 | smallint is_f0_split; |
432 | }; |
433 | struct globals2 { |
434 | uint32_t t_info; /* often used */ |
435 | uint32_t t_tclass; |
436 | char *t_string; |
437 | int t_lineno; |
438 | int t_rollback; |
439 | |
440 | var *intvar[NUM_INTERNAL_VARS]; /* often used */ |
441 | |
442 | /* former statics from various functions */ |
443 | char *split_f0__fstrings; |
444 | |
445 | uint32_t next_token__save_tclass; |
446 | uint32_t next_token__save_info; |
447 | uint32_t next_token__ltclass; |
448 | smallint next_token__concat_inserted; |
449 | |
450 | smallint next_input_file__files_happen; |
451 | rstream next_input_file__rsm; |
452 | |
453 | var *evaluate__fnargs; |
454 | unsigned evaluate__seed; |
455 | regex_t evaluate__sreg; |
456 | |
457 | var ptest__v; |
458 | |
459 | tsplitter exec_builtin__tspl; |
460 | |
461 | /* biggest and least used members go last */ |
462 | tsplitter fsplitter, rsplitter; |
463 | }; |
464 | #define G1 (ptr_to_globals[-1]) |
465 | #define G (*(struct globals2 *)ptr_to_globals) |
466 | /* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */ |
467 | /*char G1size[sizeof(G1)]; - 0x74 */ |
468 | /*char Gsize[sizeof(G)]; - 0x1c4 */ |
469 | /* Trying to keep most of members accessible with short offsets: */ |
470 | /*char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */ |
471 | #define t_double (G1.t_double ) |
472 | #define beginseq (G1.beginseq ) |
473 | #define mainseq (G1.mainseq ) |
474 | #define endseq (G1.endseq ) |
475 | #define seq (G1.seq ) |
476 | #define break_ptr (G1.break_ptr ) |
477 | #define continue_ptr (G1.continue_ptr) |
478 | #define iF (G1.iF ) |
479 | #define vhash (G1.vhash ) |
480 | #define ahash (G1.ahash ) |
481 | #define fdhash (G1.fdhash ) |
482 | #define fnhash (G1.fnhash ) |
483 | #define g_progname (G1.g_progname ) |
484 | #define g_lineno (G1.g_lineno ) |
485 | #define nfields (G1.nfields ) |
486 | #define maxfields (G1.maxfields ) |
487 | #define Fields (G1.Fields ) |
488 | #define g_cb (G1.g_cb ) |
489 | #define g_pos (G1.g_pos ) |
490 | #define g_buf (G1.g_buf ) |
491 | #define icase (G1.icase ) |
492 | #define exiting (G1.exiting ) |
493 | #define nextrec (G1.nextrec ) |
494 | #define nextfile (G1.nextfile ) |
495 | #define is_f0_split (G1.is_f0_split ) |
496 | #define t_info (G.t_info ) |
497 | #define t_tclass (G.t_tclass ) |
498 | #define t_string (G.t_string ) |
499 | #define t_lineno (G.t_lineno ) |
500 | #define t_rollback (G.t_rollback ) |
501 | #define intvar (G.intvar ) |
502 | #define fsplitter (G.fsplitter ) |
503 | #define rsplitter (G.rsplitter ) |
504 | #define INIT_G() do { \ |
505 | SET_PTR_TO_GLOBALS((char*)xzalloc(sizeof(G1)+sizeof(G)) + sizeof(G1)); \ |
506 | G.next_token__ltclass = TC_OPTERM; \ |
507 | G.evaluate__seed = 1; \ |
508 | } while (0) |
509 | |
510 | |
511 | /* function prototypes */ |
512 | static void handle_special(var *); |
513 | static node *parse_expr(uint32_t); |
514 | static void chain_group(void); |
515 | static var *evaluate(node *, var *); |
516 | static rstream *next_input_file(void); |
517 | static int fmt_num(char *, int, const char *, double, int); |
518 | static int awk_exit(int) NORETURN; |
519 | |
520 | /* ---- error handling ---- */ |
521 | |
522 | static const char EMSG_INTERNAL_ERROR[] ALIGN1 = "Internal error"; |
523 | static const char EMSG_UNEXP_EOS[] ALIGN1 = "Unexpected end of string"; |
524 | static const char EMSG_UNEXP_TOKEN[] ALIGN1 = "Unexpected token"; |
525 | static const char EMSG_DIV_BY_ZERO[] ALIGN1 = "Division by zero"; |
526 | static const char EMSG_INV_FMT[] ALIGN1 = "Invalid format specifier"; |
527 | static const char EMSG_TOO_FEW_ARGS[] ALIGN1 = "Too few arguments for builtin"; |
528 | static const char EMSG_NOT_ARRAY[] ALIGN1 = "Not an array"; |
529 | static const char EMSG_POSSIBLE_ERROR[] ALIGN1 = "Possible syntax error"; |
530 | static const char EMSG_UNDEF_FUNC[] ALIGN1 = "Call to undefined function"; |
531 | #if !ENABLE_FEATURE_AWK_LIBM |
532 | static const char EMSG_NO_MATH[] ALIGN1 = "Math support is not compiled in"; |
533 | #endif |
534 | |
535 | static void zero_out_var(var *vp) |
536 | { |
537 | memset(vp, 0, sizeof(*vp)); |
538 | } |
539 | |
540 | static void syntax_error(const char *message) NORETURN; |
541 | static void syntax_error(const char *message) |
542 | { |
543 | bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message); |
544 | } |
545 | |
546 | /* ---- hash stuff ---- */ |
547 | |
548 | static unsigned hashidx(const char *name) |
549 | { |
550 | unsigned idx = 0; |
551 | |
552 | while (*name) |
553 | idx = *name++ + (idx << 6) - idx; |
554 | return idx; |
555 | } |
556 | |
557 | /* create new hash */ |
558 | static xhash *hash_init(void) |
559 | { |
560 | xhash *newhash; |
561 | |
562 | newhash = xzalloc(sizeof(*newhash)); |
563 | newhash->csize = FIRST_PRIME; |
564 | newhash->items = xzalloc(FIRST_PRIME * sizeof(newhash->items[0])); |
565 | |
566 | return newhash; |
567 | } |
568 | |
569 | /* find item in hash, return ptr to data, NULL if not found */ |
570 | static void *hash_search(xhash *hash, const char *name) |
571 | { |
572 | hash_item *hi; |
573 | |
574 | hi = hash->items[hashidx(name) % hash->csize]; |
575 | while (hi) { |
576 | if (strcmp(hi->name, name) == 0) |
577 | return &hi->data; |
578 | hi = hi->next; |
579 | } |
580 | return NULL; |
581 | } |
582 | |
583 | /* grow hash if it becomes too big */ |
584 | static void hash_rebuild(xhash *hash) |
585 | { |
586 | unsigned newsize, i, idx; |
587 | hash_item **newitems, *hi, *thi; |
588 | |
589 | if (hash->nprime == ARRAY_SIZE(PRIMES)) |
590 | return; |
591 | |
592 | newsize = PRIMES[hash->nprime++]; |
593 | newitems = xzalloc(newsize * sizeof(newitems[0])); |
594 | |
595 | for (i = 0; i < hash->csize; i++) { |
596 | hi = hash->items[i]; |
597 | while (hi) { |
598 | thi = hi; |
599 | hi = thi->next; |
600 | idx = hashidx(thi->name) % newsize; |
601 | thi->next = newitems[idx]; |
602 | newitems[idx] = thi; |
603 | } |
604 | } |
605 | |
606 | free(hash->items); |
607 | hash->csize = newsize; |
608 | hash->items = newitems; |
609 | } |
610 | |
611 | /* find item in hash, add it if necessary. Return ptr to data */ |
612 | static void *hash_find(xhash *hash, const char *name) |
613 | { |
614 | hash_item *hi; |
615 | unsigned idx; |
616 | int l; |
617 | |
618 | hi = hash_search(hash, name); |
619 | if (!hi) { |
620 | if (++hash->nel / hash->csize > 10) |
621 | hash_rebuild(hash); |
622 | |
623 | l = strlen(name) + 1; |
624 | hi = xzalloc(sizeof(*hi) + l); |
625 | strcpy(hi->name, name); |
626 | |
627 | idx = hashidx(name) % hash->csize; |
628 | hi->next = hash->items[idx]; |
629 | hash->items[idx] = hi; |
630 | hash->glen += l; |
631 | } |
632 | return &hi->data; |
633 | } |
634 | |
635 | #define findvar(hash, name) ((var*) hash_find((hash), (name))) |
636 | #define newvar(name) ((var*) hash_find(vhash, (name))) |
637 | #define newfile(name) ((rstream*)hash_find(fdhash, (name))) |
638 | #define newfunc(name) ((func*) hash_find(fnhash, (name))) |
639 | |
640 | static void hash_remove(xhash *hash, const char *name) |
641 | { |
642 | hash_item *hi, **phi; |
643 | |
644 | phi = &hash->items[hashidx(name) % hash->csize]; |
645 | while (*phi) { |
646 | hi = *phi; |
647 | if (strcmp(hi->name, name) == 0) { |
648 | hash->glen -= (strlen(name) + 1); |
649 | hash->nel--; |
650 | *phi = hi->next; |
651 | free(hi); |
652 | break; |
653 | } |
654 | phi = &hi->next; |
655 | } |
656 | } |
657 | |
658 | /* ------ some useful functions ------ */ |
659 | |
660 | static char *skip_spaces(char *p) |
661 | { |
662 | while (1) { |
663 | if (*p == '\\' && p[1] == '\n') { |
664 | p++; |
665 | t_lineno++; |
666 | } else if (*p != ' ' && *p != '\t') { |
667 | break; |
668 | } |
669 | p++; |
670 | } |
671 | return p; |
672 | } |
673 | |
674 | /* returns old *s, advances *s past word and terminating NUL */ |
675 | static char *nextword(char **s) |
676 | { |
677 | char *p = *s; |
678 | while (*(*s)++ != '\0') |
679 | continue; |
680 | return p; |
681 | } |
682 | |
683 | static char nextchar(char **s) |
684 | { |
685 | char c, *pps; |
686 | |
687 | c = *(*s)++; |
688 | pps = *s; |
689 | if (c == '\\') |
690 | c = bb_process_escape_sequence((const char**)s); |
691 | if (c == '\\' && *s == pps) |
692 | c = *(*s)++; |
693 | return c; |
694 | } |
695 | |
696 | static ALWAYS_INLINE int isalnum_(int c) |
697 | { |
698 | return (isalnum(c) || c == '_'); |
699 | } |
700 | |
701 | static double my_strtod(char **pp) |
702 | { |
703 | #if ENABLE_DESKTOP |
704 | if ((*pp)[0] == '0' |
705 | && ((((*pp)[1] | 0x20) == 'x') || isdigit((*pp)[1])) |
706 | ) { |
707 | return strtoull(*pp, pp, 0); |
708 | } |
709 | #endif |
710 | return strtod(*pp, pp); |
711 | } |
712 | |
713 | /* -------- working with variables (set/get/copy/etc) -------- */ |
714 | |
715 | static xhash *iamarray(var *v) |
716 | { |
717 | var *a = v; |
718 | |
719 | while (a->type & VF_CHILD) |
720 | a = a->x.parent; |
721 | |
722 | if (!(a->type & VF_ARRAY)) { |
723 | a->type |= VF_ARRAY; |
724 | a->x.array = hash_init(); |
725 | } |
726 | return a->x.array; |
727 | } |
728 | |
729 | static void clear_array(xhash *array) |
730 | { |
731 | unsigned i; |
732 | hash_item *hi, *thi; |
733 | |
734 | for (i = 0; i < array->csize; i++) { |
735 | hi = array->items[i]; |
736 | while (hi) { |
737 | thi = hi; |
738 | hi = hi->next; |
739 | free(thi->data.v.string); |
740 | free(thi); |
741 | } |
742 | array->items[i] = NULL; |
743 | } |
744 | array->glen = array->nel = 0; |
745 | } |
746 | |
747 | /* clear a variable */ |
748 | static var *clrvar(var *v) |
749 | { |
750 | if (!(v->type & VF_FSTR)) |
751 | free(v->string); |
752 | |
753 | v->type &= VF_DONTTOUCH; |
754 | v->type |= VF_DIRTY; |
755 | v->string = NULL; |
756 | return v; |
757 | } |
758 | |
759 | /* assign string value to variable */ |
760 | static var *setvar_p(var *v, char *value) |
761 | { |
762 | clrvar(v); |
763 | v->string = value; |
764 | handle_special(v); |
765 | return v; |
766 | } |
767 | |
768 | /* same as setvar_p but make a copy of string */ |
769 | static var *setvar_s(var *v, const char *value) |
770 | { |
771 | return setvar_p(v, (value && *value) ? xstrdup(value) : NULL); |
772 | } |
773 | |
774 | /* same as setvar_s but sets USER flag */ |
775 | static var *setvar_u(var *v, const char *value) |
776 | { |
777 | v = setvar_s(v, value); |
778 | v->type |= VF_USER; |
779 | return v; |
780 | } |
781 | |
782 | /* set array element to user string */ |
783 | static void setari_u(var *a, int idx, const char *s) |
784 | { |
785 | var *v; |
786 | |
787 | v = findvar(iamarray(a), itoa(idx)); |
788 | setvar_u(v, s); |
789 | } |
790 | |
791 | /* assign numeric value to variable */ |
792 | static var *setvar_i(var *v, double value) |
793 | { |
794 | clrvar(v); |
795 | v->type |= VF_NUMBER; |
796 | v->number = value; |
797 | handle_special(v); |
798 | return v; |
799 | } |
800 | |
801 | static const char *getvar_s(var *v) |
802 | { |
803 | /* if v is numeric and has no cached string, convert it to string */ |
804 | if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) { |
805 | fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE); |
806 | v->string = xstrdup(g_buf); |
807 | v->type |= VF_CACHED; |
808 | } |
809 | return (v->string == NULL) ? "" : v->string; |
810 | } |
811 | |
812 | static double getvar_i(var *v) |
813 | { |
814 | char *s; |
815 | |
816 | if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) { |
817 | v->number = 0; |
818 | s = v->string; |
819 | if (s && *s) { |
820 | v->number = my_strtod(&s); |
821 | if (v->type & VF_USER) { |
822 | s = skip_spaces(s); |
823 | if (*s != '\0') |
824 | v->type &= ~VF_USER; |
825 | } |
826 | } else { |
827 | v->type &= ~VF_USER; |
828 | } |
829 | v->type |= VF_CACHED; |
830 | } |
831 | return v->number; |
832 | } |
833 | |
834 | /* Used for operands of bitwise ops */ |
835 | static unsigned long getvar_i_int(var *v) |
836 | { |
837 | double d = getvar_i(v); |
838 | |
839 | /* Casting doubles to longs is undefined for values outside |
840 | * of target type range. Try to widen it as much as possible */ |
841 | if (d >= 0) |
842 | return (unsigned long)d; |
843 | /* Why? Think about d == -4294967295.0 (assuming 32bit longs) */ |
844 | return - (long) (unsigned long) (-d); |
845 | } |
846 | |
847 | static var *copyvar(var *dest, const var *src) |
848 | { |
849 | if (dest != src) { |
850 | clrvar(dest); |
851 | dest->type |= (src->type & ~(VF_DONTTOUCH | VF_FSTR)); |
852 | dest->number = src->number; |
853 | if (src->string) |
854 | dest->string = xstrdup(src->string); |
855 | } |
856 | handle_special(dest); |
857 | return dest; |
858 | } |
859 | |
860 | static var *incvar(var *v) |
861 | { |
862 | return setvar_i(v, getvar_i(v) + 1.0); |
863 | } |
864 | |
865 | /* return true if v is number or numeric string */ |
866 | static int is_numeric(var *v) |
867 | { |
868 | getvar_i(v); |
869 | return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY)); |
870 | } |
871 | |
872 | /* return 1 when value of v corresponds to true, 0 otherwise */ |
873 | static int istrue(var *v) |
874 | { |
875 | if (is_numeric(v)) |
876 | return (v->number != 0); |
877 | return (v->string && v->string[0]); |
878 | } |
879 | |
880 | /* temporary variables allocator. Last allocated should be first freed */ |
881 | static var *nvalloc(int n) |
882 | { |
883 | nvblock *pb = NULL; |
884 | var *v, *r; |
885 | int size; |
886 | |
887 | while (g_cb) { |
888 | pb = g_cb; |
889 | if ((g_cb->pos - g_cb->nv) + n <= g_cb->size) |
890 | break; |
891 | g_cb = g_cb->next; |
892 | } |
893 | |
894 | if (!g_cb) { |
895 | size = (n <= MINNVBLOCK) ? MINNVBLOCK : n; |
896 | g_cb = xzalloc(sizeof(nvblock) + size * sizeof(var)); |
897 | g_cb->size = size; |
898 | g_cb->pos = g_cb->nv; |
899 | g_cb->prev = pb; |
900 | /*g_cb->next = NULL; - xzalloc did it */ |
901 | if (pb) |
902 | pb->next = g_cb; |
903 | } |
904 | |
905 | v = r = g_cb->pos; |
906 | g_cb->pos += n; |
907 | |
908 | while (v < g_cb->pos) { |
909 | v->type = 0; |
910 | v->string = NULL; |
911 | v++; |
912 | } |
913 | |
914 | return r; |
915 | } |
916 | |
917 | static void nvfree(var *v) |
918 | { |
919 | var *p; |
920 | |
921 | if (v < g_cb->nv || v >= g_cb->pos) |
922 | syntax_error(EMSG_INTERNAL_ERROR); |
923 | |
924 | for (p = v; p < g_cb->pos; p++) { |
925 | if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) { |
926 | clear_array(iamarray(p)); |
927 | free(p->x.array->items); |
928 | free(p->x.array); |
929 | } |
930 | if (p->type & VF_WALK) { |
931 | walker_list *n; |
932 | walker_list *w = p->x.walker; |
933 | debug_printf_walker("nvfree: freeing walker @%p\n", &p->x.walker); |
934 | p->x.walker = NULL; |
935 | while (w) { |
936 | n = w->prev; |
937 | debug_printf_walker(" free(%p)\n", w); |
938 | free(w); |
939 | w = n; |
940 | } |
941 | } |
942 | clrvar(p); |
943 | } |
944 | |
945 | g_cb->pos = v; |
946 | while (g_cb->prev && g_cb->pos == g_cb->nv) { |
947 | g_cb = g_cb->prev; |
948 | } |
949 | } |
950 | |
951 | /* ------- awk program text parsing ------- */ |
952 | |
953 | /* Parse next token pointed by global pos, place results into global ttt. |
954 | * If token isn't expected, give away. Return token class |
955 | */ |
956 | static uint32_t next_token(uint32_t expected) |
957 | { |
958 | #define concat_inserted (G.next_token__concat_inserted) |
959 | #define save_tclass (G.next_token__save_tclass) |
960 | #define save_info (G.next_token__save_info) |
961 | /* Initialized to TC_OPTERM: */ |
962 | #define ltclass (G.next_token__ltclass) |
963 | |
964 | char *p, *s; |
965 | const char *tl; |
966 | uint32_t tc; |
967 | const uint32_t *ti; |
968 | int l; |
969 | |
970 | if (t_rollback) { |
971 | t_rollback = FALSE; |
972 | |
973 | } else if (concat_inserted) { |
974 | concat_inserted = FALSE; |
975 | t_tclass = save_tclass; |
976 | t_info = save_info; |
977 | |
978 | } else { |
979 | p = g_pos; |
980 | readnext: |
981 | p = skip_spaces(p); |
982 | g_lineno = t_lineno; |
983 | if (*p == '#') |
984 | while (*p != '\n' && *p != '\0') |
985 | p++; |
986 | |
987 | if (*p == '\n') |
988 | t_lineno++; |
989 | |
990 | if (*p == '\0') { |
991 | tc = TC_EOF; |
992 | |
993 | } else if (*p == '\"') { |
994 | /* it's a string */ |
995 | t_string = s = ++p; |
996 | while (*p != '\"') { |
997 | char *pp = p; |
998 | if (*p == '\0' || *p == '\n') |
999 | syntax_error(EMSG_UNEXP_EOS); |
1000 | *s++ = nextchar(&pp); |
1001 | p = pp; |
1002 | } |
1003 | p++; |
1004 | *s = '\0'; |
1005 | tc = TC_STRING; |
1006 | |
1007 | } else if ((expected & TC_REGEXP) && *p == '/') { |
1008 | /* it's regexp */ |
1009 | t_string = s = ++p; |
1010 | while (*p != '/') { |
1011 | if (*p == '\0' || *p == '\n') |
1012 | syntax_error(EMSG_UNEXP_EOS); |
1013 | *s = *p++; |
1014 | if (*s++ == '\\') { |
1015 | char *pp = p; |
1016 | s[-1] = bb_process_escape_sequence((const char **)&pp); |
1017 | if (*p == '\\') |
1018 | *s++ = '\\'; |
1019 | if (pp == p) |
1020 | *s++ = *p++; |
1021 | else |
1022 | p = pp; |
1023 | } |
1024 | } |
1025 | p++; |
1026 | *s = '\0'; |
1027 | tc = TC_REGEXP; |
1028 | |
1029 | } else if (*p == '.' || isdigit(*p)) { |
1030 | /* it's a number */ |
1031 | char *pp = p; |
1032 | t_double = my_strtod(&pp); |
1033 | p = pp; |
1034 | if (*pp == '.') |
1035 | syntax_error(EMSG_UNEXP_TOKEN); |
1036 | tc = TC_NUMBER; |
1037 | |
1038 | } else { |
1039 | /* search for something known */ |
1040 | tl = tokenlist; |
1041 | tc = 0x00000001; |
1042 | ti = tokeninfo; |
1043 | while (*tl) { |
1044 | l = *tl++; |
1045 | if (l == NTCC) { |
1046 | tc <<= 1; |
1047 | continue; |
1048 | } |
1049 | /* if token class is expected, token |
1050 | * matches and it's not a longer word, |
1051 | * then this is what we are looking for |
1052 | */ |
1053 | if ((tc & (expected | TC_WORD | TC_NEWLINE)) |
1054 | && *tl == *p && strncmp(p, tl, l) == 0 |
1055 | && !((tc & TC_WORD) && isalnum_(p[l])) |
1056 | ) { |
1057 | t_info = *ti; |
1058 | p += l; |
1059 | break; |
1060 | } |
1061 | ti++; |
1062 | tl += l; |
1063 | } |
1064 | |
1065 | if (!*tl) { |
1066 | /* it's a name (var/array/function), |
1067 | * otherwise it's something wrong |
1068 | */ |
1069 | if (!isalnum_(*p)) |
1070 | syntax_error(EMSG_UNEXP_TOKEN); |
1071 | |
1072 | t_string = --p; |
1073 | while (isalnum_(*++p)) { |
1074 | p[-1] = *p; |
1075 | } |
1076 | p[-1] = '\0'; |
1077 | tc = TC_VARIABLE; |
1078 | /* also consume whitespace between functionname and bracket */ |
1079 | if (!(expected & TC_VARIABLE) || (expected & TC_ARRAY)) |
1080 | p = skip_spaces(p); |
1081 | if (*p == '(') { |
1082 | tc = TC_FUNCTION; |
1083 | } else { |
1084 | if (*p == '[') { |
1085 | p++; |
1086 | tc = TC_ARRAY; |
1087 | } |
1088 | } |
1089 | } |
1090 | } |
1091 | g_pos = p; |
1092 | |
1093 | /* skipping newlines in some cases */ |
1094 | if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE)) |
1095 | goto readnext; |
1096 | |
1097 | /* insert concatenation operator when needed */ |
1098 | if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) { |
1099 | concat_inserted = TRUE; |
1100 | save_tclass = tc; |
1101 | save_info = t_info; |
1102 | tc = TC_BINOP; |
1103 | t_info = OC_CONCAT | SS | P(35); |
1104 | } |
1105 | |
1106 | t_tclass = tc; |
1107 | } |
1108 | ltclass = t_tclass; |
1109 | |
1110 | /* Are we ready for this? */ |
1111 | if (!(ltclass & expected)) |
1112 | syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ? |
1113 | EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN); |
1114 | |
1115 | return ltclass; |
1116 | #undef concat_inserted |
1117 | #undef save_tclass |
1118 | #undef save_info |
1119 | #undef ltclass |
1120 | } |
1121 | |
1122 | static void rollback_token(void) |
1123 | { |
1124 | t_rollback = TRUE; |
1125 | } |
1126 | |
1127 | static node *new_node(uint32_t info) |
1128 | { |
1129 | node *n; |
1130 | |
1131 | n = xzalloc(sizeof(node)); |
1132 | n->info = info; |
1133 | n->lineno = g_lineno; |
1134 | return n; |
1135 | } |
1136 | |
1137 | static void mk_re_node(const char *s, node *n, regex_t *re) |
1138 | { |
1139 | n->info = OC_REGEXP; |
1140 | n->l.re = re; |
1141 | n->r.ire = re + 1; |
1142 | xregcomp(re, s, REG_EXTENDED); |
1143 | xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE); |
1144 | } |
1145 | |
1146 | static node *condition(void) |
1147 | { |
1148 | next_token(TC_SEQSTART); |
1149 | return parse_expr(TC_SEQTERM); |
1150 | } |
1151 | |
1152 | /* parse expression terminated by given argument, return ptr |
1153 | * to built subtree. Terminator is eaten by parse_expr */ |
1154 | static node *parse_expr(uint32_t iexp) |
1155 | { |
1156 | node sn; |
1157 | node *cn = &sn; |
1158 | node *vn, *glptr; |
1159 | uint32_t tc, xtc; |
1160 | var *v; |
1161 | |
1162 | sn.info = PRIMASK; |
1163 | sn.r.n = glptr = NULL; |
1164 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp; |
1165 | |
1166 | while (!((tc = next_token(xtc)) & iexp)) { |
1167 | if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) { |
1168 | /* input redirection (<) attached to glptr node */ |
1169 | cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37)); |
1170 | cn->a.n = glptr; |
1171 | xtc = TC_OPERAND | TC_UOPPRE; |
1172 | glptr = NULL; |
1173 | |
1174 | } else if (tc & (TC_BINOP | TC_UOPPOST)) { |
1175 | /* for binary and postfix-unary operators, jump back over |
1176 | * previous operators with higher priority */ |
1177 | vn = cn; |
1178 | while (((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2)) |
1179 | || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON)) |
1180 | ) { |
1181 | vn = vn->a.n; |
1182 | } |
1183 | if ((t_info & OPCLSMASK) == OC_TERNARY) |
1184 | t_info += P(6); |
1185 | cn = vn->a.n->r.n = new_node(t_info); |
1186 | cn->a.n = vn->a.n; |
1187 | if (tc & TC_BINOP) { |
1188 | cn->l.n = vn; |
1189 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
1190 | if ((t_info & OPCLSMASK) == OC_PGETLINE) { |
1191 | /* it's a pipe */ |
1192 | next_token(TC_GETLINE); |
1193 | /* give maximum priority to this pipe */ |
1194 | cn->info &= ~PRIMASK; |
1195 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1196 | } |
1197 | } else { |
1198 | cn->r.n = vn; |
1199 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1200 | } |
1201 | vn->a.n = cn; |
1202 | |
1203 | } else { |
1204 | /* for operands and prefix-unary operators, attach them |
1205 | * to last node */ |
1206 | vn = cn; |
1207 | cn = vn->r.n = new_node(t_info); |
1208 | cn->a.n = vn; |
1209 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
1210 | if (tc & (TC_OPERAND | TC_REGEXP)) { |
1211 | xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp; |
1212 | /* one should be very careful with switch on tclass - |
1213 | * only simple tclasses should be used! */ |
1214 | switch (tc) { |
1215 | case TC_VARIABLE: |
1216 | case TC_ARRAY: |
1217 | cn->info = OC_VAR; |
1218 | v = hash_search(ahash, t_string); |
1219 | if (v != NULL) { |
1220 | cn->info = OC_FNARG; |
1221 | cn->l.aidx = v->x.aidx; |
1222 | } else { |
1223 | cn->l.v = newvar(t_string); |
1224 | } |
1225 | if (tc & TC_ARRAY) { |
1226 | cn->info |= xS; |
1227 | cn->r.n = parse_expr(TC_ARRTERM); |
1228 | } |
1229 | break; |
1230 | |
1231 | case TC_NUMBER: |
1232 | case TC_STRING: |
1233 | cn->info = OC_VAR; |
1234 | v = cn->l.v = xzalloc(sizeof(var)); |
1235 | if (tc & TC_NUMBER) |
1236 | setvar_i(v, t_double); |
1237 | else |
1238 | setvar_s(v, t_string); |
1239 | break; |
1240 | |
1241 | case TC_REGEXP: |
1242 | mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2)); |
1243 | break; |
1244 | |
1245 | case TC_FUNCTION: |
1246 | cn->info = OC_FUNC; |
1247 | cn->r.f = newfunc(t_string); |
1248 | cn->l.n = condition(); |
1249 | break; |
1250 | |
1251 | case TC_SEQSTART: |
1252 | cn = vn->r.n = parse_expr(TC_SEQTERM); |
1253 | cn->a.n = vn; |
1254 | break; |
1255 | |
1256 | case TC_GETLINE: |
1257 | glptr = cn; |
1258 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1259 | break; |
1260 | |
1261 | case TC_BUILTIN: |
1262 | cn->l.n = condition(); |
1263 | break; |
1264 | } |
1265 | } |
1266 | } |
1267 | } |
1268 | return sn.r.n; |
1269 | } |
1270 | |
1271 | /* add node to chain. Return ptr to alloc'd node */ |
1272 | static node *chain_node(uint32_t info) |
1273 | { |
1274 | node *n; |
1275 | |
1276 | if (!seq->first) |
1277 | seq->first = seq->last = new_node(0); |
1278 | |
1279 | if (seq->programname != g_progname) { |
1280 | seq->programname = g_progname; |
1281 | n = chain_node(OC_NEWSOURCE); |
1282 | n->l.new_progname = xstrdup(g_progname); |
1283 | } |
1284 | |
1285 | n = seq->last; |
1286 | n->info = info; |
1287 | seq->last = n->a.n = new_node(OC_DONE); |
1288 | |
1289 | return n; |
1290 | } |
1291 | |
1292 | static void chain_expr(uint32_t info) |
1293 | { |
1294 | node *n; |
1295 | |
1296 | n = chain_node(info); |
1297 | n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
1298 | if (t_tclass & TC_GRPTERM) |
1299 | rollback_token(); |
1300 | } |
1301 | |
1302 | static node *chain_loop(node *nn) |
1303 | { |
1304 | node *n, *n2, *save_brk, *save_cont; |
1305 | |
1306 | save_brk = break_ptr; |
1307 | save_cont = continue_ptr; |
1308 | |
1309 | n = chain_node(OC_BR | Vx); |
1310 | continue_ptr = new_node(OC_EXEC); |
1311 | break_ptr = new_node(OC_EXEC); |
1312 | chain_group(); |
1313 | n2 = chain_node(OC_EXEC | Vx); |
1314 | n2->l.n = nn; |
1315 | n2->a.n = n; |
1316 | continue_ptr->a.n = n2; |
1317 | break_ptr->a.n = n->r.n = seq->last; |
1318 | |
1319 | continue_ptr = save_cont; |
1320 | break_ptr = save_brk; |
1321 | |
1322 | return n; |
1323 | } |
1324 | |
1325 | /* parse group and attach it to chain */ |
1326 | static void chain_group(void) |
1327 | { |
1328 | uint32_t c; |
1329 | node *n, *n2, *n3; |
1330 | |
1331 | do { |
1332 | c = next_token(TC_GRPSEQ); |
1333 | } while (c & TC_NEWLINE); |
1334 | |
1335 | if (c & TC_GRPSTART) { |
1336 | while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) { |
1337 | if (t_tclass & TC_NEWLINE) |
1338 | continue; |
1339 | rollback_token(); |
1340 | chain_group(); |
1341 | } |
1342 | } else if (c & (TC_OPSEQ | TC_OPTERM)) { |
1343 | rollback_token(); |
1344 | chain_expr(OC_EXEC | Vx); |
1345 | } else { /* TC_STATEMNT */ |
1346 | switch (t_info & OPCLSMASK) { |
1347 | case ST_IF: |
1348 | n = chain_node(OC_BR | Vx); |
1349 | n->l.n = condition(); |
1350 | chain_group(); |
1351 | n2 = chain_node(OC_EXEC); |
1352 | n->r.n = seq->last; |
1353 | if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE) == TC_ELSE) { |
1354 | chain_group(); |
1355 | n2->a.n = seq->last; |
1356 | } else { |
1357 | rollback_token(); |
1358 | } |
1359 | break; |
1360 | |
1361 | case ST_WHILE: |
1362 | n2 = condition(); |
1363 | n = chain_loop(NULL); |
1364 | n->l.n = n2; |
1365 | break; |
1366 | |
1367 | case ST_DO: |
1368 | n2 = chain_node(OC_EXEC); |
1369 | n = chain_loop(NULL); |
1370 | n2->a.n = n->a.n; |
1371 | next_token(TC_WHILE); |
1372 | n->l.n = condition(); |
1373 | break; |
1374 | |
1375 | case ST_FOR: |
1376 | next_token(TC_SEQSTART); |
1377 | n2 = parse_expr(TC_SEMICOL | TC_SEQTERM); |
1378 | if (t_tclass & TC_SEQTERM) { /* for-in */ |
1379 | if ((n2->info & OPCLSMASK) != OC_IN) |
1380 | syntax_error(EMSG_UNEXP_TOKEN); |
1381 | n = chain_node(OC_WALKINIT | VV); |
1382 | n->l.n = n2->l.n; |
1383 | n->r.n = n2->r.n; |
1384 | n = chain_loop(NULL); |
1385 | n->info = OC_WALKNEXT | Vx; |
1386 | n->l.n = n2->l.n; |
1387 | } else { /* for (;;) */ |
1388 | n = chain_node(OC_EXEC | Vx); |
1389 | n->l.n = n2; |
1390 | n2 = parse_expr(TC_SEMICOL); |
1391 | n3 = parse_expr(TC_SEQTERM); |
1392 | n = chain_loop(n3); |
1393 | n->l.n = n2; |
1394 | if (!n2) |
1395 | n->info = OC_EXEC; |
1396 | } |
1397 | break; |
1398 | |
1399 | case OC_PRINT: |
1400 | case OC_PRINTF: |
1401 | n = chain_node(t_info); |
1402 | n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM); |
1403 | if (t_tclass & TC_OUTRDR) { |
1404 | n->info |= t_info; |
1405 | n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
1406 | } |
1407 | if (t_tclass & TC_GRPTERM) |
1408 | rollback_token(); |
1409 | break; |
1410 | |
1411 | case OC_BREAK: |
1412 | n = chain_node(OC_EXEC); |
1413 | n->a.n = break_ptr; |
1414 | break; |
1415 | |
1416 | case OC_CONTINUE: |
1417 | n = chain_node(OC_EXEC); |
1418 | n->a.n = continue_ptr; |
1419 | break; |
1420 | |
1421 | /* delete, next, nextfile, return, exit */ |
1422 | default: |
1423 | chain_expr(t_info); |
1424 | } |
1425 | } |
1426 | } |
1427 | |
1428 | static void parse_program(char *p) |
1429 | { |
1430 | uint32_t tclass; |
1431 | node *cn; |
1432 | func *f; |
1433 | var *v; |
1434 | |
1435 | g_pos = p; |
1436 | t_lineno = 1; |
1437 | while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART | |
1438 | TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) { |
1439 | |
1440 | if (tclass & TC_OPTERM) |
1441 | continue; |
1442 | |
1443 | seq = &mainseq; |
1444 | if (tclass & TC_BEGIN) { |
1445 | seq = &beginseq; |
1446 | chain_group(); |
1447 | |
1448 | } else if (tclass & TC_END) { |
1449 | seq = &endseq; |
1450 | chain_group(); |
1451 | |
1452 | } else if (tclass & TC_FUNCDECL) { |
1453 | next_token(TC_FUNCTION); |
1454 | g_pos++; |
1455 | f = newfunc(t_string); |
1456 | f->body.first = NULL; |
1457 | f->nargs = 0; |
1458 | while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) { |
1459 | v = findvar(ahash, t_string); |
1460 | v->x.aidx = f->nargs++; |
1461 | |
1462 | if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM) |
1463 | break; |
1464 | } |
1465 | seq = &f->body; |
1466 | chain_group(); |
1467 | clear_array(ahash); |
1468 | |
1469 | } else if (tclass & TC_OPSEQ) { |
1470 | rollback_token(); |
1471 | cn = chain_node(OC_TEST); |
1472 | cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART); |
1473 | if (t_tclass & TC_GRPSTART) { |
1474 | rollback_token(); |
1475 | chain_group(); |
1476 | } else { |
1477 | chain_node(OC_PRINT); |
1478 | } |
1479 | cn->r.n = mainseq.last; |
1480 | |
1481 | } else /* if (tclass & TC_GRPSTART) */ { |
1482 | rollback_token(); |
1483 | chain_group(); |
1484 | } |
1485 | } |
1486 | } |
1487 | |
1488 | |
1489 | /* -------- program execution part -------- */ |
1490 | |
1491 | static node *mk_splitter(const char *s, tsplitter *spl) |
1492 | { |
1493 | regex_t *re, *ire; |
1494 | node *n; |
1495 | |
1496 | re = &spl->re[0]; |
1497 | ire = &spl->re[1]; |
1498 | n = &spl->n; |
1499 | if ((n->info & OPCLSMASK) == OC_REGEXP) { |
1500 | regfree(re); |
1501 | regfree(ire); // TODO: nuke ire, use re+1? |
1502 | } |
1503 | if (strlen(s) > 1) { |
1504 | mk_re_node(s, n, re); |
1505 | } else { |
1506 | n->info = (uint32_t) *s; |
1507 | } |
1508 | |
1509 | return n; |
1510 | } |
1511 | |
1512 | /* use node as a regular expression. Supplied with node ptr and regex_t |
1513 | * storage space. Return ptr to regex (if result points to preg, it should |
1514 | * be later regfree'd manually |
1515 | */ |
1516 | static regex_t *as_regex(node *op, regex_t *preg) |
1517 | { |
1518 | int cflags; |
1519 | var *v; |
1520 | const char *s; |
1521 | |
1522 | if ((op->info & OPCLSMASK) == OC_REGEXP) { |
1523 | return icase ? op->r.ire : op->l.re; |
1524 | } |
1525 | v = nvalloc(1); |
1526 | s = getvar_s(evaluate(op, v)); |
1527 | |
1528 | cflags = icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED; |
1529 | /* Testcase where REG_EXTENDED fails (unpaired '{'): |
1530 | * echo Hi | awk 'gsub("@(samp|code|file)\{","");' |
1531 | * gawk 3.1.5 eats this. We revert to ~REG_EXTENDED |
1532 | * (maybe gsub is not supposed to use REG_EXTENDED?). |
1533 | */ |
1534 | if (regcomp(preg, s, cflags)) { |
1535 | cflags &= ~REG_EXTENDED; |
1536 | xregcomp(preg, s, cflags); |
1537 | } |
1538 | nvfree(v); |
1539 | return preg; |
1540 | } |
1541 | |
1542 | /* gradually increasing buffer. |
1543 | * note that we reallocate even if n == old_size, |
1544 | * and thus there is at least one extra allocated byte. |
1545 | */ |
1546 | static char* qrealloc(char *b, int n, int *size) |
1547 | { |
1548 | if (!b || n >= *size) { |
1549 | *size = n + (n>>1) + 80; |
1550 | b = xrealloc(b, *size); |
1551 | } |
1552 | return b; |
1553 | } |
1554 | |
1555 | /* resize field storage space */ |
1556 | static void fsrealloc(int size) |
1557 | { |
1558 | int i; |
1559 | |
1560 | if (size >= maxfields) { |
1561 | i = maxfields; |
1562 | maxfields = size + 16; |
1563 | Fields = xrealloc(Fields, maxfields * sizeof(var)); |
1564 | for (; i < maxfields; i++) { |
1565 | Fields[i].type = VF_SPECIAL; |
1566 | Fields[i].string = NULL; |
1567 | } |
1568 | } |
1569 | |
1570 | if (size < nfields) { |
1571 | for (i = size; i < nfields; i++) { |
1572 | clrvar(Fields + i); |
1573 | } |
1574 | } |
1575 | nfields = size; |
1576 | } |
1577 | |
1578 | static int awk_split(const char *s, node *spl, char **slist) |
1579 | { |
1580 | int l, n = 0; |
1581 | char c[4]; |
1582 | char *s1; |
1583 | regmatch_t pmatch[2]; // TODO: why [2]? [1] is enough... |
1584 | |
1585 | /* in worst case, each char would be a separate field */ |
1586 | *slist = s1 = xzalloc(strlen(s) * 2 + 3); |
1587 | strcpy(s1, s); |
1588 | |
1589 | c[0] = c[1] = (char)spl->info; |
1590 | c[2] = c[3] = '\0'; |
1591 | if (*getvar_s(intvar[RS]) == '\0') |
1592 | c[2] = '\n'; |
1593 | |
1594 | if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */ |
1595 | if (!*s) |
1596 | return n; /* "": zero fields */ |
1597 | n++; /* at least one field will be there */ |
1598 | do { |
1599 | l = strcspn(s, c+2); /* len till next NUL or \n */ |
1600 | if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0 |
1601 | && pmatch[0].rm_so <= l |
1602 | ) { |
1603 | l = pmatch[0].rm_so; |
1604 | if (pmatch[0].rm_eo == 0) { |
1605 | l++; |
1606 | pmatch[0].rm_eo++; |
1607 | } |
1608 | n++; /* we saw yet another delimiter */ |
1609 | } else { |
1610 | pmatch[0].rm_eo = l; |
1611 | if (s[l]) |
1612 | pmatch[0].rm_eo++; |
1613 | } |
1614 | memcpy(s1, s, l); |
1615 | /* make sure we remove *all* of the separator chars */ |
1616 | do { |
1617 | s1[l] = '\0'; |
1618 | } while (++l < pmatch[0].rm_eo); |
1619 | nextword(&s1); |
1620 | s += pmatch[0].rm_eo; |
1621 | } while (*s); |
1622 | return n; |
1623 | } |
1624 | if (c[0] == '\0') { /* null split */ |
1625 | while (*s) { |
1626 | *s1++ = *s++; |
1627 | *s1++ = '\0'; |
1628 | n++; |
1629 | } |
1630 | return n; |
1631 | } |
1632 | if (c[0] != ' ') { /* single-character split */ |
1633 | if (icase) { |
1634 | c[0] = toupper(c[0]); |
1635 | c[1] = tolower(c[1]); |
1636 | } |
1637 | if (*s1) |
1638 | n++; |
1639 | while ((s1 = strpbrk(s1, c))) { |
1640 | *s1++ = '\0'; |
1641 | n++; |
1642 | } |
1643 | return n; |
1644 | } |
1645 | /* space split */ |
1646 | while (*s) { |
1647 | s = skip_whitespace(s); |
1648 | if (!*s) |
1649 | break; |
1650 | n++; |
1651 | while (*s && !isspace(*s)) |
1652 | *s1++ = *s++; |
1653 | *s1++ = '\0'; |
1654 | } |
1655 | return n; |
1656 | } |
1657 | |
1658 | static void split_f0(void) |
1659 | { |
1660 | /* static char *fstrings; */ |
1661 | #define fstrings (G.split_f0__fstrings) |
1662 | |
1663 | int i, n; |
1664 | char *s; |
1665 | |
1666 | if (is_f0_split) |
1667 | return; |
1668 | |
1669 | is_f0_split = TRUE; |
1670 | free(fstrings); |
1671 | fsrealloc(0); |
1672 | n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings); |
1673 | fsrealloc(n); |
1674 | s = fstrings; |
1675 | for (i = 0; i < n; i++) { |
1676 | Fields[i].string = nextword(&s); |
1677 | Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY); |
1678 | } |
1679 | |
1680 | /* set NF manually to avoid side effects */ |
1681 | clrvar(intvar[NF]); |
1682 | intvar[NF]->type = VF_NUMBER | VF_SPECIAL; |
1683 | intvar[NF]->number = nfields; |
1684 | #undef fstrings |
1685 | } |
1686 | |
1687 | /* perform additional actions when some internal variables changed */ |
1688 | static void handle_special(var *v) |
1689 | { |
1690 | int n; |
1691 | char *b; |
1692 | const char *sep, *s; |
1693 | int sl, l, len, i, bsize; |
1694 | |
1695 | if (!(v->type & VF_SPECIAL)) |
1696 | return; |
1697 | |
1698 | if (v == intvar[NF]) { |
1699 | n = (int)getvar_i(v); |
1700 | fsrealloc(n); |
1701 | |
1702 | /* recalculate $0 */ |
1703 | sep = getvar_s(intvar[OFS]); |
1704 | sl = strlen(sep); |
1705 | b = NULL; |
1706 | len = 0; |
1707 | for (i = 0; i < n; i++) { |
1708 | s = getvar_s(&Fields[i]); |
1709 | l = strlen(s); |
1710 | if (b) { |
1711 | memcpy(b+len, sep, sl); |
1712 | len += sl; |
1713 | } |
1714 | b = qrealloc(b, len+l+sl, &bsize); |
1715 | memcpy(b+len, s, l); |
1716 | len += l; |
1717 | } |
1718 | if (b) |
1719 | b[len] = '\0'; |
1720 | setvar_p(intvar[F0], b); |
1721 | is_f0_split = TRUE; |
1722 | |
1723 | } else if (v == intvar[F0]) { |
1724 | is_f0_split = FALSE; |
1725 | |
1726 | } else if (v == intvar[FS]) { |
1727 | mk_splitter(getvar_s(v), &fsplitter); |
1728 | |
1729 | } else if (v == intvar[RS]) { |
1730 | mk_splitter(getvar_s(v), &rsplitter); |
1731 | |
1732 | } else if (v == intvar[IGNORECASE]) { |
1733 | icase = istrue(v); |
1734 | |
1735 | } else { /* $n */ |
1736 | n = getvar_i(intvar[NF]); |
1737 | setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1); |
1738 | /* right here v is invalid. Just to note... */ |
1739 | } |
1740 | } |
1741 | |
1742 | /* step through func/builtin/etc arguments */ |
1743 | static node *nextarg(node **pn) |
1744 | { |
1745 | node *n; |
1746 | |
1747 | n = *pn; |
1748 | if (n && (n->info & OPCLSMASK) == OC_COMMA) { |
1749 | *pn = n->r.n; |
1750 | n = n->l.n; |
1751 | } else { |
1752 | *pn = NULL; |
1753 | } |
1754 | return n; |
1755 | } |
1756 | |
1757 | static void hashwalk_init(var *v, xhash *array) |
1758 | { |
1759 | hash_item *hi; |
1760 | unsigned i; |
1761 | walker_list *w; |
1762 | walker_list *prev_walker; |
1763 | |
1764 | if (v->type & VF_WALK) { |
1765 | prev_walker = v->x.walker; |
1766 | } else { |
1767 | v->type |= VF_WALK; |
1768 | prev_walker = NULL; |
1769 | } |
1770 | debug_printf_walker("hashwalk_init: prev_walker:%p\n", prev_walker); |
1771 | |
1772 | w = v->x.walker = xzalloc(sizeof(*w) + array->glen + 1); /* why + 1? */ |
1773 | debug_printf_walker(" walker@%p=%p\n", &v->x.walker, w); |
1774 | w->cur = w->end = w->wbuf; |
1775 | w->prev = prev_walker; |
1776 | for (i = 0; i < array->csize; i++) { |
1777 | hi = array->items[i]; |
1778 | while (hi) { |
1779 | strcpy(w->end, hi->name); |
1780 | nextword(&w->end); |
1781 | hi = hi->next; |
1782 | } |
1783 | } |
1784 | } |
1785 | |
1786 | static int hashwalk_next(var *v) |
1787 | { |
1788 | walker_list *w = v->x.walker; |
1789 | |
1790 | if (w->cur >= w->end) { |
1791 | walker_list *prev_walker = w->prev; |
1792 | |
1793 | debug_printf_walker("end of iteration, free(walker@%p:%p), prev_walker:%p\n", &v->x.walker, w, prev_walker); |
1794 | free(w); |
1795 | v->x.walker = prev_walker; |
1796 | return FALSE; |
1797 | } |
1798 | |
1799 | setvar_s(v, nextword(&w->cur)); |
1800 | return TRUE; |
1801 | } |
1802 | |
1803 | /* evaluate node, return 1 when result is true, 0 otherwise */ |
1804 | static int ptest(node *pattern) |
1805 | { |
1806 | /* ptest__v is "static": to save stack space? */ |
1807 | return istrue(evaluate(pattern, &G.ptest__v)); |
1808 | } |
1809 | |
1810 | /* read next record from stream rsm into a variable v */ |
1811 | static int awk_getline(rstream *rsm, var *v) |
1812 | { |
1813 | char *b; |
1814 | regmatch_t pmatch[2]; |
1815 | int size, a, p, pp = 0; |
1816 | int fd, so, eo, r, rp; |
1817 | char c, *m, *s; |
1818 | |
1819 | /* we're using our own buffer since we need access to accumulating |
1820 | * characters |
1821 | */ |
1822 | fd = fileno(rsm->F); |
1823 | m = rsm->buffer; |
1824 | a = rsm->adv; |
1825 | p = rsm->pos; |
1826 | size = rsm->size; |
1827 | c = (char) rsplitter.n.info; |
1828 | rp = 0; |
1829 | |
1830 | if (!m) |
1831 | m = qrealloc(m, 256, &size); |
1832 | |
1833 | do { |
1834 | b = m + a; |
1835 | so = eo = p; |
1836 | r = 1; |
1837 | if (p > 0) { |
1838 | if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) { |
1839 | if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re, |
1840 | b, 1, pmatch, 0) == 0) { |
1841 | so = pmatch[0].rm_so; |
1842 | eo = pmatch[0].rm_eo; |
1843 | if (b[eo] != '\0') |
1844 | break; |
1845 | } |
1846 | } else if (c != '\0') { |
1847 | s = strchr(b+pp, c); |
1848 | if (!s) |
1849 | s = memchr(b+pp, '\0', p - pp); |
1850 | if (s) { |
1851 | so = eo = s-b; |
1852 | eo++; |
1853 | break; |
1854 | } |
1855 | } else { |
1856 | while (b[rp] == '\n') |
1857 | rp++; |
1858 | s = strstr(b+rp, "\n\n"); |
1859 | if (s) { |
1860 | so = eo = s-b; |
1861 | while (b[eo] == '\n') |
1862 | eo++; |
1863 | if (b[eo] != '\0') |
1864 | break; |
1865 | } |
1866 | } |
1867 | } |
1868 | |
1869 | if (a > 0) { |
1870 | memmove(m, m+a, p+1); |
1871 | b = m; |
1872 | a = 0; |
1873 | } |
1874 | |
1875 | m = qrealloc(m, a+p+128, &size); |
1876 | b = m + a; |
1877 | pp = p; |
1878 | p += safe_read(fd, b+p, size-p-1); |
1879 | if (p < pp) { |
1880 | p = 0; |
1881 | r = 0; |
1882 | setvar_i(intvar[ERRNO], errno); |
1883 | } |
1884 | b[p] = '\0'; |
1885 | |
1886 | } while (p > pp); |
1887 | |
1888 | if (p == 0) { |
1889 | r--; |
1890 | } else { |
1891 | c = b[so]; b[so] = '\0'; |
1892 | setvar_s(v, b+rp); |
1893 | v->type |= VF_USER; |
1894 | b[so] = c; |
1895 | c = b[eo]; b[eo] = '\0'; |
1896 | setvar_s(intvar[RT], b+so); |
1897 | b[eo] = c; |
1898 | } |
1899 | |
1900 | rsm->buffer = m; |
1901 | rsm->adv = a + eo; |
1902 | rsm->pos = p - eo; |
1903 | rsm->size = size; |
1904 | |
1905 | return r; |
1906 | } |
1907 | |
1908 | static int fmt_num(char *b, int size, const char *format, double n, int int_as_int) |
1909 | { |
1910 | int r = 0; |
1911 | char c; |
1912 | const char *s = format; |
1913 | |
1914 | if (int_as_int && n == (int)n) { |
1915 | r = snprintf(b, size, "%d", (int)n); |
1916 | } else { |
1917 | do { c = *s; } while (c && *++s); |
1918 | if (strchr("diouxX", c)) { |
1919 | r = snprintf(b, size, format, (int)n); |
1920 | } else if (strchr("eEfgG", c)) { |
1921 | r = snprintf(b, size, format, n); |
1922 | } else { |
1923 | syntax_error(EMSG_INV_FMT); |
1924 | } |
1925 | } |
1926 | return r; |
1927 | } |
1928 | |
1929 | /* formatted output into an allocated buffer, return ptr to buffer */ |
1930 | static char *awk_printf(node *n) |
1931 | { |
1932 | char *b = NULL; |
1933 | char *fmt, *s, *f; |
1934 | const char *s1; |
1935 | int i, j, incr, bsize; |
1936 | char c, c1; |
1937 | var *v, *arg; |
1938 | |
1939 | v = nvalloc(1); |
1940 | fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v))); |
1941 | |
1942 | i = 0; |
1943 | while (*f) { |
1944 | s = f; |
1945 | while (*f && (*f != '%' || *++f == '%')) |
1946 | f++; |
1947 | while (*f && !isalpha(*f)) { |
1948 | if (*f == '*') |
1949 | syntax_error("%*x formats are not supported"); |
1950 | f++; |
1951 | } |
1952 | |
1953 | incr = (f - s) + MAXVARFMT; |
1954 | b = qrealloc(b, incr + i, &bsize); |
1955 | c = *f; |
1956 | if (c != '\0') |
1957 | f++; |
1958 | c1 = *f; |
1959 | *f = '\0'; |
1960 | arg = evaluate(nextarg(&n), v); |
1961 | |
1962 | j = i; |
1963 | if (c == 'c' || !c) { |
1964 | i += sprintf(b+i, s, is_numeric(arg) ? |
1965 | (char)getvar_i(arg) : *getvar_s(arg)); |
1966 | } else if (c == 's') { |
1967 | s1 = getvar_s(arg); |
1968 | b = qrealloc(b, incr+i+strlen(s1), &bsize); |
1969 | i += sprintf(b+i, s, s1); |
1970 | } else { |
1971 | i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE); |
1972 | } |
1973 | *f = c1; |
1974 | |
1975 | /* if there was an error while sprintf, return value is negative */ |
1976 | if (i < j) |
1977 | i = j; |
1978 | } |
1979 | |
1980 | free(fmt); |
1981 | nvfree(v); |
1982 | b = xrealloc(b, i + 1); |
1983 | b[i] = '\0'; |
1984 | return b; |
1985 | } |
1986 | |
1987 | /* Common substitution routine. |
1988 | * Replace (nm)'th substring of (src) that matches (rn) with (repl), |
1989 | * store result into (dest), return number of substitutions. |
1990 | * If nm = 0, replace all matches. |
1991 | * If src or dst is NULL, use $0. |
1992 | * If subexp != 0, enable subexpression matching (\1-\9). |
1993 | */ |
1994 | static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int subexp) |
1995 | { |
1996 | char *resbuf; |
1997 | const char *sp; |
1998 | int match_no, residx, replen, resbufsize; |
1999 | int regexec_flags; |
2000 | regmatch_t pmatch[10]; |
2001 | regex_t sreg, *regex; |
2002 | |
2003 | resbuf = NULL; |
2004 | residx = 0; |
2005 | match_no = 0; |
2006 | regexec_flags = 0; |
2007 | regex = as_regex(rn, &sreg); |
2008 | sp = getvar_s(src ? src : intvar[F0]); |
2009 | replen = strlen(repl); |
2010 | while (regexec(regex, sp, 10, pmatch, regexec_flags) == 0) { |
2011 | int so = pmatch[0].rm_so; |
2012 | int eo = pmatch[0].rm_eo; |
2013 | |
2014 | //bb_error_msg("match %u: [%u,%u] '%s'%p", match_no+1, so, eo, sp,sp); |
2015 | resbuf = qrealloc(resbuf, residx + eo + replen, &resbufsize); |
2016 | memcpy(resbuf + residx, sp, eo); |
2017 | residx += eo; |
2018 | if (++match_no >= nm) { |
2019 | const char *s; |
2020 | int nbs; |
2021 | |
2022 | /* replace */ |
2023 | residx -= (eo - so); |
2024 | nbs = 0; |
2025 | for (s = repl; *s; s++) { |
2026 | char c = resbuf[residx++] = *s; |
2027 | if (c == '\\') { |
2028 | nbs++; |
2029 | continue; |
2030 | } |
2031 | if (c == '&' || (subexp && c >= '0' && c <= '9')) { |
2032 | int j; |
2033 | residx -= ((nbs + 3) >> 1); |
2034 | j = 0; |
2035 | if (c != '&') { |
2036 | j = c - '0'; |
2037 | nbs++; |
2038 | } |
2039 | if (nbs % 2) { |
2040 | resbuf[residx++] = c; |
2041 | } else { |
2042 | int n = pmatch[j].rm_eo - pmatch[j].rm_so; |
2043 | resbuf = qrealloc(resbuf, residx + replen + n, &resbufsize); |
2044 | memcpy(resbuf + residx, sp + pmatch[j].rm_so, n); |
2045 | residx += n; |
2046 | } |
2047 | } |
2048 | nbs = 0; |
2049 | } |
2050 | } |
2051 | |
2052 | regexec_flags = REG_NOTBOL; |
2053 | sp += eo; |
2054 | if (match_no == nm) |
2055 | break; |
2056 | if (eo == so) { |
2057 | /* Empty match (e.g. "b*" will match anywhere). |
2058 | * Advance by one char. */ |
2059 | //BUG (bug 1333): |
2060 | //gsub(/\<b*/,"") on "abc" will reach this point, advance to "bc" |
2061 | //... and will erroneously match "b" even though it is NOT at the word start. |
2062 | //we need REG_NOTBOW but it does not exist... |
2063 | //TODO: if EXTRA_COMPAT=y, use GNU matching and re_search, |
2064 | //it should be able to do it correctly. |
2065 | /* Subtle: this is safe only because |
2066 | * qrealloc allocated at least one extra byte */ |
2067 | resbuf[residx] = *sp; |
2068 | if (*sp == '\0') |
2069 | goto ret; |
2070 | sp++; |
2071 | residx++; |
2072 | } |
2073 | } |
2074 | |
2075 | resbuf = qrealloc(resbuf, residx + strlen(sp), &resbufsize); |
2076 | strcpy(resbuf + residx, sp); |
2077 | ret: |
2078 | //bb_error_msg("end sp:'%s'%p", sp,sp); |
2079 | setvar_p(dest ? dest : intvar[F0], resbuf); |
2080 | if (regex == &sreg) |
2081 | regfree(regex); |
2082 | return match_no; |
2083 | } |
2084 | |
2085 | static NOINLINE int do_mktime(const char *ds) |
2086 | { |
2087 | struct tm then; |
2088 | int count; |
2089 | |
2090 | /*memset(&then, 0, sizeof(then)); - not needed */ |
2091 | then.tm_isdst = -1; /* default is unknown */ |
2092 | |
2093 | /* manpage of mktime says these fields are ints, |
2094 | * so we can sscanf stuff directly into them */ |
2095 | count = sscanf(ds, "%u %u %u %u %u %u %d", |
2096 | &then.tm_year, &then.tm_mon, &then.tm_mday, |
2097 | &then.tm_hour, &then.tm_min, &then.tm_sec, |
2098 | &then.tm_isdst); |
2099 | |
2100 | if (count < 6 |
2101 | || (unsigned)then.tm_mon < 1 |
2102 | || (unsigned)then.tm_year < 1900 |
2103 | ) { |
2104 | return -1; |
2105 | } |
2106 | |
2107 | then.tm_mon -= 1; |
2108 | then.tm_year -= 1900; |
2109 | |
2110 | return mktime(&then); |
2111 | } |
2112 | |
2113 | static NOINLINE var *exec_builtin(node *op, var *res) |
2114 | { |
2115 | #define tspl (G.exec_builtin__tspl) |
2116 | |
2117 | var *tv; |
2118 | node *an[4]; |
2119 | var *av[4]; |
2120 | const char *as[4]; |
2121 | regmatch_t pmatch[2]; |
2122 | regex_t sreg, *re; |
2123 | node *spl; |
2124 | uint32_t isr, info; |
2125 | int nargs; |
2126 | time_t tt; |
2127 | int i, l, ll, n; |
2128 | |
2129 | tv = nvalloc(4); |
2130 | isr = info = op->info; |
2131 | op = op->l.n; |
2132 | |
2133 | av[2] = av[3] = NULL; |
2134 | for (i = 0; i < 4 && op; i++) { |
2135 | an[i] = nextarg(&op); |
2136 | if (isr & 0x09000000) |
2137 | av[i] = evaluate(an[i], &tv[i]); |
2138 | if (isr & 0x08000000) |
2139 | as[i] = getvar_s(av[i]); |
2140 | isr >>= 1; |
2141 | } |
2142 | |
2143 | nargs = i; |
2144 | if ((uint32_t)nargs < (info >> 30)) |
2145 | syntax_error(EMSG_TOO_FEW_ARGS); |
2146 | |
2147 | info &= OPNMASK; |
2148 | switch (info) { |
2149 | |
2150 | case B_a2: |
2151 | #if ENABLE_FEATURE_AWK_LIBM |
2152 | setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1]))); |
2153 | #else |
2154 | syntax_error(EMSG_NO_MATH); |
2155 | #endif |
2156 | break; |
2157 | |
2158 | case B_sp: { |
2159 | char *s, *s1; |
2160 | |
2161 | if (nargs > 2) { |
2162 | spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ? |
2163 | an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl); |
2164 | } else { |
2165 | spl = &fsplitter.n; |
2166 | } |
2167 | |
2168 | n = awk_split(as[0], spl, &s); |
2169 | s1 = s; |
2170 | clear_array(iamarray(av[1])); |
2171 | for (i = 1; i <= n; i++) |
2172 | setari_u(av[1], i, nextword(&s)); |
2173 | free(s1); |
2174 | setvar_i(res, n); |
2175 | break; |
2176 | } |
2177 | |
2178 | case B_ss: { |
2179 | char *s; |
2180 | |
2181 | l = strlen(as[0]); |
2182 | i = getvar_i(av[1]) - 1; |
2183 | if (i > l) |
2184 | i = l; |
2185 | if (i < 0) |
2186 | i = 0; |
2187 | n = (nargs > 2) ? getvar_i(av[2]) : l-i; |
2188 | if (n < 0) |
2189 | n = 0; |
2190 | s = xstrndup(as[0]+i, n); |
2191 | setvar_p(res, s); |
2192 | break; |
2193 | } |
2194 | |
2195 | /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5: |
2196 | * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */ |
2197 | case B_an: |
2198 | setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1])); |
2199 | break; |
2200 | |
2201 | case B_co: |
2202 | setvar_i(res, ~getvar_i_int(av[0])); |
2203 | break; |
2204 | |
2205 | case B_ls: |
2206 | setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1])); |
2207 | break; |
2208 | |
2209 | case B_or: |
2210 | setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1])); |
2211 | break; |
2212 | |
2213 | case B_rs: |
2214 | setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1])); |
2215 | break; |
2216 | |
2217 | case B_xo: |
2218 | setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1])); |
2219 | break; |
2220 | |
2221 | case B_lo: |
2222 | case B_up: { |
2223 | char *s, *s1; |
2224 | s1 = s = xstrdup(as[0]); |
2225 | while (*s1) { |
2226 | //*s1 = (info == B_up) ? toupper(*s1) : tolower(*s1); |
2227 | if ((unsigned char)((*s1 | 0x20) - 'a') <= ('z' - 'a')) |
2228 | *s1 = (info == B_up) ? (*s1 & 0xdf) : (*s1 | 0x20); |
2229 | s1++; |
2230 | } |
2231 | setvar_p(res, s); |
2232 | break; |
2233 | } |
2234 | |
2235 | case B_ix: |
2236 | n = 0; |
2237 | ll = strlen(as[1]); |
2238 | l = strlen(as[0]) - ll; |
2239 | if (ll > 0 && l >= 0) { |
2240 | if (!icase) { |
2241 | char *s = strstr(as[0], as[1]); |
2242 | if (s) |
2243 | n = (s - as[0]) + 1; |
2244 | } else { |
2245 | /* this piece of code is terribly slow and |
2246 | * really should be rewritten |
2247 | */ |
2248 | for (i = 0; i <= l; i++) { |
2249 | if (strncasecmp(as[0]+i, as[1], ll) == 0) { |
2250 | n = i+1; |
2251 | break; |
2252 | } |
2253 | } |
2254 | } |
2255 | } |
2256 | setvar_i(res, n); |
2257 | break; |
2258 | |
2259 | case B_ti: |
2260 | if (nargs > 1) |
2261 | tt = getvar_i(av[1]); |
2262 | else |
2263 | time(&tt); |
2264 | //s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"; |
2265 | i = strftime(g_buf, MAXVARFMT, |
2266 | ((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"), |
2267 | localtime(&tt)); |
2268 | g_buf[i] = '\0'; |
2269 | setvar_s(res, g_buf); |
2270 | break; |
2271 | |
2272 | case B_mt: |
2273 | setvar_i(res, do_mktime(as[0])); |
2274 | break; |
2275 | |
2276 | case B_ma: |
2277 | re = as_regex(an[1], &sreg); |
2278 | n = regexec(re, as[0], 1, pmatch, 0); |
2279 | if (n == 0) { |
2280 | pmatch[0].rm_so++; |
2281 | pmatch[0].rm_eo++; |
2282 | } else { |
2283 | pmatch[0].rm_so = 0; |
2284 | pmatch[0].rm_eo = -1; |
2285 | } |
2286 | setvar_i(newvar("RSTART"), pmatch[0].rm_so); |
2287 | setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so); |
2288 | setvar_i(res, pmatch[0].rm_so); |
2289 | if (re == &sreg) |
2290 | regfree(re); |
2291 | break; |
2292 | |
2293 | case B_ge: |
2294 | awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE); |
2295 | break; |
2296 | |
2297 | case B_gs: |
2298 | setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE)); |
2299 | break; |
2300 | |
2301 | case B_su: |
2302 | setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE)); |
2303 | break; |
2304 | } |
2305 | |
2306 | nvfree(tv); |
2307 | return res; |
2308 | #undef tspl |
2309 | } |
2310 | |
2311 | /* |
2312 | * Evaluate node - the heart of the program. Supplied with subtree |
2313 | * and place where to store result. returns ptr to result. |
2314 | */ |
2315 | #define XC(n) ((n) >> 8) |
2316 | |
2317 | static var *evaluate(node *op, var *res) |
2318 | { |
2319 | /* This procedure is recursive so we should count every byte */ |
2320 | #define fnargs (G.evaluate__fnargs) |
2321 | /* seed is initialized to 1 */ |
2322 | #define seed (G.evaluate__seed) |
2323 | #define sreg (G.evaluate__sreg) |
2324 | |
2325 | var *v1; |
2326 | |
2327 | if (!op) |
2328 | return setvar_s(res, NULL); |
2329 | |
2330 | v1 = nvalloc(2); |
2331 | |
2332 | while (op) { |
2333 | struct { |
2334 | var *v; |
2335 | const char *s; |
2336 | } L = L; /* for compiler */ |
2337 | struct { |
2338 | var *v; |
2339 | const char *s; |
2340 | } R = R; |
2341 | double L_d = L_d; |
2342 | uint32_t opinfo; |
2343 | int opn; |
2344 | node *op1; |
2345 | |
2346 | opinfo = op->info; |
2347 | opn = (opinfo & OPNMASK); |
2348 | g_lineno = op->lineno; |
2349 | op1 = op->l.n; |
2350 | |
2351 | /* execute inevitable things */ |
2352 | if (opinfo & OF_RES1) |
2353 | L.v = evaluate(op1, v1); |
2354 | if (opinfo & OF_RES2) |
2355 | R.v = evaluate(op->r.n, v1+1); |
2356 | if (opinfo & OF_STR1) |
2357 | L.s = getvar_s(L.v); |
2358 | if (opinfo & OF_STR2) |
2359 | R.s = getvar_s(R.v); |
2360 | if (opinfo & OF_NUM1) |
2361 | L_d = getvar_i(L.v); |
2362 | |
2363 | switch (XC(opinfo & OPCLSMASK)) { |
2364 | |
2365 | /* -- iterative node type -- */ |
2366 | |
2367 | /* test pattern */ |
2368 | case XC( OC_TEST ): |
2369 | if ((op1->info & OPCLSMASK) == OC_COMMA) { |
2370 | /* it's range pattern */ |
2371 | if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) { |
2372 | op->info |= OF_CHECKED; |
2373 | if (ptest(op1->r.n)) |
2374 | op->info &= ~OF_CHECKED; |
2375 | op = op->a.n; |
2376 | } else { |
2377 | op = op->r.n; |
2378 | } |
2379 | } else { |
2380 | op = ptest(op1) ? op->a.n : op->r.n; |
2381 | } |
2382 | break; |
2383 | |
2384 | /* just evaluate an expression, also used as unconditional jump */ |
2385 | case XC( OC_EXEC ): |
2386 | break; |
2387 | |
2388 | /* branch, used in if-else and various loops */ |
2389 | case XC( OC_BR ): |
2390 | op = istrue(L.v) ? op->a.n : op->r.n; |
2391 | break; |
2392 | |
2393 | /* initialize for-in loop */ |
2394 | case XC( OC_WALKINIT ): |
2395 | hashwalk_init(L.v, iamarray(R.v)); |
2396 | break; |
2397 | |
2398 | /* get next array item */ |
2399 | case XC( OC_WALKNEXT ): |
2400 | op = hashwalk_next(L.v) ? op->a.n : op->r.n; |
2401 | break; |
2402 | |
2403 | case XC( OC_PRINT ): |
2404 | case XC( OC_PRINTF ): { |
2405 | FILE *F = stdout; |
2406 | |
2407 | if (op->r.n) { |
2408 | rstream *rsm = newfile(R.s); |
2409 | if (!rsm->F) { |
2410 | if (opn == '|') { |
2411 | rsm->F = popen(R.s, "w"); |
2412 | if (rsm->F == NULL) |
2413 | bb_perror_msg_and_die("popen"); |
2414 | rsm->is_pipe = 1; |
2415 | } else { |
2416 | rsm->F = xfopen(R.s, opn=='w' ? "w" : "a"); |
2417 | } |
2418 | } |
2419 | F = rsm->F; |
2420 | } |
2421 | |
2422 | if ((opinfo & OPCLSMASK) == OC_PRINT) { |
2423 | if (!op1) { |
2424 | fputs(getvar_s(intvar[F0]), F); |
2425 | } else { |
2426 | while (op1) { |
2427 | var *v = evaluate(nextarg(&op1), v1); |
2428 | if (v->type & VF_NUMBER) { |
2429 | fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]), |
2430 | getvar_i(v), TRUE); |
2431 | fputs(g_buf, F); |
2432 | } else { |
2433 | fputs(getvar_s(v), F); |
2434 | } |
2435 | |
2436 | if (op1) |
2437 | fputs(getvar_s(intvar[OFS]), F); |
2438 | } |
2439 | } |
2440 | fputs(getvar_s(intvar[ORS]), F); |
2441 | |
2442 | } else { /* OC_PRINTF */ |
2443 | char *s = awk_printf(op1); |
2444 | fputs(s, F); |
2445 | free(s); |
2446 | } |
2447 | fflush(F); |
2448 | break; |
2449 | } |
2450 | |
2451 | case XC( OC_DELETE ): { |
2452 | uint32_t info = op1->info & OPCLSMASK; |
2453 | var *v; |
2454 | |
2455 | if (info == OC_VAR) { |
2456 | v = op1->l.v; |
2457 | } else if (info == OC_FNARG) { |
2458 | v = &fnargs[op1->l.aidx]; |
2459 | } else { |
2460 | syntax_error(EMSG_NOT_ARRAY); |
2461 | } |
2462 | |
2463 | if (op1->r.n) { |
2464 | const char *s; |
2465 | clrvar(L.v); |
2466 | s = getvar_s(evaluate(op1->r.n, v1)); |
2467 | hash_remove(iamarray(v), s); |
2468 | } else { |
2469 | clear_array(iamarray(v)); |
2470 | } |
2471 | break; |
2472 | } |
2473 | |
2474 | case XC( OC_NEWSOURCE ): |
2475 | g_progname = op->l.new_progname; |
2476 | break; |
2477 | |
2478 | case XC( OC_RETURN ): |
2479 | copyvar(res, L.v); |
2480 | break; |
2481 | |
2482 | case XC( OC_NEXTFILE ): |
2483 | nextfile = TRUE; |
2484 | case XC( OC_NEXT ): |
2485 | nextrec = TRUE; |
2486 | case XC( OC_DONE ): |
2487 | clrvar(res); |
2488 | break; |
2489 | |
2490 | case XC( OC_EXIT ): |
2491 | awk_exit(L_d); |
2492 | |
2493 | /* -- recursive node type -- */ |
2494 | |
2495 | case XC( OC_VAR ): |
2496 | L.v = op->l.v; |
2497 | if (L.v == intvar[NF]) |
2498 | split_f0(); |
2499 | goto v_cont; |
2500 | |
2501 | case XC( OC_FNARG ): |
2502 | L.v = &fnargs[op->l.aidx]; |
2503 | v_cont: |
2504 | res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v; |
2505 | break; |
2506 | |
2507 | case XC( OC_IN ): |
2508 | setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0); |
2509 | break; |
2510 | |
2511 | case XC( OC_REGEXP ): |
2512 | op1 = op; |
2513 | L.s = getvar_s(intvar[F0]); |
2514 | goto re_cont; |
2515 | |
2516 | case XC( OC_MATCH ): |
2517 | op1 = op->r.n; |
2518 | re_cont: |
2519 | { |
2520 | regex_t *re = as_regex(op1, &sreg); |
2521 | int i = regexec(re, L.s, 0, NULL, 0); |
2522 | if (re == &sreg) |
2523 | regfree(re); |
2524 | setvar_i(res, (i == 0) ^ (opn == '!')); |
2525 | } |
2526 | break; |
2527 | |
2528 | case XC( OC_MOVE ): |
2529 | /* if source is a temporary string, jusk relink it to dest */ |
2530 | //Disabled: if R.v is numeric but happens to have cached R.v->string, |
2531 | //then L.v ends up being a string, which is wrong |
2532 | // if (R.v == v1+1 && R.v->string) { |
2533 | // res = setvar_p(L.v, R.v->string); |
2534 | // R.v->string = NULL; |
2535 | // } else { |
2536 | res = copyvar(L.v, R.v); |
2537 | // } |
2538 | break; |
2539 | |
2540 | case XC( OC_TERNARY ): |
2541 | if ((op->r.n->info & OPCLSMASK) != OC_COLON) |
2542 | syntax_error(EMSG_POSSIBLE_ERROR); |
2543 | res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res); |
2544 | break; |
2545 | |
2546 | case XC( OC_FUNC ): { |
2547 | var *vbeg, *v; |
2548 | const char *sv_progname; |
2549 | |
2550 | if (!op->r.f->body.first) |
2551 | syntax_error(EMSG_UNDEF_FUNC); |
2552 | |
2553 | vbeg = v = nvalloc(op->r.f->nargs + 1); |
2554 | while (op1) { |
2555 | var *arg = evaluate(nextarg(&op1), v1); |
2556 | copyvar(v, arg); |
2557 | v->type |= VF_CHILD; |
2558 | v->x.parent = arg; |
2559 | if (++v - vbeg >= op->r.f->nargs) |
2560 | break; |
2561 | } |
2562 | |
2563 | v = fnargs; |
2564 | fnargs = vbeg; |
2565 | sv_progname = g_progname; |
2566 | |
2567 | res = evaluate(op->r.f->body.first, res); |
2568 | |
2569 | g_progname = sv_progname; |
2570 | nvfree(fnargs); |
2571 | fnargs = v; |
2572 | |
2573 | break; |
2574 | } |
2575 | |
2576 | case XC( OC_GETLINE ): |
2577 | case XC( OC_PGETLINE ): { |
2578 | rstream *rsm; |
2579 | int i; |
2580 | |
2581 | if (op1) { |
2582 | rsm = newfile(L.s); |
2583 | if (!rsm->F) { |
2584 | if ((opinfo & OPCLSMASK) == OC_PGETLINE) { |
2585 | rsm->F = popen(L.s, "r"); |
2586 | rsm->is_pipe = TRUE; |
2587 | } else { |
2588 | rsm->F = fopen_for_read(L.s); /* not xfopen! */ |
2589 | } |
2590 | } |
2591 | } else { |
2592 | if (!iF) |
2593 | iF = next_input_file(); |
2594 | rsm = iF; |
2595 | } |
2596 | |
2597 | if (!rsm->F) { |
2598 | setvar_i(intvar[ERRNO], errno); |
2599 | setvar_i(res, -1); |
2600 | break; |
2601 | } |
2602 | |
2603 | if (!op->r.n) |
2604 | R.v = intvar[F0]; |
2605 | |
2606 | i = awk_getline(rsm, R.v); |
2607 | if (i > 0 && !op1) { |
2608 | incvar(intvar[FNR]); |
2609 | incvar(intvar[NR]); |
2610 | } |
2611 | setvar_i(res, i); |
2612 | break; |
2613 | } |
2614 | |
2615 | /* simple builtins */ |
2616 | case XC( OC_FBLTIN ): { |
2617 | int i; |
2618 | rstream *rsm; |
2619 | double R_d = R_d; /* for compiler */ |
2620 | |
2621 | switch (opn) { |
2622 | case F_in: |
2623 | R_d = (int)L_d; |
2624 | break; |
2625 | |
2626 | case F_rn: |
2627 | R_d = (double)rand() / (double)RAND_MAX; |
2628 | break; |
2629 | #if ENABLE_FEATURE_AWK_LIBM |
2630 | case F_co: |
2631 | R_d = cos(L_d); |
2632 | break; |
2633 | |
2634 | case F_ex: |
2635 | R_d = exp(L_d); |
2636 | break; |
2637 | |
2638 | case F_lg: |
2639 | R_d = log(L_d); |
2640 | break; |
2641 | |
2642 | case F_si: |
2643 | R_d = sin(L_d); |
2644 | break; |
2645 | |
2646 | case F_sq: |
2647 | R_d = sqrt(L_d); |
2648 | break; |
2649 | #else |
2650 | case F_co: |
2651 | case F_ex: |
2652 | case F_lg: |
2653 | case F_si: |
2654 | case F_sq: |
2655 | syntax_error(EMSG_NO_MATH); |
2656 | break; |
2657 | #endif |
2658 | case F_sr: |
2659 | R_d = (double)seed; |
2660 | seed = op1 ? (unsigned)L_d : (unsigned)time(NULL); |
2661 | srand(seed); |
2662 | break; |
2663 | |
2664 | case F_ti: |
2665 | R_d = time(NULL); |
2666 | break; |
2667 | |
2668 | case F_le: |
2669 | if (!op1) |
2670 | L.s = getvar_s(intvar[F0]); |
2671 | R_d = strlen(L.s); |
2672 | break; |
2673 | |
2674 | case F_sy: |
2675 | fflush_all(); |
2676 | R_d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s) |
2677 | ? (system(L.s) >> 8) : 0; |
2678 | break; |
2679 | |
2680 | case F_ff: |
2681 | if (!op1) { |
2682 | fflush(stdout); |
2683 | } else if (L.s && *L.s) { |
2684 | rsm = newfile(L.s); |
2685 | fflush(rsm->F); |
2686 | } else { |
2687 | fflush_all(); |
2688 | } |
2689 | break; |
2690 | |
2691 | case F_cl: |
2692 | i = 0; |
2693 | rsm = (rstream *)hash_search(fdhash, L.s); |
2694 | if (rsm) { |
2695 | i = rsm->is_pipe ? pclose(rsm->F) : fclose(rsm->F); |
2696 | free(rsm->buffer); |
2697 | hash_remove(fdhash, L.s); |
2698 | } |
2699 | if (i != 0) |
2700 | setvar_i(intvar[ERRNO], errno); |
2701 | R_d = (double)i; |
2702 | break; |
2703 | } |
2704 | setvar_i(res, R_d); |
2705 | break; |
2706 | } |
2707 | |
2708 | case XC( OC_BUILTIN ): |
2709 | res = exec_builtin(op, res); |
2710 | break; |
2711 | |
2712 | case XC( OC_SPRINTF ): |
2713 | setvar_p(res, awk_printf(op1)); |
2714 | break; |
2715 | |
2716 | case XC( OC_UNARY ): { |
2717 | double Ld, R_d; |
2718 | |
2719 | Ld = R_d = getvar_i(R.v); |
2720 | switch (opn) { |
2721 | case 'P': |
2722 | Ld = ++R_d; |
2723 | goto r_op_change; |
2724 | case 'p': |
2725 | R_d++; |
2726 | goto r_op_change; |
2727 | case 'M': |
2728 | Ld = --R_d; |
2729 | goto r_op_change; |
2730 | case 'm': |
2731 | R_d--; |
2732 | r_op_change: |
2733 | setvar_i(R.v, R_d); |
2734 | break; |
2735 | case '!': |
2736 | Ld = !istrue(R.v); |
2737 | break; |
2738 | case '-': |
2739 | Ld = -R_d; |
2740 | break; |
2741 | } |
2742 | setvar_i(res, Ld); |
2743 | break; |
2744 | } |
2745 | |
2746 | case XC( OC_FIELD ): { |
2747 | int i = (int)getvar_i(R.v); |
2748 | if (i == 0) { |
2749 | res = intvar[F0]; |
2750 | } else { |
2751 | split_f0(); |
2752 | if (i > nfields) |
2753 | fsrealloc(i); |
2754 | res = &Fields[i - 1]; |
2755 | } |
2756 | break; |
2757 | } |
2758 | |
2759 | /* concatenation (" ") and index joining (",") */ |
2760 | case XC( OC_CONCAT ): |
2761 | case XC( OC_COMMA ): { |
2762 | const char *sep = ""; |
2763 | if ((opinfo & OPCLSMASK) == OC_COMMA) |
2764 | sep = getvar_s(intvar[SUBSEP]); |
2765 | setvar_p(res, xasprintf("%s%s%s", L.s, sep, R.s)); |
2766 | break; |
2767 | } |
2768 | |
2769 | case XC( OC_LAND ): |
2770 | setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0); |
2771 | break; |
2772 | |
2773 | case XC( OC_LOR ): |
2774 | setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n)); |
2775 | break; |
2776 | |
2777 | case XC( OC_BINARY ): |
2778 | case XC( OC_REPLACE ): { |
2779 | double R_d = getvar_i(R.v); |
2780 | switch (opn) { |
2781 | case '+': |
2782 | L_d += R_d; |
2783 | break; |
2784 | case '-': |
2785 | L_d -= R_d; |
2786 | break; |
2787 | case '*': |
2788 | L_d *= R_d; |
2789 | break; |
2790 | case '/': |
2791 | if (R_d == 0) |
2792 | syntax_error(EMSG_DIV_BY_ZERO); |
2793 | L_d /= R_d; |
2794 | break; |
2795 | case '&': |
2796 | #if ENABLE_FEATURE_AWK_LIBM |
2797 | L_d = pow(L_d, R_d); |
2798 | #else |
2799 | syntax_error(EMSG_NO_MATH); |
2800 | #endif |
2801 | break; |
2802 | case '%': |
2803 | if (R_d == 0) |
2804 | syntax_error(EMSG_DIV_BY_ZERO); |
2805 | L_d -= (int)(L_d / R_d) * R_d; |
2806 | break; |
2807 | } |
2808 | res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : L.v, L_d); |
2809 | break; |
2810 | } |
2811 | |
2812 | case XC( OC_COMPARE ): { |
2813 | int i = i; /* for compiler */ |
2814 | double Ld; |
2815 | |
2816 | if (is_numeric(L.v) && is_numeric(R.v)) { |
2817 | Ld = getvar_i(L.v) - getvar_i(R.v); |
2818 | } else { |
2819 | const char *l = getvar_s(L.v); |
2820 | const char *r = getvar_s(R.v); |
2821 | Ld = icase ? strcasecmp(l, r) : strcmp(l, r); |
2822 | } |
2823 | switch (opn & 0xfe) { |
2824 | case 0: |
2825 | i = (Ld > 0); |
2826 | break; |
2827 | case 2: |
2828 | i = (Ld >= 0); |
2829 | break; |
2830 | case 4: |
2831 | i = (Ld == 0); |
2832 | break; |
2833 | } |
2834 | setvar_i(res, (i == 0) ^ (opn & 1)); |
2835 | break; |
2836 | } |
2837 | |
2838 | default: |
2839 | syntax_error(EMSG_POSSIBLE_ERROR); |
2840 | } |
2841 | if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS) |
2842 | op = op->a.n; |
2843 | if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS) |
2844 | break; |
2845 | if (nextrec) |
2846 | break; |
2847 | } /* while (op) */ |
2848 | |
2849 | nvfree(v1); |
2850 | return res; |
2851 | #undef fnargs |
2852 | #undef seed |
2853 | #undef sreg |
2854 | } |
2855 | |
2856 | |
2857 | /* -------- main & co. -------- */ |
2858 | |
2859 | static int awk_exit(int r) |
2860 | { |
2861 | var tv; |
2862 | unsigned i; |
2863 | hash_item *hi; |
2864 | |
2865 | zero_out_var(&tv); |
2866 | |
2867 | if (!exiting) { |
2868 | exiting = TRUE; |
2869 | nextrec = FALSE; |
2870 | evaluate(endseq.first, &tv); |
2871 | } |
2872 | |
2873 | /* waiting for children */ |
2874 | for (i = 0; i < fdhash->csize; i++) { |
2875 | hi = fdhash->items[i]; |
2876 | while (hi) { |
2877 | if (hi->data.rs.F && hi->data.rs.is_pipe) |
2878 | pclose(hi->data.rs.F); |
2879 | hi = hi->next; |
2880 | } |
2881 | } |
2882 | |
2883 | exit(r); |
2884 | } |
2885 | |
2886 | /* if expr looks like "var=value", perform assignment and return 1, |
2887 | * otherwise return 0 */ |
2888 | static int is_assignment(const char *expr) |
2889 | { |
2890 | char *exprc, *s, *s0, *s1; |
2891 | |
2892 | exprc = xstrdup(expr); |
2893 | if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) { |
2894 | free(exprc); |
2895 | return FALSE; |
2896 | } |
2897 | |
2898 | *s++ = '\0'; |
2899 | s0 = s1 = s; |
2900 | while (*s) |
2901 | *s1++ = nextchar(&s); |
2902 | |
2903 | *s1 = '\0'; |
2904 | setvar_u(newvar(exprc), s0); |
2905 | free(exprc); |
2906 | return TRUE; |
2907 | } |
2908 | |
2909 | /* switch to next input file */ |
2910 | static rstream *next_input_file(void) |
2911 | { |
2912 | #define rsm (G.next_input_file__rsm) |
2913 | #define files_happen (G.next_input_file__files_happen) |
2914 | |
2915 | FILE *F = NULL; |
2916 | const char *fname, *ind; |
2917 | |
2918 | if (rsm.F) |
2919 | fclose(rsm.F); |
2920 | rsm.F = NULL; |
2921 | rsm.pos = rsm.adv = 0; |
2922 | |
2923 | do { |
2924 | if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) { |
2925 | if (files_happen) |
2926 | return NULL; |
2927 | fname = "-"; |
2928 | F = stdin; |
2929 | } else { |
2930 | ind = getvar_s(incvar(intvar[ARGIND])); |
2931 | fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind)); |
2932 | if (fname && *fname && !is_assignment(fname)) |
2933 | F = xfopen_stdin(fname); |
2934 | } |
2935 | } while (!F); |
2936 | |
2937 | files_happen = TRUE; |
2938 | setvar_s(intvar[FILENAME], fname); |
2939 | rsm.F = F; |
2940 | return &rsm; |
2941 | #undef rsm |
2942 | #undef files_happen |
2943 | } |
2944 | |
2945 | int awk_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
2946 | int awk_main(int argc, char **argv) |
2947 | { |
2948 | unsigned opt; |
2949 | char *opt_F, *opt_W; |
2950 | llist_t *list_v = NULL; |
2951 | llist_t *list_f = NULL; |
2952 | int i, j; |
2953 | var *v; |
2954 | var tv; |
2955 | char **envp; |
2956 | char *vnames = (char *)vNames; /* cheat */ |
2957 | char *vvalues = (char *)vValues; |
2958 | |
2959 | INIT_G(); |
2960 | |
2961 | /* Undo busybox.c, or else strtod may eat ','! This breaks parsing: |
2962 | * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */ |
2963 | if (ENABLE_LOCALE_SUPPORT) |
2964 | setlocale(LC_NUMERIC, "C"); |
2965 | |
2966 | zero_out_var(&tv); |
2967 | |
2968 | /* allocate global buffer */ |
2969 | g_buf = xmalloc(MAXVARFMT + 1); |
2970 | |
2971 | vhash = hash_init(); |
2972 | ahash = hash_init(); |
2973 | fdhash = hash_init(); |
2974 | fnhash = hash_init(); |
2975 | |
2976 | /* initialize variables */ |
2977 | for (i = 0; *vnames; i++) { |
2978 | intvar[i] = v = newvar(nextword(&vnames)); |
2979 | if (*vvalues != '\377') |
2980 | setvar_s(v, nextword(&vvalues)); |
2981 | else |
2982 | setvar_i(v, 0); |
2983 | |
2984 | if (*vnames == '*') { |
2985 | v->type |= VF_SPECIAL; |
2986 | vnames++; |
2987 | } |
2988 | } |
2989 | |
2990 | handle_special(intvar[FS]); |
2991 | handle_special(intvar[RS]); |
2992 | |
2993 | newfile("/dev/stdin")->F = stdin; |
2994 | newfile("/dev/stdout")->F = stdout; |
2995 | newfile("/dev/stderr")->F = stderr; |
2996 | |
2997 | /* Huh, people report that sometimes environ is NULL. Oh well. */ |
2998 | if (environ) for (envp = environ; *envp; envp++) { |
2999 | /* environ is writable, thus we don't strdup it needlessly */ |
3000 | char *s = *envp; |
3001 | char *s1 = strchr(s, '='); |
3002 | if (s1) { |
3003 | *s1 = '\0'; |
3004 | /* Both findvar and setvar_u take const char* |
3005 | * as 2nd arg -> environment is not trashed */ |
3006 | setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1); |
3007 | *s1 = '='; |
3008 | } |
3009 | } |
3010 | opt_complementary = "v::f::"; /* -v and -f can occur multiple times */ |
3011 | opt = getopt32(argv, "F:v:f:W:", &opt_F, &list_v, &list_f, &opt_W); |
3012 | argv += optind; |
3013 | argc -= optind; |
3014 | if (opt & 0x1) |
3015 | setvar_s(intvar[FS], opt_F); // -F |
3016 | while (list_v) { /* -v */ |
3017 | if (!is_assignment(llist_pop(&list_v))) |
3018 | bb_show_usage(); |
3019 | } |
3020 | if (list_f) { /* -f */ |
3021 | do { |
3022 | char *s = NULL; |
3023 | FILE *from_file; |
3024 | |
3025 | g_progname = llist_pop(&list_f); |
3026 | from_file = xfopen_stdin(g_progname); |
3027 | /* one byte is reserved for some trick in next_token */ |
3028 | for (i = j = 1; j > 0; i += j) { |
3029 | s = xrealloc(s, i + 4096); |
3030 | j = fread(s + i, 1, 4094, from_file); |
3031 | } |
3032 | s[i] = '\0'; |
3033 | fclose(from_file); |
3034 | parse_program(s + 1); |
3035 | free(s); |
3036 | } while (list_f); |
3037 | argc++; |
3038 | } else { // no -f: take program from 1st parameter |
3039 | if (!argc) |
3040 | bb_show_usage(); |
3041 | g_progname = "cmd. line"; |
3042 | parse_program(*argv++); |
3043 | } |
3044 | if (opt & 0x8) // -W |
3045 | bb_error_msg("warning: unrecognized option '-W %s' ignored", opt_W); |
3046 | |
3047 | /* fill in ARGV array */ |
3048 | setvar_i(intvar[ARGC], argc); |
3049 | setari_u(intvar[ARGV], 0, "awk"); |
3050 | i = 0; |
3051 | while (*argv) |
3052 | setari_u(intvar[ARGV], ++i, *argv++); |
3053 | |
3054 | evaluate(beginseq.first, &tv); |
3055 | if (!mainseq.first && !endseq.first) |
3056 | awk_exit(EXIT_SUCCESS); |
3057 | |
3058 | /* input file could already be opened in BEGIN block */ |
3059 | if (!iF) |
3060 | iF = next_input_file(); |
3061 | |
3062 | /* passing through input files */ |
3063 | while (iF) { |
3064 | nextfile = FALSE; |
3065 | setvar_i(intvar[FNR], 0); |
3066 | |
3067 | while ((i = awk_getline(iF, intvar[F0])) > 0) { |
3068 | nextrec = FALSE; |
3069 | incvar(intvar[NR]); |
3070 | incvar(intvar[FNR]); |
3071 | evaluate(mainseq.first, &tv); |
3072 | |
3073 | if (nextfile) |
3074 | break; |
3075 | } |
3076 | |
3077 | if (i < 0) |
3078 | syntax_error(strerror(errno)); |
3079 | |
3080 | iF = next_input_file(); |
3081 | } |
3082 | |
3083 | awk_exit(EXIT_SUCCESS); |
3084 | /*return 0;*/ |
3085 | } |