Contents of /tags/mkinitrd-6_1_11/busybox/editors/awk.c
Parent Directory | Revision Log
Revision 928 -
(show annotations)
(download)
Wed Oct 28 13:31:19 2009 UTC (14 years, 11 months ago) by niro
File MIME type: text/plain
File size: 64854 byte(s)
Wed Oct 28 13:31:19 2009 UTC (14 years, 11 months ago) by niro
File MIME type: text/plain
File size: 64854 byte(s)
tagged 'mkinitrd-6_1_11'
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[0]; |
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_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" |
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), |
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, ARGIND, ARGC, ARGV, |
370 | ERRNO, FNR, |
371 | NR, NF, IGNORECASE, |
372 | ENVIRON, F0, NUM_INTERNAL_VARS |
373 | }; |
374 | |
375 | static const char vNames[] ALIGN1 = |
376 | "CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0" |
377 | "ORS\0" "RS\0*" "RT\0" "FILENAME\0" |
378 | "SUBSEP\0" "ARGIND\0" "ARGC\0" "ARGV\0" |
379 | "ERRNO\0" "FNR\0" |
380 | "NR\0" "NF\0*" "IGNORECASE\0*" |
381 | "ENVIRON\0" "$\0*" "\0"; |
382 | |
383 | static const char vValues[] ALIGN1 = |
384 | "%.6g\0" "%.6g\0" " \0" " \0" |
385 | "\n\0" "\n\0" "\0" "\0" |
386 | "\034\0" |
387 | "\377"; |
388 | |
389 | /* hash size may grow to these values */ |
390 | #define FIRST_PRIME 61 |
391 | static const uint16_t PRIMES[] ALIGN2 = { 251, 1021, 4093, 16381, 65521 }; |
392 | |
393 | |
394 | /* Globals. Split in two parts so that first one is addressed |
395 | * with (mostly short) negative offsets. |
396 | * NB: it's unsafe to put members of type "double" |
397 | * into globals2 (gcc may fail to align them). |
398 | */ |
399 | struct globals { |
400 | double t_double; |
401 | chain beginseq, mainseq, endseq; |
402 | chain *seq; |
403 | node *break_ptr, *continue_ptr; |
404 | rstream *iF; |
405 | xhash *vhash, *ahash, *fdhash, *fnhash; |
406 | const char *g_progname; |
407 | int g_lineno; |
408 | int nfields; |
409 | int maxfields; /* used in fsrealloc() only */ |
410 | var *Fields; |
411 | nvblock *g_cb; |
412 | char *g_pos; |
413 | char *g_buf; |
414 | smallint icase; |
415 | smallint exiting; |
416 | smallint nextrec; |
417 | smallint nextfile; |
418 | smallint is_f0_split; |
419 | }; |
420 | struct globals2 { |
421 | uint32_t t_info; /* often used */ |
422 | uint32_t t_tclass; |
423 | char *t_string; |
424 | int t_lineno; |
425 | int t_rollback; |
426 | |
427 | var *intvar[NUM_INTERNAL_VARS]; /* often used */ |
428 | |
429 | /* former statics from various functions */ |
430 | char *split_f0__fstrings; |
431 | |
432 | uint32_t next_token__save_tclass; |
433 | uint32_t next_token__save_info; |
434 | uint32_t next_token__ltclass; |
435 | smallint next_token__concat_inserted; |
436 | |
437 | smallint next_input_file__files_happen; |
438 | rstream next_input_file__rsm; |
439 | |
440 | var *evaluate__fnargs; |
441 | unsigned evaluate__seed; |
442 | regex_t evaluate__sreg; |
443 | |
444 | var ptest__v; |
445 | |
446 | tsplitter exec_builtin__tspl; |
447 | |
448 | /* biggest and least used members go last */ |
449 | tsplitter fsplitter, rsplitter; |
450 | }; |
451 | #define G1 (ptr_to_globals[-1]) |
452 | #define G (*(struct globals2 *)ptr_to_globals) |
453 | /* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */ |
454 | /*char G1size[sizeof(G1)]; - 0x74 */ |
455 | /*char Gsize[sizeof(G)]; - 0x1c4 */ |
456 | /* Trying to keep most of members accessible with short offsets: */ |
457 | /*char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */ |
458 | #define t_double (G1.t_double ) |
459 | #define beginseq (G1.beginseq ) |
460 | #define mainseq (G1.mainseq ) |
461 | #define endseq (G1.endseq ) |
462 | #define seq (G1.seq ) |
463 | #define break_ptr (G1.break_ptr ) |
464 | #define continue_ptr (G1.continue_ptr) |
465 | #define iF (G1.iF ) |
466 | #define vhash (G1.vhash ) |
467 | #define ahash (G1.ahash ) |
468 | #define fdhash (G1.fdhash ) |
469 | #define fnhash (G1.fnhash ) |
470 | #define g_progname (G1.g_progname ) |
471 | #define g_lineno (G1.g_lineno ) |
472 | #define nfields (G1.nfields ) |
473 | #define maxfields (G1.maxfields ) |
474 | #define Fields (G1.Fields ) |
475 | #define g_cb (G1.g_cb ) |
476 | #define g_pos (G1.g_pos ) |
477 | #define g_buf (G1.g_buf ) |
478 | #define icase (G1.icase ) |
479 | #define exiting (G1.exiting ) |
480 | #define nextrec (G1.nextrec ) |
481 | #define nextfile (G1.nextfile ) |
482 | #define is_f0_split (G1.is_f0_split ) |
483 | #define t_info (G.t_info ) |
484 | #define t_tclass (G.t_tclass ) |
485 | #define t_string (G.t_string ) |
486 | #define t_lineno (G.t_lineno ) |
487 | #define t_rollback (G.t_rollback ) |
488 | #define intvar (G.intvar ) |
489 | #define fsplitter (G.fsplitter ) |
490 | #define rsplitter (G.rsplitter ) |
491 | #define INIT_G() do { \ |
492 | SET_PTR_TO_GLOBALS(xzalloc(sizeof(G1) + sizeof(G)) + sizeof(G1)); \ |
493 | G.next_token__ltclass = TC_OPTERM; \ |
494 | G.evaluate__seed = 1; \ |
495 | } while (0) |
496 | |
497 | |
498 | /* function prototypes */ |
499 | static void handle_special(var *); |
500 | static node *parse_expr(uint32_t); |
501 | static void chain_group(void); |
502 | static var *evaluate(node *, var *); |
503 | static rstream *next_input_file(void); |
504 | static int fmt_num(char *, int, const char *, double, int); |
505 | static int awk_exit(int) NORETURN; |
506 | |
507 | /* ---- error handling ---- */ |
508 | |
509 | static const char EMSG_INTERNAL_ERROR[] ALIGN1 = "Internal error"; |
510 | static const char EMSG_UNEXP_EOS[] ALIGN1 = "Unexpected end of string"; |
511 | static const char EMSG_UNEXP_TOKEN[] ALIGN1 = "Unexpected token"; |
512 | static const char EMSG_DIV_BY_ZERO[] ALIGN1 = "Division by zero"; |
513 | static const char EMSG_INV_FMT[] ALIGN1 = "Invalid format specifier"; |
514 | static const char EMSG_TOO_FEW_ARGS[] ALIGN1 = "Too few arguments for builtin"; |
515 | static const char EMSG_NOT_ARRAY[] ALIGN1 = "Not an array"; |
516 | static const char EMSG_POSSIBLE_ERROR[] ALIGN1 = "Possible syntax error"; |
517 | static const char EMSG_UNDEF_FUNC[] ALIGN1 = "Call to undefined function"; |
518 | #if !ENABLE_FEATURE_AWK_LIBM |
519 | static const char EMSG_NO_MATH[] ALIGN1 = "Math support is not compiled in"; |
520 | #endif |
521 | |
522 | static void zero_out_var(var * vp) |
523 | { |
524 | memset(vp, 0, sizeof(*vp)); |
525 | } |
526 | |
527 | static void syntax_error(const char *const message) NORETURN; |
528 | static void syntax_error(const char *const message) |
529 | { |
530 | bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message); |
531 | } |
532 | |
533 | /* ---- hash stuff ---- */ |
534 | |
535 | static unsigned hashidx(const char *name) |
536 | { |
537 | unsigned idx = 0; |
538 | |
539 | while (*name) idx = *name++ + (idx << 6) - idx; |
540 | return idx; |
541 | } |
542 | |
543 | /* create new hash */ |
544 | static xhash *hash_init(void) |
545 | { |
546 | xhash *newhash; |
547 | |
548 | newhash = xzalloc(sizeof(xhash)); |
549 | newhash->csize = FIRST_PRIME; |
550 | newhash->items = xzalloc(newhash->csize * sizeof(hash_item *)); |
551 | |
552 | return newhash; |
553 | } |
554 | |
555 | /* find item in hash, return ptr to data, NULL if not found */ |
556 | static void *hash_search(xhash *hash, const char *name) |
557 | { |
558 | hash_item *hi; |
559 | |
560 | hi = hash->items [ hashidx(name) % hash->csize ]; |
561 | while (hi) { |
562 | if (strcmp(hi->name, name) == 0) |
563 | return &(hi->data); |
564 | hi = hi->next; |
565 | } |
566 | return NULL; |
567 | } |
568 | |
569 | /* grow hash if it becomes too big */ |
570 | static void hash_rebuild(xhash *hash) |
571 | { |
572 | unsigned newsize, i, idx; |
573 | hash_item **newitems, *hi, *thi; |
574 | |
575 | if (hash->nprime == ARRAY_SIZE(PRIMES)) |
576 | return; |
577 | |
578 | newsize = PRIMES[hash->nprime++]; |
579 | newitems = xzalloc(newsize * sizeof(hash_item *)); |
580 | |
581 | for (i = 0; i < hash->csize; i++) { |
582 | hi = hash->items[i]; |
583 | while (hi) { |
584 | thi = hi; |
585 | hi = thi->next; |
586 | idx = hashidx(thi->name) % newsize; |
587 | thi->next = newitems[idx]; |
588 | newitems[idx] = thi; |
589 | } |
590 | } |
591 | |
592 | free(hash->items); |
593 | hash->csize = newsize; |
594 | hash->items = newitems; |
595 | } |
596 | |
597 | /* find item in hash, add it if necessary. Return ptr to data */ |
598 | static void *hash_find(xhash *hash, const char *name) |
599 | { |
600 | hash_item *hi; |
601 | unsigned idx; |
602 | int l; |
603 | |
604 | hi = hash_search(hash, name); |
605 | if (!hi) { |
606 | if (++hash->nel / hash->csize > 10) |
607 | hash_rebuild(hash); |
608 | |
609 | l = strlen(name) + 1; |
610 | hi = xzalloc(sizeof(hash_item) + l); |
611 | memcpy(hi->name, name, l); |
612 | |
613 | idx = hashidx(name) % hash->csize; |
614 | hi->next = hash->items[idx]; |
615 | hash->items[idx] = hi; |
616 | hash->glen += l; |
617 | } |
618 | return &(hi->data); |
619 | } |
620 | |
621 | #define findvar(hash, name) ((var*) hash_find((hash), (name))) |
622 | #define newvar(name) ((var*) hash_find(vhash, (name))) |
623 | #define newfile(name) ((rstream*)hash_find(fdhash, (name))) |
624 | #define newfunc(name) ((func*) hash_find(fnhash, (name))) |
625 | |
626 | static void hash_remove(xhash *hash, const char *name) |
627 | { |
628 | hash_item *hi, **phi; |
629 | |
630 | phi = &(hash->items[hashidx(name) % hash->csize]); |
631 | while (*phi) { |
632 | hi = *phi; |
633 | if (strcmp(hi->name, name) == 0) { |
634 | hash->glen -= (strlen(name) + 1); |
635 | hash->nel--; |
636 | *phi = hi->next; |
637 | free(hi); |
638 | break; |
639 | } |
640 | phi = &(hi->next); |
641 | } |
642 | } |
643 | |
644 | /* ------ some useful functions ------ */ |
645 | |
646 | static void skip_spaces(char **s) |
647 | { |
648 | char *p = *s; |
649 | |
650 | while (1) { |
651 | if (*p == '\\' && p[1] == '\n') { |
652 | p++; |
653 | t_lineno++; |
654 | } else if (*p != ' ' && *p != '\t') { |
655 | break; |
656 | } |
657 | p++; |
658 | } |
659 | *s = p; |
660 | } |
661 | |
662 | static char *nextword(char **s) |
663 | { |
664 | char *p = *s; |
665 | |
666 | while (*(*s)++) /* */; |
667 | |
668 | return p; |
669 | } |
670 | |
671 | static char nextchar(char **s) |
672 | { |
673 | char c, *pps; |
674 | |
675 | c = *((*s)++); |
676 | pps = *s; |
677 | if (c == '\\') c = bb_process_escape_sequence((const char**)s); |
678 | if (c == '\\' && *s == pps) c = *((*s)++); |
679 | return c; |
680 | } |
681 | |
682 | static ALWAYS_INLINE int isalnum_(int c) |
683 | { |
684 | return (isalnum(c) || c == '_'); |
685 | } |
686 | |
687 | static double my_strtod(char **pp) |
688 | { |
689 | #if ENABLE_DESKTOP |
690 | if ((*pp)[0] == '0' |
691 | && ((((*pp)[1] | 0x20) == 'x') || isdigit((*pp)[1])) |
692 | ) { |
693 | return strtoull(*pp, pp, 0); |
694 | } |
695 | #endif |
696 | return strtod(*pp, pp); |
697 | } |
698 | |
699 | /* -------- working with variables (set/get/copy/etc) -------- */ |
700 | |
701 | static xhash *iamarray(var *v) |
702 | { |
703 | var *a = v; |
704 | |
705 | while (a->type & VF_CHILD) |
706 | a = a->x.parent; |
707 | |
708 | if (!(a->type & VF_ARRAY)) { |
709 | a->type |= VF_ARRAY; |
710 | a->x.array = hash_init(); |
711 | } |
712 | return a->x.array; |
713 | } |
714 | |
715 | static void clear_array(xhash *array) |
716 | { |
717 | unsigned i; |
718 | hash_item *hi, *thi; |
719 | |
720 | for (i = 0; i < array->csize; i++) { |
721 | hi = array->items[i]; |
722 | while (hi) { |
723 | thi = hi; |
724 | hi = hi->next; |
725 | free(thi->data.v.string); |
726 | free(thi); |
727 | } |
728 | array->items[i] = NULL; |
729 | } |
730 | array->glen = array->nel = 0; |
731 | } |
732 | |
733 | /* clear a variable */ |
734 | static var *clrvar(var *v) |
735 | { |
736 | if (!(v->type & VF_FSTR)) |
737 | free(v->string); |
738 | |
739 | v->type &= VF_DONTTOUCH; |
740 | v->type |= VF_DIRTY; |
741 | v->string = NULL; |
742 | return v; |
743 | } |
744 | |
745 | /* assign string value to variable */ |
746 | static var *setvar_p(var *v, char *value) |
747 | { |
748 | clrvar(v); |
749 | v->string = value; |
750 | handle_special(v); |
751 | return v; |
752 | } |
753 | |
754 | /* same as setvar_p but make a copy of string */ |
755 | static var *setvar_s(var *v, const char *value) |
756 | { |
757 | return setvar_p(v, (value && *value) ? xstrdup(value) : NULL); |
758 | } |
759 | |
760 | /* same as setvar_s but set USER flag */ |
761 | static var *setvar_u(var *v, const char *value) |
762 | { |
763 | setvar_s(v, value); |
764 | v->type |= VF_USER; |
765 | return v; |
766 | } |
767 | |
768 | /* set array element to user string */ |
769 | static void setari_u(var *a, int idx, const char *s) |
770 | { |
771 | char sidx[sizeof(int)*3 + 1]; |
772 | var *v; |
773 | |
774 | sprintf(sidx, "%d", idx); |
775 | v = findvar(iamarray(a), sidx); |
776 | setvar_u(v, s); |
777 | } |
778 | |
779 | /* assign numeric value to variable */ |
780 | static var *setvar_i(var *v, double value) |
781 | { |
782 | clrvar(v); |
783 | v->type |= VF_NUMBER; |
784 | v->number = value; |
785 | handle_special(v); |
786 | return v; |
787 | } |
788 | |
789 | static const char *getvar_s(var *v) |
790 | { |
791 | /* if v is numeric and has no cached string, convert it to string */ |
792 | if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) { |
793 | fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE); |
794 | v->string = xstrdup(g_buf); |
795 | v->type |= VF_CACHED; |
796 | } |
797 | return (v->string == NULL) ? "" : v->string; |
798 | } |
799 | |
800 | static double getvar_i(var *v) |
801 | { |
802 | char *s; |
803 | |
804 | if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) { |
805 | v->number = 0; |
806 | s = v->string; |
807 | if (s && *s) { |
808 | v->number = my_strtod(&s); |
809 | if (v->type & VF_USER) { |
810 | skip_spaces(&s); |
811 | if (*s != '\0') |
812 | v->type &= ~VF_USER; |
813 | } |
814 | } else { |
815 | v->type &= ~VF_USER; |
816 | } |
817 | v->type |= VF_CACHED; |
818 | } |
819 | return v->number; |
820 | } |
821 | |
822 | /* Used for operands of bitwise ops */ |
823 | static unsigned long getvar_i_int(var *v) |
824 | { |
825 | double d = getvar_i(v); |
826 | |
827 | /* Casting doubles to longs is undefined for values outside |
828 | * of target type range. Try to widen it as much as possible */ |
829 | if (d >= 0) |
830 | return (unsigned long)d; |
831 | /* Why? Think about d == -4294967295.0 (assuming 32bit longs) */ |
832 | return - (long) (unsigned long) (-d); |
833 | } |
834 | |
835 | static var *copyvar(var *dest, const var *src) |
836 | { |
837 | if (dest != src) { |
838 | clrvar(dest); |
839 | dest->type |= (src->type & ~(VF_DONTTOUCH | VF_FSTR)); |
840 | dest->number = src->number; |
841 | if (src->string) |
842 | dest->string = xstrdup(src->string); |
843 | } |
844 | handle_special(dest); |
845 | return dest; |
846 | } |
847 | |
848 | static var *incvar(var *v) |
849 | { |
850 | return setvar_i(v, getvar_i(v) + 1.); |
851 | } |
852 | |
853 | /* return true if v is number or numeric string */ |
854 | static int is_numeric(var *v) |
855 | { |
856 | getvar_i(v); |
857 | return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY)); |
858 | } |
859 | |
860 | /* return 1 when value of v corresponds to true, 0 otherwise */ |
861 | static int istrue(var *v) |
862 | { |
863 | if (is_numeric(v)) |
864 | return (v->number == 0) ? 0 : 1; |
865 | return (v->string && *(v->string)) ? 1 : 0; |
866 | } |
867 | |
868 | /* temporary variables allocator. Last allocated should be first freed */ |
869 | static var *nvalloc(int n) |
870 | { |
871 | nvblock *pb = NULL; |
872 | var *v, *r; |
873 | int size; |
874 | |
875 | while (g_cb) { |
876 | pb = g_cb; |
877 | if ((g_cb->pos - g_cb->nv) + n <= g_cb->size) break; |
878 | g_cb = g_cb->next; |
879 | } |
880 | |
881 | if (!g_cb) { |
882 | size = (n <= MINNVBLOCK) ? MINNVBLOCK : n; |
883 | g_cb = xzalloc(sizeof(nvblock) + size * sizeof(var)); |
884 | g_cb->size = size; |
885 | g_cb->pos = g_cb->nv; |
886 | g_cb->prev = pb; |
887 | /*g_cb->next = NULL; - xzalloc did it */ |
888 | if (pb) pb->next = g_cb; |
889 | } |
890 | |
891 | v = r = g_cb->pos; |
892 | g_cb->pos += n; |
893 | |
894 | while (v < g_cb->pos) { |
895 | v->type = 0; |
896 | v->string = NULL; |
897 | v++; |
898 | } |
899 | |
900 | return r; |
901 | } |
902 | |
903 | static void nvfree(var *v) |
904 | { |
905 | var *p; |
906 | |
907 | if (v < g_cb->nv || v >= g_cb->pos) |
908 | syntax_error(EMSG_INTERNAL_ERROR); |
909 | |
910 | for (p = v; p < g_cb->pos; p++) { |
911 | if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) { |
912 | clear_array(iamarray(p)); |
913 | free(p->x.array->items); |
914 | free(p->x.array); |
915 | } |
916 | if (p->type & VF_WALK) |
917 | free(p->x.walker); |
918 | |
919 | clrvar(p); |
920 | } |
921 | |
922 | g_cb->pos = v; |
923 | while (g_cb->prev && g_cb->pos == g_cb->nv) { |
924 | g_cb = g_cb->prev; |
925 | } |
926 | } |
927 | |
928 | /* ------- awk program text parsing ------- */ |
929 | |
930 | /* Parse next token pointed by global pos, place results into global ttt. |
931 | * If token isn't expected, give away. Return token class |
932 | */ |
933 | static uint32_t next_token(uint32_t expected) |
934 | { |
935 | #define concat_inserted (G.next_token__concat_inserted) |
936 | #define save_tclass (G.next_token__save_tclass) |
937 | #define save_info (G.next_token__save_info) |
938 | /* Initialized to TC_OPTERM: */ |
939 | #define ltclass (G.next_token__ltclass) |
940 | |
941 | char *p, *pp, *s; |
942 | const char *tl; |
943 | uint32_t tc; |
944 | const uint32_t *ti; |
945 | int l; |
946 | |
947 | if (t_rollback) { |
948 | t_rollback = FALSE; |
949 | |
950 | } else if (concat_inserted) { |
951 | concat_inserted = FALSE; |
952 | t_tclass = save_tclass; |
953 | t_info = save_info; |
954 | |
955 | } else { |
956 | p = g_pos; |
957 | readnext: |
958 | skip_spaces(&p); |
959 | g_lineno = t_lineno; |
960 | if (*p == '#') |
961 | while (*p != '\n' && *p != '\0') |
962 | p++; |
963 | |
964 | if (*p == '\n') |
965 | t_lineno++; |
966 | |
967 | if (*p == '\0') { |
968 | tc = TC_EOF; |
969 | |
970 | } else if (*p == '\"') { |
971 | /* it's a string */ |
972 | t_string = s = ++p; |
973 | while (*p != '\"') { |
974 | if (*p == '\0' || *p == '\n') |
975 | syntax_error(EMSG_UNEXP_EOS); |
976 | *(s++) = nextchar(&p); |
977 | } |
978 | p++; |
979 | *s = '\0'; |
980 | tc = TC_STRING; |
981 | |
982 | } else if ((expected & TC_REGEXP) && *p == '/') { |
983 | /* it's regexp */ |
984 | t_string = s = ++p; |
985 | while (*p != '/') { |
986 | if (*p == '\0' || *p == '\n') |
987 | syntax_error(EMSG_UNEXP_EOS); |
988 | *s = *p++; |
989 | if (*s++ == '\\') { |
990 | pp = p; |
991 | *(s-1) = bb_process_escape_sequence((const char **)&p); |
992 | if (*pp == '\\') |
993 | *s++ = '\\'; |
994 | if (p == pp) |
995 | *s++ = *p++; |
996 | } |
997 | } |
998 | p++; |
999 | *s = '\0'; |
1000 | tc = TC_REGEXP; |
1001 | |
1002 | } else if (*p == '.' || isdigit(*p)) { |
1003 | /* it's a number */ |
1004 | t_double = my_strtod(&p); |
1005 | if (*p == '.') |
1006 | syntax_error(EMSG_UNEXP_TOKEN); |
1007 | tc = TC_NUMBER; |
1008 | |
1009 | } else { |
1010 | /* search for something known */ |
1011 | tl = tokenlist; |
1012 | tc = 0x00000001; |
1013 | ti = tokeninfo; |
1014 | while (*tl) { |
1015 | l = *(tl++); |
1016 | if (l == NTCC) { |
1017 | tc <<= 1; |
1018 | continue; |
1019 | } |
1020 | /* if token class is expected, token |
1021 | * matches and it's not a longer word, |
1022 | * then this is what we are looking for |
1023 | */ |
1024 | if ((tc & (expected | TC_WORD | TC_NEWLINE)) |
1025 | && *tl == *p && strncmp(p, tl, l) == 0 |
1026 | && !((tc & TC_WORD) && isalnum_(p[l])) |
1027 | ) { |
1028 | t_info = *ti; |
1029 | p += l; |
1030 | break; |
1031 | } |
1032 | ti++; |
1033 | tl += l; |
1034 | } |
1035 | |
1036 | if (!*tl) { |
1037 | /* it's a name (var/array/function), |
1038 | * otherwise it's something wrong |
1039 | */ |
1040 | if (!isalnum_(*p)) |
1041 | syntax_error(EMSG_UNEXP_TOKEN); |
1042 | |
1043 | t_string = --p; |
1044 | while (isalnum_(*(++p))) { |
1045 | *(p-1) = *p; |
1046 | } |
1047 | *(p-1) = '\0'; |
1048 | tc = TC_VARIABLE; |
1049 | /* also consume whitespace between functionname and bracket */ |
1050 | if (!(expected & TC_VARIABLE)) |
1051 | skip_spaces(&p); |
1052 | if (*p == '(') { |
1053 | tc = TC_FUNCTION; |
1054 | } else { |
1055 | if (*p == '[') { |
1056 | p++; |
1057 | tc = TC_ARRAY; |
1058 | } |
1059 | } |
1060 | } |
1061 | } |
1062 | g_pos = p; |
1063 | |
1064 | /* skipping newlines in some cases */ |
1065 | if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE)) |
1066 | goto readnext; |
1067 | |
1068 | /* insert concatenation operator when needed */ |
1069 | if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) { |
1070 | concat_inserted = TRUE; |
1071 | save_tclass = tc; |
1072 | save_info = t_info; |
1073 | tc = TC_BINOP; |
1074 | t_info = OC_CONCAT | SS | P(35); |
1075 | } |
1076 | |
1077 | t_tclass = tc; |
1078 | } |
1079 | ltclass = t_tclass; |
1080 | |
1081 | /* Are we ready for this? */ |
1082 | if (!(ltclass & expected)) |
1083 | syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ? |
1084 | EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN); |
1085 | |
1086 | return ltclass; |
1087 | #undef concat_inserted |
1088 | #undef save_tclass |
1089 | #undef save_info |
1090 | #undef ltclass |
1091 | } |
1092 | |
1093 | static void rollback_token(void) |
1094 | { |
1095 | t_rollback = TRUE; |
1096 | } |
1097 | |
1098 | static node *new_node(uint32_t info) |
1099 | { |
1100 | node *n; |
1101 | |
1102 | n = xzalloc(sizeof(node)); |
1103 | n->info = info; |
1104 | n->lineno = g_lineno; |
1105 | return n; |
1106 | } |
1107 | |
1108 | static node *mk_re_node(const char *s, node *n, regex_t *re) |
1109 | { |
1110 | n->info = OC_REGEXP; |
1111 | n->l.re = re; |
1112 | n->r.ire = re + 1; |
1113 | xregcomp(re, s, REG_EXTENDED); |
1114 | xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE); |
1115 | |
1116 | return n; |
1117 | } |
1118 | |
1119 | static node *condition(void) |
1120 | { |
1121 | next_token(TC_SEQSTART); |
1122 | return parse_expr(TC_SEQTERM); |
1123 | } |
1124 | |
1125 | /* parse expression terminated by given argument, return ptr |
1126 | * to built subtree. Terminator is eaten by parse_expr */ |
1127 | static node *parse_expr(uint32_t iexp) |
1128 | { |
1129 | node sn; |
1130 | node *cn = &sn; |
1131 | node *vn, *glptr; |
1132 | uint32_t tc, xtc; |
1133 | var *v; |
1134 | |
1135 | sn.info = PRIMASK; |
1136 | sn.r.n = glptr = NULL; |
1137 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp; |
1138 | |
1139 | while (!((tc = next_token(xtc)) & iexp)) { |
1140 | if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) { |
1141 | /* input redirection (<) attached to glptr node */ |
1142 | cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37)); |
1143 | cn->a.n = glptr; |
1144 | xtc = TC_OPERAND | TC_UOPPRE; |
1145 | glptr = NULL; |
1146 | |
1147 | } else if (tc & (TC_BINOP | TC_UOPPOST)) { |
1148 | /* for binary and postfix-unary operators, jump back over |
1149 | * previous operators with higher priority */ |
1150 | vn = cn; |
1151 | while ( ((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2)) |
1152 | || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON)) ) |
1153 | vn = vn->a.n; |
1154 | if ((t_info & OPCLSMASK) == OC_TERNARY) |
1155 | t_info += P(6); |
1156 | cn = vn->a.n->r.n = new_node(t_info); |
1157 | cn->a.n = vn->a.n; |
1158 | if (tc & TC_BINOP) { |
1159 | cn->l.n = vn; |
1160 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
1161 | if ((t_info & OPCLSMASK) == OC_PGETLINE) { |
1162 | /* it's a pipe */ |
1163 | next_token(TC_GETLINE); |
1164 | /* give maximum priority to this pipe */ |
1165 | cn->info &= ~PRIMASK; |
1166 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1167 | } |
1168 | } else { |
1169 | cn->r.n = vn; |
1170 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1171 | } |
1172 | vn->a.n = cn; |
1173 | |
1174 | } else { |
1175 | /* for operands and prefix-unary operators, attach them |
1176 | * to last node */ |
1177 | vn = cn; |
1178 | cn = vn->r.n = new_node(t_info); |
1179 | cn->a.n = vn; |
1180 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
1181 | if (tc & (TC_OPERAND | TC_REGEXP)) { |
1182 | xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp; |
1183 | /* one should be very careful with switch on tclass - |
1184 | * only simple tclasses should be used! */ |
1185 | switch (tc) { |
1186 | case TC_VARIABLE: |
1187 | case TC_ARRAY: |
1188 | cn->info = OC_VAR; |
1189 | v = hash_search(ahash, t_string); |
1190 | if (v != NULL) { |
1191 | cn->info = OC_FNARG; |
1192 | cn->l.i = v->x.aidx; |
1193 | } else { |
1194 | cn->l.v = newvar(t_string); |
1195 | } |
1196 | if (tc & TC_ARRAY) { |
1197 | cn->info |= xS; |
1198 | cn->r.n = parse_expr(TC_ARRTERM); |
1199 | } |
1200 | break; |
1201 | |
1202 | case TC_NUMBER: |
1203 | case TC_STRING: |
1204 | cn->info = OC_VAR; |
1205 | v = cn->l.v = xzalloc(sizeof(var)); |
1206 | if (tc & TC_NUMBER) |
1207 | setvar_i(v, t_double); |
1208 | else |
1209 | setvar_s(v, t_string); |
1210 | break; |
1211 | |
1212 | case TC_REGEXP: |
1213 | mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2)); |
1214 | break; |
1215 | |
1216 | case TC_FUNCTION: |
1217 | cn->info = OC_FUNC; |
1218 | cn->r.f = newfunc(t_string); |
1219 | cn->l.n = condition(); |
1220 | break; |
1221 | |
1222 | case TC_SEQSTART: |
1223 | cn = vn->r.n = parse_expr(TC_SEQTERM); |
1224 | cn->a.n = vn; |
1225 | break; |
1226 | |
1227 | case TC_GETLINE: |
1228 | glptr = cn; |
1229 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
1230 | break; |
1231 | |
1232 | case TC_BUILTIN: |
1233 | cn->l.n = condition(); |
1234 | break; |
1235 | } |
1236 | } |
1237 | } |
1238 | } |
1239 | return sn.r.n; |
1240 | } |
1241 | |
1242 | /* add node to chain. Return ptr to alloc'd node */ |
1243 | static node *chain_node(uint32_t info) |
1244 | { |
1245 | node *n; |
1246 | |
1247 | if (!seq->first) |
1248 | seq->first = seq->last = new_node(0); |
1249 | |
1250 | if (seq->programname != g_progname) { |
1251 | seq->programname = g_progname; |
1252 | n = chain_node(OC_NEWSOURCE); |
1253 | n->l.s = xstrdup(g_progname); |
1254 | } |
1255 | |
1256 | n = seq->last; |
1257 | n->info = info; |
1258 | seq->last = n->a.n = new_node(OC_DONE); |
1259 | |
1260 | return n; |
1261 | } |
1262 | |
1263 | static void chain_expr(uint32_t info) |
1264 | { |
1265 | node *n; |
1266 | |
1267 | n = chain_node(info); |
1268 | n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
1269 | if (t_tclass & TC_GRPTERM) |
1270 | rollback_token(); |
1271 | } |
1272 | |
1273 | static node *chain_loop(node *nn) |
1274 | { |
1275 | node *n, *n2, *save_brk, *save_cont; |
1276 | |
1277 | save_brk = break_ptr; |
1278 | save_cont = continue_ptr; |
1279 | |
1280 | n = chain_node(OC_BR | Vx); |
1281 | continue_ptr = new_node(OC_EXEC); |
1282 | break_ptr = new_node(OC_EXEC); |
1283 | chain_group(); |
1284 | n2 = chain_node(OC_EXEC | Vx); |
1285 | n2->l.n = nn; |
1286 | n2->a.n = n; |
1287 | continue_ptr->a.n = n2; |
1288 | break_ptr->a.n = n->r.n = seq->last; |
1289 | |
1290 | continue_ptr = save_cont; |
1291 | break_ptr = save_brk; |
1292 | |
1293 | return n; |
1294 | } |
1295 | |
1296 | /* parse group and attach it to chain */ |
1297 | static void chain_group(void) |
1298 | { |
1299 | uint32_t c; |
1300 | node *n, *n2, *n3; |
1301 | |
1302 | do { |
1303 | c = next_token(TC_GRPSEQ); |
1304 | } while (c & TC_NEWLINE); |
1305 | |
1306 | if (c & TC_GRPSTART) { |
1307 | while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) { |
1308 | if (t_tclass & TC_NEWLINE) continue; |
1309 | rollback_token(); |
1310 | chain_group(); |
1311 | } |
1312 | } else if (c & (TC_OPSEQ | TC_OPTERM)) { |
1313 | rollback_token(); |
1314 | chain_expr(OC_EXEC | Vx); |
1315 | } else { /* TC_STATEMNT */ |
1316 | switch (t_info & OPCLSMASK) { |
1317 | case ST_IF: |
1318 | n = chain_node(OC_BR | Vx); |
1319 | n->l.n = condition(); |
1320 | chain_group(); |
1321 | n2 = chain_node(OC_EXEC); |
1322 | n->r.n = seq->last; |
1323 | if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE) == TC_ELSE) { |
1324 | chain_group(); |
1325 | n2->a.n = seq->last; |
1326 | } else { |
1327 | rollback_token(); |
1328 | } |
1329 | break; |
1330 | |
1331 | case ST_WHILE: |
1332 | n2 = condition(); |
1333 | n = chain_loop(NULL); |
1334 | n->l.n = n2; |
1335 | break; |
1336 | |
1337 | case ST_DO: |
1338 | n2 = chain_node(OC_EXEC); |
1339 | n = chain_loop(NULL); |
1340 | n2->a.n = n->a.n; |
1341 | next_token(TC_WHILE); |
1342 | n->l.n = condition(); |
1343 | break; |
1344 | |
1345 | case ST_FOR: |
1346 | next_token(TC_SEQSTART); |
1347 | n2 = parse_expr(TC_SEMICOL | TC_SEQTERM); |
1348 | if (t_tclass & TC_SEQTERM) { /* for-in */ |
1349 | if ((n2->info & OPCLSMASK) != OC_IN) |
1350 | syntax_error(EMSG_UNEXP_TOKEN); |
1351 | n = chain_node(OC_WALKINIT | VV); |
1352 | n->l.n = n2->l.n; |
1353 | n->r.n = n2->r.n; |
1354 | n = chain_loop(NULL); |
1355 | n->info = OC_WALKNEXT | Vx; |
1356 | n->l.n = n2->l.n; |
1357 | } else { /* for (;;) */ |
1358 | n = chain_node(OC_EXEC | Vx); |
1359 | n->l.n = n2; |
1360 | n2 = parse_expr(TC_SEMICOL); |
1361 | n3 = parse_expr(TC_SEQTERM); |
1362 | n = chain_loop(n3); |
1363 | n->l.n = n2; |
1364 | if (!n2) |
1365 | n->info = OC_EXEC; |
1366 | } |
1367 | break; |
1368 | |
1369 | case OC_PRINT: |
1370 | case OC_PRINTF: |
1371 | n = chain_node(t_info); |
1372 | n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM); |
1373 | if (t_tclass & TC_OUTRDR) { |
1374 | n->info |= t_info; |
1375 | n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
1376 | } |
1377 | if (t_tclass & TC_GRPTERM) |
1378 | rollback_token(); |
1379 | break; |
1380 | |
1381 | case OC_BREAK: |
1382 | n = chain_node(OC_EXEC); |
1383 | n->a.n = break_ptr; |
1384 | break; |
1385 | |
1386 | case OC_CONTINUE: |
1387 | n = chain_node(OC_EXEC); |
1388 | n->a.n = continue_ptr; |
1389 | break; |
1390 | |
1391 | /* delete, next, nextfile, return, exit */ |
1392 | default: |
1393 | chain_expr(t_info); |
1394 | } |
1395 | } |
1396 | } |
1397 | |
1398 | static void parse_program(char *p) |
1399 | { |
1400 | uint32_t tclass; |
1401 | node *cn; |
1402 | func *f; |
1403 | var *v; |
1404 | |
1405 | g_pos = p; |
1406 | t_lineno = 1; |
1407 | while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART | |
1408 | TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) { |
1409 | |
1410 | if (tclass & TC_OPTERM) |
1411 | continue; |
1412 | |
1413 | seq = &mainseq; |
1414 | if (tclass & TC_BEGIN) { |
1415 | seq = &beginseq; |
1416 | chain_group(); |
1417 | |
1418 | } else if (tclass & TC_END) { |
1419 | seq = &endseq; |
1420 | chain_group(); |
1421 | |
1422 | } else if (tclass & TC_FUNCDECL) { |
1423 | next_token(TC_FUNCTION); |
1424 | g_pos++; |
1425 | f = newfunc(t_string); |
1426 | f->body.first = NULL; |
1427 | f->nargs = 0; |
1428 | while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) { |
1429 | v = findvar(ahash, t_string); |
1430 | v->x.aidx = (f->nargs)++; |
1431 | |
1432 | if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM) |
1433 | break; |
1434 | } |
1435 | seq = &(f->body); |
1436 | chain_group(); |
1437 | clear_array(ahash); |
1438 | |
1439 | } else if (tclass & TC_OPSEQ) { |
1440 | rollback_token(); |
1441 | cn = chain_node(OC_TEST); |
1442 | cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART); |
1443 | if (t_tclass & TC_GRPSTART) { |
1444 | rollback_token(); |
1445 | chain_group(); |
1446 | } else { |
1447 | chain_node(OC_PRINT); |
1448 | } |
1449 | cn->r.n = mainseq.last; |
1450 | |
1451 | } else /* if (tclass & TC_GRPSTART) */ { |
1452 | rollback_token(); |
1453 | chain_group(); |
1454 | } |
1455 | } |
1456 | } |
1457 | |
1458 | |
1459 | /* -------- program execution part -------- */ |
1460 | |
1461 | static node *mk_splitter(const char *s, tsplitter *spl) |
1462 | { |
1463 | regex_t *re, *ire; |
1464 | node *n; |
1465 | |
1466 | re = &spl->re[0]; |
1467 | ire = &spl->re[1]; |
1468 | n = &spl->n; |
1469 | if ((n->info & OPCLSMASK) == OC_REGEXP) { |
1470 | regfree(re); |
1471 | regfree(ire); // TODO: nuke ire, use re+1? |
1472 | } |
1473 | if (strlen(s) > 1) { |
1474 | mk_re_node(s, n, re); |
1475 | } else { |
1476 | n->info = (uint32_t) *s; |
1477 | } |
1478 | |
1479 | return n; |
1480 | } |
1481 | |
1482 | /* use node as a regular expression. Supplied with node ptr and regex_t |
1483 | * storage space. Return ptr to regex (if result points to preg, it should |
1484 | * be later regfree'd manually |
1485 | */ |
1486 | static regex_t *as_regex(node *op, regex_t *preg) |
1487 | { |
1488 | var *v; |
1489 | const char *s; |
1490 | |
1491 | if ((op->info & OPCLSMASK) == OC_REGEXP) { |
1492 | return icase ? op->r.ire : op->l.re; |
1493 | } |
1494 | v = nvalloc(1); |
1495 | s = getvar_s(evaluate(op, v)); |
1496 | xregcomp(preg, s, icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED); |
1497 | nvfree(v); |
1498 | return preg; |
1499 | } |
1500 | |
1501 | /* gradually increasing buffer */ |
1502 | static void qrealloc(char **b, int n, int *size) |
1503 | { |
1504 | if (!*b || n >= *size) { |
1505 | *size = n + (n>>1) + 80; |
1506 | *b = xrealloc(*b, *size); |
1507 | } |
1508 | } |
1509 | |
1510 | /* resize field storage space */ |
1511 | static void fsrealloc(int size) |
1512 | { |
1513 | int i; |
1514 | |
1515 | if (size >= maxfields) { |
1516 | i = maxfields; |
1517 | maxfields = size + 16; |
1518 | Fields = xrealloc(Fields, maxfields * sizeof(var)); |
1519 | for (; i < maxfields; i++) { |
1520 | Fields[i].type = VF_SPECIAL; |
1521 | Fields[i].string = NULL; |
1522 | } |
1523 | } |
1524 | |
1525 | if (size < nfields) { |
1526 | for (i = size; i < nfields; i++) { |
1527 | clrvar(Fields + i); |
1528 | } |
1529 | } |
1530 | nfields = size; |
1531 | } |
1532 | |
1533 | static int awk_split(const char *s, node *spl, char **slist) |
1534 | { |
1535 | int l, n = 0; |
1536 | char c[4]; |
1537 | char *s1; |
1538 | regmatch_t pmatch[2]; // TODO: why [2]? [1] is enough... |
1539 | |
1540 | /* in worst case, each char would be a separate field */ |
1541 | *slist = s1 = xzalloc(strlen(s) * 2 + 3); |
1542 | strcpy(s1, s); |
1543 | |
1544 | c[0] = c[1] = (char)spl->info; |
1545 | c[2] = c[3] = '\0'; |
1546 | if (*getvar_s(intvar[RS]) == '\0') |
1547 | c[2] = '\n'; |
1548 | |
1549 | if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */ |
1550 | if (!*s) |
1551 | return n; /* "": zero fields */ |
1552 | n++; /* at least one field will be there */ |
1553 | do { |
1554 | l = strcspn(s, c+2); /* len till next NUL or \n */ |
1555 | if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0 |
1556 | && pmatch[0].rm_so <= l |
1557 | ) { |
1558 | l = pmatch[0].rm_so; |
1559 | if (pmatch[0].rm_eo == 0) { |
1560 | l++; |
1561 | pmatch[0].rm_eo++; |
1562 | } |
1563 | n++; /* we saw yet another delimiter */ |
1564 | } else { |
1565 | pmatch[0].rm_eo = l; |
1566 | if (s[l]) pmatch[0].rm_eo++; |
1567 | } |
1568 | memcpy(s1, s, l); |
1569 | s1[l] = '\0'; |
1570 | nextword(&s1); |
1571 | s += pmatch[0].rm_eo; |
1572 | } while (*s); |
1573 | return n; |
1574 | } |
1575 | if (c[0] == '\0') { /* null split */ |
1576 | while (*s) { |
1577 | *s1++ = *s++; |
1578 | *s1++ = '\0'; |
1579 | n++; |
1580 | } |
1581 | return n; |
1582 | } |
1583 | if (c[0] != ' ') { /* single-character split */ |
1584 | if (icase) { |
1585 | c[0] = toupper(c[0]); |
1586 | c[1] = tolower(c[1]); |
1587 | } |
1588 | if (*s1) n++; |
1589 | while ((s1 = strpbrk(s1, c))) { |
1590 | *s1++ = '\0'; |
1591 | n++; |
1592 | } |
1593 | return n; |
1594 | } |
1595 | /* space split */ |
1596 | while (*s) { |
1597 | s = skip_whitespace(s); |
1598 | if (!*s) break; |
1599 | n++; |
1600 | while (*s && !isspace(*s)) |
1601 | *s1++ = *s++; |
1602 | *s1++ = '\0'; |
1603 | } |
1604 | return n; |
1605 | } |
1606 | |
1607 | static void split_f0(void) |
1608 | { |
1609 | /* static char *fstrings; */ |
1610 | #define fstrings (G.split_f0__fstrings) |
1611 | |
1612 | int i, n; |
1613 | char *s; |
1614 | |
1615 | if (is_f0_split) |
1616 | return; |
1617 | |
1618 | is_f0_split = TRUE; |
1619 | free(fstrings); |
1620 | fsrealloc(0); |
1621 | n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings); |
1622 | fsrealloc(n); |
1623 | s = fstrings; |
1624 | for (i = 0; i < n; i++) { |
1625 | Fields[i].string = nextword(&s); |
1626 | Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY); |
1627 | } |
1628 | |
1629 | /* set NF manually to avoid side effects */ |
1630 | clrvar(intvar[NF]); |
1631 | intvar[NF]->type = VF_NUMBER | VF_SPECIAL; |
1632 | intvar[NF]->number = nfields; |
1633 | #undef fstrings |
1634 | } |
1635 | |
1636 | /* perform additional actions when some internal variables changed */ |
1637 | static void handle_special(var *v) |
1638 | { |
1639 | int n; |
1640 | char *b; |
1641 | const char *sep, *s; |
1642 | int sl, l, len, i, bsize; |
1643 | |
1644 | if (!(v->type & VF_SPECIAL)) |
1645 | return; |
1646 | |
1647 | if (v == intvar[NF]) { |
1648 | n = (int)getvar_i(v); |
1649 | fsrealloc(n); |
1650 | |
1651 | /* recalculate $0 */ |
1652 | sep = getvar_s(intvar[OFS]); |
1653 | sl = strlen(sep); |
1654 | b = NULL; |
1655 | len = 0; |
1656 | for (i = 0; i < n; i++) { |
1657 | s = getvar_s(&Fields[i]); |
1658 | l = strlen(s); |
1659 | if (b) { |
1660 | memcpy(b+len, sep, sl); |
1661 | len += sl; |
1662 | } |
1663 | qrealloc(&b, len+l+sl, &bsize); |
1664 | memcpy(b+len, s, l); |
1665 | len += l; |
1666 | } |
1667 | if (b) |
1668 | b[len] = '\0'; |
1669 | setvar_p(intvar[F0], b); |
1670 | is_f0_split = TRUE; |
1671 | |
1672 | } else if (v == intvar[F0]) { |
1673 | is_f0_split = FALSE; |
1674 | |
1675 | } else if (v == intvar[FS]) { |
1676 | mk_splitter(getvar_s(v), &fsplitter); |
1677 | |
1678 | } else if (v == intvar[RS]) { |
1679 | mk_splitter(getvar_s(v), &rsplitter); |
1680 | |
1681 | } else if (v == intvar[IGNORECASE]) { |
1682 | icase = istrue(v); |
1683 | |
1684 | } else { /* $n */ |
1685 | n = getvar_i(intvar[NF]); |
1686 | setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1); |
1687 | /* right here v is invalid. Just to note... */ |
1688 | } |
1689 | } |
1690 | |
1691 | /* step through func/builtin/etc arguments */ |
1692 | static node *nextarg(node **pn) |
1693 | { |
1694 | node *n; |
1695 | |
1696 | n = *pn; |
1697 | if (n && (n->info & OPCLSMASK) == OC_COMMA) { |
1698 | *pn = n->r.n; |
1699 | n = n->l.n; |
1700 | } else { |
1701 | *pn = NULL; |
1702 | } |
1703 | return n; |
1704 | } |
1705 | |
1706 | static void hashwalk_init(var *v, xhash *array) |
1707 | { |
1708 | char **w; |
1709 | hash_item *hi; |
1710 | unsigned i; |
1711 | |
1712 | if (v->type & VF_WALK) |
1713 | free(v->x.walker); |
1714 | |
1715 | v->type |= VF_WALK; |
1716 | w = v->x.walker = xzalloc(2 + 2*sizeof(char *) + array->glen); |
1717 | w[0] = w[1] = (char *)(w + 2); |
1718 | for (i = 0; i < array->csize; i++) { |
1719 | hi = array->items[i]; |
1720 | while (hi) { |
1721 | strcpy(*w, hi->name); |
1722 | nextword(w); |
1723 | hi = hi->next; |
1724 | } |
1725 | } |
1726 | } |
1727 | |
1728 | static int hashwalk_next(var *v) |
1729 | { |
1730 | char **w; |
1731 | |
1732 | w = v->x.walker; |
1733 | if (w[1] == w[0]) |
1734 | return FALSE; |
1735 | |
1736 | setvar_s(v, nextword(w+1)); |
1737 | return TRUE; |
1738 | } |
1739 | |
1740 | /* evaluate node, return 1 when result is true, 0 otherwise */ |
1741 | static int ptest(node *pattern) |
1742 | { |
1743 | /* ptest__v is "static": to save stack space? */ |
1744 | return istrue(evaluate(pattern, &G.ptest__v)); |
1745 | } |
1746 | |
1747 | /* read next record from stream rsm into a variable v */ |
1748 | static int awk_getline(rstream *rsm, var *v) |
1749 | { |
1750 | char *b; |
1751 | regmatch_t pmatch[2]; |
1752 | int a, p, pp=0, size; |
1753 | int fd, so, eo, r, rp; |
1754 | char c, *m, *s; |
1755 | |
1756 | /* we're using our own buffer since we need access to accumulating |
1757 | * characters |
1758 | */ |
1759 | fd = fileno(rsm->F); |
1760 | m = rsm->buffer; |
1761 | a = rsm->adv; |
1762 | p = rsm->pos; |
1763 | size = rsm->size; |
1764 | c = (char) rsplitter.n.info; |
1765 | rp = 0; |
1766 | |
1767 | if (!m) qrealloc(&m, 256, &size); |
1768 | do { |
1769 | b = m + a; |
1770 | so = eo = p; |
1771 | r = 1; |
1772 | if (p > 0) { |
1773 | if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) { |
1774 | if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re, |
1775 | b, 1, pmatch, 0) == 0) { |
1776 | so = pmatch[0].rm_so; |
1777 | eo = pmatch[0].rm_eo; |
1778 | if (b[eo] != '\0') |
1779 | break; |
1780 | } |
1781 | } else if (c != '\0') { |
1782 | s = strchr(b+pp, c); |
1783 | if (!s) s = memchr(b+pp, '\0', p - pp); |
1784 | if (s) { |
1785 | so = eo = s-b; |
1786 | eo++; |
1787 | break; |
1788 | } |
1789 | } else { |
1790 | while (b[rp] == '\n') |
1791 | rp++; |
1792 | s = strstr(b+rp, "\n\n"); |
1793 | if (s) { |
1794 | so = eo = s-b; |
1795 | while (b[eo] == '\n') eo++; |
1796 | if (b[eo] != '\0') |
1797 | break; |
1798 | } |
1799 | } |
1800 | } |
1801 | |
1802 | if (a > 0) { |
1803 | memmove(m, (const void *)(m+a), p+1); |
1804 | b = m; |
1805 | a = 0; |
1806 | } |
1807 | |
1808 | qrealloc(&m, a+p+128, &size); |
1809 | b = m + a; |
1810 | pp = p; |
1811 | p += safe_read(fd, b+p, size-p-1); |
1812 | if (p < pp) { |
1813 | p = 0; |
1814 | r = 0; |
1815 | setvar_i(intvar[ERRNO], errno); |
1816 | } |
1817 | b[p] = '\0'; |
1818 | |
1819 | } while (p > pp); |
1820 | |
1821 | if (p == 0) { |
1822 | r--; |
1823 | } else { |
1824 | c = b[so]; b[so] = '\0'; |
1825 | setvar_s(v, b+rp); |
1826 | v->type |= VF_USER; |
1827 | b[so] = c; |
1828 | c = b[eo]; b[eo] = '\0'; |
1829 | setvar_s(intvar[RT], b+so); |
1830 | b[eo] = c; |
1831 | } |
1832 | |
1833 | rsm->buffer = m; |
1834 | rsm->adv = a + eo; |
1835 | rsm->pos = p - eo; |
1836 | rsm->size = size; |
1837 | |
1838 | return r; |
1839 | } |
1840 | |
1841 | static int fmt_num(char *b, int size, const char *format, double n, int int_as_int) |
1842 | { |
1843 | int r = 0; |
1844 | char c; |
1845 | const char *s = format; |
1846 | |
1847 | if (int_as_int && n == (int)n) { |
1848 | r = snprintf(b, size, "%d", (int)n); |
1849 | } else { |
1850 | do { c = *s; } while (c && *++s); |
1851 | if (strchr("diouxX", c)) { |
1852 | r = snprintf(b, size, format, (int)n); |
1853 | } else if (strchr("eEfgG", c)) { |
1854 | r = snprintf(b, size, format, n); |
1855 | } else { |
1856 | syntax_error(EMSG_INV_FMT); |
1857 | } |
1858 | } |
1859 | return r; |
1860 | } |
1861 | |
1862 | |
1863 | /* formatted output into an allocated buffer, return ptr to buffer */ |
1864 | static char *awk_printf(node *n) |
1865 | { |
1866 | char *b = NULL; |
1867 | char *fmt, *s, *f; |
1868 | const char *s1; |
1869 | int i, j, incr, bsize; |
1870 | char c, c1; |
1871 | var *v, *arg; |
1872 | |
1873 | v = nvalloc(1); |
1874 | fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v))); |
1875 | |
1876 | i = 0; |
1877 | while (*f) { |
1878 | s = f; |
1879 | while (*f && (*f != '%' || *(++f) == '%')) |
1880 | f++; |
1881 | while (*f && !isalpha(*f)) { |
1882 | if (*f == '*') |
1883 | syntax_error("%*x formats are not supported"); |
1884 | f++; |
1885 | } |
1886 | |
1887 | incr = (f - s) + MAXVARFMT; |
1888 | qrealloc(&b, incr + i, &bsize); |
1889 | c = *f; |
1890 | if (c != '\0') f++; |
1891 | c1 = *f; |
1892 | *f = '\0'; |
1893 | arg = evaluate(nextarg(&n), v); |
1894 | |
1895 | j = i; |
1896 | if (c == 'c' || !c) { |
1897 | i += sprintf(b+i, s, is_numeric(arg) ? |
1898 | (char)getvar_i(arg) : *getvar_s(arg)); |
1899 | } else if (c == 's') { |
1900 | s1 = getvar_s(arg); |
1901 | qrealloc(&b, incr+i+strlen(s1), &bsize); |
1902 | i += sprintf(b+i, s, s1); |
1903 | } else { |
1904 | i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE); |
1905 | } |
1906 | *f = c1; |
1907 | |
1908 | /* if there was an error while sprintf, return value is negative */ |
1909 | if (i < j) i = j; |
1910 | } |
1911 | |
1912 | b = xrealloc(b, i + 1); |
1913 | free(fmt); |
1914 | nvfree(v); |
1915 | b[i] = '\0'; |
1916 | return b; |
1917 | } |
1918 | |
1919 | /* common substitution routine |
1920 | * replace (nm) substring of (src) that match (n) with (repl), store |
1921 | * result into (dest), return number of substitutions. If nm=0, replace |
1922 | * all matches. If src or dst is NULL, use $0. If ex=TRUE, enable |
1923 | * subexpression matching (\1-\9) |
1924 | */ |
1925 | static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int ex) |
1926 | { |
1927 | char *ds = NULL; |
1928 | const char *s; |
1929 | const char *sp; |
1930 | int c, i, j, di, rl, so, eo, nbs, n, dssize; |
1931 | regmatch_t pmatch[10]; |
1932 | regex_t sreg, *re; |
1933 | |
1934 | re = as_regex(rn, &sreg); |
1935 | if (!src) src = intvar[F0]; |
1936 | if (!dest) dest = intvar[F0]; |
1937 | |
1938 | i = di = 0; |
1939 | sp = getvar_s(src); |
1940 | rl = strlen(repl); |
1941 | while (regexec(re, sp, 10, pmatch, sp==getvar_s(src) ? 0 : REG_NOTBOL) == 0) { |
1942 | so = pmatch[0].rm_so; |
1943 | eo = pmatch[0].rm_eo; |
1944 | |
1945 | qrealloc(&ds, di + eo + rl, &dssize); |
1946 | memcpy(ds + di, sp, eo); |
1947 | di += eo; |
1948 | if (++i >= nm) { |
1949 | /* replace */ |
1950 | di -= (eo - so); |
1951 | nbs = 0; |
1952 | for (s = repl; *s; s++) { |
1953 | ds[di++] = c = *s; |
1954 | if (c == '\\') { |
1955 | nbs++; |
1956 | continue; |
1957 | } |
1958 | if (c == '&' || (ex && c >= '0' && c <= '9')) { |
1959 | di -= ((nbs + 3) >> 1); |
1960 | j = 0; |
1961 | if (c != '&') { |
1962 | j = c - '0'; |
1963 | nbs++; |
1964 | } |
1965 | if (nbs % 2) { |
1966 | ds[di++] = c; |
1967 | } else { |
1968 | n = pmatch[j].rm_eo - pmatch[j].rm_so; |
1969 | qrealloc(&ds, di + rl + n, &dssize); |
1970 | memcpy(ds + di, sp + pmatch[j].rm_so, n); |
1971 | di += n; |
1972 | } |
1973 | } |
1974 | nbs = 0; |
1975 | } |
1976 | } |
1977 | |
1978 | sp += eo; |
1979 | if (i == nm) break; |
1980 | if (eo == so) { |
1981 | ds[di] = *sp++; |
1982 | if (!ds[di++]) break; |
1983 | } |
1984 | } |
1985 | |
1986 | qrealloc(&ds, di + strlen(sp), &dssize); |
1987 | strcpy(ds + di, sp); |
1988 | setvar_p(dest, ds); |
1989 | if (re == &sreg) regfree(re); |
1990 | return i; |
1991 | } |
1992 | |
1993 | static var *exec_builtin(node *op, var *res) |
1994 | { |
1995 | #define tspl (G.exec_builtin__tspl) |
1996 | |
1997 | int (*to_xxx)(int); |
1998 | var *tv; |
1999 | node *an[4]; |
2000 | var *av[4]; |
2001 | const char *as[4]; |
2002 | regmatch_t pmatch[2]; |
2003 | regex_t sreg, *re; |
2004 | node *spl; |
2005 | uint32_t isr, info; |
2006 | int nargs; |
2007 | time_t tt; |
2008 | char *s, *s1; |
2009 | int i, l, ll, n; |
2010 | |
2011 | tv = nvalloc(4); |
2012 | isr = info = op->info; |
2013 | op = op->l.n; |
2014 | |
2015 | av[2] = av[3] = NULL; |
2016 | for (i = 0; i < 4 && op; i++) { |
2017 | an[i] = nextarg(&op); |
2018 | if (isr & 0x09000000) av[i] = evaluate(an[i], &tv[i]); |
2019 | if (isr & 0x08000000) as[i] = getvar_s(av[i]); |
2020 | isr >>= 1; |
2021 | } |
2022 | |
2023 | nargs = i; |
2024 | if ((uint32_t)nargs < (info >> 30)) |
2025 | syntax_error(EMSG_TOO_FEW_ARGS); |
2026 | |
2027 | switch (info & OPNMASK) { |
2028 | |
2029 | case B_a2: |
2030 | #if ENABLE_FEATURE_AWK_LIBM |
2031 | setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1]))); |
2032 | #else |
2033 | syntax_error(EMSG_NO_MATH); |
2034 | #endif |
2035 | break; |
2036 | |
2037 | case B_sp: |
2038 | if (nargs > 2) { |
2039 | spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ? |
2040 | an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl); |
2041 | } else { |
2042 | spl = &fsplitter.n; |
2043 | } |
2044 | |
2045 | n = awk_split(as[0], spl, &s); |
2046 | s1 = s; |
2047 | clear_array(iamarray(av[1])); |
2048 | for (i = 1; i <= n; i++) |
2049 | setari_u(av[1], i, nextword(&s1)); |
2050 | free(s); |
2051 | setvar_i(res, n); |
2052 | break; |
2053 | |
2054 | case B_ss: |
2055 | l = strlen(as[0]); |
2056 | i = getvar_i(av[1]) - 1; |
2057 | if (i > l) i = l; |
2058 | if (i < 0) i = 0; |
2059 | n = (nargs > 2) ? getvar_i(av[2]) : l-i; |
2060 | if (n < 0) n = 0; |
2061 | s = xstrndup(as[0]+i, n); |
2062 | setvar_p(res, s); |
2063 | break; |
2064 | |
2065 | /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5: |
2066 | * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */ |
2067 | case B_an: |
2068 | setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1])); |
2069 | break; |
2070 | |
2071 | case B_co: |
2072 | setvar_i(res, ~getvar_i_int(av[0])); |
2073 | break; |
2074 | |
2075 | case B_ls: |
2076 | setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1])); |
2077 | break; |
2078 | |
2079 | case B_or: |
2080 | setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1])); |
2081 | break; |
2082 | |
2083 | case B_rs: |
2084 | setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1])); |
2085 | break; |
2086 | |
2087 | case B_xo: |
2088 | setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1])); |
2089 | break; |
2090 | |
2091 | case B_lo: |
2092 | to_xxx = tolower; |
2093 | goto lo_cont; |
2094 | |
2095 | case B_up: |
2096 | to_xxx = toupper; |
2097 | lo_cont: |
2098 | s1 = s = xstrdup(as[0]); |
2099 | while (*s1) { |
2100 | *s1 = (*to_xxx)(*s1); |
2101 | s1++; |
2102 | } |
2103 | setvar_p(res, s); |
2104 | break; |
2105 | |
2106 | case B_ix: |
2107 | n = 0; |
2108 | ll = strlen(as[1]); |
2109 | l = strlen(as[0]) - ll; |
2110 | if (ll > 0 && l >= 0) { |
2111 | if (!icase) { |
2112 | s = strstr(as[0], as[1]); |
2113 | if (s) n = (s - as[0]) + 1; |
2114 | } else { |
2115 | /* this piece of code is terribly slow and |
2116 | * really should be rewritten |
2117 | */ |
2118 | for (i=0; i<=l; i++) { |
2119 | if (strncasecmp(as[0]+i, as[1], ll) == 0) { |
2120 | n = i+1; |
2121 | break; |
2122 | } |
2123 | } |
2124 | } |
2125 | } |
2126 | setvar_i(res, n); |
2127 | break; |
2128 | |
2129 | case B_ti: |
2130 | if (nargs > 1) |
2131 | tt = getvar_i(av[1]); |
2132 | else |
2133 | time(&tt); |
2134 | //s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"; |
2135 | i = strftime(g_buf, MAXVARFMT, |
2136 | ((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"), |
2137 | localtime(&tt)); |
2138 | g_buf[i] = '\0'; |
2139 | setvar_s(res, g_buf); |
2140 | break; |
2141 | |
2142 | case B_ma: |
2143 | re = as_regex(an[1], &sreg); |
2144 | n = regexec(re, as[0], 1, pmatch, 0); |
2145 | if (n == 0) { |
2146 | pmatch[0].rm_so++; |
2147 | pmatch[0].rm_eo++; |
2148 | } else { |
2149 | pmatch[0].rm_so = 0; |
2150 | pmatch[0].rm_eo = -1; |
2151 | } |
2152 | setvar_i(newvar("RSTART"), pmatch[0].rm_so); |
2153 | setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so); |
2154 | setvar_i(res, pmatch[0].rm_so); |
2155 | if (re == &sreg) regfree(re); |
2156 | break; |
2157 | |
2158 | case B_ge: |
2159 | awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE); |
2160 | break; |
2161 | |
2162 | case B_gs: |
2163 | setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE)); |
2164 | break; |
2165 | |
2166 | case B_su: |
2167 | setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE)); |
2168 | break; |
2169 | } |
2170 | |
2171 | nvfree(tv); |
2172 | return res; |
2173 | #undef tspl |
2174 | } |
2175 | |
2176 | /* |
2177 | * Evaluate node - the heart of the program. Supplied with subtree |
2178 | * and place where to store result. returns ptr to result. |
2179 | */ |
2180 | #define XC(n) ((n) >> 8) |
2181 | |
2182 | static var *evaluate(node *op, var *res) |
2183 | { |
2184 | /* This procedure is recursive so we should count every byte */ |
2185 | #define fnargs (G.evaluate__fnargs) |
2186 | /* seed is initialized to 1 */ |
2187 | #define seed (G.evaluate__seed) |
2188 | #define sreg (G.evaluate__sreg) |
2189 | |
2190 | node *op1; |
2191 | var *v1; |
2192 | union { |
2193 | var *v; |
2194 | const char *s; |
2195 | double d; |
2196 | int i; |
2197 | } L, R; |
2198 | uint32_t opinfo; |
2199 | int opn; |
2200 | union { |
2201 | char *s; |
2202 | rstream *rsm; |
2203 | FILE *F; |
2204 | var *v; |
2205 | regex_t *re; |
2206 | uint32_t info; |
2207 | } X; |
2208 | |
2209 | if (!op) |
2210 | return setvar_s(res, NULL); |
2211 | |
2212 | v1 = nvalloc(2); |
2213 | |
2214 | while (op) { |
2215 | opinfo = op->info; |
2216 | opn = (opinfo & OPNMASK); |
2217 | g_lineno = op->lineno; |
2218 | |
2219 | /* execute inevitable things */ |
2220 | op1 = op->l.n; |
2221 | if (opinfo & OF_RES1) X.v = L.v = evaluate(op1, v1); |
2222 | if (opinfo & OF_RES2) R.v = evaluate(op->r.n, v1+1); |
2223 | if (opinfo & OF_STR1) L.s = getvar_s(L.v); |
2224 | if (opinfo & OF_STR2) R.s = getvar_s(R.v); |
2225 | if (opinfo & OF_NUM1) L.d = getvar_i(L.v); |
2226 | |
2227 | switch (XC(opinfo & OPCLSMASK)) { |
2228 | |
2229 | /* -- iterative node type -- */ |
2230 | |
2231 | /* test pattern */ |
2232 | case XC( OC_TEST ): |
2233 | if ((op1->info & OPCLSMASK) == OC_COMMA) { |
2234 | /* it's range pattern */ |
2235 | if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) { |
2236 | op->info |= OF_CHECKED; |
2237 | if (ptest(op1->r.n)) |
2238 | op->info &= ~OF_CHECKED; |
2239 | |
2240 | op = op->a.n; |
2241 | } else { |
2242 | op = op->r.n; |
2243 | } |
2244 | } else { |
2245 | op = (ptest(op1)) ? op->a.n : op->r.n; |
2246 | } |
2247 | break; |
2248 | |
2249 | /* just evaluate an expression, also used as unconditional jump */ |
2250 | case XC( OC_EXEC ): |
2251 | break; |
2252 | |
2253 | /* branch, used in if-else and various loops */ |
2254 | case XC( OC_BR ): |
2255 | op = istrue(L.v) ? op->a.n : op->r.n; |
2256 | break; |
2257 | |
2258 | /* initialize for-in loop */ |
2259 | case XC( OC_WALKINIT ): |
2260 | hashwalk_init(L.v, iamarray(R.v)); |
2261 | break; |
2262 | |
2263 | /* get next array item */ |
2264 | case XC( OC_WALKNEXT ): |
2265 | op = hashwalk_next(L.v) ? op->a.n : op->r.n; |
2266 | break; |
2267 | |
2268 | case XC( OC_PRINT ): |
2269 | case XC( OC_PRINTF ): |
2270 | X.F = stdout; |
2271 | if (op->r.n) { |
2272 | X.rsm = newfile(R.s); |
2273 | if (!X.rsm->F) { |
2274 | if (opn == '|') { |
2275 | X.rsm->F = popen(R.s, "w"); |
2276 | if (X.rsm->F == NULL) |
2277 | bb_perror_msg_and_die("popen"); |
2278 | X.rsm->is_pipe = 1; |
2279 | } else { |
2280 | X.rsm->F = xfopen(R.s, opn=='w' ? "w" : "a"); |
2281 | } |
2282 | } |
2283 | X.F = X.rsm->F; |
2284 | } |
2285 | |
2286 | if ((opinfo & OPCLSMASK) == OC_PRINT) { |
2287 | if (!op1) { |
2288 | fputs(getvar_s(intvar[F0]), X.F); |
2289 | } else { |
2290 | while (op1) { |
2291 | L.v = evaluate(nextarg(&op1), v1); |
2292 | if (L.v->type & VF_NUMBER) { |
2293 | fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]), |
2294 | getvar_i(L.v), TRUE); |
2295 | fputs(g_buf, X.F); |
2296 | } else { |
2297 | fputs(getvar_s(L.v), X.F); |
2298 | } |
2299 | |
2300 | if (op1) fputs(getvar_s(intvar[OFS]), X.F); |
2301 | } |
2302 | } |
2303 | fputs(getvar_s(intvar[ORS]), X.F); |
2304 | |
2305 | } else { /* OC_PRINTF */ |
2306 | L.s = awk_printf(op1); |
2307 | fputs(L.s, X.F); |
2308 | free((char*)L.s); |
2309 | } |
2310 | fflush(X.F); |
2311 | break; |
2312 | |
2313 | case XC( OC_DELETE ): |
2314 | X.info = op1->info & OPCLSMASK; |
2315 | if (X.info == OC_VAR) { |
2316 | R.v = op1->l.v; |
2317 | } else if (X.info == OC_FNARG) { |
2318 | R.v = &fnargs[op1->l.i]; |
2319 | } else { |
2320 | syntax_error(EMSG_NOT_ARRAY); |
2321 | } |
2322 | |
2323 | if (op1->r.n) { |
2324 | clrvar(L.v); |
2325 | L.s = getvar_s(evaluate(op1->r.n, v1)); |
2326 | hash_remove(iamarray(R.v), L.s); |
2327 | } else { |
2328 | clear_array(iamarray(R.v)); |
2329 | } |
2330 | break; |
2331 | |
2332 | case XC( OC_NEWSOURCE ): |
2333 | g_progname = op->l.s; |
2334 | break; |
2335 | |
2336 | case XC( OC_RETURN ): |
2337 | copyvar(res, L.v); |
2338 | break; |
2339 | |
2340 | case XC( OC_NEXTFILE ): |
2341 | nextfile = TRUE; |
2342 | case XC( OC_NEXT ): |
2343 | nextrec = TRUE; |
2344 | case XC( OC_DONE ): |
2345 | clrvar(res); |
2346 | break; |
2347 | |
2348 | case XC( OC_EXIT ): |
2349 | awk_exit(L.d); |
2350 | |
2351 | /* -- recursive node type -- */ |
2352 | |
2353 | case XC( OC_VAR ): |
2354 | L.v = op->l.v; |
2355 | if (L.v == intvar[NF]) |
2356 | split_f0(); |
2357 | goto v_cont; |
2358 | |
2359 | case XC( OC_FNARG ): |
2360 | L.v = &fnargs[op->l.i]; |
2361 | v_cont: |
2362 | res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v; |
2363 | break; |
2364 | |
2365 | case XC( OC_IN ): |
2366 | setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0); |
2367 | break; |
2368 | |
2369 | case XC( OC_REGEXP ): |
2370 | op1 = op; |
2371 | L.s = getvar_s(intvar[F0]); |
2372 | goto re_cont; |
2373 | |
2374 | case XC( OC_MATCH ): |
2375 | op1 = op->r.n; |
2376 | re_cont: |
2377 | X.re = as_regex(op1, &sreg); |
2378 | R.i = regexec(X.re, L.s, 0, NULL, 0); |
2379 | if (X.re == &sreg) regfree(X.re); |
2380 | setvar_i(res, (R.i == 0 ? 1 : 0) ^ (opn == '!' ? 1 : 0)); |
2381 | break; |
2382 | |
2383 | case XC( OC_MOVE ): |
2384 | /* if source is a temporary string, jusk relink it to dest */ |
2385 | if (R.v == v1+1 && R.v->string) { |
2386 | res = setvar_p(L.v, R.v->string); |
2387 | R.v->string = NULL; |
2388 | } else { |
2389 | res = copyvar(L.v, R.v); |
2390 | } |
2391 | break; |
2392 | |
2393 | case XC( OC_TERNARY ): |
2394 | if ((op->r.n->info & OPCLSMASK) != OC_COLON) |
2395 | syntax_error(EMSG_POSSIBLE_ERROR); |
2396 | res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res); |
2397 | break; |
2398 | |
2399 | case XC( OC_FUNC ): |
2400 | if (!op->r.f->body.first) |
2401 | syntax_error(EMSG_UNDEF_FUNC); |
2402 | |
2403 | X.v = R.v = nvalloc(op->r.f->nargs+1); |
2404 | while (op1) { |
2405 | L.v = evaluate(nextarg(&op1), v1); |
2406 | copyvar(R.v, L.v); |
2407 | R.v->type |= VF_CHILD; |
2408 | R.v->x.parent = L.v; |
2409 | if (++R.v - X.v >= op->r.f->nargs) |
2410 | break; |
2411 | } |
2412 | |
2413 | R.v = fnargs; |
2414 | fnargs = X.v; |
2415 | |
2416 | L.s = g_progname; |
2417 | res = evaluate(op->r.f->body.first, res); |
2418 | g_progname = L.s; |
2419 | |
2420 | nvfree(fnargs); |
2421 | fnargs = R.v; |
2422 | break; |
2423 | |
2424 | case XC( OC_GETLINE ): |
2425 | case XC( OC_PGETLINE ): |
2426 | if (op1) { |
2427 | X.rsm = newfile(L.s); |
2428 | if (!X.rsm->F) { |
2429 | if ((opinfo & OPCLSMASK) == OC_PGETLINE) { |
2430 | X.rsm->F = popen(L.s, "r"); |
2431 | X.rsm->is_pipe = TRUE; |
2432 | } else { |
2433 | X.rsm->F = fopen_for_read(L.s); /* not xfopen! */ |
2434 | } |
2435 | } |
2436 | } else { |
2437 | if (!iF) iF = next_input_file(); |
2438 | X.rsm = iF; |
2439 | } |
2440 | |
2441 | if (!X.rsm->F) { |
2442 | setvar_i(intvar[ERRNO], errno); |
2443 | setvar_i(res, -1); |
2444 | break; |
2445 | } |
2446 | |
2447 | if (!op->r.n) |
2448 | R.v = intvar[F0]; |
2449 | |
2450 | L.i = awk_getline(X.rsm, R.v); |
2451 | if (L.i > 0) { |
2452 | if (!op1) { |
2453 | incvar(intvar[FNR]); |
2454 | incvar(intvar[NR]); |
2455 | } |
2456 | } |
2457 | setvar_i(res, L.i); |
2458 | break; |
2459 | |
2460 | /* simple builtins */ |
2461 | case XC( OC_FBLTIN ): |
2462 | switch (opn) { |
2463 | |
2464 | case F_in: |
2465 | R.d = (int)L.d; |
2466 | break; |
2467 | |
2468 | case F_rn: |
2469 | R.d = (double)rand() / (double)RAND_MAX; |
2470 | break; |
2471 | #if ENABLE_FEATURE_AWK_LIBM |
2472 | case F_co: |
2473 | R.d = cos(L.d); |
2474 | break; |
2475 | |
2476 | case F_ex: |
2477 | R.d = exp(L.d); |
2478 | break; |
2479 | |
2480 | case F_lg: |
2481 | R.d = log(L.d); |
2482 | break; |
2483 | |
2484 | case F_si: |
2485 | R.d = sin(L.d); |
2486 | break; |
2487 | |
2488 | case F_sq: |
2489 | R.d = sqrt(L.d); |
2490 | break; |
2491 | #else |
2492 | case F_co: |
2493 | case F_ex: |
2494 | case F_lg: |
2495 | case F_si: |
2496 | case F_sq: |
2497 | syntax_error(EMSG_NO_MATH); |
2498 | break; |
2499 | #endif |
2500 | case F_sr: |
2501 | R.d = (double)seed; |
2502 | seed = op1 ? (unsigned)L.d : (unsigned)time(NULL); |
2503 | srand(seed); |
2504 | break; |
2505 | |
2506 | case F_ti: |
2507 | R.d = time(NULL); |
2508 | break; |
2509 | |
2510 | case F_le: |
2511 | if (!op1) |
2512 | L.s = getvar_s(intvar[F0]); |
2513 | R.d = strlen(L.s); |
2514 | break; |
2515 | |
2516 | case F_sy: |
2517 | fflush(NULL); |
2518 | R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s) |
2519 | ? (system(L.s) >> 8) : 0; |
2520 | break; |
2521 | |
2522 | case F_ff: |
2523 | if (!op1) |
2524 | fflush(stdout); |
2525 | else { |
2526 | if (L.s && *L.s) { |
2527 | X.rsm = newfile(L.s); |
2528 | fflush(X.rsm->F); |
2529 | } else { |
2530 | fflush(NULL); |
2531 | } |
2532 | } |
2533 | break; |
2534 | |
2535 | case F_cl: |
2536 | X.rsm = (rstream *)hash_search(fdhash, L.s); |
2537 | if (X.rsm) { |
2538 | R.i = X.rsm->is_pipe ? pclose(X.rsm->F) : fclose(X.rsm->F); |
2539 | free(X.rsm->buffer); |
2540 | hash_remove(fdhash, L.s); |
2541 | } |
2542 | if (R.i != 0) |
2543 | setvar_i(intvar[ERRNO], errno); |
2544 | R.d = (double)R.i; |
2545 | break; |
2546 | } |
2547 | setvar_i(res, R.d); |
2548 | break; |
2549 | |
2550 | case XC( OC_BUILTIN ): |
2551 | res = exec_builtin(op, res); |
2552 | break; |
2553 | |
2554 | case XC( OC_SPRINTF ): |
2555 | setvar_p(res, awk_printf(op1)); |
2556 | break; |
2557 | |
2558 | case XC( OC_UNARY ): |
2559 | X.v = R.v; |
2560 | L.d = R.d = getvar_i(R.v); |
2561 | switch (opn) { |
2562 | case 'P': |
2563 | L.d = ++R.d; |
2564 | goto r_op_change; |
2565 | case 'p': |
2566 | R.d++; |
2567 | goto r_op_change; |
2568 | case 'M': |
2569 | L.d = --R.d; |
2570 | goto r_op_change; |
2571 | case 'm': |
2572 | R.d--; |
2573 | goto r_op_change; |
2574 | case '!': |
2575 | L.d = istrue(X.v) ? 0 : 1; |
2576 | break; |
2577 | case '-': |
2578 | L.d = -R.d; |
2579 | break; |
2580 | r_op_change: |
2581 | setvar_i(X.v, R.d); |
2582 | } |
2583 | setvar_i(res, L.d); |
2584 | break; |
2585 | |
2586 | case XC( OC_FIELD ): |
2587 | R.i = (int)getvar_i(R.v); |
2588 | if (R.i == 0) { |
2589 | res = intvar[F0]; |
2590 | } else { |
2591 | split_f0(); |
2592 | if (R.i > nfields) |
2593 | fsrealloc(R.i); |
2594 | res = &Fields[R.i - 1]; |
2595 | } |
2596 | break; |
2597 | |
2598 | /* concatenation (" ") and index joining (",") */ |
2599 | case XC( OC_CONCAT ): |
2600 | case XC( OC_COMMA ): |
2601 | opn = strlen(L.s) + strlen(R.s) + 2; |
2602 | X.s = xmalloc(opn); |
2603 | strcpy(X.s, L.s); |
2604 | if ((opinfo & OPCLSMASK) == OC_COMMA) { |
2605 | L.s = getvar_s(intvar[SUBSEP]); |
2606 | X.s = xrealloc(X.s, opn + strlen(L.s)); |
2607 | strcat(X.s, L.s); |
2608 | } |
2609 | strcat(X.s, R.s); |
2610 | setvar_p(res, X.s); |
2611 | break; |
2612 | |
2613 | case XC( OC_LAND ): |
2614 | setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0); |
2615 | break; |
2616 | |
2617 | case XC( OC_LOR ): |
2618 | setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n)); |
2619 | break; |
2620 | |
2621 | case XC( OC_BINARY ): |
2622 | case XC( OC_REPLACE ): |
2623 | R.d = getvar_i(R.v); |
2624 | switch (opn) { |
2625 | case '+': |
2626 | L.d += R.d; |
2627 | break; |
2628 | case '-': |
2629 | L.d -= R.d; |
2630 | break; |
2631 | case '*': |
2632 | L.d *= R.d; |
2633 | break; |
2634 | case '/': |
2635 | if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO); |
2636 | L.d /= R.d; |
2637 | break; |
2638 | case '&': |
2639 | #if ENABLE_FEATURE_AWK_LIBM |
2640 | L.d = pow(L.d, R.d); |
2641 | #else |
2642 | syntax_error(EMSG_NO_MATH); |
2643 | #endif |
2644 | break; |
2645 | case '%': |
2646 | if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO); |
2647 | L.d -= (int)(L.d / R.d) * R.d; |
2648 | break; |
2649 | } |
2650 | res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : X.v, L.d); |
2651 | break; |
2652 | |
2653 | case XC( OC_COMPARE ): |
2654 | if (is_numeric(L.v) && is_numeric(R.v)) { |
2655 | L.d = getvar_i(L.v) - getvar_i(R.v); |
2656 | } else { |
2657 | L.s = getvar_s(L.v); |
2658 | R.s = getvar_s(R.v); |
2659 | L.d = icase ? strcasecmp(L.s, R.s) : strcmp(L.s, R.s); |
2660 | } |
2661 | switch (opn & 0xfe) { |
2662 | case 0: |
2663 | R.i = (L.d > 0); |
2664 | break; |
2665 | case 2: |
2666 | R.i = (L.d >= 0); |
2667 | break; |
2668 | case 4: |
2669 | R.i = (L.d == 0); |
2670 | break; |
2671 | } |
2672 | setvar_i(res, (opn & 0x1 ? R.i : !R.i) ? 1 : 0); |
2673 | break; |
2674 | |
2675 | default: |
2676 | syntax_error(EMSG_POSSIBLE_ERROR); |
2677 | } |
2678 | if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS) |
2679 | op = op->a.n; |
2680 | if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS) |
2681 | break; |
2682 | if (nextrec) |
2683 | break; |
2684 | } |
2685 | nvfree(v1); |
2686 | return res; |
2687 | #undef fnargs |
2688 | #undef seed |
2689 | #undef sreg |
2690 | } |
2691 | |
2692 | |
2693 | /* -------- main & co. -------- */ |
2694 | |
2695 | static int awk_exit(int r) |
2696 | { |
2697 | var tv; |
2698 | unsigned i; |
2699 | hash_item *hi; |
2700 | |
2701 | zero_out_var(&tv); |
2702 | |
2703 | if (!exiting) { |
2704 | exiting = TRUE; |
2705 | nextrec = FALSE; |
2706 | evaluate(endseq.first, &tv); |
2707 | } |
2708 | |
2709 | /* waiting for children */ |
2710 | for (i = 0; i < fdhash->csize; i++) { |
2711 | hi = fdhash->items[i]; |
2712 | while (hi) { |
2713 | if (hi->data.rs.F && hi->data.rs.is_pipe) |
2714 | pclose(hi->data.rs.F); |
2715 | hi = hi->next; |
2716 | } |
2717 | } |
2718 | |
2719 | exit(r); |
2720 | } |
2721 | |
2722 | /* if expr looks like "var=value", perform assignment and return 1, |
2723 | * otherwise return 0 */ |
2724 | static int is_assignment(const char *expr) |
2725 | { |
2726 | char *exprc, *s, *s0, *s1; |
2727 | |
2728 | exprc = xstrdup(expr); |
2729 | if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) { |
2730 | free(exprc); |
2731 | return FALSE; |
2732 | } |
2733 | |
2734 | *(s++) = '\0'; |
2735 | s0 = s1 = s; |
2736 | while (*s) |
2737 | *(s1++) = nextchar(&s); |
2738 | |
2739 | *s1 = '\0'; |
2740 | setvar_u(newvar(exprc), s0); |
2741 | free(exprc); |
2742 | return TRUE; |
2743 | } |
2744 | |
2745 | /* switch to next input file */ |
2746 | static rstream *next_input_file(void) |
2747 | { |
2748 | #define rsm (G.next_input_file__rsm) |
2749 | #define files_happen (G.next_input_file__files_happen) |
2750 | |
2751 | FILE *F = NULL; |
2752 | const char *fname, *ind; |
2753 | |
2754 | if (rsm.F) fclose(rsm.F); |
2755 | rsm.F = NULL; |
2756 | rsm.pos = rsm.adv = 0; |
2757 | |
2758 | do { |
2759 | if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) { |
2760 | if (files_happen) |
2761 | return NULL; |
2762 | fname = "-"; |
2763 | F = stdin; |
2764 | } else { |
2765 | ind = getvar_s(incvar(intvar[ARGIND])); |
2766 | fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind)); |
2767 | if (fname && *fname && !is_assignment(fname)) |
2768 | F = xfopen_stdin(fname); |
2769 | } |
2770 | } while (!F); |
2771 | |
2772 | files_happen = TRUE; |
2773 | setvar_s(intvar[FILENAME], fname); |
2774 | rsm.F = F; |
2775 | return &rsm; |
2776 | #undef rsm |
2777 | #undef files_happen |
2778 | } |
2779 | |
2780 | int awk_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
2781 | int awk_main(int argc, char **argv) |
2782 | { |
2783 | unsigned opt; |
2784 | char *opt_F, *opt_W; |
2785 | llist_t *list_v = NULL; |
2786 | llist_t *list_f = NULL; |
2787 | int i, j; |
2788 | var *v; |
2789 | var tv; |
2790 | char **envp; |
2791 | char *vnames = (char *)vNames; /* cheat */ |
2792 | char *vvalues = (char *)vValues; |
2793 | |
2794 | INIT_G(); |
2795 | |
2796 | /* Undo busybox.c, or else strtod may eat ','! This breaks parsing: |
2797 | * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */ |
2798 | if (ENABLE_LOCALE_SUPPORT) |
2799 | setlocale(LC_NUMERIC, "C"); |
2800 | |
2801 | zero_out_var(&tv); |
2802 | |
2803 | /* allocate global buffer */ |
2804 | g_buf = xmalloc(MAXVARFMT + 1); |
2805 | |
2806 | vhash = hash_init(); |
2807 | ahash = hash_init(); |
2808 | fdhash = hash_init(); |
2809 | fnhash = hash_init(); |
2810 | |
2811 | /* initialize variables */ |
2812 | for (i = 0; *vnames; i++) { |
2813 | intvar[i] = v = newvar(nextword(&vnames)); |
2814 | if (*vvalues != '\377') |
2815 | setvar_s(v, nextword(&vvalues)); |
2816 | else |
2817 | setvar_i(v, 0); |
2818 | |
2819 | if (*vnames == '*') { |
2820 | v->type |= VF_SPECIAL; |
2821 | vnames++; |
2822 | } |
2823 | } |
2824 | |
2825 | handle_special(intvar[FS]); |
2826 | handle_special(intvar[RS]); |
2827 | |
2828 | newfile("/dev/stdin")->F = stdin; |
2829 | newfile("/dev/stdout")->F = stdout; |
2830 | newfile("/dev/stderr")->F = stderr; |
2831 | |
2832 | /* Huh, people report that sometimes environ is NULL. Oh well. */ |
2833 | if (environ) for (envp = environ; *envp; envp++) { |
2834 | /* environ is writable, thus we don't strdup it needlessly */ |
2835 | char *s = *envp; |
2836 | char *s1 = strchr(s, '='); |
2837 | if (s1) { |
2838 | *s1 = '\0'; |
2839 | /* Both findvar and setvar_u take const char* |
2840 | * as 2nd arg -> environment is not trashed */ |
2841 | setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1); |
2842 | *s1 = '='; |
2843 | } |
2844 | } |
2845 | opt_complementary = "v::f::"; /* -v and -f can occur multiple times */ |
2846 | opt = getopt32(argv, "F:v:f:W:", &opt_F, &list_v, &list_f, &opt_W); |
2847 | argv += optind; |
2848 | argc -= optind; |
2849 | if (opt & 0x1) |
2850 | setvar_s(intvar[FS], opt_F); // -F |
2851 | while (list_v) { /* -v */ |
2852 | if (!is_assignment(llist_pop(&list_v))) |
2853 | bb_show_usage(); |
2854 | } |
2855 | if (list_f) { /* -f */ |
2856 | do { |
2857 | char *s = NULL; |
2858 | FILE *from_file; |
2859 | |
2860 | g_progname = llist_pop(&list_f); |
2861 | from_file = xfopen_stdin(g_progname); |
2862 | /* one byte is reserved for some trick in next_token */ |
2863 | for (i = j = 1; j > 0; i += j) { |
2864 | s = xrealloc(s, i + 4096); |
2865 | j = fread(s + i, 1, 4094, from_file); |
2866 | } |
2867 | s[i] = '\0'; |
2868 | fclose(from_file); |
2869 | parse_program(s + 1); |
2870 | free(s); |
2871 | } while (list_f); |
2872 | } else { // no -f: take program from 1st parameter |
2873 | if (!argc) |
2874 | bb_show_usage(); |
2875 | g_progname = "cmd. line"; |
2876 | parse_program(*argv++); |
2877 | argc--; |
2878 | } |
2879 | if (opt & 0x8) // -W |
2880 | bb_error_msg("warning: unrecognized option '-W %s' ignored", opt_W); |
2881 | |
2882 | /* fill in ARGV array */ |
2883 | setvar_i(intvar[ARGC], argc + 1); |
2884 | setari_u(intvar[ARGV], 0, "awk"); |
2885 | i = 0; |
2886 | while (*argv) |
2887 | setari_u(intvar[ARGV], ++i, *argv++); |
2888 | |
2889 | evaluate(beginseq.first, &tv); |
2890 | if (!mainseq.first && !endseq.first) |
2891 | awk_exit(EXIT_SUCCESS); |
2892 | |
2893 | /* input file could already be opened in BEGIN block */ |
2894 | if (!iF) iF = next_input_file(); |
2895 | |
2896 | /* passing through input files */ |
2897 | while (iF) { |
2898 | nextfile = FALSE; |
2899 | setvar_i(intvar[FNR], 0); |
2900 | |
2901 | while ((i = awk_getline(iF, intvar[F0])) > 0) { |
2902 | nextrec = FALSE; |
2903 | incvar(intvar[NR]); |
2904 | incvar(intvar[FNR]); |
2905 | evaluate(mainseq.first, &tv); |
2906 | |
2907 | if (nextfile) |
2908 | break; |
2909 | } |
2910 | |
2911 | if (i < 0) |
2912 | syntax_error(strerror(errno)); |
2913 | |
2914 | iF = next_input_file(); |
2915 | } |
2916 | |
2917 | awk_exit(EXIT_SUCCESS); |
2918 | /*return 0;*/ |
2919 | } |