blob: 397d70f4cbb830c95918f809b7843ac05a97379c [file] [log] [blame]
Glenn L McGrath545106f2002-11-11 06:21:00 +00001/* 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 *
Bernhard Reutner-Fischer86f5c992006-01-22 22:55:11 +00007 * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
Glenn L McGrath545106f2002-11-11 06:21:00 +00008 */
9
Denis Vlasenkob6adbf12007-05-26 19:00:18 +000010#include "libbb.h"
Rob Landleyd921b2e2006-08-03 15:41:12 +000011#include "xregex.h"
12#include <math.h>
Glenn L McGrath545106f2002-11-11 06:21:00 +000013
Denis Vlasenko99912ca2007-04-10 15:43:37 +000014/* This is a NOEXEC applet. Be very careful! */
15
Glenn L McGrath545106f2002-11-11 06:21:00 +000016
Denys Vlasenkoda62b092010-03-11 12:13:18 +010017/* 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
Denis Vlasenko629563b2007-02-24 17:05:52 +000027#define MAXVARFMT 240
28#define MINNVBLOCK 64
Glenn L McGrath545106f2002-11-11 06:21:00 +000029
30/* variable flags */
Denis Vlasenko629563b2007-02-24 17:05:52 +000031#define VF_NUMBER 0x0001 /* 1 = primary type is number */
32#define VF_ARRAY 0x0002 /* 1 = it's an array */
Glenn L McGrath545106f2002-11-11 06:21:00 +000033
Denis Vlasenko629563b2007-02-24 17:05:52 +000034#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 */
Glenn L McGrath545106f2002-11-11 06:21:00 +000041
42/* these flags are static, don't change them when value is changed */
Denis Vlasenko629563b2007-02-24 17:05:52 +000043#define VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY)
Glenn L McGrath545106f2002-11-11 06:21:00 +000044
Denys Vlasenkoda62b092010-03-11 12:13:18 +010045typedef struct walker_list {
46 char *end;
47 char *cur;
48 struct walker_list *prev;
49 char wbuf[1];
50} walker_list;
51
Glenn L McGrath545106f2002-11-11 06:21:00 +000052/* Variable */
53typedef struct var_s {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +000054 unsigned type; /* flags */
Glenn L McGrath545106f2002-11-11 06:21:00 +000055 double number;
56 char *string;
57 union {
Denis Vlasenko629563b2007-02-24 17:05:52 +000058 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 */
Denys Vlasenkoda62b092010-03-11 12:13:18 +010061 walker_list *walker; /* list of array elements (for..in) */
Glenn L McGrath545106f2002-11-11 06:21:00 +000062 } x;
63} var;
64
65/* Node chain (pattern-action chain, BEGIN, END, function bodies) */
66typedef struct chain_s {
67 struct node_s *first;
68 struct node_s *last;
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +000069 const char *programname;
Glenn L McGrath545106f2002-11-11 06:21:00 +000070} chain;
71
72/* Function */
73typedef struct func_s {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +000074 unsigned nargs;
Glenn L McGrath545106f2002-11-11 06:21:00 +000075 struct chain_s body;
76} func;
77
78/* I/O stream */
79typedef struct rstream_s {
80 FILE *F;
81 char *buffer;
Glenn L McGrath00ed36f2003-10-30 13:36:39 +000082 int adv;
Glenn L McGrath545106f2002-11-11 06:21:00 +000083 int size;
84 int pos;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +000085 smallint is_pipe;
Glenn L McGrath545106f2002-11-11 06:21:00 +000086} rstream;
87
88typedef struct hash_item_s {
89 union {
Denis Vlasenkoffba9412007-05-17 23:03:35 +000090 struct var_s v; /* variable/array hash */
91 struct rstream_s rs; /* redirect streams hash */
92 struct func_s f; /* functions hash */
Glenn L McGrath545106f2002-11-11 06:21:00 +000093 } data;
Denis Vlasenkoffba9412007-05-17 23:03:35 +000094 struct hash_item_s *next; /* next in chain */
95 char name[1]; /* really it's longer */
Glenn L McGrath545106f2002-11-11 06:21:00 +000096} hash_item;
97
98typedef struct xhash_s {
Denis Vlasenkoffba9412007-05-17 23:03:35 +000099 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 */
Glenn L McGrath545106f2002-11-11 06:21:00 +0000103 struct hash_item_s **items;
104} xhash;
105
106/* Tree node */
107typedef struct node_s {
Mike Frysingerf87b3e32005-09-27 04:16:22 +0000108 uint32_t info;
Denis Vlasenkocd5c7862007-05-17 16:37:22 +0000109 unsigned lineno;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000110 union {
111 struct node_s *n;
112 var *v;
113 int i;
114 char *s;
115 regex_t *re;
116 } l;
117 union {
118 struct node_s *n;
119 regex_t *ire;
120 func *f;
121 int argno;
122 } r;
123 union {
124 struct node_s *n;
125 } a;
126} node;
127
128/* Block of temporary variables */
129typedef struct nvblock_s {
130 int size;
131 var *pos;
132 struct nvblock_s *prev;
133 struct nvblock_s *next;
Denys Vlasenkod069e532009-09-09 23:12:10 +0200134 var nv[];
Glenn L McGrath545106f2002-11-11 06:21:00 +0000135} nvblock;
136
137typedef struct tsplitter_s {
138 node n;
139 regex_t re[2];
140} tsplitter;
141
142/* simple token classes */
143/* Order and hex values are very important!!! See next_token() */
144#define TC_SEQSTART 1 /* ( */
145#define TC_SEQTERM (1 << 1) /* ) */
146#define TC_REGEXP (1 << 2) /* /.../ */
147#define TC_OUTRDR (1 << 3) /* | > >> */
148#define TC_UOPPOST (1 << 4) /* unary postfix operator */
149#define TC_UOPPRE1 (1 << 5) /* unary prefix operator */
150#define TC_BINOPX (1 << 6) /* two-opnd operator */
151#define TC_IN (1 << 7)
152#define TC_COMMA (1 << 8)
153#define TC_PIPE (1 << 9) /* input redirection pipe */
154#define TC_UOPPRE2 (1 << 10) /* unary prefix operator */
155#define TC_ARRTERM (1 << 11) /* ] */
156#define TC_GRPSTART (1 << 12) /* { */
157#define TC_GRPTERM (1 << 13) /* } */
158#define TC_SEMICOL (1 << 14)
159#define TC_NEWLINE (1 << 15)
160#define TC_STATX (1 << 16) /* ctl statement (for, next...) */
161#define TC_WHILE (1 << 17)
162#define TC_ELSE (1 << 18)
163#define TC_BUILTIN (1 << 19)
164#define TC_GETLINE (1 << 20)
165#define TC_FUNCDECL (1 << 21) /* `function' `func' */
166#define TC_BEGIN (1 << 22)
167#define TC_END (1 << 23)
168#define TC_EOF (1 << 24)
169#define TC_VARIABLE (1 << 25)
170#define TC_ARRAY (1 << 26)
171#define TC_FUNCTION (1 << 27)
172#define TC_STRING (1 << 28)
173#define TC_NUMBER (1 << 29)
174
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000175#define TC_UOPPRE (TC_UOPPRE1 | TC_UOPPRE2)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000176
177/* combined token classes */
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000178#define TC_BINOP (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN)
179#define TC_UNARYOP (TC_UOPPRE | TC_UOPPOST)
180#define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION \
181 | TC_BUILTIN | TC_GETLINE | TC_SEQSTART | TC_STRING | TC_NUMBER)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000182
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000183#define TC_STATEMNT (TC_STATX | TC_WHILE)
184#define TC_OPTERM (TC_SEMICOL | TC_NEWLINE)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000185
186/* word tokens, cannot mean something else if not expected */
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000187#define TC_WORD (TC_IN | TC_STATEMNT | TC_ELSE | TC_BUILTIN \
188 | TC_GETLINE | TC_FUNCDECL | TC_BEGIN | TC_END)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000189
190/* discard newlines after these */
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000191#define TC_NOTERM (TC_COMMA | TC_GRPSTART | TC_GRPTERM \
192 | TC_BINOP | TC_OPTERM)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000193
194/* what can expression begin with */
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000195#define TC_OPSEQ (TC_OPERAND | TC_UOPPRE | TC_REGEXP)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000196/* what can group begin with */
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000197#define TC_GRPSEQ (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000198
199/* if previous token class is CONCAT1 and next is CONCAT2, concatenation */
200/* operator is inserted between them */
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000201#define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM \
202 | TC_STRING | TC_NUMBER | TC_UOPPOST)
203#define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000204
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000205#define OF_RES1 0x010000
206#define OF_RES2 0x020000
207#define OF_STR1 0x040000
208#define OF_STR2 0x080000
209#define OF_NUM1 0x100000
210#define OF_CHECKED 0x200000
Glenn L McGrath545106f2002-11-11 06:21:00 +0000211
212/* combined operator flags */
213#define xx 0
214#define xV OF_RES2
215#define xS (OF_RES2 | OF_STR2)
216#define Vx OF_RES1
217#define VV (OF_RES1 | OF_RES2)
218#define Nx (OF_RES1 | OF_NUM1)
219#define NV (OF_RES1 | OF_NUM1 | OF_RES2)
220#define Sx (OF_RES1 | OF_STR1)
221#define SV (OF_RES1 | OF_STR1 | OF_RES2)
222#define SS (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2)
223
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000224#define OPCLSMASK 0xFF00
225#define OPNMASK 0x007F
Glenn L McGrath545106f2002-11-11 06:21:00 +0000226
227/* operator priority is a highest byte (even: r->l, odd: l->r grouping)
228 * For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1,
229 * n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string
230 */
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000231#define P(x) (x << 24)
232#define PRIMASK 0x7F000000
233#define PRIMASK2 0x7E000000
Glenn L McGrath545106f2002-11-11 06:21:00 +0000234
235/* Operation classes */
236
237#define SHIFT_TIL_THIS 0x0600
238#define RECUR_FROM_THIS 0x1000
239
240enum {
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000241 OC_DELETE = 0x0100, OC_EXEC = 0x0200, OC_NEWSOURCE = 0x0300,
242 OC_PRINT = 0x0400, OC_PRINTF = 0x0500, OC_WALKINIT = 0x0600,
Glenn L McGrath545106f2002-11-11 06:21:00 +0000243
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000244 OC_BR = 0x0700, OC_BREAK = 0x0800, OC_CONTINUE = 0x0900,
245 OC_EXIT = 0x0a00, OC_NEXT = 0x0b00, OC_NEXTFILE = 0x0c00,
246 OC_TEST = 0x0d00, OC_WALKNEXT = 0x0e00,
Glenn L McGrath545106f2002-11-11 06:21:00 +0000247
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000248 OC_BINARY = 0x1000, OC_BUILTIN = 0x1100, OC_COLON = 0x1200,
249 OC_COMMA = 0x1300, OC_COMPARE = 0x1400, OC_CONCAT = 0x1500,
250 OC_FBLTIN = 0x1600, OC_FIELD = 0x1700, OC_FNARG = 0x1800,
251 OC_FUNC = 0x1900, OC_GETLINE = 0x1a00, OC_IN = 0x1b00,
252 OC_LAND = 0x1c00, OC_LOR = 0x1d00, OC_MATCH = 0x1e00,
253 OC_MOVE = 0x1f00, OC_PGETLINE = 0x2000, OC_REGEXP = 0x2100,
254 OC_REPLACE = 0x2200, OC_RETURN = 0x2300, OC_SPRINTF = 0x2400,
255 OC_TERNARY = 0x2500, OC_UNARY = 0x2600, OC_VAR = 0x2700,
256 OC_DONE = 0x2800,
Glenn L McGrath545106f2002-11-11 06:21:00 +0000257
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000258 ST_IF = 0x3000, ST_DO = 0x3100, ST_FOR = 0x3200,
259 ST_WHILE = 0x3300
Glenn L McGrath545106f2002-11-11 06:21:00 +0000260};
261
262/* simple builtins */
263enum {
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000264 F_in, F_rn, F_co, F_ex, F_lg, F_si, F_sq, F_sr,
Glenn L McGrath545106f2002-11-11 06:21:00 +0000265 F_ti, F_le, F_sy, F_ff, F_cl
266};
267
268/* builtins */
269enum {
Leonid Lisovskiy46a0be52009-09-21 04:08:08 +0200270 B_a2, B_ix, B_ma, B_sp, B_ss, B_ti, B_mt, B_lo, B_up,
Denis Vlasenkoe175ff22006-09-26 17:41:00 +0000271 B_ge, B_gs, B_su,
272 B_an, B_co, B_ls, B_or, B_rs, B_xo,
Glenn L McGrath545106f2002-11-11 06:21:00 +0000273};
274
275/* tokens and their corresponding info values */
276
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000277#define NTC "\377" /* switch to next token class (tc<<1) */
278#define NTCC '\377'
Glenn L McGrath545106f2002-11-11 06:21:00 +0000279
280#define OC_B OC_BUILTIN
281
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000282static const char tokenlist[] ALIGN1 =
Denis Vlasenkof782f522007-01-01 23:51:30 +0000283 "\1(" NTC
284 "\1)" NTC
285 "\1/" NTC /* REGEXP */
286 "\2>>" "\1>" "\1|" NTC /* OUTRDR */
287 "\2++" "\2--" NTC /* UOPPOST */
288 "\2++" "\2--" "\1$" NTC /* UOPPRE1 */
289 "\2==" "\1=" "\2+=" "\2-=" /* BINOPX */
290 "\2*=" "\2/=" "\2%=" "\2^="
291 "\1+" "\1-" "\3**=" "\2**"
292 "\1/" "\1%" "\1^" "\1*"
293 "\2!=" "\2>=" "\2<=" "\1>"
294 "\1<" "\2!~" "\1~" "\2&&"
295 "\2||" "\1?" "\1:" NTC
296 "\2in" NTC
297 "\1," NTC
298 "\1|" NTC
299 "\1+" "\1-" "\1!" NTC /* UOPPRE2 */
300 "\1]" NTC
301 "\1{" NTC
302 "\1}" NTC
303 "\1;" NTC
304 "\1\n" NTC
305 "\2if" "\2do" "\3for" "\5break" /* STATX */
306 "\10continue" "\6delete" "\5print"
307 "\6printf" "\4next" "\10nextfile"
308 "\6return" "\4exit" NTC
309 "\5while" NTC
310 "\4else" NTC
Glenn L McGrath545106f2002-11-11 06:21:00 +0000311
Denis Vlasenkof782f522007-01-01 23:51:30 +0000312 "\3and" "\5compl" "\6lshift" "\2or"
313 "\6rshift" "\3xor"
314 "\5close" "\6system" "\6fflush" "\5atan2" /* BUILTIN */
315 "\3cos" "\3exp" "\3int" "\3log"
316 "\4rand" "\3sin" "\4sqrt" "\5srand"
317 "\6gensub" "\4gsub" "\5index" "\6length"
318 "\5match" "\5split" "\7sprintf" "\3sub"
Leonid Lisovskiy46a0be52009-09-21 04:08:08 +0200319 "\6substr" "\7systime" "\10strftime" "\6mktime"
Denis Vlasenkof782f522007-01-01 23:51:30 +0000320 "\7tolower" "\7toupper" NTC
321 "\7getline" NTC
322 "\4func" "\10function" NTC
323 "\5BEGIN" NTC
324 "\3END" "\0"
Glenn L McGrath545106f2002-11-11 06:21:00 +0000325 ;
326
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000327static const uint32_t tokeninfo[] = {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000328 0,
329 0,
330 OC_REGEXP,
Denis Vlasenkof782f522007-01-01 23:51:30 +0000331 xS|'a', xS|'w', xS|'|',
332 OC_UNARY|xV|P(9)|'p', OC_UNARY|xV|P(9)|'m',
333 OC_UNARY|xV|P(9)|'P', OC_UNARY|xV|P(9)|'M',
334 OC_FIELD|xV|P(5),
335 OC_COMPARE|VV|P(39)|5, OC_MOVE|VV|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_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&',
339 OC_BINARY|NV|P(29)|'+', OC_BINARY|NV|P(29)|'-',
340 OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&',
341 OC_BINARY|NV|P(25)|'/', OC_BINARY|NV|P(25)|'%',
342 OC_BINARY|NV|P(15)|'&', OC_BINARY|NV|P(25)|'*',
343 OC_COMPARE|VV|P(39)|4, OC_COMPARE|VV|P(39)|3,
344 OC_COMPARE|VV|P(39)|0, OC_COMPARE|VV|P(39)|1,
345 OC_COMPARE|VV|P(39)|2, OC_MATCH|Sx|P(45)|'!',
346 OC_MATCH|Sx|P(45)|'~', OC_LAND|Vx|P(55),
347 OC_LOR|Vx|P(59), OC_TERNARY|Vx|P(64)|'?',
348 OC_COLON|xx|P(67)|':',
Glenn L McGrath545106f2002-11-11 06:21:00 +0000349 OC_IN|SV|P(49),
350 OC_COMMA|SS|P(80),
351 OC_PGETLINE|SV|P(37),
Denis Vlasenkof782f522007-01-01 23:51:30 +0000352 OC_UNARY|xV|P(19)|'+', OC_UNARY|xV|P(19)|'-',
353 OC_UNARY|xV|P(19)|'!',
Glenn L McGrath545106f2002-11-11 06:21:00 +0000354 0,
355 0,
356 0,
357 0,
358 0,
Denis Vlasenkof782f522007-01-01 23:51:30 +0000359 ST_IF, ST_DO, ST_FOR, OC_BREAK,
360 OC_CONTINUE, OC_DELETE|Vx, OC_PRINT,
361 OC_PRINTF, OC_NEXT, OC_NEXTFILE,
362 OC_RETURN|Vx, OC_EXIT|Nx,
Glenn L McGrath545106f2002-11-11 06:21:00 +0000363 ST_WHILE,
364 0,
365
Denis Vlasenkoe175ff22006-09-26 17:41:00 +0000366 OC_B|B_an|P(0x83), OC_B|B_co|P(0x41), OC_B|B_ls|P(0x83), OC_B|B_or|P(0x83),
367 OC_B|B_rs|P(0x83), OC_B|B_xo|P(0x83),
Glenn L McGrath545106f2002-11-11 06:21:00 +0000368 OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83),
369 OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg,
370 OC_FBLTIN|F_rn, OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr,
371 OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), OC_FBLTIN|Sx|F_le,
372 OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF, OC_B|B_su|P(0xb6),
Leonid Lisovskiy46a0be52009-09-21 04:08:08 +0200373 OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b), OC_B|B_mt|P(0x0b),
Glenn L McGrath545106f2002-11-11 06:21:00 +0000374 OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49),
375 OC_GETLINE|SV|P(0),
376 0, 0,
377 0,
378 0
379};
380
381/* internal variable names and their initial values */
Eric Andersenc7bda1c2004-03-15 08:29:22 +0000382/* asterisk marks SPECIAL vars; $ is just no-named Field0 */
Glenn L McGrath545106f2002-11-11 06:21:00 +0000383enum {
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000384 CONVFMT, OFMT, FS, OFS,
Denis Vlasenkof782f522007-01-01 23:51:30 +0000385 ORS, RS, RT, FILENAME,
Denis Vlasenko41d5ebe2009-01-25 01:00:15 +0000386 SUBSEP, F0, ARGIND, ARGC,
387 ARGV, ERRNO, FNR, NR,
388 NF, IGNORECASE, ENVIRON, NUM_INTERNAL_VARS
Glenn L McGrath545106f2002-11-11 06:21:00 +0000389};
390
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000391static const char vNames[] ALIGN1 =
Denis Vlasenkof782f522007-01-01 23:51:30 +0000392 "CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0"
393 "ORS\0" "RS\0*" "RT\0" "FILENAME\0"
Denis Vlasenko41d5ebe2009-01-25 01:00:15 +0000394 "SUBSEP\0" "$\0*" "ARGIND\0" "ARGC\0"
395 "ARGV\0" "ERRNO\0" "FNR\0" "NR\0"
396 "NF\0*" "IGNORECASE\0*" "ENVIRON\0" "\0";
Glenn L McGrath545106f2002-11-11 06:21:00 +0000397
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000398static const char vValues[] ALIGN1 =
Denis Vlasenkof782f522007-01-01 23:51:30 +0000399 "%.6g\0" "%.6g\0" " \0" " \0"
400 "\n\0" "\n\0" "\0" "\0"
Denis Vlasenko41d5ebe2009-01-25 01:00:15 +0000401 "\034\0" "\0" "\377";
Glenn L McGrath545106f2002-11-11 06:21:00 +0000402
403/* hash size may grow to these values */
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000404#define FIRST_PRIME 61
405static const uint16_t PRIMES[] ALIGN2 = { 251, 1021, 4093, 16381, 65521 };
Glenn L McGrath545106f2002-11-11 06:21:00 +0000406
Glenn L McGrath545106f2002-11-11 06:21:00 +0000407
Denis Vlasenkoe16e6e72007-06-07 13:14:53 +0000408/* Globals. Split in two parts so that first one is addressed
Denis Vlasenko9aa5c652009-02-26 11:21:04 +0000409 * with (mostly short) negative offsets.
410 * NB: it's unsafe to put members of type "double"
411 * into globals2 (gcc may fail to align them).
412 */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000413struct globals {
Denis Vlasenko9aa5c652009-02-26 11:21:04 +0000414 double t_double;
Denis Vlasenko574f2f42008-02-27 18:41:59 +0000415 chain beginseq, mainseq, endseq;
416 chain *seq;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000417 node *break_ptr, *continue_ptr;
418 rstream *iF;
419 xhash *vhash, *ahash, *fdhash, *fnhash;
420 const char *g_progname;
421 int g_lineno;
422 int nfields;
423 int maxfields; /* used in fsrealloc() only */
424 var *Fields;
425 nvblock *g_cb;
426 char *g_pos;
427 char *g_buf;
Denis Vlasenkoe16e6e72007-06-07 13:14:53 +0000428 smallint icase;
429 smallint exiting;
430 smallint nextrec;
431 smallint nextfile;
432 smallint is_f0_split;
433};
434struct globals2 {
435 uint32_t t_info; /* often used */
436 uint32_t t_tclass;
437 char *t_string;
438 int t_lineno;
439 int t_rollback;
440
441 var *intvar[NUM_INTERNAL_VARS]; /* often used */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000442
443 /* former statics from various functions */
444 char *split_f0__fstrings;
445
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000446 uint32_t next_token__save_tclass;
447 uint32_t next_token__save_info;
448 uint32_t next_token__ltclass;
Denis Vlasenkoe16e6e72007-06-07 13:14:53 +0000449 smallint next_token__concat_inserted;
450
451 smallint next_input_file__files_happen;
452 rstream next_input_file__rsm;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000453
454 var *evaluate__fnargs;
455 unsigned evaluate__seed;
456 regex_t evaluate__sreg;
457
458 var ptest__v;
459
460 tsplitter exec_builtin__tspl;
461
Denis Vlasenkoe16e6e72007-06-07 13:14:53 +0000462 /* biggest and least used members go last */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000463 tsplitter fsplitter, rsplitter;
Denys Vlasenko3dbc5a92010-02-05 14:54:22 +0100464};
Denis Vlasenkoe16e6e72007-06-07 13:14:53 +0000465#define G1 (ptr_to_globals[-1])
Denis Vlasenko574f2f42008-02-27 18:41:59 +0000466#define G (*(struct globals2 *)ptr_to_globals)
Denis Vlasenkoe16e6e72007-06-07 13:14:53 +0000467/* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */
Denis Vlasenko9aa5c652009-02-26 11:21:04 +0000468/*char G1size[sizeof(G1)]; - 0x74 */
469/*char Gsize[sizeof(G)]; - 0x1c4 */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000470/* Trying to keep most of members accessible with short offsets: */
Denis Vlasenko9aa5c652009-02-26 11:21:04 +0000471/*char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */
472#define t_double (G1.t_double )
Denis Vlasenkoe16e6e72007-06-07 13:14:53 +0000473#define beginseq (G1.beginseq )
474#define mainseq (G1.mainseq )
475#define endseq (G1.endseq )
476#define seq (G1.seq )
477#define break_ptr (G1.break_ptr )
478#define continue_ptr (G1.continue_ptr)
479#define iF (G1.iF )
480#define vhash (G1.vhash )
481#define ahash (G1.ahash )
482#define fdhash (G1.fdhash )
483#define fnhash (G1.fnhash )
484#define g_progname (G1.g_progname )
485#define g_lineno (G1.g_lineno )
486#define nfields (G1.nfields )
487#define maxfields (G1.maxfields )
488#define Fields (G1.Fields )
489#define g_cb (G1.g_cb )
490#define g_pos (G1.g_pos )
491#define g_buf (G1.g_buf )
492#define icase (G1.icase )
493#define exiting (G1.exiting )
494#define nextrec (G1.nextrec )
495#define nextfile (G1.nextfile )
496#define is_f0_split (G1.is_f0_split )
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000497#define t_info (G.t_info )
498#define t_tclass (G.t_tclass )
499#define t_string (G.t_string )
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000500#define t_lineno (G.t_lineno )
501#define t_rollback (G.t_rollback )
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000502#define intvar (G.intvar )
503#define fsplitter (G.fsplitter )
504#define rsplitter (G.rsplitter )
505#define INIT_G() do { \
Denys Vlasenko90a99042009-09-06 02:36:23 +0200506 SET_PTR_TO_GLOBALS((char*)xzalloc(sizeof(G1)+sizeof(G)) + sizeof(G1)); \
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000507 G.next_token__ltclass = TC_OPTERM; \
508 G.evaluate__seed = 1; \
509} while (0)
510
Glenn L McGrath545106f2002-11-11 06:21:00 +0000511
512/* function prototypes */
Glenn L McGrath545106f2002-11-11 06:21:00 +0000513static void handle_special(var *);
Mike Frysingerf87b3e32005-09-27 04:16:22 +0000514static node *parse_expr(uint32_t);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000515static void chain_group(void);
516static var *evaluate(node *, var *);
517static rstream *next_input_file(void);
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000518static int fmt_num(char *, int, const char *, double, int);
Denis Vlasenkoa60f84e2008-07-05 09:18:54 +0000519static int awk_exit(int) NORETURN;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000520
521/* ---- error handling ---- */
522
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000523static const char EMSG_INTERNAL_ERROR[] ALIGN1 = "Internal error";
524static const char EMSG_UNEXP_EOS[] ALIGN1 = "Unexpected end of string";
525static const char EMSG_UNEXP_TOKEN[] ALIGN1 = "Unexpected token";
526static const char EMSG_DIV_BY_ZERO[] ALIGN1 = "Division by zero";
527static const char EMSG_INV_FMT[] ALIGN1 = "Invalid format specifier";
528static const char EMSG_TOO_FEW_ARGS[] ALIGN1 = "Too few arguments for builtin";
529static const char EMSG_NOT_ARRAY[] ALIGN1 = "Not an array";
530static const char EMSG_POSSIBLE_ERROR[] ALIGN1 = "Possible syntax error";
531static const char EMSG_UNDEF_FUNC[] ALIGN1 = "Call to undefined function";
Denis Vlasenko2d5bd802008-10-24 10:49:49 +0000532#if !ENABLE_FEATURE_AWK_LIBM
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000533static const char EMSG_NO_MATH[] ALIGN1 = "Math support is not compiled in";
Glenn L McGrath545106f2002-11-11 06:21:00 +0000534#endif
535
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100536static void zero_out_var(var *vp)
Denis Vlasenkof782f522007-01-01 23:51:30 +0000537{
538 memset(vp, 0, sizeof(*vp));
539}
540
Denis Vlasenkoc7cc5a92009-04-19 01:27:20 +0000541static void syntax_error(const char *message) NORETURN;
542static void syntax_error(const char *message)
Glenn L McGrathd4036f82002-11-28 09:30:40 +0000543{
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000544 bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000545}
546
Glenn L McGrath545106f2002-11-11 06:21:00 +0000547/* ---- hash stuff ---- */
548
Denis Vlasenkof782f522007-01-01 23:51:30 +0000549static unsigned hashidx(const char *name)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000550{
Denis Vlasenkof782f522007-01-01 23:51:30 +0000551 unsigned idx = 0;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000552
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100553 while (*name)
554 idx = *name++ + (idx << 6) - idx;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000555 return idx;
556}
557
558/* create new hash */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000559static xhash *hash_init(void)
560{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000561 xhash *newhash;
Eric Andersenc7bda1c2004-03-15 08:29:22 +0000562
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100563 newhash = xzalloc(sizeof(*newhash));
Glenn L McGrath545106f2002-11-11 06:21:00 +0000564 newhash->csize = FIRST_PRIME;
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100565 newhash->items = xzalloc(FIRST_PRIME * sizeof(newhash->items[0]));
Glenn L McGrath545106f2002-11-11 06:21:00 +0000566
567 return newhash;
568}
569
570/* find item in hash, return ptr to data, NULL if not found */
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000571static void *hash_search(xhash *hash, const char *name)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000572{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000573 hash_item *hi;
574
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100575 hi = hash->items[hashidx(name) % hash->csize];
Glenn L McGrath545106f2002-11-11 06:21:00 +0000576 while (hi) {
577 if (strcmp(hi->name, name) == 0)
578 return &(hi->data);
579 hi = hi->next;
580 }
581 return NULL;
582}
583
584/* grow hash if it becomes too big */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000585static void hash_rebuild(xhash *hash)
586{
Denis Vlasenkof782f522007-01-01 23:51:30 +0000587 unsigned newsize, i, idx;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000588 hash_item **newitems, *hi, *thi;
589
Denis Vlasenko80b8b392007-06-25 10:55:35 +0000590 if (hash->nprime == ARRAY_SIZE(PRIMES))
Glenn L McGrath545106f2002-11-11 06:21:00 +0000591 return;
592
593 newsize = PRIMES[hash->nprime++];
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100594 newitems = xzalloc(newsize * sizeof(newitems[0]));
Glenn L McGrath545106f2002-11-11 06:21:00 +0000595
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000596 for (i = 0; i < hash->csize; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000597 hi = hash->items[i];
598 while (hi) {
599 thi = hi;
600 hi = thi->next;
601 idx = hashidx(thi->name) % newsize;
602 thi->next = newitems[idx];
603 newitems[idx] = thi;
604 }
605 }
606
607 free(hash->items);
608 hash->csize = newsize;
609 hash->items = newitems;
610}
611
612/* find item in hash, add it if necessary. Return ptr to data */
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000613static void *hash_find(xhash *hash, const char *name)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000614{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000615 hash_item *hi;
Denis Vlasenkof782f522007-01-01 23:51:30 +0000616 unsigned idx;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000617 int l;
618
619 hi = hash_search(hash, name);
Denis Vlasenkob78c7822007-07-18 18:31:11 +0000620 if (!hi) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000621 if (++hash->nel / hash->csize > 10)
622 hash_rebuild(hash);
623
Rob Landleya3896512006-05-07 20:20:34 +0000624 l = strlen(name) + 1;
Denis Vlasenko7a676642009-03-15 22:20:31 +0000625 hi = xzalloc(sizeof(*hi) + l);
626 strcpy(hi->name, name);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000627
628 idx = hashidx(name) % hash->csize;
629 hi->next = hash->items[idx];
630 hash->items[idx] = hi;
631 hash->glen += l;
632 }
633 return &(hi->data);
634}
635
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000636#define findvar(hash, name) ((var*) hash_find((hash), (name)))
637#define newvar(name) ((var*) hash_find(vhash, (name)))
638#define newfile(name) ((rstream*)hash_find(fdhash, (name)))
639#define newfunc(name) ((func*) hash_find(fnhash, (name)))
Glenn L McGrath545106f2002-11-11 06:21:00 +0000640
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000641static void hash_remove(xhash *hash, const char *name)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000642{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000643 hash_item *hi, **phi;
644
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000645 phi = &(hash->items[hashidx(name) % hash->csize]);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000646 while (*phi) {
647 hi = *phi;
648 if (strcmp(hi->name, name) == 0) {
Rob Landleya3896512006-05-07 20:20:34 +0000649 hash->glen -= (strlen(name) + 1);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000650 hash->nel--;
651 *phi = hi->next;
652 free(hi);
653 break;
654 }
655 phi = &(hi->next);
656 }
657}
658
659/* ------ some useful functions ------ */
660
Mike Frysinger10a11e22005-09-27 02:23:02 +0000661static void skip_spaces(char **s)
662{
"Robert P. J. Day"68229832006-07-01 13:08:46 +0000663 char *p = *s;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000664
Denis Vlasenkocd5c7862007-05-17 16:37:22 +0000665 while (1) {
666 if (*p == '\\' && p[1] == '\n') {
667 p++;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000668 t_lineno++;
Denis Vlasenkocd5c7862007-05-17 16:37:22 +0000669 } else if (*p != ' ' && *p != '\t') {
670 break;
671 }
Mike Frysingerde2b9382005-09-27 03:18:00 +0000672 p++;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000673 }
674 *s = p;
675}
676
Mike Frysinger10a11e22005-09-27 02:23:02 +0000677static char *nextword(char **s)
678{
"Robert P. J. Day"68229832006-07-01 13:08:46 +0000679 char *p = *s;
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100680 while (*(*s)++)
681 continue;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000682 return p;
683}
684
Mike Frysinger10a11e22005-09-27 02:23:02 +0000685static char nextchar(char **s)
686{
"Robert P. J. Day"68229832006-07-01 13:08:46 +0000687 char c, *pps;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000688
689 c = *((*s)++);
690 pps = *s;
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100691 if (c == '\\')
692 c = bb_process_escape_sequence((const char**)s);
693 if (c == '\\' && *s == pps)
694 c = *((*s)++);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000695 return c;
696}
697
Denis Vlasenko77ad97f2008-05-13 02:27:31 +0000698static ALWAYS_INLINE int isalnum_(int c)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000699{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000700 return (isalnum(c) || c == '_');
701}
702
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +0000703static double my_strtod(char **pp)
704{
705#if ENABLE_DESKTOP
706 if ((*pp)[0] == '0'
707 && ((((*pp)[1] | 0x20) == 'x') || isdigit((*pp)[1]))
708 ) {
709 return strtoull(*pp, pp, 0);
710 }
711#endif
712 return strtod(*pp, pp);
713}
714
Glenn L McGrath545106f2002-11-11 06:21:00 +0000715/* -------- working with variables (set/get/copy/etc) -------- */
716
Mike Frysinger10a11e22005-09-27 02:23:02 +0000717static xhash *iamarray(var *v)
718{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000719 var *a = v;
720
721 while (a->type & VF_CHILD)
722 a = a->x.parent;
723
Denis Vlasenkocd5c7862007-05-17 16:37:22 +0000724 if (!(a->type & VF_ARRAY)) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000725 a->type |= VF_ARRAY;
726 a->x.array = hash_init();
727 }
728 return a->x.array;
729}
730
Mike Frysinger10a11e22005-09-27 02:23:02 +0000731static void clear_array(xhash *array)
732{
Denis Vlasenkof782f522007-01-01 23:51:30 +0000733 unsigned i;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000734 hash_item *hi, *thi;
735
Denis Vlasenkocd5c7862007-05-17 16:37:22 +0000736 for (i = 0; i < array->csize; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000737 hi = array->items[i];
738 while (hi) {
739 thi = hi;
740 hi = hi->next;
Aaron Lehmanna170e1c2002-11-28 11:27:31 +0000741 free(thi->data.v.string);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000742 free(thi);
743 }
744 array->items[i] = NULL;
745 }
746 array->glen = array->nel = 0;
747}
748
749/* clear a variable */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000750static var *clrvar(var *v)
751{
Aaron Lehmanna170e1c2002-11-28 11:27:31 +0000752 if (!(v->type & VF_FSTR))
Glenn L McGrath545106f2002-11-11 06:21:00 +0000753 free(v->string);
754
755 v->type &= VF_DONTTOUCH;
756 v->type |= VF_DIRTY;
757 v->string = NULL;
758 return v;
759}
760
761/* assign string value to variable */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000762static var *setvar_p(var *v, char *value)
763{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000764 clrvar(v);
765 v->string = value;
766 handle_special(v);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000767 return v;
768}
769
770/* same as setvar_p but make a copy of string */
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000771static var *setvar_s(var *v, const char *value)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000772{
Rob Landleyd921b2e2006-08-03 15:41:12 +0000773 return setvar_p(v, (value && *value) ? xstrdup(value) : NULL);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000774}
775
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100776/* same as setvar_s but sets USER flag */
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000777static var *setvar_u(var *v, const char *value)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000778{
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100779 v = setvar_s(v, value);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000780 v->type |= VF_USER;
781 return v;
782}
783
784/* set array element to user string */
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000785static void setari_u(var *a, int idx, const char *s)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000786{
"Robert P. J. Day"68229832006-07-01 13:08:46 +0000787 var *v;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000788
Denys Vlasenko7bb346f2009-10-06 22:09:50 +0200789 v = findvar(iamarray(a), itoa(idx));
Glenn L McGrath545106f2002-11-11 06:21:00 +0000790 setvar_u(v, s);
791}
792
793/* assign numeric value to variable */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000794static var *setvar_i(var *v, double value)
795{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000796 clrvar(v);
797 v->type |= VF_NUMBER;
798 v->number = value;
799 handle_special(v);
800 return v;
801}
802
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +0000803static const char *getvar_s(var *v)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000804{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000805 /* if v is numeric and has no cached string, convert it to string */
806 if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000807 fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE);
808 v->string = xstrdup(g_buf);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000809 v->type |= VF_CACHED;
810 }
811 return (v->string == NULL) ? "" : v->string;
812}
813
Mike Frysinger10a11e22005-09-27 02:23:02 +0000814static double getvar_i(var *v)
815{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000816 char *s;
817
818 if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) {
819 v->number = 0;
820 s = v->string;
821 if (s && *s) {
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +0000822 v->number = my_strtod(&s);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000823 if (v->type & VF_USER) {
824 skip_spaces(&s);
825 if (*s != '\0')
826 v->type &= ~VF_USER;
827 }
828 } else {
829 v->type &= ~VF_USER;
830 }
831 v->type |= VF_CACHED;
832 }
833 return v->number;
834}
835
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +0000836/* Used for operands of bitwise ops */
837static unsigned long getvar_i_int(var *v)
838{
839 double d = getvar_i(v);
840
841 /* Casting doubles to longs is undefined for values outside
842 * of target type range. Try to widen it as much as possible */
843 if (d >= 0)
844 return (unsigned long)d;
Denis Vlasenko665eaff2008-09-05 04:59:02 +0000845 /* Why? Think about d == -4294967295.0 (assuming 32bit longs) */
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +0000846 return - (long) (unsigned long) (-d);
847}
848
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +0000849static var *copyvar(var *dest, const var *src)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000850{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000851 if (dest != src) {
852 clrvar(dest);
Denis Vlasenko629563b2007-02-24 17:05:52 +0000853 dest->type |= (src->type & ~(VF_DONTTOUCH | VF_FSTR));
Glenn L McGrath545106f2002-11-11 06:21:00 +0000854 dest->number = src->number;
855 if (src->string)
Rob Landleyd921b2e2006-08-03 15:41:12 +0000856 dest->string = xstrdup(src->string);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000857 }
858 handle_special(dest);
859 return dest;
860}
861
Mike Frysinger10a11e22005-09-27 02:23:02 +0000862static var *incvar(var *v)
863{
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100864 return setvar_i(v, getvar_i(v) + 1.0);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000865}
866
867/* return true if v is number or numeric string */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000868static int is_numeric(var *v)
869{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000870 getvar_i(v);
871 return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY));
872}
873
874/* return 1 when value of v corresponds to true, 0 otherwise */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000875static int istrue(var *v)
876{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000877 if (is_numeric(v))
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100878 return (v->number != 0);
879 return (v->string && v->string[0]);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000880}
881
Eric Andersenaff114c2004-04-14 17:51:38 +0000882/* temporary variables allocator. Last allocated should be first freed */
Mike Frysinger10a11e22005-09-27 02:23:02 +0000883static var *nvalloc(int n)
884{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000885 nvblock *pb = NULL;
886 var *v, *r;
887 int size;
888
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000889 while (g_cb) {
890 pb = g_cb;
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100891 if ((g_cb->pos - g_cb->nv) + n <= g_cb->size)
892 break;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000893 g_cb = g_cb->next;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000894 }
895
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000896 if (!g_cb) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000897 size = (n <= MINNVBLOCK) ? MINNVBLOCK : n;
Denis Vlasenkoe0a7fc52008-07-02 11:14:59 +0000898 g_cb = xzalloc(sizeof(nvblock) + size * sizeof(var));
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000899 g_cb->size = size;
900 g_cb->pos = g_cb->nv;
901 g_cb->prev = pb;
Denis Vlasenkoe0a7fc52008-07-02 11:14:59 +0000902 /*g_cb->next = NULL; - xzalloc did it */
Denys Vlasenkocdeda162009-11-30 01:14:16 +0100903 if (pb)
904 pb->next = g_cb;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000905 }
906
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000907 v = r = g_cb->pos;
908 g_cb->pos += n;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000909
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000910 while (v < g_cb->pos) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000911 v->type = 0;
912 v->string = NULL;
913 v++;
914 }
915
916 return r;
917}
918
Mike Frysinger10a11e22005-09-27 02:23:02 +0000919static void nvfree(var *v)
920{
Glenn L McGrath545106f2002-11-11 06:21:00 +0000921 var *p;
922
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000923 if (v < g_cb->nv || v >= g_cb->pos)
924 syntax_error(EMSG_INTERNAL_ERROR);
Glenn L McGrath545106f2002-11-11 06:21:00 +0000925
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000926 for (p = v; p < g_cb->pos; p++) {
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000927 if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000928 clear_array(iamarray(p));
929 free(p->x.array->items);
930 free(p->x.array);
931 }
Denys Vlasenko3cb60c32010-03-10 19:20:32 +0100932 if (p->type & VF_WALK) {
Denys Vlasenkoda62b092010-03-11 12:13:18 +0100933 walker_list *n;
934 walker_list *w = p->x.walker;
935 debug_printf_walker("nvfree: freeing walker @%p\n", &p->x.walker);
936 p->x.walker = NULL;
937 while (w) {
938 n = w->prev;
939 debug_printf_walker(" free(%p)\n", w);
940 free(w);
941 w = n;
942 }
Denys Vlasenko3cb60c32010-03-10 19:20:32 +0100943 }
Glenn L McGrath545106f2002-11-11 06:21:00 +0000944 clrvar(p);
945 }
946
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000947 g_cb->pos = v;
948 while (g_cb->prev && g_cb->pos == g_cb->nv) {
949 g_cb = g_cb->prev;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000950 }
951}
952
953/* ------- awk program text parsing ------- */
954
Denis Vlasenkocd5c7862007-05-17 16:37:22 +0000955/* Parse next token pointed by global pos, place results into global ttt.
Glenn L McGrath545106f2002-11-11 06:21:00 +0000956 * If token isn't expected, give away. Return token class
957 */
Mike Frysingerf87b3e32005-09-27 04:16:22 +0000958static uint32_t next_token(uint32_t expected)
Mike Frysinger10a11e22005-09-27 02:23:02 +0000959{
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000960#define concat_inserted (G.next_token__concat_inserted)
961#define save_tclass (G.next_token__save_tclass)
962#define save_info (G.next_token__save_info)
963/* Initialized to TC_OPTERM: */
964#define ltclass (G.next_token__ltclass)
Glenn L McGrath545106f2002-11-11 06:21:00 +0000965
Denis Vlasenkof782f522007-01-01 23:51:30 +0000966 char *p, *pp, *s;
967 const char *tl;
968 uint32_t tc;
969 const uint32_t *ti;
970 int l;
971
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000972 if (t_rollback) {
973 t_rollback = FALSE;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000974
975 } else if (concat_inserted) {
Glenn L McGrath545106f2002-11-11 06:21:00 +0000976 concat_inserted = FALSE;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000977 t_tclass = save_tclass;
978 t_info = save_info;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000979
980 } else {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000981 p = g_pos;
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +0000982 readnext:
Glenn L McGrath545106f2002-11-11 06:21:00 +0000983 skip_spaces(&p);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000984 g_lineno = t_lineno;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000985 if (*p == '#')
Denis Vlasenkoffba9412007-05-17 23:03:35 +0000986 while (*p != '\n' && *p != '\0')
987 p++;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000988
989 if (*p == '\n')
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000990 t_lineno++;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000991
992 if (*p == '\0') {
993 tc = TC_EOF;
994
995 } else if (*p == '\"') {
996 /* it's a string */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +0000997 t_string = s = ++p;
Glenn L McGrath545106f2002-11-11 06:21:00 +0000998 while (*p != '\"') {
999 if (*p == '\0' || *p == '\n')
1000 syntax_error(EMSG_UNEXP_EOS);
1001 *(s++) = nextchar(&p);
1002 }
1003 p++;
1004 *s = '\0';
1005 tc = TC_STRING;
1006
1007 } else if ((expected & TC_REGEXP) && *p == '/') {
1008 /* it's regexp */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001009 t_string = s = ++p;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001010 while (*p != '/') {
1011 if (*p == '\0' || *p == '\n')
1012 syntax_error(EMSG_UNEXP_EOS);
Denis Vlasenkod9b5ab82007-05-18 07:30:43 +00001013 *s = *p++;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001014 if (*s++ == '\\') {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001015 pp = p;
Manuel Novoa III cad53642003-03-19 09:13:01 +00001016 *(s-1) = bb_process_escape_sequence((const char **)&p);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001017 if (*pp == '\\')
1018 *s++ = '\\';
1019 if (p == pp)
1020 *s++ = *p++;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001021 }
1022 }
1023 p++;
1024 *s = '\0';
1025 tc = TC_REGEXP;
1026
1027 } else if (*p == '.' || isdigit(*p)) {
1028 /* it's a number */
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00001029 t_double = my_strtod(&p);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001030 if (*p == '.')
1031 syntax_error(EMSG_UNEXP_TOKEN);
1032 tc = TC_NUMBER;
1033
1034 } else {
1035 /* search for something known */
1036 tl = tokenlist;
1037 tc = 0x00000001;
1038 ti = tokeninfo;
1039 while (*tl) {
1040 l = *(tl++);
1041 if (l == NTCC) {
1042 tc <<= 1;
1043 continue;
1044 }
1045 /* if token class is expected, token
1046 * matches and it's not a longer word,
1047 * then this is what we are looking for
1048 */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001049 if ((tc & (expected | TC_WORD | TC_NEWLINE))
1050 && *tl == *p && strncmp(p, tl, l) == 0
1051 && !((tc & TC_WORD) && isalnum_(p[l]))
1052 ) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001053 t_info = *ti;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001054 p += l;
1055 break;
1056 }
1057 ti++;
1058 tl += l;
1059 }
1060
Denis Vlasenkof782f522007-01-01 23:51:30 +00001061 if (!*tl) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001062 /* it's a name (var/array/function),
1063 * otherwise it's something wrong
1064 */
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00001065 if (!isalnum_(*p))
Glenn L McGrath545106f2002-11-11 06:21:00 +00001066 syntax_error(EMSG_UNEXP_TOKEN);
1067
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001068 t_string = --p;
Denis Vlasenkobf0a2012006-12-26 10:42:51 +00001069 while (isalnum_(*(++p))) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001070 *(p-1) = *p;
1071 }
1072 *(p-1) = '\0';
1073 tc = TC_VARIABLE;
Bernhard Reutner-Fischerbb204622005-10-17 14:21:06 +00001074 /* also consume whitespace between functionname and bracket */
Alexander Shishkind03cd3b2010-02-25 17:55:40 +02001075 if (!(expected & TC_VARIABLE) || (expected & TC_ARRAY))
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001076 skip_spaces(&p);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001077 if (*p == '(') {
1078 tc = TC_FUNCTION;
1079 } else {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001080 if (*p == '[') {
1081 p++;
1082 tc = TC_ARRAY;
1083 }
1084 }
1085 }
1086 }
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001087 g_pos = p;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001088
1089 /* skipping newlines in some cases */
1090 if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE))
1091 goto readnext;
1092
1093 /* insert concatenation operator when needed */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001094 if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001095 concat_inserted = TRUE;
1096 save_tclass = tc;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001097 save_info = t_info;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001098 tc = TC_BINOP;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001099 t_info = OC_CONCAT | SS | P(35);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001100 }
1101
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001102 t_tclass = tc;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001103 }
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001104 ltclass = t_tclass;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001105
1106 /* Are we ready for this? */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001107 if (!(ltclass & expected))
Glenn L McGrath545106f2002-11-11 06:21:00 +00001108 syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ?
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001109 EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001110
1111 return ltclass;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001112#undef concat_inserted
1113#undef save_tclass
1114#undef save_info
1115#undef ltclass
Glenn L McGrath545106f2002-11-11 06:21:00 +00001116}
1117
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00001118static void rollback_token(void)
1119{
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001120 t_rollback = TRUE;
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00001121}
Glenn L McGrath545106f2002-11-11 06:21:00 +00001122
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001123static node *new_node(uint32_t info)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001124{
"Robert P. J. Day"68229832006-07-01 13:08:46 +00001125 node *n;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001126
Denis Vlasenko4cccc032006-12-22 18:37:07 +00001127 n = xzalloc(sizeof(node));
Glenn L McGrath545106f2002-11-11 06:21:00 +00001128 n->info = info;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001129 n->lineno = g_lineno;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001130 return n;
1131}
1132
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001133static node *mk_re_node(const char *s, node *n, regex_t *re)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001134{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001135 n->info = OC_REGEXP;
1136 n->l.re = re;
1137 n->r.ire = re + 1;
1138 xregcomp(re, s, REG_EXTENDED);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001139 xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001140
1141 return n;
1142}
1143
Mike Frysinger10a11e22005-09-27 02:23:02 +00001144static node *condition(void)
1145{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001146 next_token(TC_SEQSTART);
1147 return parse_expr(TC_SEQTERM);
1148}
1149
1150/* parse expression terminated by given argument, return ptr
1151 * to built subtree. Terminator is eaten by parse_expr */
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001152static node *parse_expr(uint32_t iexp)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001153{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001154 node sn;
1155 node *cn = &sn;
1156 node *vn, *glptr;
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001157 uint32_t tc, xtc;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001158 var *v;
1159
1160 sn.info = PRIMASK;
1161 sn.r.n = glptr = NULL;
1162 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp;
1163
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00001164 while (!((tc = next_token(xtc)) & iexp)) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001165 if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001166 /* input redirection (<) attached to glptr node */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001167 cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37));
Glenn L McGrath4bded582004-02-22 11:55:09 +00001168 cn->a.n = glptr;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001169 xtc = TC_OPERAND | TC_UOPPRE;
1170 glptr = NULL;
1171
1172 } else if (tc & (TC_BINOP | TC_UOPPOST)) {
1173 /* for binary and postfix-unary operators, jump back over
1174 * previous operators with higher priority */
1175 vn = cn;
Denys Vlasenkocdeda162009-11-30 01:14:16 +01001176 while (((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2))
1177 || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON))
1178 ) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001179 vn = vn->a.n;
Denys Vlasenkocdeda162009-11-30 01:14:16 +01001180 }
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001181 if ((t_info & OPCLSMASK) == OC_TERNARY)
1182 t_info += P(6);
1183 cn = vn->a.n->r.n = new_node(t_info);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001184 cn->a.n = vn->a.n;
1185 if (tc & TC_BINOP) {
1186 cn->l.n = vn;
1187 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001188 if ((t_info & OPCLSMASK) == OC_PGETLINE) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001189 /* it's a pipe */
1190 next_token(TC_GETLINE);
1191 /* give maximum priority to this pipe */
1192 cn->info &= ~PRIMASK;
1193 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1194 }
1195 } else {
1196 cn->r.n = vn;
1197 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1198 }
1199 vn->a.n = cn;
1200
1201 } else {
1202 /* for operands and prefix-unary operators, attach them
1203 * to last node */
1204 vn = cn;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001205 cn = vn->r.n = new_node(t_info);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001206 cn->a.n = vn;
1207 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
1208 if (tc & (TC_OPERAND | TC_REGEXP)) {
Rob Landleyed830e82005-06-07 02:43:52 +00001209 xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp;
Eric Andersenc7bda1c2004-03-15 08:29:22 +00001210 /* one should be very careful with switch on tclass -
Glenn L McGrath545106f2002-11-11 06:21:00 +00001211 * only simple tclasses should be used! */
1212 switch (tc) {
Denis Vlasenkof782f522007-01-01 23:51:30 +00001213 case TC_VARIABLE:
1214 case TC_ARRAY:
Glenn L McGrath545106f2002-11-11 06:21:00 +00001215 cn->info = OC_VAR;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001216 v = hash_search(ahash, t_string);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001217 if (v != NULL) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001218 cn->info = OC_FNARG;
1219 cn->l.i = v->x.aidx;
1220 } else {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001221 cn->l.v = newvar(t_string);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001222 }
1223 if (tc & TC_ARRAY) {
1224 cn->info |= xS;
1225 cn->r.n = parse_expr(TC_ARRTERM);
1226 }
Glenn L McGrath545106f2002-11-11 06:21:00 +00001227 break;
Mike Frysingerde2b9382005-09-27 03:18:00 +00001228
Denis Vlasenkof782f522007-01-01 23:51:30 +00001229 case TC_NUMBER:
1230 case TC_STRING:
Glenn L McGrath545106f2002-11-11 06:21:00 +00001231 cn->info = OC_VAR;
Rob Landley9ffd4232006-05-21 18:30:35 +00001232 v = cn->l.v = xzalloc(sizeof(var));
Glenn L McGrath545106f2002-11-11 06:21:00 +00001233 if (tc & TC_NUMBER)
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001234 setvar_i(v, t_double);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001235 else
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001236 setvar_s(v, t_string);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001237 break;
1238
Denis Vlasenkof782f522007-01-01 23:51:30 +00001239 case TC_REGEXP:
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001240 mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2));
Glenn L McGrath545106f2002-11-11 06:21:00 +00001241 break;
1242
Denis Vlasenkof782f522007-01-01 23:51:30 +00001243 case TC_FUNCTION:
Mike Frysingerde2b9382005-09-27 03:18:00 +00001244 cn->info = OC_FUNC;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001245 cn->r.f = newfunc(t_string);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001246 cn->l.n = condition();
1247 break;
1248
Denis Vlasenkof782f522007-01-01 23:51:30 +00001249 case TC_SEQSTART:
Glenn L McGrath545106f2002-11-11 06:21:00 +00001250 cn = vn->r.n = parse_expr(TC_SEQTERM);
1251 cn->a.n = vn;
1252 break;
1253
Denis Vlasenkof782f522007-01-01 23:51:30 +00001254 case TC_GETLINE:
Glenn L McGrath545106f2002-11-11 06:21:00 +00001255 glptr = cn;
1256 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1257 break;
1258
Denis Vlasenkof782f522007-01-01 23:51:30 +00001259 case TC_BUILTIN:
Glenn L McGrath545106f2002-11-11 06:21:00 +00001260 cn->l.n = condition();
1261 break;
1262 }
1263 }
1264 }
1265 }
1266 return sn.r.n;
1267}
1268
1269/* add node to chain. Return ptr to alloc'd node */
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001270static node *chain_node(uint32_t info)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001271{
"Robert P. J. Day"68229832006-07-01 13:08:46 +00001272 node *n;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001273
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00001274 if (!seq->first)
Glenn L McGrath545106f2002-11-11 06:21:00 +00001275 seq->first = seq->last = new_node(0);
1276
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001277 if (seq->programname != g_progname) {
1278 seq->programname = g_progname;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001279 n = chain_node(OC_NEWSOURCE);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001280 n->l.s = xstrdup(g_progname);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001281 }
1282
1283 n = seq->last;
1284 n->info = info;
1285 seq->last = n->a.n = new_node(OC_DONE);
1286
1287 return n;
1288}
1289
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001290static void chain_expr(uint32_t info)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001291{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001292 node *n;
1293
1294 n = chain_node(info);
1295 n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001296 if (t_tclass & TC_GRPTERM)
Glenn L McGrath545106f2002-11-11 06:21:00 +00001297 rollback_token();
1298}
1299
Mike Frysinger10a11e22005-09-27 02:23:02 +00001300static node *chain_loop(node *nn)
1301{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001302 node *n, *n2, *save_brk, *save_cont;
1303
1304 save_brk = break_ptr;
1305 save_cont = continue_ptr;
1306
1307 n = chain_node(OC_BR | Vx);
1308 continue_ptr = new_node(OC_EXEC);
1309 break_ptr = new_node(OC_EXEC);
1310 chain_group();
1311 n2 = chain_node(OC_EXEC | Vx);
1312 n2->l.n = nn;
1313 n2->a.n = n;
1314 continue_ptr->a.n = n2;
1315 break_ptr->a.n = n->r.n = seq->last;
1316
1317 continue_ptr = save_cont;
1318 break_ptr = save_brk;
1319
1320 return n;
1321}
1322
1323/* parse group and attach it to chain */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001324static void chain_group(void)
1325{
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001326 uint32_t c;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001327 node *n, *n2, *n3;
1328
1329 do {
1330 c = next_token(TC_GRPSEQ);
1331 } while (c & TC_NEWLINE);
1332
1333 if (c & TC_GRPSTART) {
Denis Vlasenkobf0a2012006-12-26 10:42:51 +00001334 while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001335 if (t_tclass & TC_NEWLINE) continue;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001336 rollback_token();
1337 chain_group();
1338 }
1339 } else if (c & (TC_OPSEQ | TC_OPTERM)) {
1340 rollback_token();
1341 chain_expr(OC_EXEC | Vx);
1342 } else { /* TC_STATEMNT */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001343 switch (t_info & OPCLSMASK) {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001344 case ST_IF:
1345 n = chain_node(OC_BR | Vx);
1346 n->l.n = condition();
1347 chain_group();
1348 n2 = chain_node(OC_EXEC);
1349 n->r.n = seq->last;
1350 if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE) == TC_ELSE) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001351 chain_group();
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001352 n2->a.n = seq->last;
1353 } else {
1354 rollback_token();
1355 }
1356 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001357
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001358 case ST_WHILE:
1359 n2 = condition();
1360 n = chain_loop(NULL);
1361 n->l.n = n2;
1362 break;
1363
1364 case ST_DO:
1365 n2 = chain_node(OC_EXEC);
1366 n = chain_loop(NULL);
1367 n2->a.n = n->a.n;
1368 next_token(TC_WHILE);
1369 n->l.n = condition();
1370 break;
1371
1372 case ST_FOR:
1373 next_token(TC_SEQSTART);
1374 n2 = parse_expr(TC_SEMICOL | TC_SEQTERM);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001375 if (t_tclass & TC_SEQTERM) { /* for-in */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001376 if ((n2->info & OPCLSMASK) != OC_IN)
1377 syntax_error(EMSG_UNEXP_TOKEN);
1378 n = chain_node(OC_WALKINIT | VV);
1379 n->l.n = n2->l.n;
1380 n->r.n = n2->r.n;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001381 n = chain_loop(NULL);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001382 n->info = OC_WALKNEXT | Vx;
1383 n->l.n = n2->l.n;
1384 } else { /* for (;;) */
1385 n = chain_node(OC_EXEC | Vx);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001386 n->l.n = n2;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001387 n2 = parse_expr(TC_SEMICOL);
1388 n3 = parse_expr(TC_SEQTERM);
1389 n = chain_loop(n3);
1390 n->l.n = n2;
Denis Vlasenkob78c7822007-07-18 18:31:11 +00001391 if (!n2)
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001392 n->info = OC_EXEC;
1393 }
1394 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001395
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001396 case OC_PRINT:
1397 case OC_PRINTF:
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001398 n = chain_node(t_info);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001399 n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001400 if (t_tclass & TC_OUTRDR) {
1401 n->info |= t_info;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001402 n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM);
1403 }
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001404 if (t_tclass & TC_GRPTERM)
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001405 rollback_token();
1406 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001407
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001408 case OC_BREAK:
1409 n = chain_node(OC_EXEC);
1410 n->a.n = break_ptr;
1411 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001412
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001413 case OC_CONTINUE:
1414 n = chain_node(OC_EXEC);
1415 n->a.n = continue_ptr;
1416 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001417
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001418 /* delete, next, nextfile, return, exit */
1419 default:
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001420 chain_expr(t_info);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001421 }
1422 }
1423}
1424
Mike Frysinger10a11e22005-09-27 02:23:02 +00001425static void parse_program(char *p)
1426{
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001427 uint32_t tclass;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001428 node *cn;
1429 func *f;
1430 var *v;
1431
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001432 g_pos = p;
1433 t_lineno = 1;
Denis Vlasenkobf0a2012006-12-26 10:42:51 +00001434 while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART |
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001435 TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001436
1437 if (tclass & TC_OPTERM)
1438 continue;
1439
1440 seq = &mainseq;
1441 if (tclass & TC_BEGIN) {
1442 seq = &beginseq;
1443 chain_group();
1444
1445 } else if (tclass & TC_END) {
1446 seq = &endseq;
1447 chain_group();
1448
1449 } else if (tclass & TC_FUNCDECL) {
1450 next_token(TC_FUNCTION);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001451 g_pos++;
1452 f = newfunc(t_string);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001453 f->body.first = NULL;
1454 f->nargs = 0;
Denis Vlasenkobf0a2012006-12-26 10:42:51 +00001455 while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001456 v = findvar(ahash, t_string);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001457 v->x.aidx = (f->nargs)++;
1458
1459 if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM)
1460 break;
1461 }
1462 seq = &(f->body);
1463 chain_group();
1464 clear_array(ahash);
1465
1466 } else if (tclass & TC_OPSEQ) {
1467 rollback_token();
1468 cn = chain_node(OC_TEST);
1469 cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001470 if (t_tclass & TC_GRPSTART) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001471 rollback_token();
1472 chain_group();
1473 } else {
1474 chain_node(OC_PRINT);
1475 }
1476 cn->r.n = mainseq.last;
1477
1478 } else /* if (tclass & TC_GRPSTART) */ {
1479 rollback_token();
1480 chain_group();
1481 }
1482 }
1483}
1484
1485
1486/* -------- program execution part -------- */
1487
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001488static node *mk_splitter(const char *s, tsplitter *spl)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001489{
"Robert P. J. Day"68229832006-07-01 13:08:46 +00001490 regex_t *re, *ire;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001491 node *n;
1492
1493 re = &spl->re[0];
1494 ire = &spl->re[1];
1495 n = &spl->n;
Denis Vlasenko890ac9d2006-10-07 15:16:19 +00001496 if ((n->info & OPCLSMASK) == OC_REGEXP) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001497 regfree(re);
Denis Vlasenkob78c7822007-07-18 18:31:11 +00001498 regfree(ire); // TODO: nuke ire, use re+1?
Glenn L McGrath545106f2002-11-11 06:21:00 +00001499 }
Rob Landleya3896512006-05-07 20:20:34 +00001500 if (strlen(s) > 1) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001501 mk_re_node(s, n, re);
1502 } else {
Mike Frysingerf87b3e32005-09-27 04:16:22 +00001503 n->info = (uint32_t) *s;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001504 }
1505
1506 return n;
1507}
1508
1509/* use node as a regular expression. Supplied with node ptr and regex_t
Eric Andersenaff114c2004-04-14 17:51:38 +00001510 * storage space. Return ptr to regex (if result points to preg, it should
Glenn L McGrath545106f2002-11-11 06:21:00 +00001511 * be later regfree'd manually
1512 */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001513static regex_t *as_regex(node *op, regex_t *preg)
1514{
Denis Vlasenko7a676642009-03-15 22:20:31 +00001515 int cflags;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001516 var *v;
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001517 const char *s;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001518
1519 if ((op->info & OPCLSMASK) == OC_REGEXP) {
1520 return icase ? op->r.ire : op->l.re;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001521 }
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001522 v = nvalloc(1);
1523 s = getvar_s(evaluate(op, v));
Denis Vlasenko7a676642009-03-15 22:20:31 +00001524
1525 cflags = icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED;
1526 /* Testcase where REG_EXTENDED fails (unpaired '{'):
1527 * echo Hi | awk 'gsub("@(samp|code|file)\{","");'
1528 * gawk 3.1.5 eats this. We revert to ~REG_EXTENDED
1529 * (maybe gsub is not supposed to use REG_EXTENDED?).
1530 */
1531 if (regcomp(preg, s, cflags)) {
1532 cflags &= ~REG_EXTENDED;
1533 xregcomp(preg, s, cflags);
1534 }
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001535 nvfree(v);
1536 return preg;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001537}
1538
1539/* gradually increasing buffer */
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001540static char* qrealloc(char *b, int n, int *size)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001541{
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001542 if (!b || n >= *size) {
Denis Vlasenkodeeed592008-07-08 05:14:36 +00001543 *size = n + (n>>1) + 80;
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001544 b = xrealloc(b, *size);
Denis Vlasenkodeeed592008-07-08 05:14:36 +00001545 }
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001546 return b;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001547}
1548
1549/* resize field storage space */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001550static void fsrealloc(int size)
1551{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001552 int i;
1553
1554 if (size >= maxfields) {
1555 i = maxfields;
1556 maxfields = size + 16;
Denis Vlasenkof782f522007-01-01 23:51:30 +00001557 Fields = xrealloc(Fields, maxfields * sizeof(var));
1558 for (; i < maxfields; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001559 Fields[i].type = VF_SPECIAL;
1560 Fields[i].string = NULL;
1561 }
1562 }
1563
1564 if (size < nfields) {
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00001565 for (i = size; i < nfields; i++) {
1566 clrvar(Fields + i);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001567 }
1568 }
1569 nfields = size;
1570}
1571
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001572static int awk_split(const char *s, node *spl, char **slist)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001573{
Denis Vlasenkof782f522007-01-01 23:51:30 +00001574 int l, n = 0;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001575 char c[4];
1576 char *s1;
Denis Vlasenkob78c7822007-07-18 18:31:11 +00001577 regmatch_t pmatch[2]; // TODO: why [2]? [1] is enough...
Glenn L McGrath545106f2002-11-11 06:21:00 +00001578
1579 /* in worst case, each char would be a separate field */
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001580 *slist = s1 = xzalloc(strlen(s) * 2 + 3);
1581 strcpy(s1, s);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001582
1583 c[0] = c[1] = (char)spl->info;
1584 c[2] = c[3] = '\0';
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001585 if (*getvar_s(intvar[RS]) == '\0')
1586 c[2] = '\n';
Glenn L McGrath545106f2002-11-11 06:21:00 +00001587
Denis Vlasenkoaf1bd092007-07-18 18:32:25 +00001588 if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */
1589 if (!*s)
1590 return n; /* "": zero fields */
1591 n++; /* at least one field will be there */
1592 do {
1593 l = strcspn(s, c+2); /* len till next NUL or \n */
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001594 if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0
1595 && pmatch[0].rm_so <= l
1596 ) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001597 l = pmatch[0].rm_so;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001598 if (pmatch[0].rm_eo == 0) {
1599 l++;
1600 pmatch[0].rm_eo++;
1601 }
Denis Vlasenkoaf1bd092007-07-18 18:32:25 +00001602 n++; /* we saw yet another delimiter */
Glenn L McGrath545106f2002-11-11 06:21:00 +00001603 } else {
1604 pmatch[0].rm_eo = l;
Denys Vlasenkoe4244232009-05-18 23:50:03 +02001605 if (s[l])
1606 pmatch[0].rm_eo++;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001607 }
Glenn L McGrath545106f2002-11-11 06:21:00 +00001608 memcpy(s1, s, l);
Denis Vlasenko67b5eeb2009-04-12 13:54:13 +00001609 /* make sure we remove *all* of the separator chars */
Denys Vlasenkoe4244232009-05-18 23:50:03 +02001610 do {
1611 s1[l] = '\0';
1612 } while (++l < pmatch[0].rm_eo);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001613 nextword(&s1);
1614 s += pmatch[0].rm_eo;
Denis Vlasenkoaf1bd092007-07-18 18:32:25 +00001615 } while (*s);
1616 return n;
1617 }
1618 if (c[0] == '\0') { /* null split */
Denis Vlasenkobf0a2012006-12-26 10:42:51 +00001619 while (*s) {
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00001620 *s1++ = *s++;
1621 *s1++ = '\0';
Glenn L McGrath545106f2002-11-11 06:21:00 +00001622 n++;
1623 }
Denis Vlasenkoaf1bd092007-07-18 18:32:25 +00001624 return n;
1625 }
1626 if (c[0] != ' ') { /* single-character split */
Glenn L McGrath545106f2002-11-11 06:21:00 +00001627 if (icase) {
1628 c[0] = toupper(c[0]);
1629 c[1] = tolower(c[1]);
1630 }
1631 if (*s1) n++;
1632 while ((s1 = strpbrk(s1, c))) {
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00001633 *s1++ = '\0';
Glenn L McGrath545106f2002-11-11 06:21:00 +00001634 n++;
1635 }
Denis Vlasenkoaf1bd092007-07-18 18:32:25 +00001636 return n;
1637 }
1638 /* space split */
1639 while (*s) {
1640 s = skip_whitespace(s);
1641 if (!*s) break;
1642 n++;
1643 while (*s && !isspace(*s))
1644 *s1++ = *s++;
1645 *s1++ = '\0';
Glenn L McGrath545106f2002-11-11 06:21:00 +00001646 }
1647 return n;
1648}
1649
Mike Frysinger10a11e22005-09-27 02:23:02 +00001650static void split_f0(void)
1651{
Denis Vlasenkoaf1bd092007-07-18 18:32:25 +00001652/* static char *fstrings; */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001653#define fstrings (G.split_f0__fstrings)
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001654
Glenn L McGrath545106f2002-11-11 06:21:00 +00001655 int i, n;
1656 char *s;
1657
1658 if (is_f0_split)
1659 return;
1660
1661 is_f0_split = TRUE;
Aaron Lehmanna170e1c2002-11-28 11:27:31 +00001662 free(fstrings);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001663 fsrealloc(0);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001664 n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001665 fsrealloc(n);
1666 s = fstrings;
Denis Vlasenkof782f522007-01-01 23:51:30 +00001667 for (i = 0; i < n; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001668 Fields[i].string = nextword(&s);
1669 Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY);
1670 }
1671
1672 /* set NF manually to avoid side effects */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001673 clrvar(intvar[NF]);
1674 intvar[NF]->type = VF_NUMBER | VF_SPECIAL;
1675 intvar[NF]->number = nfields;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001676#undef fstrings
Glenn L McGrath545106f2002-11-11 06:21:00 +00001677}
1678
1679/* perform additional actions when some internal variables changed */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001680static void handle_special(var *v)
1681{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001682 int n;
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001683 char *b;
1684 const char *sep, *s;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001685 int sl, l, len, i, bsize;
1686
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00001687 if (!(v->type & VF_SPECIAL))
Glenn L McGrath545106f2002-11-11 06:21:00 +00001688 return;
1689
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001690 if (v == intvar[NF]) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001691 n = (int)getvar_i(v);
1692 fsrealloc(n);
1693
1694 /* recalculate $0 */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001695 sep = getvar_s(intvar[OFS]);
Rob Landleya3896512006-05-07 20:20:34 +00001696 sl = strlen(sep);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001697 b = NULL;
1698 len = 0;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001699 for (i = 0; i < n; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001700 s = getvar_s(&Fields[i]);
Rob Landleya3896512006-05-07 20:20:34 +00001701 l = strlen(s);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001702 if (b) {
1703 memcpy(b+len, sep, sl);
1704 len += sl;
1705 }
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001706 b = qrealloc(b, len+l+sl, &bsize);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001707 memcpy(b+len, s, l);
1708 len += l;
1709 }
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00001710 if (b)
1711 b[len] = '\0';
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001712 setvar_p(intvar[F0], b);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001713 is_f0_split = TRUE;
1714
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001715 } else if (v == intvar[F0]) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001716 is_f0_split = FALSE;
1717
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001718 } else if (v == intvar[FS]) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001719 mk_splitter(getvar_s(v), &fsplitter);
1720
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001721 } else if (v == intvar[RS]) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001722 mk_splitter(getvar_s(v), &rsplitter);
1723
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001724 } else if (v == intvar[IGNORECASE]) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001725 icase = istrue(v);
1726
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00001727 } else { /* $n */
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001728 n = getvar_i(intvar[NF]);
1729 setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001730 /* right here v is invalid. Just to note... */
1731 }
1732}
1733
1734/* step through func/builtin/etc arguments */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001735static node *nextarg(node **pn)
1736{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001737 node *n;
1738
1739 n = *pn;
1740 if (n && (n->info & OPCLSMASK) == OC_COMMA) {
1741 *pn = n->r.n;
1742 n = n->l.n;
1743 } else {
1744 *pn = NULL;
1745 }
1746 return n;
1747}
1748
Mike Frysinger10a11e22005-09-27 02:23:02 +00001749static void hashwalk_init(var *v, xhash *array)
1750{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001751 hash_item *hi;
Denis Vlasenko77ad97f2008-05-13 02:27:31 +00001752 unsigned i;
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001753 walker_list *w;
1754 walker_list *prev_walker;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001755
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001756 if (v->type & VF_WALK) {
1757 prev_walker = v->x.walker;
1758 } else {
1759 v->type |= VF_WALK;
1760 prev_walker = NULL;
1761 }
1762 debug_printf_walker("hashwalk_init: prev_walker:%p\n", prev_walker);
Denys Vlasenko3cb60c32010-03-10 19:20:32 +01001763
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001764 w = v->x.walker = xzalloc(sizeof(*w) + array->glen + 1); /* why + 1? */
1765 debug_printf_walker(" walker@%p=%p\n", &v->x.walker, w);
1766 w->cur = w->end = w->wbuf;
1767 w->prev = prev_walker;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001768 for (i = 0; i < array->csize; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001769 hi = array->items[i];
Denis Vlasenkobf0a2012006-12-26 10:42:51 +00001770 while (hi) {
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001771 strcpy(w->end, hi->name);
1772 nextword(&w->end);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001773 hi = hi->next;
1774 }
1775 }
1776}
1777
Mike Frysinger10a11e22005-09-27 02:23:02 +00001778static int hashwalk_next(var *v)
1779{
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001780 walker_list *w = v->x.walker;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001781
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001782 if (w->cur >= w->end) {
1783 walker_list *prev_walker = w->prev;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001784
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001785 debug_printf_walker("end of iteration, free(walker@%p:%p), prev_walker:%p\n", &v->x.walker, w, prev_walker);
1786 free(w);
Denys Vlasenko3cb60c32010-03-10 19:20:32 +01001787 v->x.walker = prev_walker;
1788 return FALSE;
1789 }
1790
Denys Vlasenkoda62b092010-03-11 12:13:18 +01001791 setvar_s(v, nextword(&w->cur));
Glenn L McGrath545106f2002-11-11 06:21:00 +00001792 return TRUE;
1793}
1794
1795/* evaluate node, return 1 when result is true, 0 otherwise */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001796static int ptest(node *pattern)
1797{
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001798 /* ptest__v is "static": to save stack space? */
1799 return istrue(evaluate(pattern, &G.ptest__v));
Glenn L McGrath545106f2002-11-11 06:21:00 +00001800}
1801
1802/* read next record from stream rsm into a variable v */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001803static int awk_getline(rstream *rsm, var *v)
1804{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001805 char *b;
1806 regmatch_t pmatch[2];
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001807 int a, p, pp=0, size;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001808 int fd, so, eo, r, rp;
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001809 char c, *m, *s;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001810
1811 /* we're using our own buffer since we need access to accumulating
1812 * characters
1813 */
1814 fd = fileno(rsm->F);
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001815 m = rsm->buffer;
1816 a = rsm->adv;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001817 p = rsm->pos;
1818 size = rsm->size;
1819 c = (char) rsplitter.n.info;
1820 rp = 0;
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001821
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001822 if (!m)
1823 m = qrealloc(m, 256, &size);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001824 do {
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001825 b = m + a;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001826 so = eo = p;
1827 r = 1;
1828 if (p > 0) {
1829 if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) {
1830 if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re,
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001831 b, 1, pmatch, 0) == 0) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001832 so = pmatch[0].rm_so;
1833 eo = pmatch[0].rm_eo;
1834 if (b[eo] != '\0')
1835 break;
1836 }
1837 } else if (c != '\0') {
1838 s = strchr(b+pp, c);
Denis Vlasenkob78c7822007-07-18 18:31:11 +00001839 if (!s) s = memchr(b+pp, '\0', p - pp);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001840 if (s) {
1841 so = eo = s-b;
1842 eo++;
1843 break;
1844 }
1845 } else {
1846 while (b[rp] == '\n')
1847 rp++;
1848 s = strstr(b+rp, "\n\n");
1849 if (s) {
1850 so = eo = s-b;
1851 while (b[eo] == '\n') eo++;
1852 if (b[eo] != '\0')
1853 break;
1854 }
1855 }
1856 }
1857
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001858 if (a > 0) {
1859 memmove(m, (const void *)(m+a), p+1);
1860 b = m;
1861 a = 0;
1862 }
1863
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001864 m = qrealloc(m, a+p+128, &size);
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001865 b = m + a;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001866 pp = p;
1867 p += safe_read(fd, b+p, size-p-1);
1868 if (p < pp) {
1869 p = 0;
1870 r = 0;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001871 setvar_i(intvar[ERRNO], errno);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001872 }
1873 b[p] = '\0';
1874
1875 } while (p > pp);
1876
1877 if (p == 0) {
1878 r--;
1879 } else {
1880 c = b[so]; b[so] = '\0';
1881 setvar_s(v, b+rp);
1882 v->type |= VF_USER;
1883 b[so] = c;
1884 c = b[eo]; b[eo] = '\0';
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001885 setvar_s(intvar[RT], b+so);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001886 b[eo] = c;
1887 }
1888
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00001889 rsm->buffer = m;
1890 rsm->adv = a + eo;
1891 rsm->pos = p - eo;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001892 rsm->size = size;
1893
1894 return r;
1895}
1896
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +00001897static int fmt_num(char *b, int size, const char *format, double n, int int_as_int)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001898{
Denis Vlasenkof782f522007-01-01 23:51:30 +00001899 int r = 0;
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +00001900 char c;
Denis Vlasenkof782f522007-01-01 23:51:30 +00001901 const char *s = format;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001902
1903 if (int_as_int && n == (int)n) {
1904 r = snprintf(b, size, "%d", (int)n);
1905 } else {
Denis Vlasenkof782f522007-01-01 23:51:30 +00001906 do { c = *s; } while (c && *++s);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001907 if (strchr("diouxX", c)) {
1908 r = snprintf(b, size, format, (int)n);
1909 } else if (strchr("eEfgG", c)) {
1910 r = snprintf(b, size, format, n);
1911 } else {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00001912 syntax_error(EMSG_INV_FMT);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001913 }
1914 }
1915 return r;
1916}
1917
Glenn L McGrath545106f2002-11-11 06:21:00 +00001918/* formatted output into an allocated buffer, return ptr to buffer */
Mike Frysinger10a11e22005-09-27 02:23:02 +00001919static char *awk_printf(node *n)
1920{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001921 char *b = NULL;
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001922 char *fmt, *s, *f;
1923 const char *s1;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001924 int i, j, incr, bsize;
1925 char c, c1;
1926 var *v, *arg;
1927
1928 v = nvalloc(1);
Rob Landleyd921b2e2006-08-03 15:41:12 +00001929 fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v)));
Glenn L McGrath545106f2002-11-11 06:21:00 +00001930
1931 i = 0;
1932 while (*f) {
1933 s = f;
1934 while (*f && (*f != '%' || *(++f) == '%'))
1935 f++;
Denis Vlasenko389f9d52007-05-09 21:57:23 +00001936 while (*f && !isalpha(*f)) {
1937 if (*f == '*')
1938 syntax_error("%*x formats are not supported");
Glenn L McGrath545106f2002-11-11 06:21:00 +00001939 f++;
Denis Vlasenko389f9d52007-05-09 21:57:23 +00001940 }
Glenn L McGrath545106f2002-11-11 06:21:00 +00001941
1942 incr = (f - s) + MAXVARFMT;
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001943 b = qrealloc(b, incr + i, &bsize);
Denis Vlasenkof782f522007-01-01 23:51:30 +00001944 c = *f;
1945 if (c != '\0') f++;
1946 c1 = *f;
1947 *f = '\0';
Glenn L McGrath545106f2002-11-11 06:21:00 +00001948 arg = evaluate(nextarg(&n), v);
1949
1950 j = i;
1951 if (c == 'c' || !c) {
Denis Vlasenkof782f522007-01-01 23:51:30 +00001952 i += sprintf(b+i, s, is_numeric(arg) ?
1953 (char)getvar_i(arg) : *getvar_s(arg));
Glenn L McGrath545106f2002-11-11 06:21:00 +00001954 } else if (c == 's') {
Denis Vlasenko92758142006-10-03 19:56:34 +00001955 s1 = getvar_s(arg);
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01001956 b = qrealloc(b, incr+i+strlen(s1), &bsize);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001957 i += sprintf(b+i, s, s1);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001958 } else {
1959 i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE);
1960 }
1961 *f = c1;
1962
1963 /* if there was an error while sprintf, return value is negative */
1964 if (i < j) i = j;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001965 }
1966
Denis Vlasenkof782f522007-01-01 23:51:30 +00001967 b = xrealloc(b, i + 1);
Glenn L McGrath545106f2002-11-11 06:21:00 +00001968 free(fmt);
1969 nvfree(v);
1970 b[i] = '\0';
1971 return b;
1972}
1973
1974/* common substitution routine
1975 * replace (nm) substring of (src) that match (n) with (repl), store
1976 * result into (dest), return number of substitutions. If nm=0, replace
1977 * all matches. If src or dst is NULL, use $0. If ex=TRUE, enable
1978 * subexpression matching (\1-\9)
1979 */
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001980static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int ex)
Mike Frysinger10a11e22005-09-27 02:23:02 +00001981{
Glenn L McGrath545106f2002-11-11 06:21:00 +00001982 char *ds = NULL;
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00001983 const char *s;
1984 const char *sp;
Glenn L McGrath545106f2002-11-11 06:21:00 +00001985 int c, i, j, di, rl, so, eo, nbs, n, dssize;
1986 regmatch_t pmatch[10];
1987 regex_t sreg, *re;
1988
1989 re = as_regex(rn, &sreg);
Denis Vlasenkob78c7822007-07-18 18:31:11 +00001990 if (!src) src = intvar[F0];
1991 if (!dest) dest = intvar[F0];
Glenn L McGrath545106f2002-11-11 06:21:00 +00001992
1993 i = di = 0;
1994 sp = getvar_s(src);
Rob Landleya3896512006-05-07 20:20:34 +00001995 rl = strlen(repl);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00001996 while (regexec(re, sp, 10, pmatch, sp==getvar_s(src) ? 0 : REG_NOTBOL) == 0) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00001997 so = pmatch[0].rm_so;
1998 eo = pmatch[0].rm_eo;
1999
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01002000 ds = qrealloc(ds, di + eo + rl, &dssize);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002001 memcpy(ds + di, sp, eo);
2002 di += eo;
2003 if (++i >= nm) {
2004 /* replace */
2005 di -= (eo - so);
2006 nbs = 0;
2007 for (s = repl; *s; s++) {
2008 ds[di++] = c = *s;
2009 if (c == '\\') {
2010 nbs++;
2011 continue;
2012 }
2013 if (c == '&' || (ex && c >= '0' && c <= '9')) {
2014 di -= ((nbs + 3) >> 1);
2015 j = 0;
2016 if (c != '&') {
2017 j = c - '0';
2018 nbs++;
2019 }
2020 if (nbs % 2) {
2021 ds[di++] = c;
2022 } else {
2023 n = pmatch[j].rm_eo - pmatch[j].rm_so;
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01002024 ds = qrealloc(ds, di + rl + n, &dssize);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002025 memcpy(ds + di, sp + pmatch[j].rm_so, n);
2026 di += n;
2027 }
2028 }
2029 nbs = 0;
2030 }
2031 }
2032
2033 sp += eo;
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002034 if (i == nm)
2035 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002036 if (eo == so) {
Denis Vlasenkob78c7822007-07-18 18:31:11 +00002037 ds[di] = *sp++;
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002038 if (!ds[di++])
2039 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002040 }
2041 }
2042
Denys Vlasenkoc9955f22010-03-10 19:21:54 +01002043 ds = qrealloc(ds, di + strlen(sp), &dssize);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002044 strcpy(ds + di, sp);
2045 setvar_p(dest, ds);
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002046 if (re == &sreg)
2047 regfree(re);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002048 return i;
2049}
2050
Leonid Lisovskiy46a0be52009-09-21 04:08:08 +02002051static NOINLINE int do_mktime(const char *ds)
2052{
2053 struct tm then;
2054 int count;
2055
2056 /*memset(&then, 0, sizeof(then)); - not needed */
2057 then.tm_isdst = -1; /* default is unknown */
2058
2059 /* manpage of mktime says these fields are ints,
2060 * so we can sscanf stuff directly into them */
2061 count = sscanf(ds, "%u %u %u %u %u %u %d",
2062 &then.tm_year, &then.tm_mon, &then.tm_mday,
2063 &then.tm_hour, &then.tm_min, &then.tm_sec,
2064 &then.tm_isdst);
2065
2066 if (count < 6
2067 || (unsigned)then.tm_mon < 1
2068 || (unsigned)then.tm_year < 1900
2069 ) {
2070 return -1;
2071 }
2072
2073 then.tm_mon -= 1;
Denys Vlasenkobc3e9472009-09-21 04:16:00 +02002074 then.tm_year -= 1900;
Leonid Lisovskiy46a0be52009-09-21 04:08:08 +02002075
2076 return mktime(&then);
2077}
2078
2079static NOINLINE var *exec_builtin(node *op, var *res)
Mike Frysinger10a11e22005-09-27 02:23:02 +00002080{
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002081#define tspl (G.exec_builtin__tspl)
2082
Glenn L McGrath545106f2002-11-11 06:21:00 +00002083 var *tv;
2084 node *an[4];
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002085 var *av[4];
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00002086 const char *as[4];
Glenn L McGrath545106f2002-11-11 06:21:00 +00002087 regmatch_t pmatch[2];
2088 regex_t sreg, *re;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002089 node *spl;
Mike Frysingerf87b3e32005-09-27 04:16:22 +00002090 uint32_t isr, info;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002091 int nargs;
2092 time_t tt;
2093 char *s, *s1;
2094 int i, l, ll, n;
2095
2096 tv = nvalloc(4);
2097 isr = info = op->info;
2098 op = op->l.n;
2099
2100 av[2] = av[3] = NULL;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002101 for (i = 0; i < 4 && op; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002102 an[i] = nextarg(&op);
2103 if (isr & 0x09000000) av[i] = evaluate(an[i], &tv[i]);
2104 if (isr & 0x08000000) as[i] = getvar_s(av[i]);
2105 isr >>= 1;
2106 }
2107
2108 nargs = i;
Denis Vlasenko77ad97f2008-05-13 02:27:31 +00002109 if ((uint32_t)nargs < (info >> 30))
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002110 syntax_error(EMSG_TOO_FEW_ARGS);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002111
Denys Vlasenko56b3eec2009-10-23 13:03:59 +02002112 info &= OPNMASK;
2113 switch (info) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002114
Denis Vlasenkof782f522007-01-01 23:51:30 +00002115 case B_a2:
Denis Vlasenko2d5bd802008-10-24 10:49:49 +00002116#if ENABLE_FEATURE_AWK_LIBM
Denis Vlasenko37890e22008-10-21 12:59:34 +00002117 setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1])));
Glenn L McGrath545106f2002-11-11 06:21:00 +00002118#else
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002119 syntax_error(EMSG_NO_MATH);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002120#endif
2121 break;
2122
Denis Vlasenkof782f522007-01-01 23:51:30 +00002123 case B_sp:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002124 if (nargs > 2) {
2125 spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ?
2126 an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl);
2127 } else {
2128 spl = &fsplitter.n;
2129 }
2130
2131 n = awk_split(as[0], spl, &s);
2132 s1 = s;
2133 clear_array(iamarray(av[1]));
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00002134 for (i = 1; i <= n; i++)
Glenn L McGrath545106f2002-11-11 06:21:00 +00002135 setari_u(av[1], i, nextword(&s1));
2136 free(s);
2137 setvar_i(res, n);
2138 break;
2139
Denis Vlasenkof782f522007-01-01 23:51:30 +00002140 case B_ss:
Rob Landleya3896512006-05-07 20:20:34 +00002141 l = strlen(as[0]);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002142 i = getvar_i(av[1]) - 1;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002143 if (i > l) i = l;
2144 if (i < 0) i = 0;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002145 n = (nargs > 2) ? getvar_i(av[2]) : l-i;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002146 if (n < 0) n = 0;
Denis Vlasenko8ae5b282008-07-02 22:47:49 +00002147 s = xstrndup(as[0]+i, n);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002148 setvar_p(res, s);
2149 break;
Denis Vlasenkof7996f32007-01-11 17:20:00 +00002150
Denis Vlasenko7cbcd1c2008-08-28 23:16:58 +00002151 /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5:
2152 * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002153 case B_an:
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00002154 setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1]));
Denis Vlasenkoe175ff22006-09-26 17:41:00 +00002155 break;
Denis Vlasenkof7996f32007-01-11 17:20:00 +00002156
Denis Vlasenkof782f522007-01-01 23:51:30 +00002157 case B_co:
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00002158 setvar_i(res, ~getvar_i_int(av[0]));
Denis Vlasenkoe175ff22006-09-26 17:41:00 +00002159 break;
2160
Denis Vlasenkof782f522007-01-01 23:51:30 +00002161 case B_ls:
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00002162 setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1]));
Denis Vlasenkoe175ff22006-09-26 17:41:00 +00002163 break;
2164
Denis Vlasenkof782f522007-01-01 23:51:30 +00002165 case B_or:
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00002166 setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1]));
Denis Vlasenkoe175ff22006-09-26 17:41:00 +00002167 break;
2168
Denis Vlasenkof782f522007-01-01 23:51:30 +00002169 case B_rs:
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00002170 setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1]));
Denis Vlasenkoe175ff22006-09-26 17:41:00 +00002171 break;
2172
Denis Vlasenkof782f522007-01-01 23:51:30 +00002173 case B_xo:
Denis Vlasenkoa2e1eea2008-09-02 09:00:23 +00002174 setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1]));
Denis Vlasenkoe175ff22006-09-26 17:41:00 +00002175 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002176
Denis Vlasenkof782f522007-01-01 23:51:30 +00002177 case B_lo:
Denis Vlasenkof782f522007-01-01 23:51:30 +00002178 case B_up:
Rob Landleyd921b2e2006-08-03 15:41:12 +00002179 s1 = s = xstrdup(as[0]);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002180 while (*s1) {
Denys Vlasenko56b3eec2009-10-23 13:03:59 +02002181 //*s1 = (info == B_up) ? toupper(*s1) : tolower(*s1);
2182 if ((unsigned char)((*s1 | 0x20) - 'a') <= ('z' - 'a'))
2183 *s1 = (info == B_up) ? (*s1 & 0xdf) : (*s1 | 0x20);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002184 s1++;
2185 }
2186 setvar_p(res, s);
2187 break;
2188
Denis Vlasenkof782f522007-01-01 23:51:30 +00002189 case B_ix:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002190 n = 0;
Rob Landleya3896512006-05-07 20:20:34 +00002191 ll = strlen(as[1]);
2192 l = strlen(as[0]) - ll;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002193 if (ll > 0 && l >= 0) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002194 if (!icase) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002195 s = strstr(as[0], as[1]);
2196 if (s) n = (s - as[0]) + 1;
2197 } else {
2198 /* this piece of code is terribly slow and
2199 * really should be rewritten
2200 */
2201 for (i=0; i<=l; i++) {
2202 if (strncasecmp(as[0]+i, as[1], ll) == 0) {
2203 n = i+1;
2204 break;
2205 }
2206 }
2207 }
2208 }
2209 setvar_i(res, n);
2210 break;
2211
Denis Vlasenkof782f522007-01-01 23:51:30 +00002212 case B_ti:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002213 if (nargs > 1)
2214 tt = getvar_i(av[1]);
2215 else
2216 time(&tt);
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00002217 //s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y";
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002218 i = strftime(g_buf, MAXVARFMT,
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00002219 ((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"),
2220 localtime(&tt));
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002221 g_buf[i] = '\0';
2222 setvar_s(res, g_buf);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002223 break;
2224
Leonid Lisovskiy46a0be52009-09-21 04:08:08 +02002225 case B_mt:
2226 setvar_i(res, do_mktime(as[0]));
2227 break;
2228
Denis Vlasenkof782f522007-01-01 23:51:30 +00002229 case B_ma:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002230 re = as_regex(an[1], &sreg);
2231 n = regexec(re, as[0], 1, pmatch, 0);
2232 if (n == 0) {
2233 pmatch[0].rm_so++;
2234 pmatch[0].rm_eo++;
2235 } else {
2236 pmatch[0].rm_so = 0;
2237 pmatch[0].rm_eo = -1;
2238 }
2239 setvar_i(newvar("RSTART"), pmatch[0].rm_so);
2240 setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so);
2241 setvar_i(res, pmatch[0].rm_so);
2242 if (re == &sreg) regfree(re);
2243 break;
2244
Denis Vlasenkof782f522007-01-01 23:51:30 +00002245 case B_ge:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002246 awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE);
2247 break;
2248
Denis Vlasenkof782f522007-01-01 23:51:30 +00002249 case B_gs:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002250 setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE));
2251 break;
2252
Denis Vlasenkof782f522007-01-01 23:51:30 +00002253 case B_su:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002254 setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE));
2255 break;
2256 }
2257
2258 nvfree(tv);
2259 return res;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002260#undef tspl
Glenn L McGrath545106f2002-11-11 06:21:00 +00002261}
2262
2263/*
2264 * Evaluate node - the heart of the program. Supplied with subtree
2265 * and place where to store result. returns ptr to result.
2266 */
2267#define XC(n) ((n) >> 8)
2268
Mike Frysinger10a11e22005-09-27 02:23:02 +00002269static var *evaluate(node *op, var *res)
2270{
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002271/* This procedure is recursive so we should count every byte */
2272#define fnargs (G.evaluate__fnargs)
2273/* seed is initialized to 1 */
2274#define seed (G.evaluate__seed)
2275#define sreg (G.evaluate__sreg)
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00002276
Glenn L McGrath545106f2002-11-11 06:21:00 +00002277 node *op1;
2278 var *v1;
2279 union {
2280 var *v;
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00002281 const char *s;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002282 double d;
2283 int i;
2284 } L, R;
Mike Frysingerf87b3e32005-09-27 04:16:22 +00002285 uint32_t opinfo;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002286 int opn;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002287 union {
2288 char *s;
2289 rstream *rsm;
2290 FILE *F;
2291 var *v;
2292 regex_t *re;
Mike Frysingerf87b3e32005-09-27 04:16:22 +00002293 uint32_t info;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002294 } X;
2295
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00002296 if (!op)
Glenn L McGrath545106f2002-11-11 06:21:00 +00002297 return setvar_s(res, NULL);
2298
2299 v1 = nvalloc(2);
2300
2301 while (op) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002302 opinfo = op->info;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002303 opn = (opinfo & OPNMASK);
2304 g_lineno = op->lineno;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002305
Mike Frysingerde2b9382005-09-27 03:18:00 +00002306 /* execute inevitable things */
Glenn L McGrath545106f2002-11-11 06:21:00 +00002307 op1 = op->l.n;
2308 if (opinfo & OF_RES1) X.v = L.v = evaluate(op1, v1);
2309 if (opinfo & OF_RES2) R.v = evaluate(op->r.n, v1+1);
2310 if (opinfo & OF_STR1) L.s = getvar_s(L.v);
2311 if (opinfo & OF_STR2) R.s = getvar_s(R.v);
2312 if (opinfo & OF_NUM1) L.d = getvar_i(L.v);
2313
2314 switch (XC(opinfo & OPCLSMASK)) {
2315
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002316 /* -- iterative node type -- */
Glenn L McGrath545106f2002-11-11 06:21:00 +00002317
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002318 /* test pattern */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002319 case XC( OC_TEST ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002320 if ((op1->info & OPCLSMASK) == OC_COMMA) {
2321 /* it's range pattern */
2322 if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) {
2323 op->info |= OF_CHECKED;
2324 if (ptest(op1->r.n))
2325 op->info &= ~OF_CHECKED;
2326
2327 op = op->a.n;
2328 } else {
2329 op = op->r.n;
2330 }
2331 } else {
2332 op = (ptest(op1)) ? op->a.n : op->r.n;
2333 }
2334 break;
2335
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002336 /* just evaluate an expression, also used as unconditional jump */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002337 case XC( OC_EXEC ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002338 break;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002339
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002340 /* branch, used in if-else and various loops */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002341 case XC( OC_BR ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002342 op = istrue(L.v) ? op->a.n : op->r.n;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002343 break;
2344
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002345 /* initialize for-in loop */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002346 case XC( OC_WALKINIT ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002347 hashwalk_init(L.v, iamarray(R.v));
Glenn L McGrath545106f2002-11-11 06:21:00 +00002348 break;
2349
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002350 /* get next array item */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002351 case XC( OC_WALKNEXT ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002352 op = hashwalk_next(L.v) ? op->a.n : op->r.n;
2353 break;
2354
Denis Vlasenkof782f522007-01-01 23:51:30 +00002355 case XC( OC_PRINT ):
2356 case XC( OC_PRINTF ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002357 X.F = stdout;
Mike Frysingerde2b9382005-09-27 03:18:00 +00002358 if (op->r.n) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002359 X.rsm = newfile(R.s);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002360 if (!X.rsm->F) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002361 if (opn == '|') {
Denis Vlasenko51742f42007-04-12 00:32:05 +00002362 X.rsm->F = popen(R.s, "w");
2363 if (X.rsm->F == NULL)
Manuel Novoa III cad53642003-03-19 09:13:01 +00002364 bb_perror_msg_and_die("popen");
Glenn L McGrath545106f2002-11-11 06:21:00 +00002365 X.rsm->is_pipe = 1;
2366 } else {
Rob Landleyd921b2e2006-08-03 15:41:12 +00002367 X.rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
Glenn L McGrath545106f2002-11-11 06:21:00 +00002368 }
2369 }
2370 X.F = X.rsm->F;
2371 }
2372
2373 if ((opinfo & OPCLSMASK) == OC_PRINT) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002374 if (!op1) {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002375 fputs(getvar_s(intvar[F0]), X.F);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002376 } else {
2377 while (op1) {
2378 L.v = evaluate(nextarg(&op1), v1);
2379 if (L.v->type & VF_NUMBER) {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002380 fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]),
Denis Vlasenkob54b2082006-10-27 09:05:40 +00002381 getvar_i(L.v), TRUE);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002382 fputs(g_buf, X.F);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002383 } else {
2384 fputs(getvar_s(L.v), X.F);
2385 }
2386
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002387 if (op1) fputs(getvar_s(intvar[OFS]), X.F);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002388 }
2389 }
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002390 fputs(getvar_s(intvar[ORS]), X.F);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002391
2392 } else { /* OC_PRINTF */
2393 L.s = awk_printf(op1);
2394 fputs(L.s, X.F);
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00002395 free((char*)L.s);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002396 }
2397 fflush(X.F);
2398 break;
2399
Denis Vlasenkof782f522007-01-01 23:51:30 +00002400 case XC( OC_DELETE ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002401 X.info = op1->info & OPCLSMASK;
2402 if (X.info == OC_VAR) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002403 R.v = op1->l.v;
2404 } else if (X.info == OC_FNARG) {
2405 R.v = &fnargs[op1->l.i];
2406 } else {
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002407 syntax_error(EMSG_NOT_ARRAY);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002408 }
2409
Mike Frysingerde2b9382005-09-27 03:18:00 +00002410 if (op1->r.n) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002411 clrvar(L.v);
2412 L.s = getvar_s(evaluate(op1->r.n, v1));
2413 hash_remove(iamarray(R.v), L.s);
2414 } else {
2415 clear_array(iamarray(R.v));
2416 }
2417 break;
2418
Denis Vlasenkof782f522007-01-01 23:51:30 +00002419 case XC( OC_NEWSOURCE ):
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002420 g_progname = op->l.s;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002421 break;
2422
Denis Vlasenkof782f522007-01-01 23:51:30 +00002423 case XC( OC_RETURN ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002424 copyvar(res, L.v);
2425 break;
2426
Denis Vlasenkof782f522007-01-01 23:51:30 +00002427 case XC( OC_NEXTFILE ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002428 nextfile = TRUE;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002429 case XC( OC_NEXT ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002430 nextrec = TRUE;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002431 case XC( OC_DONE ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002432 clrvar(res);
2433 break;
2434
Denis Vlasenkof782f522007-01-01 23:51:30 +00002435 case XC( OC_EXIT ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002436 awk_exit(L.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002437
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002438 /* -- recursive node type -- */
Glenn L McGrath545106f2002-11-11 06:21:00 +00002439
Denis Vlasenkof782f522007-01-01 23:51:30 +00002440 case XC( OC_VAR ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002441 L.v = op->l.v;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002442 if (L.v == intvar[NF])
Glenn L McGrath545106f2002-11-11 06:21:00 +00002443 split_f0();
2444 goto v_cont;
2445
Denis Vlasenkof782f522007-01-01 23:51:30 +00002446 case XC( OC_FNARG ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002447 L.v = &fnargs[op->l.i];
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00002448 v_cont:
2449 res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002450 break;
2451
Denis Vlasenkof782f522007-01-01 23:51:30 +00002452 case XC( OC_IN ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002453 setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0);
2454 break;
2455
Denis Vlasenkof782f522007-01-01 23:51:30 +00002456 case XC( OC_REGEXP ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002457 op1 = op;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002458 L.s = getvar_s(intvar[F0]);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002459 goto re_cont;
2460
Denis Vlasenkof782f522007-01-01 23:51:30 +00002461 case XC( OC_MATCH ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002462 op1 = op->r.n;
Denis Vlasenkoe1d3e032007-01-01 23:53:52 +00002463 re_cont:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002464 X.re = as_regex(op1, &sreg);
2465 R.i = regexec(X.re, L.s, 0, NULL, 0);
2466 if (X.re == &sreg) regfree(X.re);
Denys Vlasenko12847742009-11-30 01:15:04 +01002467 setvar_i(res, (R.i == 0) ^ (opn == '!'));
Glenn L McGrath545106f2002-11-11 06:21:00 +00002468 break;
2469
Denis Vlasenkof782f522007-01-01 23:51:30 +00002470 case XC( OC_MOVE ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002471 /* if source is a temporary string, jusk relink it to dest */
Denys Vlasenko12847742009-11-30 01:15:04 +01002472//Disabled: if R.v is numeric but happens to have cached R.v->string,
2473//then L.v ends up being a string, which is wrong
2474// if (R.v == v1+1 && R.v->string) {
2475// res = setvar_p(L.v, R.v->string);
2476// R.v->string = NULL;
2477// } else {
Mike Frysingerde2b9382005-09-27 03:18:00 +00002478 res = copyvar(L.v, R.v);
Denys Vlasenko12847742009-11-30 01:15:04 +01002479// }
Glenn L McGrath545106f2002-11-11 06:21:00 +00002480 break;
2481
Denis Vlasenkof782f522007-01-01 23:51:30 +00002482 case XC( OC_TERNARY ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002483 if ((op->r.n->info & OPCLSMASK) != OC_COLON)
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002484 syntax_error(EMSG_POSSIBLE_ERROR);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002485 res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res);
2486 break;
2487
Denis Vlasenkof782f522007-01-01 23:51:30 +00002488 case XC( OC_FUNC ):
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002489 if (!op->r.f->body.first)
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002490 syntax_error(EMSG_UNDEF_FUNC);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002491
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002492 X.v = R.v = nvalloc(op->r.f->nargs + 1);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002493 while (op1) {
2494 L.v = evaluate(nextarg(&op1), v1);
2495 copyvar(R.v, L.v);
2496 R.v->type |= VF_CHILD;
2497 R.v->x.parent = L.v;
2498 if (++R.v - X.v >= op->r.f->nargs)
2499 break;
2500 }
2501
2502 R.v = fnargs;
2503 fnargs = X.v;
2504
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002505 L.s = g_progname;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002506 res = evaluate(op->r.f->body.first, res);
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002507 g_progname = L.s;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002508
2509 nvfree(fnargs);
2510 fnargs = R.v;
2511 break;
2512
Denis Vlasenkof782f522007-01-01 23:51:30 +00002513 case XC( OC_GETLINE ):
2514 case XC( OC_PGETLINE ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002515 if (op1) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002516 X.rsm = newfile(L.s);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002517 if (!X.rsm->F) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002518 if ((opinfo & OPCLSMASK) == OC_PGETLINE) {
2519 X.rsm->F = popen(L.s, "r");
2520 X.rsm->is_pipe = TRUE;
2521 } else {
Denis Vlasenko5415c852008-07-21 23:05:26 +00002522 X.rsm->F = fopen_for_read(L.s); /* not xfopen! */
Glenn L McGrath545106f2002-11-11 06:21:00 +00002523 }
2524 }
2525 } else {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002526 if (!iF) iF = next_input_file();
Glenn L McGrath545106f2002-11-11 06:21:00 +00002527 X.rsm = iF;
2528 }
2529
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002530 if (!X.rsm->F) {
2531 setvar_i(intvar[ERRNO], errno);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002532 setvar_i(res, -1);
2533 break;
2534 }
2535
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002536 if (!op->r.n)
2537 R.v = intvar[F0];
Glenn L McGrath545106f2002-11-11 06:21:00 +00002538
2539 L.i = awk_getline(X.rsm, R.v);
2540 if (L.i > 0) {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002541 if (!op1) {
2542 incvar(intvar[FNR]);
2543 incvar(intvar[NR]);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002544 }
2545 }
2546 setvar_i(res, L.i);
2547 break;
2548
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002549 /* simple builtins */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002550 case XC( OC_FBLTIN ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002551 switch (opn) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002552
Denis Vlasenkof782f522007-01-01 23:51:30 +00002553 case F_in:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002554 R.d = (int)L.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002555 break;
2556
Denis Vlasenkof782f522007-01-01 23:51:30 +00002557 case F_rn:
2558 R.d = (double)rand() / (double)RAND_MAX;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002559 break;
Denis Vlasenko2d5bd802008-10-24 10:49:49 +00002560#if ENABLE_FEATURE_AWK_LIBM
Denis Vlasenkof782f522007-01-01 23:51:30 +00002561 case F_co:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002562 R.d = cos(L.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002563 break;
2564
Denis Vlasenkof782f522007-01-01 23:51:30 +00002565 case F_ex:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002566 R.d = exp(L.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002567 break;
2568
Denis Vlasenkof782f522007-01-01 23:51:30 +00002569 case F_lg:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002570 R.d = log(L.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002571 break;
2572
Denis Vlasenkof782f522007-01-01 23:51:30 +00002573 case F_si:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002574 R.d = sin(L.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002575 break;
2576
Denis Vlasenkof782f522007-01-01 23:51:30 +00002577 case F_sq:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002578 R.d = sqrt(L.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002579 break;
2580#else
Denis Vlasenkof782f522007-01-01 23:51:30 +00002581 case F_co:
2582 case F_ex:
2583 case F_lg:
2584 case F_si:
2585 case F_sq:
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002586 syntax_error(EMSG_NO_MATH);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002587 break;
2588#endif
Denis Vlasenkof782f522007-01-01 23:51:30 +00002589 case F_sr:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002590 R.d = (double)seed;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002591 seed = op1 ? (unsigned)L.d : (unsigned)time(NULL);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002592 srand(seed);
2593 break;
2594
Denis Vlasenkof782f522007-01-01 23:51:30 +00002595 case F_ti:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002596 R.d = time(NULL);
2597 break;
2598
Denis Vlasenkof782f522007-01-01 23:51:30 +00002599 case F_le:
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002600 if (!op1)
2601 L.s = getvar_s(intvar[F0]);
Rob Landleya3896512006-05-07 20:20:34 +00002602 R.d = strlen(L.s);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002603 break;
2604
Denis Vlasenkof782f522007-01-01 23:51:30 +00002605 case F_sy:
Denys Vlasenko8131eea2009-11-02 14:19:51 +01002606 fflush_all();
Denis Vlasenko249fabf2006-12-19 00:29:22 +00002607 R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
2608 ? (system(L.s) >> 8) : 0;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002609 break;
2610
Denis Vlasenkof782f522007-01-01 23:51:30 +00002611 case F_ff:
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002612 if (!op1)
Glenn L McGrath545106f2002-11-11 06:21:00 +00002613 fflush(stdout);
2614 else {
2615 if (L.s && *L.s) {
2616 X.rsm = newfile(L.s);
2617 fflush(X.rsm->F);
2618 } else {
Denys Vlasenko8131eea2009-11-02 14:19:51 +01002619 fflush_all();
Glenn L McGrath545106f2002-11-11 06:21:00 +00002620 }
2621 }
2622 break;
2623
Denis Vlasenkof782f522007-01-01 23:51:30 +00002624 case F_cl:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002625 X.rsm = (rstream *)hash_search(fdhash, L.s);
2626 if (X.rsm) {
2627 R.i = X.rsm->is_pipe ? pclose(X.rsm->F) : fclose(X.rsm->F);
Aaron Lehmanna170e1c2002-11-28 11:27:31 +00002628 free(X.rsm->buffer);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002629 hash_remove(fdhash, L.s);
2630 }
2631 if (R.i != 0)
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002632 setvar_i(intvar[ERRNO], errno);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002633 R.d = (double)R.i;
2634 break;
2635 }
2636 setvar_i(res, R.d);
2637 break;
2638
Denis Vlasenkof782f522007-01-01 23:51:30 +00002639 case XC( OC_BUILTIN ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002640 res = exec_builtin(op, res);
2641 break;
2642
Denis Vlasenkof782f522007-01-01 23:51:30 +00002643 case XC( OC_SPRINTF ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002644 setvar_p(res, awk_printf(op1));
Glenn L McGrath545106f2002-11-11 06:21:00 +00002645 break;
2646
Denis Vlasenkof782f522007-01-01 23:51:30 +00002647 case XC( OC_UNARY ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002648 X.v = R.v;
2649 L.d = R.d = getvar_i(R.v);
2650 switch (opn) {
Denis Vlasenkof782f522007-01-01 23:51:30 +00002651 case 'P':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002652 L.d = ++R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002653 goto r_op_change;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002654 case 'p':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002655 R.d++;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002656 goto r_op_change;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002657 case 'M':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002658 L.d = --R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002659 goto r_op_change;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002660 case 'm':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002661 R.d--;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002662 goto r_op_change;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002663 case '!':
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002664 L.d = !istrue(X.v);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002665 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002666 case '-':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002667 L.d = -R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002668 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002669 r_op_change:
Glenn L McGrath545106f2002-11-11 06:21:00 +00002670 setvar_i(X.v, R.d);
2671 }
2672 setvar_i(res, L.d);
2673 break;
2674
Denis Vlasenkof782f522007-01-01 23:51:30 +00002675 case XC( OC_FIELD ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002676 R.i = (int)getvar_i(R.v);
2677 if (R.i == 0) {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002678 res = intvar[F0];
Glenn L McGrath545106f2002-11-11 06:21:00 +00002679 } else {
2680 split_f0();
2681 if (R.i > nfields)
2682 fsrealloc(R.i);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002683 res = &Fields[R.i - 1];
Glenn L McGrath545106f2002-11-11 06:21:00 +00002684 }
2685 break;
2686
Denis Vlasenkocd5c7862007-05-17 16:37:22 +00002687 /* concatenation (" ") and index joining (",") */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002688 case XC( OC_CONCAT ):
2689 case XC( OC_COMMA ):
Rob Landleya3896512006-05-07 20:20:34 +00002690 opn = strlen(L.s) + strlen(R.s) + 2;
Denis Vlasenkob95636c2006-12-19 23:36:04 +00002691 X.s = xmalloc(opn);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002692 strcpy(X.s, L.s);
2693 if ((opinfo & OPCLSMASK) == OC_COMMA) {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002694 L.s = getvar_s(intvar[SUBSEP]);
Denis Vlasenkof782f522007-01-01 23:51:30 +00002695 X.s = xrealloc(X.s, opn + strlen(L.s));
Glenn L McGrath545106f2002-11-11 06:21:00 +00002696 strcat(X.s, L.s);
2697 }
2698 strcat(X.s, R.s);
2699 setvar_p(res, X.s);
2700 break;
2701
Denis Vlasenkof782f522007-01-01 23:51:30 +00002702 case XC( OC_LAND ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002703 setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0);
2704 break;
2705
Denis Vlasenkof782f522007-01-01 23:51:30 +00002706 case XC( OC_LOR ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002707 setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n));
2708 break;
2709
Denis Vlasenkof782f522007-01-01 23:51:30 +00002710 case XC( OC_BINARY ):
2711 case XC( OC_REPLACE ):
Mike Frysingerde2b9382005-09-27 03:18:00 +00002712 R.d = getvar_i(R.v);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002713 switch (opn) {
Denis Vlasenkof782f522007-01-01 23:51:30 +00002714 case '+':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002715 L.d += R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002716 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002717 case '-':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002718 L.d -= R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002719 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002720 case '*':
Mike Frysingerde2b9382005-09-27 03:18:00 +00002721 L.d *= R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002722 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002723 case '/':
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002724 if (R.d == 0)
2725 syntax_error(EMSG_DIV_BY_ZERO);
Mike Frysingerde2b9382005-09-27 03:18:00 +00002726 L.d /= R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002727 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002728 case '&':
Denis Vlasenko2d5bd802008-10-24 10:49:49 +00002729#if ENABLE_FEATURE_AWK_LIBM
Mike Frysingerde2b9382005-09-27 03:18:00 +00002730 L.d = pow(L.d, R.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002731#else
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002732 syntax_error(EMSG_NO_MATH);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002733#endif
2734 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002735 case '%':
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002736 if (R.d == 0)
2737 syntax_error(EMSG_DIV_BY_ZERO);
Mike Frysingerde2b9382005-09-27 03:18:00 +00002738 L.d -= (int)(L.d / R.d) * R.d;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002739 break;
2740 }
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002741 res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : X.v, L.d);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002742 break;
2743
Denis Vlasenkof782f522007-01-01 23:51:30 +00002744 case XC( OC_COMPARE ):
Glenn L McGrath545106f2002-11-11 06:21:00 +00002745 if (is_numeric(L.v) && is_numeric(R.v)) {
2746 L.d = getvar_i(L.v) - getvar_i(R.v);
2747 } else {
2748 L.s = getvar_s(L.v);
2749 R.s = getvar_s(R.v);
2750 L.d = icase ? strcasecmp(L.s, R.s) : strcmp(L.s, R.s);
2751 }
2752 switch (opn & 0xfe) {
Denis Vlasenkof782f522007-01-01 23:51:30 +00002753 case 0:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002754 R.i = (L.d > 0);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002755 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002756 case 2:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002757 R.i = (L.d >= 0);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002758 break;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002759 case 4:
Mike Frysingerde2b9382005-09-27 03:18:00 +00002760 R.i = (L.d == 0);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002761 break;
2762 }
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002763 setvar_i(res, (opn & 1 ? R.i : !R.i) ? 1 : 0);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002764 break;
2765
Denis Vlasenkof782f522007-01-01 23:51:30 +00002766 default:
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002767 syntax_error(EMSG_POSSIBLE_ERROR);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002768 }
2769 if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS)
2770 op = op->a.n;
2771 if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS)
2772 break;
2773 if (nextrec)
2774 break;
2775 }
2776 nvfree(v1);
2777 return res;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002778#undef fnargs
2779#undef seed
2780#undef sreg
Glenn L McGrath545106f2002-11-11 06:21:00 +00002781}
2782
2783
2784/* -------- main & co. -------- */
2785
Mike Frysinger10a11e22005-09-27 02:23:02 +00002786static int awk_exit(int r)
2787{
Denis Vlasenkof782f522007-01-01 23:51:30 +00002788 var tv;
2789 unsigned i;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002790 hash_item *hi;
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00002791
Denis Vlasenkof782f522007-01-01 23:51:30 +00002792 zero_out_var(&tv);
2793
2794 if (!exiting) {
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00002795 exiting = TRUE;
Glenn L McGrathca29ffc2004-09-24 09:24:27 +00002796 nextrec = FALSE;
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00002797 evaluate(endseq.first, &tv);
2798 }
Glenn L McGrath545106f2002-11-11 06:21:00 +00002799
2800 /* waiting for children */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002801 for (i = 0; i < fdhash->csize; i++) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002802 hi = fdhash->items[i];
Denis Vlasenkobf0a2012006-12-26 10:42:51 +00002803 while (hi) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002804 if (hi->data.rs.F && hi->data.rs.is_pipe)
2805 pclose(hi->data.rs.F);
2806 hi = hi->next;
2807 }
2808 }
2809
2810 exit(r);
2811}
2812
2813/* if expr looks like "var=value", perform assignment and return 1,
2814 * otherwise return 0 */
"Vladimir N. Oleynik"5cf9a032005-10-19 09:21:51 +00002815static int is_assignment(const char *expr)
Mike Frysinger10a11e22005-09-27 02:23:02 +00002816{
Glenn L McGrath545106f2002-11-11 06:21:00 +00002817 char *exprc, *s, *s0, *s1;
2818
Rob Landleyd921b2e2006-08-03 15:41:12 +00002819 exprc = xstrdup(expr);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002820 if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) {
2821 free(exprc);
2822 return FALSE;
2823 }
2824
2825 *(s++) = '\0';
2826 s0 = s1 = s;
2827 while (*s)
2828 *(s1++) = nextchar(&s);
2829
2830 *s1 = '\0';
2831 setvar_u(newvar(exprc), s0);
2832 free(exprc);
2833 return TRUE;
2834}
2835
2836/* switch to next input file */
Mike Frysinger10a11e22005-09-27 02:23:02 +00002837static rstream *next_input_file(void)
2838{
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002839#define rsm (G.next_input_file__rsm)
2840#define files_happen (G.next_input_file__files_happen)
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002841
Glenn L McGrath545106f2002-11-11 06:21:00 +00002842 FILE *F = NULL;
Denis Vlasenkoa41fdf32007-01-29 22:51:00 +00002843 const char *fname, *ind;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002844
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002845 if (rsm.F)
2846 fclose(rsm.F);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002847 rsm.F = NULL;
Glenn L McGrath00ed36f2003-10-30 13:36:39 +00002848 rsm.pos = rsm.adv = 0;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002849
2850 do {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002851 if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002852 if (files_happen)
2853 return NULL;
2854 fname = "-";
2855 F = stdin;
2856 } else {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002857 ind = getvar_s(incvar(intvar[ARGIND]));
2858 fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind));
Glenn L McGrath545106f2002-11-11 06:21:00 +00002859 if (fname && *fname && !is_assignment(fname))
Denis Vlasenko3bb2bbd2008-07-01 01:57:36 +00002860 F = xfopen_stdin(fname);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002861 }
2862 } while (!F);
2863
2864 files_happen = TRUE;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002865 setvar_s(intvar[FILENAME], fname);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002866 rsm.F = F;
2867 return &rsm;
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002868#undef rsm
2869#undef files_happen
Glenn L McGrath545106f2002-11-11 06:21:00 +00002870}
2871
Denis Vlasenko9b49a5e2007-10-11 10:05:36 +00002872int awk_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Rob Landleydfba7412006-03-06 20:47:33 +00002873int awk_main(int argc, char **argv)
Mike Frysinger10a11e22005-09-27 02:23:02 +00002874{
Denis Vlasenko67b23e62006-10-03 21:00:06 +00002875 unsigned opt;
Denis Vlasenkobe644a82007-03-10 17:22:14 +00002876 char *opt_F, *opt_W;
Denis Vlasenko3bb2bbd2008-07-01 01:57:36 +00002877 llist_t *list_v = NULL;
2878 llist_t *list_f = NULL;
2879 int i, j;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002880 var *v;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002881 var tv;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002882 char **envp;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002883 char *vnames = (char *)vNames; /* cheat */
2884 char *vvalues = (char *)vValues;
2885
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002886 INIT_G();
2887
Denis Vlasenko150f4022007-01-13 21:06:21 +00002888 /* Undo busybox.c, or else strtod may eat ','! This breaks parsing:
Denis Vlasenko6dc6ebb2007-01-01 23:53:12 +00002889 * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */
2890 if (ENABLE_LOCALE_SUPPORT)
2891 setlocale(LC_NUMERIC, "C");
2892
Denis Vlasenkof782f522007-01-01 23:51:30 +00002893 zero_out_var(&tv);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002894
2895 /* allocate global buffer */
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002896 g_buf = xmalloc(MAXVARFMT + 1);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002897
2898 vhash = hash_init();
2899 ahash = hash_init();
2900 fdhash = hash_init();
2901 fnhash = hash_init();
2902
2903 /* initialize variables */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002904 for (i = 0; *vnames; i++) {
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002905 intvar[i] = v = newvar(nextword(&vnames));
Denis Vlasenkof782f522007-01-01 23:51:30 +00002906 if (*vvalues != '\377')
2907 setvar_s(v, nextword(&vvalues));
Glenn L McGrath545106f2002-11-11 06:21:00 +00002908 else
2909 setvar_i(v, 0);
2910
Denis Vlasenkof782f522007-01-01 23:51:30 +00002911 if (*vnames == '*') {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002912 v->type |= VF_SPECIAL;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002913 vnames++;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002914 }
2915 }
2916
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002917 handle_special(intvar[FS]);
2918 handle_special(intvar[RS]);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002919
Denis Vlasenkof782f522007-01-01 23:51:30 +00002920 newfile("/dev/stdin")->F = stdin;
2921 newfile("/dev/stdout")->F = stdout;
2922 newfile("/dev/stderr")->F = stderr;
Glenn L McGrath545106f2002-11-11 06:21:00 +00002923
Denis Vlasenkof71d9162007-05-03 22:57:56 +00002924 /* Huh, people report that sometimes environ is NULL. Oh well. */
2925 if (environ) for (envp = environ; *envp; envp++) {
Denis Vlasenkob78c7822007-07-18 18:31:11 +00002926 /* environ is writable, thus we don't strdup it needlessly */
2927 char *s = *envp;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002928 char *s1 = strchr(s, '=');
2929 if (s1) {
Denis Vlasenkob78c7822007-07-18 18:31:11 +00002930 *s1 = '\0';
2931 /* Both findvar and setvar_u take const char*
2932 * as 2nd arg -> environment is not trashed */
2933 setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1);
2934 *s1 = '=';
Eric Andersen67776be2004-07-30 23:52:08 +00002935 }
Glenn L McGrath545106f2002-11-11 06:21:00 +00002936 }
Denis Vlasenko3bb2bbd2008-07-01 01:57:36 +00002937 opt_complementary = "v::f::"; /* -v and -f can occur multiple times */
2938 opt = getopt32(argv, "F:v:f:W:", &opt_F, &list_v, &list_f, &opt_W);
Denis Vlasenkof782f522007-01-01 23:51:30 +00002939 argv += optind;
2940 argc -= optind;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002941 if (opt & 0x1)
2942 setvar_s(intvar[FS], opt_F); // -F
Denis Vlasenko3bb2bbd2008-07-01 01:57:36 +00002943 while (list_v) { /* -v */
2944 if (!is_assignment(llist_pop(&list_v)))
Denis Vlasenkobe644a82007-03-10 17:22:14 +00002945 bb_show_usage();
2946 }
Denis Vlasenko3bb2bbd2008-07-01 01:57:36 +00002947 if (list_f) { /* -f */
2948 do {
2949 char *s = NULL;
2950 FILE *from_file;
2951
2952 g_progname = llist_pop(&list_f);
2953 from_file = xfopen_stdin(g_progname);
2954 /* one byte is reserved for some trick in next_token */
Denis Vlasenkof782f522007-01-01 23:51:30 +00002955 for (i = j = 1; j > 0; i += j) {
2956 s = xrealloc(s, i + 4096);
2957 j = fread(s + i, 1, 4094, from_file);
Denis Vlasenko099efbf2006-09-22 09:02:30 +00002958 }
Denis Vlasenko3bb2bbd2008-07-01 01:57:36 +00002959 s[i] = '\0';
2960 fclose(from_file);
2961 parse_program(s + 1);
2962 free(s);
2963 } while (list_f);
Denis Vlasenko41d5ebe2009-01-25 01:00:15 +00002964 argc++;
Denis Vlasenkof782f522007-01-01 23:51:30 +00002965 } else { // no -f: take program from 1st parameter
2966 if (!argc)
2967 bb_show_usage();
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00002968 g_progname = "cmd. line";
Denis Vlasenkof782f522007-01-01 23:51:30 +00002969 parse_program(*argv++);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002970 }
Denis Vlasenko099efbf2006-09-22 09:02:30 +00002971 if (opt & 0x8) // -W
Denis Vlasenkoe1a0d482006-10-20 13:28:22 +00002972 bb_error_msg("warning: unrecognized option '-W %s' ignored", opt_W);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002973
Glenn L McGrath545106f2002-11-11 06:21:00 +00002974 /* fill in ARGV array */
Denis Vlasenko41d5ebe2009-01-25 01:00:15 +00002975 setvar_i(intvar[ARGC], argc);
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002976 setari_u(intvar[ARGV], 0, "awk");
Denis Vlasenkof782f522007-01-01 23:51:30 +00002977 i = 0;
2978 while (*argv)
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002979 setari_u(intvar[ARGV], ++i, *argv++);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002980
2981 evaluate(beginseq.first, &tv);
Denis Vlasenkof782f522007-01-01 23:51:30 +00002982 if (!mainseq.first && !endseq.first)
Glenn L McGrath545106f2002-11-11 06:21:00 +00002983 awk_exit(EXIT_SUCCESS);
2984
2985 /* input file could already be opened in BEGIN block */
Denys Vlasenkocdeda162009-11-30 01:14:16 +01002986 if (!iF)
2987 iF = next_input_file();
Glenn L McGrath545106f2002-11-11 06:21:00 +00002988
2989 /* passing through input files */
2990 while (iF) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002991 nextfile = FALSE;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002992 setvar_i(intvar[FNR], 0);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002993
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002994 while ((i = awk_getline(iF, intvar[F0])) > 0) {
Glenn L McGrath545106f2002-11-11 06:21:00 +00002995 nextrec = FALSE;
Denis Vlasenkoffba9412007-05-17 23:03:35 +00002996 incvar(intvar[NR]);
2997 incvar(intvar[FNR]);
Glenn L McGrath545106f2002-11-11 06:21:00 +00002998 evaluate(mainseq.first, &tv);
2999
3000 if (nextfile)
3001 break;
3002 }
3003
Denis Vlasenkof782f522007-01-01 23:51:30 +00003004 if (i < 0)
Denis Vlasenkoae5a8aa2007-06-06 17:01:00 +00003005 syntax_error(strerror(errno));
Glenn L McGrath545106f2002-11-11 06:21:00 +00003006
3007 iF = next_input_file();
Glenn L McGrath545106f2002-11-11 06:21:00 +00003008 }
3009
Glenn L McGrath545106f2002-11-11 06:21:00 +00003010 awk_exit(EXIT_SUCCESS);
Denis Vlasenkof782f522007-01-01 23:51:30 +00003011 /*return 0;*/
Glenn L McGrath545106f2002-11-11 06:21:00 +00003012}