source: MondoRescue/branches/3.3/mindi-busybox/editors/awk.c@ 3621

Last change on this file since 3621 was 3621, checked in by Bruno Cornec, 7 years ago

New 3?3 banch for incorporation of latest busybox 1.25. Changing minor version to handle potential incompatibilities.
File size: 76.5 KB
Line 
1/* vi: set sw=4 ts=4: */
2/*
3 * awk implementation for busybox
4 *
5 * Copyright (C) 2002 by Dmitry Zakharov <dmit@crp.bank.gov.ua>
6 *
7 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
8 */
9
10//config:config AWK
11//config: bool "awk"
12//config: default y
13//config: help
14//config: Awk is used as a pattern scanning and processing language. This is
15//config: the BusyBox implementation of that programming language.
16//config:
17//config:config FEATURE_AWK_LIBM
18//config: bool "Enable math functions (requires libm)"
19//config: default y
20//config: depends on AWK
21//config: help
22//config: Enable math functions of the Awk programming language.
23//config: NOTE: This will require libm to be present for linking.
24//config:
25//config:config FEATURE_AWK_GNU_EXTENSIONS
26//config: bool "Enable a few GNU extensions"
27//config: default y
28//config: depends on AWK
29//config: help
30//config: Enable a few features from gawk:
31//config: * command line option -e AWK_PROGRAM
32//config: * simultaneous use of -f and -e on the command line.
33//config: This enables the use of awk library files.
34//config: Ex: awk -f mylib.awk -e '{print myfunction($1);}' ...
35
36//applet:IF_AWK(APPLET_NOEXEC(awk, awk, BB_DIR_USR_BIN, BB_SUID_DROP, awk))
37
38//kbuild:lib-$(CONFIG_AWK) += awk.o
39
40//usage:#define awk_trivial_usage
41//usage: "[OPTIONS] [AWK_PROGRAM] [FILE]..."
42//usage:#define awk_full_usage "\n\n"
43//usage: " -v VAR=VAL Set variable"
44//usage: "\n -F SEP Use SEP as field separator"
45//usage: "\n -f FILE Read program from FILE"
46//usage: IF_FEATURE_AWK_GNU_EXTENSIONS(
47//usage: "\n -e AWK_PROGRAM"
48//usage: )
49
50#include "libbb.h"
51#include "xregex.h"
52#include <math.h>
53
54/* This is a NOEXEC applet. Be very careful! */
55
56
57/* If you comment out one of these below, it will be #defined later
58 * to perform debug printfs to stderr: */
59#define debug_printf_walker(...) do {} while (0)
60#define debug_printf_eval(...) do {} while (0)
61#define debug_printf_parse(...) do {} while (0)
62
63#ifndef debug_printf_walker
64# define debug_printf_walker(...) (fprintf(stderr, __VA_ARGS__))
65#endif
66#ifndef debug_printf_eval
67# define debug_printf_eval(...) (fprintf(stderr, __VA_ARGS__))
68#endif
69#ifndef debug_printf_parse
70# define debug_printf_parse(...) (fprintf(stderr, __VA_ARGS__))
71#endif
72
73
74#define OPTSTR_AWK \
75 "F:v:f:" \
76 IF_FEATURE_AWK_GNU_EXTENSIONS("e:") \
77 "W:"
78#define OPTCOMPLSTR_AWK \
79 "v::f::" \
80 IF_FEATURE_AWK_GNU_EXTENSIONS("e::")
81enum {
82 OPTBIT_F, /* define field separator */
83 OPTBIT_v, /* define variable */
84 OPTBIT_f, /* pull in awk program from file */
85 IF_FEATURE_AWK_GNU_EXTENSIONS(OPTBIT_e,) /* -e AWK_PROGRAM */
86 OPTBIT_W, /* -W ignored */
87 OPT_F = 1 << OPTBIT_F,
88 OPT_v = 1 << OPTBIT_v,
89 OPT_f = 1 << OPTBIT_f,
90 OPT_e = IF_FEATURE_AWK_GNU_EXTENSIONS((1 << OPTBIT_e)) + 0,
91 OPT_W = 1 << OPTBIT_W
92};
93
94#define MAXVARFMT 240
95#define MINNVBLOCK 64
96
97/* variable flags */
98#define VF_NUMBER 0x0001 /* 1 = primary type is number */
99#define VF_ARRAY 0x0002 /* 1 = it's an array */
100
101#define VF_CACHED 0x0100 /* 1 = num/str value has cached str/num eq */
102#define VF_USER 0x0200 /* 1 = user input (may be numeric string) */
103#define VF_SPECIAL 0x0400 /* 1 = requires extra handling when changed */
104#define VF_WALK 0x0800 /* 1 = variable has alloc'd x.walker list */
105#define VF_FSTR 0x1000 /* 1 = var::string points to fstring buffer */
106#define VF_CHILD 0x2000 /* 1 = function arg; x.parent points to source */
107#define VF_DIRTY 0x4000 /* 1 = variable was set explicitly */
108
109/* these flags are static, don't change them when value is changed */
110#define VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY)
111
112typedef struct walker_list {
113 char *end;
114 char *cur;
115 struct walker_list *prev;
116 char wbuf[1];
117} walker_list;
118
119/* Variable */
120typedef struct var_s {
121 unsigned type; /* flags */
122 double number;
123 char *string;
124 union {
125 int aidx; /* func arg idx (for compilation stage) */
126 struct xhash_s *array; /* array ptr */
127 struct var_s *parent; /* for func args, ptr to actual parameter */
128 walker_list *walker; /* list of array elements (for..in) */
129 } x;
130} var;
131
132/* Node chain (pattern-action chain, BEGIN, END, function bodies) */
133typedef struct chain_s {
134 struct node_s *first;
135 struct node_s *last;
136 const char *programname;
137} chain;
138
139/* Function */
140typedef struct func_s {
141 unsigned nargs;
142 struct chain_s body;
143} func;
144
145/* I/O stream */
146typedef struct rstream_s {
147 FILE *F;
148 char *buffer;
149 int adv;
150 int size;
151 int pos;
152 smallint is_pipe;
153} rstream;
154
155typedef struct hash_item_s {
156 union {
157 struct var_s v; /* variable/array hash */
158 struct rstream_s rs; /* redirect streams hash */
159 struct func_s f; /* functions hash */
160 } data;
161 struct hash_item_s *next; /* next in chain */
162 char name[1]; /* really it's longer */
163} hash_item;
164
165typedef struct xhash_s {
166 unsigned nel; /* num of elements */
167 unsigned csize; /* current hash size */
168 unsigned nprime; /* next hash size in PRIMES[] */
169 unsigned glen; /* summary length of item names */
170 struct hash_item_s **items;
171} xhash;
172
173/* Tree node */
174typedef struct node_s {
175 uint32_t info;
176 unsigned lineno;
177 union {
178 struct node_s *n;
179 var *v;
180 int aidx;
181 char *new_progname;
182 regex_t *re;
183 } l;
184 union {
185 struct node_s *n;
186 regex_t *ire;
187 func *f;
188 } r;
189 union {
190 struct node_s *n;
191 } a;
192} node;
193
194/* Block of temporary variables */
195typedef struct nvblock_s {
196 int size;
197 var *pos;
198 struct nvblock_s *prev;
199 struct nvblock_s *next;
200 var nv[];
201} nvblock;
202
203typedef struct tsplitter_s {
204 node n;
205 regex_t re[2];
206} tsplitter;
207
208/* simple token classes */
209/* Order and hex values are very important!!! See next_token() */
210#define TC_SEQSTART (1 << 0) /* ( */
211#define TC_SEQTERM (1 << 1) /* ) */
212#define TC_REGEXP (1 << 2) /* /.../ */
213#define TC_OUTRDR (1 << 3) /* | > >> */
214#define TC_UOPPOST (1 << 4) /* unary postfix operator */
215#define TC_UOPPRE1 (1 << 5) /* unary prefix operator */
216#define TC_BINOPX (1 << 6) /* two-opnd operator */
217#define TC_IN (1 << 7)
218#define TC_COMMA (1 << 8)
219#define TC_PIPE (1 << 9) /* input redirection pipe */
220#define TC_UOPPRE2 (1 << 10) /* unary prefix operator */
221#define TC_ARRTERM (1 << 11) /* ] */
222#define TC_GRPSTART (1 << 12) /* { */
223#define TC_GRPTERM (1 << 13) /* } */
224#define TC_SEMICOL (1 << 14)
225#define TC_NEWLINE (1 << 15)
226#define TC_STATX (1 << 16) /* ctl statement (for, next...) */
227#define TC_WHILE (1 << 17)
228#define TC_ELSE (1 << 18)
229#define TC_BUILTIN (1 << 19)
230/* This costs ~50 bytes of code.
231 * A separate class to support deprecated "length" form. If we don't need that
232 * (i.e. if we demand that only "length()" with () is valid), then TC_LENGTH
233 * can be merged with TC_BUILTIN:
234 */
235#define TC_LENGTH (1 << 20)
236#define TC_GETLINE (1 << 21)
237#define TC_FUNCDECL (1 << 22) /* `function' `func' */
238#define TC_BEGIN (1 << 23)
239#define TC_END (1 << 24)
240#define TC_EOF (1 << 25)
241#define TC_VARIABLE (1 << 26)
242#define TC_ARRAY (1 << 27)
243#define TC_FUNCTION (1 << 28)
244#define TC_STRING (1 << 29)
245#define TC_NUMBER (1 << 30)
246
247#define TC_UOPPRE (TC_UOPPRE1 | TC_UOPPRE2)
248
249/* combined token classes */
250#define TC_BINOP (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN)
251//#define TC_UNARYOP (TC_UOPPRE | TC_UOPPOST)
252#define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION \
253 | TC_BUILTIN | TC_LENGTH | TC_GETLINE \
254 | TC_SEQSTART | TC_STRING | TC_NUMBER)
255
256#define TC_STATEMNT (TC_STATX | TC_WHILE)
257#define TC_OPTERM (TC_SEMICOL | TC_NEWLINE)
258
259/* word tokens, cannot mean something else if not expected */
260#define TC_WORD (TC_IN | TC_STATEMNT | TC_ELSE \
261 | TC_BUILTIN | TC_LENGTH | TC_GETLINE \
262 | TC_FUNCDECL | TC_BEGIN | TC_END)
263
264/* discard newlines after these */
265#define TC_NOTERM (TC_COMMA | TC_GRPSTART | TC_GRPTERM \
266 | TC_BINOP | TC_OPTERM)
267
268/* what can expression begin with */
269#define TC_OPSEQ (TC_OPERAND | TC_UOPPRE | TC_REGEXP)
270/* what can group begin with */
271#define TC_GRPSEQ (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART)
272
273/* if previous token class is CONCAT1 and next is CONCAT2, concatenation */
274/* operator is inserted between them */
275#define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM \
276 | TC_STRING | TC_NUMBER | TC_UOPPOST)
277#define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE)
278
279#define OF_RES1 0x010000
280#define OF_RES2 0x020000
281#define OF_STR1 0x040000
282#define OF_STR2 0x080000
283#define OF_NUM1 0x100000
284#define OF_CHECKED 0x200000
285
286/* combined operator flags */
287#define xx 0
288#define xV OF_RES2
289#define xS (OF_RES2 | OF_STR2)
290#define Vx OF_RES1
291#define VV (OF_RES1 | OF_RES2)
292#define Nx (OF_RES1 | OF_NUM1)
293#define NV (OF_RES1 | OF_NUM1 | OF_RES2)
294#define Sx (OF_RES1 | OF_STR1)
295#define SV (OF_RES1 | OF_STR1 | OF_RES2)
296#define SS (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2)
297
298#define OPCLSMASK 0xFF00
299#define OPNMASK 0x007F
300
301/* operator priority is a highest byte (even: r->l, odd: l->r grouping)
302 * For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1,
303 * n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string
304 */
305#undef P
306#undef PRIMASK
307#undef PRIMASK2
308#define P(x) (x << 24)
309#define PRIMASK 0x7F000000
310#define PRIMASK2 0x7E000000
311
312/* Operation classes */
313
314#define SHIFT_TIL_THIS 0x0600
315#define RECUR_FROM_THIS 0x1000
316
317enum {
318 OC_DELETE = 0x0100, OC_EXEC = 0x0200, OC_NEWSOURCE = 0x0300,
319 OC_PRINT = 0x0400, OC_PRINTF = 0x0500, OC_WALKINIT = 0x0600,
320
321 OC_BR = 0x0700, OC_BREAK = 0x0800, OC_CONTINUE = 0x0900,
322 OC_EXIT = 0x0a00, OC_NEXT = 0x0b00, OC_NEXTFILE = 0x0c00,
323 OC_TEST = 0x0d00, OC_WALKNEXT = 0x0e00,
324
325 OC_BINARY = 0x1000, OC_BUILTIN = 0x1100, OC_COLON = 0x1200,
326 OC_COMMA = 0x1300, OC_COMPARE = 0x1400, OC_CONCAT = 0x1500,
327 OC_FBLTIN = 0x1600, OC_FIELD = 0x1700, OC_FNARG = 0x1800,
328 OC_FUNC = 0x1900, OC_GETLINE = 0x1a00, OC_IN = 0x1b00,
329 OC_LAND = 0x1c00, OC_LOR = 0x1d00, OC_MATCH = 0x1e00,
330 OC_MOVE = 0x1f00, OC_PGETLINE = 0x2000, OC_REGEXP = 0x2100,
331 OC_REPLACE = 0x2200, OC_RETURN = 0x2300, OC_SPRINTF = 0x2400,
332 OC_TERNARY = 0x2500, OC_UNARY = 0x2600, OC_VAR = 0x2700,
333 OC_DONE = 0x2800,
334
335 ST_IF = 0x3000, ST_DO = 0x3100, ST_FOR = 0x3200,
336 ST_WHILE = 0x3300
337};
338
339/* simple builtins */
340enum {
341 F_in, F_rn, F_co, F_ex, F_lg, F_si, F_sq, F_sr,
342 F_ti, F_le, F_sy, F_ff, F_cl
343};
344
345/* builtins */
346enum {
347 B_a2, B_ix, B_ma, B_sp, B_ss, B_ti, B_mt, B_lo, B_up,
348 B_ge, B_gs, B_su,
349 B_an, B_co, B_ls, B_or, B_rs, B_xo,
350};
351
352/* tokens and their corresponding info values */
353
354#define NTC "\377" /* switch to next token class (tc<<1) */
355#define NTCC '\377'
356
357static const char tokenlist[] ALIGN1 =
358 "\1(" NTC /* TC_SEQSTART */
359 "\1)" NTC /* TC_SEQTERM */
360 "\1/" NTC /* TC_REGEXP */
361 "\2>>" "\1>" "\1|" NTC /* TC_OUTRDR */
362 "\2++" "\2--" NTC /* TC_UOPPOST */
363 "\2++" "\2--" "\1$" NTC /* TC_UOPPRE1 */
364 "\2==" "\1=" "\2+=" "\2-=" /* TC_BINOPX */
365 "\2*=" "\2/=" "\2%=" "\2^="
366 "\1+" "\1-" "\3**=" "\2**"
367 "\1/" "\1%" "\1^" "\1*"
368 "\2!=" "\2>=" "\2<=" "\1>"
369 "\1<" "\2!~" "\1~" "\2&&"
370 "\2||" "\1?" "\1:" NTC
371 "\2in" NTC /* TC_IN */
372 "\1," NTC /* TC_COMMA */
373 "\1|" NTC /* TC_PIPE */
374 "\1+" "\1-" "\1!" NTC /* TC_UOPPRE2 */
375 "\1]" NTC /* TC_ARRTERM */
376 "\1{" NTC /* TC_GRPSTART */
377 "\1}" NTC /* TC_GRPTERM */
378 "\1;" NTC /* TC_SEMICOL */
379 "\1\n" NTC /* TC_NEWLINE */
380 "\2if" "\2do" "\3for" "\5break" /* TC_STATX */
381 "\10continue" "\6delete" "\5print"
382 "\6printf" "\4next" "\10nextfile"
383 "\6return" "\4exit" NTC
384 "\5while" NTC /* TC_WHILE */
385 "\4else" NTC /* TC_ELSE */
386 "\3and" "\5compl" "\6lshift" "\2or" /* TC_BUILTIN */
387 "\6rshift" "\3xor"
388 "\5close" "\6system" "\6fflush" "\5atan2"
389 "\3cos" "\3exp" "\3int" "\3log"
390 "\4rand" "\3sin" "\4sqrt" "\5srand"
391 "\6gensub" "\4gsub" "\5index" /* "\6length" was here */
392 "\5match" "\5split" "\7sprintf" "\3sub"
393 "\6substr" "\7systime" "\10strftime" "\6mktime"
394 "\7tolower" "\7toupper" NTC
395 "\6length" NTC /* TC_LENGTH */
396 "\7getline" NTC /* TC_GETLINE */
397 "\4func" "\10function" NTC /* TC_FUNCDECL */
398 "\5BEGIN" NTC /* TC_BEGIN */
399 "\3END" /* TC_END */
400 /* compiler adds trailing "\0" */
401 ;
402
403#define OC_B OC_BUILTIN
404
405static const uint32_t tokeninfo[] = {
406 0,
407 0,
408 OC_REGEXP,
409 xS|'a', xS|'w', xS|'|',
410 OC_UNARY|xV|P(9)|'p', OC_UNARY|xV|P(9)|'m',
411 OC_UNARY|xV|P(9)|'P', OC_UNARY|xV|P(9)|'M', OC_FIELD|xV|P(5),
412 OC_COMPARE|VV|P(39)|5, OC_MOVE|VV|P(74), OC_REPLACE|NV|P(74)|'+', OC_REPLACE|NV|P(74)|'-',
413 OC_REPLACE|NV|P(74)|'*', OC_REPLACE|NV|P(74)|'/', OC_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&',
414 OC_BINARY|NV|P(29)|'+', OC_BINARY|NV|P(29)|'-', OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&',
415 OC_BINARY|NV|P(25)|'/', OC_BINARY|NV|P(25)|'%', OC_BINARY|NV|P(15)|'&', OC_BINARY|NV|P(25)|'*',
416 OC_COMPARE|VV|P(39)|4, OC_COMPARE|VV|P(39)|3, OC_COMPARE|VV|P(39)|0, OC_COMPARE|VV|P(39)|1,
417 OC_COMPARE|VV|P(39)|2, OC_MATCH|Sx|P(45)|'!', OC_MATCH|Sx|P(45)|'~', OC_LAND|Vx|P(55),
418 OC_LOR|Vx|P(59), OC_TERNARY|Vx|P(64)|'?', OC_COLON|xx|P(67)|':',
419 OC_IN|SV|P(49), /* TC_IN */
420 OC_COMMA|SS|P(80),
421 OC_PGETLINE|SV|P(37),
422 OC_UNARY|xV|P(19)|'+', OC_UNARY|xV|P(19)|'-', OC_UNARY|xV|P(19)|'!',
423 0, /* ] */
424 0,
425 0,
426 0,
427 0, /* \n */
428 ST_IF, ST_DO, ST_FOR, OC_BREAK,
429 OC_CONTINUE, OC_DELETE|Vx, OC_PRINT,
430 OC_PRINTF, OC_NEXT, OC_NEXTFILE,
431 OC_RETURN|Vx, OC_EXIT|Nx,
432 ST_WHILE,
433 0, /* else */
434 OC_B|B_an|P(0x83), OC_B|B_co|P(0x41), OC_B|B_ls|P(0x83), OC_B|B_or|P(0x83),
435 OC_B|B_rs|P(0x83), OC_B|B_xo|P(0x83),
436 OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83),
437 OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg,
438 OC_FBLTIN|F_rn, OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr,
439 OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), /* OC_FBLTIN|Sx|F_le, was here */
440 OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF, OC_B|B_su|P(0xb6),
441 OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b), OC_B|B_mt|P(0x0b),
442 OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49),
443 OC_FBLTIN|Sx|F_le, /* TC_LENGTH */
444 OC_GETLINE|SV|P(0),
445 0, 0,
446 0,
447 0 /* TC_END */
448};
449
450/* internal variable names and their initial values */
451/* asterisk marks SPECIAL vars; $ is just no-named Field0 */
452enum {
453 CONVFMT, OFMT, FS, OFS,
454 ORS, RS, RT, FILENAME,
455 SUBSEP, F0, ARGIND, ARGC,
456 ARGV, ERRNO, FNR, NR,
457 NF, IGNORECASE, ENVIRON, NUM_INTERNAL_VARS
458};
459
460static const char vNames[] ALIGN1 =
461 "CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0"
462 "ORS\0" "RS\0*" "RT\0" "FILENAME\0"
463 "SUBSEP\0" "$\0*" "ARGIND\0" "ARGC\0"
464 "ARGV\0" "ERRNO\0" "FNR\0" "NR\0"
465 "NF\0*" "IGNORECASE\0*" "ENVIRON\0" "\0";
466
467static const char vValues[] ALIGN1 =
468 "%.6g\0" "%.6g\0" " \0" " \0"
469 "\n\0" "\n\0" "\0" "\0"
470 "\034\0" "\0" "\377";
471
472/* hash size may grow to these values */
473#define FIRST_PRIME 61
474static const uint16_t PRIMES[] ALIGN2 = { 251, 1021, 4093, 16381, 65521 };
475
476
477/* Globals. Split in two parts so that first one is addressed
478 * with (mostly short) negative offsets.
479 * NB: it's unsafe to put members of type "double"
480 * into globals2 (gcc may fail to align them).
481 */
482struct globals {
483 double t_double;
484 chain beginseq, mainseq, endseq;
485 chain *seq;
486 node *break_ptr, *continue_ptr;
487 rstream *iF;
488 xhash *vhash, *ahash, *fdhash, *fnhash;
489 const char *g_progname;
490 int g_lineno;
491 int nfields;
492 int maxfields; /* used in fsrealloc() only */
493 var *Fields;
494 nvblock *g_cb;
495 char *g_pos;
496 char *g_buf;
497 smallint icase;
498 smallint exiting;
499 smallint nextrec;
500 smallint nextfile;
501 smallint is_f0_split;
502 smallint t_rollback;
503};
504struct globals2 {
505 uint32_t t_info; /* often used */
506 uint32_t t_tclass;
507 char *t_string;
508 int t_lineno;
509
510 var *intvar[NUM_INTERNAL_VARS]; /* often used */
511
512 /* former statics from various functions */
513 char *split_f0__fstrings;
514
515 uint32_t next_token__save_tclass;
516 uint32_t next_token__save_info;
517 uint32_t next_token__ltclass;
518 smallint next_token__concat_inserted;
519
520 smallint next_input_file__files_happen;
521 rstream next_input_file__rsm;
522
523 var *evaluate__fnargs;
524 unsigned evaluate__seed;
525 regex_t evaluate__sreg;
526
527 var ptest__v;
528
529 tsplitter exec_builtin__tspl;
530
531 /* biggest and least used members go last */
532 tsplitter fsplitter, rsplitter;
533};
534#define G1 (ptr_to_globals[-1])
535#define G (*(struct globals2 *)ptr_to_globals)
536/* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */
537/*char G1size[sizeof(G1)]; - 0x74 */
538/*char Gsize[sizeof(G)]; - 0x1c4 */
539/* Trying to keep most of members accessible with short offsets: */
540/*char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */
541#define t_double (G1.t_double )
542#define beginseq (G1.beginseq )
543#define mainseq (G1.mainseq )
544#define endseq (G1.endseq )
545#define seq (G1.seq )
546#define break_ptr (G1.break_ptr )
547#define continue_ptr (G1.continue_ptr)
548#define iF (G1.iF )
549#define vhash (G1.vhash )
550#define ahash (G1.ahash )
551#define fdhash (G1.fdhash )
552#define fnhash (G1.fnhash )
553#define g_progname (G1.g_progname )
554#define g_lineno (G1.g_lineno )
555#define nfields (G1.nfields )
556#define maxfields (G1.maxfields )
557#define Fields (G1.Fields )
558#define g_cb (G1.g_cb )
559#define g_pos (G1.g_pos )
560#define g_buf (G1.g_buf )
561#define icase (G1.icase )
562#define exiting (G1.exiting )
563#define nextrec (G1.nextrec )
564#define nextfile (G1.nextfile )
565#define is_f0_split (G1.is_f0_split )
566#define t_rollback (G1.t_rollback )
567#define t_info (G.t_info )
568#define t_tclass (G.t_tclass )
569#define t_string (G.t_string )
570#define t_lineno (G.t_lineno )
571#define intvar (G.intvar )
572#define fsplitter (G.fsplitter )
573#define rsplitter (G.rsplitter )
574#define INIT_G() do { \
575 SET_PTR_TO_GLOBALS((char*)xzalloc(sizeof(G1)+sizeof(G)) + sizeof(G1)); \
576 G.next_token__ltclass = TC_OPTERM; \
577 G.evaluate__seed = 1; \
578} while (0)
579
580
581/* function prototypes */
582static void handle_special(var *);
583static node *parse_expr(uint32_t);
584static void chain_group(void);
585static var *evaluate(node *, var *);
586static rstream *next_input_file(void);
587static int fmt_num(char *, int, const char *, double, int);
588static int awk_exit(int) NORETURN;
589
590/* ---- error handling ---- */
591
592static const char EMSG_INTERNAL_ERROR[] ALIGN1 = "Internal error";
593static const char EMSG_UNEXP_EOS[] ALIGN1 = "Unexpected end of string";
594static const char EMSG_UNEXP_TOKEN[] ALIGN1 = "Unexpected token";
595static const char EMSG_DIV_BY_ZERO[] ALIGN1 = "Division by zero";
596static const char EMSG_INV_FMT[] ALIGN1 = "Invalid format specifier";
597static const char EMSG_TOO_FEW_ARGS[] ALIGN1 = "Too few arguments for builtin";
598static const char EMSG_NOT_ARRAY[] ALIGN1 = "Not an array";
599static const char EMSG_POSSIBLE_ERROR[] ALIGN1 = "Possible syntax error";
600static const char EMSG_UNDEF_FUNC[] ALIGN1 = "Call to undefined function";
601static const char EMSG_NO_MATH[] ALIGN1 = "Math support is not compiled in";
602
603static void zero_out_var(var *vp)
604{
605 memset(vp, 0, sizeof(*vp));
606}
607
608static void syntax_error(const char *message) NORETURN;
609static void syntax_error(const char *message)
610{
611 bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message);
612}
613
614/* ---- hash stuff ---- */
615
616static unsigned hashidx(const char *name)
617{
618 unsigned idx = 0;
619
620 while (*name)
621 idx = *name++ + (idx << 6) - idx;
622 return idx;
623}
624
625/* create new hash */
626static xhash *hash_init(void)
627{
628 xhash *newhash;
629
630 newhash = xzalloc(sizeof(*newhash));
631 newhash->csize = FIRST_PRIME;
632 newhash->items = xzalloc(FIRST_PRIME * sizeof(newhash->items[0]));
633
634 return newhash;
635}
636
637/* find item in hash, return ptr to data, NULL if not found */
638static void *hash_search(xhash *hash, const char *name)
639{
640 hash_item *hi;
641
642 hi = hash->items[hashidx(name) % hash->csize];
643 while (hi) {
644 if (strcmp(hi->name, name) == 0)
645 return &hi->data;
646 hi = hi->next;
647 }
648 return NULL;
649}
650
651/* grow hash if it becomes too big */
652static void hash_rebuild(xhash *hash)
653{
654 unsigned newsize, i, idx;
655 hash_item **newitems, *hi, *thi;
656
657 if (hash->nprime == ARRAY_SIZE(PRIMES))
658 return;
659
660 newsize = PRIMES[hash->nprime++];
661 newitems = xzalloc(newsize * sizeof(newitems[0]));
662
663 for (i = 0; i < hash->csize; i++) {
664 hi = hash->items[i];
665 while (hi) {
666 thi = hi;
667 hi = thi->next;
668 idx = hashidx(thi->name) % newsize;
669 thi->next = newitems[idx];
670 newitems[idx] = thi;
671 }
672 }
673
674 free(hash->items);
675 hash->csize = newsize;
676 hash->items = newitems;
677}
678
679/* find item in hash, add it if necessary. Return ptr to data */
680static void *hash_find(xhash *hash, const char *name)
681{
682 hash_item *hi;
683 unsigned idx;
684 int l;
685
686 hi = hash_search(hash, name);
687 if (!hi) {
688 if (++hash->nel / hash->csize > 10)
689 hash_rebuild(hash);
690
691 l = strlen(name) + 1;
692 hi = xzalloc(sizeof(*hi) + l);
693 strcpy(hi->name, name);
694
695 idx = hashidx(name) % hash->csize;
696 hi->next = hash->items[idx];
697 hash->items[idx] = hi;
698 hash->glen += l;
699 }
700 return &hi->data;
701}
702
703#define findvar(hash, name) ((var*) hash_find((hash), (name)))
704#define newvar(name) ((var*) hash_find(vhash, (name)))
705#define newfile(name) ((rstream*)hash_find(fdhash, (name)))
706#define newfunc(name) ((func*) hash_find(fnhash, (name)))
707
708static void hash_remove(xhash *hash, const char *name)
709{
710 hash_item *hi, **phi;
711
712 phi = &hash->items[hashidx(name) % hash->csize];
713 while (*phi) {
714 hi = *phi;
715 if (strcmp(hi->name, name) == 0) {
716 hash->glen -= (strlen(name) + 1);
717 hash->nel--;
718 *phi = hi->next;
719 free(hi);
720 break;
721 }
722 phi = &hi->next;
723 }
724}
725
726/* ------ some useful functions ------ */
727
728static char *skip_spaces(char *p)
729{
730 while (1) {
731 if (*p == '\\' && p[1] == '\n') {
732 p++;
733 t_lineno++;
734 } else if (*p != ' ' && *p != '\t') {
735 break;
736 }
737 p++;
738 }
739 return p;
740}
741
742/* returns old *s, advances *s past word and terminating NUL */
743static char *nextword(char **s)
744{
745 char *p = *s;
746 while (*(*s)++ != '\0')
747 continue;
748 return p;
749}
750
751static char nextchar(char **s)
752{
753 char c, *pps;
754
755 c = *(*s)++;
756 pps = *s;
757 if (c == '\\')
758 c = bb_process_escape_sequence((const char**)s);
759 /* Example awk statement:
760 * s = "abc\"def"
761 * we must treat \" as "
762 */
763 if (c == '\\' && *s == pps) { /* unrecognized \z? */
764 c = *(*s); /* yes, fetch z */
765 if (c)
766 (*s)++; /* advance unless z = NUL */
767 }
768 return c;
769}
770
771/* TODO: merge with strcpy_and_process_escape_sequences()?
772 */
773static void unescape_string_in_place(char *s1)
774{
775 char *s = s1;
776 while ((*s1 = nextchar(&s)) != '\0')
777 s1++;
778}
779
780static ALWAYS_INLINE int isalnum_(int c)
781{
782 return (isalnum(c) || c == '_');
783}
784
785static double my_strtod(char **pp)
786{
787 char *cp = *pp;
788 if (ENABLE_DESKTOP && cp[0] == '0') {
789 /* Might be hex or octal integer: 0x123abc or 07777 */
790 char c = (cp[1] | 0x20);
791 if (c == 'x' || isdigit(cp[1])) {
792 unsigned long long ull = strtoull(cp, pp, 0);
793 if (c == 'x')
794 return ull;
795 c = **pp;
796 if (!isdigit(c) && c != '.')
797 return ull;
798 /* else: it may be a floating number. Examples:
799 * 009.123 (*pp points to '9')
800 * 000.123 (*pp points to '.')
801 * fall through to strtod.
802 */
803 }
804 }
805 return strtod(cp, pp);
806}
807
808/* -------- working with variables (set/get/copy/etc) -------- */
809
810static xhash *iamarray(var *v)
811{
812 var *a = v;
813
814 while (a->type & VF_CHILD)
815 a = a->x.parent;
816
817 if (!(a->type & VF_ARRAY)) {
818 a->type |= VF_ARRAY;
819 a->x.array = hash_init();
820 }
821 return a->x.array;
822}
823
824static void clear_array(xhash *array)
825{
826 unsigned i;
827 hash_item *hi, *thi;
828
829 for (i = 0; i < array->csize; i++) {
830 hi = array->items[i];
831 while (hi) {
832 thi = hi;
833 hi = hi->next;
834 free(thi->data.v.string);
835 free(thi);
836 }
837 array->items[i] = NULL;
838 }
839 array->glen = array->nel = 0;
840}
841
842/* clear a variable */
843static var *clrvar(var *v)
844{
845 if (!(v->type & VF_FSTR))
846 free(v->string);
847
848 v->type &= VF_DONTTOUCH;
849 v->type |= VF_DIRTY;
850 v->string = NULL;
851 return v;
852}
853
854/* assign string value to variable */
855static var *setvar_p(var *v, char *value)
856{
857 clrvar(v);
858 v->string = value;
859 handle_special(v);
860 return v;
861}
862
863/* same as setvar_p but make a copy of string */
864static var *setvar_s(var *v, const char *value)
865{
866 return setvar_p(v, (value && *value) ? xstrdup(value) : NULL);
867}
868
869/* same as setvar_s but sets USER flag */
870static var *setvar_u(var *v, const char *value)
871{
872 v = setvar_s(v, value);
873 v->type |= VF_USER;
874 return v;
875}
876
877/* set array element to user string */
878static void setari_u(var *a, int idx, const char *s)
879{
880 var *v;
881
882 v = findvar(iamarray(a), itoa(idx));
883 setvar_u(v, s);
884}
885
886/* assign numeric value to variable */
887static var *setvar_i(var *v, double value)
888{
889 clrvar(v);
890 v->type |= VF_NUMBER;
891 v->number = value;
892 handle_special(v);
893 return v;
894}
895
896static const char *getvar_s(var *v)
897{
898 /* if v is numeric and has no cached string, convert it to string */
899 if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) {
900 fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE);
901 v->string = xstrdup(g_buf);
902 v->type |= VF_CACHED;
903 }
904 return (v->string == NULL) ? "" : v->string;
905}
906
907static double getvar_i(var *v)
908{
909 char *s;
910
911 if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) {
912 v->number = 0;
913 s = v->string;
914 if (s && *s) {
915 debug_printf_eval("getvar_i: '%s'->", s);
916 v->number = my_strtod(&s);
917 debug_printf_eval("%f (s:'%s')\n", v->number, s);
918 if (v->type & VF_USER) {
919 s = skip_spaces(s);
920 if (*s != '\0')
921 v->type &= ~VF_USER;
922 }
923 } else {
924 debug_printf_eval("getvar_i: '%s'->zero\n", s);
925 v->type &= ~VF_USER;
926 }
927 v->type |= VF_CACHED;
928 }
929 debug_printf_eval("getvar_i: %f\n", v->number);
930 return v->number;
931}
932
933/* Used for operands of bitwise ops */
934static unsigned long getvar_i_int(var *v)
935{
936 double d = getvar_i(v);
937
938 /* Casting doubles to longs is undefined for values outside
939 * of target type range. Try to widen it as much as possible */
940 if (d >= 0)
941 return (unsigned long)d;
942 /* Why? Think about d == -4294967295.0 (assuming 32bit longs) */
943 return - (long) (unsigned long) (-d);
944}
945
946static var *copyvar(var *dest, const var *src)
947{
948 if (dest != src) {
949 clrvar(dest);
950 dest->type |= (src->type & ~(VF_DONTTOUCH | VF_FSTR));
951 debug_printf_eval("copyvar: number:%f string:'%s'\n", src->number, src->string);
952 dest->number = src->number;
953 if (src->string)
954 dest->string = xstrdup(src->string);
955 }
956 handle_special(dest);
957 return dest;
958}
959
960static var *incvar(var *v)
961{
962 return setvar_i(v, getvar_i(v) + 1.0);
963}
964
965/* return true if v is number or numeric string */
966static int is_numeric(var *v)
967{
968 getvar_i(v);
969 return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY));
970}
971
972/* return 1 when value of v corresponds to true, 0 otherwise */
973static int istrue(var *v)
974{
975 if (is_numeric(v))
976 return (v->number != 0);
977 return (v->string && v->string[0]);
978}
979
980/* temporary variables allocator. Last allocated should be first freed */
981static var *nvalloc(int n)
982{
983 nvblock *pb = NULL;
984 var *v, *r;
985 int size;
986
987 while (g_cb) {
988 pb = g_cb;
989 if ((g_cb->pos - g_cb->nv) + n <= g_cb->size)
990 break;
991 g_cb = g_cb->next;
992 }
993
994 if (!g_cb) {
995 size = (n <= MINNVBLOCK) ? MINNVBLOCK : n;
996 g_cb = xzalloc(sizeof(nvblock) + size * sizeof(var));
997 g_cb->size = size;
998 g_cb->pos = g_cb->nv;
999 g_cb->prev = pb;
1000 /*g_cb->next = NULL; - xzalloc did it */
1001 if (pb)
1002 pb->next = g_cb;
1003 }
1004
1005 v = r = g_cb->pos;
1006 g_cb->pos += n;
1007
1008 while (v < g_cb->pos) {
1009 v->type = 0;
1010 v->string = NULL;
1011 v++;
1012 }
1013
1014 return r;
1015}
1016
1017static void nvfree(var *v)
1018{
1019 var *p;
1020
1021 if (v < g_cb->nv || v >= g_cb->pos)
1022 syntax_error(EMSG_INTERNAL_ERROR);
1023
1024 for (p = v; p < g_cb->pos; p++) {
1025 if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) {
1026 clear_array(iamarray(p));
1027 free(p->x.array->items);
1028 free(p->x.array);
1029 }
1030 if (p->type & VF_WALK) {
1031 walker_list *n;
1032 walker_list *w = p->x.walker;
1033 debug_printf_walker("nvfree: freeing walker @%p\n", &p->x.walker);
1034 p->x.walker = NULL;
1035 while (w) {
1036 n = w->prev;
1037 debug_printf_walker(" free(%p)\n", w);
1038 free(w);
1039 w = n;
1040 }
1041 }
1042 clrvar(p);
1043 }
1044
1045 g_cb->pos = v;
1046 while (g_cb->prev && g_cb->pos == g_cb->nv) {
1047 g_cb = g_cb->prev;
1048 }
1049}
1050
1051/* ------- awk program text parsing ------- */
1052
1053/* Parse next token pointed by global pos, place results into global ttt.
1054 * If token isn't expected, give away. Return token class
1055 */
1056static uint32_t next_token(uint32_t expected)
1057{
1058#define concat_inserted (G.next_token__concat_inserted)
1059#define save_tclass (G.next_token__save_tclass)
1060#define save_info (G.next_token__save_info)
1061/* Initialized to TC_OPTERM: */
1062#define ltclass (G.next_token__ltclass)
1063
1064 char *p, *s;
1065 const char *tl;
1066 uint32_t tc;
1067 const uint32_t *ti;
1068
1069 if (t_rollback) {
1070 t_rollback = FALSE;
1071 } else if (concat_inserted) {
1072 concat_inserted = FALSE;
1073 t_tclass = save_tclass;
1074 t_info = save_info;
1075 } else {
1076 p = g_pos;
1077 readnext:
1078 p = skip_spaces(p);
1079 g_lineno = t_lineno;
1080 if (*p == '#')
1081 while (*p != '\n' && *p != '\0')
1082 p++;
1083
1084 if (*p == '\n')
1085 t_lineno++;
1086
1087 if (*p == '\0') {
1088 tc = TC_EOF;
1089 debug_printf_parse("%s: token found: TC_EOF\n", __func__);
1090 } else if (*p == '\"') {
1091 /* it's a string */
1092 t_string = s = ++p;
1093 while (*p != '\"') {
1094 char *pp;
1095 if (*p == '\0' || *p == '\n')
1096 syntax_error(EMSG_UNEXP_EOS);
1097 pp = p;
1098 *s++ = nextchar(&pp);
1099 p = pp;
1100 }
1101 p++;
1102 *s = '\0';
1103 tc = TC_STRING;
1104 debug_printf_parse("%s: token found:'%s' TC_STRING\n", __func__, t_string);
1105 } else if ((expected & TC_REGEXP) && *p == '/') {
1106 /* it's regexp */
1107 t_string = s = ++p;
1108 while (*p != '/') {
1109 if (*p == '\0' || *p == '\n')
1110 syntax_error(EMSG_UNEXP_EOS);
1111 *s = *p++;
1112 if (*s++ == '\\') {
1113 char *pp = p;
1114 s[-1] = bb_process_escape_sequence((const char **)&pp);
1115 if (*p == '\\')
1116 *s++ = '\\';
1117 if (pp == p)
1118 *s++ = *p++;
1119 else
1120 p = pp;
1121 }
1122 }
1123 p++;
1124 *s = '\0';
1125 tc = TC_REGEXP;
1126 debug_printf_parse("%s: token found:'%s' TC_REGEXP\n", __func__, t_string);
1127
1128 } else if (*p == '.' || isdigit(*p)) {
1129 /* it's a number */
1130 char *pp = p;
1131 t_double = my_strtod(&pp);
1132 p = pp;
1133 if (*p == '.')
1134 syntax_error(EMSG_UNEXP_TOKEN);
1135 tc = TC_NUMBER;
1136 debug_printf_parse("%s: token found:%f TC_NUMBER\n", __func__, t_double);
1137 } else {
1138 /* search for something known */
1139 tl = tokenlist;
1140 tc = 0x00000001;
1141 ti = tokeninfo;
1142 while (*tl) {
1143 int l = (unsigned char) *tl++;
1144 if (l == (unsigned char) NTCC) {
1145 tc <<= 1;
1146 continue;
1147 }
1148 /* if token class is expected,
1149 * token matches,
1150 * and it's not a longer word,
1151 */
1152 if ((tc & (expected | TC_WORD | TC_NEWLINE))
1153 && strncmp(p, tl, l) == 0
1154 && !((tc & TC_WORD) && isalnum_(p[l]))
1155 ) {
1156 /* then this is what we are looking for */
1157 t_info = *ti;
1158 debug_printf_parse("%s: token found:'%.*s' t_info:%x\n", __func__, l, p, t_info);
1159 p += l;
1160 goto token_found;
1161 }
1162 ti++;
1163 tl += l;
1164 }
1165 /* not a known token */
1166
1167 /* is it a name? (var/array/function) */
1168 if (!isalnum_(*p))
1169 syntax_error(EMSG_UNEXP_TOKEN); /* no */
1170 /* yes */
1171 t_string = --p;
1172 while (isalnum_(*++p)) {
1173 p[-1] = *p;
1174 }
1175 p[-1] = '\0';
1176 tc = TC_VARIABLE;
1177 /* also consume whitespace between functionname and bracket */
1178 if (!(expected & TC_VARIABLE) || (expected & TC_ARRAY))
1179 p = skip_spaces(p);
1180 if (*p == '(') {
1181 tc = TC_FUNCTION;
1182 debug_printf_parse("%s: token found:'%s' TC_FUNCTION\n", __func__, t_string);
1183 } else {
1184 if (*p == '[') {
1185 p++;
1186 tc = TC_ARRAY;
1187 debug_printf_parse("%s: token found:'%s' TC_ARRAY\n", __func__, t_string);
1188 } else
1189 debug_printf_parse("%s: token found:'%s' TC_VARIABLE\n", __func__, t_string);
1190 }
1191 }
1192 token_found:
1193 g_pos = p;
1194
1195 /* skipping newlines in some cases */
1196 if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE))
1197 goto readnext;
1198
1199 /* insert concatenation operator when needed */
1200 if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) {
1201 concat_inserted = TRUE;
1202 save_tclass = tc;
1203 save_info = t_info;
1204 tc = TC_BINOP;
1205 t_info = OC_CONCAT | SS | P(35);
1206 }
1207
1208 t_tclass = tc;
1209 }
1210 ltclass = t_tclass;
1211
1212 /* Are we ready for this? */
1213 if (!(ltclass & expected)) {
1214 syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ?
1215 EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN);
1216 }
1217
1218 return ltclass;
1219#undef concat_inserted
1220#undef save_tclass
1221#undef save_info
1222#undef ltclass
1223}
1224
1225static void rollback_token(void)
1226{
1227 t_rollback = TRUE;
1228}
1229
1230static node *new_node(uint32_t info)
1231{
1232 node *n;
1233
1234 n = xzalloc(sizeof(node));
1235 n->info = info;
1236 n->lineno = g_lineno;
1237 return n;
1238}
1239
1240static void mk_re_node(const char *s, node *n, regex_t *re)
1241{
1242 n->info = OC_REGEXP;
1243 n->l.re = re;
1244 n->r.ire = re + 1;
1245 xregcomp(re, s, REG_EXTENDED);
1246 xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE);
1247}
1248
1249static node *condition(void)
1250{
1251 next_token(TC_SEQSTART);
1252 return parse_expr(TC_SEQTERM);
1253}
1254
1255/* parse expression terminated by given argument, return ptr
1256 * to built subtree. Terminator is eaten by parse_expr */
1257static node *parse_expr(uint32_t iexp)
1258{
1259 node sn;
1260 node *cn = &sn;
1261 node *vn, *glptr;
1262 uint32_t tc, xtc;
1263 var *v;
1264
1265 debug_printf_parse("%s(%x)\n", __func__, iexp);
1266
1267 sn.info = PRIMASK;
1268 sn.r.n = glptr = NULL;
1269 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp;
1270
1271 while (!((tc = next_token(xtc)) & iexp)) {
1272
1273 if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) {
1274 /* input redirection (<) attached to glptr node */
1275 debug_printf_parse("%s: input redir\n", __func__);
1276 cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37));
1277 cn->a.n = glptr;
1278 xtc = TC_OPERAND | TC_UOPPRE;
1279 glptr = NULL;
1280
1281 } else if (tc & (TC_BINOP | TC_UOPPOST)) {
1282 debug_printf_parse("%s: TC_BINOP | TC_UOPPOST\n", __func__);
1283 /* for binary and postfix-unary operators, jump back over
1284 * previous operators with higher priority */
1285 vn = cn;
1286 while (((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2))
1287 || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON))
1288 ) {
1289 vn = vn->a.n;
1290 }
1291 if ((t_info & OPCLSMASK) == OC_TERNARY)
1292 t_info += P(6);
1293 cn = vn->a.n->r.n = new_node(t_info);
1294 cn->a.n = vn->a.n;
1295 if (tc & TC_BINOP) {
1296 cn->l.n = vn;
1297 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
1298 if ((t_info & OPCLSMASK) == OC_PGETLINE) {
1299 /* it's a pipe */
1300 next_token(TC_GETLINE);
1301 /* give maximum priority to this pipe */
1302 cn->info &= ~PRIMASK;
1303 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1304 }
1305 } else {
1306 cn->r.n = vn;
1307 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1308 }
1309 vn->a.n = cn;
1310
1311 } else {
1312 debug_printf_parse("%s: other\n", __func__);
1313 /* for operands and prefix-unary operators, attach them
1314 * to last node */
1315 vn = cn;
1316 cn = vn->r.n = new_node(t_info);
1317 cn->a.n = vn;
1318 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
1319 if (tc & (TC_OPERAND | TC_REGEXP)) {
1320 debug_printf_parse("%s: TC_OPERAND | TC_REGEXP\n", __func__);
1321 xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp;
1322 /* one should be very careful with switch on tclass -
1323 * only simple tclasses should be used! */
1324 switch (tc) {
1325 case TC_VARIABLE:
1326 case TC_ARRAY:
1327 debug_printf_parse("%s: TC_VARIABLE | TC_ARRAY\n", __func__);
1328 cn->info = OC_VAR;
1329 v = hash_search(ahash, t_string);
1330 if (v != NULL) {
1331 cn->info = OC_FNARG;
1332 cn->l.aidx = v->x.aidx;
1333 } else {
1334 cn->l.v = newvar(t_string);
1335 }
1336 if (tc & TC_ARRAY) {
1337 cn->info |= xS;
1338 cn->r.n = parse_expr(TC_ARRTERM);
1339 }
1340 break;
1341
1342 case TC_NUMBER:
1343 case TC_STRING:
1344 debug_printf_parse("%s: TC_NUMBER | TC_STRING\n", __func__);
1345 cn->info = OC_VAR;
1346 v = cn->l.v = xzalloc(sizeof(var));
1347 if (tc & TC_NUMBER)
1348 setvar_i(v, t_double);
1349 else
1350 setvar_s(v, t_string);
1351 break;
1352
1353 case TC_REGEXP:
1354 debug_printf_parse("%s: TC_REGEXP\n", __func__);
1355 mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2));
1356 break;
1357
1358 case TC_FUNCTION:
1359 debug_printf_parse("%s: TC_FUNCTION\n", __func__);
1360 cn->info = OC_FUNC;
1361 cn->r.f = newfunc(t_string);
1362 cn->l.n = condition();
1363 break;
1364
1365 case TC_SEQSTART:
1366 debug_printf_parse("%s: TC_SEQSTART\n", __func__);
1367 cn = vn->r.n = parse_expr(TC_SEQTERM);
1368 if (!cn)
1369 syntax_error("Empty sequence");
1370 cn->a.n = vn;
1371 break;
1372
1373 case TC_GETLINE:
1374 debug_printf_parse("%s: TC_GETLINE\n", __func__);
1375 glptr = cn;
1376 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1377 break;
1378
1379 case TC_BUILTIN:
1380 debug_printf_parse("%s: TC_BUILTIN\n", __func__);
1381 cn->l.n = condition();
1382 break;
1383
1384 case TC_LENGTH:
1385 debug_printf_parse("%s: TC_LENGTH\n", __func__);
1386 next_token(TC_SEQSTART | TC_OPTERM | TC_GRPTERM);
1387 rollback_token();
1388 if (t_tclass & TC_SEQSTART) {
1389 /* It was a "(" token. Handle just like TC_BUILTIN */
1390 cn->l.n = condition();
1391 }
1392 break;
1393 }
1394 }
1395 }
1396 }
1397
1398 debug_printf_parse("%s() returns %p\n", __func__, sn.r.n);
1399 return sn.r.n;
1400}
1401
1402/* add node to chain. Return ptr to alloc'd node */
1403static node *chain_node(uint32_t info)
1404{
1405 node *n;
1406
1407 if (!seq->first)
1408 seq->first = seq->last = new_node(0);
1409
1410 if (seq->programname != g_progname) {
1411 seq->programname = g_progname;
1412 n = chain_node(OC_NEWSOURCE);
1413 n->l.new_progname = xstrdup(g_progname);
1414 }
1415
1416 n = seq->last;
1417 n->info = info;
1418 seq->last = n->a.n = new_node(OC_DONE);
1419
1420 return n;
1421}
1422
1423static void chain_expr(uint32_t info)
1424{
1425 node *n;
1426
1427 n = chain_node(info);
1428 n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM);
1429 if (t_tclass & TC_GRPTERM)
1430 rollback_token();
1431}
1432
1433static node *chain_loop(node *nn)
1434{
1435 node *n, *n2, *save_brk, *save_cont;
1436
1437 save_brk = break_ptr;
1438 save_cont = continue_ptr;
1439
1440 n = chain_node(OC_BR | Vx);
1441 continue_ptr = new_node(OC_EXEC);
1442 break_ptr = new_node(OC_EXEC);
1443 chain_group();
1444 n2 = chain_node(OC_EXEC | Vx);
1445 n2->l.n = nn;
1446 n2->a.n = n;
1447 continue_ptr->a.n = n2;
1448 break_ptr->a.n = n->r.n = seq->last;
1449
1450 continue_ptr = save_cont;
1451 break_ptr = save_brk;
1452
1453 return n;
1454}
1455
1456/* parse group and attach it to chain */
1457static void chain_group(void)
1458{
1459 uint32_t c;
1460 node *n, *n2, *n3;
1461
1462 do {
1463 c = next_token(TC_GRPSEQ);
1464 } while (c & TC_NEWLINE);
1465
1466 if (c & TC_GRPSTART) {
1467 debug_printf_parse("%s: TC_GRPSTART\n", __func__);
1468 while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) {
1469 debug_printf_parse("%s: !TC_GRPTERM\n", __func__);
1470 if (t_tclass & TC_NEWLINE)
1471 continue;
1472 rollback_token();
1473 chain_group();
1474 }
1475 debug_printf_parse("%s: TC_GRPTERM\n", __func__);
1476 } else if (c & (TC_OPSEQ | TC_OPTERM)) {
1477 debug_printf_parse("%s: TC_OPSEQ | TC_OPTERM\n", __func__);
1478 rollback_token();
1479 chain_expr(OC_EXEC | Vx);
1480 } else {
1481 /* TC_STATEMNT */
1482 debug_printf_parse("%s: TC_STATEMNT(?)\n", __func__);
1483 switch (t_info & OPCLSMASK) {
1484 case ST_IF:
1485 debug_printf_parse("%s: ST_IF\n", __func__);
1486 n = chain_node(OC_BR | Vx);
1487 n->l.n = condition();
1488 chain_group();
1489 n2 = chain_node(OC_EXEC);
1490 n->r.n = seq->last;
1491 if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE) == TC_ELSE) {
1492 chain_group();
1493 n2->a.n = seq->last;
1494 } else {
1495 rollback_token();
1496 }
1497 break;
1498
1499 case ST_WHILE:
1500 debug_printf_parse("%s: ST_WHILE\n", __func__);
1501 n2 = condition();
1502 n = chain_loop(NULL);
1503 n->l.n = n2;
1504 break;
1505
1506 case ST_DO:
1507 debug_printf_parse("%s: ST_DO\n", __func__);
1508 n2 = chain_node(OC_EXEC);
1509 n = chain_loop(NULL);
1510 n2->a.n = n->a.n;
1511 next_token(TC_WHILE);
1512 n->l.n = condition();
1513 break;
1514
1515 case ST_FOR:
1516 debug_printf_parse("%s: ST_FOR\n", __func__);
1517 next_token(TC_SEQSTART);
1518 n2 = parse_expr(TC_SEMICOL | TC_SEQTERM);
1519 if (t_tclass & TC_SEQTERM) { /* for-in */
1520 if ((n2->info & OPCLSMASK) != OC_IN)
1521 syntax_error(EMSG_UNEXP_TOKEN);
1522 n = chain_node(OC_WALKINIT | VV);
1523 n->l.n = n2->l.n;
1524 n->r.n = n2->r.n;
1525 n = chain_loop(NULL);
1526 n->info = OC_WALKNEXT | Vx;
1527 n->l.n = n2->l.n;
1528 } else { /* for (;;) */
1529 n = chain_node(OC_EXEC | Vx);
1530 n->l.n = n2;
1531 n2 = parse_expr(TC_SEMICOL);
1532 n3 = parse_expr(TC_SEQTERM);
1533 n = chain_loop(n3);
1534 n->l.n = n2;
1535 if (!n2)
1536 n->info = OC_EXEC;
1537 }
1538 break;
1539
1540 case OC_PRINT:
1541 case OC_PRINTF:
1542 debug_printf_parse("%s: OC_PRINT[F]\n", __func__);
1543 n = chain_node(t_info);
1544 n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM);
1545 if (t_tclass & TC_OUTRDR) {
1546 n->info |= t_info;
1547 n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM);
1548 }
1549 if (t_tclass & TC_GRPTERM)
1550 rollback_token();
1551 break;
1552
1553 case OC_BREAK:
1554 debug_printf_parse("%s: OC_BREAK\n", __func__);
1555 n = chain_node(OC_EXEC);
1556 n->a.n = break_ptr;
1557 chain_expr(t_info);
1558 break;
1559
1560 case OC_CONTINUE:
1561 debug_printf_parse("%s: OC_CONTINUE\n", __func__);
1562 n = chain_node(OC_EXEC);
1563 n->a.n = continue_ptr;
1564 chain_expr(t_info);
1565 break;
1566
1567 /* delete, next, nextfile, return, exit */
1568 default:
1569 debug_printf_parse("%s: default\n", __func__);
1570 chain_expr(t_info);
1571 }
1572 }
1573}
1574
1575static void parse_program(char *p)
1576{
1577 uint32_t tclass;
1578 node *cn;
1579 func *f;
1580 var *v;
1581
1582 g_pos = p;
1583 t_lineno = 1;
1584 while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART |
1585 TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) {
1586
1587 if (tclass & TC_OPTERM) {
1588 debug_printf_parse("%s: TC_OPTERM\n", __func__);
1589 continue;
1590 }
1591
1592 seq = &mainseq;
1593 if (tclass & TC_BEGIN) {
1594 debug_printf_parse("%s: TC_BEGIN\n", __func__);
1595 seq = &beginseq;
1596 chain_group();
1597 } else if (tclass & TC_END) {
1598 debug_printf_parse("%s: TC_END\n", __func__);
1599 seq = &endseq;
1600 chain_group();
1601 } else if (tclass & TC_FUNCDECL) {
1602 debug_printf_parse("%s: TC_FUNCDECL\n", __func__);
1603 next_token(TC_FUNCTION);
1604 g_pos++;
1605 f = newfunc(t_string);
1606 f->body.first = NULL;
1607 f->nargs = 0;
1608 while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) {
1609 v = findvar(ahash, t_string);
1610 v->x.aidx = f->nargs++;
1611
1612 if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM)
1613 break;
1614 }
1615 seq = &f->body;
1616 chain_group();
1617 clear_array(ahash);
1618 } else if (tclass & TC_OPSEQ) {
1619 debug_printf_parse("%s: TC_OPSEQ\n", __func__);
1620 rollback_token();
1621 cn = chain_node(OC_TEST);
1622 cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART);
1623 if (t_tclass & TC_GRPSTART) {
1624 debug_printf_parse("%s: TC_GRPSTART\n", __func__);
1625 rollback_token();
1626 chain_group();
1627 } else {
1628 debug_printf_parse("%s: !TC_GRPSTART\n", __func__);
1629 chain_node(OC_PRINT);
1630 }
1631 cn->r.n = mainseq.last;
1632 } else /* if (tclass & TC_GRPSTART) */ {
1633 debug_printf_parse("%s: TC_GRPSTART(?)\n", __func__);
1634 rollback_token();
1635 chain_group();
1636 }
1637 }
1638 debug_printf_parse("%s: TC_EOF\n", __func__);
1639}
1640
1641
1642/* -------- program execution part -------- */
1643
1644static node *mk_splitter(const char *s, tsplitter *spl)
1645{
1646 regex_t *re, *ire;
1647 node *n;
1648
1649 re = &spl->re[0];
1650 ire = &spl->re[1];
1651 n = &spl->n;
1652 if ((n->info & OPCLSMASK) == OC_REGEXP) {
1653 regfree(re);
1654 regfree(ire); // TODO: nuke ire, use re+1?
1655 }
1656 if (s[0] && s[1]) { /* strlen(s) > 1 */
1657 mk_re_node(s, n, re);
1658 } else {
1659 n->info = (uint32_t) s[0];
1660 }
1661
1662 return n;
1663}
1664
1665/* use node as a regular expression. Supplied with node ptr and regex_t
1666 * storage space. Return ptr to regex (if result points to preg, it should
1667 * be later regfree'd manually
1668 */
1669static regex_t *as_regex(node *op, regex_t *preg)
1670{
1671 int cflags;
1672 var *v;
1673 const char *s;
1674
1675 if ((op->info & OPCLSMASK) == OC_REGEXP) {
1676 return icase ? op->r.ire : op->l.re;
1677 }
1678 v = nvalloc(1);
1679 s = getvar_s(evaluate(op, v));
1680
1681 cflags = icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED;
1682 /* Testcase where REG_EXTENDED fails (unpaired '{'):
1683 * echo Hi | awk 'gsub("@(samp|code|file)\{","");'
1684 * gawk 3.1.5 eats this. We revert to ~REG_EXTENDED
1685 * (maybe gsub is not supposed to use REG_EXTENDED?).
1686 */
1687 if (regcomp(preg, s, cflags)) {
1688 cflags &= ~REG_EXTENDED;
1689 xregcomp(preg, s, cflags);
1690 }
1691 nvfree(v);
1692 return preg;
1693}
1694
1695/* gradually increasing buffer.
1696 * note that we reallocate even if n == old_size,
1697 * and thus there is at least one extra allocated byte.
1698 */
1699static char* qrealloc(char *b, int n, int *size)
1700{
1701 if (!b || n >= *size) {
1702 *size = n + (n>>1) + 80;
1703 b = xrealloc(b, *size);
1704 }
1705 return b;
1706}
1707
1708/* resize field storage space */
1709static void fsrealloc(int size)
1710{
1711 int i;
1712
1713 if (size >= maxfields) {
1714 i = maxfields;
1715 maxfields = size + 16;
1716 Fields = xrealloc(Fields, maxfields * sizeof(Fields[0]));
1717 for (; i < maxfields; i++) {
1718 Fields[i].type = VF_SPECIAL;
1719 Fields[i].string = NULL;
1720 }
1721 }
1722 /* if size < nfields, clear extra field variables */
1723 for (i = size; i < nfields; i++) {
1724 clrvar(Fields + i);
1725 }
1726 nfields = size;
1727}
1728
1729static int awk_split(const char *s, node *spl, char **slist)
1730{
1731 int l, n;
1732 char c[4];
1733 char *s1;
1734 regmatch_t pmatch[2]; // TODO: why [2]? [1] is enough...
1735
1736 /* in worst case, each char would be a separate field */
1737 *slist = s1 = xzalloc(strlen(s) * 2 + 3);
1738 strcpy(s1, s);
1739
1740 c[0] = c[1] = (char)spl->info;
1741 c[2] = c[3] = '\0';
1742 if (*getvar_s(intvar[RS]) == '\0')
1743 c[2] = '\n';
1744
1745 n = 0;
1746 if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */
1747 if (!*s)
1748 return n; /* "": zero fields */
1749 n++; /* at least one field will be there */
1750 do {
1751 l = strcspn(s, c+2); /* len till next NUL or \n */
1752 if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0
1753 && pmatch[0].rm_so <= l
1754 ) {
1755 l = pmatch[0].rm_so;
1756 if (pmatch[0].rm_eo == 0) {
1757 l++;
1758 pmatch[0].rm_eo++;
1759 }
1760 n++; /* we saw yet another delimiter */
1761 } else {
1762 pmatch[0].rm_eo = l;
1763 if (s[l])
1764 pmatch[0].rm_eo++;
1765 }
1766 memcpy(s1, s, l);
1767 /* make sure we remove *all* of the separator chars */
1768 do {
1769 s1[l] = '\0';
1770 } while (++l < pmatch[0].rm_eo);
1771 nextword(&s1);
1772 s += pmatch[0].rm_eo;
1773 } while (*s);
1774 return n;
1775 }
1776 if (c[0] == '\0') { /* null split */
1777 while (*s) {
1778 *s1++ = *s++;
1779 *s1++ = '\0';
1780 n++;
1781 }
1782 return n;
1783 }
1784 if (c[0] != ' ') { /* single-character split */
1785 if (icase) {
1786 c[0] = toupper(c[0]);
1787 c[1] = tolower(c[1]);
1788 }
1789 if (*s1)
1790 n++;
1791 while ((s1 = strpbrk(s1, c)) != NULL) {
1792 *s1++ = '\0';
1793 n++;
1794 }
1795 return n;
1796 }
1797 /* space split */
1798 while (*s) {
1799 s = skip_whitespace(s);
1800 if (!*s)
1801 break;
1802 n++;
1803 while (*s && !isspace(*s))
1804 *s1++ = *s++;
1805 *s1++ = '\0';
1806 }
1807 return n;
1808}
1809
1810static void split_f0(void)
1811{
1812/* static char *fstrings; */
1813#define fstrings (G.split_f0__fstrings)
1814
1815 int i, n;
1816 char *s;
1817
1818 if (is_f0_split)
1819 return;
1820
1821 is_f0_split = TRUE;
1822 free(fstrings);
1823 fsrealloc(0);
1824 n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings);
1825 fsrealloc(n);
1826 s = fstrings;
1827 for (i = 0; i < n; i++) {
1828 Fields[i].string = nextword(&s);
1829 Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY);
1830 }
1831
1832 /* set NF manually to avoid side effects */
1833 clrvar(intvar[NF]);
1834 intvar[NF]->type = VF_NUMBER | VF_SPECIAL;
1835 intvar[NF]->number = nfields;
1836#undef fstrings
1837}
1838
1839/* perform additional actions when some internal variables changed */
1840static void handle_special(var *v)
1841{
1842 int n;
1843 char *b;
1844 const char *sep, *s;
1845 int sl, l, len, i, bsize;
1846
1847 if (!(v->type & VF_SPECIAL))
1848 return;
1849
1850 if (v == intvar[NF]) {
1851 n = (int)getvar_i(v);
1852 fsrealloc(n);
1853
1854 /* recalculate $0 */
1855 sep = getvar_s(intvar[OFS]);
1856 sl = strlen(sep);
1857 b = NULL;
1858 len = 0;
1859 for (i = 0; i < n; i++) {
1860 s = getvar_s(&Fields[i]);
1861 l = strlen(s);
1862 if (b) {
1863 memcpy(b+len, sep, sl);
1864 len += sl;
1865 }
1866 b = qrealloc(b, len+l+sl, &bsize);
1867 memcpy(b+len, s, l);
1868 len += l;
1869 }
1870 if (b)
1871 b[len] = '\0';
1872 setvar_p(intvar[F0], b);
1873 is_f0_split = TRUE;
1874
1875 } else if (v == intvar[F0]) {
1876 is_f0_split = FALSE;
1877
1878 } else if (v == intvar[FS]) {
1879 /*
1880 * The POSIX-2008 standard says that changing FS should have no effect on the
1881 * current input line, but only on the next one. The language is:
1882 *
1883 * > Before the first reference to a field in the record is evaluated, the record
1884 * > shall be split into fields, according to the rules in Regular Expressions,
1885 * > using the value of FS that was current at the time the record was read.
1886 *
1887 * So, split up current line before assignment to FS:
1888 */
1889 split_f0();
1890
1891 mk_splitter(getvar_s(v), &fsplitter);
1892 } else if (v == intvar[RS]) {
1893 mk_splitter(getvar_s(v), &rsplitter);
1894 } else if (v == intvar[IGNORECASE]) {
1895 icase = istrue(v);
1896 } else { /* $n */
1897 n = getvar_i(intvar[NF]);
1898 setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1);
1899 /* right here v is invalid. Just to note... */
1900 }
1901}
1902
1903/* step through func/builtin/etc arguments */
1904static node *nextarg(node **pn)
1905{
1906 node *n;
1907
1908 n = *pn;
1909 if (n && (n->info & OPCLSMASK) == OC_COMMA) {
1910 *pn = n->r.n;
1911 n = n->l.n;
1912 } else {
1913 *pn = NULL;
1914 }
1915 return n;
1916}
1917
1918static void hashwalk_init(var *v, xhash *array)
1919{
1920 hash_item *hi;
1921 unsigned i;
1922 walker_list *w;
1923 walker_list *prev_walker;
1924
1925 if (v->type & VF_WALK) {
1926 prev_walker = v->x.walker;
1927 } else {
1928 v->type |= VF_WALK;
1929 prev_walker = NULL;
1930 }
1931 debug_printf_walker("hashwalk_init: prev_walker:%p\n", prev_walker);
1932
1933 w = v->x.walker = xzalloc(sizeof(*w) + array->glen + 1); /* why + 1? */
1934 debug_printf_walker(" walker@%p=%p\n", &v->x.walker, w);
1935 w->cur = w->end = w->wbuf;
1936 w->prev = prev_walker;
1937 for (i = 0; i < array->csize; i++) {
1938 hi = array->items[i];
1939 while (hi) {
1940 strcpy(w->end, hi->name);
1941 nextword(&w->end);
1942 hi = hi->next;
1943 }
1944 }
1945}
1946
1947static int hashwalk_next(var *v)
1948{
1949 walker_list *w = v->x.walker;
1950
1951 if (w->cur >= w->end) {
1952 walker_list *prev_walker = w->prev;
1953
1954 debug_printf_walker("end of iteration, free(walker@%p:%p), prev_walker:%p\n", &v->x.walker, w, prev_walker);
1955 free(w);
1956 v->x.walker = prev_walker;
1957 return FALSE;
1958 }
1959
1960 setvar_s(v, nextword(&w->cur));
1961 return TRUE;
1962}
1963
1964/* evaluate node, return 1 when result is true, 0 otherwise */
1965static int ptest(node *pattern)
1966{
1967 /* ptest__v is "static": to save stack space? */
1968 return istrue(evaluate(pattern, &G.ptest__v));
1969}
1970
1971/* read next record from stream rsm into a variable v */
1972static int awk_getline(rstream *rsm, var *v)
1973{
1974 char *b;
1975 regmatch_t pmatch[2];
1976 int size, a, p, pp = 0;
1977 int fd, so, eo, r, rp;
1978 char c, *m, *s;
1979
1980 debug_printf_eval("entered %s()\n", __func__);
1981
1982 /* we're using our own buffer since we need access to accumulating
1983 * characters
1984 */
1985 fd = fileno(rsm->F);
1986 m = rsm->buffer;
1987 a = rsm->adv;
1988 p = rsm->pos;
1989 size = rsm->size;
1990 c = (char) rsplitter.n.info;
1991 rp = 0;
1992
1993 if (!m)
1994 m = qrealloc(m, 256, &size);
1995
1996 do {
1997 b = m + a;
1998 so = eo = p;
1999 r = 1;
2000 if (p > 0) {
2001 if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) {
2002 if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re,
2003 b, 1, pmatch, 0) == 0) {
2004 so = pmatch[0].rm_so;
2005 eo = pmatch[0].rm_eo;
2006 if (b[eo] != '\0')
2007 break;
2008 }
2009 } else if (c != '\0') {
2010 s = strchr(b+pp, c);
2011 if (!s)
2012 s = memchr(b+pp, '\0', p - pp);
2013 if (s) {
2014 so = eo = s-b;
2015 eo++;
2016 break;
2017 }
2018 } else {
2019 while (b[rp] == '\n')
2020 rp++;
2021 s = strstr(b+rp, "\n\n");
2022 if (s) {
2023 so = eo = s-b;
2024 while (b[eo] == '\n')
2025 eo++;
2026 if (b[eo] != '\0')
2027 break;
2028 }
2029 }
2030 }
2031
2032 if (a > 0) {
2033 memmove(m, m+a, p+1);
2034 b = m;
2035 a = 0;
2036 }
2037
2038 m = qrealloc(m, a+p+128, &size);
2039 b = m + a;
2040 pp = p;
2041 p += safe_read(fd, b+p, size-p-1);
2042 if (p < pp) {
2043 p = 0;
2044 r = 0;
2045 setvar_i(intvar[ERRNO], errno);
2046 }
2047 b[p] = '\0';
2048
2049 } while (p > pp);
2050
2051 if (p == 0) {
2052 r--;
2053 } else {
2054 c = b[so]; b[so] = '\0';
2055 setvar_s(v, b+rp);
2056 v->type |= VF_USER;
2057 b[so] = c;
2058 c = b[eo]; b[eo] = '\0';
2059 setvar_s(intvar[RT], b+so);
2060 b[eo] = c;
2061 }
2062
2063 rsm->buffer = m;
2064 rsm->adv = a + eo;
2065 rsm->pos = p - eo;
2066 rsm->size = size;
2067
2068 debug_printf_eval("returning from %s(): %d\n", __func__, r);
2069
2070 return r;
2071}
2072
2073static int fmt_num(char *b, int size, const char *format, double n, int int_as_int)
2074{
2075 int r = 0;
2076 char c;
2077 const char *s = format;
2078
2079 if (int_as_int && n == (long long)n) {
2080 r = snprintf(b, size, "%lld", (long long)n);
2081 } else {
2082 do { c = *s; } while (c && *++s);
2083 if (strchr("diouxX", c)) {
2084 r = snprintf(b, size, format, (int)n);
2085 } else if (strchr("eEfgG", c)) {
2086 r = snprintf(b, size, format, n);
2087 } else {
2088 syntax_error(EMSG_INV_FMT);
2089 }
2090 }
2091 return r;
2092}
2093
2094/* formatted output into an allocated buffer, return ptr to buffer */
2095static char *awk_printf(node *n)
2096{
2097 char *b = NULL;
2098 char *fmt, *s, *f;
2099 const char *s1;
2100 int i, j, incr, bsize;
2101 char c, c1;
2102 var *v, *arg;
2103
2104 v = nvalloc(1);
2105 fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v)));
2106
2107 i = 0;
2108 while (*f) {
2109 s = f;
2110 while (*f && (*f != '%' || *++f == '%'))
2111 f++;
2112 while (*f && !isalpha(*f)) {
2113 if (*f == '*')
2114 syntax_error("%*x formats are not supported");
2115 f++;
2116 }
2117
2118 incr = (f - s) + MAXVARFMT;
2119 b = qrealloc(b, incr + i, &bsize);
2120 c = *f;
2121 if (c != '\0')
2122 f++;
2123 c1 = *f;
2124 *f = '\0';
2125 arg = evaluate(nextarg(&n), v);
2126
2127 j = i;
2128 if (c == 'c' || !c) {
2129 i += sprintf(b+i, s, is_numeric(arg) ?
2130 (char)getvar_i(arg) : *getvar_s(arg));
2131 } else if (c == 's') {
2132 s1 = getvar_s(arg);
2133 b = qrealloc(b, incr+i+strlen(s1), &bsize);
2134 i += sprintf(b+i, s, s1);
2135 } else {
2136 i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE);
2137 }
2138 *f = c1;
2139
2140 /* if there was an error while sprintf, return value is negative */
2141 if (i < j)
2142 i = j;
2143 }
2144
2145 free(fmt);
2146 nvfree(v);
2147 b = xrealloc(b, i + 1);
2148 b[i] = '\0';
2149 return b;
2150}
2151
2152/* Common substitution routine.
2153 * Replace (nm)'th substring of (src) that matches (rn) with (repl),
2154 * store result into (dest), return number of substitutions.
2155 * If nm = 0, replace all matches.
2156 * If src or dst is NULL, use $0.
2157 * If subexp != 0, enable subexpression matching (\1-\9).
2158 */
2159static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int subexp)
2160{
2161 char *resbuf;
2162 const char *sp;
2163 int match_no, residx, replen, resbufsize;
2164 int regexec_flags;
2165 regmatch_t pmatch[10];
2166 regex_t sreg, *regex;
2167
2168 resbuf = NULL;
2169 residx = 0;
2170 match_no = 0;
2171 regexec_flags = 0;
2172 regex = as_regex(rn, &sreg);
2173 sp = getvar_s(src ? src : intvar[F0]);
2174 replen = strlen(repl);
2175 while (regexec(regex, sp, 10, pmatch, regexec_flags) == 0) {
2176 int so = pmatch[0].rm_so;
2177 int eo = pmatch[0].rm_eo;
2178
2179 //bb_error_msg("match %u: [%u,%u] '%s'%p", match_no+1, so, eo, sp,sp);
2180 resbuf = qrealloc(resbuf, residx + eo + replen, &resbufsize);
2181 memcpy(resbuf + residx, sp, eo);
2182 residx += eo;
2183 if (++match_no >= nm) {
2184 const char *s;
2185 int nbs;
2186
2187 /* replace */
2188 residx -= (eo - so);
2189 nbs = 0;
2190 for (s = repl; *s; s++) {
2191 char c = resbuf[residx++] = *s;
2192 if (c == '\\') {
2193 nbs++;
2194 continue;
2195 }
2196 if (c == '&' || (subexp && c >= '0' && c <= '9')) {
2197 int j;
2198 residx -= ((nbs + 3) >> 1);
2199 j = 0;
2200 if (c != '&') {
2201 j = c - '0';
2202 nbs++;
2203 }
2204 if (nbs % 2) {
2205 resbuf[residx++] = c;
2206 } else {
2207 int n = pmatch[j].rm_eo - pmatch[j].rm_so;
2208 resbuf = qrealloc(resbuf, residx + replen + n, &resbufsize);
2209 memcpy(resbuf + residx, sp + pmatch[j].rm_so, n);
2210 residx += n;
2211 }
2212 }
2213 nbs = 0;
2214 }
2215 }
2216
2217 regexec_flags = REG_NOTBOL;
2218 sp += eo;
2219 if (match_no == nm)
2220 break;
2221 if (eo == so) {
2222 /* Empty match (e.g. "b*" will match anywhere).
2223 * Advance by one char. */
2224//BUG (bug 1333):
2225//gsub(/\<b*/,"") on "abc" will reach this point, advance to "bc"
2226//... and will erroneously match "b" even though it is NOT at the word start.
2227//we need REG_NOTBOW but it does not exist...
2228//TODO: if EXTRA_COMPAT=y, use GNU matching and re_search,
2229//it should be able to do it correctly.
2230 /* Subtle: this is safe only because
2231 * qrealloc allocated at least one extra byte */
2232 resbuf[residx] = *sp;
2233 if (*sp == '\0')
2234 goto ret;
2235 sp++;
2236 residx++;
2237 }
2238 }
2239
2240 resbuf = qrealloc(resbuf, residx + strlen(sp), &resbufsize);
2241 strcpy(resbuf + residx, sp);
2242 ret:
2243 //bb_error_msg("end sp:'%s'%p", sp,sp);
2244 setvar_p(dest ? dest : intvar[F0], resbuf);
2245 if (regex == &sreg)
2246 regfree(regex);
2247 return match_no;
2248}
2249
2250static NOINLINE int do_mktime(const char *ds)
2251{
2252 struct tm then;
2253 int count;
2254
2255 /*memset(&then, 0, sizeof(then)); - not needed */
2256 then.tm_isdst = -1; /* default is unknown */
2257
2258 /* manpage of mktime says these fields are ints,
2259 * so we can sscanf stuff directly into them */
2260 count = sscanf(ds, "%u %u %u %u %u %u %d",
2261 &then.tm_year, &then.tm_mon, &then.tm_mday,
2262 &then.tm_hour, &then.tm_min, &then.tm_sec,
2263 &then.tm_isdst);
2264
2265 if (count < 6
2266 || (unsigned)then.tm_mon < 1
2267 || (unsigned)then.tm_year < 1900
2268 ) {
2269 return -1;
2270 }
2271
2272 then.tm_mon -= 1;
2273 then.tm_year -= 1900;
2274
2275 return mktime(&then);
2276}
2277
2278static NOINLINE var *exec_builtin(node *op, var *res)
2279{
2280#define tspl (G.exec_builtin__tspl)
2281
2282 var *tv;
2283 node *an[4];
2284 var *av[4];
2285 const char *as[4];
2286 regmatch_t pmatch[2];
2287 regex_t sreg, *re;
2288 node *spl;
2289 uint32_t isr, info;
2290 int nargs;
2291 time_t tt;
2292 int i, l, ll, n;
2293
2294 tv = nvalloc(4);
2295 isr = info = op->info;
2296 op = op->l.n;
2297
2298 av[2] = av[3] = NULL;
2299 for (i = 0; i < 4 && op; i++) {
2300 an[i] = nextarg(&op);
2301 if (isr & 0x09000000)
2302 av[i] = evaluate(an[i], &tv[i]);
2303 if (isr & 0x08000000)
2304 as[i] = getvar_s(av[i]);
2305 isr >>= 1;
2306 }
2307
2308 nargs = i;
2309 if ((uint32_t)nargs < (info >> 30))
2310 syntax_error(EMSG_TOO_FEW_ARGS);
2311
2312 info &= OPNMASK;
2313 switch (info) {
2314
2315 case B_a2:
2316 if (ENABLE_FEATURE_AWK_LIBM)
2317 setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1])));
2318 else
2319 syntax_error(EMSG_NO_MATH);
2320 break;
2321
2322 case B_sp: {
2323 char *s, *s1;
2324
2325 if (nargs > 2) {
2326 spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ?
2327 an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl);
2328 } else {
2329 spl = &fsplitter.n;
2330 }
2331
2332 n = awk_split(as[0], spl, &s);
2333 s1 = s;
2334 clear_array(iamarray(av[1]));
2335 for (i = 1; i <= n; i++)
2336 setari_u(av[1], i, nextword(&s));
2337 free(s1);
2338 setvar_i(res, n);
2339 break;
2340 }
2341
2342 case B_ss: {
2343 char *s;
2344
2345 l = strlen(as[0]);
2346 i = getvar_i(av[1]) - 1;
2347 if (i > l)
2348 i = l;
2349 if (i < 0)
2350 i = 0;
2351 n = (nargs > 2) ? getvar_i(av[2]) : l-i;
2352 if (n < 0)
2353 n = 0;
2354 s = xstrndup(as[0]+i, n);
2355 setvar_p(res, s);
2356 break;
2357 }
2358
2359 /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5:
2360 * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */
2361 case B_an:
2362 setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1]));
2363 break;
2364
2365 case B_co:
2366 setvar_i(res, ~getvar_i_int(av[0]));
2367 break;
2368
2369 case B_ls:
2370 setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1]));
2371 break;
2372
2373 case B_or:
2374 setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1]));
2375 break;
2376
2377 case B_rs:
2378 setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1]));
2379 break;
2380
2381 case B_xo:
2382 setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1]));
2383 break;
2384
2385 case B_lo:
2386 case B_up: {
2387 char *s, *s1;
2388 s1 = s = xstrdup(as[0]);
2389 while (*s1) {
2390 //*s1 = (info == B_up) ? toupper(*s1) : tolower(*s1);
2391 if ((unsigned char)((*s1 | 0x20) - 'a') <= ('z' - 'a'))
2392 *s1 = (info == B_up) ? (*s1 & 0xdf) : (*s1 | 0x20);
2393 s1++;
2394 }
2395 setvar_p(res, s);
2396 break;
2397 }
2398
2399 case B_ix:
2400 n = 0;
2401 ll = strlen(as[1]);
2402 l = strlen(as[0]) - ll;
2403 if (ll > 0 && l >= 0) {
2404 if (!icase) {
2405 char *s = strstr(as[0], as[1]);
2406 if (s)
2407 n = (s - as[0]) + 1;
2408 } else {
2409 /* this piece of code is terribly slow and
2410 * really should be rewritten
2411 */
2412 for (i = 0; i <= l; i++) {
2413 if (strncasecmp(as[0]+i, as[1], ll) == 0) {
2414 n = i+1;
2415 break;
2416 }
2417 }
2418 }
2419 }
2420 setvar_i(res, n);
2421 break;
2422
2423 case B_ti:
2424 if (nargs > 1)
2425 tt = getvar_i(av[1]);
2426 else
2427 time(&tt);
2428 //s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y";
2429 i = strftime(g_buf, MAXVARFMT,
2430 ((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"),
2431 localtime(&tt));
2432 g_buf[i] = '\0';
2433 setvar_s(res, g_buf);
2434 break;
2435
2436 case B_mt:
2437 setvar_i(res, do_mktime(as[0]));
2438 break;
2439
2440 case B_ma:
2441 re = as_regex(an[1], &sreg);
2442 n = regexec(re, as[0], 1, pmatch, 0);
2443 if (n == 0) {
2444 pmatch[0].rm_so++;
2445 pmatch[0].rm_eo++;
2446 } else {
2447 pmatch[0].rm_so = 0;
2448 pmatch[0].rm_eo = -1;
2449 }
2450 setvar_i(newvar("RSTART"), pmatch[0].rm_so);
2451 setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so);
2452 setvar_i(res, pmatch[0].rm_so);
2453 if (re == &sreg)
2454 regfree(re);
2455 break;
2456
2457 case B_ge:
2458 awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE);
2459 break;
2460
2461 case B_gs:
2462 setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE));
2463 break;
2464
2465 case B_su:
2466 setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE));
2467 break;
2468 }
2469
2470 nvfree(tv);
2471 return res;
2472#undef tspl
2473}
2474
2475/*
2476 * Evaluate node - the heart of the program. Supplied with subtree
2477 * and place where to store result. returns ptr to result.
2478 */
2479#define XC(n) ((n) >> 8)
2480
2481static var *evaluate(node *op, var *res)
2482{
2483/* This procedure is recursive so we should count every byte */
2484#define fnargs (G.evaluate__fnargs)
2485/* seed is initialized to 1 */
2486#define seed (G.evaluate__seed)
2487#define sreg (G.evaluate__sreg)
2488
2489 var *v1;
2490
2491 if (!op)
2492 return setvar_s(res, NULL);
2493
2494 debug_printf_eval("entered %s()\n", __func__);
2495
2496 v1 = nvalloc(2);
2497
2498 while (op) {
2499 struct {
2500 var *v;
2501 const char *s;
2502 } L = L; /* for compiler */
2503 struct {
2504 var *v;
2505 const char *s;
2506 } R = R;
2507 double L_d = L_d;
2508 uint32_t opinfo;
2509 int opn;
2510 node *op1;
2511
2512 opinfo = op->info;
2513 opn = (opinfo & OPNMASK);
2514 g_lineno = op->lineno;
2515 op1 = op->l.n;
2516 debug_printf_eval("opinfo:%08x opn:%08x\n", opinfo, opn);
2517
2518 /* execute inevitable things */
2519 if (opinfo & OF_RES1)
2520 L.v = evaluate(op1, v1);
2521 if (opinfo & OF_RES2)
2522 R.v = evaluate(op->r.n, v1+1);
2523 if (opinfo & OF_STR1) {
2524 L.s = getvar_s(L.v);
2525 debug_printf_eval("L.s:'%s'\n", L.s);
2526 }
2527 if (opinfo & OF_STR2) {
2528 R.s = getvar_s(R.v);
2529 debug_printf_eval("R.s:'%s'\n", R.s);
2530 }
2531 if (opinfo & OF_NUM1) {
2532 L_d = getvar_i(L.v);
2533 debug_printf_eval("L_d:%f\n", L_d);
2534 }
2535
2536 debug_printf_eval("switch(0x%x)\n", XC(opinfo & OPCLSMASK));
2537 switch (XC(opinfo & OPCLSMASK)) {
2538
2539 /* -- iterative node type -- */
2540
2541 /* test pattern */
2542 case XC( OC_TEST ):
2543 if ((op1->info & OPCLSMASK) == OC_COMMA) {
2544 /* it's range pattern */
2545 if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) {
2546 op->info |= OF_CHECKED;
2547 if (ptest(op1->r.n))
2548 op->info &= ~OF_CHECKED;
2549 op = op->a.n;
2550 } else {
2551 op = op->r.n;
2552 }
2553 } else {
2554 op = ptest(op1) ? op->a.n : op->r.n;
2555 }
2556 break;
2557
2558 /* just evaluate an expression, also used as unconditional jump */
2559 case XC( OC_EXEC ):
2560 break;
2561
2562 /* branch, used in if-else and various loops */
2563 case XC( OC_BR ):
2564 op = istrue(L.v) ? op->a.n : op->r.n;
2565 break;
2566
2567 /* initialize for-in loop */
2568 case XC( OC_WALKINIT ):
2569 hashwalk_init(L.v, iamarray(R.v));
2570 break;
2571
2572 /* get next array item */
2573 case XC( OC_WALKNEXT ):
2574 op = hashwalk_next(L.v) ? op->a.n : op->r.n;
2575 break;
2576
2577 case XC( OC_PRINT ):
2578 case XC( OC_PRINTF ): {
2579 FILE *F = stdout;
2580
2581 if (op->r.n) {
2582 rstream *rsm = newfile(R.s);
2583 if (!rsm->F) {
2584 if (opn == '|') {
2585 rsm->F = popen(R.s, "w");
2586 if (rsm->F == NULL)
2587 bb_perror_msg_and_die("popen");
2588 rsm->is_pipe = 1;
2589 } else {
2590 rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
2591 }
2592 }
2593 F = rsm->F;
2594 }
2595
2596 if ((opinfo & OPCLSMASK) == OC_PRINT) {
2597 if (!op1) {
2598 fputs(getvar_s(intvar[F0]), F);
2599 } else {
2600 while (op1) {
2601 var *v = evaluate(nextarg(&op1), v1);
2602 if (v->type & VF_NUMBER) {
2603 fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]),
2604 getvar_i(v), TRUE);
2605 fputs(g_buf, F);
2606 } else {
2607 fputs(getvar_s(v), F);
2608 }
2609
2610 if (op1)
2611 fputs(getvar_s(intvar[OFS]), F);
2612 }
2613 }
2614 fputs(getvar_s(intvar[ORS]), F);
2615
2616 } else { /* OC_PRINTF */
2617 char *s = awk_printf(op1);
2618 fputs(s, F);
2619 free(s);
2620 }
2621 fflush(F);
2622 break;
2623 }
2624
2625 case XC( OC_DELETE ): {
2626 uint32_t info = op1->info & OPCLSMASK;
2627 var *v;
2628
2629 if (info == OC_VAR) {
2630 v = op1->l.v;
2631 } else if (info == OC_FNARG) {
2632 v = &fnargs[op1->l.aidx];
2633 } else {
2634 syntax_error(EMSG_NOT_ARRAY);
2635 }
2636
2637 if (op1->r.n) {
2638 const char *s;
2639 clrvar(L.v);
2640 s = getvar_s(evaluate(op1->r.n, v1));
2641 hash_remove(iamarray(v), s);
2642 } else {
2643 clear_array(iamarray(v));
2644 }
2645 break;
2646 }
2647
2648 case XC( OC_NEWSOURCE ):
2649 g_progname = op->l.new_progname;
2650 break;
2651
2652 case XC( OC_RETURN ):
2653 copyvar(res, L.v);
2654 break;
2655
2656 case XC( OC_NEXTFILE ):
2657 nextfile = TRUE;
2658 case XC( OC_NEXT ):
2659 nextrec = TRUE;
2660 case XC( OC_DONE ):
2661 clrvar(res);
2662 break;
2663
2664 case XC( OC_EXIT ):
2665 awk_exit(L_d);
2666
2667 /* -- recursive node type -- */
2668
2669 case XC( OC_VAR ):
2670 L.v = op->l.v;
2671 if (L.v == intvar[NF])
2672 split_f0();
2673 goto v_cont;
2674
2675 case XC( OC_FNARG ):
2676 L.v = &fnargs[op->l.aidx];
2677 v_cont:
2678 res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v;
2679 break;
2680
2681 case XC( OC_IN ):
2682 setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0);
2683 break;
2684
2685 case XC( OC_REGEXP ):
2686 op1 = op;
2687 L.s = getvar_s(intvar[F0]);
2688 goto re_cont;
2689
2690 case XC( OC_MATCH ):
2691 op1 = op->r.n;
2692 re_cont:
2693 {
2694 regex_t *re = as_regex(op1, &sreg);
2695 int i = regexec(re, L.s, 0, NULL, 0);
2696 if (re == &sreg)
2697 regfree(re);
2698 setvar_i(res, (i == 0) ^ (opn == '!'));
2699 }
2700 break;
2701
2702 case XC( OC_MOVE ):
2703 debug_printf_eval("MOVE\n");
2704 /* if source is a temporary string, jusk relink it to dest */
2705//Disabled: if R.v is numeric but happens to have cached R.v->string,
2706//then L.v ends up being a string, which is wrong
2707// if (R.v == v1+1 && R.v->string) {
2708// res = setvar_p(L.v, R.v->string);
2709// R.v->string = NULL;
2710// } else {
2711 res = copyvar(L.v, R.v);
2712// }
2713 break;
2714
2715 case XC( OC_TERNARY ):
2716 if ((op->r.n->info & OPCLSMASK) != OC_COLON)
2717 syntax_error(EMSG_POSSIBLE_ERROR);
2718 res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res);
2719 break;
2720
2721 case XC( OC_FUNC ): {
2722 var *vbeg, *v;
2723 const char *sv_progname;
2724
2725 /* The body might be empty, still has to eval the args */
2726 if (!op->r.n->info && !op->r.f->body.first)
2727 syntax_error(EMSG_UNDEF_FUNC);
2728
2729 vbeg = v = nvalloc(op->r.f->nargs + 1);
2730 while (op1) {
2731 var *arg = evaluate(nextarg(&op1), v1);
2732 copyvar(v, arg);
2733 v->type |= VF_CHILD;
2734 v->x.parent = arg;
2735 if (++v - vbeg >= op->r.f->nargs)
2736 break;
2737 }
2738
2739 v = fnargs;
2740 fnargs = vbeg;
2741 sv_progname = g_progname;
2742
2743 res = evaluate(op->r.f->body.first, res);
2744
2745 g_progname = sv_progname;
2746 nvfree(fnargs);
2747 fnargs = v;
2748
2749 break;
2750 }
2751
2752 case XC( OC_GETLINE ):
2753 case XC( OC_PGETLINE ): {
2754 rstream *rsm;
2755 int i;
2756
2757 if (op1) {
2758 rsm = newfile(L.s);
2759 if (!rsm->F) {
2760 if ((opinfo & OPCLSMASK) == OC_PGETLINE) {
2761 rsm->F = popen(L.s, "r");
2762 rsm->is_pipe = TRUE;
2763 } else {
2764 rsm->F = fopen_for_read(L.s); /* not xfopen! */
2765 }
2766 }
2767 } else {
2768 if (!iF)
2769 iF = next_input_file();
2770 rsm = iF;
2771 }
2772
2773 if (!rsm || !rsm->F) {
2774 setvar_i(intvar[ERRNO], errno);
2775 setvar_i(res, -1);
2776 break;
2777 }
2778
2779 if (!op->r.n)
2780 R.v = intvar[F0];
2781
2782 i = awk_getline(rsm, R.v);
2783 if (i > 0 && !op1) {
2784 incvar(intvar[FNR]);
2785 incvar(intvar[NR]);
2786 }
2787 setvar_i(res, i);
2788 break;
2789 }
2790
2791 /* simple builtins */
2792 case XC( OC_FBLTIN ): {
2793 double R_d = R_d; /* for compiler */
2794
2795 switch (opn) {
2796 case F_in:
2797 R_d = (long long)L_d;
2798 break;
2799
2800 case F_rn:
2801 R_d = (double)rand() / (double)RAND_MAX;
2802 break;
2803
2804 case F_co:
2805 if (ENABLE_FEATURE_AWK_LIBM) {
2806 R_d = cos(L_d);
2807 break;
2808 }
2809
2810 case F_ex:
2811 if (ENABLE_FEATURE_AWK_LIBM) {
2812 R_d = exp(L_d);
2813 break;
2814 }
2815
2816 case F_lg:
2817 if (ENABLE_FEATURE_AWK_LIBM) {
2818 R_d = log(L_d);
2819 break;
2820 }
2821
2822 case F_si:
2823 if (ENABLE_FEATURE_AWK_LIBM) {
2824 R_d = sin(L_d);
2825 break;
2826 }
2827
2828 case F_sq:
2829 if (ENABLE_FEATURE_AWK_LIBM) {
2830 R_d = sqrt(L_d);
2831 break;
2832 }
2833
2834 syntax_error(EMSG_NO_MATH);
2835 break;
2836
2837 case F_sr:
2838 R_d = (double)seed;
2839 seed = op1 ? (unsigned)L_d : (unsigned)time(NULL);
2840 srand(seed);
2841 break;
2842
2843 case F_ti:
2844 R_d = time(NULL);
2845 break;
2846
2847 case F_le:
2848 debug_printf_eval("length: L.s:'%s'\n", L.s);
2849 if (!op1) {
2850 L.s = getvar_s(intvar[F0]);
2851 debug_printf_eval("length: L.s='%s'\n", L.s);
2852 }
2853 else if (L.v->type & VF_ARRAY) {
2854 R_d = L.v->x.array->nel;
2855 debug_printf_eval("length: array_len:%d\n", L.v->x.array->nel);
2856 break;
2857 }
2858 R_d = strlen(L.s);
2859 break;
2860
2861 case F_sy:
2862 fflush_all();
2863 R_d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
2864 ? (system(L.s) >> 8) : 0;
2865 break;
2866
2867 case F_ff:
2868 if (!op1) {
2869 fflush(stdout);
2870 } else if (L.s && *L.s) {
2871 rstream *rsm = newfile(L.s);
2872 fflush(rsm->F);
2873 } else {
2874 fflush_all();
2875 }
2876 break;
2877
2878 case F_cl: {
2879 rstream *rsm;
2880 int err = 0;
2881 rsm = (rstream *)hash_search(fdhash, L.s);
2882 debug_printf_eval("OC_FBLTIN F_cl rsm:%p\n", rsm);
2883 if (rsm) {
2884 debug_printf_eval("OC_FBLTIN F_cl "
2885 "rsm->is_pipe:%d, ->F:%p\n",
2886 rsm->is_pipe, rsm->F);
2887 /* Can be NULL if open failed. Example:
2888 * getline line <"doesnt_exist";
2889 * close("doesnt_exist"); <--- here rsm->F is NULL
2890 */
2891 if (rsm->F)
2892 err = rsm->is_pipe ? pclose(rsm->F) : fclose(rsm->F);
2893 free(rsm->buffer);
2894 hash_remove(fdhash, L.s);
2895 }
2896 if (err)
2897 setvar_i(intvar[ERRNO], errno);
2898 R_d = (double)err;
2899 break;
2900 }
2901 } /* switch */
2902 setvar_i(res, R_d);
2903 break;
2904 }
2905
2906 case XC( OC_BUILTIN ):
2907 res = exec_builtin(op, res);
2908 break;
2909
2910 case XC( OC_SPRINTF ):
2911 setvar_p(res, awk_printf(op1));
2912 break;
2913
2914 case XC( OC_UNARY ): {
2915 double Ld, R_d;
2916
2917 Ld = R_d = getvar_i(R.v);
2918 switch (opn) {
2919 case 'P':
2920 Ld = ++R_d;
2921 goto r_op_change;
2922 case 'p':
2923 R_d++;
2924 goto r_op_change;
2925 case 'M':
2926 Ld = --R_d;
2927 goto r_op_change;
2928 case 'm':
2929 R_d--;
2930 r_op_change:
2931 setvar_i(R.v, R_d);
2932 break;
2933 case '!':
2934 Ld = !istrue(R.v);
2935 break;
2936 case '-':
2937 Ld = -R_d;
2938 break;
2939 }
2940 setvar_i(res, Ld);
2941 break;
2942 }
2943
2944 case XC( OC_FIELD ): {
2945 int i = (int)getvar_i(R.v);
2946 if (i == 0) {
2947 res = intvar[F0];
2948 } else {
2949 split_f0();
2950 if (i > nfields)
2951 fsrealloc(i);
2952 res = &Fields[i - 1];
2953 }
2954 break;
2955 }
2956
2957 /* concatenation (" ") and index joining (",") */
2958 case XC( OC_CONCAT ):
2959 case XC( OC_COMMA ): {
2960 const char *sep = "";
2961 if ((opinfo & OPCLSMASK) == OC_COMMA)
2962 sep = getvar_s(intvar[SUBSEP]);
2963 setvar_p(res, xasprintf("%s%s%s", L.s, sep, R.s));
2964 break;
2965 }
2966
2967 case XC( OC_LAND ):
2968 setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0);
2969 break;
2970
2971 case XC( OC_LOR ):
2972 setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n));
2973 break;
2974
2975 case XC( OC_BINARY ):
2976 case XC( OC_REPLACE ): {
2977 double R_d = getvar_i(R.v);
2978 debug_printf_eval("BINARY/REPLACE: R_d:%f opn:%c\n", R_d, opn);
2979 switch (opn) {
2980 case '+':
2981 L_d += R_d;
2982 break;
2983 case '-':
2984 L_d -= R_d;
2985 break;
2986 case '*':
2987 L_d *= R_d;
2988 break;
2989 case '/':
2990 if (R_d == 0)
2991 syntax_error(EMSG_DIV_BY_ZERO);
2992 L_d /= R_d;
2993 break;
2994 case '&':
2995 if (ENABLE_FEATURE_AWK_LIBM)
2996 L_d = pow(L_d, R_d);
2997 else
2998 syntax_error(EMSG_NO_MATH);
2999 break;
3000 case '%':
3001 if (R_d == 0)
3002 syntax_error(EMSG_DIV_BY_ZERO);
3003 L_d -= (long long)(L_d / R_d) * R_d;
3004 break;
3005 }
3006 debug_printf_eval("BINARY/REPLACE result:%f\n", L_d);
3007 res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : L.v, L_d);
3008 break;
3009 }
3010
3011 case XC( OC_COMPARE ): {
3012 int i = i; /* for compiler */
3013 double Ld;
3014
3015 if (is_numeric(L.v) && is_numeric(R.v)) {
3016 Ld = getvar_i(L.v) - getvar_i(R.v);
3017 } else {
3018 const char *l = getvar_s(L.v);
3019 const char *r = getvar_s(R.v);
3020 Ld = icase ? strcasecmp(l, r) : strcmp(l, r);
3021 }
3022 switch (opn & 0xfe) {
3023 case 0:
3024 i = (Ld > 0);
3025 break;
3026 case 2:
3027 i = (Ld >= 0);
3028 break;
3029 case 4:
3030 i = (Ld == 0);
3031 break;
3032 }
3033 setvar_i(res, (i == 0) ^ (opn & 1));
3034 break;
3035 }
3036
3037 default:
3038 syntax_error(EMSG_POSSIBLE_ERROR);
3039 }
3040 if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS)
3041 op = op->a.n;
3042 if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS)
3043 break;
3044 if (nextrec)
3045 break;
3046 } /* while (op) */
3047
3048 nvfree(v1);
3049 debug_printf_eval("returning from %s(): %p\n", __func__, res);
3050 return res;
3051#undef fnargs
3052#undef seed
3053#undef sreg
3054}
3055
3056
3057/* -------- main & co. -------- */
3058
3059static int awk_exit(int r)
3060{
3061 var tv;
3062 unsigned i;
3063 hash_item *hi;
3064
3065 zero_out_var(&tv);
3066
3067 if (!exiting) {
3068 exiting = TRUE;
3069 nextrec = FALSE;
3070 evaluate(endseq.first, &tv);
3071 }
3072
3073 /* waiting for children */
3074 for (i = 0; i < fdhash->csize; i++) {
3075 hi = fdhash->items[i];
3076 while (hi) {
3077 if (hi->data.rs.F && hi->data.rs.is_pipe)
3078 pclose(hi->data.rs.F);
3079 hi = hi->next;
3080 }
3081 }
3082
3083 exit(r);
3084}
3085
3086/* if expr looks like "var=value", perform assignment and return 1,
3087 * otherwise return 0 */
3088static int is_assignment(const char *expr)
3089{
3090 char *exprc, *val;
3091
3092 if (!isalnum_(*expr) || (val = strchr(expr, '=')) == NULL) {
3093 return FALSE;
3094 }
3095
3096 exprc = xstrdup(expr);
3097 val = exprc + (val - expr);
3098 *val++ = '\0';
3099
3100 unescape_string_in_place(val);
3101 setvar_u(newvar(exprc), val);
3102 free(exprc);
3103 return TRUE;
3104}
3105
3106/* switch to next input file */
3107static rstream *next_input_file(void)
3108{
3109#define rsm (G.next_input_file__rsm)
3110#define files_happen (G.next_input_file__files_happen)
3111
3112 FILE *F;
3113 const char *fname, *ind;
3114
3115 if (rsm.F)
3116 fclose(rsm.F);
3117 rsm.F = NULL;
3118 rsm.pos = rsm.adv = 0;
3119
3120 for (;;) {
3121 if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) {
3122 if (files_happen)
3123 return NULL;
3124 fname = "-";
3125 F = stdin;
3126 break;
3127 }
3128 ind = getvar_s(incvar(intvar[ARGIND]));
3129 fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind));
3130 if (fname && *fname && !is_assignment(fname)) {
3131 F = xfopen_stdin(fname);
3132 break;
3133 }
3134 }
3135
3136 files_happen = TRUE;
3137 setvar_s(intvar[FILENAME], fname);
3138 rsm.F = F;
3139 return &rsm;
3140#undef rsm
3141#undef files_happen
3142}
3143
3144int awk_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
3145int awk_main(int argc, char **argv)
3146{
3147 unsigned opt;
3148 char *opt_F;
3149 llist_t *list_v = NULL;
3150 llist_t *list_f = NULL;
3151#if ENABLE_FEATURE_AWK_GNU_EXTENSIONS
3152 llist_t *list_e = NULL;
3153#endif
3154 int i, j;
3155 var *v;
3156 var tv;
3157 char **envp;
3158 char *vnames = (char *)vNames; /* cheat */
3159 char *vvalues = (char *)vValues;
3160
3161 INIT_G();
3162
3163 /* Undo busybox.c, or else strtod may eat ','! This breaks parsing:
3164 * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */
3165 if (ENABLE_LOCALE_SUPPORT)
3166 setlocale(LC_NUMERIC, "C");
3167
3168 zero_out_var(&tv);
3169
3170 /* allocate global buffer */
3171 g_buf = xmalloc(MAXVARFMT + 1);
3172
3173 vhash = hash_init();
3174 ahash = hash_init();
3175 fdhash = hash_init();
3176 fnhash = hash_init();
3177
3178 /* initialize variables */
3179 for (i = 0; *vnames; i++) {
3180 intvar[i] = v = newvar(nextword(&vnames));
3181 if (*vvalues != '\377')
3182 setvar_s(v, nextword(&vvalues));
3183 else
3184 setvar_i(v, 0);
3185
3186 if (*vnames == '*') {
3187 v->type |= VF_SPECIAL;
3188 vnames++;
3189 }
3190 }
3191
3192 handle_special(intvar[FS]);
3193 handle_special(intvar[RS]);
3194
3195 newfile("/dev/stdin")->F = stdin;
3196 newfile("/dev/stdout")->F = stdout;
3197 newfile("/dev/stderr")->F = stderr;
3198
3199 /* Huh, people report that sometimes environ is NULL. Oh well. */
3200 if (environ) for (envp = environ; *envp; envp++) {
3201 /* environ is writable, thus we don't strdup it needlessly */
3202 char *s = *envp;
3203 char *s1 = strchr(s, '=');
3204 if (s1) {
3205 *s1 = '\0';
3206 /* Both findvar and setvar_u take const char*
3207 * as 2nd arg -> environment is not trashed */
3208 setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1);
3209 *s1 = '=';
3210 }
3211 }
3212 opt_complementary = OPTCOMPLSTR_AWK;
3213 opt = getopt32(argv, OPTSTR_AWK, &opt_F, &list_v, &list_f, IF_FEATURE_AWK_GNU_EXTENSIONS(&list_e,) NULL);
3214 argv += optind;
3215 argc -= optind;
3216 if (opt & OPT_W)
3217 bb_error_msg("warning: option -W is ignored");
3218 if (opt & OPT_F) {
3219 unescape_string_in_place(opt_F);
3220 setvar_s(intvar[FS], opt_F);
3221 }
3222 while (list_v) {
3223 if (!is_assignment(llist_pop(&list_v)))
3224 bb_show_usage();
3225 }
3226 while (list_f) {
3227 char *s = NULL;
3228 FILE *from_file;
3229
3230 g_progname = llist_pop(&list_f);
3231 from_file = xfopen_stdin(g_progname);
3232 /* one byte is reserved for some trick in next_token */
3233 for (i = j = 1; j > 0; i += j) {
3234 s = xrealloc(s, i + 4096);
3235 j = fread(s + i, 1, 4094, from_file);
3236 }
3237 s[i] = '\0';
3238 fclose(from_file);
3239 parse_program(s + 1);
3240 free(s);
3241 }
3242 g_progname = "cmd. line";
3243#if ENABLE_FEATURE_AWK_GNU_EXTENSIONS
3244 while (list_e) {
3245 parse_program(llist_pop(&list_e));
3246 }
3247#endif
3248 if (!(opt & (OPT_f | OPT_e))) {
3249 if (!*argv)
3250 bb_show_usage();
3251 parse_program(*argv++);
3252 argc--;
3253 }
3254
3255 /* fill in ARGV array */
3256 setvar_i(intvar[ARGC], argc + 1);
3257 setari_u(intvar[ARGV], 0, "awk");
3258 i = 0;
3259 while (*argv)
3260 setari_u(intvar[ARGV], ++i, *argv++);
3261
3262 evaluate(beginseq.first, &tv);
3263 if (!mainseq.first && !endseq.first)
3264 awk_exit(EXIT_SUCCESS);
3265
3266 /* input file could already be opened in BEGIN block */
3267 if (!iF)
3268 iF = next_input_file();
3269
3270 /* passing through input files */
3271 while (iF) {
3272 nextfile = FALSE;
3273 setvar_i(intvar[FNR], 0);
3274
3275 while ((i = awk_getline(iF, intvar[F0])) > 0) {
3276 nextrec = FALSE;
3277 incvar(intvar[NR]);
3278 incvar(intvar[FNR]);
3279 evaluate(mainseq.first, &tv);
3280
3281 if (nextfile)
3282 break;
3283 }
3284
3285 if (i < 0)
3286 syntax_error(strerror(errno));
3287
3288 iF = next_input_file();
3289 }
3290
3291 awk_exit(EXIT_SUCCESS);
3292 /*return 0;*/
3293}
Note: See TracBrowser for help on using the repository browser.