1 /* vi: set sw=4 ts=4: */
3 * awk implementation for busybox
5 * Copyright (C) 2002 by Dmitry Zakharov <dmit@crp.bank.gov.ua>
7 * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
19 #define VF_NUMBER 0x0001 /* 1 = primary type is number */
20 #define VF_ARRAY 0x0002 /* 1 = it's an array */
22 #define VF_CACHED 0x0100 /* 1 = num/str value has cached str/num eq */
23 #define VF_USER 0x0200 /* 1 = user input (may be numeric string) */
24 #define VF_SPECIAL 0x0400 /* 1 = requires extra handling when changed */
25 #define VF_WALK 0x0800 /* 1 = variable has alloc'd x.walker list */
26 #define VF_FSTR 0x1000 /* 1 = string points to fstring buffer */
27 #define VF_CHILD 0x2000 /* 1 = function arg; x.parent points to source */
28 #define VF_DIRTY 0x4000 /* 1 = variable was set explicitly */
30 /* these flags are static, don't change them when value is changed */
31 #define VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY)
34 typedef struct var_s {
35 unsigned short type; /* flags */
39 int aidx; /* func arg idx (for compilation stage) */
40 struct xhash_s *array; /* array ptr */
41 struct var_s *parent; /* for func args, ptr to actual parameter */
42 char **walker; /* list of array elements (for..in) */
46 /* Node chain (pattern-action chain, BEGIN, END, function bodies) */
47 typedef struct chain_s {
54 typedef struct func_s {
60 typedef struct rstream_s {
66 unsigned short is_pipe;
69 typedef struct hash_item_s {
71 struct var_s v; /* variable/array hash */
72 struct rstream_s rs; /* redirect streams hash */
73 struct func_s f; /* functions hash */
75 struct hash_item_s *next; /* next in chain */
76 char name[1]; /* really it's longer */
79 typedef struct xhash_s {
80 unsigned nel; /* num of elements */
81 unsigned csize; /* current hash size */
82 unsigned nprime; /* next hash size in PRIMES[] */
83 unsigned glen; /* summary length of item names */
84 struct hash_item_s **items;
88 typedef struct node_s {
90 unsigned short lineno;
109 /* Block of temporary variables */
110 typedef struct nvblock_s {
113 struct nvblock_s *prev;
114 struct nvblock_s *next;
118 typedef struct tsplitter_s {
123 /* simple token classes */
124 /* Order and hex values are very important!!! See next_token() */
125 #define TC_SEQSTART 1 /* ( */
126 #define TC_SEQTERM (1 << 1) /* ) */
127 #define TC_REGEXP (1 << 2) /* /.../ */
128 #define TC_OUTRDR (1 << 3) /* | > >> */
129 #define TC_UOPPOST (1 << 4) /* unary postfix operator */
130 #define TC_UOPPRE1 (1 << 5) /* unary prefix operator */
131 #define TC_BINOPX (1 << 6) /* two-opnd operator */
132 #define TC_IN (1 << 7)
133 #define TC_COMMA (1 << 8)
134 #define TC_PIPE (1 << 9) /* input redirection pipe */
135 #define TC_UOPPRE2 (1 << 10) /* unary prefix operator */
136 #define TC_ARRTERM (1 << 11) /* ] */
137 #define TC_GRPSTART (1 << 12) /* { */
138 #define TC_GRPTERM (1 << 13) /* } */
139 #define TC_SEMICOL (1 << 14)
140 #define TC_NEWLINE (1 << 15)
141 #define TC_STATX (1 << 16) /* ctl statement (for, next...) */
142 #define TC_WHILE (1 << 17)
143 #define TC_ELSE (1 << 18)
144 #define TC_BUILTIN (1 << 19)
145 #define TC_GETLINE (1 << 20)
146 #define TC_FUNCDECL (1 << 21) /* `function' `func' */
147 #define TC_BEGIN (1 << 22)
148 #define TC_END (1 << 23)
149 #define TC_EOF (1 << 24)
150 #define TC_VARIABLE (1 << 25)
151 #define TC_ARRAY (1 << 26)
152 #define TC_FUNCTION (1 << 27)
153 #define TC_STRING (1 << 28)
154 #define TC_NUMBER (1 << 29)
156 #define TC_UOPPRE (TC_UOPPRE1 | TC_UOPPRE2)
158 /* combined token classes */
159 #define TC_BINOP (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN)
160 #define TC_UNARYOP (TC_UOPPRE | TC_UOPPOST)
161 #define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION | \
162 TC_BUILTIN | TC_GETLINE | TC_SEQSTART | TC_STRING | TC_NUMBER)
164 #define TC_STATEMNT (TC_STATX | TC_WHILE)
165 #define TC_OPTERM (TC_SEMICOL | TC_NEWLINE)
167 /* word tokens, cannot mean something else if not expected */
168 #define TC_WORD (TC_IN | TC_STATEMNT | TC_ELSE | TC_BUILTIN | \
169 TC_GETLINE | TC_FUNCDECL | TC_BEGIN | TC_END)
171 /* discard newlines after these */
172 #define TC_NOTERM (TC_COMMA | TC_GRPSTART | TC_GRPTERM | \
173 TC_BINOP | TC_OPTERM)
175 /* what can expression begin with */
176 #define TC_OPSEQ (TC_OPERAND | TC_UOPPRE | TC_REGEXP)
177 /* what can group begin with */
178 #define TC_GRPSEQ (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART)
180 /* if previous token class is CONCAT1 and next is CONCAT2, concatenation */
181 /* operator is inserted between them */
182 #define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM | \
183 TC_STRING | TC_NUMBER | TC_UOPPOST)
184 #define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE)
186 #define OF_RES1 0x010000
187 #define OF_RES2 0x020000
188 #define OF_STR1 0x040000
189 #define OF_STR2 0x080000
190 #define OF_NUM1 0x100000
191 #define OF_CHECKED 0x200000
193 /* combined operator flags */
196 #define xS (OF_RES2 | OF_STR2)
198 #define VV (OF_RES1 | OF_RES2)
199 #define Nx (OF_RES1 | OF_NUM1)
200 #define NV (OF_RES1 | OF_NUM1 | OF_RES2)
201 #define Sx (OF_RES1 | OF_STR1)
202 #define SV (OF_RES1 | OF_STR1 | OF_RES2)
203 #define SS (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2)
205 #define OPCLSMASK 0xFF00
206 #define OPNMASK 0x007F
208 /* operator priority is a highest byte (even: r->l, odd: l->r grouping)
209 * For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1,
210 * n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string
212 #define P(x) (x << 24)
213 #define PRIMASK 0x7F000000
214 #define PRIMASK2 0x7E000000
216 /* Operation classes */
218 #define SHIFT_TIL_THIS 0x0600
219 #define RECUR_FROM_THIS 0x1000
222 OC_DELETE=0x0100, OC_EXEC=0x0200, OC_NEWSOURCE=0x0300,
223 OC_PRINT=0x0400, OC_PRINTF=0x0500, OC_WALKINIT=0x0600,
225 OC_BR=0x0700, OC_BREAK=0x0800, OC_CONTINUE=0x0900,
226 OC_EXIT=0x0a00, OC_NEXT=0x0b00, OC_NEXTFILE=0x0c00,
227 OC_TEST=0x0d00, OC_WALKNEXT=0x0e00,
229 OC_BINARY=0x1000, OC_BUILTIN=0x1100, OC_COLON=0x1200,
230 OC_COMMA=0x1300, OC_COMPARE=0x1400, OC_CONCAT=0x1500,
231 OC_FBLTIN=0x1600, OC_FIELD=0x1700, OC_FNARG=0x1800,
232 OC_FUNC=0x1900, OC_GETLINE=0x1a00, OC_IN=0x1b00,
233 OC_LAND=0x1c00, OC_LOR=0x1d00, OC_MATCH=0x1e00,
234 OC_MOVE=0x1f00, OC_PGETLINE=0x2000, OC_REGEXP=0x2100,
235 OC_REPLACE=0x2200, OC_RETURN=0x2300, OC_SPRINTF=0x2400,
236 OC_TERNARY=0x2500, OC_UNARY=0x2600, OC_VAR=0x2700,
239 ST_IF=0x3000, ST_DO=0x3100, ST_FOR=0x3200,
243 /* simple builtins */
245 F_in=0, F_rn, F_co, F_ex, F_lg, F_si, F_sq, F_sr,
246 F_ti, F_le, F_sy, F_ff, F_cl
251 B_a2=0, B_ix, B_ma, B_sp, B_ss, B_ti, B_lo, B_up,
253 B_an, B_co, B_ls, B_or, B_rs, B_xo,
256 /* tokens and their corresponding info values */
258 #define NTC "\377" /* switch to next token class (tc<<1) */
261 #define OC_B OC_BUILTIN
263 static const char tokenlist[] =
266 "\1/" NTC /* REGEXP */
267 "\2>>" "\1>" "\1|" NTC /* OUTRDR */
268 "\2++" "\2--" NTC /* UOPPOST */
269 "\2++" "\2--" "\1$" NTC /* UOPPRE1 */
270 "\2==" "\1=" "\2+=" "\2-=" /* BINOPX */
271 "\2*=" "\2/=" "\2%=" "\2^="
272 "\1+" "\1-" "\3**=" "\2**"
273 "\1/" "\1%" "\1^" "\1*"
274 "\2!=" "\2>=" "\2<=" "\1>"
275 "\1<" "\2!~" "\1~" "\2&&"
276 "\2||" "\1?" "\1:" NTC
280 "\1+" "\1-" "\1!" NTC /* UOPPRE2 */
286 "\2if" "\2do" "\3for" "\5break" /* STATX */
287 "\10continue" "\6delete" "\5print"
288 "\6printf" "\4next" "\10nextfile"
289 "\6return" "\4exit" NTC
293 "\3and" "\5compl" "\6lshift" "\2or"
295 "\5close" "\6system" "\6fflush" "\5atan2" /* BUILTIN */
296 "\3cos" "\3exp" "\3int" "\3log"
297 "\4rand" "\3sin" "\4sqrt" "\5srand"
298 "\6gensub" "\4gsub" "\5index" "\6length"
299 "\5match" "\5split" "\7sprintf" "\3sub"
300 "\6substr" "\7systime" "\10strftime"
301 "\7tolower" "\7toupper" NTC
303 "\4func" "\10function" NTC
308 static const uint32_t tokeninfo[] = {
312 xS|'a', xS|'w', xS|'|',
313 OC_UNARY|xV|P(9)|'p', OC_UNARY|xV|P(9)|'m',
314 OC_UNARY|xV|P(9)|'P', OC_UNARY|xV|P(9)|'M',
316 OC_COMPARE|VV|P(39)|5, OC_MOVE|VV|P(74),
317 OC_REPLACE|NV|P(74)|'+', OC_REPLACE|NV|P(74)|'-',
318 OC_REPLACE|NV|P(74)|'*', OC_REPLACE|NV|P(74)|'/',
319 OC_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&',
320 OC_BINARY|NV|P(29)|'+', OC_BINARY|NV|P(29)|'-',
321 OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&',
322 OC_BINARY|NV|P(25)|'/', OC_BINARY|NV|P(25)|'%',
323 OC_BINARY|NV|P(15)|'&', OC_BINARY|NV|P(25)|'*',
324 OC_COMPARE|VV|P(39)|4, OC_COMPARE|VV|P(39)|3,
325 OC_COMPARE|VV|P(39)|0, OC_COMPARE|VV|P(39)|1,
326 OC_COMPARE|VV|P(39)|2, OC_MATCH|Sx|P(45)|'!',
327 OC_MATCH|Sx|P(45)|'~', OC_LAND|Vx|P(55),
328 OC_LOR|Vx|P(59), OC_TERNARY|Vx|P(64)|'?',
329 OC_COLON|xx|P(67)|':',
332 OC_PGETLINE|SV|P(37),
333 OC_UNARY|xV|P(19)|'+', OC_UNARY|xV|P(19)|'-',
334 OC_UNARY|xV|P(19)|'!',
340 ST_IF, ST_DO, ST_FOR, OC_BREAK,
341 OC_CONTINUE, OC_DELETE|Vx, OC_PRINT,
342 OC_PRINTF, OC_NEXT, OC_NEXTFILE,
343 OC_RETURN|Vx, OC_EXIT|Nx,
347 OC_B|B_an|P(0x83), OC_B|B_co|P(0x41), OC_B|B_ls|P(0x83), OC_B|B_or|P(0x83),
348 OC_B|B_rs|P(0x83), OC_B|B_xo|P(0x83),
349 OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83),
350 OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg,
351 OC_FBLTIN|F_rn, OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr,
352 OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), OC_FBLTIN|Sx|F_le,
353 OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF, OC_B|B_su|P(0xb6),
354 OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b),
355 OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49),
362 /* internal variable names and their initial values */
363 /* asterisk marks SPECIAL vars; $ is just no-named Field0 */
365 CONVFMT=0, OFMT, FS, OFS,
366 ORS, RS, RT, FILENAME,
367 SUBSEP, ARGIND, ARGC, ARGV,
370 ENVIRON, F0, _intvarcount_
373 static const char vNames[] =
374 "CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0"
375 "ORS\0" "RS\0*" "RT\0" "FILENAME\0"
376 "SUBSEP\0" "ARGIND\0" "ARGC\0" "ARGV\0"
378 "NR\0" "NF\0*" "IGNORECASE\0*"
379 "ENVIRON\0" "$\0*" "\0";
381 static const char vValues[] =
382 "%.6g\0" "%.6g\0" " \0" " \0"
383 "\n\0" "\n\0" "\0" "\0"
387 /* hash size may grow to these values */
388 #define FIRST_PRIME 61;
389 static const unsigned PRIMES[] = { 251, 1021, 4093, 16381, 65521 };
390 enum { NPRIMES = sizeof(PRIMES) / sizeof(unsigned) };
394 extern char **environ;
396 static var * V[_intvarcount_];
397 static chain beginseq, mainseq, endseq, *seq;
398 static int nextrec, nextfile;
399 static node *break_ptr, *continue_ptr;
401 static xhash *vhash, *ahash, *fdhash, *fnhash;
402 static char *programname;
404 static int is_f0_split;
407 static tsplitter fsplitter, rsplitter;
423 /* function prototypes */
424 static void handle_special(var *);
425 static node *parse_expr(uint32_t);
426 static void chain_group(void);
427 static var *evaluate(node *, var *);
428 static rstream *next_input_file(void);
429 static int fmt_num(char *, int, const char *, double, int);
430 static int awk_exit(int) ATTRIBUTE_NORETURN;
432 /* ---- error handling ---- */
434 static const char EMSG_INTERNAL_ERROR[] = "Internal error";
435 static const char EMSG_UNEXP_EOS[] = "Unexpected end of string";
436 static const char EMSG_UNEXP_TOKEN[] = "Unexpected token";
437 static const char EMSG_DIV_BY_ZERO[] = "Division by zero";
438 static const char EMSG_INV_FMT[] = "Invalid format specifier";
439 static const char EMSG_TOO_FEW_ARGS[] = "Too few arguments for builtin";
440 static const char EMSG_NOT_ARRAY[] = "Not an array";
441 static const char EMSG_POSSIBLE_ERROR[] = "Possible syntax error";
442 static const char EMSG_UNDEF_FUNC[] = "Call to undefined function";
443 #if !ENABLE_FEATURE_AWK_MATH
444 static const char EMSG_NO_MATH[] = "Math support is not compiled in";
447 static void zero_out_var(var * vp)
449 memset(vp, 0, sizeof(*vp));
452 static void syntax_error(const char * const message) ATTRIBUTE_NORETURN;
453 static void syntax_error(const char * const message)
455 bb_error_msg_and_die("%s:%i: %s", programname, lineno, message);
458 #define runtime_error(x) syntax_error(x)
461 /* ---- hash stuff ---- */
463 static unsigned hashidx(const char *name)
467 while (*name) idx = *name++ + (idx << 6) - idx;
471 /* create new hash */
472 static xhash *hash_init(void)
476 newhash = xzalloc(sizeof(xhash));
477 newhash->csize = FIRST_PRIME;
478 newhash->items = xzalloc(newhash->csize * sizeof(hash_item *));
483 /* find item in hash, return ptr to data, NULL if not found */
484 static void *hash_search(xhash *hash, const char *name)
488 hi = hash->items [ hashidx(name) % hash->csize ];
490 if (strcmp(hi->name, name) == 0)
497 /* grow hash if it becomes too big */
498 static void hash_rebuild(xhash *hash)
500 unsigned newsize, i, idx;
501 hash_item **newitems, *hi, *thi;
503 if (hash->nprime == NPRIMES)
506 newsize = PRIMES[hash->nprime++];
507 newitems = xzalloc(newsize * sizeof(hash_item *));
509 for (i=0; i<hash->csize; i++) {
514 idx = hashidx(thi->name) % newsize;
515 thi->next = newitems[idx];
521 hash->csize = newsize;
522 hash->items = newitems;
525 /* find item in hash, add it if necessary. Return ptr to data */
526 static void *hash_find(xhash *hash, const char *name)
532 hi = hash_search(hash, name);
534 if (++hash->nel / hash->csize > 10)
537 l = strlen(name) + 1;
538 hi = xzalloc(sizeof(hash_item) + l);
539 memcpy(hi->name, name, l);
541 idx = hashidx(name) % hash->csize;
542 hi->next = hash->items[idx];
543 hash->items[idx] = hi;
549 #define findvar(hash, name) ((var*) hash_find((hash) , (name)))
550 #define newvar(name) ((var*) hash_find(vhash , (name)))
551 #define newfile(name) ((rstream*)hash_find(fdhash ,(name)))
552 #define newfunc(name) ((func*) hash_find(fnhash , (name)))
554 static void hash_remove(xhash *hash, const char *name)
556 hash_item *hi, **phi;
558 phi = &(hash->items[ hashidx(name) % hash->csize ]);
561 if (strcmp(hi->name, name) == 0) {
562 hash->glen -= (strlen(name) + 1);
572 /* ------ some useful functions ------ */
574 static void skip_spaces(char **s)
578 while (*p == ' ' || *p == '\t' ||
579 (*p == '\\' && *(p+1) == '\n' && (++p, ++t.lineno))) {
585 static char *nextword(char **s)
589 while (*(*s)++) /* */;
594 static char nextchar(char **s)
600 if (c == '\\') c = bb_process_escape_sequence((const char**)s);
601 if (c == '\\' && *s == pps) c = *((*s)++);
605 static int ATTRIBUTE_ALWAYS_INLINE isalnum_(int c)
607 return (isalnum(c) || c == '_');
610 static FILE *afopen(const char *path, const char *mode)
612 return (*path == '-' && *(path+1) == '\0') ? stdin : xfopen(path, mode);
615 /* -------- working with variables (set/get/copy/etc) -------- */
617 static xhash *iamarray(var *v)
621 while (a->type & VF_CHILD)
624 if (! (a->type & VF_ARRAY)) {
626 a->x.array = hash_init();
631 static void clear_array(xhash *array)
636 for (i=0; i<array->csize; i++) {
637 hi = array->items[i];
641 free(thi->data.v.string);
644 array->items[i] = NULL;
646 array->glen = array->nel = 0;
649 /* clear a variable */
650 static var *clrvar(var *v)
652 if (!(v->type & VF_FSTR))
655 v->type &= VF_DONTTOUCH;
661 /* assign string value to variable */
662 static var *setvar_p(var *v, char *value)
671 /* same as setvar_p but make a copy of string */
672 static var *setvar_s(var *v, const char *value)
674 return setvar_p(v, (value && *value) ? xstrdup(value) : NULL);
677 /* same as setvar_s but set USER flag */
678 static var *setvar_u(var *v, const char *value)
685 /* set array element to user string */
686 static void setari_u(var *a, int idx, const char *s)
689 static char sidx[12];
691 sprintf(sidx, "%d", idx);
692 v = findvar(iamarray(a), sidx);
696 /* assign numeric value to variable */
697 static var *setvar_i(var *v, double value)
700 v->type |= VF_NUMBER;
706 static char *getvar_s(var *v)
708 /* if v is numeric and has no cached string, convert it to string */
709 if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) {
710 fmt_num(buf, MAXVARFMT, getvar_s(V[CONVFMT]), v->number, TRUE);
711 v->string = xstrdup(buf);
712 v->type |= VF_CACHED;
714 return (v->string == NULL) ? "" : v->string;
717 static double getvar_i(var *v)
721 if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) {
725 v->number = strtod(s, &s);
726 if (v->type & VF_USER) {
734 v->type |= VF_CACHED;
739 static var *copyvar(var *dest, const var *src)
743 dest->type |= (src->type & ~VF_DONTTOUCH);
744 dest->number = src->number;
746 dest->string = xstrdup(src->string);
748 handle_special(dest);
752 static var *incvar(var *v)
754 return setvar_i(v, getvar_i(v)+1.);
757 /* return true if v is number or numeric string */
758 static int is_numeric(var *v)
761 return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY));
764 /* return 1 when value of v corresponds to true, 0 otherwise */
765 static int istrue(var *v)
768 return (v->number == 0) ? 0 : 1;
770 return (v->string && *(v->string)) ? 1 : 0;
773 /* temporary variables allocator. Last allocated should be first freed */
774 static var *nvalloc(int n)
782 if ((cb->pos - cb->nv) + n <= cb->size) break;
787 size = (n <= MINNVBLOCK) ? MINNVBLOCK : n;
788 cb = xmalloc(sizeof(nvblock) + size * sizeof(var));
793 if (pb) pb->next = cb;
799 while (v < cb->pos) {
808 static void nvfree(var *v)
812 if (v < cb->nv || v >= cb->pos)
813 runtime_error(EMSG_INTERNAL_ERROR);
815 for (p=v; p<cb->pos; p++) {
816 if ((p->type & (VF_ARRAY|VF_CHILD)) == VF_ARRAY) {
817 clear_array(iamarray(p));
818 free(p->x.array->items);
821 if (p->type & VF_WALK)
828 while (cb->prev && cb->pos == cb->nv) {
833 /* ------- awk program text parsing ------- */
835 /* Parse next token pointed by global pos, place results into global t.
836 * If token isn't expected, give away. Return token class
838 static uint32_t next_token(uint32_t expected)
840 static int concat_inserted;
841 static uint32_t save_tclass, save_info;
842 static uint32_t ltclass = TC_OPTERM;
853 } else if (concat_inserted) {
854 concat_inserted = FALSE;
855 t.tclass = save_tclass;
864 while (*p != '\n' && *p != '\0') p++;
872 } else if (*p == '\"') {
876 if (*p == '\0' || *p == '\n')
877 syntax_error(EMSG_UNEXP_EOS);
878 *(s++) = nextchar(&p);
884 } else if ((expected & TC_REGEXP) && *p == '/') {
888 if (*p == '\0' || *p == '\n')
889 syntax_error(EMSG_UNEXP_EOS);
890 if ((*s++ = *p++) == '\\') {
892 *(s-1) = bb_process_escape_sequence((const char **)&p);
893 if (*pp == '\\') *s++ = '\\';
894 if (p == pp) *s++ = *p++;
901 } else if (*p == '.' || isdigit(*p)) {
903 t.number = strtod(p, &p);
905 syntax_error(EMSG_UNEXP_TOKEN);
909 /* search for something known */
919 /* if token class is expected, token
920 * matches and it's not a longer word,
921 * then this is what we are looking for
923 if ((tc & (expected | TC_WORD | TC_NEWLINE)) &&
924 *tl == *p && strncmp(p, tl, l) == 0 &&
925 !((tc & TC_WORD) && isalnum_(*(p + l)))) {
935 /* it's a name (var/array/function),
936 * otherwise it's something wrong
939 syntax_error(EMSG_UNEXP_TOKEN);
942 while (isalnum_(*(++p))) {
947 /* also consume whitespace between functionname and bracket */
948 if (!(expected & TC_VARIABLE)) skip_spaces(&p);
961 /* skipping newlines in some cases */
962 if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE))
965 /* insert concatenation operator when needed */
966 if ((ltclass&TC_CONCAT1) && (tc&TC_CONCAT2) && (expected&TC_BINOP)) {
967 concat_inserted = TRUE;
971 t.info = OC_CONCAT | SS | P(35);
978 /* Are we ready for this? */
979 if (! (ltclass & expected))
980 syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ?
981 EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN);
986 static void rollback_token(void) { t.rollback = TRUE; }
988 static node *new_node(uint32_t info)
992 n = xzalloc(sizeof(node));
998 static node *mk_re_node(char *s, node *n, regex_t *re)
1000 n->info = OC_REGEXP;
1003 xregcomp(re, s, REG_EXTENDED);
1004 xregcomp(re+1, s, REG_EXTENDED | REG_ICASE);
1009 static node *condition(void)
1011 next_token(TC_SEQSTART);
1012 return parse_expr(TC_SEQTERM);
1015 /* parse expression terminated by given argument, return ptr
1016 * to built subtree. Terminator is eaten by parse_expr */
1017 static node *parse_expr(uint32_t iexp)
1026 sn.r.n = glptr = NULL;
1027 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp;
1029 while (! ((tc = next_token(xtc)) & iexp)) {
1030 if (glptr && (t.info == (OC_COMPARE|VV|P(39)|2))) {
1031 /* input redirection (<) attached to glptr node */
1032 cn = glptr->l.n = new_node(OC_CONCAT|SS|P(37));
1034 xtc = TC_OPERAND | TC_UOPPRE;
1037 } else if (tc & (TC_BINOP | TC_UOPPOST)) {
1038 /* for binary and postfix-unary operators, jump back over
1039 * previous operators with higher priority */
1041 while ( ((t.info & PRIMASK) > (vn->a.n->info & PRIMASK2)) ||
1042 ((t.info == vn->info) && ((t.info & OPCLSMASK) == OC_COLON)) )
1044 if ((t.info & OPCLSMASK) == OC_TERNARY)
1046 cn = vn->a.n->r.n = new_node(t.info);
1048 if (tc & TC_BINOP) {
1050 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
1051 if ((t.info & OPCLSMASK) == OC_PGETLINE) {
1053 next_token(TC_GETLINE);
1054 /* give maximum priority to this pipe */
1055 cn->info &= ~PRIMASK;
1056 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1060 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1065 /* for operands and prefix-unary operators, attach them
1068 cn = vn->r.n = new_node(t.info);
1070 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
1071 if (tc & (TC_OPERAND | TC_REGEXP)) {
1072 xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp;
1073 /* one should be very careful with switch on tclass -
1074 * only simple tclasses should be used! */
1079 if ((v = hash_search(ahash, t.string)) != NULL) {
1080 cn->info = OC_FNARG;
1081 cn->l.i = v->x.aidx;
1083 cn->l.v = newvar(t.string);
1085 if (tc & TC_ARRAY) {
1087 cn->r.n = parse_expr(TC_ARRTERM);
1094 v = cn->l.v = xzalloc(sizeof(var));
1096 setvar_i(v, t.number);
1098 setvar_s(v, t.string);
1102 mk_re_node(t.string, cn, xzalloc(sizeof(regex_t)*2));
1107 cn->r.f = newfunc(t.string);
1108 cn->l.n = condition();
1112 cn = vn->r.n = parse_expr(TC_SEQTERM);
1118 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1122 cn->l.n = condition();
1131 /* add node to chain. Return ptr to alloc'd node */
1132 static node *chain_node(uint32_t info)
1137 seq->first = seq->last = new_node(0);
1139 if (seq->programname != programname) {
1140 seq->programname = programname;
1141 n = chain_node(OC_NEWSOURCE);
1142 n->l.s = xstrdup(programname);
1147 seq->last = n->a.n = new_node(OC_DONE);
1152 static void chain_expr(uint32_t info)
1156 n = chain_node(info);
1157 n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM);
1158 if (t.tclass & TC_GRPTERM)
1162 static node *chain_loop(node *nn)
1164 node *n, *n2, *save_brk, *save_cont;
1166 save_brk = break_ptr;
1167 save_cont = continue_ptr;
1169 n = chain_node(OC_BR | Vx);
1170 continue_ptr = new_node(OC_EXEC);
1171 break_ptr = new_node(OC_EXEC);
1173 n2 = chain_node(OC_EXEC | Vx);
1176 continue_ptr->a.n = n2;
1177 break_ptr->a.n = n->r.n = seq->last;
1179 continue_ptr = save_cont;
1180 break_ptr = save_brk;
1185 /* parse group and attach it to chain */
1186 static void chain_group(void)
1192 c = next_token(TC_GRPSEQ);
1193 } while (c & TC_NEWLINE);
1195 if (c & TC_GRPSTART) {
1196 while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) {
1197 if (t.tclass & TC_NEWLINE) continue;
1201 } else if (c & (TC_OPSEQ | TC_OPTERM)) {
1203 chain_expr(OC_EXEC | Vx);
1204 } else { /* TC_STATEMNT */
1205 switch (t.info & OPCLSMASK) {
1207 n = chain_node(OC_BR | Vx);
1208 n->l.n = condition();
1210 n2 = chain_node(OC_EXEC);
1212 if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE)==TC_ELSE) {
1214 n2->a.n = seq->last;
1222 n = chain_loop(NULL);
1227 n2 = chain_node(OC_EXEC);
1228 n = chain_loop(NULL);
1230 next_token(TC_WHILE);
1231 n->l.n = condition();
1235 next_token(TC_SEQSTART);
1236 n2 = parse_expr(TC_SEMICOL | TC_SEQTERM);
1237 if (t.tclass & TC_SEQTERM) { /* for-in */
1238 if ((n2->info & OPCLSMASK) != OC_IN)
1239 syntax_error(EMSG_UNEXP_TOKEN);
1240 n = chain_node(OC_WALKINIT | VV);
1243 n = chain_loop(NULL);
1244 n->info = OC_WALKNEXT | Vx;
1246 } else { /* for (;;) */
1247 n = chain_node(OC_EXEC | Vx);
1249 n2 = parse_expr(TC_SEMICOL);
1250 n3 = parse_expr(TC_SEQTERM);
1260 n = chain_node(t.info);
1261 n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM);
1262 if (t.tclass & TC_OUTRDR) {
1264 n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM);
1266 if (t.tclass & TC_GRPTERM)
1271 n = chain_node(OC_EXEC);
1276 n = chain_node(OC_EXEC);
1277 n->a.n = continue_ptr;
1280 /* delete, next, nextfile, return, exit */
1287 static void parse_program(char *p)
1296 while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART |
1297 TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) {
1299 if (tclass & TC_OPTERM)
1303 if (tclass & TC_BEGIN) {
1307 } else if (tclass & TC_END) {
1311 } else if (tclass & TC_FUNCDECL) {
1312 next_token(TC_FUNCTION);
1314 f = newfunc(t.string);
1315 f->body.first = NULL;
1317 while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) {
1318 v = findvar(ahash, t.string);
1319 v->x.aidx = (f->nargs)++;
1321 if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM)
1328 } else if (tclass & TC_OPSEQ) {
1330 cn = chain_node(OC_TEST);
1331 cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART);
1332 if (t.tclass & TC_GRPSTART) {
1336 chain_node(OC_PRINT);
1338 cn->r.n = mainseq.last;
1340 } else /* if (tclass & TC_GRPSTART) */ {
1348 /* -------- program execution part -------- */
1350 static node *mk_splitter(char *s, tsplitter *spl)
1358 if ((n->info & OPCLSMASK) == OC_REGEXP) {
1362 if (strlen(s) > 1) {
1363 mk_re_node(s, n, re);
1365 n->info = (uint32_t) *s;
1371 /* use node as a regular expression. Supplied with node ptr and regex_t
1372 * storage space. Return ptr to regex (if result points to preg, it should
1373 * be later regfree'd manually
1375 static regex_t *as_regex(node *op, regex_t *preg)
1380 if ((op->info & OPCLSMASK) == OC_REGEXP) {
1381 return icase ? op->r.ire : op->l.re;
1384 s = getvar_s(evaluate(op, v));
1385 xregcomp(preg, s, icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED);
1391 /* gradually increasing buffer */
1392 static void qrealloc(char **b, int n, int *size)
1394 if (!*b || n >= *size)
1395 *b = xrealloc(*b, *size = n + (n>>1) + 80);
1398 /* resize field storage space */
1399 static void fsrealloc(int size)
1401 static int maxfields; /* = 0;*/
1404 if (size >= maxfields) {
1406 maxfields = size + 16;
1407 Fields = xrealloc(Fields, maxfields * sizeof(var));
1408 for (; i < maxfields; i++) {
1409 Fields[i].type = VF_SPECIAL;
1410 Fields[i].string = NULL;
1414 if (size < nfields) {
1415 for (i = size; i < nfields; i++) {
1422 static int awk_split(char *s, node *spl, char **slist)
1427 regmatch_t pmatch[2];
1429 /* in worst case, each char would be a separate field */
1430 *slist = s1 = xstrndup(s, strlen(s) * 2 + 3);
1432 c[0] = c[1] = (char)spl->info;
1434 if (*getvar_s(V[RS]) == '\0') c[2] = '\n';
1436 if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */
1438 l = strcspn(s, c+2);
1439 if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0 &&
1440 pmatch[0].rm_so <= l) {
1441 l = pmatch[0].rm_so;
1442 if (pmatch[0].rm_eo == 0) { l++; pmatch[0].rm_eo++; }
1444 pmatch[0].rm_eo = l;
1445 if (s[l]) pmatch[0].rm_eo++;
1451 s += pmatch[0].rm_eo;
1454 } else if (c[0] == '\0') { /* null split */
1460 } else if (c[0] != ' ') { /* single-character split */
1462 c[0] = toupper(c[0]);
1463 c[1] = tolower(c[1]);
1466 while ((s1 = strpbrk(s1, c))) {
1470 } else { /* space split */
1472 s = skip_whitespace(s);
1475 while (*s && !isspace(*s))
1483 static void split_f0(void)
1485 static char *fstrings = NULL;
1495 n = awk_split(getvar_s(V[F0]), &fsplitter.n, &fstrings);
1498 for (i = 0; i < n; i++) {
1499 Fields[i].string = nextword(&s);
1500 Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY);
1503 /* set NF manually to avoid side effects */
1505 V[NF]->type = VF_NUMBER | VF_SPECIAL;
1506 V[NF]->number = nfields;
1509 /* perform additional actions when some internal variables changed */
1510 static void handle_special(var *v)
1514 int sl, l, len, i, bsize;
1516 if (!(v->type & VF_SPECIAL))
1520 n = (int)getvar_i(v);
1523 /* recalculate $0 */
1524 sep = getvar_s(V[OFS]);
1528 for (i=0; i<n; i++) {
1529 s = getvar_s(&Fields[i]);
1532 memcpy(b+len, sep, sl);
1535 qrealloc(&b, len+l+sl, &bsize);
1536 memcpy(b+len, s, l);
1544 } else if (v == V[F0]) {
1545 is_f0_split = FALSE;
1547 } else if (v == V[FS]) {
1548 mk_splitter(getvar_s(v), &fsplitter);
1550 } else if (v == V[RS]) {
1551 mk_splitter(getvar_s(v), &rsplitter);
1553 } else if (v == V[IGNORECASE]) {
1557 n = getvar_i(V[NF]);
1558 setvar_i(V[NF], n > v-Fields ? n : v-Fields+1);
1559 /* right here v is invalid. Just to note... */
1563 /* step through func/builtin/etc arguments */
1564 static node *nextarg(node **pn)
1569 if (n && (n->info & OPCLSMASK) == OC_COMMA) {
1578 static void hashwalk_init(var *v, xhash *array)
1584 if (v->type & VF_WALK)
1588 w = v->x.walker = xzalloc(2 + 2*sizeof(char *) + array->glen);
1589 *w = *(w+1) = (char *)(w + 2);
1590 for (i=0; i<array->csize; i++) {
1591 hi = array->items[i];
1593 strcpy(*w, hi->name);
1600 static int hashwalk_next(var *v)
1608 setvar_s(v, nextword(w+1));
1612 /* evaluate node, return 1 when result is true, 0 otherwise */
1613 static int ptest(node *pattern)
1615 static var v; /* static: to save stack space? */
1617 return istrue(evaluate(pattern, &v));
1620 /* read next record from stream rsm into a variable v */
1621 static int awk_getline(rstream *rsm, var *v)
1624 regmatch_t pmatch[2];
1625 int a, p, pp=0, size;
1626 int fd, so, eo, r, rp;
1629 /* we're using our own buffer since we need access to accumulating
1632 fd = fileno(rsm->F);
1637 c = (char) rsplitter.n.info;
1640 if (! m) qrealloc(&m, 256, &size);
1646 if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) {
1647 if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re,
1648 b, 1, pmatch, 0) == 0) {
1649 so = pmatch[0].rm_so;
1650 eo = pmatch[0].rm_eo;
1654 } else if (c != '\0') {
1655 s = strchr(b+pp, c);
1656 if (! s) s = memchr(b+pp, '\0', p - pp);
1663 while (b[rp] == '\n')
1665 s = strstr(b+rp, "\n\n");
1668 while (b[eo] == '\n') eo++;
1676 memmove(m, (const void *)(m+a), p+1);
1681 qrealloc(&m, a+p+128, &size);
1684 p += safe_read(fd, b+p, size-p-1);
1688 setvar_i(V[ERRNO], errno);
1697 c = b[so]; b[so] = '\0';
1701 c = b[eo]; b[eo] = '\0';
1702 setvar_s(V[RT], b+so);
1714 static int fmt_num(char *b, int size, const char *format, double n, int int_as_int)
1718 const char *s = format;
1720 if (int_as_int && n == (int)n) {
1721 r = snprintf(b, size, "%d", (int)n);
1723 do { c = *s; } while (c && *++s);
1724 if (strchr("diouxX", c)) {
1725 r = snprintf(b, size, format, (int)n);
1726 } else if (strchr("eEfgG", c)) {
1727 r = snprintf(b, size, format, n);
1729 runtime_error(EMSG_INV_FMT);
1736 /* formatted output into an allocated buffer, return ptr to buffer */
1737 static char *awk_printf(node *n)
1740 char *fmt, *s, *s1, *f;
1741 int i, j, incr, bsize;
1746 fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v)));
1751 while (*f && (*f != '%' || *(++f) == '%'))
1753 while (*f && !isalpha(*f))
1756 incr = (f - s) + MAXVARFMT;
1757 qrealloc(&b, incr + i, &bsize);
1762 arg = evaluate(nextarg(&n), v);
1765 if (c == 'c' || !c) {
1766 i += sprintf(b+i, s, is_numeric(arg) ?
1767 (char)getvar_i(arg) : *getvar_s(arg));
1769 } else if (c == 's') {
1771 qrealloc(&b, incr+i+strlen(s1), &bsize);
1772 i += sprintf(b+i, s, s1);
1775 i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE);
1779 /* if there was an error while sprintf, return value is negative */
1783 b = xrealloc(b, i + 1);
1790 /* common substitution routine
1791 * replace (nm) substring of (src) that match (n) with (repl), store
1792 * result into (dest), return number of substitutions. If nm=0, replace
1793 * all matches. If src or dst is NULL, use $0. If ex=TRUE, enable
1794 * subexpression matching (\1-\9)
1796 static int awk_sub(node *rn, char *repl, int nm, var *src, var *dest, int ex)
1800 int c, i, j, di, rl, so, eo, nbs, n, dssize;
1801 regmatch_t pmatch[10];
1804 re = as_regex(rn, &sreg);
1805 if (! src) src = V[F0];
1806 if (! dest) dest = V[F0];
1811 while (regexec(re, sp, 10, pmatch, sp==getvar_s(src) ? 0:REG_NOTBOL) == 0) {
1812 so = pmatch[0].rm_so;
1813 eo = pmatch[0].rm_eo;
1815 qrealloc(&ds, di + eo + rl, &dssize);
1816 memcpy(ds + di, sp, eo);
1822 for (s = repl; *s; s++) {
1828 if (c == '&' || (ex && c >= '0' && c <= '9')) {
1829 di -= ((nbs + 3) >> 1);
1838 n = pmatch[j].rm_eo - pmatch[j].rm_so;
1839 qrealloc(&ds, di + rl + n, &dssize);
1840 memcpy(ds + di, sp + pmatch[j].rm_so, n);
1851 if (! (ds[di++] = *sp++)) break;
1855 qrealloc(&ds, di + strlen(sp), &dssize);
1856 strcpy(ds + di, sp);
1858 if (re == &sreg) regfree(re);
1862 static var *exec_builtin(node *op, var *res)
1869 regmatch_t pmatch[2];
1871 static tsplitter tspl;
1880 isr = info = op->info;
1883 av[2] = av[3] = NULL;
1884 for (i=0 ; i<4 && op ; i++) {
1885 an[i] = nextarg(&op);
1886 if (isr & 0x09000000) av[i] = evaluate(an[i], &tv[i]);
1887 if (isr & 0x08000000) as[i] = getvar_s(av[i]);
1892 if (nargs < (info >> 30))
1893 runtime_error(EMSG_TOO_FEW_ARGS);
1895 switch (info & OPNMASK) {
1898 #if ENABLE_FEATURE_AWK_MATH
1899 setvar_i(res, atan2(getvar_i(av[i]), getvar_i(av[1])));
1901 runtime_error(EMSG_NO_MATH);
1907 spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ?
1908 an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl);
1913 n = awk_split(as[0], spl, &s);
1915 clear_array(iamarray(av[1]));
1916 for (i=1; i<=n; i++)
1917 setari_u(av[1], i, nextword(&s1));
1924 i = getvar_i(av[1]) - 1;
1925 if (i>l) i=l; if (i<0) i=0;
1926 n = (nargs > 2) ? getvar_i(av[2]) : l-i;
1929 strncpy(s, as[0]+i, n);
1935 setvar_i(res, (long)getvar_i(av[0]) & (long)getvar_i(av[1]));
1939 setvar_i(res, ~(long)getvar_i(av[0]));
1943 setvar_i(res, (long)getvar_i(av[0]) << (long)getvar_i(av[1]));
1947 setvar_i(res, (long)getvar_i(av[0]) | (long)getvar_i(av[1]));
1951 setvar_i(res, (long)((unsigned long)getvar_i(av[0]) >> (unsigned long)getvar_i(av[1])));
1955 setvar_i(res, (long)getvar_i(av[0]) ^ (long)getvar_i(av[1]));
1965 s1 = s = xstrdup(as[0]);
1967 *s1 = (*to_xxx)(*s1);
1976 l = strlen(as[0]) - ll;
1977 if (ll > 0 && l >= 0) {
1979 s = strstr(as[0], as[1]);
1980 if (s) n = (s - as[0]) + 1;
1982 /* this piece of code is terribly slow and
1983 * really should be rewritten
1985 for (i=0; i<=l; i++) {
1986 if (strncasecmp(as[0]+i, as[1], ll) == 0) {
1998 tt = getvar_i(av[1]);
2001 s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y";
2002 i = strftime(buf, MAXVARFMT, s, localtime(&tt));
2008 re = as_regex(an[1], &sreg);
2009 n = regexec(re, as[0], 1, pmatch, 0);
2014 pmatch[0].rm_so = 0;
2015 pmatch[0].rm_eo = -1;
2017 setvar_i(newvar("RSTART"), pmatch[0].rm_so);
2018 setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so);
2019 setvar_i(res, pmatch[0].rm_so);
2020 if (re == &sreg) regfree(re);
2024 awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE);
2028 setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE));
2032 setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE));
2041 * Evaluate node - the heart of the program. Supplied with subtree
2042 * and place where to store result. returns ptr to result.
2044 #define XC(n) ((n) >> 8)
2046 static var *evaluate(node *op, var *res)
2048 /* This procedure is recursive so we should count every byte */
2049 static var *fnargs = NULL;
2050 static unsigned seed = 1;
2051 static regex_t sreg;
2073 return setvar_s(res, NULL);
2080 opn = (short)(opinfo & OPNMASK);
2081 lineno = op->lineno;
2083 /* execute inevitable things */
2085 if (opinfo & OF_RES1) X.v = L.v = evaluate(op1, v1);
2086 if (opinfo & OF_RES2) R.v = evaluate(op->r.n, v1+1);
2087 if (opinfo & OF_STR1) L.s = getvar_s(L.v);
2088 if (opinfo & OF_STR2) R.s = getvar_s(R.v);
2089 if (opinfo & OF_NUM1) L.d = getvar_i(L.v);
2091 switch (XC(opinfo & OPCLSMASK)) {
2093 /* -- iterative node type -- */
2097 if ((op1->info & OPCLSMASK) == OC_COMMA) {
2098 /* it's range pattern */
2099 if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) {
2100 op->info |= OF_CHECKED;
2101 if (ptest(op1->r.n))
2102 op->info &= ~OF_CHECKED;
2109 op = (ptest(op1)) ? op->a.n : op->r.n;
2113 /* just evaluate an expression, also used as unconditional jump */
2117 /* branch, used in if-else and various loops */
2119 op = istrue(L.v) ? op->a.n : op->r.n;
2122 /* initialize for-in loop */
2123 case XC( OC_WALKINIT ):
2124 hashwalk_init(L.v, iamarray(R.v));
2127 /* get next array item */
2128 case XC( OC_WALKNEXT ):
2129 op = hashwalk_next(L.v) ? op->a.n : op->r.n;
2132 case XC( OC_PRINT ):
2133 case XC( OC_PRINTF ):
2136 X.rsm = newfile(R.s);
2139 if((X.rsm->F = popen(R.s, "w")) == NULL)
2140 bb_perror_msg_and_die("popen");
2143 X.rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
2149 if ((opinfo & OPCLSMASK) == OC_PRINT) {
2151 fputs(getvar_s(V[F0]), X.F);
2154 L.v = evaluate(nextarg(&op1), v1);
2155 if (L.v->type & VF_NUMBER) {
2156 fmt_num(buf, MAXVARFMT, getvar_s(V[OFMT]),
2157 getvar_i(L.v), TRUE);
2160 fputs(getvar_s(L.v), X.F);
2163 if (op1) fputs(getvar_s(V[OFS]), X.F);
2166 fputs(getvar_s(V[ORS]), X.F);
2168 } else { /* OC_PRINTF */
2169 L.s = awk_printf(op1);
2176 case XC( OC_DELETE ):
2177 X.info = op1->info & OPCLSMASK;
2178 if (X.info == OC_VAR) {
2180 } else if (X.info == OC_FNARG) {
2181 R.v = &fnargs[op1->l.i];
2183 runtime_error(EMSG_NOT_ARRAY);
2188 L.s = getvar_s(evaluate(op1->r.n, v1));
2189 hash_remove(iamarray(R.v), L.s);
2191 clear_array(iamarray(R.v));
2195 case XC( OC_NEWSOURCE ):
2196 programname = op->l.s;
2199 case XC( OC_RETURN ):
2203 case XC( OC_NEXTFILE ):
2214 /* -- recursive node type -- */
2222 case XC( OC_FNARG ):
2223 L.v = &fnargs[op->l.i];
2225 res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v;
2229 setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0);
2232 case XC( OC_REGEXP ):
2234 L.s = getvar_s(V[F0]);
2237 case XC( OC_MATCH ):
2240 X.re = as_regex(op1, &sreg);
2241 R.i = regexec(X.re, L.s, 0, NULL, 0);
2242 if (X.re == &sreg) regfree(X.re);
2243 setvar_i(res, (R.i == 0 ? 1 : 0) ^ (opn == '!' ? 1 : 0));
2247 /* if source is a temporary string, jusk relink it to dest */
2248 if (R.v == v1+1 && R.v->string) {
2249 res = setvar_p(L.v, R.v->string);
2252 res = copyvar(L.v, R.v);
2256 case XC( OC_TERNARY ):
2257 if ((op->r.n->info & OPCLSMASK) != OC_COLON)
2258 runtime_error(EMSG_POSSIBLE_ERROR);
2259 res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res);
2263 if (! op->r.f->body.first)
2264 runtime_error(EMSG_UNDEF_FUNC);
2266 X.v = R.v = nvalloc(op->r.f->nargs+1);
2268 L.v = evaluate(nextarg(&op1), v1);
2270 R.v->type |= VF_CHILD;
2271 R.v->x.parent = L.v;
2272 if (++R.v - X.v >= op->r.f->nargs)
2280 res = evaluate(op->r.f->body.first, res);
2287 case XC( OC_GETLINE ):
2288 case XC( OC_PGETLINE ):
2290 X.rsm = newfile(L.s);
2292 if ((opinfo & OPCLSMASK) == OC_PGETLINE) {
2293 X.rsm->F = popen(L.s, "r");
2294 X.rsm->is_pipe = TRUE;
2296 X.rsm->F = fopen(L.s, "r"); /* not xfopen! */
2300 if (! iF) iF = next_input_file();
2305 setvar_i(V[ERRNO], errno);
2313 L.i = awk_getline(X.rsm, R.v);
2323 /* simple builtins */
2324 case XC( OC_FBLTIN ):
2332 R.d = (double)rand() / (double)RAND_MAX;
2335 #if ENABLE_FEATURE_AWK_MATH
2361 runtime_error(EMSG_NO_MATH);
2367 seed = op1 ? (unsigned)L.d : (unsigned)time(NULL);
2377 L.s = getvar_s(V[F0]);
2383 R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
2384 ? (system(L.s) >> 8) : 0;
2392 X.rsm = newfile(L.s);
2401 X.rsm = (rstream *)hash_search(fdhash, L.s);
2403 R.i = X.rsm->is_pipe ? pclose(X.rsm->F) : fclose(X.rsm->F);
2404 free(X.rsm->buffer);
2405 hash_remove(fdhash, L.s);
2408 setvar_i(V[ERRNO], errno);
2415 case XC( OC_BUILTIN ):
2416 res = exec_builtin(op, res);
2419 case XC( OC_SPRINTF ):
2420 setvar_p(res, awk_printf(op1));
2423 case XC( OC_UNARY ):
2425 L.d = R.d = getvar_i(R.v);
2440 L.d = istrue(X.v) ? 0 : 1;
2451 case XC( OC_FIELD ):
2452 R.i = (int)getvar_i(R.v);
2460 res = &Fields[R.i-1];
2464 /* concatenation (" ") and index joining (",") */
2465 case XC( OC_CONCAT ):
2466 case XC( OC_COMMA ):
2467 opn = strlen(L.s) + strlen(R.s) + 2;
2470 if ((opinfo & OPCLSMASK) == OC_COMMA) {
2471 L.s = getvar_s(V[SUBSEP]);
2472 X.s = xrealloc(X.s, opn + strlen(L.s));
2480 setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0);
2484 setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n));
2487 case XC( OC_BINARY ):
2488 case XC( OC_REPLACE ):
2489 R.d = getvar_i(R.v);
2501 if (R.d == 0) runtime_error(EMSG_DIV_BY_ZERO);
2505 #if ENABLE_FEATURE_AWK_MATH
2506 L.d = pow(L.d, R.d);
2508 runtime_error(EMSG_NO_MATH);
2512 if (R.d == 0) runtime_error(EMSG_DIV_BY_ZERO);
2513 L.d -= (int)(L.d / R.d) * R.d;
2516 res = setvar_i(((opinfo&OPCLSMASK) == OC_BINARY) ? res : X.v, L.d);
2519 case XC( OC_COMPARE ):
2520 if (is_numeric(L.v) && is_numeric(R.v)) {
2521 L.d = getvar_i(L.v) - getvar_i(R.v);
2523 L.s = getvar_s(L.v);
2524 R.s = getvar_s(R.v);
2525 L.d = icase ? strcasecmp(L.s, R.s) : strcmp(L.s, R.s);
2527 switch (opn & 0xfe) {
2538 setvar_i(res, (opn & 0x1 ? R.i : !R.i) ? 1 : 0);
2542 runtime_error(EMSG_POSSIBLE_ERROR);
2544 if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS)
2546 if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS)
2556 /* -------- main & co. -------- */
2558 static int awk_exit(int r)
2569 evaluate(endseq.first, &tv);
2572 /* waiting for children */
2573 for (i = 0; i < fdhash->csize; i++) {
2574 hi = fdhash->items[i];
2576 if (hi->data.rs.F && hi->data.rs.is_pipe)
2577 pclose(hi->data.rs.F);
2585 /* if expr looks like "var=value", perform assignment and return 1,
2586 * otherwise return 0 */
2587 static int is_assignment(const char *expr)
2589 char *exprc, *s, *s0, *s1;
2591 exprc = xstrdup(expr);
2592 if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) {
2600 *(s1++) = nextchar(&s);
2603 setvar_u(newvar(exprc), s0);
2608 /* switch to next input file */
2609 static rstream *next_input_file(void)
2614 static int files_happen = FALSE;
2616 if (rsm.F) fclose(rsm.F);
2618 rsm.pos = rsm.adv = 0;
2621 if (getvar_i(V[ARGIND])+1 >= getvar_i(V[ARGC])) {
2627 ind = getvar_s(incvar(V[ARGIND]));
2628 fname = getvar_s(findvar(iamarray(V[ARGV]), ind));
2629 if (fname && *fname && !is_assignment(fname))
2630 F = afopen(fname, "r");
2634 files_happen = TRUE;
2635 setvar_s(V[FILENAME], fname);
2640 int awk_main(int argc, char **argv)
2643 char *opt_F, *opt_v, *opt_W;
2648 char *vnames = (char *)vNames; /* cheat */
2649 char *vvalues = (char *)vValues;
2651 /* Undo busybox.c, or else strtod may eat ','! This breaks parsing:
2652 * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */
2653 if (ENABLE_LOCALE_SUPPORT)
2654 setlocale(LC_NUMERIC, "C");
2658 /* allocate global buffer */
2659 buf = xmalloc(MAXVARFMT + 1);
2661 vhash = hash_init();
2662 ahash = hash_init();
2663 fdhash = hash_init();
2664 fnhash = hash_init();
2666 /* initialize variables */
2667 for (i = 0; *vnames; i++) {
2668 V[i] = v = newvar(nextword(&vnames));
2669 if (*vvalues != '\377')
2670 setvar_s(v, nextword(&vvalues));
2674 if (*vnames == '*') {
2675 v->type |= VF_SPECIAL;
2680 handle_special(V[FS]);
2681 handle_special(V[RS]);
2683 newfile("/dev/stdin")->F = stdin;
2684 newfile("/dev/stdout")->F = stdout;
2685 newfile("/dev/stderr")->F = stderr;
2687 for (envp = environ; *envp; envp++) {
2688 char *s = xstrdup(*envp);
2689 char *s1 = strchr(s, '=');
2692 setvar_u(findvar(iamarray(V[ENVIRON]), s), s1);
2697 opt = getopt32(argc, argv, "F:v:f:W:", &opt_F, &opt_v, &programname, &opt_W);
2700 if (opt & 0x1) setvar_s(V[FS], opt_F); // -F
2701 if (opt & 0x2) if (!is_assignment(opt_v)) bb_show_usage(); // -v
2702 if (opt & 0x4) { // -f
2703 char *s = s; /* die, gcc, die */
2704 FILE *from_file = afopen(programname, "r");
2705 /* one byte is reserved for some trick in next_token */
2706 if (fseek(from_file, 0, SEEK_END) == 0) {
2707 flen = ftell(from_file);
2708 s = xmalloc(flen + 4);
2709 fseek(from_file, 0, SEEK_SET);
2710 i = 1 + fread(s + 1, 1, flen, from_file);
2712 for (i = j = 1; j > 0; i += j) {
2713 s = xrealloc(s, i + 4096);
2714 j = fread(s + i, 1, 4094, from_file);
2719 parse_program(s + 1);
2721 } else { // no -f: take program from 1st parameter
2724 programname = "cmd. line";
2725 parse_program(*argv++);
2728 if (opt & 0x8) // -W
2729 bb_error_msg("warning: unrecognized option '-W %s' ignored", opt_W);
2731 /* fill in ARGV array */
2732 setvar_i(V[ARGC], argc + 1);
2733 setari_u(V[ARGV], 0, "awk");
2736 setari_u(V[ARGV], ++i, *argv++);
2738 evaluate(beginseq.first, &tv);
2739 if (!mainseq.first && !endseq.first)
2740 awk_exit(EXIT_SUCCESS);
2742 /* input file could already be opened in BEGIN block */
2743 if (!iF) iF = next_input_file();
2745 /* passing through input files */
2748 setvar_i(V[FNR], 0);
2750 while ((i = awk_getline(iF, V[F0])) > 0) {
2754 evaluate(mainseq.first, &tv);
2761 runtime_error(strerror(errno));
2763 iF = next_input_file();
2766 awk_exit(EXIT_SUCCESS);