editors/awk.c

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