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