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 *const message) NORETURN;
 525 static void syntax_error(const char *const 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]) pmatch[0].rm_eo++;
1575                         }
1576                         memcpy(s1, s, l);
1577                         s1[l] = '\0';
1578                         nextword(&s1);
1579                         s += pmatch[0].rm_eo;
1580                 } while (*s);
1581                 return n;
1582         }
1583         if (c[0] == '\0') {  /* null split */
1584                 while (*s) {
1585                         *s1++ = *s++;
1586                         *s1++ = '\0';
1587                         n++;
1588                 }
1589                 return n;
1590         }
1591         if (c[0] != ' ') {  /* single-character split */
1592                 if (icase) {
1593                         c[0] = toupper(c[0]);
1594                         c[1] = tolower(c[1]);
1595                 }
1596                 if (*s1) n++;
1597                 while ((s1 = strpbrk(s1, c))) {
1598                         *s1++ = '\0';
1599                         n++;
1600                 }
1601                 return n;
1602         }
1603         /* space split */
1604         while (*s) {
1605                 s = skip_whitespace(s);
1606                 if (!*s) break;
1607                 n++;
1608                 while (*s && !isspace(*s))
1609                         *s1++ = *s++;
1610                 *s1++ = '\0';
1611         }
1612         return n;
1613 }
1614
1615 static void split_f0(void)
1616 {
1617 /* static char *fstrings; */
1618 #define fstrings (G.split_f0__fstrings)
1619
1620         int i, n;
1621         char *s;
1622
1623         if (is_f0_split)
1624                 return;
1625
1626         is_f0_split = TRUE;
1627         free(fstrings);
1628         fsrealloc(0);
1629         n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings);
1630         fsrealloc(n);
1631         s = fstrings;
1632         for (i = 0; i < n; i++) {
1633                 Fields[i].string = nextword(&s);
1634                 Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY);
1635         }
1636
1637         /* set NF manually to avoid side effects */
1638         clrvar(intvar[NF]);
1639         intvar[NF]->type = VF_NUMBER | VF_SPECIAL;
1640         intvar[NF]->number = nfields;
1641 #undef fstrings
1642 }
1643
1644 /* perform additional actions when some internal variables changed */
1645 static void handle_special(var *v)
1646 {
1647         int n;
1648         char *b;
1649         const char *sep, *s;
1650         int sl, l, len, i, bsize;
1651
1652         if (!(v->type & VF_SPECIAL))
1653                 return;
1654
1655         if (v == intvar[NF]) {
1656                 n = (int)getvar_i(v);
1657                 fsrealloc(n);
1658
1659                 /* recalculate $0 */
1660                 sep = getvar_s(intvar[OFS]);
1661                 sl = strlen(sep);
1662                 b = NULL;
1663                 len = 0;
1664                 for (i = 0; i < n; i++) {
1665                         s = getvar_s(&Fields[i]);
1666                         l = strlen(s);
1667                         if (b) {
1668                                 memcpy(b+len, sep, sl);
1669                                 len += sl;
1670                         }
1671                         qrealloc(&b, len+l+sl, &bsize);
1672                         memcpy(b+len, s, l);
1673                         len += l;
1674                 }
1675                 if (b)
1676                         b[len] = '\0';
1677                 setvar_p(intvar[F0], b);
1678                 is_f0_split = TRUE;
1679
1680         } else if (v == intvar[F0]) {
1681                 is_f0_split = FALSE;
1682
1683         } else if (v == intvar[FS]) {
1684                 mk_splitter(getvar_s(v), &fsplitter);
1685
1686         } else if (v == intvar[RS]) {
1687                 mk_splitter(getvar_s(v), &rsplitter);
1688
1689         } else if (v == intvar[IGNORECASE]) {
1690                 icase = istrue(v);
1691
1692         } else {                                /* $n */
1693                 n = getvar_i(intvar[NF]);
1694                 setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1);
1695                 /* right here v is invalid. Just to note... */
1696         }
1697 }
1698
1699 /* step through func/builtin/etc arguments */
1700 static node *nextarg(node **pn)
1701 {
1702         node *n;
1703
1704         n = *pn;
1705         if (n && (n->info & OPCLSMASK) == OC_COMMA) {
1706                 *pn = n->r.n;
1707                 n = n->l.n;
1708         } else {
1709                 *pn = NULL;
1710         }
1711         return n;
1712 }
1713
1714 static void hashwalk_init(var *v, xhash *array)
1715 {
1716         char **w;
1717         hash_item *hi;
1718         unsigned i;
1719
1720         if (v->type & VF_WALK)
1721                 free(v->x.walker);
1722
1723         v->type |= VF_WALK;
1724         w = v->x.walker = xzalloc(2 + 2*sizeof(char *) + array->glen);
1725         w[0] = w[1] = (char *)(w + 2);
1726         for (i = 0; i < array->csize; i++) {
1727                 hi = array->items[i];
1728                 while (hi) {
1729                         strcpy(*w, hi->name);
1730                         nextword(w);
1731                         hi = hi->next;
1732                 }
1733         }
1734 }
1735
1736 static int hashwalk_next(var *v)
1737 {
1738         char **w;
1739
1740         w = v->x.walker;
1741         if (w[1] == w[0])
1742                 return FALSE;
1743
1744         setvar_s(v, nextword(w+1));
1745         return TRUE;
1746 }
1747
1748 /* evaluate node, return 1 when result is true, 0 otherwise */
1749 static int ptest(node *pattern)
1750 {
1751         /* ptest__v is "static": to save stack space? */
1752         return istrue(evaluate(pattern, &G.ptest__v));
1753 }
1754
1755 /* read next record from stream rsm into a variable v */
1756 static int awk_getline(rstream *rsm, var *v)
1757 {
1758         char *b;
1759         regmatch_t pmatch[2];
1760         int a, p, pp=0, size;
1761         int fd, so, eo, r, rp;
1762         char c, *m, *s;
1763
1764         /* we're using our own buffer since we need access to accumulating
1765          * characters
1766          */
1767         fd = fileno(rsm->F);
1768         m = rsm->buffer;
1769         a = rsm->adv;
1770         p = rsm->pos;
1771         size = rsm->size;
1772         c = (char) rsplitter.n.info;
1773         rp = 0;
1774
1775         if (!m) qrealloc(&m, 256, &size);
1776         do {
1777                 b = m + a;
1778                 so = eo = p;
1779                 r = 1;
1780                 if (p > 0) {
1781                         if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) {
1782                                 if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re,
1783                                                         b, 1, pmatch, 0) == 0) {
1784                                         so = pmatch[0].rm_so;
1785                                         eo = pmatch[0].rm_eo;
1786                                         if (b[eo] != '\0')
1787                                                 break;
1788                                 }
1789                         } else if (c != '\0') {
1790                                 s = strchr(b+pp, c);
1791                                 if (!s) s = memchr(b+pp, '\0', p - pp);
1792                                 if (s) {
1793                                         so = eo = s-b;
1794                                         eo++;
1795                                         break;
1796                                 }
1797                         } else {
1798                                 while (b[rp] == '\n')
1799                                         rp++;
1800                                 s = strstr(b+rp, "\n\n");
1801                                 if (s) {
1802                                         so = eo = s-b;
1803                                         while (b[eo] == '\n') eo++;
1804                                         if (b[eo] != '\0')
1805                                                 break;
1806                                 }
1807                         }
1808                 }
1809
1810                 if (a > 0) {
1811                         memmove(m, (const void *)(m+a), p+1);
1812                         b = m;
1813                         a = 0;
1814                 }
1815
1816                 qrealloc(&m, a+p+128, &size);
1817                 b = m + a;
1818                 pp = p;
1819                 p += safe_read(fd, b+p, size-p-1);
1820                 if (p < pp) {
1821                         p = 0;
1822                         r = 0;
1823                         setvar_i(intvar[ERRNO], errno);
1824                 }
1825                 b[p] = '\0';
1826
1827         } while (p > pp);
1828
1829         if (p == 0) {
1830                 r--;
1831         } else {
1832                 c = b[so]; b[so] = '\0';
1833                 setvar_s(v, b+rp);
1834                 v->type |= VF_USER;
1835                 b[so] = c;
1836                 c = b[eo]; b[eo] = '\0';
1837                 setvar_s(intvar[RT], b+so);
1838                 b[eo] = c;
1839         }
1840
1841         rsm->buffer = m;
1842         rsm->adv = a + eo;
1843         rsm->pos = p - eo;
1844         rsm->size = size;
1845
1846         return r;
1847 }
1848
1849 static int fmt_num(char *b, int size, const char *format, double n, int int_as_int)
1850 {
1851         int r = 0;
1852         char c;
1853         const char *s = format;
1854
1855         if (int_as_int && n == (int)n) {
1856                 r = snprintf(b, size, "%d", (int)n);
1857         } else {
1858                 do { c = *s; } while (c && *++s);
1859                 if (strchr("diouxX", c)) {
1860                         r = snprintf(b, size, format, (int)n);
1861                 } else if (strchr("eEfgG", c)) {
1862                         r = snprintf(b, size, format, n);
1863                 } else {
1864                         syntax_error(EMSG_INV_FMT);
1865                 }
1866         }
1867         return r;
1868 }
1869
1870 /* formatted output into an allocated buffer, return ptr to buffer */
1871 static char *awk_printf(node *n)
1872 {
1873         char *b = NULL;
1874         char *fmt, *s, *f;
1875         const char *s1;
1876         int i, j, incr, bsize;
1877         char c, c1;
1878         var *v, *arg;
1879
1880         v = nvalloc(1);
1881         fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v)));
1882
1883         i = 0;
1884         while (*f) {
1885                 s = f;
1886                 while (*f && (*f != '%' || *(++f) == '%'))
1887                         f++;
1888                 while (*f && !isalpha(*f)) {
1889                         if (*f == '*')
1890                                 syntax_error("%*x formats are not supported");
1891                         f++;
1892                 }
1893
1894                 incr = (f - s) + MAXVARFMT;
1895                 qrealloc(&b, incr + i, &bsize);
1896                 c = *f;
1897                 if (c != '\0') f++;
1898                 c1 = *f;
1899                 *f = '\0';
1900                 arg = evaluate(nextarg(&n), v);
1901
1902                 j = i;
1903                 if (c == 'c' || !c) {
1904                         i += sprintf(b+i, s, is_numeric(arg) ?
1905                                         (char)getvar_i(arg) : *getvar_s(arg));
1906                 } else if (c == 's') {
1907                         s1 = getvar_s(arg);
1908                         qrealloc(&b, incr+i+strlen(s1), &bsize);
1909                         i += sprintf(b+i, s, s1);
1910                 } else {
1911                         i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE);
1912                 }
1913                 *f = c1;
1914
1915                 /* if there was an error while sprintf, return value is negative */
1916                 if (i < j) i = j;
1917         }
1918
1919         b = xrealloc(b, i + 1);
1920         free(fmt);
1921         nvfree(v);
1922         b[i] = '\0';
1923         return b;
1924 }
1925
1926 /* common substitution routine
1927  * replace (nm) substring of (src) that match (n) with (repl), store
1928  * result into (dest), return number of substitutions. If nm=0, replace
1929  * all matches. If src or dst is NULL, use $0. If ex=TRUE, enable
1930  * subexpression matching (\1-\9)
1931  */
1932 static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int ex)
1933 {
1934         char *ds = NULL;
1935         const char *s;
1936         const char *sp;
1937         int c, i, j, di, rl, so, eo, nbs, n, dssize;
1938         regmatch_t pmatch[10];
1939         regex_t sreg, *re;
1940
1941         re = as_regex(rn, &sreg);
1942         if (!src) src = intvar[F0];
1943         if (!dest) dest = intvar[F0];
1944
1945         i = di = 0;
1946         sp = getvar_s(src);
1947         rl = strlen(repl);
1948         while (regexec(re, sp, 10, pmatch, sp==getvar_s(src) ? 0 : REG_NOTBOL) == 0) {
1949                 so = pmatch[0].rm_so;
1950                 eo = pmatch[0].rm_eo;
1951
1952                 qrealloc(&ds, di + eo + rl, &dssize);
1953                 memcpy(ds + di, sp, eo);
1954                 di += eo;
1955                 if (++i >= nm) {
1956                         /* replace */
1957                         di -= (eo - so);
1958                         nbs = 0;
1959                         for (s = repl; *s; s++) {
1960                                 ds[di++] = c = *s;
1961                                 if (c == '\\') {
1962                                         nbs++;
1963                                         continue;
1964                                 }
1965                                 if (c == '&' || (ex && c >= '0' && c <= '9')) {
1966                                         di -= ((nbs + 3) >> 1);
1967                                         j = 0;
1968                                         if (c != '&') {
1969                                                 j = c - '0';
1970                                                 nbs++;
1971                                         }
1972                                         if (nbs % 2) {
1973                                                 ds[di++] = c;
1974                                         } else {
1975                                                 n = pmatch[j].rm_eo - pmatch[j].rm_so;
1976                                                 qrealloc(&ds, di + rl + n, &dssize);
1977                                                 memcpy(ds + di, sp + pmatch[j].rm_so, n);
1978                                                 di += n;
1979                                         }
1980                                 }
1981                                 nbs = 0;
1982                         }
1983                 }
1984
1985                 sp += eo;
1986                 if (i == nm) break;
1987                 if (eo == so) {
1988                         ds[di] = *sp++;
1989                         if (!ds[di++]) break;
1990                 }
1991         }
1992
1993         qrealloc(&ds, di + strlen(sp), &dssize);
1994         strcpy(ds + di, sp);
1995         setvar_p(dest, ds);
1996         if (re == &sreg) regfree(re);
1997         return i;
1998 }
1999
2000 static var *exec_builtin(node *op, var *res)
2001 {
2002 #define tspl (G.exec_builtin__tspl)
2003
2004         int (*to_xxx)(int);
2005         var *tv;
2006         node *an[4];
2007         var *av[4];
2008         const char *as[4];
2009         regmatch_t pmatch[2];
2010         regex_t sreg, *re;
2011         node *spl;
2012         uint32_t isr, info;
2013         int nargs;
2014         time_t tt;
2015         char *s, *s1;
2016         int i, l, ll, n;
2017
2018         tv = nvalloc(4);
2019         isr = info = op->info;
2020         op = op->l.n;
2021
2022         av[2] = av[3] = NULL;
2023         for (i = 0; i < 4 && op; i++) {
2024                 an[i] = nextarg(&op);
2025                 if (isr & 0x09000000) av[i] = evaluate(an[i], &tv[i]);
2026                 if (isr & 0x08000000) as[i] = getvar_s(av[i]);
2027                 isr >>= 1;
2028         }
2029
2030         nargs = i;
2031         if ((uint32_t)nargs < (info >> 30))
2032                 syntax_error(EMSG_TOO_FEW_ARGS);
2033
2034         switch (info & OPNMASK) {
2035
2036         case B_a2:
2037 #if ENABLE_FEATURE_AWK_LIBM
2038                 setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1])));
2039 #else
2040                 syntax_error(EMSG_NO_MATH);
2041 #endif
2042                 break;
2043
2044         case B_sp:
2045                 if (nargs > 2) {
2046                         spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ?
2047                                 an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl);
2048                 } else {
2049                         spl = &fsplitter.n;
2050                 }
2051
2052                 n = awk_split(as[0], spl, &s);
2053                 s1 = s;
2054                 clear_array(iamarray(av[1]));
2055                 for (i = 1; i <= n; i++)
2056                         setari_u(av[1], i, nextword(&s1));
2057                 free(s);
2058                 setvar_i(res, n);
2059                 break;
2060
2061         case B_ss:
2062                 l = strlen(as[0]);
2063                 i = getvar_i(av[1]) - 1;
2064                 if (i > l) i = l;
2065                 if (i < 0) i = 0;
2066                 n = (nargs > 2) ? getvar_i(av[2]) : l-i;
2067                 if (n < 0) n = 0;
2068                 s = xstrndup(as[0]+i, n);
2069                 setvar_p(res, s);
2070                 break;
2071
2072         /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5:
2073          * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */
2074         case B_an:
2075                 setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1]));
2076                 break;
2077
2078         case B_co:
2079                 setvar_i(res, ~getvar_i_int(av[0]));
2080                 break;
2081
2082         case B_ls:
2083                 setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1]));
2084                 break;
2085
2086         case B_or:
2087                 setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1]));
2088                 break;
2089
2090         case B_rs:
2091                 setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1]));
2092                 break;
2093
2094         case B_xo:
2095                 setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1]));
2096                 break;
2097
2098         case B_lo:
2099                 to_xxx = tolower;
2100                 goto lo_cont;
2101
2102         case B_up:
2103                 to_xxx = toupper;
2104  lo_cont:
2105                 s1 = s = xstrdup(as[0]);
2106                 while (*s1) {
2107                         *s1 = (*to_xxx)(*s1);
2108                         s1++;
2109                 }
2110                 setvar_p(res, s);
2111                 break;
2112
2113         case B_ix:
2114                 n = 0;
2115                 ll = strlen(as[1]);
2116                 l = strlen(as[0]) - ll;
2117                 if (ll > 0 && l >= 0) {
2118                         if (!icase) {
2119                                 s = strstr(as[0], as[1]);
2120                                 if (s) n = (s - as[0]) + 1;
2121                         } else {
2122                                 /* this piece of code is terribly slow and
2123                                  * really should be rewritten
2124                                  */
2125                                 for (i=0; i<=l; i++) {
2126                                         if (strncasecmp(as[0]+i, as[1], ll) == 0) {
2127                                                 n = i+1;
2128                                                 break;
2129                                         }
2130                                 }
2131                         }
2132                 }
2133                 setvar_i(res, n);
2134                 break;
2135
2136         case B_ti:
2137                 if (nargs > 1)
2138                         tt = getvar_i(av[1]);
2139                 else
2140                         time(&tt);
2141                 //s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y";
2142                 i = strftime(g_buf, MAXVARFMT,
2143                         ((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"),
2144                         localtime(&tt));
2145                 g_buf[i] = '\0';
2146                 setvar_s(res, g_buf);
2147                 break;
2148
2149         case B_ma:
2150                 re = as_regex(an[1], &sreg);
2151                 n = regexec(re, as[0], 1, pmatch, 0);
2152                 if (n == 0) {
2153                         pmatch[0].rm_so++;
2154                         pmatch[0].rm_eo++;
2155                 } else {
2156                         pmatch[0].rm_so = 0;
2157                         pmatch[0].rm_eo = -1;
2158                 }
2159                 setvar_i(newvar("RSTART"), pmatch[0].rm_so);
2160                 setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so);
2161                 setvar_i(res, pmatch[0].rm_so);
2162                 if (re == &sreg) regfree(re);
2163                 break;
2164
2165         case B_ge:
2166                 awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE);
2167                 break;
2168
2169         case B_gs:
2170                 setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE));
2171                 break;
2172
2173         case B_su:
2174                 setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE));
2175                 break;
2176         }
2177
2178         nvfree(tv);
2179         return res;
2180 #undef tspl
2181 }
2182
2183 /*
2184  * Evaluate node - the heart of the program. Supplied with subtree
2185  * and place where to store result. returns ptr to result.
2186  */
2187 #define XC(n) ((n) >> 8)
2188
2189 static var *evaluate(node *op, var *res)
2190 {
2191 /* This procedure is recursive so we should count every byte */
2192 #define fnargs (G.evaluate__fnargs)
2193 /* seed is initialized to 1 */
2194 #define seed   (G.evaluate__seed)
2195 #define sreg   (G.evaluate__sreg)
2196
2197         node *op1;
2198         var *v1;
2199         union {
2200                 var *v;
2201                 const char *s;
2202                 double d;
2203                 int i;
2204         } L, R;
2205         uint32_t opinfo;
2206         int opn;
2207         union {
2208                 char *s;
2209                 rstream *rsm;
2210                 FILE *F;
2211                 var *v;
2212                 regex_t *re;
2213                 uint32_t info;
2214         } X;
2215
2216         if (!op)
2217                 return setvar_s(res, NULL);
2218
2219         v1 = nvalloc(2);
2220
2221         while (op) {
2222                 opinfo = op->info;
2223                 opn = (opinfo & OPNMASK);
2224                 g_lineno = op->lineno;
2225
2226                 /* execute inevitable things */
2227                 op1 = op->l.n;
2228                 if (opinfo & OF_RES1) X.v = L.v = evaluate(op1, v1);
2229                 if (opinfo & OF_RES2) R.v = evaluate(op->r.n, v1+1);
2230                 if (opinfo & OF_STR1) L.s = getvar_s(L.v);
2231                 if (opinfo & OF_STR2) R.s = getvar_s(R.v);
2232                 if (opinfo & OF_NUM1) L.d = getvar_i(L.v);
2233
2234                 switch (XC(opinfo & OPCLSMASK)) {
2235
2236                 /* -- iterative node type -- */
2237
2238                 /* test pattern */
2239                 case XC( OC_TEST ):
2240                         if ((op1->info & OPCLSMASK) == OC_COMMA) {
2241                                 /* it's range pattern */
2242                                 if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) {
2243                                         op->info |= OF_CHECKED;
2244                                         if (ptest(op1->r.n))
2245                                                 op->info &= ~OF_CHECKED;
2246
2247                                         op = op->a.n;
2248                                 } else {
2249                                         op = op->r.n;
2250                                 }
2251                         } else {
2252                                 op = (ptest(op1)) ? op->a.n : op->r.n;
2253                         }
2254                         break;
2255
2256                 /* just evaluate an expression, also used as unconditional jump */
2257                 case XC( OC_EXEC ):
2258                         break;
2259
2260                 /* branch, used in if-else and various loops */
2261                 case XC( OC_BR ):
2262                         op = istrue(L.v) ? op->a.n : op->r.n;
2263                         break;
2264
2265                 /* initialize for-in loop */
2266                 case XC( OC_WALKINIT ):
2267                         hashwalk_init(L.v, iamarray(R.v));
2268                         break;
2269
2270                 /* get next array item */
2271                 case XC( OC_WALKNEXT ):
2272                         op = hashwalk_next(L.v) ? op->a.n : op->r.n;
2273                         break;
2274
2275                 case XC( OC_PRINT ):
2276                 case XC( OC_PRINTF ):
2277                         X.F = stdout;
2278                         if (op->r.n) {
2279                                 X.rsm = newfile(R.s);
2280                                 if (!X.rsm->F) {
2281                                         if (opn == '|') {
2282                                                 X.rsm->F = popen(R.s, "w");
2283                                                 if (X.rsm->F == NULL)
2284                                                         bb_perror_msg_and_die("popen");
2285                                                 X.rsm->is_pipe = 1;
2286                                         } else {
2287                                                 X.rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
2288                                         }
2289                                 }
2290                                 X.F = X.rsm->F;
2291                         }
2292
2293                         if ((opinfo & OPCLSMASK) == OC_PRINT) {
2294                                 if (!op1) {
2295                                         fputs(getvar_s(intvar[F0]), X.F);
2296                                 } else {
2297                                         while (op1) {
2298                                                 L.v = evaluate(nextarg(&op1), v1);
2299                                                 if (L.v->type & VF_NUMBER) {
2300                                                         fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]),
2301                                                                         getvar_i(L.v), TRUE);
2302                                                         fputs(g_buf, X.F);
2303                                                 } else {
2304                                                         fputs(getvar_s(L.v), X.F);
2305                                                 }
2306
2307                                                 if (op1) fputs(getvar_s(intvar[OFS]), X.F);
2308                                         }
2309                                 }
2310                                 fputs(getvar_s(intvar[ORS]), X.F);
2311
2312                         } else {        /* OC_PRINTF */
2313                                 L.s = awk_printf(op1);
2314                                 fputs(L.s, X.F);
2315                                 free((char*)L.s);
2316                         }
2317                         fflush(X.F);
2318                         break;
2319
2320                 case XC( OC_DELETE ):
2321                         X.info = op1->info & OPCLSMASK;
2322                         if (X.info == OC_VAR) {
2323                                 R.v = op1->l.v;
2324                         } else if (X.info == OC_FNARG) {
2325                                 R.v = &fnargs[op1->l.i];
2326                         } else {
2327                                 syntax_error(EMSG_NOT_ARRAY);
2328                         }
2329
2330                         if (op1->r.n) {
2331                                 clrvar(L.v);
2332                                 L.s = getvar_s(evaluate(op1->r.n, v1));
2333                                 hash_remove(iamarray(R.v), L.s);
2334                         } else {
2335                                 clear_array(iamarray(R.v));
2336                         }
2337                         break;
2338
2339                 case XC( OC_NEWSOURCE ):
2340                         g_progname = op->l.s;
2341                         break;
2342
2343                 case XC( OC_RETURN ):
2344                         copyvar(res, L.v);
2345                         break;
2346
2347                 case XC( OC_NEXTFILE ):
2348                         nextfile = TRUE;
2349                 case XC( OC_NEXT ):
2350                         nextrec = TRUE;
2351                 case XC( OC_DONE ):
2352                         clrvar(res);
2353                         break;
2354
2355                 case XC( OC_EXIT ):
2356                         awk_exit(L.d);
2357
2358                 /* -- recursive node type -- */
2359
2360                 case XC( OC_VAR ):
2361                         L.v = op->l.v;
2362                         if (L.v == intvar[NF])
2363                                 split_f0();
2364                         goto v_cont;
2365
2366                 case XC( OC_FNARG ):
2367                         L.v = &fnargs[op->l.i];
2368  v_cont:
2369                         res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v;
2370                         break;
2371
2372                 case XC( OC_IN ):
2373                         setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0);
2374                         break;
2375
2376                 case XC( OC_REGEXP ):
2377                         op1 = op;
2378                         L.s = getvar_s(intvar[F0]);
2379                         goto re_cont;
2380
2381                 case XC( OC_MATCH ):
2382                         op1 = op->r.n;
2383  re_cont:
2384                         X.re = as_regex(op1, &sreg);
2385                         R.i = regexec(X.re, L.s, 0, NULL, 0);
2386                         if (X.re == &sreg) regfree(X.re);
2387                         setvar_i(res, (R.i == 0 ? 1 : 0) ^ (opn == '!' ? 1 : 0));
2388                         break;
2389
2390                 case XC( OC_MOVE ):
2391                         /* if source is a temporary string, jusk relink it to dest */
2392                         if (R.v == v1+1 && R.v->string) {
2393                                 res = setvar_p(L.v, R.v->string);
2394                                 R.v->string = NULL;
2395                         } else {
2396                                 res = copyvar(L.v, R.v);
2397                         }
2398                         break;
2399
2400                 case XC( OC_TERNARY ):
2401                         if ((op->r.n->info & OPCLSMASK) != OC_COLON)
2402                                 syntax_error(EMSG_POSSIBLE_ERROR);
2403                         res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res);
2404                         break;
2405
2406                 case XC( OC_FUNC ):
2407                         if (!op->r.f->body.first)
2408                                 syntax_error(EMSG_UNDEF_FUNC);
2409
2410                         X.v = R.v = nvalloc(op->r.f->nargs+1);
2411                         while (op1) {
2412                                 L.v = evaluate(nextarg(&op1), v1);
2413                                 copyvar(R.v, L.v);
2414                                 R.v->type |= VF_CHILD;
2415                                 R.v->x.parent = L.v;
2416                                 if (++R.v - X.v >= op->r.f->nargs)
2417                                         break;
2418                         }
2419
2420                         R.v = fnargs;
2421                         fnargs = X.v;
2422
2423                         L.s = g_progname;
2424                         res = evaluate(op->r.f->body.first, res);
2425                         g_progname = L.s;
2426
2427                         nvfree(fnargs);
2428                         fnargs = R.v;
2429                         break;
2430
2431                 case XC( OC_GETLINE ):
2432                 case XC( OC_PGETLINE ):
2433                         if (op1) {
2434                                 X.rsm = newfile(L.s);
2435                                 if (!X.rsm->F) {
2436                                         if ((opinfo & OPCLSMASK) == OC_PGETLINE) {
2437                                                 X.rsm->F = popen(L.s, "r");
2438                                                 X.rsm->is_pipe = TRUE;
2439                                         } else {
2440                                                 X.rsm->F = fopen_for_read(L.s);         /* not xfopen! */
2441                                         }
2442                                 }
2443                         } else {
2444                                 if (!iF) iF = next_input_file();
2445                                 X.rsm = iF;
2446                         }
2447
2448                         if (!X.rsm->F) {
2449                                 setvar_i(intvar[ERRNO], errno);
2450                                 setvar_i(res, -1);
2451                                 break;
2452                         }
2453
2454                         if (!op->r.n)
2455                                 R.v = intvar[F0];
2456
2457                         L.i = awk_getline(X.rsm, R.v);
2458                         if (L.i > 0) {
2459                                 if (!op1) {
2460                                         incvar(intvar[FNR]);
2461                                         incvar(intvar[NR]);
2462                                 }
2463                         }
2464                         setvar_i(res, L.i);
2465                         break;
2466
2467                 /* simple builtins */
2468                 case XC( OC_FBLTIN ):
2469                         switch (opn) {
2470
2471                         case F_in:
2472                                 R.d = (int)L.d;
2473                                 break;
2474
2475                         case F_rn:
2476                                 R.d = (double)rand() / (double)RAND_MAX;
2477                                 break;
2478 #if ENABLE_FEATURE_AWK_LIBM
2479                         case F_co:
2480                                 R.d = cos(L.d);
2481                                 break;
2482
2483                         case F_ex:
2484                                 R.d = exp(L.d);
2485                                 break;
2486
2487                         case F_lg:
2488                                 R.d = log(L.d);
2489                                 break;
2490
2491                         case F_si:
2492                                 R.d = sin(L.d);
2493                                 break;
2494
2495                         case F_sq:
2496                                 R.d = sqrt(L.d);
2497                                 break;
2498 #else
2499                         case F_co:
2500                         case F_ex:
2501                         case F_lg:
2502                         case F_si:
2503                         case F_sq:
2504                                 syntax_error(EMSG_NO_MATH);
2505                                 break;
2506 #endif
2507                         case F_sr:
2508                                 R.d = (double)seed;
2509                                 seed = op1 ? (unsigned)L.d : (unsigned)time(NULL);
2510                                 srand(seed);
2511                                 break;
2512
2513                         case F_ti:
2514                                 R.d = time(NULL);
2515                                 break;
2516
2517                         case F_le:
2518                                 if (!op1)
2519                                         L.s = getvar_s(intvar[F0]);
2520                                 R.d = strlen(L.s);
2521                                 break;
2522
2523                         case F_sy:
2524                                 fflush(NULL);
2525                                 R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
2526                                                 ? (system(L.s) >> 8) : 0;
2527                                 break;
2528
2529                         case F_ff:
2530                                 if (!op1)
2531                                         fflush(stdout);
2532                                 else {
2533                                         if (L.s && *L.s) {
2534                                                 X.rsm = newfile(L.s);
2535                                                 fflush(X.rsm->F);
2536                                         } else {
2537                                                 fflush(NULL);
2538                                         }
2539                                 }
2540                                 break;
2541
2542                         case F_cl:
2543                                 X.rsm = (rstream *)hash_search(fdhash, L.s);
2544                                 if (X.rsm) {
2545                                         R.i = X.rsm->is_pipe ? pclose(X.rsm->F) : fclose(X.rsm->F);
2546                                         free(X.rsm->buffer);
2547                                         hash_remove(fdhash, L.s);
2548                                 }
2549                                 if (R.i != 0)
2550                                         setvar_i(intvar[ERRNO], errno);
2551                                 R.d = (double)R.i;
2552                                 break;
2553                         }
2554                         setvar_i(res, R.d);
2555                         break;
2556
2557                 case XC( OC_BUILTIN ):
2558                         res = exec_builtin(op, res);
2559                         break;
2560
2561                 case XC( OC_SPRINTF ):
2562                         setvar_p(res, awk_printf(op1));
2563                         break;
2564
2565                 case XC( OC_UNARY ):
2566                         X.v = R.v;
2567                         L.d = R.d = getvar_i(R.v);
2568                         switch (opn) {
2569                         case 'P':
2570                                 L.d = ++R.d;
2571                                 goto r_op_change;
2572                         case 'p':
2573                                 R.d++;
2574                                 goto r_op_change;
2575                         case 'M':
2576                                 L.d = --R.d;
2577                                 goto r_op_change;
2578                         case 'm':
2579                                 R.d--;
2580                                 goto r_op_change;
2581                         case '!':
2582                                 L.d = istrue(X.v) ? 0 : 1;
2583                                 break;
2584                         case '-':
2585                                 L.d = -R.d;
2586                                 break;
2587  r_op_change:
2588                                 setvar_i(X.v, R.d);
2589                         }
2590                         setvar_i(res, L.d);
2591                         break;
2592
2593                 case XC( OC_FIELD ):
2594                         R.i = (int)getvar_i(R.v);
2595                         if (R.i == 0) {
2596                                 res = intvar[F0];
2597                         } else {
2598                                 split_f0();
2599                                 if (R.i > nfields)
2600                                         fsrealloc(R.i);
2601                                 res = &Fields[R.i - 1];
2602                         }
2603                         break;
2604
2605                 /* concatenation (" ") and index joining (",") */
2606                 case XC( OC_CONCAT ):
2607                 case XC( OC_COMMA ):
2608                         opn = strlen(L.s) + strlen(R.s) + 2;
2609                         X.s = xmalloc(opn);
2610                         strcpy(X.s, L.s);
2611                         if ((opinfo & OPCLSMASK) == OC_COMMA) {
2612                                 L.s = getvar_s(intvar[SUBSEP]);
2613                                 X.s = xrealloc(X.s, opn + strlen(L.s));
2614                                 strcat(X.s, L.s);
2615                         }
2616                         strcat(X.s, R.s);
2617                         setvar_p(res, X.s);
2618                         break;
2619
2620                 case XC( OC_LAND ):
2621                         setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0);
2622                         break;
2623
2624                 case XC( OC_LOR ):
2625                         setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n));
2626                         break;
2627
2628                 case XC( OC_BINARY ):
2629                 case XC( OC_REPLACE ):
2630                         R.d = getvar_i(R.v);
2631                         switch (opn) {
2632                         case '+':
2633                                 L.d += R.d;
2634                                 break;
2635                         case '-':
2636                                 L.d -= R.d;
2637                                 break;
2638                         case '*':
2639                                 L.d *= R.d;
2640                                 break;
2641                         case '/':
2642                                 if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
2643                                 L.d /= R.d;
2644                                 break;
2645                         case '&':
2646 #if ENABLE_FEATURE_AWK_LIBM
2647                                 L.d = pow(L.d, R.d);
2648 #else
2649                                 syntax_error(EMSG_NO_MATH);
2650 #endif
2651                                 break;
2652                         case '%':
2653                                 if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
2654                                 L.d -= (int)(L.d / R.d) * R.d;
2655                                 break;
2656                         }
2657                         res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : X.v, L.d);
2658                         break;
2659
2660                 case XC( OC_COMPARE ):
2661                         if (is_numeric(L.v) && is_numeric(R.v)) {
2662                                 L.d = getvar_i(L.v) - getvar_i(R.v);
2663                         } else {
2664                                 L.s = getvar_s(L.v);
2665                                 R.s = getvar_s(R.v);
2666                                 L.d = icase ? strcasecmp(L.s, R.s) : strcmp(L.s, R.s);
2667                         }
2668                         switch (opn & 0xfe) {
2669                         case 0:
2670                                 R.i = (L.d > 0);
2671                                 break;
2672                         case 2:
2673                                 R.i = (L.d >= 0);
2674                                 break;
2675                         case 4:
2676                                 R.i = (L.d == 0);
2677                                 break;
2678                         }
2679                         setvar_i(res, (opn & 0x1 ? R.i : !R.i) ? 1 : 0);
2680                         break;
2681
2682                 default:
2683                         syntax_error(EMSG_POSSIBLE_ERROR);
2684                 }
2685                 if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS)
2686                         op = op->a.n;
2687                 if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS)
2688                         break;
2689                 if (nextrec)
2690                         break;
2691         }
2692         nvfree(v1);
2693         return res;
2694 #undef fnargs
2695 #undef seed
2696 #undef sreg
2697 }
2698
2699
2700 /* -------- main & co. -------- */
2701
2702 static int awk_exit(int r)
2703 {
2704         var tv;
2705         unsigned i;
2706         hash_item *hi;
2707
2708         zero_out_var(&tv);
2709
2710         if (!exiting) {
2711                 exiting = TRUE;
2712                 nextrec = FALSE;
2713                 evaluate(endseq.first, &tv);
2714         }
2715
2716         /* waiting for children */
2717         for (i = 0; i < fdhash->csize; i++) {
2718                 hi = fdhash->items[i];
2719                 while (hi) {
2720                         if (hi->data.rs.F && hi->data.rs.is_pipe)
2721                                 pclose(hi->data.rs.F);
2722                         hi = hi->next;
2723                 }
2724         }
2725
2726         exit(r);
2727 }
2728
2729 /* if expr looks like "var=value", perform assignment and return 1,
2730  * otherwise return 0 */
2731 static int is_assignment(const char *expr)
2732 {
2733         char *exprc, *s, *s0, *s1;
2734
2735         exprc = xstrdup(expr);
2736         if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) {
2737                 free(exprc);
2738                 return FALSE;
2739         }
2740
2741         *(s++) = '\0';
2742         s0 = s1 = s;
2743         while (*s)
2744                 *(s1++) = nextchar(&s);
2745
2746         *s1 = '\0';
2747         setvar_u(newvar(exprc), s0);
2748         free(exprc);
2749         return TRUE;
2750 }
2751
2752 /* switch to next input file */
2753 static rstream *next_input_file(void)
2754 {
2755 #define rsm          (G.next_input_file__rsm)
2756 #define files_happen (G.next_input_file__files_happen)
2757
2758         FILE *F = NULL;
2759         const char *fname, *ind;
2760
2761         if (rsm.F) fclose(rsm.F);
2762         rsm.F = NULL;
2763         rsm.pos = rsm.adv = 0;
2764
2765         do {
2766                 if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) {
2767                         if (files_happen)
2768                                 return NULL;
2769                         fname = "-";
2770                         F = stdin;
2771                 } else {
2772                         ind = getvar_s(incvar(intvar[ARGIND]));
2773                         fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind));
2774                         if (fname && *fname && !is_assignment(fname))
2775                                 F = xfopen_stdin(fname);
2776                 }
2777         } while (!F);
2778
2779         files_happen = TRUE;
2780         setvar_s(intvar[FILENAME], fname);
2781         rsm.F = F;
2782         return &rsm;
2783 #undef rsm
2784 #undef files_happen
2785 }
2786
2787 int awk_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
2788 int awk_main(int argc, char **argv)
2789 {
2790         unsigned opt;
2791         char *opt_F, *opt_W;
2792         llist_t *list_v = NULL;
2793         llist_t *list_f = NULL;
2794         int i, j;
2795         var *v;
2796         var tv;
2797         char **envp;
2798         char *vnames = (char *)vNames; /* cheat */
2799         char *vvalues = (char *)vValues;
2800
2801         INIT_G();
2802
2803         /* Undo busybox.c, or else strtod may eat ','! This breaks parsing:
2804          * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */
2805         if (ENABLE_LOCALE_SUPPORT)
2806                 setlocale(LC_NUMERIC, "C");
2807
2808         zero_out_var(&tv);
2809
2810         /* allocate global buffer */
2811         g_buf = xmalloc(MAXVARFMT + 1);
2812
2813         vhash = hash_init();
2814         ahash = hash_init();
2815         fdhash = hash_init();
2816         fnhash = hash_init();
2817
2818         /* initialize variables */
2819         for (i = 0; *vnames; i++) {
2820                 intvar[i] = v = newvar(nextword(&vnames));
2821                 if (*vvalues != '\377')
2822                         setvar_s(v, nextword(&vvalues));
2823                 else
2824                         setvar_i(v, 0);
2825
2826                 if (*vnames == '*') {
2827                         v->type |= VF_SPECIAL;
2828                         vnames++;
2829                 }
2830         }
2831
2832         handle_special(intvar[FS]);
2833         handle_special(intvar[RS]);
2834
2835         newfile("/dev/stdin")->F = stdin;
2836         newfile("/dev/stdout")->F = stdout;
2837         newfile("/dev/stderr")->F = stderr;
2838
2839         /* Huh, people report that sometimes environ is NULL. Oh well. */
2840         if (environ) for (envp = environ; *envp; envp++) {
2841                 /* environ is writable, thus we don't strdup it needlessly */
2842                 char *s = *envp;
2843                 char *s1 = strchr(s, '=');
2844                 if (s1) {
2845                         *s1 = '\0';
2846                         /* Both findvar and setvar_u take const char*
2847                          * as 2nd arg -> environment is not trashed */
2848                         setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1);
2849                         *s1 = '=';
2850                 }
2851         }
2852         opt_complementary = "v::f::"; /* -v and -f can occur multiple times */
2853         opt = getopt32(argv, "F:v:f:W:", &opt_F, &list_v, &list_f, &opt_W);
2854         argv += optind;
2855         argc -= optind;
2856         if (opt & 0x1)
2857                 setvar_s(intvar[FS], opt_F); // -F
2858         while (list_v) { /* -v */
2859                 if (!is_assignment(llist_pop(&list_v)))
2860                         bb_show_usage();
2861         }
2862         if (list_f) { /* -f */
2863                 do {
2864                         char *s = NULL;
2865                         FILE *from_file;
2866
2867                         g_progname = llist_pop(&list_f);
2868                         from_file = xfopen_stdin(g_progname);
2869                         /* one byte is reserved for some trick in next_token */
2870                         for (i = j = 1; j > 0; i += j) {
2871                                 s = xrealloc(s, i + 4096);
2872                                 j = fread(s + i, 1, 4094, from_file);
2873                         }
2874                         s[i] = '\0';
2875                         fclose(from_file);
2876                         parse_program(s + 1);
2877                         free(s);
2878                 } while (list_f);
2879                 argc++;
2880         } else { // no -f: take program from 1st parameter
2881                 if (!argc)
2882                         bb_show_usage();
2883                 g_progname = "cmd. line";
2884                 parse_program(*argv++);
2885         }
2886         if (opt & 0x8) // -W
2887                 bb_error_msg("warning: unrecognized option '-W %s' ignored", opt_W);
2888
2889         /* fill in ARGV array */
2890         setvar_i(intvar[ARGC], argc);
2891         setari_u(intvar[ARGV], 0, "awk");
2892         i = 0;
2893         while (*argv)
2894                 setari_u(intvar[ARGV], ++i, *argv++);
2895
2896         evaluate(beginseq.first, &tv);
2897         if (!mainseq.first && !endseq.first)
2898                 awk_exit(EXIT_SUCCESS);
2899
2900         /* input file could already be opened in BEGIN block */
2901         if (!iF) iF = next_input_file();
2902
2903         /* passing through input files */
2904         while (iF) {
2905                 nextfile = FALSE;
2906                 setvar_i(intvar[FNR], 0);
2907
2908                 while ((i = awk_getline(iF, intvar[F0])) > 0) {
2909                         nextrec = FALSE;
2910                         incvar(intvar[NR]);
2911                         incvar(intvar[FNR]);
2912                         evaluate(mainseq.first, &tv);
2913
2914                         if (nextfile)
2915                                 break;
2916                 }
2917
2918                 if (i < 0)
2919                         syntax_error(strerror(errno));
2920
2921                 iF = next_input_file();
2922         }
2923
2924         awk_exit(EXIT_SUCCESS);
2925         /*return 0;*/
2926 }