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 GPLv2 or later, see file LICENSE in this source tree.
   8  */
   9 //config:config AWK
  10 //config:       bool "awk (22 kb)"
  11 //config:       default y
  12 //config:       help
  13 //config:       Awk is used as a pattern scanning and processing language.
  14 //config:
  15 //config:config FEATURE_AWK_LIBM
  16 //config:       bool "Enable math functions (requires libm)"
  17 //config:       default y
  18 //config:       depends on AWK
  19 //config:       help
  20 //config:       Enable math functions of the Awk programming language.
  21 //config:       NOTE: This requires libm to be present for linking.
  22 //config:
  23 //config:config FEATURE_AWK_GNU_EXTENSIONS
  24 //config:       bool "Enable a few GNU extensions"
  25 //config:       default y
  26 //config:       depends on AWK
  27 //config:       help
  28 //config:       Enable a few features from gawk:
  29 //config:       * command line option -e AWK_PROGRAM
  30 //config:       * simultaneous use of -f and -e on the command line.
  31 //config:       This enables the use of awk library files.
  32 //config:       Example: awk -f mylib.awk -e '{print myfunction($1);}' ...
  33
  34 //applet:IF_AWK(APPLET_NOEXEC(awk, awk, BB_DIR_USR_BIN, BB_SUID_DROP, awk))
  35
  36 //kbuild:lib-$(CONFIG_AWK) += awk.o
  37
  38 //usage:#define awk_trivial_usage
  39 //usage:       "[OPTIONS] [AWK_PROGRAM] [FILE]..."
  40 //usage:#define awk_full_usage "\n\n"
  41 //usage:       "        -v VAR=VAL      Set variable"
  42 //usage:     "\n        -F SEP          Use SEP as field separator"
  43 //usage:     "\n        -f FILE         Read program from FILE"
  44 //usage:        IF_FEATURE_AWK_GNU_EXTENSIONS(
  45 //usage:     "\n        -e AWK_PROGRAM"
  46 //usage:        )
  47
  48 #include "libbb.h"
  49 #include "xregex.h"
  50 #include <math.h>
  51
  52 /* This is a NOEXEC applet. Be very careful! */
  53
  54
  55 /* If you comment out one of these below, it will be #defined later
  56  * to perform debug printfs to stderr: */
  57 #define debug_printf_walker(...)  do {} while (0)
  58 #define debug_printf_eval(...)  do {} while (0)
  59 #define debug_printf_parse(...)  do {} while (0)
  60
  61 #ifndef debug_printf_walker
  62 # define debug_printf_walker(...) (fprintf(stderr, __VA_ARGS__))
  63 #endif
  64 #ifndef debug_printf_eval
  65 # define debug_printf_eval(...) (fprintf(stderr, __VA_ARGS__))
  66 #endif
  67 #ifndef debug_printf_parse
  68 # define debug_printf_parse(...) (fprintf(stderr, __VA_ARGS__))
  69 #endif
  70
  71
  72 /* "+": stop on first non-option:
  73  * $ awk 'BEGIN { for(i=1; i<ARGC; ++i) { print i ": " ARGV[i] }}' -argz
  74  * 1: -argz
  75  */
  76 #define OPTSTR_AWK "+" \
  77         "F:v:*f:*" \
  78         IF_FEATURE_AWK_GNU_EXTENSIONS("e:*") \
  79         "W:"
  80 enum {
  81         OPTBIT_F,       /* define field separator */
  82         OPTBIT_v,       /* define variable */
  83         OPTBIT_f,       /* pull in awk program from file */
  84         IF_FEATURE_AWK_GNU_EXTENSIONS(OPTBIT_e,) /* -e AWK_PROGRAM */
  85         OPTBIT_W,       /* -W ignored */
  86         OPT_F = 1 << OPTBIT_F,
  87         OPT_v = 1 << OPTBIT_v,
  88         OPT_f = 1 << OPTBIT_f,
  89         OPT_e = IF_FEATURE_AWK_GNU_EXTENSIONS((1 << OPTBIT_e)) + 0,
  90         OPT_W = 1 << OPTBIT_W
  91 };
  92
  93 #define MAXVARFMT       240
  94 #define MINNVBLOCK      64
  95
  96 /* variable flags */
  97 #define VF_NUMBER       0x0001  /* 1 = primary type is number */
  98 #define VF_ARRAY        0x0002  /* 1 = it's an array */
  99
 100 #define VF_CACHED       0x0100  /* 1 = num/str value has cached str/num eq */
 101 #define VF_USER         0x0200  /* 1 = user input (may be numeric string) */
 102 #define VF_SPECIAL      0x0400  /* 1 = requires extra handling when changed */
 103 #define VF_WALK         0x0800  /* 1 = variable has alloc'd x.walker list */
 104 #define VF_FSTR         0x1000  /* 1 = var::string points to fstring buffer */
 105 #define VF_CHILD        0x2000  /* 1 = function arg; x.parent points to source */
 106 #define VF_DIRTY        0x4000  /* 1 = variable was set explicitly */
 107
 108 /* these flags are static, don't change them when value is changed */
 109 #define VF_DONTTOUCH    (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY)
 110
 111 typedef struct walker_list {
 112         char *end;
 113         char *cur;
 114         struct walker_list *prev;
 115         char wbuf[1];
 116 } walker_list;
 117
 118 /* Variable */
 119 typedef struct var_s {
 120         unsigned type;            /* flags */
 121         double number;
 122         char *string;
 123         union {
 124                 int aidx;               /* func arg idx (for compilation stage) */
 125                 struct xhash_s *array;  /* array ptr */
 126                 struct var_s *parent;   /* for func args, ptr to actual parameter */
 127                 walker_list *walker;    /* list of array elements (for..in) */
 128         } x;
 129 } var;
 130
 131 /* Node chain (pattern-action chain, BEGIN, END, function bodies) */
 132 typedef struct chain_s {
 133         struct node_s *first;
 134         struct node_s *last;
 135         const char *programname;
 136 } chain;
 137
 138 /* Function */
 139 typedef struct func_s {
 140         unsigned nargs;
 141         struct chain_s body;
 142 } func;
 143
 144 /* I/O stream */
 145 typedef struct rstream_s {
 146         FILE *F;
 147         char *buffer;
 148         int adv;
 149         int size;
 150         int pos;
 151         smallint is_pipe;
 152 } rstream;
 153
 154 typedef struct hash_item_s {
 155         union {
 156                 struct var_s v;         /* variable/array hash */
 157                 struct rstream_s rs;    /* redirect streams hash */
 158                 struct func_s f;        /* functions hash */
 159         } data;
 160         struct hash_item_s *next;       /* next in chain */
 161         char name[1];                   /* really it's longer */
 162 } hash_item;
 163
 164 typedef struct xhash_s {
 165         unsigned nel;           /* num of elements */
 166         unsigned csize;         /* current hash size */
 167         unsigned nprime;        /* next hash size in PRIMES[] */
 168         unsigned glen;          /* summary length of item names */
 169         struct hash_item_s **items;
 170 } xhash;
 171
 172 /* Tree node */
 173 typedef struct node_s {
 174         uint32_t info;
 175         unsigned lineno;
 176         union {
 177                 struct node_s *n;
 178                 var *v;
 179                 int aidx;
 180                 char *new_progname;
 181                 regex_t *re;
 182         } l;
 183         union {
 184                 struct node_s *n;
 185                 regex_t *ire;
 186                 func *f;
 187         } r;
 188         union {
 189                 struct node_s *n;
 190         } a;
 191 } node;
 192
 193 /* Block of temporary variables */
 194 typedef struct nvblock_s {
 195         int size;
 196         var *pos;
 197         struct nvblock_s *prev;
 198         struct nvblock_s *next;
 199         var nv[];
 200 } nvblock;
 201
 202 typedef struct tsplitter_s {
 203         node n;
 204         regex_t re[2];
 205 } tsplitter;
 206
 207 /* simple token classes */
 208 /* Order and hex values are very important!!!  See next_token() */
 209 #define TC_SEQSTART     (1 << 0)                /* ( */
 210 #define TC_SEQTERM      (1 << 1)                /* ) */
 211 #define TC_REGEXP       (1 << 2)                /* /.../ */
 212 #define TC_OUTRDR       (1 << 3)                /* | > >> */
 213 #define TC_UOPPOST      (1 << 4)                /* unary postfix operator */
 214 #define TC_UOPPRE1      (1 << 5)                /* unary prefix operator */
 215 #define TC_BINOPX       (1 << 6)                /* two-opnd operator */
 216 #define TC_IN           (1 << 7)
 217 #define TC_COMMA        (1 << 8)
 218 #define TC_PIPE         (1 << 9)                /* input redirection pipe */
 219 #define TC_UOPPRE2      (1 << 10)               /* unary prefix operator */
 220 #define TC_ARRTERM      (1 << 11)               /* ] */
 221 #define TC_GRPSTART     (1 << 12)               /* { */
 222 #define TC_GRPTERM      (1 << 13)               /* } */
 223 #define TC_SEMICOL      (1 << 14)
 224 #define TC_NEWLINE      (1 << 15)
 225 #define TC_STATX        (1 << 16)               /* ctl statement (for, next...) */
 226 #define TC_WHILE        (1 << 17)
 227 #define TC_ELSE         (1 << 18)
 228 #define TC_BUILTIN      (1 << 19)
 229 /* This costs ~50 bytes of code.
 230  * A separate class to support deprecated "length" form. If we don't need that
 231  * (i.e. if we demand that only "length()" with () is valid), then TC_LENGTH
 232  * can be merged with TC_BUILTIN:
 233  */
 234 #define TC_LENGTH       (1 << 20)
 235 #define TC_GETLINE      (1 << 21)
 236 #define TC_FUNCDECL     (1 << 22)               /* 'function' 'func' */
 237 #define TC_BEGIN        (1 << 23)
 238 #define TC_END          (1 << 24)
 239 #define TC_EOF          (1 << 25)
 240 #define TC_VARIABLE     (1 << 26)
 241 #define TC_ARRAY        (1 << 27)
 242 #define TC_FUNCTION     (1 << 28)
 243 #define TC_STRING       (1 << 29)
 244 #define TC_NUMBER       (1 << 30)
 245
 246 #define TC_UOPPRE  (TC_UOPPRE1 | TC_UOPPRE2)
 247
 248 /* combined token classes */
 249 #define TC_BINOP   (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN)
 250 //#define       TC_UNARYOP (TC_UOPPRE | TC_UOPPOST)
 251 #define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION \
 252                    | TC_BUILTIN | TC_LENGTH | TC_GETLINE \
 253                    | TC_SEQSTART | TC_STRING | TC_NUMBER)
 254
 255 #define TC_STATEMNT (TC_STATX | TC_WHILE)
 256 #define TC_OPTERM  (TC_SEMICOL | TC_NEWLINE)
 257
 258 /* word tokens, cannot mean something else if not expected */
 259 #define TC_WORD    (TC_IN | TC_STATEMNT | TC_ELSE \
 260                    | TC_BUILTIN | TC_LENGTH | TC_GETLINE \
 261                    | TC_FUNCDECL | TC_BEGIN | TC_END)
 262
 263 /* discard newlines after these */
 264 #define TC_NOTERM  (TC_COMMA | TC_GRPSTART | TC_GRPTERM \
 265                    | TC_BINOP | TC_OPTERM)
 266
 267 /* what can expression begin with */
 268 #define TC_OPSEQ   (TC_OPERAND | TC_UOPPRE | TC_REGEXP)
 269 /* what can group begin with */
 270 #define TC_GRPSEQ  (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART)
 271
 272 /* if previous token class is CONCAT1 and next is CONCAT2, concatenation */
 273 /* operator is inserted between them */
 274 #define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM \
 275                    | TC_STRING | TC_NUMBER | TC_UOPPOST)
 276 #define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE)
 277
 278 #define OF_RES1    0x010000
 279 #define OF_RES2    0x020000
 280 #define OF_STR1    0x040000
 281 #define OF_STR2    0x080000
 282 #define OF_NUM1    0x100000
 283 #define OF_CHECKED 0x200000
 284
 285 /* combined operator flags */
 286 #define xx      0
 287 #define xV      OF_RES2
 288 #define xS      (OF_RES2 | OF_STR2)
 289 #define Vx      OF_RES1
 290 #define VV      (OF_RES1 | OF_RES2)
 291 #define Nx      (OF_RES1 | OF_NUM1)
 292 #define NV      (OF_RES1 | OF_NUM1 | OF_RES2)
 293 #define Sx      (OF_RES1 | OF_STR1)
 294 #define SV      (OF_RES1 | OF_STR1 | OF_RES2)
 295 #define SS      (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2)
 296
 297 #define OPCLSMASK 0xFF00
 298 #define OPNMASK   0x007F
 299
 300 /* operator priority is a highest byte (even: r->l, odd: l->r grouping)
 301  * For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1,
 302  * n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string
 303  */
 304 #undef P
 305 #undef PRIMASK
 306 #undef PRIMASK2
 307 #define P(x)      (x << 24)
 308 #define PRIMASK   0x7F000000
 309 #define PRIMASK2  0x7E000000
 310
 311 /* Operation classes */
 312
 313 #define SHIFT_TIL_THIS  0x0600
 314 #define RECUR_FROM_THIS 0x1000
 315
 316 enum {
 317         OC_DELETE = 0x0100,     OC_EXEC = 0x0200,       OC_NEWSOURCE = 0x0300,
 318         OC_PRINT = 0x0400,      OC_PRINTF = 0x0500,     OC_WALKINIT = 0x0600,
 319
 320         OC_BR = 0x0700,         OC_BREAK = 0x0800,      OC_CONTINUE = 0x0900,
 321         OC_EXIT = 0x0a00,       OC_NEXT = 0x0b00,       OC_NEXTFILE = 0x0c00,
 322         OC_TEST = 0x0d00,       OC_WALKNEXT = 0x0e00,
 323
 324         OC_BINARY = 0x1000,     OC_BUILTIN = 0x1100,    OC_COLON = 0x1200,
 325         OC_COMMA = 0x1300,      OC_COMPARE = 0x1400,    OC_CONCAT = 0x1500,
 326         OC_FBLTIN = 0x1600,     OC_FIELD = 0x1700,      OC_FNARG = 0x1800,
 327         OC_FUNC = 0x1900,       OC_GETLINE = 0x1a00,    OC_IN = 0x1b00,
 328         OC_LAND = 0x1c00,       OC_LOR = 0x1d00,        OC_MATCH = 0x1e00,
 329         OC_MOVE = 0x1f00,       OC_PGETLINE = 0x2000,   OC_REGEXP = 0x2100,
 330         OC_REPLACE = 0x2200,    OC_RETURN = 0x2300,     OC_SPRINTF = 0x2400,
 331         OC_TERNARY = 0x2500,    OC_UNARY = 0x2600,      OC_VAR = 0x2700,
 332         OC_DONE = 0x2800,
 333
 334         ST_IF = 0x3000,         ST_DO = 0x3100,         ST_FOR = 0x3200,
 335         ST_WHILE = 0x3300
 336 };
 337
 338 /* simple builtins */
 339 enum {
 340         F_in,   F_rn,   F_co,   F_ex,   F_lg,   F_si,   F_sq,   F_sr,
 341         F_ti,   F_le,   F_sy,   F_ff,   F_cl
 342 };
 343
 344 /* builtins */
 345 enum {
 346         B_a2,   B_ix,   B_ma,   B_sp,   B_ss,   B_ti,   B_mt,   B_lo,   B_up,
 347         B_ge,   B_gs,   B_su,
 348         B_an,   B_co,   B_ls,   B_or,   B_rs,   B_xo,
 349 };
 350
 351 /* tokens and their corresponding info values */
 352
 353 #define NTC     "\377"  /* switch to next token class (tc<<1) */
 354 #define NTCC    '\377'
 355
 356 static const char tokenlist[] ALIGN1 =
 357         "\1("         NTC                                   /* TC_SEQSTART */
 358         "\1)"         NTC                                   /* TC_SEQTERM */
 359         "\1/"         NTC                                   /* TC_REGEXP */
 360         "\2>>"        "\1>"         "\1|"       NTC         /* TC_OUTRDR */
 361         "\2++"        "\2--"        NTC                     /* TC_UOPPOST */
 362         "\2++"        "\2--"        "\1$"       NTC         /* TC_UOPPRE1 */
 363         "\2=="        "\1="         "\2+="      "\2-="      /* TC_BINOPX */
 364         "\2*="        "\2/="        "\2%="      "\2^="
 365         "\1+"         "\1-"         "\3**="     "\2**"
 366         "\1/"         "\1%"         "\1^"       "\1*"
 367         "\2!="        "\2>="        "\2<="      "\1>"
 368         "\1<"         "\2!~"        "\1~"       "\2&&"
 369         "\2||"        "\1?"         "\1:"       NTC
 370         "\2in"        NTC                                   /* TC_IN */
 371         "\1,"         NTC                                   /* TC_COMMA */
 372         "\1|"         NTC                                   /* TC_PIPE */
 373         "\1+"         "\1-"         "\1!"       NTC         /* TC_UOPPRE2 */
 374         "\1]"         NTC                                   /* TC_ARRTERM */
 375         "\1{"         NTC                                   /* TC_GRPSTART */
 376         "\1}"         NTC                                   /* TC_GRPTERM */
 377         "\1;"         NTC                                   /* TC_SEMICOL */
 378         "\1\n"        NTC                                   /* TC_NEWLINE */
 379         "\2if"        "\2do"        "\3for"     "\5break"   /* TC_STATX */
 380         "\10continue" "\6delete"    "\5print"
 381         "\6printf"    "\4next"      "\10nextfile"
 382         "\6return"    "\4exit"      NTC
 383         "\5while"     NTC                                   /* TC_WHILE */
 384         "\4else"      NTC                                   /* TC_ELSE */
 385         "\3and"       "\5compl"     "\6lshift"  "\2or"      /* TC_BUILTIN */
 386         "\6rshift"    "\3xor"
 387         "\5close"     "\6system"    "\6fflush"  "\5atan2"
 388         "\3cos"       "\3exp"       "\3int"     "\3log"
 389         "\4rand"      "\3sin"       "\4sqrt"    "\5srand"
 390         "\6gensub"    "\4gsub"      "\5index"   /* "\6length" was here */
 391         "\5match"     "\5split"     "\7sprintf" "\3sub"
 392         "\6substr"    "\7systime"   "\10strftime" "\6mktime"
 393         "\7tolower"   "\7toupper"   NTC
 394         "\6length"    NTC                                   /* TC_LENGTH */
 395         "\7getline"   NTC                                   /* TC_GETLINE */
 396         "\4func"      "\10function" NTC                     /* TC_FUNCDECL */
 397         "\5BEGIN"     NTC                                   /* TC_BEGIN */
 398         "\3END"                                             /* TC_END */
 399         /* compiler adds trailing "\0" */
 400         ;
 401
 402 #define OC_B  OC_BUILTIN
 403
 404 static const uint32_t tokeninfo[] = {
 405         0,
 406         0,
 407         OC_REGEXP,
 408         xS|'a',                  xS|'w',                  xS|'|',
 409         OC_UNARY|xV|P(9)|'p',    OC_UNARY|xV|P(9)|'m',
 410         OC_UNARY|xV|P(9)|'P',    OC_UNARY|xV|P(9)|'M',    OC_FIELD|xV|P(5),
 411         OC_COMPARE|VV|P(39)|5,   OC_MOVE|VV|P(74),        OC_REPLACE|NV|P(74)|'+', OC_REPLACE|NV|P(74)|'-',
 412         OC_REPLACE|NV|P(74)|'*', OC_REPLACE|NV|P(74)|'/', OC_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&',
 413         OC_BINARY|NV|P(29)|'+',  OC_BINARY|NV|P(29)|'-',  OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&',
 414         OC_BINARY|NV|P(25)|'/',  OC_BINARY|NV|P(25)|'%',  OC_BINARY|NV|P(15)|'&',  OC_BINARY|NV|P(25)|'*',
 415         OC_COMPARE|VV|P(39)|4,   OC_COMPARE|VV|P(39)|3,   OC_COMPARE|VV|P(39)|0,   OC_COMPARE|VV|P(39)|1,
 416         OC_COMPARE|VV|P(39)|2,   OC_MATCH|Sx|P(45)|'!',   OC_MATCH|Sx|P(45)|'~',   OC_LAND|Vx|P(55),
 417         OC_LOR|Vx|P(59),         OC_TERNARY|Vx|P(64)|'?', OC_COLON|xx|P(67)|':',
 418         OC_IN|SV|P(49), /* TC_IN */
 419         OC_COMMA|SS|P(80),
 420         OC_PGETLINE|SV|P(37),
 421         OC_UNARY|xV|P(19)|'+',   OC_UNARY|xV|P(19)|'-',   OC_UNARY|xV|P(19)|'!',
 422         0, /* ] */
 423         0,
 424         0,
 425         0,
 426         0, /* \n */
 427         ST_IF,        ST_DO,        ST_FOR,      OC_BREAK,
 428         OC_CONTINUE,  OC_DELETE|Vx, OC_PRINT,
 429         OC_PRINTF,    OC_NEXT,      OC_NEXTFILE,
 430         OC_RETURN|Vx, OC_EXIT|Nx,
 431         ST_WHILE,
 432         0, /* else */
 433         OC_B|B_an|P(0x83), OC_B|B_co|P(0x41), OC_B|B_ls|P(0x83), OC_B|B_or|P(0x83),
 434         OC_B|B_rs|P(0x83), OC_B|B_xo|P(0x83),
 435         OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83),
 436         OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg,
 437         OC_FBLTIN|F_rn,    OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr,
 438         OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), /* OC_FBLTIN|Sx|F_le, was here */
 439         OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF,        OC_B|B_su|P(0xb6),
 440         OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti,    OC_B|B_ti|P(0x0b), OC_B|B_mt|P(0x0b),
 441         OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49),
 442         OC_FBLTIN|Sx|F_le, /* TC_LENGTH */
 443         OC_GETLINE|SV|P(0),
 444         0,                 0,
 445         0,
 446         0 /* TC_END */
 447 };
 448
 449 /* internal variable names and their initial values       */
 450 /* asterisk marks SPECIAL vars; $ is just no-named Field0 */
 451 enum {
 452         CONVFMT,    OFMT,       FS,         OFS,
 453         ORS,        RS,         RT,         FILENAME,
 454         SUBSEP,     F0,         ARGIND,     ARGC,
 455         ARGV,       ERRNO,      FNR,        NR,
 456         NF,         IGNORECASE, ENVIRON,    NUM_INTERNAL_VARS
 457 };
 458
 459 static const char vNames[] ALIGN1 =
 460         "CONVFMT\0" "OFMT\0"    "FS\0*"     "OFS\0"
 461         "ORS\0"     "RS\0*"     "RT\0"      "FILENAME\0"
 462         "SUBSEP\0"  "$\0*"      "ARGIND\0"  "ARGC\0"
 463         "ARGV\0"    "ERRNO\0"   "FNR\0"     "NR\0"
 464         "NF\0*"     "IGNORECASE\0*" "ENVIRON\0" "\0";
 465
 466 static const char vValues[] ALIGN1 =
 467         "%.6g\0"    "%.6g\0"    " \0"       " \0"
 468         "\n\0"      "\n\0"      "\0"        "\0"
 469         "\034\0"    "\0"        "\377";
 470
 471 /* hash size may grow to these values */
 472 #define FIRST_PRIME 61
 473 static const uint16_t PRIMES[] ALIGN2 = { 251, 1021, 4093, 16381, 65521 };
 474
 475
 476 /* Globals. Split in two parts so that first one is addressed
 477  * with (mostly short) negative offsets.
 478  * NB: it's unsafe to put members of type "double"
 479  * into globals2 (gcc may fail to align them).
 480  */
 481 struct globals {
 482         double t_double;
 483         chain beginseq, mainseq, endseq;
 484         chain *seq;
 485         node *break_ptr, *continue_ptr;
 486         rstream *iF;
 487         xhash *vhash, *ahash, *fdhash, *fnhash;
 488         const char *g_progname;
 489         int g_lineno;
 490         int nfields;
 491         int maxfields; /* used in fsrealloc() only */
 492         var *Fields;
 493         nvblock *g_cb;
 494         char *g_pos;
 495         char *g_buf;
 496         smallint icase;
 497         smallint exiting;
 498         smallint nextrec;
 499         smallint nextfile;
 500         smallint is_f0_split;
 501         smallint t_rollback;
 502 };
 503 struct globals2 {
 504         uint32_t t_info; /* often used */
 505         uint32_t t_tclass;
 506         char *t_string;
 507         int t_lineno;
 508
 509         var *intvar[NUM_INTERNAL_VARS]; /* often used */
 510
 511         /* former statics from various functions */
 512         char *split_f0__fstrings;
 513
 514         uint32_t next_token__save_tclass;
 515         uint32_t next_token__save_info;
 516         uint32_t next_token__ltclass;
 517         smallint next_token__concat_inserted;
 518
 519         smallint next_input_file__files_happen;
 520         rstream next_input_file__rsm;
 521
 522         var *evaluate__fnargs;
 523         unsigned evaluate__seed;
 524         regex_t evaluate__sreg;
 525
 526         var ptest__v;
 527
 528         tsplitter exec_builtin__tspl;
 529
 530         /* biggest and least used members go last */
 531         tsplitter fsplitter, rsplitter;
 532 };
 533 #define G1 (ptr_to_globals[-1])
 534 #define G (*(struct globals2 *)ptr_to_globals)
 535 /* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */
 536 /*char G1size[sizeof(G1)]; - 0x74 */
 537 /*char Gsize[sizeof(G)]; - 0x1c4 */
 538 /* Trying to keep most of members accessible with short offsets: */
 539 /*char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */
 540 #define t_double     (G1.t_double    )
 541 #define beginseq     (G1.beginseq    )
 542 #define mainseq      (G1.mainseq     )
 543 #define endseq       (G1.endseq      )
 544 #define seq          (G1.seq         )
 545 #define break_ptr    (G1.break_ptr   )
 546 #define continue_ptr (G1.continue_ptr)
 547 #define iF           (G1.iF          )
 548 #define vhash        (G1.vhash       )
 549 #define ahash        (G1.ahash       )
 550 #define fdhash       (G1.fdhash      )
 551 #define fnhash       (G1.fnhash      )
 552 #define g_progname   (G1.g_progname  )
 553 #define g_lineno     (G1.g_lineno    )
 554 #define nfields      (G1.nfields     )
 555 #define maxfields    (G1.maxfields   )
 556 #define Fields       (G1.Fields      )
 557 #define g_cb         (G1.g_cb        )
 558 #define g_pos        (G1.g_pos       )
 559 #define g_buf        (G1.g_buf       )
 560 #define icase        (G1.icase       )
 561 #define exiting      (G1.exiting     )
 562 #define nextrec      (G1.nextrec     )
 563 #define nextfile     (G1.nextfile    )
 564 #define is_f0_split  (G1.is_f0_split )
 565 #define t_rollback   (G1.t_rollback  )
 566 #define t_info       (G.t_info      )
 567 #define t_tclass     (G.t_tclass    )
 568 #define t_string     (G.t_string    )
 569 #define t_lineno     (G.t_lineno    )
 570 #define intvar       (G.intvar      )
 571 #define fsplitter    (G.fsplitter   )
 572 #define rsplitter    (G.rsplitter   )
 573 #define INIT_G() do { \
 574         SET_PTR_TO_GLOBALS((char*)xzalloc(sizeof(G1)+sizeof(G)) + sizeof(G1)); \
 575         G.next_token__ltclass = TC_OPTERM; \
 576         G.evaluate__seed = 1; \
 577 } while (0)
 578
 579
 580 /* function prototypes */
 581 static void handle_special(var *);
 582 static node *parse_expr(uint32_t);
 583 static void chain_group(void);
 584 static var *evaluate(node *, var *);
 585 static rstream *next_input_file(void);
 586 static int fmt_num(char *, int, const char *, double, int);
 587 static int awk_exit(int) NORETURN;
 588
 589 /* ---- error handling ---- */
 590
 591 static const char EMSG_INTERNAL_ERROR[] ALIGN1 = "Internal error";
 592 static const char EMSG_UNEXP_EOS[] ALIGN1 = "Unexpected end of string";
 593 static const char EMSG_UNEXP_TOKEN[] ALIGN1 = "Unexpected token";
 594 static const char EMSG_DIV_BY_ZERO[] ALIGN1 = "Division by zero";
 595 static const char EMSG_INV_FMT[] ALIGN1 = "Invalid format specifier";
 596 static const char EMSG_TOO_FEW_ARGS[] ALIGN1 = "Too few arguments for builtin";
 597 static const char EMSG_NOT_ARRAY[] ALIGN1 = "Not an array";
 598 static const char EMSG_POSSIBLE_ERROR[] ALIGN1 = "Possible syntax error";
 599 static const char EMSG_UNDEF_FUNC[] ALIGN1 = "Call to undefined function";
 600 static const char EMSG_NO_MATH[] ALIGN1 = "Math support is not compiled in";
 601 static const char EMSG_NEGATIVE_FIELD[] ALIGN1 = "Access to negative field";
 602
 603 static void zero_out_var(var *vp)
 604 {
 605         memset(vp, 0, sizeof(*vp));
 606 }
 607
 608 static void syntax_error(const char *message) NORETURN;
 609 static void syntax_error(const char *message)
 610 {
 611         bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message);
 612 }
 613
 614 /* ---- hash stuff ---- */
 615
 616 static unsigned hashidx(const char *name)
 617 {
 618         unsigned idx = 0;
 619
 620         while (*name)
 621                 idx = *name++ + (idx << 6) - idx;
 622         return idx;
 623 }
 624
 625 /* create new hash */
 626 static xhash *hash_init(void)
 627 {
 628         xhash *newhash;
 629
 630         newhash = xzalloc(sizeof(*newhash));
 631         newhash->csize = FIRST_PRIME;
 632         newhash->items = xzalloc(FIRST_PRIME * sizeof(newhash->items[0]));
 633
 634         return newhash;
 635 }
 636
 637 /* find item in hash, return ptr to data, NULL if not found */
 638 static void *hash_search(xhash *hash, const char *name)
 639 {
 640         hash_item *hi;
 641
 642         hi = hash->items[hashidx(name) % hash->csize];
 643         while (hi) {
 644                 if (strcmp(hi->name, name) == 0)
 645                         return &hi->data;
 646                 hi = hi->next;
 647         }
 648         return NULL;
 649 }
 650
 651 /* grow hash if it becomes too big */
 652 static void hash_rebuild(xhash *hash)
 653 {
 654         unsigned newsize, i, idx;
 655         hash_item **newitems, *hi, *thi;
 656
 657         if (hash->nprime == ARRAY_SIZE(PRIMES))
 658                 return;
 659
 660         newsize = PRIMES[hash->nprime++];
 661         newitems = xzalloc(newsize * sizeof(newitems[0]));
 662
 663         for (i = 0; i < hash->csize; i++) {
 664                 hi = hash->items[i];
 665                 while (hi) {
 666                         thi = hi;
 667                         hi = thi->next;
 668                         idx = hashidx(thi->name) % newsize;
 669                         thi->next = newitems[idx];
 670                         newitems[idx] = thi;
 671                 }
 672         }
 673
 674         free(hash->items);
 675         hash->csize = newsize;
 676         hash->items = newitems;
 677 }
 678
 679 /* find item in hash, add it if necessary. Return ptr to data */
 680 static void *hash_find(xhash *hash, const char *name)
 681 {
 682         hash_item *hi;
 683         unsigned idx;
 684         int l;
 685
 686         hi = hash_search(hash, name);
 687         if (!hi) {
 688                 if (++hash->nel / hash->csize > 10)
 689                         hash_rebuild(hash);
 690
 691                 l = strlen(name) + 1;
 692                 hi = xzalloc(sizeof(*hi) + l);
 693                 strcpy(hi->name, name);
 694
 695                 idx = hashidx(name) % hash->csize;
 696                 hi->next = hash->items[idx];
 697                 hash->items[idx] = hi;
 698                 hash->glen += l;
 699         }
 700         return &hi->data;
 701 }
 702
 703 #define findvar(hash, name) ((var*)    hash_find((hash), (name)))
 704 #define newvar(name)        ((var*)    hash_find(vhash, (name)))
 705 #define newfile(name)       ((rstream*)hash_find(fdhash, (name)))
 706 #define newfunc(name)       ((func*)   hash_find(fnhash, (name)))
 707
 708 static void hash_remove(xhash *hash, const char *name)
 709 {
 710         hash_item *hi, **phi;
 711
 712         phi = &hash->items[hashidx(name) % hash->csize];
 713         while (*phi) {
 714                 hi = *phi;
 715                 if (strcmp(hi->name, name) == 0) {
 716                         hash->glen -= (strlen(name) + 1);
 717                         hash->nel--;
 718                         *phi = hi->next;
 719                         free(hi);
 720                         break;
 721                 }
 722                 phi = &hi->next;
 723         }
 724 }
 725
 726 /* ------ some useful functions ------ */
 727
 728 static char *skip_spaces(char *p)
 729 {
 730         while (1) {
 731                 if (*p == '\\' && p[1] == '\n') {
 732                         p++;
 733                         t_lineno++;
 734                 } else if (*p != ' ' && *p != '\t') {
 735                         break;
 736                 }
 737                 p++;
 738         }
 739         return p;
 740 }
 741
 742 /* returns old *s, advances *s past word and terminating NUL */
 743 static char *nextword(char **s)
 744 {
 745         char *p = *s;
 746         while (*(*s)++ != '\0')
 747                 continue;
 748         return p;
 749 }
 750
 751 static char nextchar(char **s)
 752 {
 753         char c, *pps;
 754
 755         c = *(*s)++;
 756         pps = *s;
 757         if (c == '\\')
 758                 c = bb_process_escape_sequence((const char**)s);
 759         /* Example awk statement:
 760          * s = "abc\"def"
 761          * we must treat \" as "
 762          */
 763         if (c == '\\' && *s == pps) { /* unrecognized \z? */
 764                 c = *(*s); /* yes, fetch z */
 765                 if (c)
 766                         (*s)++; /* advance unless z = NUL */
 767         }
 768         return c;
 769 }
 770
 771 /* TODO: merge with strcpy_and_process_escape_sequences()?
 772  */
 773 static void unescape_string_in_place(char *s1)
 774 {
 775         char *s = s1;
 776         while ((*s1 = nextchar(&s)) != '\0')
 777                 s1++;
 778 }
 779
 780 static ALWAYS_INLINE int isalnum_(int c)
 781 {
 782         return (isalnum(c) || c == '_');
 783 }
 784
 785 static double my_strtod(char **pp)
 786 {
 787         char *cp = *pp;
 788         if (ENABLE_DESKTOP && cp[0] == '0') {
 789                 /* Might be hex or octal integer: 0x123abc or 07777 */
 790                 char c = (cp[1] | 0x20);
 791                 if (c == 'x' || isdigit(cp[1])) {
 792                         unsigned long long ull = strtoull(cp, pp, 0);
 793                         if (c == 'x')
 794                                 return ull;
 795                         c = **pp;
 796                         if (!isdigit(c) && c != '.')
 797                                 return ull;
 798                         /* else: it may be a floating number. Examples:
 799                          * 009.123 (*pp points to '9')
 800                          * 000.123 (*pp points to '.')
 801                          * fall through to strtod.
 802                          */
 803                 }
 804         }
 805         return strtod(cp, pp);
 806 }
 807
 808 /* -------- working with variables (set/get/copy/etc) -------- */
 809
 810 static xhash *iamarray(var *v)
 811 {
 812         var *a = v;
 813
 814         while (a->type & VF_CHILD)
 815                 a = a->x.parent;
 816
 817         if (!(a->type & VF_ARRAY)) {
 818                 a->type |= VF_ARRAY;
 819                 a->x.array = hash_init();
 820         }
 821         return a->x.array;
 822 }
 823
 824 static void clear_array(xhash *array)
 825 {
 826         unsigned i;
 827         hash_item *hi, *thi;
 828
 829         for (i = 0; i < array->csize; i++) {
 830                 hi = array->items[i];
 831                 while (hi) {
 832                         thi = hi;
 833                         hi = hi->next;
 834                         free(thi->data.v.string);
 835                         free(thi);
 836                 }
 837                 array->items[i] = NULL;
 838         }
 839         array->glen = array->nel = 0;
 840 }
 841
 842 /* clear a variable */
 843 static var *clrvar(var *v)
 844 {
 845         if (!(v->type & VF_FSTR))
 846                 free(v->string);
 847
 848         v->type &= VF_DONTTOUCH;
 849         v->type |= VF_DIRTY;
 850         v->string = NULL;
 851         return v;
 852 }
 853
 854 /* assign string value to variable */
 855 static var *setvar_p(var *v, char *value)
 856 {
 857         clrvar(v);
 858         v->string = value;
 859         handle_special(v);
 860         return v;
 861 }
 862
 863 /* same as setvar_p but make a copy of string */
 864 static var *setvar_s(var *v, const char *value)
 865 {
 866         return setvar_p(v, (value && *value) ? xstrdup(value) : NULL);
 867 }
 868
 869 /* same as setvar_s but sets USER flag */
 870 static var *setvar_u(var *v, const char *value)
 871 {
 872         v = setvar_s(v, value);
 873         v->type |= VF_USER;
 874         return v;
 875 }
 876
 877 /* set array element to user string */
 878 static void setari_u(var *a, int idx, const char *s)
 879 {
 880         var *v;
 881
 882         v = findvar(iamarray(a), itoa(idx));
 883         setvar_u(v, s);
 884 }
 885
 886 /* assign numeric value to variable */
 887 static var *setvar_i(var *v, double value)
 888 {
 889         clrvar(v);
 890         v->type |= VF_NUMBER;
 891         v->number = value;
 892         handle_special(v);
 893         return v;
 894 }
 895
 896 static const char *getvar_s(var *v)
 897 {
 898         /* if v is numeric and has no cached string, convert it to string */
 899         if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) {
 900                 fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE);
 901                 v->string = xstrdup(g_buf);
 902                 v->type |= VF_CACHED;
 903         }
 904         return (v->string == NULL) ? "" : v->string;
 905 }
 906
 907 static double getvar_i(var *v)
 908 {
 909         char *s;
 910
 911         if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) {
 912                 v->number = 0;
 913                 s = v->string;
 914                 if (s && *s) {
 915                         debug_printf_eval("getvar_i: '%s'->", s);
 916                         v->number = my_strtod(&s);
 917                         debug_printf_eval("%f (s:'%s')\n", v->number, s);
 918                         if (v->type & VF_USER) {
 919                                 s = skip_spaces(s);
 920                                 if (*s != '\0')
 921                                         v->type &= ~VF_USER;
 922                         }
 923                 } else {
 924                         debug_printf_eval("getvar_i: '%s'->zero\n", s);
 925                         v->type &= ~VF_USER;
 926                 }
 927                 v->type |= VF_CACHED;
 928         }
 929         debug_printf_eval("getvar_i: %f\n", v->number);
 930         return v->number;
 931 }
 932
 933 /* Used for operands of bitwise ops */
 934 static unsigned long getvar_i_int(var *v)
 935 {
 936         double d = getvar_i(v);
 937
 938         /* Casting doubles to longs is undefined for values outside
 939          * of target type range. Try to widen it as much as possible */
 940         if (d >= 0)
 941                 return (unsigned long)d;
 942         /* Why? Think about d == -4294967295.0 (assuming 32bit longs) */
 943         return - (long) (unsigned long) (-d);
 944 }
 945
 946 static var *copyvar(var *dest, const var *src)
 947 {
 948         if (dest != src) {
 949                 clrvar(dest);
 950                 dest->type |= (src->type & ~(VF_DONTTOUCH | VF_FSTR));
 951                 debug_printf_eval("copyvar: number:%f string:'%s'\n", src->number, src->string);
 952                 dest->number = src->number;
 953                 if (src->string)
 954                         dest->string = xstrdup(src->string);
 955         }
 956         handle_special(dest);
 957         return dest;
 958 }
 959
 960 static var *incvar(var *v)
 961 {
 962         return setvar_i(v, getvar_i(v) + 1.0);
 963 }
 964
 965 /* return true if v is number or numeric string */
 966 static int is_numeric(var *v)
 967 {
 968         getvar_i(v);
 969         return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY));
 970 }
 971
 972 /* return 1 when value of v corresponds to true, 0 otherwise */
 973 static int istrue(var *v)
 974 {
 975         if (is_numeric(v))
 976                 return (v->number != 0);
 977         return (v->string && v->string[0]);
 978 }
 979
 980 /* temporary variables allocator. Last allocated should be first freed */
 981 static var *nvalloc(int n)
 982 {
 983         nvblock *pb = NULL;
 984         var *v, *r;
 985         int size;
 986
 987         while (g_cb) {
 988                 pb = g_cb;
 989                 if ((g_cb->pos - g_cb->nv) + n <= g_cb->size)
 990                         break;
 991                 g_cb = g_cb->next;
 992         }
 993
 994         if (!g_cb) {
 995                 size = (n <= MINNVBLOCK) ? MINNVBLOCK : n;
 996                 g_cb = xzalloc(sizeof(nvblock) + size * sizeof(var));
 997                 g_cb->size = size;
 998                 g_cb->pos = g_cb->nv;
 999                 g_cb->prev = pb;
1000                 /*g_cb->next = NULL; - xzalloc did it */
1001                 if (pb)
1002                         pb->next = g_cb;
1003         }
1004
1005         v = r = g_cb->pos;
1006         g_cb->pos += n;
1007
1008         while (v < g_cb->pos) {
1009                 v->type = 0;
1010                 v->string = NULL;
1011                 v++;
1012         }
1013
1014         return r;
1015 }
1016
1017 static void nvfree(var *v)
1018 {
1019         var *p;
1020
1021         if (v < g_cb->nv || v >= g_cb->pos)
1022                 syntax_error(EMSG_INTERNAL_ERROR);
1023
1024         for (p = v; p < g_cb->pos; p++) {
1025                 if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) {
1026                         clear_array(iamarray(p));
1027                         free(p->x.array->items);
1028                         free(p->x.array);
1029                 }
1030                 if (p->type & VF_WALK) {
1031                         walker_list *n;
1032                         walker_list *w = p->x.walker;
1033                         debug_printf_walker("nvfree: freeing walker @%p\n", &p->x.walker);
1034                         p->x.walker = NULL;
1035                         while (w) {
1036                                 n = w->prev;
1037                                 debug_printf_walker(" free(%p)\n", w);
1038                                 free(w);
1039                                 w = n;
1040                         }
1041                 }
1042                 clrvar(p);
1043         }
1044
1045         g_cb->pos = v;
1046         while (g_cb->prev && g_cb->pos == g_cb->nv) {
1047                 g_cb = g_cb->prev;
1048         }
1049 }
1050
1051 /* ------- awk program text parsing ------- */
1052
1053 /* Parse next token pointed by global pos, place results into global ttt.
1054  * If token isn't expected, give away. Return token class
1055  */
1056 static uint32_t next_token(uint32_t expected)
1057 {
1058 #define concat_inserted (G.next_token__concat_inserted)
1059 #define save_tclass     (G.next_token__save_tclass)
1060 #define save_info       (G.next_token__save_info)
1061 /* Initialized to TC_OPTERM: */
1062 #define ltclass         (G.next_token__ltclass)
1063
1064         char *p, *s;
1065         const char *tl;
1066         uint32_t tc;
1067         const uint32_t *ti;
1068
1069         if (t_rollback) {
1070                 t_rollback = FALSE;
1071         } else if (concat_inserted) {
1072                 concat_inserted = FALSE;
1073                 t_tclass = save_tclass;
1074                 t_info = save_info;
1075         } else {
1076                 p = g_pos;
1077  readnext:
1078                 p = skip_spaces(p);
1079                 g_lineno = t_lineno;
1080                 if (*p == '#')
1081                         while (*p != '\n' && *p != '\0')
1082                                 p++;
1083
1084                 if (*p == '\n')
1085                         t_lineno++;
1086
1087                 if (*p == '\0') {
1088                         tc = TC_EOF;
1089                         debug_printf_parse("%s: token found: TC_EOF\n", __func__);
1090                 } else if (*p == '\"') {
1091                         /* it's a string */
1092                         t_string = s = ++p;
1093                         while (*p != '\"') {
1094                                 char *pp;
1095                                 if (*p == '\0' || *p == '\n')
1096                                         syntax_error(EMSG_UNEXP_EOS);
1097                                 pp = p;
1098                                 *s++ = nextchar(&pp);
1099                                 p = pp;
1100                         }
1101                         p++;
1102                         *s = '\0';
1103                         tc = TC_STRING;
1104                         debug_printf_parse("%s: token found:'%s' TC_STRING\n", __func__, t_string);
1105                 } else if ((expected & TC_REGEXP) && *p == '/') {
1106                         /* it's regexp */
1107                         t_string = s = ++p;
1108                         while (*p != '/') {
1109                                 if (*p == '\0' || *p == '\n')
1110                                         syntax_error(EMSG_UNEXP_EOS);
1111                                 *s = *p++;
1112                                 if (*s++ == '\\') {
1113                                         char *pp = p;
1114                                         s[-1] = bb_process_escape_sequence((const char **)&pp);
1115                                         if (*p == '\\')
1116                                                 *s++ = '\\';
1117                                         if (pp == p)
1118                                                 *s++ = *p++;
1119                                         else
1120                                                 p = pp;
1121                                 }
1122                         }
1123                         p++;
1124                         *s = '\0';
1125                         tc = TC_REGEXP;
1126                         debug_printf_parse("%s: token found:'%s' TC_REGEXP\n", __func__, t_string);
1127
1128                 } else if (*p == '.' || isdigit(*p)) {
1129                         /* it's a number */
1130                         char *pp = p;
1131                         t_double = my_strtod(&pp);
1132                         p = pp;
1133                         if (*p == '.')
1134                                 syntax_error(EMSG_UNEXP_TOKEN);
1135                         tc = TC_NUMBER;
1136                         debug_printf_parse("%s: token found:%f TC_NUMBER\n", __func__, t_double);
1137                 } else {
1138                         /* search for something known */
1139                         tl = tokenlist;
1140                         tc = 0x00000001;
1141                         ti = tokeninfo;
1142                         while (*tl) {
1143                                 int l = (unsigned char) *tl++;
1144                                 if (l == (unsigned char) NTCC) {
1145                                         tc <<= 1;
1146                                         continue;
1147                                 }
1148                                 /* if token class is expected,
1149                                  * token matches,
1150                                  * and it's not a longer word,
1151                                  */
1152                                 if ((tc & (expected | TC_WORD | TC_NEWLINE))
1153                                  && strncmp(p, tl, l) == 0
1154                                  && !((tc & TC_WORD) && isalnum_(p[l]))
1155                                 ) {
1156                                         /* then this is what we are looking for */
1157                                         t_info = *ti;
1158                                         debug_printf_parse("%s: token found:'%.*s' t_info:%x\n", __func__, l, p, t_info);
1159                                         p += l;
1160                                         goto token_found;
1161                                 }
1162                                 ti++;
1163                                 tl += l;
1164                         }
1165                         /* not a known token */
1166
1167                         /* is it a name? (var/array/function) */
1168                         if (!isalnum_(*p))
1169                                 syntax_error(EMSG_UNEXP_TOKEN); /* no */
1170                         /* yes */
1171                         t_string = --p;
1172                         while (isalnum_(*++p)) {
1173                                 p[-1] = *p;
1174                         }
1175                         p[-1] = '\0';
1176                         tc = TC_VARIABLE;
1177                         /* also consume whitespace between functionname and bracket */
1178                         if (!(expected & TC_VARIABLE) || (expected & TC_ARRAY))
1179                                 p = skip_spaces(p);
1180                         if (*p == '(') {
1181                                 tc = TC_FUNCTION;
1182                                 debug_printf_parse("%s: token found:'%s' TC_FUNCTION\n", __func__, t_string);
1183                         } else {
1184                                 if (*p == '[') {
1185                                         p++;
1186                                         tc = TC_ARRAY;
1187                                         debug_printf_parse("%s: token found:'%s' TC_ARRAY\n", __func__, t_string);
1188                                 } else
1189                                         debug_printf_parse("%s: token found:'%s' TC_VARIABLE\n", __func__, t_string);
1190                         }
1191                 }
1192  token_found:
1193                 g_pos = p;
1194
1195                 /* skipping newlines in some cases */
1196                 if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE))
1197                         goto readnext;
1198
1199                 /* insert concatenation operator when needed */
1200                 if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) {
1201                         concat_inserted = TRUE;
1202                         save_tclass = tc;
1203                         save_info = t_info;
1204                         tc = TC_BINOP;
1205                         t_info = OC_CONCAT | SS | P(35);
1206                 }
1207
1208                 t_tclass = tc;
1209         }
1210         ltclass = t_tclass;
1211
1212         /* Are we ready for this? */
1213         if (!(ltclass & expected)) {
1214                 syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ?
1215                                 EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN);
1216         }
1217
1218         return ltclass;
1219 #undef concat_inserted
1220 #undef save_tclass
1221 #undef save_info
1222 #undef ltclass
1223 }
1224
1225 static void rollback_token(void)
1226 {
1227         t_rollback = TRUE;
1228 }
1229
1230 static node *new_node(uint32_t info)
1231 {
1232         node *n;
1233
1234         n = xzalloc(sizeof(node));
1235         n->info = info;
1236         n->lineno = g_lineno;
1237         return n;
1238 }
1239
1240 static void mk_re_node(const char *s, node *n, regex_t *re)
1241 {
1242         n->info = OC_REGEXP;
1243         n->l.re = re;
1244         n->r.ire = re + 1;
1245         xregcomp(re, s, REG_EXTENDED);
1246         xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE);
1247 }
1248
1249 static node *condition(void)
1250 {
1251         next_token(TC_SEQSTART);
1252         return parse_expr(TC_SEQTERM);
1253 }
1254
1255 /* parse expression terminated by given argument, return ptr
1256  * to built subtree. Terminator is eaten by parse_expr */
1257 static node *parse_expr(uint32_t iexp)
1258 {
1259         node sn;
1260         node *cn = &sn;
1261         node *vn, *glptr;
1262         uint32_t tc, xtc;
1263         var *v;
1264
1265         debug_printf_parse("%s(%x)\n", __func__, iexp);
1266
1267         sn.info = PRIMASK;
1268         sn.r.n = glptr = NULL;
1269         xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp;
1270
1271         while (!((tc = next_token(xtc)) & iexp)) {
1272
1273                 if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) {
1274                         /* input redirection (<) attached to glptr node */
1275                         debug_printf_parse("%s: input redir\n", __func__);
1276                         cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37));
1277                         cn->a.n = glptr;
1278                         xtc = TC_OPERAND | TC_UOPPRE;
1279                         glptr = NULL;
1280
1281                 } else if (tc & (TC_BINOP | TC_UOPPOST)) {
1282                         debug_printf_parse("%s: TC_BINOP | TC_UOPPOST\n", __func__);
1283                         /* for binary and postfix-unary operators, jump back over
1284                          * previous operators with higher priority */
1285                         vn = cn;
1286                         while (((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2))
1287                             || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON))
1288                         ) {
1289                                 vn = vn->a.n;
1290                         }
1291                         if ((t_info & OPCLSMASK) == OC_TERNARY)
1292                                 t_info += P(6);
1293                         cn = vn->a.n->r.n = new_node(t_info);
1294                         cn->a.n = vn->a.n;
1295                         if (tc & TC_BINOP) {
1296                                 cn->l.n = vn;
1297                                 xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
1298                                 if ((t_info & OPCLSMASK) == OC_PGETLINE) {
1299                                         /* it's a pipe */
1300                                         next_token(TC_GETLINE);
1301                                         /* give maximum priority to this pipe */
1302                                         cn->info &= ~PRIMASK;
1303                                         xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1304                                 }
1305                         } else {
1306                                 cn->r.n = vn;
1307                                 xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1308                         }
1309                         vn->a.n = cn;
1310
1311                 } else {
1312                         debug_printf_parse("%s: other\n", __func__);
1313                         /* for operands and prefix-unary operators, attach them
1314                          * to last node */
1315                         vn = cn;
1316                         cn = vn->r.n = new_node(t_info);
1317                         cn->a.n = vn;
1318                         xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
1319                         if (tc & (TC_OPERAND | TC_REGEXP)) {
1320                                 debug_printf_parse("%s: TC_OPERAND | TC_REGEXP\n", __func__);
1321                                 xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp;
1322                                 /* one should be very careful with switch on tclass -
1323                                  * only simple tclasses should be used! */
1324                                 switch (tc) {
1325                                 case TC_VARIABLE:
1326                                 case TC_ARRAY:
1327                                         debug_printf_parse("%s: TC_VARIABLE | TC_ARRAY\n", __func__);
1328                                         cn->info = OC_VAR;
1329                                         v = hash_search(ahash, t_string);
1330                                         if (v != NULL) {
1331                                                 cn->info = OC_FNARG;
1332                                                 cn->l.aidx = v->x.aidx;
1333                                         } else {
1334                                                 cn->l.v = newvar(t_string);
1335                                         }
1336                                         if (tc & TC_ARRAY) {
1337                                                 cn->info |= xS;
1338                                                 cn->r.n = parse_expr(TC_ARRTERM);
1339                                         }
1340                                         break;
1341
1342                                 case TC_NUMBER:
1343                                 case TC_STRING:
1344                                         debug_printf_parse("%s: TC_NUMBER | TC_STRING\n", __func__);
1345                                         cn->info = OC_VAR;
1346                                         v = cn->l.v = xzalloc(sizeof(var));
1347                                         if (tc & TC_NUMBER)
1348                                                 setvar_i(v, t_double);
1349                                         else
1350                                                 setvar_s(v, t_string);
1351                                         break;
1352
1353                                 case TC_REGEXP:
1354                                         debug_printf_parse("%s: TC_REGEXP\n", __func__);
1355                                         mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2));
1356                                         break;
1357
1358                                 case TC_FUNCTION:
1359                                         debug_printf_parse("%s: TC_FUNCTION\n", __func__);
1360                                         cn->info = OC_FUNC;
1361                                         cn->r.f = newfunc(t_string);
1362                                         cn->l.n = condition();
1363                                         break;
1364
1365                                 case TC_SEQSTART:
1366                                         debug_printf_parse("%s: TC_SEQSTART\n", __func__);
1367                                         cn = vn->r.n = parse_expr(TC_SEQTERM);
1368                                         if (!cn)
1369                                                 syntax_error("Empty sequence");
1370                                         cn->a.n = vn;
1371                                         break;
1372
1373                                 case TC_GETLINE:
1374                                         debug_printf_parse("%s: TC_GETLINE\n", __func__);
1375                                         glptr = cn;
1376                                         xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
1377                                         break;
1378
1379                                 case TC_BUILTIN:
1380                                         debug_printf_parse("%s: TC_BUILTIN\n", __func__);
1381                                         cn->l.n = condition();
1382                                         break;
1383
1384                                 case TC_LENGTH:
1385                                         debug_printf_parse("%s: TC_LENGTH\n", __func__);
1386                                         next_token(TC_SEQSTART | TC_OPTERM | TC_GRPTERM);
1387                                         rollback_token();
1388                                         if (t_tclass & TC_SEQSTART) {
1389                                                 /* It was a "(" token. Handle just like TC_BUILTIN */
1390                                                 cn->l.n = condition();
1391                                         }
1392                                         break;
1393                                 }
1394                         }
1395                 }
1396         }
1397
1398         debug_printf_parse("%s() returns %p\n", __func__, sn.r.n);
1399         return sn.r.n;
1400 }
1401
1402 /* add node to chain. Return ptr to alloc'd node */
1403 static node *chain_node(uint32_t info)
1404 {
1405         node *n;
1406
1407         if (!seq->first)
1408                 seq->first = seq->last = new_node(0);
1409
1410         if (seq->programname != g_progname) {
1411                 seq->programname = g_progname;
1412                 n = chain_node(OC_NEWSOURCE);
1413                 n->l.new_progname = xstrdup(g_progname);
1414         }
1415
1416         n = seq->last;
1417         n->info = info;
1418         seq->last = n->a.n = new_node(OC_DONE);
1419
1420         return n;
1421 }
1422
1423 static void chain_expr(uint32_t info)
1424 {
1425         node *n;
1426
1427         n = chain_node(info);
1428         n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM);
1429         if (t_tclass & TC_GRPTERM)
1430                 rollback_token();
1431 }
1432
1433 static node *chain_loop(node *nn)
1434 {
1435         node *n, *n2, *save_brk, *save_cont;
1436
1437         save_brk = break_ptr;
1438         save_cont = continue_ptr;
1439
1440         n = chain_node(OC_BR | Vx);
1441         continue_ptr = new_node(OC_EXEC);
1442         break_ptr = new_node(OC_EXEC);
1443         chain_group();
1444         n2 = chain_node(OC_EXEC | Vx);
1445         n2->l.n = nn;
1446         n2->a.n = n;
1447         continue_ptr->a.n = n2;
1448         break_ptr->a.n = n->r.n = seq->last;
1449
1450         continue_ptr = save_cont;
1451         break_ptr = save_brk;
1452
1453         return n;
1454 }
1455
1456 /* parse group and attach it to chain */
1457 static void chain_group(void)
1458 {
1459         uint32_t c;
1460         node *n, *n2, *n3;
1461
1462         do {
1463                 c = next_token(TC_GRPSEQ);
1464         } while (c & TC_NEWLINE);
1465
1466         if (c & TC_GRPSTART) {
1467                 debug_printf_parse("%s: TC_GRPSTART\n", __func__);
1468                 while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) {
1469                         debug_printf_parse("%s: !TC_GRPTERM\n", __func__);
1470                         if (t_tclass & TC_NEWLINE)
1471                                 continue;
1472                         rollback_token();
1473                         chain_group();
1474                 }
1475                 debug_printf_parse("%s: TC_GRPTERM\n", __func__);
1476         } else if (c & (TC_OPSEQ | TC_OPTERM)) {
1477                 debug_printf_parse("%s: TC_OPSEQ | TC_OPTERM\n", __func__);
1478                 rollback_token();
1479                 chain_expr(OC_EXEC | Vx);
1480         } else {
1481                 /* TC_STATEMNT */
1482                 debug_printf_parse("%s: TC_STATEMNT(?)\n", __func__);
1483                 switch (t_info & OPCLSMASK) {
1484                 case ST_IF:
1485                         debug_printf_parse("%s: ST_IF\n", __func__);
1486                         n = chain_node(OC_BR | Vx);
1487                         n->l.n = condition();
1488                         chain_group();
1489                         n2 = chain_node(OC_EXEC);
1490                         n->r.n = seq->last;
1491                         if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE) == TC_ELSE) {
1492                                 chain_group();
1493                                 n2->a.n = seq->last;
1494                         } else {
1495                                 rollback_token();
1496                         }
1497                         break;
1498
1499                 case ST_WHILE:
1500                         debug_printf_parse("%s: ST_WHILE\n", __func__);
1501                         n2 = condition();
1502                         n = chain_loop(NULL);
1503                         n->l.n = n2;
1504                         break;
1505
1506                 case ST_DO:
1507                         debug_printf_parse("%s: ST_DO\n", __func__);
1508                         n2 = chain_node(OC_EXEC);
1509                         n = chain_loop(NULL);
1510                         n2->a.n = n->a.n;
1511                         next_token(TC_WHILE);
1512                         n->l.n = condition();
1513                         break;
1514
1515                 case ST_FOR:
1516                         debug_printf_parse("%s: ST_FOR\n", __func__);
1517                         next_token(TC_SEQSTART);
1518                         n2 = parse_expr(TC_SEMICOL | TC_SEQTERM);
1519                         if (t_tclass & TC_SEQTERM) {    /* for-in */
1520                                 if (!n2 || (n2->info & OPCLSMASK) != OC_IN)
1521                                         syntax_error(EMSG_UNEXP_TOKEN);
1522                                 n = chain_node(OC_WALKINIT | VV);
1523                                 n->l.n = n2->l.n;
1524                                 n->r.n = n2->r.n;
1525                                 n = chain_loop(NULL);
1526                                 n->info = OC_WALKNEXT | Vx;
1527                                 n->l.n = n2->l.n;
1528                         } else {                        /* for (;;) */
1529                                 n = chain_node(OC_EXEC | Vx);
1530                                 n->l.n = n2;
1531                                 n2 = parse_expr(TC_SEMICOL);
1532                                 n3 = parse_expr(TC_SEQTERM);
1533                                 n = chain_loop(n3);
1534                                 n->l.n = n2;
1535                                 if (!n2)
1536                                         n->info = OC_EXEC;
1537                         }
1538                         break;
1539
1540                 case OC_PRINT:
1541                 case OC_PRINTF:
1542                         debug_printf_parse("%s: OC_PRINT[F]\n", __func__);
1543                         n = chain_node(t_info);
1544                         n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM);
1545                         if (t_tclass & TC_OUTRDR) {
1546                                 n->info |= t_info;
1547                                 n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM);
1548                         }
1549                         if (t_tclass & TC_GRPTERM)
1550                                 rollback_token();
1551                         break;
1552
1553                 case OC_BREAK:
1554                         debug_printf_parse("%s: OC_BREAK\n", __func__);
1555                         n = chain_node(OC_EXEC);
1556                         n->a.n = break_ptr;
1557                         chain_expr(t_info);
1558                         break;
1559
1560                 case OC_CONTINUE:
1561                         debug_printf_parse("%s: OC_CONTINUE\n", __func__);
1562                         n = chain_node(OC_EXEC);
1563                         n->a.n = continue_ptr;
1564                         chain_expr(t_info);
1565                         break;
1566
1567                 /* delete, next, nextfile, return, exit */
1568                 default:
1569                         debug_printf_parse("%s: default\n", __func__);
1570                         chain_expr(t_info);
1571                 }
1572         }
1573 }
1574
1575 static void parse_program(char *p)
1576 {
1577         uint32_t tclass;
1578         node *cn;
1579         func *f;
1580         var *v;
1581
1582         g_pos = p;
1583         t_lineno = 1;
1584         while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART |
1585                         TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) {
1586
1587                 if (tclass & TC_OPTERM) {
1588                         debug_printf_parse("%s: TC_OPTERM\n", __func__);
1589                         continue;
1590                 }
1591
1592                 seq = &mainseq;
1593                 if (tclass & TC_BEGIN) {
1594                         debug_printf_parse("%s: TC_BEGIN\n", __func__);
1595                         seq = &beginseq;
1596                         chain_group();
1597                 } else if (tclass & TC_END) {
1598                         debug_printf_parse("%s: TC_END\n", __func__);
1599                         seq = &endseq;
1600                         chain_group();
1601                 } else if (tclass & TC_FUNCDECL) {
1602                         debug_printf_parse("%s: TC_FUNCDECL\n", __func__);
1603                         next_token(TC_FUNCTION);
1604                         g_pos++;
1605                         f = newfunc(t_string);
1606                         f->body.first = NULL;
1607                         f->nargs = 0;
1608                         while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) {
1609                                 v = findvar(ahash, t_string);
1610                                 v->x.aidx = f->nargs++;
1611
1612                                 if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM)
1613                                         break;
1614                         }
1615                         seq = &f->body;
1616                         chain_group();
1617                         clear_array(ahash);
1618                 } else if (tclass & TC_OPSEQ) {
1619                         debug_printf_parse("%s: TC_OPSEQ\n", __func__);
1620                         rollback_token();
1621                         cn = chain_node(OC_TEST);
1622                         cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART);
1623                         if (t_tclass & TC_GRPSTART) {
1624                                 debug_printf_parse("%s: TC_GRPSTART\n", __func__);
1625                                 rollback_token();
1626                                 chain_group();
1627                         } else {
1628                                 debug_printf_parse("%s: !TC_GRPSTART\n", __func__);
1629                                 chain_node(OC_PRINT);
1630                         }
1631                         cn->r.n = mainseq.last;
1632                 } else /* if (tclass & TC_GRPSTART) */ {
1633                         debug_printf_parse("%s: TC_GRPSTART(?)\n", __func__);
1634                         rollback_token();
1635                         chain_group();
1636                 }
1637         }
1638         debug_printf_parse("%s: TC_EOF\n", __func__);
1639 }
1640
1641
1642 /* -------- program execution part -------- */
1643
1644 static node *mk_splitter(const char *s, tsplitter *spl)
1645 {
1646         regex_t *re, *ire;
1647         node *n;
1648
1649         re = &spl->re[0];
1650         ire = &spl->re[1];
1651         n = &spl->n;
1652         if ((n->info & OPCLSMASK) == OC_REGEXP) {
1653                 regfree(re);
1654                 regfree(ire); // TODO: nuke ire, use re+1?
1655         }
1656         if (s[0] && s[1]) { /* strlen(s) > 1 */
1657                 mk_re_node(s, n, re);
1658         } else {
1659                 n->info = (uint32_t) s[0];
1660         }
1661
1662         return n;
1663 }
1664
1665 /* use node as a regular expression. Supplied with node ptr and regex_t
1666  * storage space. Return ptr to regex (if result points to preg, it should
1667  * be later regfree'd manually
1668  */
1669 static regex_t *as_regex(node *op, regex_t *preg)
1670 {
1671         int cflags;
1672         var *v;
1673         const char *s;
1674
1675         if ((op->info & OPCLSMASK) == OC_REGEXP) {
1676                 return icase ? op->r.ire : op->l.re;
1677         }
1678         v = nvalloc(1);
1679         s = getvar_s(evaluate(op, v));
1680
1681         cflags = icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED;
1682         /* Testcase where REG_EXTENDED fails (unpaired '{'):
1683          * echo Hi | awk 'gsub("@(samp|code|file)\{","");'
1684          * gawk 3.1.5 eats this. We revert to ~REG_EXTENDED
1685          * (maybe gsub is not supposed to use REG_EXTENDED?).
1686          */
1687         if (regcomp(preg, s, cflags)) {
1688                 cflags &= ~REG_EXTENDED;
1689                 xregcomp(preg, s, cflags);
1690         }
1691         nvfree(v);
1692         return preg;
1693 }
1694
1695 /* gradually increasing buffer.
1696  * note that we reallocate even if n == old_size,
1697  * and thus there is at least one extra allocated byte.
1698  */
1699 static char* qrealloc(char *b, int n, int *size)
1700 {
1701         if (!b || n >= *size) {
1702                 *size = n + (n>>1) + 80;
1703                 b = xrealloc(b, *size);
1704         }
1705         return b;
1706 }
1707
1708 /* resize field storage space */
1709 static void fsrealloc(int size)
1710 {
1711         int i;
1712
1713         if (size >= maxfields) {
1714                 i = maxfields;
1715                 maxfields = size + 16;
1716                 Fields = xrealloc(Fields, maxfields * sizeof(Fields[0]));
1717                 for (; i < maxfields; i++) {
1718                         Fields[i].type = VF_SPECIAL;
1719                         Fields[i].string = NULL;
1720                 }
1721         }
1722         /* if size < nfields, clear extra field variables */
1723         for (i = size; i < nfields; i++) {
1724                 clrvar(Fields + i);
1725         }
1726         nfields = size;
1727 }
1728
1729 static int awk_split(const char *s, node *spl, char **slist)
1730 {
1731         int l, n;
1732         char c[4];
1733         char *s1;
1734         regmatch_t pmatch[2]; // TODO: why [2]? [1] is enough...
1735
1736         /* in worst case, each char would be a separate field */
1737         *slist = s1 = xzalloc(strlen(s) * 2 + 3);
1738         strcpy(s1, s);
1739
1740         c[0] = c[1] = (char)spl->info;
1741         c[2] = c[3] = '\0';
1742         if (*getvar_s(intvar[RS]) == '\0')
1743                 c[2] = '\n';
1744
1745         n = 0;
1746         if ((spl->info & OPCLSMASK) == OC_REGEXP) {  /* regex split */
1747                 if (!*s)
1748                         return n; /* "": zero fields */
1749                 n++; /* at least one field will be there */
1750                 do {
1751                         l = strcspn(s, c+2); /* len till next NUL or \n */
1752                         if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0
1753                          && pmatch[0].rm_so <= l
1754                         ) {
1755                                 l = pmatch[0].rm_so;
1756                                 if (pmatch[0].rm_eo == 0) {
1757                                         l++;
1758                                         pmatch[0].rm_eo++;
1759                                 }
1760                                 n++; /* we saw yet another delimiter */
1761                         } else {
1762                                 pmatch[0].rm_eo = l;
1763                                 if (s[l])
1764                                         pmatch[0].rm_eo++;
1765                         }
1766                         memcpy(s1, s, l);
1767                         /* make sure we remove *all* of the separator chars */
1768                         do {
1769                                 s1[l] = '\0';
1770                         } while (++l < pmatch[0].rm_eo);
1771                         nextword(&s1);
1772                         s += pmatch[0].rm_eo;
1773                 } while (*s);
1774                 return n;
1775         }
1776         if (c[0] == '\0') {  /* null split */
1777                 while (*s) {
1778                         *s1++ = *s++;
1779                         *s1++ = '\0';
1780                         n++;
1781                 }
1782                 return n;
1783         }
1784         if (c[0] != ' ') {  /* single-character split */
1785                 if (icase) {
1786                         c[0] = toupper(c[0]);
1787                         c[1] = tolower(c[1]);
1788                 }
1789                 if (*s1)
1790                         n++;
1791                 while ((s1 = strpbrk(s1, c)) != NULL) {
1792                         *s1++ = '\0';
1793                         n++;
1794                 }
1795                 return n;
1796         }
1797         /* space split */
1798         while (*s) {
1799                 s = skip_whitespace(s);
1800                 if (!*s)
1801                         break;
1802                 n++;
1803                 while (*s && !isspace(*s))
1804                         *s1++ = *s++;
1805                 *s1++ = '\0';
1806         }
1807         return n;
1808 }
1809
1810 static void split_f0(void)
1811 {
1812 /* static char *fstrings; */
1813 #define fstrings (G.split_f0__fstrings)
1814
1815         int i, n;
1816         char *s;
1817
1818         if (is_f0_split)
1819                 return;
1820
1821         is_f0_split = TRUE;
1822         free(fstrings);
1823         fsrealloc(0);
1824         n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings);
1825         fsrealloc(n);
1826         s = fstrings;
1827         for (i = 0; i < n; i++) {
1828                 Fields[i].string = nextword(&s);
1829                 Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY);
1830         }
1831
1832         /* set NF manually to avoid side effects */
1833         clrvar(intvar[NF]);
1834         intvar[NF]->type = VF_NUMBER | VF_SPECIAL;
1835         intvar[NF]->number = nfields;
1836 #undef fstrings
1837 }
1838
1839 /* perform additional actions when some internal variables changed */
1840 static void handle_special(var *v)
1841 {
1842         int n;
1843         char *b;
1844         const char *sep, *s;
1845         int sl, l, len, i, bsize;
1846
1847         if (!(v->type & VF_SPECIAL))
1848                 return;
1849
1850         if (v == intvar[NF]) {
1851                 n = (int)getvar_i(v);
1852                 if (n < 0)
1853                         syntax_error("NF set to negative value");
1854                 fsrealloc(n);
1855
1856                 /* recalculate $0 */
1857                 sep = getvar_s(intvar[OFS]);
1858                 sl = strlen(sep);
1859                 b = NULL;
1860                 len = 0;
1861                 for (i = 0; i < n; i++) {
1862                         s = getvar_s(&Fields[i]);
1863                         l = strlen(s);
1864                         if (b) {
1865                                 memcpy(b+len, sep, sl);
1866                                 len += sl;
1867                         }
1868                         b = qrealloc(b, len+l+sl, &bsize);
1869                         memcpy(b+len, s, l);
1870                         len += l;
1871                 }
1872                 if (b)
1873                         b[len] = '\0';
1874                 setvar_p(intvar[F0], b);
1875                 is_f0_split = TRUE;
1876
1877         } else if (v == intvar[F0]) {
1878                 is_f0_split = FALSE;
1879
1880         } else if (v == intvar[FS]) {
1881                 /*
1882                  * The POSIX-2008 standard says that changing FS should have no effect on the
1883                  * current input line, but only on the next one. The language is:
1884                  *
1885                  * > Before the first reference to a field in the record is evaluated, the record
1886                  * > shall be split into fields, according to the rules in Regular Expressions,
1887                  * > using the value of FS that was current at the time the record was read.
1888                  *
1889                  * So, split up current line before assignment to FS:
1890                  */
1891                 split_f0();
1892
1893                 mk_splitter(getvar_s(v), &fsplitter);
1894         } else if (v == intvar[RS]) {
1895                 mk_splitter(getvar_s(v), &rsplitter);
1896         } else if (v == intvar[IGNORECASE]) {
1897                 icase = istrue(v);
1898         } else {                                /* $n */
1899                 n = getvar_i(intvar[NF]);
1900                 setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1);
1901                 /* right here v is invalid. Just to note... */
1902         }
1903 }
1904
1905 /* step through func/builtin/etc arguments */
1906 static node *nextarg(node **pn)
1907 {
1908         node *n;
1909
1910         n = *pn;
1911         if (n && (n->info & OPCLSMASK) == OC_COMMA) {
1912                 *pn = n->r.n;
1913                 n = n->l.n;
1914         } else {
1915                 *pn = NULL;
1916         }
1917         return n;
1918 }
1919
1920 static void hashwalk_init(var *v, xhash *array)
1921 {
1922         hash_item *hi;
1923         unsigned i;
1924         walker_list *w;
1925         walker_list *prev_walker;
1926
1927         if (v->type & VF_WALK) {
1928                 prev_walker = v->x.walker;
1929         } else {
1930                 v->type |= VF_WALK;
1931                 prev_walker = NULL;
1932         }
1933         debug_printf_walker("hashwalk_init: prev_walker:%p\n", prev_walker);
1934
1935         w = v->x.walker = xzalloc(sizeof(*w) + array->glen + 1); /* why + 1? */
1936         debug_printf_walker(" walker@%p=%p\n", &v->x.walker, w);
1937         w->cur = w->end = w->wbuf;
1938         w->prev = prev_walker;
1939         for (i = 0; i < array->csize; i++) {
1940                 hi = array->items[i];
1941                 while (hi) {
1942                         strcpy(w->end, hi->name);
1943                         nextword(&w->end);
1944                         hi = hi->next;
1945                 }
1946         }
1947 }
1948
1949 static int hashwalk_next(var *v)
1950 {
1951         walker_list *w = v->x.walker;
1952
1953         if (w->cur >= w->end) {
1954                 walker_list *prev_walker = w->prev;
1955
1956                 debug_printf_walker("end of iteration, free(walker@%p:%p), prev_walker:%p\n", &v->x.walker, w, prev_walker);
1957                 free(w);
1958                 v->x.walker = prev_walker;
1959                 return FALSE;
1960         }
1961
1962         setvar_s(v, nextword(&w->cur));
1963         return TRUE;
1964 }
1965
1966 /* evaluate node, return 1 when result is true, 0 otherwise */
1967 static int ptest(node *pattern)
1968 {
1969         /* ptest__v is "static": to save stack space? */
1970         return istrue(evaluate(pattern, &G.ptest__v));
1971 }
1972
1973 /* read next record from stream rsm into a variable v */
1974 static int awk_getline(rstream *rsm, var *v)
1975 {
1976         char *b;
1977         regmatch_t pmatch[2];
1978         int size, a, p, pp = 0;
1979         int fd, so, eo, r, rp;
1980         char c, *m, *s;
1981
1982         debug_printf_eval("entered %s()\n", __func__);
1983
1984         /* we're using our own buffer since we need access to accumulating
1985          * characters
1986          */
1987         fd = fileno(rsm->F);
1988         m = rsm->buffer;
1989         a = rsm->adv;
1990         p = rsm->pos;
1991         size = rsm->size;
1992         c = (char) rsplitter.n.info;
1993         rp = 0;
1994
1995         if (!m)
1996                 m = qrealloc(m, 256, &size);
1997
1998         do {
1999                 b = m + a;
2000                 so = eo = p;
2001                 r = 1;
2002                 if (p > 0) {
2003                         if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) {
2004                                 if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re,
2005                                                         b, 1, pmatch, 0) == 0) {
2006                                         so = pmatch[0].rm_so;
2007                                         eo = pmatch[0].rm_eo;
2008                                         if (b[eo] != '\0')
2009                                                 break;
2010                                 }
2011                         } else if (c != '\0') {
2012                                 s = strchr(b+pp, c);
2013                                 if (!s)
2014                                         s = memchr(b+pp, '\0', p - pp);
2015                                 if (s) {
2016                                         so = eo = s-b;
2017                                         eo++;
2018                                         break;
2019                                 }
2020                         } else {
2021                                 while (b[rp] == '\n')
2022                                         rp++;
2023                                 s = strstr(b+rp, "\n\n");
2024                                 if (s) {
2025                                         so = eo = s-b;
2026                                         while (b[eo] == '\n')
2027                                                 eo++;
2028                                         if (b[eo] != '\0')
2029                                                 break;
2030                                 }
2031                         }
2032                 }
2033
2034                 if (a > 0) {
2035                         memmove(m, m+a, p+1);
2036                         b = m;
2037                         a = 0;
2038                 }
2039
2040                 m = qrealloc(m, a+p+128, &size);
2041                 b = m + a;
2042                 pp = p;
2043                 p += safe_read(fd, b+p, size-p-1);
2044                 if (p < pp) {
2045                         p = 0;
2046                         r = 0;
2047                         setvar_i(intvar[ERRNO], errno);
2048                 }
2049                 b[p] = '\0';
2050
2051         } while (p > pp);
2052
2053         if (p == 0) {
2054                 r--;
2055         } else {
2056                 c = b[so]; b[so] = '\0';
2057                 setvar_s(v, b+rp);
2058                 v->type |= VF_USER;
2059                 b[so] = c;
2060                 c = b[eo]; b[eo] = '\0';
2061                 setvar_s(intvar[RT], b+so);
2062                 b[eo] = c;
2063         }
2064
2065         rsm->buffer = m;
2066         rsm->adv = a + eo;
2067         rsm->pos = p - eo;
2068         rsm->size = size;
2069
2070         debug_printf_eval("returning from %s(): %d\n", __func__, r);
2071
2072         return r;
2073 }
2074
2075 static int fmt_num(char *b, int size, const char *format, double n, int int_as_int)
2076 {
2077         int r = 0;
2078         char c;
2079         const char *s = format;
2080
2081         if (int_as_int && n == (long long)n) {
2082                 r = snprintf(b, size, "%lld", (long long)n);
2083         } else {
2084                 do { c = *s; } while (c && *++s);
2085                 if (strchr("diouxX", c)) {
2086                         r = snprintf(b, size, format, (int)n);
2087                 } else if (strchr("eEfgG", c)) {
2088                         r = snprintf(b, size, format, n);
2089                 } else {
2090                         syntax_error(EMSG_INV_FMT);
2091                 }
2092         }
2093         return r;
2094 }
2095
2096 /* formatted output into an allocated buffer, return ptr to buffer */
2097 static char *awk_printf(node *n)
2098 {
2099         char *b = NULL;
2100         char *fmt, *s, *f;
2101         const char *s1;
2102         int i, j, incr, bsize;
2103         char c, c1;
2104         var *v, *arg;
2105
2106         v = nvalloc(1);
2107         fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v)));
2108
2109         i = 0;
2110         while (*f) {
2111                 s = f;
2112                 while (*f && (*f != '%' || *++f == '%'))
2113                         f++;
2114                 while (*f && !isalpha(*f)) {
2115                         if (*f == '*')
2116                                 syntax_error("%*x formats are not supported");
2117                         f++;
2118                 }
2119
2120                 incr = (f - s) + MAXVARFMT;
2121                 b = qrealloc(b, incr + i, &bsize);
2122                 c = *f;
2123                 if (c != '\0')
2124                         f++;
2125                 c1 = *f;
2126                 *f = '\0';
2127                 arg = evaluate(nextarg(&n), v);
2128
2129                 j = i;
2130                 if (c == 'c' || !c) {
2131                         i += sprintf(b+i, s, is_numeric(arg) ?
2132                                         (char)getvar_i(arg) : *getvar_s(arg));
2133                 } else if (c == 's') {
2134                         s1 = getvar_s(arg);
2135                         b = qrealloc(b, incr+i+strlen(s1), &bsize);
2136                         i += sprintf(b+i, s, s1);
2137                 } else {
2138                         i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE);
2139                 }
2140                 *f = c1;
2141
2142                 /* if there was an error while sprintf, return value is negative */
2143                 if (i < j)
2144                         i = j;
2145         }
2146
2147         free(fmt);
2148         nvfree(v);
2149         b = xrealloc(b, i + 1);
2150         b[i] = '\0';
2151         return b;
2152 }
2153
2154 /* Common substitution routine.
2155  * Replace (nm)'th substring of (src) that matches (rn) with (repl),
2156  * store result into (dest), return number of substitutions.
2157  * If nm = 0, replace all matches.
2158  * If src or dst is NULL, use $0.
2159  * If subexp != 0, enable subexpression matching (\1-\9).
2160  */
2161 static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int subexp)
2162 {
2163         char *resbuf;
2164         const char *sp;
2165         int match_no, residx, replen, resbufsize;
2166         int regexec_flags;
2167         regmatch_t pmatch[10];
2168         regex_t sreg, *regex;
2169
2170         resbuf = NULL;
2171         residx = 0;
2172         match_no = 0;
2173         regexec_flags = 0;
2174         regex = as_regex(rn, &sreg);
2175         sp = getvar_s(src ? src : intvar[F0]);
2176         replen = strlen(repl);
2177         while (regexec(regex, sp, 10, pmatch, regexec_flags) == 0) {
2178                 int so = pmatch[0].rm_so;
2179                 int eo = pmatch[0].rm_eo;
2180
2181                 //bb_error_msg("match %u: [%u,%u] '%s'%p", match_no+1, so, eo, sp,sp);
2182                 resbuf = qrealloc(resbuf, residx + eo + replen, &resbufsize);
2183                 memcpy(resbuf + residx, sp, eo);
2184                 residx += eo;
2185                 if (++match_no >= nm) {
2186                         const char *s;
2187                         int nbs;
2188
2189                         /* replace */
2190                         residx -= (eo - so);
2191                         nbs = 0;
2192                         for (s = repl; *s; s++) {
2193                                 char c = resbuf[residx++] = *s;
2194                                 if (c == '\\') {
2195                                         nbs++;
2196                                         continue;
2197                                 }
2198                                 if (c == '&' || (subexp && c >= '0' && c <= '9')) {
2199                                         int j;
2200                                         residx -= ((nbs + 3) >> 1);
2201                                         j = 0;
2202                                         if (c != '&') {
2203                                                 j = c - '0';
2204                                                 nbs++;
2205                                         }
2206                                         if (nbs % 2) {
2207                                                 resbuf[residx++] = c;
2208                                         } else {
2209                                                 int n = pmatch[j].rm_eo - pmatch[j].rm_so;
2210                                                 resbuf = qrealloc(resbuf, residx + replen + n, &resbufsize);
2211                                                 memcpy(resbuf + residx, sp + pmatch[j].rm_so, n);
2212                                                 residx += n;
2213                                         }
2214                                 }
2215                                 nbs = 0;
2216                         }
2217                 }
2218
2219                 regexec_flags = REG_NOTBOL;
2220                 sp += eo;
2221                 if (match_no == nm)
2222                         break;
2223                 if (eo == so) {
2224                         /* Empty match (e.g. "b*" will match anywhere).
2225                          * Advance by one char. */
2226 //BUG (bug 1333):
2227 //gsub(/\<b*/,"") on "abc" will reach this point, advance to "bc"
2228 //... and will erroneously match "b" even though it is NOT at the word start.
2229 //we need REG_NOTBOW but it does not exist...
2230 //TODO: if EXTRA_COMPAT=y, use GNU matching and re_search,
2231 //it should be able to do it correctly.
2232                         /* Subtle: this is safe only because
2233                          * qrealloc allocated at least one extra byte */
2234                         resbuf[residx] = *sp;
2235                         if (*sp == '\0')
2236                                 goto ret;
2237                         sp++;
2238                         residx++;
2239                 }
2240         }
2241
2242         resbuf = qrealloc(resbuf, residx + strlen(sp), &resbufsize);
2243         strcpy(resbuf + residx, sp);
2244  ret:
2245         //bb_error_msg("end sp:'%s'%p", sp,sp);
2246         setvar_p(dest ? dest : intvar[F0], resbuf);
2247         if (regex == &sreg)
2248                 regfree(regex);
2249         return match_no;
2250 }
2251
2252 static NOINLINE int do_mktime(const char *ds)
2253 {
2254         struct tm then;
2255         int count;
2256
2257         /*memset(&then, 0, sizeof(then)); - not needed */
2258         then.tm_isdst = -1; /* default is unknown */
2259
2260         /* manpage of mktime says these fields are ints,
2261          * so we can sscanf stuff directly into them */
2262         count = sscanf(ds, "%u %u %u %u %u %u %d",
2263                 &then.tm_year, &then.tm_mon, &then.tm_mday,
2264                 &then.tm_hour, &then.tm_min, &then.tm_sec,
2265                 &then.tm_isdst);
2266
2267         if (count < 6
2268          || (unsigned)then.tm_mon < 1
2269          || (unsigned)then.tm_year < 1900
2270         ) {
2271                 return -1;
2272         }
2273
2274         then.tm_mon -= 1;
2275         then.tm_year -= 1900;
2276
2277         return mktime(&then);
2278 }
2279
2280 static NOINLINE var *exec_builtin(node *op, var *res)
2281 {
2282 #define tspl (G.exec_builtin__tspl)
2283
2284         var *tv;
2285         node *an[4];
2286         var *av[4];
2287         const char *as[4];
2288         regmatch_t pmatch[2];
2289         regex_t sreg, *re;
2290         node *spl;
2291         uint32_t isr, info;
2292         int nargs;
2293         time_t tt;
2294         int i, l, ll, n;
2295
2296         tv = nvalloc(4);
2297         isr = info = op->info;
2298         op = op->l.n;
2299
2300         av[2] = av[3] = NULL;
2301         for (i = 0; i < 4 && op; i++) {
2302                 an[i] = nextarg(&op);
2303                 if (isr & 0x09000000)
2304                         av[i] = evaluate(an[i], &tv[i]);
2305                 if (isr & 0x08000000)
2306                         as[i] = getvar_s(av[i]);
2307                 isr >>= 1;
2308         }
2309
2310         nargs = i;
2311         if ((uint32_t)nargs < (info >> 30))
2312                 syntax_error(EMSG_TOO_FEW_ARGS);
2313
2314         info &= OPNMASK;
2315         switch (info) {
2316
2317         case B_a2:
2318                 if (ENABLE_FEATURE_AWK_LIBM)
2319                         setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1])));
2320                 else
2321                         syntax_error(EMSG_NO_MATH);
2322                 break;
2323
2324         case B_sp: {
2325                 char *s, *s1;
2326
2327                 if (nargs > 2) {
2328                         spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ?
2329                                 an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl);
2330                 } else {
2331                         spl = &fsplitter.n;
2332                 }
2333
2334                 n = awk_split(as[0], spl, &s);
2335                 s1 = s;
2336                 clear_array(iamarray(av[1]));
2337                 for (i = 1; i <= n; i++)
2338                         setari_u(av[1], i, nextword(&s));
2339                 free(s1);
2340                 setvar_i(res, n);
2341                 break;
2342         }
2343
2344         case B_ss: {
2345                 char *s;
2346
2347                 l = strlen(as[0]);
2348                 i = getvar_i(av[1]) - 1;
2349                 if (i > l)
2350                         i = l;
2351                 if (i < 0)
2352                         i = 0;
2353                 n = (nargs > 2) ? getvar_i(av[2]) : l-i;
2354                 if (n < 0)
2355                         n = 0;
2356                 s = xstrndup(as[0]+i, n);
2357                 setvar_p(res, s);
2358                 break;
2359         }
2360
2361         /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5:
2362          * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */
2363         case B_an:
2364                 setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1]));
2365                 break;
2366
2367         case B_co:
2368                 setvar_i(res, ~getvar_i_int(av[0]));
2369                 break;
2370
2371         case B_ls:
2372                 setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1]));
2373                 break;
2374
2375         case B_or:
2376                 setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1]));
2377                 break;
2378
2379         case B_rs:
2380                 setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1]));
2381                 break;
2382
2383         case B_xo:
2384                 setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1]));
2385                 break;
2386
2387         case B_lo:
2388         case B_up: {
2389                 char *s, *s1;
2390                 s1 = s = xstrdup(as[0]);
2391                 while (*s1) {
2392                         //*s1 = (info == B_up) ? toupper(*s1) : tolower(*s1);
2393                         if ((unsigned char)((*s1 | 0x20) - 'a') <= ('z' - 'a'))
2394                                 *s1 = (info == B_up) ? (*s1 & 0xdf) : (*s1 | 0x20);
2395                         s1++;
2396                 }
2397                 setvar_p(res, s);
2398                 break;
2399         }
2400
2401         case B_ix:
2402                 n = 0;
2403                 ll = strlen(as[1]);
2404                 l = strlen(as[0]) - ll;
2405                 if (ll > 0 && l >= 0) {
2406                         if (!icase) {
2407                                 char *s = strstr(as[0], as[1]);
2408                                 if (s)
2409                                         n = (s - as[0]) + 1;
2410                         } else {
2411                                 /* this piece of code is terribly slow and
2412                                  * really should be rewritten
2413                                  */
2414                                 for (i = 0; i <= l; i++) {
2415                                         if (strncasecmp(as[0]+i, as[1], ll) == 0) {
2416                                                 n = i+1;
2417                                                 break;
2418                                         }
2419                                 }
2420                         }
2421                 }
2422                 setvar_i(res, n);
2423                 break;
2424
2425         case B_ti:
2426                 if (nargs > 1)
2427                         tt = getvar_i(av[1]);
2428                 else
2429                         time(&tt);
2430                 //s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y";
2431                 i = strftime(g_buf, MAXVARFMT,
2432                         ((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"),
2433                         localtime(&tt));
2434                 g_buf[i] = '\0';
2435                 setvar_s(res, g_buf);
2436                 break;
2437
2438         case B_mt:
2439                 setvar_i(res, do_mktime(as[0]));
2440                 break;
2441
2442         case B_ma:
2443                 re = as_regex(an[1], &sreg);
2444                 n = regexec(re, as[0], 1, pmatch, 0);
2445                 if (n == 0) {
2446                         pmatch[0].rm_so++;
2447                         pmatch[0].rm_eo++;
2448                 } else {
2449                         pmatch[0].rm_so = 0;
2450                         pmatch[0].rm_eo = -1;
2451                 }
2452                 setvar_i(newvar("RSTART"), pmatch[0].rm_so);
2453                 setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so);
2454                 setvar_i(res, pmatch[0].rm_so);
2455                 if (re == &sreg)
2456                         regfree(re);
2457                 break;
2458
2459         case B_ge:
2460                 awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE);
2461                 break;
2462
2463         case B_gs:
2464                 setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE));
2465                 break;
2466
2467         case B_su:
2468                 setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE));
2469                 break;
2470         }
2471
2472         nvfree(tv);
2473         return res;
2474 #undef tspl
2475 }
2476
2477 /*
2478  * Evaluate node - the heart of the program. Supplied with subtree
2479  * and place where to store result. returns ptr to result.
2480  */
2481 #define XC(n) ((n) >> 8)
2482
2483 static var *evaluate(node *op, var *res)
2484 {
2485 /* This procedure is recursive so we should count every byte */
2486 #define fnargs (G.evaluate__fnargs)
2487 /* seed is initialized to 1 */
2488 #define seed   (G.evaluate__seed)
2489 #define sreg   (G.evaluate__sreg)
2490
2491         var *v1;
2492
2493         if (!op)
2494                 return setvar_s(res, NULL);
2495
2496         debug_printf_eval("entered %s()\n", __func__);
2497
2498         v1 = nvalloc(2);
2499
2500         while (op) {
2501                 struct {
2502                         var *v;
2503                         const char *s;
2504                 } L = L; /* for compiler */
2505                 struct {
2506                         var *v;
2507                         const char *s;
2508                 } R = R;
2509                 double L_d = L_d;
2510                 uint32_t opinfo;
2511                 int opn;
2512                 node *op1;
2513
2514                 opinfo = op->info;
2515                 opn = (opinfo & OPNMASK);
2516                 g_lineno = op->lineno;
2517                 op1 = op->l.n;
2518                 debug_printf_eval("opinfo:%08x opn:%08x\n", opinfo, opn);
2519
2520                 /* "delete" is special:
2521                  * "delete array[var--]" must evaluate index expr only once,
2522                  * must not evaluate it in "execute inevitable things" part.
2523                  */
2524                 if (XC(opinfo & OPCLSMASK) == XC(OC_DELETE)) {
2525                         uint32_t info = op1->info & OPCLSMASK;
2526                         var *v;
2527
2528                         debug_printf_eval("DELETE\n");
2529                         if (info == OC_VAR) {
2530                                 v = op1->l.v;
2531                         } else if (info == OC_FNARG) {
2532                                 v = &fnargs[op1->l.aidx];
2533                         } else {
2534                                 syntax_error(EMSG_NOT_ARRAY);
2535                         }
2536                         if (op1->r.n) { /* array ref? */
2537                                 const char *s;
2538                                 s = getvar_s(evaluate(op1->r.n, v1));
2539                                 hash_remove(iamarray(v), s);
2540                         } else {
2541                                 clear_array(iamarray(v));
2542                         }
2543                         goto next;
2544                 }
2545
2546                 /* execute inevitable things */
2547                 if (opinfo & OF_RES1)
2548                         L.v = evaluate(op1, v1);
2549                 if (opinfo & OF_RES2)
2550                         R.v = evaluate(op->r.n, v1+1);
2551                 if (opinfo & OF_STR1) {
2552                         L.s = getvar_s(L.v);
2553                         debug_printf_eval("L.s:'%s'\n", L.s);
2554                 }
2555                 if (opinfo & OF_STR2) {
2556                         R.s = getvar_s(R.v);
2557                         debug_printf_eval("R.s:'%s'\n", R.s);
2558                 }
2559                 if (opinfo & OF_NUM1) {
2560                         L_d = getvar_i(L.v);
2561                         debug_printf_eval("L_d:%f\n", L_d);
2562                 }
2563
2564                 debug_printf_eval("switch(0x%x)\n", XC(opinfo & OPCLSMASK));
2565                 switch (XC(opinfo & OPCLSMASK)) {
2566
2567                 /* -- iterative node type -- */
2568
2569                 /* test pattern */
2570                 case XC( OC_TEST ):
2571                         if ((op1->info & OPCLSMASK) == OC_COMMA) {
2572                                 /* it's range pattern */
2573                                 if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) {
2574                                         op->info |= OF_CHECKED;
2575                                         if (ptest(op1->r.n))
2576                                                 op->info &= ~OF_CHECKED;
2577                                         op = op->a.n;
2578                                 } else {
2579                                         op = op->r.n;
2580                                 }
2581                         } else {
2582                                 op = ptest(op1) ? op->a.n : op->r.n;
2583                         }
2584                         break;
2585
2586                 /* just evaluate an expression, also used as unconditional jump */
2587                 case XC( OC_EXEC ):
2588                         break;
2589
2590                 /* branch, used in if-else and various loops */
2591                 case XC( OC_BR ):
2592                         op = istrue(L.v) ? op->a.n : op->r.n;
2593                         break;
2594
2595                 /* initialize for-in loop */
2596                 case XC( OC_WALKINIT ):
2597                         hashwalk_init(L.v, iamarray(R.v));
2598                         break;
2599
2600                 /* get next array item */
2601                 case XC( OC_WALKNEXT ):
2602                         op = hashwalk_next(L.v) ? op->a.n : op->r.n;
2603                         break;
2604
2605                 case XC( OC_PRINT ):
2606                 case XC( OC_PRINTF ): {
2607                         FILE *F = stdout;
2608
2609                         if (op->r.n) {
2610                                 rstream *rsm = newfile(R.s);
2611                                 if (!rsm->F) {
2612                                         if (opn == '|') {
2613                                                 rsm->F = popen(R.s, "w");
2614                                                 if (rsm->F == NULL)
2615                                                         bb_perror_msg_and_die("popen");
2616                                                 rsm->is_pipe = 1;
2617                                         } else {
2618                                                 rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
2619                                         }
2620                                 }
2621                                 F = rsm->F;
2622                         }
2623
2624                         if ((opinfo & OPCLSMASK) == OC_PRINT) {
2625                                 if (!op1) {
2626                                         fputs(getvar_s(intvar[F0]), F);
2627                                 } else {
2628                                         while (op1) {
2629                                                 var *v = evaluate(nextarg(&op1), v1);
2630                                                 if (v->type & VF_NUMBER) {
2631                                                         fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]),
2632                                                                         getvar_i(v), TRUE);
2633                                                         fputs(g_buf, F);
2634                                                 } else {
2635                                                         fputs(getvar_s(v), F);
2636                                                 }
2637
2638                                                 if (op1)
2639                                                         fputs(getvar_s(intvar[OFS]), F);
2640                                         }
2641                                 }
2642                                 fputs(getvar_s(intvar[ORS]), F);
2643
2644                         } else {        /* OC_PRINTF */
2645                                 char *s = awk_printf(op1);
2646                                 fputs(s, F);
2647                                 free(s);
2648                         }
2649                         fflush(F);
2650                         break;
2651                 }
2652
2653                 /* case XC( OC_DELETE ): - moved to happen before arg evaluation */
2654
2655                 case XC( OC_NEWSOURCE ):
2656                         g_progname = op->l.new_progname;
2657                         break;
2658
2659                 case XC( OC_RETURN ):
2660                         copyvar(res, L.v);
2661                         break;
2662
2663                 case XC( OC_NEXTFILE ):
2664                         nextfile = TRUE;
2665                 case XC( OC_NEXT ):
2666                         nextrec = TRUE;
2667                 case XC( OC_DONE ):
2668                         clrvar(res);
2669                         break;
2670
2671                 case XC( OC_EXIT ):
2672                         awk_exit(L_d);
2673
2674                 /* -- recursive node type -- */
2675
2676                 case XC( OC_VAR ):
2677                         debug_printf_eval("VAR\n");
2678                         L.v = op->l.v;
2679                         if (L.v == intvar[NF])
2680                                 split_f0();
2681                         goto v_cont;
2682
2683                 case XC( OC_FNARG ):
2684                         debug_printf_eval("FNARG[%d]\n", op->l.aidx);
2685                         L.v = &fnargs[op->l.aidx];
2686  v_cont:
2687                         res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v;
2688                         break;
2689
2690                 case XC( OC_IN ):
2691                         setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0);
2692                         break;
2693
2694                 case XC( OC_REGEXP ):
2695                         op1 = op;
2696                         L.s = getvar_s(intvar[F0]);
2697                         goto re_cont;
2698
2699                 case XC( OC_MATCH ):
2700                         op1 = op->r.n;
2701  re_cont:
2702                         {
2703                                 regex_t *re = as_regex(op1, &sreg);
2704                                 int i = regexec(re, L.s, 0, NULL, 0);
2705                                 if (re == &sreg)
2706                                         regfree(re);
2707                                 setvar_i(res, (i == 0) ^ (opn == '!'));
2708                         }
2709                         break;
2710
2711                 case XC( OC_MOVE ):
2712                         debug_printf_eval("MOVE\n");
2713                         /* if source is a temporary string, jusk relink it to dest */
2714 //Disabled: if R.v is numeric but happens to have cached R.v->string,
2715 //then L.v ends up being a string, which is wrong
2716 //                      if (R.v == v1+1 && R.v->string) {
2717 //                              res = setvar_p(L.v, R.v->string);
2718 //                              R.v->string = NULL;
2719 //                      } else {
2720                                 res = copyvar(L.v, R.v);
2721 //                      }
2722                         break;
2723
2724                 case XC( OC_TERNARY ):
2725                         if ((op->r.n->info & OPCLSMASK) != OC_COLON)
2726                                 syntax_error(EMSG_POSSIBLE_ERROR);
2727                         res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res);
2728                         break;
2729
2730                 case XC( OC_FUNC ): {
2731                         var *vbeg, *v;
2732                         const char *sv_progname;
2733
2734                         /* The body might be empty, still has to eval the args */
2735                         if (!op->r.n->info && !op->r.f->body.first)
2736                                 syntax_error(EMSG_UNDEF_FUNC);
2737
2738                         vbeg = v = nvalloc(op->r.f->nargs + 1);
2739                         while (op1) {
2740                                 var *arg = evaluate(nextarg(&op1), v1);
2741                                 copyvar(v, arg);
2742                                 v->type |= VF_CHILD;
2743                                 v->x.parent = arg;
2744                                 if (++v - vbeg >= op->r.f->nargs)
2745                                         break;
2746                         }
2747
2748                         v = fnargs;
2749                         fnargs = vbeg;
2750                         sv_progname = g_progname;
2751
2752                         res = evaluate(op->r.f->body.first, res);
2753
2754                         g_progname = sv_progname;
2755                         nvfree(fnargs);
2756                         fnargs = v;
2757
2758                         break;
2759                 }
2760
2761                 case XC( OC_GETLINE ):
2762                 case XC( OC_PGETLINE ): {
2763                         rstream *rsm;
2764                         int i;
2765
2766                         if (op1) {
2767                                 rsm = newfile(L.s);
2768                                 if (!rsm->F) {
2769                                         if ((opinfo & OPCLSMASK) == OC_PGETLINE) {
2770                                                 rsm->F = popen(L.s, "r");
2771                                                 rsm->is_pipe = TRUE;
2772                                         } else {
2773                                                 rsm->F = fopen_for_read(L.s);  /* not xfopen! */
2774                                         }
2775                                 }
2776                         } else {
2777                                 if (!iF)
2778                                         iF = next_input_file();
2779                                 rsm = iF;
2780                         }
2781
2782                         if (!rsm || !rsm->F) {
2783                                 setvar_i(intvar[ERRNO], errno);
2784                                 setvar_i(res, -1);
2785                                 break;
2786                         }
2787
2788                         if (!op->r.n)
2789                                 R.v = intvar[F0];
2790
2791                         i = awk_getline(rsm, R.v);
2792                         if (i > 0 && !op1) {
2793                                 incvar(intvar[FNR]);
2794                                 incvar(intvar[NR]);
2795                         }
2796                         setvar_i(res, i);
2797                         break;
2798                 }
2799
2800                 /* simple builtins */
2801                 case XC( OC_FBLTIN ): {
2802                         double R_d = R_d; /* for compiler */
2803
2804                         switch (opn) {
2805                         case F_in:
2806                                 R_d = (long long)L_d;
2807                                 break;
2808
2809                         case F_rn:
2810                                 R_d = (double)rand() / (double)RAND_MAX;
2811                                 break;
2812
2813                         case F_co:
2814                                 if (ENABLE_FEATURE_AWK_LIBM) {
2815                                         R_d = cos(L_d);
2816                                         break;
2817                                 }
2818
2819                         case F_ex:
2820                                 if (ENABLE_FEATURE_AWK_LIBM) {
2821                                         R_d = exp(L_d);
2822                                         break;
2823                                 }
2824
2825                         case F_lg:
2826                                 if (ENABLE_FEATURE_AWK_LIBM) {
2827                                         R_d = log(L_d);
2828                                         break;
2829                                 }
2830
2831                         case F_si:
2832                                 if (ENABLE_FEATURE_AWK_LIBM) {
2833                                         R_d = sin(L_d);
2834                                         break;
2835                                 }
2836
2837                         case F_sq:
2838                                 if (ENABLE_FEATURE_AWK_LIBM) {
2839                                         R_d = sqrt(L_d);
2840                                         break;
2841                                 }
2842
2843                                 syntax_error(EMSG_NO_MATH);
2844                                 break;
2845
2846                         case F_sr:
2847                                 R_d = (double)seed;
2848                                 seed = op1 ? (unsigned)L_d : (unsigned)time(NULL);
2849                                 srand(seed);
2850                                 break;
2851
2852                         case F_ti:
2853                                 R_d = time(NULL);
2854                                 break;
2855
2856                         case F_le:
2857                                 debug_printf_eval("length: L.s:'%s'\n", L.s);
2858                                 if (!op1) {
2859                                         L.s = getvar_s(intvar[F0]);
2860                                         debug_printf_eval("length: L.s='%s'\n", L.s);
2861                                 }
2862                                 else if (L.v->type & VF_ARRAY) {
2863                                         R_d = L.v->x.array->nel;
2864                                         debug_printf_eval("length: array_len:%d\n", L.v->x.array->nel);
2865                                         break;
2866                                 }
2867                                 R_d = strlen(L.s);
2868                                 break;
2869
2870                         case F_sy:
2871                                 fflush_all();
2872                                 R_d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
2873                                                 ? (system(L.s) >> 8) : 0;
2874                                 break;
2875
2876                         case F_ff:
2877                                 if (!op1) {
2878                                         fflush(stdout);
2879                                 } else if (L.s && *L.s) {
2880                                         rstream *rsm = newfile(L.s);
2881                                         fflush(rsm->F);
2882                                 } else {
2883                                         fflush_all();
2884                                 }
2885                                 break;
2886
2887                         case F_cl: {
2888                                 rstream *rsm;
2889                                 int err = 0;
2890                                 rsm = (rstream *)hash_search(fdhash, L.s);
2891                                 debug_printf_eval("OC_FBLTIN F_cl rsm:%p\n", rsm);
2892                                 if (rsm) {
2893                                         debug_printf_eval("OC_FBLTIN F_cl "
2894                                                 "rsm->is_pipe:%d, ->F:%p\n",
2895                                                 rsm->is_pipe, rsm->F);
2896                                         /* Can be NULL if open failed. Example:
2897                                          * getline line <"doesnt_exist";
2898                                          * close("doesnt_exist"); <--- here rsm->F is NULL
2899                                          */
2900                                         if (rsm->F)
2901                                                 err = rsm->is_pipe ? pclose(rsm->F) : fclose(rsm->F);
2902                                         free(rsm->buffer);
2903                                         hash_remove(fdhash, L.s);
2904                                 }
2905                                 if (err)
2906                                         setvar_i(intvar[ERRNO], errno);
2907                                 R_d = (double)err;
2908                                 break;
2909                         }
2910                         } /* switch */
2911                         setvar_i(res, R_d);
2912                         break;
2913                 }
2914
2915                 case XC( OC_BUILTIN ):
2916                         res = exec_builtin(op, res);
2917                         break;
2918
2919                 case XC( OC_SPRINTF ):
2920                         setvar_p(res, awk_printf(op1));
2921                         break;
2922
2923                 case XC( OC_UNARY ): {
2924                         double Ld, R_d;
2925
2926                         Ld = R_d = getvar_i(R.v);
2927                         switch (opn) {
2928                         case 'P':
2929                                 Ld = ++R_d;
2930                                 goto r_op_change;
2931                         case 'p':
2932                                 R_d++;
2933                                 goto r_op_change;
2934                         case 'M':
2935                                 Ld = --R_d;
2936                                 goto r_op_change;
2937                         case 'm':
2938                                 R_d--;
2939  r_op_change:
2940                                 setvar_i(R.v, R_d);
2941                                 break;
2942                         case '!':
2943                                 Ld = !istrue(R.v);
2944                                 break;
2945                         case '-':
2946                                 Ld = -R_d;
2947                                 break;
2948                         }
2949                         setvar_i(res, Ld);
2950                         break;
2951                 }
2952
2953                 case XC( OC_FIELD ): {
2954                         int i = (int)getvar_i(R.v);
2955                         if (i < 0)
2956                                 syntax_error(EMSG_NEGATIVE_FIELD);
2957                         if (i == 0) {
2958                                 res = intvar[F0];
2959                         } else {
2960                                 split_f0();
2961                                 if (i > nfields)
2962                                         fsrealloc(i);
2963                                 res = &Fields[i - 1];
2964                         }
2965                         break;
2966                 }
2967
2968                 /* concatenation (" ") and index joining (",") */
2969                 case XC( OC_CONCAT ):
2970                 case XC( OC_COMMA ): {
2971                         const char *sep = "";
2972                         if ((opinfo & OPCLSMASK) == OC_COMMA)
2973                                 sep = getvar_s(intvar[SUBSEP]);
2974                         setvar_p(res, xasprintf("%s%s%s", L.s, sep, R.s));
2975                         break;
2976                 }
2977
2978                 case XC( OC_LAND ):
2979                         setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0);
2980                         break;
2981
2982                 case XC( OC_LOR ):
2983                         setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n));
2984                         break;
2985
2986                 case XC( OC_BINARY ):
2987                 case XC( OC_REPLACE ): {
2988                         double R_d = getvar_i(R.v);
2989                         debug_printf_eval("BINARY/REPLACE: R_d:%f opn:%c\n", R_d, opn);
2990                         switch (opn) {
2991                         case '+':
2992                                 L_d += R_d;
2993                                 break;
2994                         case '-':
2995                                 L_d -= R_d;
2996                                 break;
2997                         case '*':
2998                                 L_d *= R_d;
2999                                 break;
3000                         case '/':
3001                                 if (R_d == 0)
3002                                         syntax_error(EMSG_DIV_BY_ZERO);
3003                                 L_d /= R_d;
3004                                 break;
3005                         case '&':
3006                                 if (ENABLE_FEATURE_AWK_LIBM)
3007                                         L_d = pow(L_d, R_d);
3008                                 else
3009                                         syntax_error(EMSG_NO_MATH);
3010                                 break;
3011                         case '%':
3012                                 if (R_d == 0)
3013                                         syntax_error(EMSG_DIV_BY_ZERO);
3014                                 L_d -= (long long)(L_d / R_d) * R_d;
3015                                 break;
3016                         }
3017                         debug_printf_eval("BINARY/REPLACE result:%f\n", L_d);
3018                         res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : L.v, L_d);
3019                         break;
3020                 }
3021
3022                 case XC( OC_COMPARE ): {
3023                         int i = i; /* for compiler */
3024                         double Ld;
3025
3026                         if (is_numeric(L.v) && is_numeric(R.v)) {
3027                                 Ld = getvar_i(L.v) - getvar_i(R.v);
3028                         } else {
3029                                 const char *l = getvar_s(L.v);
3030                                 const char *r = getvar_s(R.v);
3031                                 Ld = icase ? strcasecmp(l, r) : strcmp(l, r);
3032                         }
3033                         switch (opn & 0xfe) {
3034                         case 0:
3035                                 i = (Ld > 0);
3036                                 break;
3037                         case 2:
3038                                 i = (Ld >= 0);
3039                                 break;
3040                         case 4:
3041                                 i = (Ld == 0);
3042                                 break;
3043                         }
3044                         setvar_i(res, (i == 0) ^ (opn & 1));
3045                         break;
3046                 }
3047
3048                 default:
3049                         syntax_error(EMSG_POSSIBLE_ERROR);
3050                 } /* switch */
3051  next:
3052                 if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS)
3053                         op = op->a.n;
3054                 if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS)
3055                         break;
3056                 if (nextrec)
3057                         break;
3058         } /* while (op) */
3059
3060         nvfree(v1);
3061         debug_printf_eval("returning from %s(): %p\n", __func__, res);
3062         return res;
3063 #undef fnargs
3064 #undef seed
3065 #undef sreg
3066 }
3067
3068
3069 /* -------- main & co. -------- */
3070
3071 static int awk_exit(int r)
3072 {
3073         var tv;
3074         unsigned i;
3075         hash_item *hi;
3076
3077         zero_out_var(&tv);
3078
3079         if (!exiting) {
3080                 exiting = TRUE;
3081                 nextrec = FALSE;
3082                 evaluate(endseq.first, &tv);
3083         }
3084
3085         /* waiting for children */
3086         for (i = 0; i < fdhash->csize; i++) {
3087                 hi = fdhash->items[i];
3088                 while (hi) {
3089                         if (hi->data.rs.F && hi->data.rs.is_pipe)
3090                                 pclose(hi->data.rs.F);
3091                         hi = hi->next;
3092                 }
3093         }
3094
3095         exit(r);
3096 }
3097
3098 /* if expr looks like "var=value", perform assignment and return 1,
3099  * otherwise return 0 */
3100 static int is_assignment(const char *expr)
3101 {
3102         char *exprc, *val;
3103
3104         if (!isalnum_(*expr) || (val = strchr(expr, '=')) == NULL) {
3105                 return FALSE;
3106         }
3107
3108         exprc = xstrdup(expr);
3109         val = exprc + (val - expr);
3110         *val++ = '\0';
3111
3112         unescape_string_in_place(val);
3113         setvar_u(newvar(exprc), val);
3114         free(exprc);
3115         return TRUE;
3116 }
3117
3118 /* switch to next input file */
3119 static rstream *next_input_file(void)
3120 {
3121 #define rsm          (G.next_input_file__rsm)
3122 #define files_happen (G.next_input_file__files_happen)
3123
3124         FILE *F;
3125         const char *fname, *ind;
3126
3127         if (rsm.F)
3128                 fclose(rsm.F);
3129         rsm.F = NULL;
3130         rsm.pos = rsm.adv = 0;
3131
3132         for (;;) {
3133                 if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) {
3134                         if (files_happen)
3135                                 return NULL;
3136                         fname = "-";
3137                         F = stdin;
3138                         break;
3139                 }
3140                 ind = getvar_s(incvar(intvar[ARGIND]));
3141                 fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind));
3142                 if (fname && *fname && !is_assignment(fname)) {
3143                         F = xfopen_stdin(fname);
3144                         break;
3145                 }
3146         }
3147
3148         files_happen = TRUE;
3149         setvar_s(intvar[FILENAME], fname);
3150         rsm.F = F;
3151         return &rsm;
3152 #undef rsm
3153 #undef files_happen
3154 }
3155
3156 int awk_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
3157 int awk_main(int argc UNUSED_PARAM, char **argv)
3158 {
3159         unsigned opt;
3160         char *opt_F;
3161         llist_t *list_v = NULL;
3162         llist_t *list_f = NULL;
3163 #if ENABLE_FEATURE_AWK_GNU_EXTENSIONS
3164         llist_t *list_e = NULL;
3165 #endif
3166         int i, j;
3167         var *v;
3168         var tv;
3169         char **envp;
3170         char *vnames = (char *)vNames; /* cheat */
3171         char *vvalues = (char *)vValues;
3172
3173         INIT_G();
3174
3175         /* Undo busybox.c, or else strtod may eat ','! This breaks parsing:
3176          * $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */
3177         if (ENABLE_LOCALE_SUPPORT)
3178                 setlocale(LC_NUMERIC, "C");
3179
3180         zero_out_var(&tv);
3181
3182         /* allocate global buffer */
3183         g_buf = xmalloc(MAXVARFMT + 1);
3184
3185         vhash = hash_init();
3186         ahash = hash_init();
3187         fdhash = hash_init();
3188         fnhash = hash_init();
3189
3190         /* initialize variables */
3191         for (i = 0; *vnames; i++) {
3192                 intvar[i] = v = newvar(nextword(&vnames));
3193                 if (*vvalues != '\377')
3194                         setvar_s(v, nextword(&vvalues));
3195                 else
3196                         setvar_i(v, 0);
3197
3198                 if (*vnames == '*') {
3199                         v->type |= VF_SPECIAL;
3200                         vnames++;
3201                 }
3202         }
3203
3204         handle_special(intvar[FS]);
3205         handle_special(intvar[RS]);
3206
3207         newfile("/dev/stdin")->F = stdin;
3208         newfile("/dev/stdout")->F = stdout;
3209         newfile("/dev/stderr")->F = stderr;
3210
3211         /* Huh, people report that sometimes environ is NULL. Oh well. */
3212         if (environ) for (envp = environ; *envp; envp++) {
3213                 /* environ is writable, thus we don't strdup it needlessly */
3214                 char *s = *envp;
3215                 char *s1 = strchr(s, '=');
3216                 if (s1) {
3217                         *s1 = '\0';
3218                         /* Both findvar and setvar_u take const char*
3219                          * as 2nd arg -> environment is not trashed */
3220                         setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1);
3221                         *s1 = '=';
3222                 }
3223         }
3224         opt = getopt32(argv, OPTSTR_AWK, &opt_F, &list_v, &list_f, IF_FEATURE_AWK_GNU_EXTENSIONS(&list_e,) NULL);
3225         argv += optind;
3226         //argc -= optind;
3227         if (opt & OPT_W)
3228                 bb_error_msg("warning: option -W is ignored");
3229         if (opt & OPT_F) {
3230                 unescape_string_in_place(opt_F);
3231                 setvar_s(intvar[FS], opt_F);
3232         }
3233         while (list_v) {
3234                 if (!is_assignment(llist_pop(&list_v)))
3235                         bb_show_usage();
3236         }
3237         while (list_f) {
3238                 char *s = NULL;
3239                 FILE *from_file;
3240
3241                 g_progname = llist_pop(&list_f);
3242                 from_file = xfopen_stdin(g_progname);
3243                 /* one byte is reserved for some trick in next_token */
3244                 for (i = j = 1; j > 0; i += j) {
3245                         s = xrealloc(s, i + 4096);
3246                         j = fread(s + i, 1, 4094, from_file);
3247                 }
3248                 s[i] = '\0';
3249                 fclose(from_file);
3250                 parse_program(s + 1);
3251                 free(s);
3252         }
3253         g_progname = "cmd. line";
3254 #if ENABLE_FEATURE_AWK_GNU_EXTENSIONS
3255         while (list_e) {
3256                 parse_program(llist_pop(&list_e));
3257         }
3258 #endif
3259         if (!(opt & (OPT_f | OPT_e))) {
3260                 if (!*argv)
3261                         bb_show_usage();
3262                 parse_program(*argv++);
3263         }
3264
3265         /* fill in ARGV array */
3266         setari_u(intvar[ARGV], 0, "awk");
3267         i = 0;
3268         while (*argv)
3269                 setari_u(intvar[ARGV], ++i, *argv++);
3270         setvar_i(intvar[ARGC], i + 1);
3271
3272         evaluate(beginseq.first, &tv);
3273         if (!mainseq.first && !endseq.first)
3274                 awk_exit(EXIT_SUCCESS);
3275
3276         /* input file could already be opened in BEGIN block */
3277         if (!iF)
3278                 iF = next_input_file();
3279
3280         /* passing through input files */
3281         while (iF) {
3282                 nextfile = FALSE;
3283                 setvar_i(intvar[FNR], 0);
3284
3285                 while ((i = awk_getline(iF, intvar[F0])) > 0) {
3286                         nextrec = FALSE;
3287                         incvar(intvar[NR]);
3288                         incvar(intvar[FNR]);
3289                         evaluate(mainseq.first, &tv);
3290
3291                         if (nextfile)
3292                                 break;
3293                 }
3294
3295                 if (i < 0)
3296                         syntax_error(strerror(errno));
3297
3298                 iF = next_input_file();
3299         }
3300
3301         awk_exit(EXIT_SUCCESS);
3302         /*return 0;*/
3303 }