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