busybox/editors/awk.c

2885 lines
62 KiB
C
Raw Normal View History

/* vi: set sw=4 ts=4: */
/*
* awk implementation for busybox
*
* Copyright (C) 2002 by Dmitry Zakharov <dmit@crp.bank.gov.ua>
*
* Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
*/
#include "libbb.h"
#include "xregex.h"
#include <math.h>
extern char **environ;
/* This is a NOEXEC applet. Be very careful! */
#define MAXVARFMT 240
#define MINNVBLOCK 64
/* variable flags */
#define VF_NUMBER 0x0001 /* 1 = primary type is number */
#define VF_ARRAY 0x0002 /* 1 = it's an array */
#define VF_CACHED 0x0100 /* 1 = num/str value has cached str/num eq */
#define VF_USER 0x0200 /* 1 = user input (may be numeric string) */
#define VF_SPECIAL 0x0400 /* 1 = requires extra handling when changed */
#define VF_WALK 0x0800 /* 1 = variable has alloc'd x.walker list */
#define VF_FSTR 0x1000 /* 1 = var::string points to fstring buffer */
#define VF_CHILD 0x2000 /* 1 = function arg; x.parent points to source */
#define VF_DIRTY 0x4000 /* 1 = variable was set explicitly */
/* these flags are static, don't change them when value is changed */
#define VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY)
/* Variable */
typedef struct var_s {
unsigned type; /* flags */
double number;
char *string;
union {
int aidx; /* func arg idx (for compilation stage) */
struct xhash_s *array; /* array ptr */
struct var_s *parent; /* for func args, ptr to actual parameter */
char **walker; /* list of array elements (for..in) */
} x;
} var;
/* Node chain (pattern-action chain, BEGIN, END, function bodies) */
typedef struct chain_s {
struct node_s *first;
struct node_s *last;
const char *programname;
} chain;
/* Function */
typedef struct func_s {
unsigned nargs;
struct chain_s body;
} func;
/* I/O stream */
typedef struct rstream_s {
FILE *F;
char *buffer;
int adv;
int size;
int pos;
smallint is_pipe;
} rstream;
typedef struct hash_item_s {
union {
2007-05-18 04:33:35 +05:30
struct var_s v; /* variable/array hash */
struct rstream_s rs; /* redirect streams hash */
struct func_s f; /* functions hash */
} data;
2007-05-18 04:33:35 +05:30
struct hash_item_s *next; /* next in chain */
char name[1]; /* really it's longer */
} hash_item;
typedef struct xhash_s {
2007-05-18 04:33:35 +05:30
unsigned nel; /* num of elements */
unsigned csize; /* current hash size */
unsigned nprime; /* next hash size in PRIMES[] */
unsigned glen; /* summary length of item names */
struct hash_item_s **items;
} xhash;
/* Tree node */
typedef struct node_s {
uint32_t info;
unsigned lineno;
union {
struct node_s *n;
var *v;
int i;
char *s;
regex_t *re;
} l;
union {
struct node_s *n;
regex_t *ire;
func *f;
int argno;
} r;
union {
struct node_s *n;
} a;
} node;
/* Block of temporary variables */
typedef struct nvblock_s {
int size;
var *pos;
struct nvblock_s *prev;
struct nvblock_s *next;
var nv[0];
} nvblock;
typedef struct tsplitter_s {
node n;
regex_t re[2];
} tsplitter;
/* simple token classes */
/* Order and hex values are very important!!! See next_token() */
#define TC_SEQSTART 1 /* ( */
#define TC_SEQTERM (1 << 1) /* ) */
#define TC_REGEXP (1 << 2) /* /.../ */
#define TC_OUTRDR (1 << 3) /* | > >> */
#define TC_UOPPOST (1 << 4) /* unary postfix operator */
#define TC_UOPPRE1 (1 << 5) /* unary prefix operator */
#define TC_BINOPX (1 << 6) /* two-opnd operator */
#define TC_IN (1 << 7)
#define TC_COMMA (1 << 8)
#define TC_PIPE (1 << 9) /* input redirection pipe */
#define TC_UOPPRE2 (1 << 10) /* unary prefix operator */
#define TC_ARRTERM (1 << 11) /* ] */
#define TC_GRPSTART (1 << 12) /* { */
#define TC_GRPTERM (1 << 13) /* } */
#define TC_SEMICOL (1 << 14)
#define TC_NEWLINE (1 << 15)
#define TC_STATX (1 << 16) /* ctl statement (for, next...) */
#define TC_WHILE (1 << 17)
#define TC_ELSE (1 << 18)
#define TC_BUILTIN (1 << 19)
#define TC_GETLINE (1 << 20)
#define TC_FUNCDECL (1 << 21) /* `function' `func' */
#define TC_BEGIN (1 << 22)
#define TC_END (1 << 23)
#define TC_EOF (1 << 24)
#define TC_VARIABLE (1 << 25)
#define TC_ARRAY (1 << 26)
#define TC_FUNCTION (1 << 27)
#define TC_STRING (1 << 28)
#define TC_NUMBER (1 << 29)
2007-05-18 04:33:35 +05:30
#define TC_UOPPRE (TC_UOPPRE1 | TC_UOPPRE2)
/* combined token classes */
2007-05-18 04:33:35 +05:30
#define TC_BINOP (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN)
#define TC_UNARYOP (TC_UOPPRE | TC_UOPPOST)
#define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION \
| TC_BUILTIN | TC_GETLINE | TC_SEQSTART | TC_STRING | TC_NUMBER)
2007-05-18 04:33:35 +05:30
#define TC_STATEMNT (TC_STATX | TC_WHILE)
#define TC_OPTERM (TC_SEMICOL | TC_NEWLINE)
/* word tokens, cannot mean something else if not expected */
2007-05-18 04:33:35 +05:30
#define TC_WORD (TC_IN | TC_STATEMNT | TC_ELSE | TC_BUILTIN \
| TC_GETLINE | TC_FUNCDECL | TC_BEGIN | TC_END)
/* discard newlines after these */
2007-05-18 04:33:35 +05:30
#define TC_NOTERM (TC_COMMA | TC_GRPSTART | TC_GRPTERM \
| TC_BINOP | TC_OPTERM)
/* what can expression begin with */
2007-05-18 04:33:35 +05:30
#define TC_OPSEQ (TC_OPERAND | TC_UOPPRE | TC_REGEXP)
/* what can group begin with */
2007-05-18 04:33:35 +05:30
#define TC_GRPSEQ (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART)
/* if previous token class is CONCAT1 and next is CONCAT2, concatenation */
/* operator is inserted between them */
2007-05-18 04:33:35 +05:30
#define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM \
| TC_STRING | TC_NUMBER | TC_UOPPOST)
#define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE)
2007-05-18 04:33:35 +05:30
#define OF_RES1 0x010000
#define OF_RES2 0x020000
#define OF_STR1 0x040000
#define OF_STR2 0x080000
#define OF_NUM1 0x100000
#define OF_CHECKED 0x200000
/* combined operator flags */
#define xx 0
#define xV OF_RES2
#define xS (OF_RES2 | OF_STR2)
#define Vx OF_RES1
#define VV (OF_RES1 | OF_RES2)
#define Nx (OF_RES1 | OF_NUM1)
#define NV (OF_RES1 | OF_NUM1 | OF_RES2)
#define Sx (OF_RES1 | OF_STR1)
#define SV (OF_RES1 | OF_STR1 | OF_RES2)
#define SS (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2)
2007-05-18 04:33:35 +05:30
#define OPCLSMASK 0xFF00
#define OPNMASK 0x007F
/* operator priority is a highest byte (even: r->l, odd: l->r grouping)
* For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1,
* n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string
*/
2007-05-18 04:33:35 +05:30
#define P(x) (x << 24)
#define PRIMASK 0x7F000000
#define PRIMASK2 0x7E000000
/* Operation classes */
#define SHIFT_TIL_THIS 0x0600
#define RECUR_FROM_THIS 0x1000
enum {
2007-05-18 04:33:35 +05:30
OC_DELETE = 0x0100, OC_EXEC = 0x0200, OC_NEWSOURCE = 0x0300,
OC_PRINT = 0x0400, OC_PRINTF = 0x0500, OC_WALKINIT = 0x0600,
OC_BR = 0x0700, OC_BREAK = 0x0800, OC_CONTINUE = 0x0900,
OC_EXIT = 0x0a00, OC_NEXT = 0x0b00, OC_NEXTFILE = 0x0c00,
OC_TEST = 0x0d00, OC_WALKNEXT = 0x0e00,
OC_BINARY = 0x1000, OC_BUILTIN = 0x1100, OC_COLON = 0x1200,
OC_COMMA = 0x1300, OC_COMPARE = 0x1400, OC_CONCAT = 0x1500,
OC_FBLTIN = 0x1600, OC_FIELD = 0x1700, OC_FNARG = 0x1800,
OC_FUNC = 0x1900, OC_GETLINE = 0x1a00, OC_IN = 0x1b00,
OC_LAND = 0x1c00, OC_LOR = 0x1d00, OC_MATCH = 0x1e00,
OC_MOVE = 0x1f00, OC_PGETLINE = 0x2000, OC_REGEXP = 0x2100,
OC_REPLACE = 0x2200, OC_RETURN = 0x2300, OC_SPRINTF = 0x2400,
OC_TERNARY = 0x2500, OC_UNARY = 0x2600, OC_VAR = 0x2700,
OC_DONE = 0x2800,
ST_IF = 0x3000, ST_DO = 0x3100, ST_FOR = 0x3200,
ST_WHILE = 0x3300
};
/* simple builtins */
enum {
2007-05-18 04:33:35 +05:30
F_in, F_rn, F_co, F_ex, F_lg, F_si, F_sq, F_sr,
F_ti, F_le, F_sy, F_ff, F_cl
};
/* builtins */
enum {
2007-05-18 04:33:35 +05:30
B_a2, B_ix, B_ma, B_sp, B_ss, B_ti, B_lo, B_up,
2006-09-26 23:11:00 +05:30
B_ge, B_gs, B_su,
B_an, B_co, B_ls, B_or, B_rs, B_xo,
};
/* tokens and their corresponding info values */
2007-05-18 04:33:35 +05:30
#define NTC "\377" /* switch to next token class (tc<<1) */
#define NTCC '\377'
#define OC_B OC_BUILTIN
static const char tokenlist[] =
"\1(" NTC
"\1)" NTC
"\1/" NTC /* REGEXP */
"\2>>" "\1>" "\1|" NTC /* OUTRDR */
"\2++" "\2--" NTC /* UOPPOST */
"\2++" "\2--" "\1$" NTC /* UOPPRE1 */
"\2==" "\1=" "\2+=" "\2-=" /* BINOPX */
"\2*=" "\2/=" "\2%=" "\2^="
"\1+" "\1-" "\3**=" "\2**"
"\1/" "\1%" "\1^" "\1*"
"\2!=" "\2>=" "\2<=" "\1>"
"\1<" "\2!~" "\1~" "\2&&"
"\2||" "\1?" "\1:" NTC
"\2in" NTC
"\1," NTC
"\1|" NTC
"\1+" "\1-" "\1!" NTC /* UOPPRE2 */
"\1]" NTC
"\1{" NTC
"\1}" NTC
"\1;" NTC
"\1\n" NTC
"\2if" "\2do" "\3for" "\5break" /* STATX */
"\10continue" "\6delete" "\5print"
"\6printf" "\4next" "\10nextfile"
"\6return" "\4exit" NTC
"\5while" NTC
"\4else" NTC
"\3and" "\5compl" "\6lshift" "\2or"
"\6rshift" "\3xor"
"\5close" "\6system" "\6fflush" "\5atan2" /* BUILTIN */
"\3cos" "\3exp" "\3int" "\3log"
"\4rand" "\3sin" "\4sqrt" "\5srand"
"\6gensub" "\4gsub" "\5index" "\6length"
"\5match" "\5split" "\7sprintf" "\3sub"
"\6substr" "\7systime" "\10strftime"
"\7tolower" "\7toupper" NTC
"\7getline" NTC
"\4func" "\10function" NTC
"\5BEGIN" NTC
"\3END" "\0"
;
static const uint32_t tokeninfo[] = {
0,
0,
OC_REGEXP,
xS|'a', xS|'w', xS|'|',
OC_UNARY|xV|P(9)|'p', OC_UNARY|xV|P(9)|'m',
OC_UNARY|xV|P(9)|'P', OC_UNARY|xV|P(9)|'M',
OC_FIELD|xV|P(5),
OC_COMPARE|VV|P(39)|5, OC_MOVE|VV|P(74),
OC_REPLACE|NV|P(74)|'+', OC_REPLACE|NV|P(74)|'-',
OC_REPLACE|NV|P(74)|'*', OC_REPLACE|NV|P(74)|'/',
OC_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&',
OC_BINARY|NV|P(29)|'+', OC_BINARY|NV|P(29)|'-',
OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&',
OC_BINARY|NV|P(25)|'/', OC_BINARY|NV|P(25)|'%',
OC_BINARY|NV|P(15)|'&', OC_BINARY|NV|P(25)|'*',
OC_COMPARE|VV|P(39)|4, OC_COMPARE|VV|P(39)|3,
OC_COMPARE|VV|P(39)|0, OC_COMPARE|VV|P(39)|1,
OC_COMPARE|VV|P(39)|2, OC_MATCH|Sx|P(45)|'!',
OC_MATCH|Sx|P(45)|'~', OC_LAND|Vx|P(55),
OC_LOR|Vx|P(59), OC_TERNARY|Vx|P(64)|'?',
OC_COLON|xx|P(67)|':',
OC_IN|SV|P(49),
OC_COMMA|SS|P(80),
OC_PGETLINE|SV|P(37),
OC_UNARY|xV|P(19)|'+', OC_UNARY|xV|P(19)|'-',
OC_UNARY|xV|P(19)|'!',
0,
0,
0,
0,
0,
ST_IF, ST_DO, ST_FOR, OC_BREAK,
OC_CONTINUE, OC_DELETE|Vx, OC_PRINT,
OC_PRINTF, OC_NEXT, OC_NEXTFILE,
OC_RETURN|Vx, OC_EXIT|Nx,
ST_WHILE,
0,
2006-09-26 23:11:00 +05:30
OC_B|B_an|P(0x83), OC_B|B_co|P(0x41), OC_B|B_ls|P(0x83), OC_B|B_or|P(0x83),
OC_B|B_rs|P(0x83), OC_B|B_xo|P(0x83),
OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83),
OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg,
OC_FBLTIN|F_rn, OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr,
OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), OC_FBLTIN|Sx|F_le,
OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF, OC_B|B_su|P(0xb6),
OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b),
OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49),
OC_GETLINE|SV|P(0),
0, 0,
0,
0
};
/* internal variable names and their initial values */
/* asterisk marks SPECIAL vars; $ is just no-named Field0 */
enum {
2007-05-18 04:33:35 +05:30
CONVFMT, OFMT, FS, OFS,
ORS, RS, RT, FILENAME,
SUBSEP, ARGIND, ARGC, ARGV,
ERRNO, FNR,
NR, NF, IGNORECASE,
2007-05-18 04:33:35 +05:30
ENVIRON, F0, NUM_INTERNAL_VARS
};
static const char vNames[] =
"CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0"
"ORS\0" "RS\0*" "RT\0" "FILENAME\0"
"SUBSEP\0" "ARGIND\0" "ARGC\0" "ARGV\0"
"ERRNO\0" "FNR\0"
"NR\0" "NF\0*" "IGNORECASE\0*"
"ENVIRON\0" "$\0*" "\0";
static const char vValues[] =
"%.6g\0" "%.6g\0" " \0" " \0"
"\n\0" "\n\0" "\0" "\0"
"\034\0"
"\377";
/* hash size may grow to these values */
#define FIRST_PRIME 61;
static const unsigned PRIMES[] = { 251, 1021, 4093, 16381, 65521 };
2007-05-18 04:33:35 +05:30
enum { NPRIMES = sizeof(PRIMES) / sizeof(PRIMES[0]) };
/* Globals. Split in two parts so that first one is addressed
* with (mostly short) negative offsets */
struct globals {
chain beginseq, mainseq, endseq, *seq;
node *break_ptr, *continue_ptr;
rstream *iF;
xhash *vhash, *ahash, *fdhash, *fnhash;
const char *g_progname;
int g_lineno;
int nfields;
int maxfields; /* used in fsrealloc() only */
var *Fields;
nvblock *g_cb;
char *g_pos;
char *g_buf;
smallint icase;
smallint exiting;
smallint nextrec;
smallint nextfile;
smallint is_f0_split;
};
struct globals2 {
uint32_t t_info; /* often used */
uint32_t t_tclass;
char *t_string;
int t_lineno;
int t_rollback;
var *intvar[NUM_INTERNAL_VARS]; /* often used */
/* former statics from various functions */
char *split_f0__fstrings;
uint32_t next_token__save_tclass;
uint32_t next_token__save_info;
uint32_t next_token__ltclass;
smallint next_token__concat_inserted;
smallint next_input_file__files_happen;
rstream next_input_file__rsm;
var *evaluate__fnargs;
unsigned evaluate__seed;
regex_t evaluate__sreg;
var ptest__v;
tsplitter exec_builtin__tspl;
/* biggest and least used members go last */
double t_double;
tsplitter fsplitter, rsplitter;
};
#define G1 (ptr_to_globals[-1])
#define G (*(struct globals2 *const)ptr_to_globals)
/* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */
/* char G1size[sizeof(G1)]; - 0x6c */
/* char Gsize[sizeof(G)]; - 0x1cc */
/* Trying to keep most of members accessible with short offsets: */
/* char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */
#define beginseq (G1.beginseq )
#define mainseq (G1.mainseq )
#define endseq (G1.endseq )
#define seq (G1.seq )
#define break_ptr (G1.break_ptr )
#define continue_ptr (G1.continue_ptr)
#define iF (G1.iF )
#define vhash (G1.vhash )
#define ahash (G1.ahash )
#define fdhash (G1.fdhash )
#define fnhash (G1.fnhash )
#define g_progname (G1.g_progname )
#define g_lineno (G1.g_lineno )
#define nfields (G1.nfields )
#define maxfields (G1.maxfields )
#define Fields (G1.Fields )
#define g_cb (G1.g_cb )
#define g_pos (G1.g_pos )
#define g_buf (G1.g_buf )
#define icase (G1.icase )
#define exiting (G1.exiting )
#define nextrec (G1.nextrec )
#define nextfile (G1.nextfile )
#define is_f0_split (G1.is_f0_split )
#define t_info (G.t_info )
#define t_tclass (G.t_tclass )
#define t_string (G.t_string )
#define t_double (G.t_double )
#define t_lineno (G.t_lineno )
#define t_rollback (G.t_rollback )
#define intvar (G.intvar )
#define fsplitter (G.fsplitter )
#define rsplitter (G.rsplitter )
#define INIT_G() do { \
PTR_TO_GLOBALS = xzalloc(sizeof(G1) + sizeof(G)) + sizeof(G1); \
G.next_token__ltclass = TC_OPTERM; \
G.evaluate__seed = 1; \
} while (0)
/* function prototypes */
static void handle_special(var *);
static node *parse_expr(uint32_t);
static void chain_group(void);
static var *evaluate(node *, var *);
static rstream *next_input_file(void);
static int fmt_num(char *, int, const char *, double, int);
static int awk_exit(int) ATTRIBUTE_NORETURN;
/* ---- error handling ---- */
static const char EMSG_INTERNAL_ERROR[] = "Internal error";
static const char EMSG_UNEXP_EOS[] = "Unexpected end of string";
static const char EMSG_UNEXP_TOKEN[] = "Unexpected token";
static const char EMSG_DIV_BY_ZERO[] = "Division by zero";
static const char EMSG_INV_FMT[] = "Invalid format specifier";
static const char EMSG_TOO_FEW_ARGS[] = "Too few arguments for builtin";
static const char EMSG_NOT_ARRAY[] = "Not an array";
static const char EMSG_POSSIBLE_ERROR[] = "Possible syntax error";
static const char EMSG_UNDEF_FUNC[] = "Call to undefined function";
#if !ENABLE_FEATURE_AWK_MATH
static const char EMSG_NO_MATH[] = "Math support is not compiled in";
#endif
static void zero_out_var(var * vp)
{
memset(vp, 0, sizeof(*vp));
}
static void syntax_error(const char * const message) ATTRIBUTE_NORETURN;
static void syntax_error(const char * const message)
{
bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message);
}
/* ---- hash stuff ---- */
static unsigned hashidx(const char *name)
2005-09-27 07:53:02 +05:30
{
unsigned idx = 0;
while (*name) idx = *name++ + (idx << 6) - idx;
return idx;
}
/* create new hash */
2005-09-27 07:53:02 +05:30
static xhash *hash_init(void)
{
xhash *newhash;
newhash = xzalloc(sizeof(xhash));
newhash->csize = FIRST_PRIME;
newhash->items = xzalloc(newhash->csize * sizeof(hash_item *));
return newhash;
}
/* find item in hash, return ptr to data, NULL if not found */
static void *hash_search(xhash *hash, const char *name)
2005-09-27 07:53:02 +05:30
{
hash_item *hi;
hi = hash->items [ hashidx(name) % hash->csize ];
while (hi) {
if (strcmp(hi->name, name) == 0)
return &(hi->data);
hi = hi->next;
}
return NULL;
}
/* grow hash if it becomes too big */
2005-09-27 07:53:02 +05:30
static void hash_rebuild(xhash *hash)
{
unsigned newsize, i, idx;
hash_item **newitems, *hi, *thi;
if (hash->nprime == NPRIMES)
return;
newsize = PRIMES[hash->nprime++];
newitems = xzalloc(newsize * sizeof(hash_item *));
2007-05-18 04:33:35 +05:30
for (i = 0; i < hash->csize; i++) {
hi = hash->items[i];
while (hi) {
thi = hi;
hi = thi->next;
idx = hashidx(thi->name) % newsize;
thi->next = newitems[idx];
newitems[idx] = thi;
}
}
free(hash->items);
hash->csize = newsize;
hash->items = newitems;
}
/* find item in hash, add it if necessary. Return ptr to data */
static void *hash_find(xhash *hash, const char *name)
2005-09-27 07:53:02 +05:30
{
hash_item *hi;
unsigned idx;
int l;
hi = hash_search(hash, name);
if (! hi) {
if (++hash->nel / hash->csize > 10)
hash_rebuild(hash);
l = strlen(name) + 1;
hi = xzalloc(sizeof(hash_item) + l);
memcpy(hi->name, name, l);
idx = hashidx(name) % hash->csize;
hi->next = hash->items[idx];
hash->items[idx] = hi;
hash->glen += l;
}
return &(hi->data);
}
2007-05-18 04:33:35 +05:30
#define findvar(hash, name) ((var*) hash_find((hash), (name)))
#define newvar(name) ((var*) hash_find(vhash, (name)))
#define newfile(name) ((rstream*)hash_find(fdhash, (name)))
#define newfunc(name) ((func*) hash_find(fnhash, (name)))
static void hash_remove(xhash *hash, const char *name)
2005-09-27 07:53:02 +05:30
{
hash_item *hi, **phi;
2007-05-18 04:33:35 +05:30
phi = &(hash->items[hashidx(name) % hash->csize]);
while (*phi) {
hi = *phi;
if (strcmp(hi->name, name) == 0) {
hash->glen -= (strlen(name) + 1);
hash->nel--;
*phi = hi->next;
free(hi);
break;
}
phi = &(hi->next);
}
}
/* ------ some useful functions ------ */
2005-09-27 07:53:02 +05:30
static void skip_spaces(char **s)
{
char *p = *s;
while (1) {
if (*p == '\\' && p[1] == '\n') {
p++;
t_lineno++;
} else if (*p != ' ' && *p != '\t') {
break;
}
2005-09-27 08:48:00 +05:30
p++;
}
*s = p;
}
2005-09-27 07:53:02 +05:30
static char *nextword(char **s)
{
char *p = *s;
while (*(*s)++) /* */;
return p;
}
2005-09-27 07:53:02 +05:30
static char nextchar(char **s)
{
char c, *pps;
c = *((*s)++);
pps = *s;
2003-03-19 14:43:01 +05:30
if (c == '\\') c = bb_process_escape_sequence((const char**)s);
if (c == '\\' && *s == pps) c = *((*s)++);
return c;
}
static int ALWAYS_INLINE isalnum_(int c)
2005-09-27 07:53:02 +05:30
{
return (isalnum(c) || c == '_');
}
2005-09-27 07:53:02 +05:30
static FILE *afopen(const char *path, const char *mode)
{
return (*path == '-' && *(path+1) == '\0') ? stdin : xfopen(path, mode);
}
/* -------- working with variables (set/get/copy/etc) -------- */
2005-09-27 07:53:02 +05:30
static xhash *iamarray(var *v)
{
var *a = v;
while (a->type & VF_CHILD)
a = a->x.parent;
if (!(a->type & VF_ARRAY)) {
a->type |= VF_ARRAY;
a->x.array = hash_init();
}
return a->x.array;
}
2005-09-27 07:53:02 +05:30
static void clear_array(xhash *array)
{
unsigned i;
hash_item *hi, *thi;
for (i = 0; i < array->csize; i++) {
hi = array->items[i];
while (hi) {
thi = hi;
hi = hi->next;
2002-11-28 16:57:31 +05:30
free(thi->data.v.string);
free(thi);
}
array->items[i] = NULL;
}
array->glen = array->nel = 0;
}
/* clear a variable */
2005-09-27 07:53:02 +05:30
static var *clrvar(var *v)
{
2002-11-28 16:57:31 +05:30
if (!(v->type & VF_FSTR))
free(v->string);
v->type &= VF_DONTTOUCH;
v->type |= VF_DIRTY;
v->string = NULL;
return v;
}
/* assign string value to variable */
2005-09-27 07:53:02 +05:30
static var *setvar_p(var *v, char *value)
{
clrvar(v);
v->string = value;
handle_special(v);
return v;
}
/* same as setvar_p but make a copy of string */
static var *setvar_s(var *v, const char *value)
2005-09-27 07:53:02 +05:30
{
return setvar_p(v, (value && *value) ? xstrdup(value) : NULL);
}
/* same as setvar_s but set USER flag */
static var *setvar_u(var *v, const char *value)
2005-09-27 07:53:02 +05:30
{
setvar_s(v, value);
v->type |= VF_USER;
return v;
}
/* set array element to user string */
static void setari_u(var *a, int idx, const char *s)
2005-09-27 07:53:02 +05:30
{
2007-05-18 04:33:35 +05:30
char sidx[sizeof(int)*3 + 1];
var *v;
sprintf(sidx, "%d", idx);
v = findvar(iamarray(a), sidx);
setvar_u(v, s);
}
/* assign numeric value to variable */
2005-09-27 07:53:02 +05:30
static var *setvar_i(var *v, double value)
{
clrvar(v);
v->type |= VF_NUMBER;
v->number = value;
handle_special(v);
return v;
}
static const char *getvar_s(var *v)
2005-09-27 07:53:02 +05:30
{
/* if v is numeric and has no cached string, convert it to string */
if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) {
fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE);
v->string = xstrdup(g_buf);
v->type |= VF_CACHED;
}
return (v->string == NULL) ? "" : v->string;
}
2005-09-27 07:53:02 +05:30
static double getvar_i(var *v)
{
char *s;
if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) {
v->number = 0;
s = v->string;
if (s && *s) {
v->number = strtod(s, &s);
if (v->type & VF_USER) {
skip_spaces(&s);
if (*s != '\0')
v->type &= ~VF_USER;
}
} else {
v->type &= ~VF_USER;
}
v->type |= VF_CACHED;
}
return v->number;
}
static var *copyvar(var *dest, const var *src)
2005-09-27 07:53:02 +05:30
{
if (dest != src) {
clrvar(dest);
dest->type |= (src->type & ~(VF_DONTTOUCH | VF_FSTR));
dest->number = src->number;
if (src->string)
dest->string = xstrdup(src->string);
}
handle_special(dest);
return dest;
}
2005-09-27 07:53:02 +05:30
static var *incvar(var *v)
{
return setvar_i(v, getvar_i(v)+1.);
}
/* return true if v is number or numeric string */
2005-09-27 07:53:02 +05:30
static int is_numeric(var *v)
{
getvar_i(v);
return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY));
}
/* return 1 when value of v corresponds to true, 0 otherwise */
2005-09-27 07:53:02 +05:30
static int istrue(var *v)
{
if (is_numeric(v))
return (v->number == 0) ? 0 : 1;
return (v->string && *(v->string)) ? 1 : 0;
}
/* temporary variables allocator. Last allocated should be first freed */
2005-09-27 07:53:02 +05:30
static var *nvalloc(int n)
{
nvblock *pb = NULL;
var *v, *r;
int size;
while (g_cb) {
pb = g_cb;
if ((g_cb->pos - g_cb->nv) + n <= g_cb->size) break;
g_cb = g_cb->next;
}
if (!g_cb) {
size = (n <= MINNVBLOCK) ? MINNVBLOCK : n;
g_cb = xmalloc(sizeof(nvblock) + size * sizeof(var));
g_cb->size = size;
g_cb->pos = g_cb->nv;
g_cb->prev = pb;
g_cb->next = NULL;
if (pb) pb->next = g_cb;
}
v = r = g_cb->pos;
g_cb->pos += n;
while (v < g_cb->pos) {
v->type = 0;
v->string = NULL;
v++;
}
return r;
}
2005-09-27 07:53:02 +05:30
static void nvfree(var *v)
{
var *p;
if (v < g_cb->nv || v >= g_cb->pos)
syntax_error(EMSG_INTERNAL_ERROR);
for (p = v; p < g_cb->pos; p++) {
2007-05-18 04:33:35 +05:30
if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) {
clear_array(iamarray(p));
free(p->x.array->items);
free(p->x.array);
}
if (p->type & VF_WALK)
free(p->x.walker);
clrvar(p);
}
g_cb->pos = v;
while (g_cb->prev && g_cb->pos == g_cb->nv) {
g_cb = g_cb->prev;
}
}
/* ------- awk program text parsing ------- */
/* Parse next token pointed by global pos, place results into global ttt.
* If token isn't expected, give away. Return token class
*/
static uint32_t next_token(uint32_t expected)
2005-09-27 07:53:02 +05:30
{
#define concat_inserted (G.next_token__concat_inserted)
#define save_tclass (G.next_token__save_tclass)
#define save_info (G.next_token__save_info)
/* Initialized to TC_OPTERM: */
#define ltclass (G.next_token__ltclass)
char *p, *pp, *s;
const char *tl;
uint32_t tc;
const uint32_t *ti;
int l;
if (t_rollback) {
t_rollback = FALSE;
} else if (concat_inserted) {
concat_inserted = FALSE;
t_tclass = save_tclass;
t_info = save_info;
} else {
p = g_pos;
2007-01-02 05:23:52 +05:30
readnext:
skip_spaces(&p);
g_lineno = t_lineno;
if (*p == '#')
2007-05-18 04:33:35 +05:30
while (*p != '\n' && *p != '\0')
p++;
if (*p == '\n')
t_lineno++;
if (*p == '\0') {
tc = TC_EOF;
} else if (*p == '\"') {
/* it's a string */
t_string = s = ++p;
while (*p != '\"') {
if (*p == '\0' || *p == '\n')
syntax_error(EMSG_UNEXP_EOS);
*(s++) = nextchar(&p);
}
p++;
*s = '\0';
tc = TC_STRING;
} else if ((expected & TC_REGEXP) && *p == '/') {
/* it's regexp */
t_string = s = ++p;
while (*p != '/') {
if (*p == '\0' || *p == '\n')
syntax_error(EMSG_UNEXP_EOS);
*s = *p++;
2007-05-18 04:33:35 +05:30
if (*s++ == '\\') {
pp = p;
2003-03-19 14:43:01 +05:30
*(s-1) = bb_process_escape_sequence((const char **)&p);
2007-05-18 04:33:35 +05:30
if (*pp == '\\')
*s++ = '\\';
if (p == pp)
*s++ = *p++;
}
}
p++;
*s = '\0';
tc = TC_REGEXP;
} else if (*p == '.' || isdigit(*p)) {
/* it's a number */
t_double = strtod(p, &p);
if (*p == '.')
syntax_error(EMSG_UNEXP_TOKEN);
tc = TC_NUMBER;
} else {
/* search for something known */
tl = tokenlist;
tc = 0x00000001;
ti = tokeninfo;
while (*tl) {
l = *(tl++);
if (l == NTCC) {
tc <<= 1;
continue;
}
/* if token class is expected, token
* matches and it's not a longer word,
* then this is what we are looking for
*/
2007-05-18 04:33:35 +05:30
if ((tc & (expected | TC_WORD | TC_NEWLINE))
&& *tl == *p && strncmp(p, tl, l) == 0
&& !((tc & TC_WORD) && isalnum_(p[l]))
) {
t_info = *ti;
p += l;
break;
}
ti++;
tl += l;
}
if (!*tl) {
/* it's a name (var/array/function),
* otherwise it's something wrong
*/
2007-01-02 05:23:52 +05:30
if (!isalnum_(*p))
syntax_error(EMSG_UNEXP_TOKEN);
t_string = --p;
while (isalnum_(*(++p))) {
*(p-1) = *p;
}
*(p-1) = '\0';
tc = TC_VARIABLE;
/* also consume whitespace between functionname and bracket */
2007-05-18 04:33:35 +05:30
if (!(expected & TC_VARIABLE))
skip_spaces(&p);
if (*p == '(') {
tc = TC_FUNCTION;
} else {
if (*p == '[') {
p++;
tc = TC_ARRAY;
}
}
}
}
g_pos = p;
/* skipping newlines in some cases */
if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE))
goto readnext;
/* insert concatenation operator when needed */
2007-05-18 04:33:35 +05:30
if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) {
concat_inserted = TRUE;
save_tclass = tc;
save_info = t_info;
tc = TC_BINOP;
t_info = OC_CONCAT | SS | P(35);
}
t_tclass = tc;
}
ltclass = t_tclass;
/* Are we ready for this? */
2007-05-18 04:33:35 +05:30
if (!(ltclass & expected))
syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ?
2007-05-18 04:33:35 +05:30
EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN);
return ltclass;
#undef concat_inserted
#undef save_tclass
#undef save_info
#undef ltclass
}
static void rollback_token(void)
{
t_rollback = TRUE;
}
static node *new_node(uint32_t info)
2005-09-27 07:53:02 +05:30
{
node *n;
n = xzalloc(sizeof(node));
n->info = info;
n->lineno = g_lineno;
return n;
}
static node *mk_re_node(const char *s, node *n, regex_t *re)
2005-09-27 07:53:02 +05:30
{
n->info = OC_REGEXP;
n->l.re = re;
n->r.ire = re + 1;
xregcomp(re, s, REG_EXTENDED);
2007-05-18 04:33:35 +05:30
xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE);
return n;
}
2005-09-27 07:53:02 +05:30
static node *condition(void)
{
next_token(TC_SEQSTART);
return parse_expr(TC_SEQTERM);
}
/* parse expression terminated by given argument, return ptr
* to built subtree. Terminator is eaten by parse_expr */
static node *parse_expr(uint32_t iexp)
2005-09-27 07:53:02 +05:30
{
node sn;
node *cn = &sn;
node *vn, *glptr;
uint32_t tc, xtc;
var *v;
sn.info = PRIMASK;
sn.r.n = glptr = NULL;
xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp;
while (!((tc = next_token(xtc)) & iexp)) {
if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) {
/* input redirection (<) attached to glptr node */
2007-05-18 04:33:35 +05:30
cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37));
cn->a.n = glptr;
xtc = TC_OPERAND | TC_UOPPRE;
glptr = NULL;
} else if (tc & (TC_BINOP | TC_UOPPOST)) {
/* for binary and postfix-unary operators, jump back over
* previous operators with higher priority */
vn = cn;
while ( ((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2))
|| ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON)) )
vn = vn->a.n;
if ((t_info & OPCLSMASK) == OC_TERNARY)
t_info += P(6);
cn = vn->a.n->r.n = new_node(t_info);
cn->a.n = vn->a.n;
if (tc & TC_BINOP) {
cn->l.n = vn;
xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
if ((t_info & OPCLSMASK) == OC_PGETLINE) {
/* it's a pipe */
next_token(TC_GETLINE);
/* give maximum priority to this pipe */
cn->info &= ~PRIMASK;
xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
}
} else {
cn->r.n = vn;
xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
}
vn->a.n = cn;
} else {
/* for operands and prefix-unary operators, attach them
* to last node */
vn = cn;
cn = vn->r.n = new_node(t_info);
cn->a.n = vn;
xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP;
if (tc & (TC_OPERAND | TC_REGEXP)) {
xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp;
/* one should be very careful with switch on tclass -
* only simple tclasses should be used! */
switch (tc) {
case TC_VARIABLE:
case TC_ARRAY:
cn->info = OC_VAR;
v = hash_search(ahash, t_string);
2007-05-18 04:33:35 +05:30
if (v != NULL) {
cn->info = OC_FNARG;
cn->l.i = v->x.aidx;
} else {
cn->l.v = newvar(t_string);
}
if (tc & TC_ARRAY) {
cn->info |= xS;
cn->r.n = parse_expr(TC_ARRTERM);
}
break;
2005-09-27 08:48:00 +05:30
case TC_NUMBER:
case TC_STRING:
cn->info = OC_VAR;
v = cn->l.v = xzalloc(sizeof(var));
if (tc & TC_NUMBER)
setvar_i(v, t_double);
else
setvar_s(v, t_string);
break;
case TC_REGEXP:
mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2));
break;
case TC_FUNCTION:
2005-09-27 08:48:00 +05:30
cn->info = OC_FUNC;
cn->r.f = newfunc(t_string);
cn->l.n = condition();
break;
case TC_SEQSTART:
cn = vn->r.n = parse_expr(TC_SEQTERM);
cn->a.n = vn;
break;
case TC_GETLINE:
glptr = cn;
xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp;
break;
case TC_BUILTIN:
cn->l.n = condition();
break;
}
}
}
}
return sn.r.n;
}
/* add node to chain. Return ptr to alloc'd node */
static node *chain_node(uint32_t info)
2005-09-27 07:53:02 +05:30
{
node *n;
if (!seq->first)
seq->first = seq->last = new_node(0);
if (seq->programname != g_progname) {
seq->programname = g_progname;
n = chain_node(OC_NEWSOURCE);
n->l.s = xstrdup(g_progname);
}
n = seq->last;
n->info = info;
seq->last = n->a.n = new_node(OC_DONE);
return n;
}
static void chain_expr(uint32_t info)
2005-09-27 07:53:02 +05:30
{
node *n;
n = chain_node(info);
n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM);
if (t_tclass & TC_GRPTERM)
rollback_token();
}
2005-09-27 07:53:02 +05:30
static node *chain_loop(node *nn)
{
node *n, *n2, *save_brk, *save_cont;
save_brk = break_ptr;
save_cont = continue_ptr;
n = chain_node(OC_BR | Vx);
continue_ptr = new_node(OC_EXEC);
break_ptr = new_node(OC_EXEC);
chain_group();
n2 = chain_node(OC_EXEC | Vx);
n2->l.n = nn;
n2->a.n = n;
continue_ptr->a.n = n2;
break_ptr->a.n = n->r.n = seq->last;
continue_ptr = save_cont;
break_ptr = save_brk;
return n;
}
/* parse group and attach it to chain */
2005-09-27 07:53:02 +05:30
static void chain_group(void)
{
uint32_t c;
node *n, *n2, *n3;
do {
c = next_token(TC_GRPSEQ);
} while (c & TC_NEWLINE);
if (c & TC_GRPSTART) {
while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) {
if (t_tclass & TC_NEWLINE) continue;
rollback_token();
chain_group();
}
} else if (c & (TC_OPSEQ | TC_OPTERM)) {
rollback_token();
chain_expr(OC_EXEC | Vx);
} else { /* TC_STATEMNT */
switch (t_info & OPCLSMASK) {
2007-05-18 04:33:35 +05:30
case ST_IF:
n = chain_node(OC_BR | Vx);
n->l.n = condition();
chain_group();
n2 = chain_node(OC_EXEC);
n->r.n = seq->last;
if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE) == TC_ELSE) {
chain_group();
2007-05-18 04:33:35 +05:30
n2->a.n = seq->last;
} else {
rollback_token();
}
break;
2007-05-18 04:33:35 +05:30
case ST_WHILE:
n2 = condition();
n = chain_loop(NULL);
n->l.n = n2;
break;
2007-05-18 04:33:35 +05:30
case ST_DO:
n2 = chain_node(OC_EXEC);
n = chain_loop(NULL);
n2->a.n = n->a.n;
next_token(TC_WHILE);
n->l.n = condition();
break;
2007-05-18 04:33:35 +05:30
case ST_FOR:
next_token(TC_SEQSTART);
n2 = parse_expr(TC_SEMICOL | TC_SEQTERM);
if (t_tclass & TC_SEQTERM) { /* for-in */
2007-05-18 04:33:35 +05:30
if ((n2->info & OPCLSMASK) != OC_IN)
syntax_error(EMSG_UNEXP_TOKEN);
n = chain_node(OC_WALKINIT | VV);
n->l.n = n2->l.n;
n->r.n = n2->r.n;
n = chain_loop(NULL);
n->info = OC_WALKNEXT | Vx;
n->l.n = n2->l.n;
} else { /* for (;;) */
n = chain_node(OC_EXEC | Vx);
n->l.n = n2;
n2 = parse_expr(TC_SEMICOL);
n3 = parse_expr(TC_SEQTERM);
n = chain_loop(n3);
n->l.n = n2;
if (! n2)
n->info = OC_EXEC;
}
break;
2007-05-18 04:33:35 +05:30
case OC_PRINT:
case OC_PRINTF:
n = chain_node(t_info);
2007-05-18 04:33:35 +05:30
n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM);
if (t_tclass & TC_OUTRDR) {
n->info |= t_info;
2007-05-18 04:33:35 +05:30
n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM);
}
if (t_tclass & TC_GRPTERM)
2007-05-18 04:33:35 +05:30
rollback_token();
break;
2007-05-18 04:33:35 +05:30
case OC_BREAK:
n = chain_node(OC_EXEC);
n->a.n = break_ptr;
break;
2007-05-18 04:33:35 +05:30
case OC_CONTINUE:
n = chain_node(OC_EXEC);
n->a.n = continue_ptr;
break;
2007-05-18 04:33:35 +05:30
/* delete, next, nextfile, return, exit */
default:
chain_expr(t_info);
}
}
}
2005-09-27 07:53:02 +05:30
static void parse_program(char *p)
{
uint32_t tclass;
node *cn;
func *f;
var *v;
g_pos = p;
t_lineno = 1;
while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART |
2007-05-18 04:33:35 +05:30
TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) {
if (tclass & TC_OPTERM)
continue;
seq = &mainseq;
if (tclass & TC_BEGIN) {
seq = &beginseq;
chain_group();
} else if (tclass & TC_END) {
seq = &endseq;
chain_group();
} else if (tclass & TC_FUNCDECL) {
next_token(TC_FUNCTION);
g_pos++;
f = newfunc(t_string);
f->body.first = NULL;
f->nargs = 0;
while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) {
v = findvar(ahash, t_string);
v->x.aidx = (f->nargs)++;
if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM)
break;
}
seq = &(f->body);
chain_group();
clear_array(ahash);
} else if (tclass & TC_OPSEQ) {
rollback_token();
cn = chain_node(OC_TEST);
cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART);
if (t_tclass & TC_GRPSTART) {
rollback_token();
chain_group();
} else {
chain_node(OC_PRINT);
}
cn->r.n = mainseq.last;
} else /* if (tclass & TC_GRPSTART) */ {
rollback_token();
chain_group();
}
}
}
/* -------- program execution part -------- */
static node *mk_splitter(const char *s, tsplitter *spl)
2005-09-27 07:53:02 +05:30
{
regex_t *re, *ire;
node *n;
re = &spl->re[0];
ire = &spl->re[1];
n = &spl->n;
if ((n->info & OPCLSMASK) == OC_REGEXP) {
regfree(re);
regfree(ire);
}
if (strlen(s) > 1) {
mk_re_node(s, n, re);
} else {
n->info = (uint32_t) *s;
}
return n;
}
/* use node as a regular expression. Supplied with node ptr and regex_t
* storage space. Return ptr to regex (if result points to preg, it should
* be later regfree'd manually
*/
2005-09-27 07:53:02 +05:30
static regex_t *as_regex(node *op, regex_t *preg)
{
var *v;
const char *s;
if ((op->info & OPCLSMASK) == OC_REGEXP) {
return icase ? op->r.ire : op->l.re;
}
2007-05-18 04:33:35 +05:30
v = nvalloc(1);
s = getvar_s(evaluate(op, v));
xregcomp(preg, s, icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED);
nvfree(v);
return preg;
}
/* gradually increasing buffer */
2005-09-27 07:53:02 +05:30
static void qrealloc(char **b, int n, int *size)
{
2007-01-02 05:23:52 +05:30
if (!*b || n >= *size)
*b = xrealloc(*b, *size = n + (n>>1) + 80);
}
/* resize field storage space */
2005-09-27 07:53:02 +05:30
static void fsrealloc(int size)
{
int i;
if (size >= maxfields) {
i = maxfields;
maxfields = size + 16;
Fields = xrealloc(Fields, maxfields * sizeof(var));
for (; i < maxfields; i++) {
Fields[i].type = VF_SPECIAL;
Fields[i].string = NULL;
}
}
if (size < nfields) {
2007-01-02 05:23:52 +05:30
for (i = size; i < nfields; i++) {
clrvar(Fields + i);
}
}
nfields = size;
}
static int awk_split(const char *s, node *spl, char **slist)
2005-09-27 07:53:02 +05:30
{
int l, n = 0;
char c[4];
char *s1;
regmatch_t pmatch[2];
/* in worst case, each char would be a separate field */
*slist = s1 = xzalloc(strlen(s) * 2 + 3);
strcpy(s1, s);
c[0] = c[1] = (char)spl->info;
c[2] = c[3] = '\0';
2007-05-18 04:33:35 +05:30
if (*getvar_s(intvar[RS]) == '\0')
c[2] = '\n';
if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */
while (*s) {
l = strcspn(s, c+2);
if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0
&& pmatch[0].rm_so <= l
) {
l = pmatch[0].rm_so;
2007-05-18 04:33:35 +05:30
if (pmatch[0].rm_eo == 0) {
l++;
pmatch[0].rm_eo++;
}
} else {
pmatch[0].rm_eo = l;
if (s[l]) pmatch[0].rm_eo++;
}
memcpy(s1, s, l);
s1[l] = '\0';
nextword(&s1);
s += pmatch[0].rm_eo;
n++;
}
} else if (c[0] == '\0') { /* null split */
while (*s) {
2007-01-02 05:23:52 +05:30
*s1++ = *s++;
*s1++ = '\0';
n++;
}
} else if (c[0] != ' ') { /* single-character split */
if (icase) {
c[0] = toupper(c[0]);
c[1] = tolower(c[1]);
}
if (*s1) n++;
while ((s1 = strpbrk(s1, c))) {
2007-01-02 05:23:52 +05:30
*s1++ = '\0';
n++;
}
} else { /* space split */
while (*s) {
2006-10-25 18:16:03 +05:30
s = skip_whitespace(s);
2007-01-02 05:23:52 +05:30
if (!*s) break;
n++;
while (*s && !isspace(*s))
2007-01-02 05:23:52 +05:30
*s1++ = *s++;
*s1++ = '\0';
}
}
return n;
}
2005-09-27 07:53:02 +05:30
static void split_f0(void)
{
#define fstrings (G.split_f0__fstrings)
2007-05-18 04:33:35 +05:30
int i, n;
char *s;
if (is_f0_split)
return;
is_f0_split = TRUE;
2002-11-28 16:57:31 +05:30
free(fstrings);
fsrealloc(0);
2007-05-18 04:33:35 +05:30
n = awk_split(getvar_s(intvar[F0]), &fsplitter.n, &fstrings);
fsrealloc(n);
s = fstrings;
for (i = 0; i < n; i++) {
Fields[i].string = nextword(&s);
Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY);
}
/* set NF manually to avoid side effects */
2007-05-18 04:33:35 +05:30
clrvar(intvar[NF]);
intvar[NF]->type = VF_NUMBER | VF_SPECIAL;
intvar[NF]->number = nfields;
#undef fstrings
}
/* perform additional actions when some internal variables changed */
2005-09-27 07:53:02 +05:30
static void handle_special(var *v)
{
int n;
char *b;
const char *sep, *s;
int sl, l, len, i, bsize;
2007-01-02 05:23:52 +05:30
if (!(v->type & VF_SPECIAL))
return;
2007-05-18 04:33:35 +05:30
if (v == intvar[NF]) {
n = (int)getvar_i(v);
fsrealloc(n);
/* recalculate $0 */
2007-05-18 04:33:35 +05:30
sep = getvar_s(intvar[OFS]);
sl = strlen(sep);
b = NULL;
len = 0;
2007-05-18 04:33:35 +05:30
for (i = 0; i < n; i++) {
s = getvar_s(&Fields[i]);
l = strlen(s);
if (b) {
memcpy(b+len, sep, sl);
len += sl;
}
qrealloc(&b, len+l+sl, &bsize);
memcpy(b+len, s, l);
len += l;
}
2007-01-02 05:23:52 +05:30
if (b)
b[len] = '\0';
2007-05-18 04:33:35 +05:30
setvar_p(intvar[F0], b);
is_f0_split = TRUE;
2007-05-18 04:33:35 +05:30
} else if (v == intvar[F0]) {
is_f0_split = FALSE;
2007-05-18 04:33:35 +05:30
} else if (v == intvar[FS]) {
mk_splitter(getvar_s(v), &fsplitter);
2007-05-18 04:33:35 +05:30
} else if (v == intvar[RS]) {
mk_splitter(getvar_s(v), &rsplitter);
2007-05-18 04:33:35 +05:30
} else if (v == intvar[IGNORECASE]) {
icase = istrue(v);
2007-01-02 05:23:52 +05:30
} else { /* $n */
2007-05-18 04:33:35 +05:30
n = getvar_i(intvar[NF]);
setvar_i(intvar[NF], n > v-Fields ? n : v-Fields+1);
/* right here v is invalid. Just to note... */
}
}
/* step through func/builtin/etc arguments */
2005-09-27 07:53:02 +05:30
static node *nextarg(node **pn)
{
node *n;
n = *pn;
if (n && (n->info & OPCLSMASK) == OC_COMMA) {
*pn = n->r.n;
n = n->l.n;
} else {
*pn = NULL;
}
return n;
}
2005-09-27 07:53:02 +05:30
static void hashwalk_init(var *v, xhash *array)
{
char **w;
hash_item *hi;
int i;
if (v->type & VF_WALK)
free(v->x.walker);
v->type |= VF_WALK;
w = v->x.walker = xzalloc(2 + 2*sizeof(char *) + array->glen);
2007-05-18 04:33:35 +05:30
w[0] = w[1] = (char *)(w + 2);
for (i = 0; i < array->csize; i++) {
hi = array->items[i];
while (hi) {
strcpy(*w, hi->name);
nextword(w);
hi = hi->next;
}
}
}
2005-09-27 07:53:02 +05:30
static int hashwalk_next(var *v)
{
char **w;
w = v->x.walker;
2007-05-18 04:33:35 +05:30
if (w[1] == w[0])
return FALSE;
setvar_s(v, nextword(w+1));
return TRUE;
}
/* evaluate node, return 1 when result is true, 0 otherwise */
2005-09-27 07:53:02 +05:30
static int ptest(node *pattern)
{
/* ptest__v is "static": to save stack space? */
return istrue(evaluate(pattern, &G.ptest__v));
}
/* read next record from stream rsm into a variable v */
2005-09-27 07:53:02 +05:30
static int awk_getline(rstream *rsm, var *v)
{
char *b;
regmatch_t pmatch[2];
int a, p, pp=0, size;
int fd, so, eo, r, rp;
char c, *m, *s;
/* we're using our own buffer since we need access to accumulating
* characters
*/
fd = fileno(rsm->F);
m = rsm->buffer;
a = rsm->adv;
p = rsm->pos;
size = rsm->size;
c = (char) rsplitter.n.info;
rp = 0;
if (! m) qrealloc(&m, 256, &size);
do {
b = m + a;
so = eo = p;
r = 1;
if (p > 0) {
if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) {
if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re,
2007-05-18 04:33:35 +05:30
b, 1, pmatch, 0) == 0) {
so = pmatch[0].rm_so;
eo = pmatch[0].rm_eo;
if (b[eo] != '\0')
break;
}
} else if (c != '\0') {
s = strchr(b+pp, c);
if (! s) s = memchr(b+pp, '\0', p - pp);
if (s) {
so = eo = s-b;
eo++;
break;
}
} else {
while (b[rp] == '\n')
rp++;
s = strstr(b+rp, "\n\n");
if (s) {
so = eo = s-b;
while (b[eo] == '\n') eo++;
if (b[eo] != '\0')
break;
}
}
}
if (a > 0) {
memmove(m, (const void *)(m+a), p+1);
b = m;
a = 0;
}
qrealloc(&m, a+p+128, &size);
b = m + a;
pp = p;
p += safe_read(fd, b+p, size-p-1);
if (p < pp) {
p = 0;
r = 0;
2007-05-18 04:33:35 +05:30
setvar_i(intvar[ERRNO], errno);
}
b[p] = '\0';
} while (p > pp);
if (p == 0) {
r--;
} else {
c = b[so]; b[so] = '\0';
setvar_s(v, b+rp);
v->type |= VF_USER;
b[so] = c;
c = b[eo]; b[eo] = '\0';
2007-05-18 04:33:35 +05:30
setvar_s(intvar[RT], b+so);
b[eo] = c;
}
rsm->buffer = m;
rsm->adv = a + eo;
rsm->pos = p - eo;
rsm->size = size;
return r;
}
static int fmt_num(char *b, int size, const char *format, double n, int int_as_int)
2005-09-27 07:53:02 +05:30
{
int r = 0;
char c;
const char *s = format;
if (int_as_int && n == (int)n) {
r = snprintf(b, size, "%d", (int)n);
} else {
do { c = *s; } while (c && *++s);
if (strchr("diouxX", c)) {
r = snprintf(b, size, format, (int)n);
} else if (strchr("eEfgG", c)) {
r = snprintf(b, size, format, n);
} else {
syntax_error(EMSG_INV_FMT);
}
}
return r;
}
/* formatted output into an allocated buffer, return ptr to buffer */
2005-09-27 07:53:02 +05:30
static char *awk_printf(node *n)
{
char *b = NULL;
char *fmt, *s, *f;
const char *s1;
int i, j, incr, bsize;
char c, c1;
var *v, *arg;
v = nvalloc(1);
fmt = f = xstrdup(getvar_s(evaluate(nextarg(&n), v)));
i = 0;
while (*f) {
s = f;
while (*f && (*f != '%' || *(++f) == '%'))
f++;
while (*f && !isalpha(*f)) {
if (*f == '*')
syntax_error("%*x formats are not supported");
f++;
}
incr = (f - s) + MAXVARFMT;
qrealloc(&b, incr + i, &bsize);
c = *f;
if (c != '\0') f++;
c1 = *f;
*f = '\0';
arg = evaluate(nextarg(&n), v);
j = i;
if (c == 'c' || !c) {
i += sprintf(b+i, s, is_numeric(arg) ?
(char)getvar_i(arg) : *getvar_s(arg));
} else if (c == 's') {
2006-10-04 01:26:34 +05:30
s1 = getvar_s(arg);
qrealloc(&b, incr+i+strlen(s1), &bsize);
i += sprintf(b+i, s, s1);
} else {
i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE);
}
*f = c1;
/* if there was an error while sprintf, return value is negative */
if (i < j) i = j;
}
b = xrealloc(b, i + 1);
free(fmt);
nvfree(v);
b[i] = '\0';
return b;
}
/* common substitution routine
* replace (nm) substring of (src) that match (n) with (repl), store
* result into (dest), return number of substitutions. If nm=0, replace
* all matches. If src or dst is NULL, use $0. If ex=TRUE, enable
* subexpression matching (\1-\9)
*/
static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int ex)
2005-09-27 07:53:02 +05:30
{
char *ds = NULL;
const char *s;
const char *sp;
int c, i, j, di, rl, so, eo, nbs, n, dssize;
regmatch_t pmatch[10];
regex_t sreg, *re;
re = as_regex(rn, &sreg);
2007-05-18 04:33:35 +05:30
if (! src) src = intvar[F0];
if (! dest) dest = intvar[F0];
i = di = 0;
sp = getvar_s(src);
rl = strlen(repl);
2007-05-18 04:33:35 +05:30
while (regexec(re, sp, 10, pmatch, sp==getvar_s(src) ? 0 : REG_NOTBOL) == 0) {
so = pmatch[0].rm_so;
eo = pmatch[0].rm_eo;
qrealloc(&ds, di + eo + rl, &dssize);
memcpy(ds + di, sp, eo);
di += eo;
if (++i >= nm) {
/* replace */
di -= (eo - so);
nbs = 0;
for (s = repl; *s; s++) {
ds[di++] = c = *s;
if (c == '\\') {
nbs++;
continue;
}
if (c == '&' || (ex && c >= '0' && c <= '9')) {
di -= ((nbs + 3) >> 1);
j = 0;
if (c != '&') {
j = c - '0';
nbs++;
}
if (nbs % 2) {
ds[di++] = c;
} else {
n = pmatch[j].rm_eo - pmatch[j].rm_so;
qrealloc(&ds, di + rl + n, &dssize);
memcpy(ds + di, sp + pmatch[j].rm_so, n);
di += n;
}
}
nbs = 0;
}
}
sp += eo;
if (i == nm) break;
if (eo == so) {
if (! (ds[di++] = *sp++)) break;
}
}
qrealloc(&ds, di + strlen(sp), &dssize);
strcpy(ds + di, sp);
setvar_p(dest, ds);
if (re == &sreg) regfree(re);
return i;
}
2005-09-27 07:53:02 +05:30
static var *exec_builtin(node *op, var *res)
{
#define tspl (G.exec_builtin__tspl)
int (*to_xxx)(int);
var *tv;
node *an[4];
2007-05-18 04:33:35 +05:30
var *av[4];
const char *as[4];
regmatch_t pmatch[2];
regex_t sreg, *re;
node *spl;
uint32_t isr, info;
int nargs;
time_t tt;
char *s, *s1;
int i, l, ll, n;
tv = nvalloc(4);
isr = info = op->info;
op = op->l.n;
av[2] = av[3] = NULL;
2007-05-18 04:33:35 +05:30
for (i = 0; i < 4 && op; i++) {
an[i] = nextarg(&op);
if (isr & 0x09000000) av[i] = evaluate(an[i], &tv[i]);
if (isr & 0x08000000) as[i] = getvar_s(av[i]);
isr >>= 1;
}
nargs = i;
if (nargs < (info >> 30))
syntax_error(EMSG_TOO_FEW_ARGS);
switch (info & OPNMASK) {
case B_a2:
#if ENABLE_FEATURE_AWK_MATH
setvar_i(res, atan2(getvar_i(av[i]), getvar_i(av[1])));
#else
syntax_error(EMSG_NO_MATH);
#endif
break;
case B_sp:
if (nargs > 2) {
spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ?
an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl);
} else {
spl = &fsplitter.n;
}
n = awk_split(as[0], spl, &s);
s1 = s;
clear_array(iamarray(av[1]));
for (i=1; i<=n; i++)
setari_u(av[1], i, nextword(&s1));
free(s);
setvar_i(res, n);
break;
case B_ss:
l = strlen(as[0]);
i = getvar_i(av[1]) - 1;
2007-05-18 04:33:35 +05:30
if (i > l) i = l;
if (i < 0) i = 0;
n = (nargs > 2) ? getvar_i(av[2]) : l-i;
2007-05-18 04:33:35 +05:30
if (n < 0) n = 0;
s = xmalloc(n+1);
strncpy(s, as[0]+i, n);
s[n] = '\0';
setvar_p(res, s);
break;
case B_an:
2006-09-26 23:11:00 +05:30
setvar_i(res, (long)getvar_i(av[0]) & (long)getvar_i(av[1]));
break;
case B_co:
2006-09-26 23:11:00 +05:30
setvar_i(res, ~(long)getvar_i(av[0]));
break;
case B_ls:
2006-09-26 23:11:00 +05:30
setvar_i(res, (long)getvar_i(av[0]) << (long)getvar_i(av[1]));
break;
case B_or:
2006-09-26 23:11:00 +05:30
setvar_i(res, (long)getvar_i(av[0]) | (long)getvar_i(av[1]));
break;
case B_rs:
2006-09-26 23:11:00 +05:30
setvar_i(res, (long)((unsigned long)getvar_i(av[0]) >> (unsigned long)getvar_i(av[1])));
break;
case B_xo:
2006-09-26 23:11:00 +05:30
setvar_i(res, (long)getvar_i(av[0]) ^ (long)getvar_i(av[1]));
break;
case B_lo:
to_xxx = tolower;
goto lo_cont;
case B_up:
to_xxx = toupper;
2007-01-02 05:23:52 +05:30
lo_cont:
s1 = s = xstrdup(as[0]);
while (*s1) {
*s1 = (*to_xxx)(*s1);
s1++;
}
setvar_p(res, s);
break;
case B_ix:
n = 0;
ll = strlen(as[1]);
l = strlen(as[0]) - ll;
if (ll > 0 && l >= 0) {
if (!icase) {
s = strstr(as[0], as[1]);
if (s) n = (s - as[0]) + 1;
} else {
/* this piece of code is terribly slow and
* really should be rewritten
*/
for (i=0; i<=l; i++) {
if (strncasecmp(as[0]+i, as[1], ll) == 0) {
n = i+1;
break;
}
}
}
}
setvar_i(res, n);
break;
case B_ti:
if (nargs > 1)
tt = getvar_i(av[1]);
else
time(&tt);
//s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y";
i = strftime(g_buf, MAXVARFMT,
((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"),
localtime(&tt));
g_buf[i] = '\0';
setvar_s(res, g_buf);
break;
case B_ma:
re = as_regex(an[1], &sreg);
n = regexec(re, as[0], 1, pmatch, 0);
if (n == 0) {
pmatch[0].rm_so++;
pmatch[0].rm_eo++;
} else {
pmatch[0].rm_so = 0;
pmatch[0].rm_eo = -1;
}
setvar_i(newvar("RSTART"), pmatch[0].rm_so);
setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so);
setvar_i(res, pmatch[0].rm_so);
if (re == &sreg) regfree(re);
break;
case B_ge:
awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE);
break;
case B_gs:
setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE));
break;
case B_su:
setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE));
break;
}
nvfree(tv);
return res;
#undef tspl
}
/*
* Evaluate node - the heart of the program. Supplied with subtree
* and place where to store result. returns ptr to result.
*/
#define XC(n) ((n) >> 8)
2005-09-27 07:53:02 +05:30
static var *evaluate(node *op, var *res)
{
/* This procedure is recursive so we should count every byte */
#define fnargs (G.evaluate__fnargs)
/* seed is initialized to 1 */
#define seed (G.evaluate__seed)
#define sreg (G.evaluate__sreg)
2007-01-02 05:23:52 +05:30
node *op1;
var *v1;
union {
var *v;
const char *s;
double d;
int i;
} L, R;
uint32_t opinfo;
int opn;
union {
char *s;
rstream *rsm;
FILE *F;
var *v;
regex_t *re;
uint32_t info;
} X;
2007-01-02 05:23:52 +05:30
if (!op)
return setvar_s(res, NULL);
v1 = nvalloc(2);
while (op) {
opinfo = op->info;
opn = (opinfo & OPNMASK);
g_lineno = op->lineno;
2005-09-27 08:48:00 +05:30
/* execute inevitable things */
op1 = op->l.n;
if (opinfo & OF_RES1) X.v = L.v = evaluate(op1, v1);
if (opinfo & OF_RES2) R.v = evaluate(op->r.n, v1+1);
if (opinfo & OF_STR1) L.s = getvar_s(L.v);
if (opinfo & OF_STR2) R.s = getvar_s(R.v);
if (opinfo & OF_NUM1) L.d = getvar_i(L.v);
switch (XC(opinfo & OPCLSMASK)) {
/* -- iterative node type -- */
/* test pattern */
case XC( OC_TEST ):
if ((op1->info & OPCLSMASK) == OC_COMMA) {
/* it's range pattern */
if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) {
op->info |= OF_CHECKED;
if (ptest(op1->r.n))
op->info &= ~OF_CHECKED;
op = op->a.n;
} else {
op = op->r.n;
}
} else {
op = (ptest(op1)) ? op->a.n : op->r.n;
}
break;
/* just evaluate an expression, also used as unconditional jump */
case XC( OC_EXEC ):
2005-09-27 08:48:00 +05:30
break;
/* branch, used in if-else and various loops */
case XC( OC_BR ):
2005-09-27 08:48:00 +05:30
op = istrue(L.v) ? op->a.n : op->r.n;
break;
/* initialize for-in loop */
case XC( OC_WALKINIT ):
2005-09-27 08:48:00 +05:30
hashwalk_init(L.v, iamarray(R.v));
break;
/* get next array item */
case XC( OC_WALKNEXT ):
op = hashwalk_next(L.v) ? op->a.n : op->r.n;
break;
case XC( OC_PRINT ):
case XC( OC_PRINTF ):
X.F = stdout;
2005-09-27 08:48:00 +05:30
if (op->r.n) {
X.rsm = newfile(R.s);
if (!X.rsm->F) {
if (opn == '|') {
2007-04-12 06:02:05 +05:30
X.rsm->F = popen(R.s, "w");
if (X.rsm->F == NULL)
2003-03-19 14:43:01 +05:30
bb_perror_msg_and_die("popen");
X.rsm->is_pipe = 1;
} else {
X.rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
}
}
X.F = X.rsm->F;
}
if ((opinfo & OPCLSMASK) == OC_PRINT) {
if (!op1) {
2007-05-18 04:33:35 +05:30
fputs(getvar_s(intvar[F0]), X.F);
} else {
while (op1) {
L.v = evaluate(nextarg(&op1), v1);
if (L.v->type & VF_NUMBER) {
fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]),
getvar_i(L.v), TRUE);
fputs(g_buf, X.F);
} else {
fputs(getvar_s(L.v), X.F);
}
2007-05-18 04:33:35 +05:30
if (op1) fputs(getvar_s(intvar[OFS]), X.F);
}
}
2007-05-18 04:33:35 +05:30
fputs(getvar_s(intvar[ORS]), X.F);
} else { /* OC_PRINTF */
L.s = awk_printf(op1);
fputs(L.s, X.F);
free((char*)L.s);
}
fflush(X.F);
break;
case XC( OC_DELETE ):
2005-09-27 08:48:00 +05:30
X.info = op1->info & OPCLSMASK;
if (X.info == OC_VAR) {
R.v = op1->l.v;
} else if (X.info == OC_FNARG) {
R.v = &fnargs[op1->l.i];
} else {
syntax_error(EMSG_NOT_ARRAY);
}
2005-09-27 08:48:00 +05:30
if (op1->r.n) {
clrvar(L.v);
L.s = getvar_s(evaluate(op1->r.n, v1));
hash_remove(iamarray(R.v), L.s);
} else {
clear_array(iamarray(R.v));
}
break;
case XC( OC_NEWSOURCE ):
g_progname = op->l.s;
break;
case XC( OC_RETURN ):
copyvar(res, L.v);
break;
case XC( OC_NEXTFILE ):
2005-09-27 08:48:00 +05:30
nextfile = TRUE;
case XC( OC_NEXT ):
2005-09-27 08:48:00 +05:30
nextrec = TRUE;
case XC( OC_DONE ):
clrvar(res);
break;
case XC( OC_EXIT ):
2005-09-27 08:48:00 +05:30
awk_exit(L.d);
/* -- recursive node type -- */
case XC( OC_VAR ):
2005-09-27 08:48:00 +05:30
L.v = op->l.v;
2007-05-18 04:33:35 +05:30
if (L.v == intvar[NF])
split_f0();
goto v_cont;
case XC( OC_FNARG ):
2005-09-27 08:48:00 +05:30
L.v = &fnargs[op->l.i];
2007-01-02 05:23:52 +05:30
v_cont:
res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v;
break;
case XC( OC_IN ):
setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0);
break;
case XC( OC_REGEXP ):
2005-09-27 08:48:00 +05:30
op1 = op;
2007-05-18 04:33:35 +05:30
L.s = getvar_s(intvar[F0]);
goto re_cont;
case XC( OC_MATCH ):
2005-09-27 08:48:00 +05:30
op1 = op->r.n;
2007-01-02 05:23:52 +05:30
re_cont:
X.re = as_regex(op1, &sreg);
R.i = regexec(X.re, L.s, 0, NULL, 0);
if (X.re == &sreg) regfree(X.re);
setvar_i(res, (R.i == 0 ? 1 : 0) ^ (opn == '!' ? 1 : 0));
break;
case XC( OC_MOVE ):
2005-09-27 08:48:00 +05:30
/* if source is a temporary string, jusk relink it to dest */
if (R.v == v1+1 && R.v->string) {
res = setvar_p(L.v, R.v->string);
R.v->string = NULL;
} else {
2005-09-27 08:48:00 +05:30
res = copyvar(L.v, R.v);
}
break;
case XC( OC_TERNARY ):
2005-09-27 08:48:00 +05:30
if ((op->r.n->info & OPCLSMASK) != OC_COLON)
syntax_error(EMSG_POSSIBLE_ERROR);
res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res);
break;
case XC( OC_FUNC ):
2007-05-18 04:33:35 +05:30
if (!op->r.f->body.first)
syntax_error(EMSG_UNDEF_FUNC);
X.v = R.v = nvalloc(op->r.f->nargs+1);
while (op1) {
L.v = evaluate(nextarg(&op1), v1);
copyvar(R.v, L.v);
R.v->type |= VF_CHILD;
R.v->x.parent = L.v;
if (++R.v - X.v >= op->r.f->nargs)
break;
}
R.v = fnargs;
fnargs = X.v;
L.s = g_progname;
res = evaluate(op->r.f->body.first, res);
g_progname = L.s;
nvfree(fnargs);
fnargs = R.v;
break;
case XC( OC_GETLINE ):
case XC( OC_PGETLINE ):
2005-09-27 08:48:00 +05:30
if (op1) {
X.rsm = newfile(L.s);
2007-05-18 04:33:35 +05:30
if (!X.rsm->F) {
if ((opinfo & OPCLSMASK) == OC_PGETLINE) {
X.rsm->F = popen(L.s, "r");
X.rsm->is_pipe = TRUE;
} else {
X.rsm->F = fopen(L.s, "r"); /* not xfopen! */
}
}
} else {
2007-05-18 04:33:35 +05:30
if (!iF) iF = next_input_file();
X.rsm = iF;
}
2007-05-18 04:33:35 +05:30
if (!X.rsm->F) {
setvar_i(intvar[ERRNO], errno);
setvar_i(res, -1);
break;
}
2007-05-18 04:33:35 +05:30
if (!op->r.n)
R.v = intvar[F0];
L.i = awk_getline(X.rsm, R.v);
if (L.i > 0) {
2007-05-18 04:33:35 +05:30
if (!op1) {
incvar(intvar[FNR]);
incvar(intvar[NR]);
}
}
setvar_i(res, L.i);
break;
/* simple builtins */
case XC( OC_FBLTIN ):
2005-09-27 08:48:00 +05:30
switch (opn) {
case F_in:
2005-09-27 08:48:00 +05:30
R.d = (int)L.d;
break;
case F_rn:
R.d = (double)rand() / (double)RAND_MAX;
break;
#if ENABLE_FEATURE_AWK_MATH
case F_co:
2005-09-27 08:48:00 +05:30
R.d = cos(L.d);
break;
case F_ex:
2005-09-27 08:48:00 +05:30
R.d = exp(L.d);
break;
case F_lg:
2005-09-27 08:48:00 +05:30
R.d = log(L.d);
break;
case F_si:
2005-09-27 08:48:00 +05:30
R.d = sin(L.d);
break;
case F_sq:
2005-09-27 08:48:00 +05:30
R.d = sqrt(L.d);
break;
#else
case F_co:
case F_ex:
case F_lg:
case F_si:
case F_sq:
syntax_error(EMSG_NO_MATH);
break;
#endif
case F_sr:
R.d = (double)seed;
seed = op1 ? (unsigned)L.d : (unsigned)time(NULL);
srand(seed);
break;
case F_ti:
R.d = time(NULL);
break;
case F_le:
2007-05-18 04:33:35 +05:30
if (!op1)
L.s = getvar_s(intvar[F0]);
R.d = strlen(L.s);
break;
case F_sy:
fflush(NULL);
R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
? (system(L.s) >> 8) : 0;
break;
case F_ff:
2007-05-18 04:33:35 +05:30
if (!op1)
fflush(stdout);
else {
if (L.s && *L.s) {
X.rsm = newfile(L.s);
fflush(X.rsm->F);
} else {
fflush(NULL);
}
}
break;
case F_cl:
X.rsm = (rstream *)hash_search(fdhash, L.s);
if (X.rsm) {
R.i = X.rsm->is_pipe ? pclose(X.rsm->F) : fclose(X.rsm->F);
2002-11-28 16:57:31 +05:30
free(X.rsm->buffer);
hash_remove(fdhash, L.s);
}
if (R.i != 0)
2007-05-18 04:33:35 +05:30
setvar_i(intvar[ERRNO], errno);
R.d = (double)R.i;
break;
}
setvar_i(res, R.d);
break;
case XC( OC_BUILTIN ):
res = exec_builtin(op, res);
break;
case XC( OC_SPRINTF ):
2005-09-27 08:48:00 +05:30
setvar_p(res, awk_printf(op1));
break;
case XC( OC_UNARY ):
2005-09-27 08:48:00 +05:30
X.v = R.v;
L.d = R.d = getvar_i(R.v);
switch (opn) {
case 'P':
2005-09-27 08:48:00 +05:30
L.d = ++R.d;
goto r_op_change;
case 'p':
2005-09-27 08:48:00 +05:30
R.d++;
goto r_op_change;
case 'M':
2005-09-27 08:48:00 +05:30
L.d = --R.d;
goto r_op_change;
case 'm':
2005-09-27 08:48:00 +05:30
R.d--;
goto r_op_change;
case '!':
2006-10-04 01:26:34 +05:30
L.d = istrue(X.v) ? 0 : 1;
break;
case '-':
2005-09-27 08:48:00 +05:30
L.d = -R.d;
break;
r_op_change:
setvar_i(X.v, R.d);
}
setvar_i(res, L.d);
break;
case XC( OC_FIELD ):
2005-09-27 08:48:00 +05:30
R.i = (int)getvar_i(R.v);
if (R.i == 0) {
2007-05-18 04:33:35 +05:30
res = intvar[F0];
} else {
split_f0();
if (R.i > nfields)
fsrealloc(R.i);
2007-05-18 04:33:35 +05:30
res = &Fields[R.i - 1];
}
break;
/* concatenation (" ") and index joining (",") */
case XC( OC_CONCAT ):
case XC( OC_COMMA ):
opn = strlen(L.s) + strlen(R.s) + 2;
2006-12-20 05:06:04 +05:30
X.s = xmalloc(opn);
strcpy(X.s, L.s);
if ((opinfo & OPCLSMASK) == OC_COMMA) {
2007-05-18 04:33:35 +05:30
L.s = getvar_s(intvar[SUBSEP]);
X.s = xrealloc(X.s, opn + strlen(L.s));
strcat(X.s, L.s);
}
strcat(X.s, R.s);
setvar_p(res, X.s);
break;
case XC( OC_LAND ):
setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0);
break;
case XC( OC_LOR ):
setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n));
break;
case XC( OC_BINARY ):
case XC( OC_REPLACE ):
2005-09-27 08:48:00 +05:30
R.d = getvar_i(R.v);
switch (opn) {
case '+':
2005-09-27 08:48:00 +05:30
L.d += R.d;
break;
case '-':
2005-09-27 08:48:00 +05:30
L.d -= R.d;
break;
case '*':
2005-09-27 08:48:00 +05:30
L.d *= R.d;
break;
case '/':
if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
2005-09-27 08:48:00 +05:30
L.d /= R.d;
break;
case '&':
#if ENABLE_FEATURE_AWK_MATH
2005-09-27 08:48:00 +05:30
L.d = pow(L.d, R.d);
#else
syntax_error(EMSG_NO_MATH);
#endif
break;
case '%':
if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
2005-09-27 08:48:00 +05:30
L.d -= (int)(L.d / R.d) * R.d;
break;
}
2007-05-18 04:33:35 +05:30
res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : X.v, L.d);
break;
case XC( OC_COMPARE ):
if (is_numeric(L.v) && is_numeric(R.v)) {
L.d = getvar_i(L.v) - getvar_i(R.v);
} else {
L.s = getvar_s(L.v);
R.s = getvar_s(R.v);
L.d = icase ? strcasecmp(L.s, R.s) : strcmp(L.s, R.s);
}
switch (opn & 0xfe) {
case 0:
2005-09-27 08:48:00 +05:30
R.i = (L.d > 0);
break;
case 2:
2005-09-27 08:48:00 +05:30
R.i = (L.d >= 0);
break;
case 4:
2005-09-27 08:48:00 +05:30
R.i = (L.d == 0);
break;
}
setvar_i(res, (opn & 0x1 ? R.i : !R.i) ? 1 : 0);
break;
default:
syntax_error(EMSG_POSSIBLE_ERROR);
}
if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS)
op = op->a.n;
if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS)
break;
if (nextrec)
break;
}
nvfree(v1);
return res;
#undef fnargs
#undef seed
#undef sreg
}
/* -------- main & co. -------- */
2005-09-27 07:53:02 +05:30
static int awk_exit(int r)
{
var tv;
unsigned i;
hash_item *hi;
zero_out_var(&tv);
if (!exiting) {
exiting = TRUE;
nextrec = FALSE;
evaluate(endseq.first, &tv);
}
/* waiting for children */
for (i = 0; i < fdhash->csize; i++) {
hi = fdhash->items[i];
while (hi) {
if (hi->data.rs.F && hi->data.rs.is_pipe)
pclose(hi->data.rs.F);
hi = hi->next;
}
}
exit(r);
}
/* if expr looks like "var=value", perform assignment and return 1,
* otherwise return 0 */
static int is_assignment(const char *expr)
2005-09-27 07:53:02 +05:30
{
char *exprc, *s, *s0, *s1;
exprc = xstrdup(expr);
if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) {
free(exprc);
return FALSE;
}
*(s++) = '\0';
s0 = s1 = s;
while (*s)
*(s1++) = nextchar(&s);
*s1 = '\0';
setvar_u(newvar(exprc), s0);
free(exprc);
return TRUE;
}
/* switch to next input file */
2005-09-27 07:53:02 +05:30
static rstream *next_input_file(void)
{
#define rsm (G.next_input_file__rsm)
#define files_happen (G.next_input_file__files_happen)
2007-05-18 04:33:35 +05:30
FILE *F = NULL;
const char *fname, *ind;
if (rsm.F) fclose(rsm.F);
rsm.F = NULL;
rsm.pos = rsm.adv = 0;
do {
2007-05-18 04:33:35 +05:30
if (getvar_i(intvar[ARGIND])+1 >= getvar_i(intvar[ARGC])) {
if (files_happen)
return NULL;
fname = "-";
F = stdin;
} else {
2007-05-18 04:33:35 +05:30
ind = getvar_s(incvar(intvar[ARGIND]));
fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind));
if (fname && *fname && !is_assignment(fname))
F = afopen(fname, "r");
}
} while (!F);
files_happen = TRUE;
2007-05-18 04:33:35 +05:30
setvar_s(intvar[FILENAME], fname);
rsm.F = F;
return &rsm;
#undef rsm
#undef files_happen
}
int awk_main(int argc, char **argv);
int awk_main(int argc, char **argv)
2005-09-27 07:53:02 +05:30
{
unsigned opt;
2007-03-10 22:52:14 +05:30
char *opt_F, *opt_W;
llist_t *opt_v = NULL;
int i, j, flen;
var *v;
var tv;
char **envp;
char *vnames = (char *)vNames; /* cheat */
char *vvalues = (char *)vValues;
INIT_G();
/* Undo busybox.c, or else strtod may eat ','! This breaks parsing:
* $1,$2 == '$1,' '$2', NOT '$1' ',' '$2' */
if (ENABLE_LOCALE_SUPPORT)
setlocale(LC_NUMERIC, "C");
zero_out_var(&tv);
/* allocate global buffer */
g_buf = xmalloc(MAXVARFMT + 1);
vhash = hash_init();
ahash = hash_init();
fdhash = hash_init();
fnhash = hash_init();
/* initialize variables */
for (i = 0; *vnames; i++) {
2007-05-18 04:33:35 +05:30
intvar[i] = v = newvar(nextword(&vnames));
if (*vvalues != '\377')
setvar_s(v, nextword(&vvalues));
else
setvar_i(v, 0);
if (*vnames == '*') {
v->type |= VF_SPECIAL;
vnames++;
}
}
2007-05-18 04:33:35 +05:30
handle_special(intvar[FS]);
handle_special(intvar[RS]);
newfile("/dev/stdin")->F = stdin;
newfile("/dev/stdout")->F = stdout;
newfile("/dev/stderr")->F = stderr;
2007-05-04 04:27:56 +05:30
/* Huh, people report that sometimes environ is NULL. Oh well. */
if (environ) for (envp = environ; *envp; envp++) {
char *s = xstrdup(*envp);
char *s1 = strchr(s, '=');
if (s1) {
*s1++ = '\0';
2007-05-18 04:33:35 +05:30
setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1);
}
free(s);
}
2007-03-10 22:52:14 +05:30
opt_complementary = "v::";
opt = getopt32(argc, argv, "F:v:f:W:", &opt_F, &opt_v, &g_progname, &opt_W);
argv += optind;
argc -= optind;
2007-05-18 04:33:35 +05:30
if (opt & 0x1)
setvar_s(intvar[FS], opt_F); // -F
2007-03-10 22:52:14 +05:30
while (opt_v) { /* -v */
if (!is_assignment(llist_pop(&opt_v)))
bb_show_usage();
}
2006-09-22 14:32:30 +05:30
if (opt & 0x4) { // -f
char *s = s; /* die, gcc, die */
FILE *from_file = afopen(g_progname, "r");
2006-09-22 14:32:30 +05:30
/* one byte is reserved for some trick in next_token */
if (fseek(from_file, 0, SEEK_END) == 0) {
flen = ftell(from_file);
s = xmalloc(flen + 4);
fseek(from_file, 0, SEEK_SET);
i = 1 + fread(s + 1, 1, flen, from_file);
2006-09-22 14:32:30 +05:30
} else {
for (i = j = 1; j > 0; i += j) {
s = xrealloc(s, i + 4096);
j = fread(s + i, 1, 4094, from_file);
2006-09-22 14:32:30 +05:30
}
}
2006-09-22 14:32:30 +05:30
s[i] = '\0';
fclose(from_file);
parse_program(s + 1);
2006-09-22 14:32:30 +05:30
free(s);
} else { // no -f: take program from 1st parameter
if (!argc)
2003-03-19 14:43:01 +05:30
bb_show_usage();
g_progname = "cmd. line";
parse_program(*argv++);
argc--;
}
if (opt & 0x8) // -W
bb_error_msg("warning: unrecognized option '-W %s' ignored", opt_W);
/* fill in ARGV array */
2007-05-18 04:33:35 +05:30
setvar_i(intvar[ARGC], argc + 1);
setari_u(intvar[ARGV], 0, "awk");
i = 0;
while (*argv)
2007-05-18 04:33:35 +05:30
setari_u(intvar[ARGV], ++i, *argv++);
evaluate(beginseq.first, &tv);
if (!mainseq.first && !endseq.first)
awk_exit(EXIT_SUCCESS);
/* input file could already be opened in BEGIN block */
if (!iF) iF = next_input_file();
/* passing through input files */
while (iF) {
nextfile = FALSE;
2007-05-18 04:33:35 +05:30
setvar_i(intvar[FNR], 0);
2007-05-18 04:33:35 +05:30
while ((i = awk_getline(iF, intvar[F0])) > 0) {
nextrec = FALSE;
2007-05-18 04:33:35 +05:30
incvar(intvar[NR]);
incvar(intvar[FNR]);
evaluate(mainseq.first, &tv);
if (nextfile)
break;
}
if (i < 0)
syntax_error(strerror(errno));
iF = next_input_file();
}
awk_exit(EXIT_SUCCESS);
/*return 0;*/
}