#include "libbb.h"
#include "xregex.h"
#include <math.h>
-extern char **environ;
/* This is a NOEXEC applet. Be very careful! */
/* Variable */
typedef struct var_s {
- unsigned short type; /* flags */
+ unsigned type; /* flags */
double number;
char *string;
union {
/* Function */
typedef struct func_s {
- unsigned short nargs;
+ unsigned nargs;
struct chain_s body;
} func;
int adv;
int size;
int pos;
- unsigned short is_pipe;
+ smallint is_pipe;
} rstream;
typedef struct hash_item_s {
#define OC_B OC_BUILTIN
-static const char tokenlist[] =
+static const char tokenlist[] ALIGN1 =
"\1(" NTC
"\1)" NTC
"\1/" NTC /* REGEXP */
enum {
CONVFMT, OFMT, FS, OFS,
ORS, RS, RT, FILENAME,
- SUBSEP, ARGIND, ARGC, ARGV,
- ERRNO, FNR,
- NR, NF, IGNORECASE,
- ENVIRON, F0, NUM_INTERNAL_VARS
+ SUBSEP, F0, ARGIND, ARGC,
+ ARGV, ERRNO, FNR, NR,
+ NF, IGNORECASE, ENVIRON, NUM_INTERNAL_VARS
};
-static const char vNames[] =
+static const char vNames[] ALIGN1 =
"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" "$\0*" "\0";
+ "SUBSEP\0" "$\0*" "ARGIND\0" "ARGC\0"
+ "ARGV\0" "ERRNO\0" "FNR\0" "NR\0"
+ "NF\0*" "IGNORECASE\0*" "ENVIRON\0" "\0";
-static const char vValues[] =
+static const char vValues[] ALIGN1 =
"%.6g\0" "%.6g\0" " \0" " \0"
"\n\0" "\n\0" "\0" "\0"
- "\034\0"
- "\377";
+ "\034\0" "\0" "\377";
/* hash size may grow to these values */
-#define FIRST_PRIME 61;
-static const unsigned PRIMES[] = { 251, 1021, 4093, 16381, 65521 };
-enum { NPRIMES = sizeof(PRIMES) / sizeof(PRIMES[0]) };
-
-/* globals */
-
-static var *intvar[NUM_INTERNAL_VARS];
-static chain beginseq, mainseq, endseq, *seq;
-static int nextrec, nextfile;
-static node *break_ptr, *continue_ptr;
-static rstream *iF;
-static xhash *vhash, *ahash, *fdhash, *fnhash;
-static const char *programname;
-static int lineno;
-static int is_f0_split;
-static int nfields;
-static var *Fields;
-static tsplitter fsplitter, rsplitter;
-static nvblock *cb;
-static char *pos;
-static char *buf;
-static int icase;
-static int exiting;
-
-static struct {
- uint32_t tclass;
- uint32_t info;
- char *string;
- double number;
- int lineno;
- int rollback;
-} ttt;
-/* It had even better name: 't'. Whoever knows what is it, please rename! */
-/* (actually it looks like unrelated stuff lumped together...) */
+#define FIRST_PRIME 61
+static const uint16_t PRIMES[] ALIGN2 = { 251, 1021, 4093, 16381, 65521 };
+
+
+/* Globals. Split in two parts so that first one is addressed
+ * with (mostly short) negative offsets.
+ * NB: it's unsafe to put members of type "double"
+ * into globals2 (gcc may fail to align them).
+ */
+struct globals {
+ double t_double;
+ chain beginseq, mainseq, endseq;
+ chain *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 */
+ tsplitter fsplitter, rsplitter;
+};
+#define G1 (ptr_to_globals[-1])
+#define G (*(struct globals2 *)ptr_to_globals)
+/* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */
+/*char G1size[sizeof(G1)]; - 0x74 */
+/*char Gsize[sizeof(G)]; - 0x1c4 */
+/* Trying to keep most of members accessible with short offsets: */
+/*char Gofs_seed[offsetof(struct globals2, evaluate__seed)]; - 0x90 */
+#define t_double (G1.t_double )
+#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_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 { \
+ SET_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 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;
+static int awk_exit(int) 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";
+static const char EMSG_INTERNAL_ERROR[] ALIGN1 = "Internal error";
+static const char EMSG_UNEXP_EOS[] ALIGN1 = "Unexpected end of string";
+static const char EMSG_UNEXP_TOKEN[] ALIGN1 = "Unexpected token";
+static const char EMSG_DIV_BY_ZERO[] ALIGN1 = "Division by zero";
+static const char EMSG_INV_FMT[] ALIGN1 = "Invalid format specifier";
+static const char EMSG_TOO_FEW_ARGS[] ALIGN1 = "Too few arguments for builtin";
+static const char EMSG_NOT_ARRAY[] ALIGN1 = "Not an array";
+static const char EMSG_POSSIBLE_ERROR[] ALIGN1 = "Possible syntax error";
+static const char EMSG_UNDEF_FUNC[] ALIGN1 = "Call to undefined function";
+#if !ENABLE_FEATURE_AWK_LIBM
+static const char EMSG_NO_MATH[] ALIGN1 = "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)
+static void syntax_error(const char *message) NORETURN;
+static void syntax_error(const char *message)
{
- bb_error_msg_and_die("%s:%i: %s", programname, lineno, message);
+ bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message);
}
-#define runtime_error(x) syntax_error(x)
-
-
/* ---- hash stuff ---- */
static unsigned hashidx(const char *name)
unsigned newsize, i, idx;
hash_item **newitems, *hi, *thi;
- if (hash->nprime == NPRIMES)
+ if (hash->nprime == ARRAY_SIZE(PRIMES))
return;
newsize = PRIMES[hash->nprime++];
int l;
hi = hash_search(hash, name);
- if (! hi) {
+ 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);
+ hi = xzalloc(sizeof(*hi) + l);
+ strcpy(hi->name, name);
idx = hashidx(name) % hash->csize;
hi->next = hash->items[idx];
while (1) {
if (*p == '\\' && p[1] == '\n') {
p++;
- ttt.lineno++;
+ t_lineno++;
} else if (*p != ' ' && *p != '\t') {
break;
}
return c;
}
-static int ATTRIBUTE_ALWAYS_INLINE isalnum_(int c)
+static ALWAYS_INLINE int isalnum_(int c)
{
return (isalnum(c) || c == '_');
}
-static FILE *afopen(const char *path, const char *mode)
+static double my_strtod(char **pp)
{
- return (*path == '-' && *(path+1) == '\0') ? stdin : xfopen(path, mode);
+#if ENABLE_DESKTOP
+ if ((*pp)[0] == '0'
+ && ((((*pp)[1] | 0x20) == 'x') || isdigit((*pp)[1]))
+ ) {
+ return strtoull(*pp, pp, 0);
+ }
+#endif
+ return strtod(*pp, pp);
}
/* -------- working with variables (set/get/copy/etc) -------- */
{
/* if v is numeric and has no cached string, convert it to string */
if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) {
- fmt_num(buf, MAXVARFMT, getvar_s(intvar[CONVFMT]), v->number, TRUE);
- v->string = xstrdup(buf);
+ 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;
v->number = 0;
s = v->string;
if (s && *s) {
- v->number = strtod(s, &s);
+ v->number = my_strtod(&s);
if (v->type & VF_USER) {
skip_spaces(&s);
if (*s != '\0')
return v->number;
}
+/* Used for operands of bitwise ops */
+static unsigned long getvar_i_int(var *v)
+{
+ double d = getvar_i(v);
+
+ /* Casting doubles to longs is undefined for values outside
+ * of target type range. Try to widen it as much as possible */
+ if (d >= 0)
+ return (unsigned long)d;
+ /* Why? Think about d == -4294967295.0 (assuming 32bit longs) */
+ return - (long) (unsigned long) (-d);
+}
+
static var *copyvar(var *dest, const var *src)
{
if (dest != src) {
static var *incvar(var *v)
{
- return setvar_i(v, getvar_i(v)+1.);
+ return setvar_i(v, getvar_i(v) + 1.);
}
/* return true if v is number or numeric string */
{
if (is_numeric(v))
return (v->number == 0) ? 0 : 1;
- else
- return (v->string && *(v->string)) ? 1 : 0;
+ return (v->string && *(v->string)) ? 1 : 0;
}
/* temporary variables allocator. Last allocated should be first freed */
var *v, *r;
int size;
- while (cb) {
- pb = cb;
- if ((cb->pos - cb->nv) + n <= cb->size) break;
- cb = cb->next;
+ while (g_cb) {
+ pb = g_cb;
+ if ((g_cb->pos - g_cb->nv) + n <= g_cb->size) break;
+ g_cb = g_cb->next;
}
- if (! cb) {
+ if (!g_cb) {
size = (n <= MINNVBLOCK) ? MINNVBLOCK : n;
- cb = xmalloc(sizeof(nvblock) + size * sizeof(var));
- cb->size = size;
- cb->pos = cb->nv;
- cb->prev = pb;
- cb->next = NULL;
- if (pb) pb->next = cb;
+ g_cb = xzalloc(sizeof(nvblock) + size * sizeof(var));
+ g_cb->size = size;
+ g_cb->pos = g_cb->nv;
+ g_cb->prev = pb;
+ /*g_cb->next = NULL; - xzalloc did it */
+ if (pb) pb->next = g_cb;
}
- v = r = cb->pos;
- cb->pos += n;
+ v = r = g_cb->pos;
+ g_cb->pos += n;
- while (v < cb->pos) {
+ while (v < g_cb->pos) {
v->type = 0;
v->string = NULL;
v++;
{
var *p;
- if (v < cb->nv || v >= cb->pos)
- runtime_error(EMSG_INTERNAL_ERROR);
+ if (v < g_cb->nv || v >= g_cb->pos)
+ syntax_error(EMSG_INTERNAL_ERROR);
- for (p = v; p < cb->pos; p++) {
+ for (p = v; p < g_cb->pos; p++) {
if ((p->type & (VF_ARRAY | VF_CHILD)) == VF_ARRAY) {
clear_array(iamarray(p));
free(p->x.array->items);
clrvar(p);
}
- cb->pos = v;
- while (cb->prev && cb->pos == cb->nv) {
- cb = cb->prev;
+ g_cb->pos = v;
+ while (g_cb->prev && g_cb->pos == g_cb->nv) {
+ g_cb = g_cb->prev;
}
}
*/
static uint32_t next_token(uint32_t expected)
{
- static int concat_inserted;
- static uint32_t save_tclass, save_info;
- static uint32_t ltclass = TC_OPTERM;
+#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;
const uint32_t *ti;
int l;
- if (ttt.rollback) {
- ttt.rollback = FALSE;
+ if (t_rollback) {
+ t_rollback = FALSE;
} else if (concat_inserted) {
concat_inserted = FALSE;
- ttt.tclass = save_tclass;
- ttt.info = save_info;
+ t_tclass = save_tclass;
+ t_info = save_info;
} else {
- p = pos;
+ p = g_pos;
readnext:
skip_spaces(&p);
- lineno = ttt.lineno;
+ g_lineno = t_lineno;
if (*p == '#')
while (*p != '\n' && *p != '\0')
p++;
if (*p == '\n')
- ttt.lineno++;
+ t_lineno++;
if (*p == '\0') {
tc = TC_EOF;
} else if (*p == '\"') {
/* it's a string */
- ttt.string = s = ++p;
+ t_string = s = ++p;
while (*p != '\"') {
if (*p == '\0' || *p == '\n')
syntax_error(EMSG_UNEXP_EOS);
} else if ((expected & TC_REGEXP) && *p == '/') {
/* it's regexp */
- ttt.string = s = ++p;
+ t_string = s = ++p;
while (*p != '/') {
if (*p == '\0' || *p == '\n')
syntax_error(EMSG_UNEXP_EOS);
} else if (*p == '.' || isdigit(*p)) {
/* it's a number */
- ttt.number = strtod(p, &p);
+ t_double = my_strtod(&p);
if (*p == '.')
syntax_error(EMSG_UNEXP_TOKEN);
tc = TC_NUMBER;
&& *tl == *p && strncmp(p, tl, l) == 0
&& !((tc & TC_WORD) && isalnum_(p[l]))
) {
- ttt.info = *ti;
+ t_info = *ti;
p += l;
break;
}
if (!isalnum_(*p))
syntax_error(EMSG_UNEXP_TOKEN);
- ttt.string = --p;
+ t_string = --p;
while (isalnum_(*(++p))) {
*(p-1) = *p;
}
}
}
}
- pos = p;
+ g_pos = p;
/* skipping newlines in some cases */
if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE))
if ((ltclass & TC_CONCAT1) && (tc & TC_CONCAT2) && (expected & TC_BINOP)) {
concat_inserted = TRUE;
save_tclass = tc;
- save_info = ttt.info;
+ save_info = t_info;
tc = TC_BINOP;
- ttt.info = OC_CONCAT | SS | P(35);
+ t_info = OC_CONCAT | SS | P(35);
}
- ttt.tclass = tc;
+ t_tclass = tc;
}
- ltclass = ttt.tclass;
+ ltclass = t_tclass;
/* Are we ready for this? */
if (!(ltclass & expected))
EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN);
return ltclass;
+#undef concat_inserted
+#undef save_tclass
+#undef save_info
+#undef ltclass
}
static void rollback_token(void)
{
- ttt.rollback = TRUE;
+ t_rollback = TRUE;
}
static node *new_node(uint32_t info)
n = xzalloc(sizeof(node));
n->info = info;
- n->lineno = lineno;
+ n->lineno = g_lineno;
return n;
}
xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp;
while (!((tc = next_token(xtc)) & iexp)) {
- if (glptr && (ttt.info == (OC_COMPARE | VV | P(39) | 2))) {
+ if (glptr && (t_info == (OC_COMPARE | VV | P(39) | 2))) {
/* input redirection (<) attached to glptr node */
cn = glptr->l.n = new_node(OC_CONCAT | SS | P(37));
cn->a.n = glptr;
/* for binary and postfix-unary operators, jump back over
* previous operators with higher priority */
vn = cn;
- while ( ((ttt.info & PRIMASK) > (vn->a.n->info & PRIMASK2))
- || ((ttt.info == vn->info) && ((ttt.info & OPCLSMASK) == OC_COLON)) )
+ while ( ((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2))
+ || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON)) )
vn = vn->a.n;
- if ((ttt.info & OPCLSMASK) == OC_TERNARY)
- ttt.info += P(6);
- cn = vn->a.n->r.n = new_node(ttt.info);
+ 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 ((ttt.info & OPCLSMASK) == OC_PGETLINE) {
+ if ((t_info & OPCLSMASK) == OC_PGETLINE) {
/* it's a pipe */
next_token(TC_GETLINE);
/* give maximum priority to this pipe */
/* for operands and prefix-unary operators, attach them
* to last node */
vn = cn;
- cn = vn->r.n = new_node(ttt.info);
+ 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)) {
case TC_VARIABLE:
case TC_ARRAY:
cn->info = OC_VAR;
- v = hash_search(ahash, ttt.string);
+ v = hash_search(ahash, t_string);
if (v != NULL) {
cn->info = OC_FNARG;
cn->l.i = v->x.aidx;
} else {
- cn->l.v = newvar(ttt.string);
+ cn->l.v = newvar(t_string);
}
if (tc & TC_ARRAY) {
cn->info |= xS;
cn->info = OC_VAR;
v = cn->l.v = xzalloc(sizeof(var));
if (tc & TC_NUMBER)
- setvar_i(v, ttt.number);
+ setvar_i(v, t_double);
else
- setvar_s(v, ttt.string);
+ setvar_s(v, t_string);
break;
case TC_REGEXP:
- mk_re_node(ttt.string, cn, xzalloc(sizeof(regex_t)*2));
+ mk_re_node(t_string, cn, xzalloc(sizeof(regex_t)*2));
break;
case TC_FUNCTION:
cn->info = OC_FUNC;
- cn->r.f = newfunc(ttt.string);
+ cn->r.f = newfunc(t_string);
cn->l.n = condition();
break;
if (!seq->first)
seq->first = seq->last = new_node(0);
- if (seq->programname != programname) {
- seq->programname = programname;
+ if (seq->programname != g_progname) {
+ seq->programname = g_progname;
n = chain_node(OC_NEWSOURCE);
- n->l.s = xstrdup(programname);
+ n->l.s = xstrdup(g_progname);
}
n = seq->last;
n = chain_node(info);
n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM);
- if (ttt.tclass & TC_GRPTERM)
+ if (t_tclass & TC_GRPTERM)
rollback_token();
}
if (c & TC_GRPSTART) {
while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) {
- if (ttt.tclass & TC_NEWLINE) continue;
+ if (t_tclass & TC_NEWLINE) continue;
rollback_token();
chain_group();
}
rollback_token();
chain_expr(OC_EXEC | Vx);
} else { /* TC_STATEMNT */
- switch (ttt.info & OPCLSMASK) {
+ switch (t_info & OPCLSMASK) {
case ST_IF:
n = chain_node(OC_BR | Vx);
n->l.n = condition();
case ST_FOR:
next_token(TC_SEQSTART);
n2 = parse_expr(TC_SEMICOL | TC_SEQTERM);
- if (ttt.tclass & TC_SEQTERM) { /* for-in */
+ if (t_tclass & TC_SEQTERM) { /* for-in */
if ((n2->info & OPCLSMASK) != OC_IN)
syntax_error(EMSG_UNEXP_TOKEN);
n = chain_node(OC_WALKINIT | VV);
n3 = parse_expr(TC_SEQTERM);
n = chain_loop(n3);
n->l.n = n2;
- if (! n2)
+ if (!n2)
n->info = OC_EXEC;
}
break;
case OC_PRINT:
case OC_PRINTF:
- n = chain_node(ttt.info);
+ n = chain_node(t_info);
n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM);
- if (ttt.tclass & TC_OUTRDR) {
- n->info |= ttt.info;
+ if (t_tclass & TC_OUTRDR) {
+ n->info |= t_info;
n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM);
}
- if (ttt.tclass & TC_GRPTERM)
+ if (t_tclass & TC_GRPTERM)
rollback_token();
break;
/* delete, next, nextfile, return, exit */
default:
- chain_expr(ttt.info);
+ chain_expr(t_info);
}
}
}
func *f;
var *v;
- pos = p;
- ttt.lineno = 1;
+ g_pos = p;
+ t_lineno = 1;
while ((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART |
TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) {
} else if (tclass & TC_FUNCDECL) {
next_token(TC_FUNCTION);
- pos++;
- f = newfunc(ttt.string);
+ 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, ttt.string);
+ v = findvar(ahash, t_string);
v->x.aidx = (f->nargs)++;
if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM)
rollback_token();
cn = chain_node(OC_TEST);
cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART);
- if (ttt.tclass & TC_GRPSTART) {
+ if (t_tclass & TC_GRPSTART) {
rollback_token();
chain_group();
} else {
n = &spl->n;
if ((n->info & OPCLSMASK) == OC_REGEXP) {
regfree(re);
- regfree(ire);
+ regfree(ire); // TODO: nuke ire, use re+1?
}
if (strlen(s) > 1) {
mk_re_node(s, n, re);
*/
static regex_t *as_regex(node *op, regex_t *preg)
{
+ int cflags;
var *v;
const char *s;
}
v = nvalloc(1);
s = getvar_s(evaluate(op, v));
- xregcomp(preg, s, icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED);
+
+ cflags = icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED;
+ /* Testcase where REG_EXTENDED fails (unpaired '{'):
+ * echo Hi | awk 'gsub("@(samp|code|file)\{","");'
+ * gawk 3.1.5 eats this. We revert to ~REG_EXTENDED
+ * (maybe gsub is not supposed to use REG_EXTENDED?).
+ */
+ if (regcomp(preg, s, cflags)) {
+ cflags &= ~REG_EXTENDED;
+ xregcomp(preg, s, cflags);
+ }
nvfree(v);
return preg;
}
/* gradually increasing buffer */
static void qrealloc(char **b, int n, int *size)
{
- if (!*b || n >= *size)
- *b = xrealloc(*b, *size = n + (n>>1) + 80);
+ if (!*b || n >= *size) {
+ *size = n + (n>>1) + 80;
+ *b = xrealloc(*b, *size);
+ }
}
/* resize field storage space */
static void fsrealloc(int size)
{
- static int maxfields; /* = 0;*/
int i;
if (size >= maxfields) {
int l, n = 0;
char c[4];
char *s1;
- regmatch_t pmatch[2];
+ regmatch_t pmatch[2]; // TODO: why [2]? [1] is enough...
/* in worst case, each char would be a separate field */
*slist = s1 = xzalloc(strlen(s) * 2 + 3);
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 ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */
+ if (!*s)
+ return n; /* "": zero fields */
+ n++; /* at least one field will be there */
+ do {
+ l = strcspn(s, c+2); /* len till next NUL or \n */
if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0
&& pmatch[0].rm_so <= l
) {
l++;
pmatch[0].rm_eo++;
}
+ n++; /* we saw yet another delimiter */
} else {
pmatch[0].rm_eo = l;
- if (s[l]) pmatch[0].rm_eo++;
+ if (s[l])
+ pmatch[0].rm_eo++;
}
-
memcpy(s1, s, l);
- s1[l] = '\0';
+ /* make sure we remove *all* of the separator chars */
+ do {
+ s1[l] = '\0';
+ } while (++l < pmatch[0].rm_eo);
nextword(&s1);
s += pmatch[0].rm_eo;
- n++;
- }
- } else if (c[0] == '\0') { /* null split */
+ } while (*s);
+ return n;
+ }
+ if (c[0] == '\0') { /* null split */
while (*s) {
*s1++ = *s++;
*s1++ = '\0';
n++;
}
- } else if (c[0] != ' ') { /* single-character split */
+ return n;
+ }
+ if (c[0] != ' ') { /* single-character split */
if (icase) {
c[0] = toupper(c[0]);
c[1] = tolower(c[1]);
*s1++ = '\0';
n++;
}
- } else { /* space split */
- while (*s) {
- s = skip_whitespace(s);
- if (!*s) break;
- n++;
- while (*s && !isspace(*s))
- *s1++ = *s++;
- *s1++ = '\0';
- }
+ return n;
+ }
+ /* space split */
+ while (*s) {
+ s = skip_whitespace(s);
+ if (!*s) break;
+ n++;
+ while (*s && !isspace(*s))
+ *s1++ = *s++;
+ *s1++ = '\0';
}
return n;
}
static void split_f0(void)
{
- static char *fstrings = NULL;
+/* static char *fstrings; */
+#define fstrings (G.split_f0__fstrings)
int i, n;
char *s;
clrvar(intvar[NF]);
intvar[NF]->type = VF_NUMBER | VF_SPECIAL;
intvar[NF]->number = nfields;
+#undef fstrings
}
/* perform additional actions when some internal variables changed */
{
char **w;
hash_item *hi;
- int i;
+ unsigned i;
if (v->type & VF_WALK)
free(v->x.walker);
/* evaluate node, return 1 when result is true, 0 otherwise */
static int ptest(node *pattern)
{
- static var v; /* static: to save stack space? */
-
- return istrue(evaluate(pattern, &v));
+ /* 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 */
c = (char) rsplitter.n.info;
rp = 0;
- if (! m) qrealloc(&m, 256, &size);
+ if (!m) qrealloc(&m, 256, &size);
do {
b = m + a;
so = eo = p;
}
} else if (c != '\0') {
s = strchr(b+pp, c);
- if (! s) s = memchr(b+pp, '\0', p - pp);
+ if (!s) s = memchr(b+pp, '\0', p - pp);
if (s) {
so = eo = s-b;
eo++;
} else if (strchr("eEfgG", c)) {
r = snprintf(b, size, format, n);
} else {
- runtime_error(EMSG_INV_FMT);
+ syntax_error(EMSG_INV_FMT);
}
}
return r;
}
-
/* formatted output into an allocated buffer, return ptr to buffer */
static char *awk_printf(node *n)
{
regex_t sreg, *re;
re = as_regex(rn, &sreg);
- if (! src) src = intvar[F0];
- if (! dest) dest = intvar[F0];
+ if (!src) src = intvar[F0];
+ if (!dest) dest = intvar[F0];
i = di = 0;
sp = getvar_s(src);
sp += eo;
if (i == nm) break;
if (eo == so) {
- if (! (ds[di++] = *sp++)) break;
+ ds[di] = *sp++;
+ if (!ds[di++]) break;
}
}
static var *exec_builtin(node *op, var *res)
{
+#define tspl (G.exec_builtin__tspl)
+
int (*to_xxx)(int);
var *tv;
node *an[4];
const char *as[4];
regmatch_t pmatch[2];
regex_t sreg, *re;
- static tsplitter tspl;
node *spl;
uint32_t isr, info;
int nargs;
}
nargs = i;
- if (nargs < (info >> 30))
- runtime_error(EMSG_TOO_FEW_ARGS);
+ if ((uint32_t)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])));
+#if ENABLE_FEATURE_AWK_LIBM
+ setvar_i(res, atan2(getvar_i(av[0]), getvar_i(av[1])));
#else
- runtime_error(EMSG_NO_MATH);
+ syntax_error(EMSG_NO_MATH);
#endif
break;
n = awk_split(as[0], spl, &s);
s1 = s;
clear_array(iamarray(av[1]));
- for (i=1; i<=n; i++)
+ for (i = 1; i <= n; i++)
setari_u(av[1], i, nextword(&s1));
free(s);
setvar_i(res, n);
if (i < 0) i = 0;
n = (nargs > 2) ? getvar_i(av[2]) : l-i;
if (n < 0) n = 0;
- s = xmalloc(n+1);
- strncpy(s, as[0]+i, n);
- s[n] = '\0';
+ s = xstrndup(as[0]+i, n);
setvar_p(res, s);
break;
+ /* Bitwise ops must assume that operands are unsigned. GNU Awk 3.1.5:
+ * awk '{ print or(-1,1) }' gives "4.29497e+09", not "-2.xxxe+09" */
case B_an:
- setvar_i(res, (long)getvar_i(av[0]) & (long)getvar_i(av[1]));
+ setvar_i(res, getvar_i_int(av[0]) & getvar_i_int(av[1]));
break;
case B_co:
- setvar_i(res, ~(long)getvar_i(av[0]));
+ setvar_i(res, ~getvar_i_int(av[0]));
break;
case B_ls:
- setvar_i(res, (long)getvar_i(av[0]) << (long)getvar_i(av[1]));
+ setvar_i(res, getvar_i_int(av[0]) << getvar_i_int(av[1]));
break;
case B_or:
- setvar_i(res, (long)getvar_i(av[0]) | (long)getvar_i(av[1]));
+ setvar_i(res, getvar_i_int(av[0]) | getvar_i_int(av[1]));
break;
case B_rs:
- setvar_i(res, (long)((unsigned long)getvar_i(av[0]) >> (unsigned long)getvar_i(av[1])));
+ setvar_i(res, getvar_i_int(av[0]) >> getvar_i_int(av[1]));
break;
case B_xo:
- setvar_i(res, (long)getvar_i(av[0]) ^ (long)getvar_i(av[1]));
+ setvar_i(res, getvar_i_int(av[0]) ^ getvar_i_int(av[1]));
break;
case B_lo:
ll = strlen(as[1]);
l = strlen(as[0]) - ll;
if (ll > 0 && l >= 0) {
- if (! icase) {
+ if (!icase) {
s = strstr(as[0], as[1]);
if (s) n = (s - as[0]) + 1;
} else {
else
time(&tt);
//s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y";
- i = strftime(buf, MAXVARFMT,
+ i = strftime(g_buf, MAXVARFMT,
((nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"),
localtime(&tt));
- buf[i] = '\0';
- setvar_s(res, buf);
+ g_buf[i] = '\0';
+ setvar_s(res, g_buf);
break;
case B_ma:
nvfree(tv);
return res;
+#undef tspl
}
/*
static var *evaluate(node *op, var *res)
{
- /* This procedure is recursive so we should count every byte */
- static var *fnargs = NULL;
- static unsigned seed = 1;
- static regex_t sreg;
+/* 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)
node *op1;
var *v1;
int i;
} L, R;
uint32_t opinfo;
- short opn;
+ int opn;
union {
char *s;
rstream *rsm;
while (op) {
opinfo = op->info;
- opn = (short)(opinfo & OPNMASK);
- lineno = op->lineno;
+ opn = (opinfo & OPNMASK);
+ g_lineno = op->lineno;
/* execute inevitable things */
op1 = op->l.n;
X.F = stdout;
if (op->r.n) {
X.rsm = newfile(R.s);
- if (! X.rsm->F) {
+ if (!X.rsm->F) {
if (opn == '|') {
X.rsm->F = popen(R.s, "w");
if (X.rsm->F == NULL)
}
if ((opinfo & OPCLSMASK) == OC_PRINT) {
- if (! op1) {
+ if (!op1) {
fputs(getvar_s(intvar[F0]), X.F);
} else {
while (op1) {
L.v = evaluate(nextarg(&op1), v1);
if (L.v->type & VF_NUMBER) {
- fmt_num(buf, MAXVARFMT, getvar_s(intvar[OFMT]),
+ fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]),
getvar_i(L.v), TRUE);
- fputs(buf, X.F);
+ fputs(g_buf, X.F);
} else {
fputs(getvar_s(L.v), X.F);
}
} else if (X.info == OC_FNARG) {
R.v = &fnargs[op1->l.i];
} else {
- runtime_error(EMSG_NOT_ARRAY);
+ syntax_error(EMSG_NOT_ARRAY);
}
if (op1->r.n) {
break;
case XC( OC_NEWSOURCE ):
- programname = op->l.s;
+ g_progname = op->l.s;
break;
case XC( OC_RETURN ):
case XC( OC_TERNARY ):
if ((op->r.n->info & OPCLSMASK) != OC_COLON)
- runtime_error(EMSG_POSSIBLE_ERROR);
+ 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 ):
if (!op->r.f->body.first)
- runtime_error(EMSG_UNDEF_FUNC);
+ syntax_error(EMSG_UNDEF_FUNC);
X.v = R.v = nvalloc(op->r.f->nargs+1);
while (op1) {
R.v = fnargs;
fnargs = X.v;
- L.s = programname;
+ L.s = g_progname;
res = evaluate(op->r.f->body.first, res);
- programname = L.s;
+ g_progname = L.s;
nvfree(fnargs);
fnargs = R.v;
X.rsm->F = popen(L.s, "r");
X.rsm->is_pipe = TRUE;
} else {
- X.rsm->F = fopen(L.s, "r"); /* not xfopen! */
+ X.rsm->F = fopen_for_read(L.s); /* not xfopen! */
}
}
} else {
case F_rn:
R.d = (double)rand() / (double)RAND_MAX;
break;
-#if ENABLE_FEATURE_AWK_MATH
+#if ENABLE_FEATURE_AWK_LIBM
case F_co:
R.d = cos(L.d);
break;
case F_lg:
case F_si:
case F_sq:
- runtime_error(EMSG_NO_MATH);
+ syntax_error(EMSG_NO_MATH);
break;
#endif
case F_sr:
L.d *= R.d;
break;
case '/':
- if (R.d == 0) runtime_error(EMSG_DIV_BY_ZERO);
+ if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
L.d /= R.d;
break;
case '&':
-#if ENABLE_FEATURE_AWK_MATH
+#if ENABLE_FEATURE_AWK_LIBM
L.d = pow(L.d, R.d);
#else
- runtime_error(EMSG_NO_MATH);
+ syntax_error(EMSG_NO_MATH);
#endif
break;
case '%':
- if (R.d == 0) runtime_error(EMSG_DIV_BY_ZERO);
+ if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
L.d -= (int)(L.d / R.d) * R.d;
break;
}
break;
default:
- runtime_error(EMSG_POSSIBLE_ERROR);
+ syntax_error(EMSG_POSSIBLE_ERROR);
}
if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS)
op = op->a.n;
}
nvfree(v1);
return res;
+#undef fnargs
+#undef seed
+#undef sreg
}
/* switch to next input file */
static rstream *next_input_file(void)
{
- static rstream rsm;
- static int files_happen = FALSE;
+#define rsm (G.next_input_file__rsm)
+#define files_happen (G.next_input_file__files_happen)
FILE *F = NULL;
const char *fname, *ind;
ind = getvar_s(incvar(intvar[ARGIND]));
fname = getvar_s(findvar(iamarray(intvar[ARGV]), ind));
if (fname && *fname && !is_assignment(fname))
- F = afopen(fname, "r");
+ F = xfopen_stdin(fname);
}
} while (!F);
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) MAIN_EXTERNALLY_VISIBLE;
int awk_main(int argc, char **argv)
{
unsigned opt;
char *opt_F, *opt_W;
- llist_t *opt_v = NULL;
- int i, j, flen;
+ llist_t *list_v = NULL;
+ llist_t *list_f = NULL;
+ int i, j;
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)
zero_out_var(&tv);
/* allocate global buffer */
- buf = xmalloc(MAXVARFMT + 1);
+ g_buf = xmalloc(MAXVARFMT + 1);
vhash = hash_init();
ahash = hash_init();
/* Huh, people report that sometimes environ is NULL. Oh well. */
if (environ) for (envp = environ; *envp; envp++) {
- char *s = xstrdup(*envp);
+ /* environ is writable, thus we don't strdup it needlessly */
+ char *s = *envp;
char *s1 = strchr(s, '=');
if (s1) {
- *s1++ = '\0';
- setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1);
+ *s1 = '\0';
+ /* Both findvar and setvar_u take const char*
+ * as 2nd arg -> environment is not trashed */
+ setvar_u(findvar(iamarray(intvar[ENVIRON]), s), s1 + 1);
+ *s1 = '=';
}
- free(s);
}
- opt_complementary = "v::";
- opt = getopt32(argc, argv, "F:v:f:W:", &opt_F, &opt_v, &programname, &opt_W);
+ opt_complementary = "v::f::"; /* -v and -f can occur multiple times */
+ opt = getopt32(argv, "F:v:f:W:", &opt_F, &list_v, &list_f, &opt_W);
argv += optind;
argc -= optind;
if (opt & 0x1)
setvar_s(intvar[FS], opt_F); // -F
- while (opt_v) { /* -v */
- if (!is_assignment(llist_pop(&opt_v)))
+ while (list_v) { /* -v */
+ if (!is_assignment(llist_pop(&list_v)))
bb_show_usage();
}
- if (opt & 0x4) { // -f
- char *s = s; /* die, gcc, die */
- FILE *from_file = afopen(programname, "r");
- /* 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);
- } else {
+ if (list_f) { /* -f */
+ do {
+ char *s = NULL;
+ FILE *from_file;
+
+ g_progname = llist_pop(&list_f);
+ from_file = xfopen_stdin(g_progname);
+ /* one byte is reserved for some trick in next_token */
for (i = j = 1; j > 0; i += j) {
s = xrealloc(s, i + 4096);
j = fread(s + i, 1, 4094, from_file);
}
- }
- s[i] = '\0';
- fclose(from_file);
- parse_program(s + 1);
- free(s);
+ s[i] = '\0';
+ fclose(from_file);
+ parse_program(s + 1);
+ free(s);
+ } while (list_f);
+ argc++;
} else { // no -f: take program from 1st parameter
if (!argc)
bb_show_usage();
- programname = "cmd. line";
+ 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 */
- setvar_i(intvar[ARGC], argc + 1);
+ setvar_i(intvar[ARGC], argc);
setari_u(intvar[ARGV], 0, "awk");
i = 0;
while (*argv)
}
if (i < 0)
- runtime_error(strerror(errno));
+ syntax_error(strerror(errno));
iF = next_input_file();
}
projects / oweals / busybox.git / blobdiff