#include "libbb.h"
#include "xregex.h"
#include <math.h>
-extern char **environ;
/* This is a NOEXEC applet. Be very careful! */
+/* If you comment out one of these below, it will be #defined later
+ * to perform debug printfs to stderr: */
+#define debug_printf_walker(...) do {} while (0)
+
+#ifndef debug_printf_walker
+# define debug_printf_walker(...) (fprintf(stderr, __VA_ARGS__))
+#endif
+
+
+
#define MAXVARFMT 240
#define MINNVBLOCK 64
/* 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)
+typedef struct walker_list {
+ char *end;
+ char *cur;
+ struct walker_list *prev;
+ char wbuf[1];
+} walker_list;
+
/* Variable */
typedef struct var_s {
unsigned type; /* flags */
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) */
+ walker_list *walker; /* list of array elements (for..in) */
} x;
} var;
union {
struct node_s *n;
var *v;
- int i;
- char *s;
+ int aidx;
+ char *new_progname;
regex_t *re;
} l;
union {
struct node_s *n;
regex_t *ire;
func *f;
- int argno;
} r;
union {
struct node_s *n;
var *pos;
struct nvblock_s *prev;
struct nvblock_s *next;
- var nv[0];
+ var nv[];
} nvblock;
typedef struct tsplitter_s {
/* builtins */
enum {
- B_a2, B_ix, B_ma, B_sp, B_ss, B_ti, B_lo, B_up,
+ B_a2, B_ix, B_ma, B_sp, B_ss, B_ti, B_mt, B_lo, B_up,
B_ge, B_gs, B_su,
B_an, B_co, B_ls, B_or, B_rs, B_xo,
};
"\4rand" "\3sin" "\4sqrt" "\5srand"
"\6gensub" "\4gsub" "\5index" "\6length"
"\5match" "\5split" "\7sprintf" "\3sub"
- "\6substr" "\7systime" "\10strftime"
+ "\6substr" "\7systime" "\10strftime" "\6mktime"
"\7tolower" "\7toupper" NTC
"\7getline" NTC
"\4func" "\10function" NTC
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_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b), OC_B|B_mt|P(0x0b),
OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49),
OC_GETLINE|SV|P(0),
0, 0,
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[] 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[] 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
/* Globals. Split in two parts so that first one is addressed
- * with (mostly short) negative offsets */
+ * 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 {
- chain beginseq, mainseq, endseq, *seq;
+ double t_double;
+ chain beginseq, mainseq, endseq;
+ chain *seq;
node *break_ptr, *continue_ptr;
rstream *iF;
xhash *vhash, *ahash, *fdhash, *fnhash;
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)
+#define G (*(struct globals2 *)ptr_to_globals)
/* For debug. nm --size-sort awk.o | grep -vi ' [tr] ' */
-/* char G1size[sizeof(G1)]; - 0x6c */
-/* char Gsize[sizeof(G)]; - 0x1cc */
+/*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 */
+/*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 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); \
+ SET_PTR_TO_GLOBALS((char*)xzalloc(sizeof(G1)+sizeof(G)) + sizeof(G1)); \
G.next_token__ltclass = TC_OPTERM; \
G.evaluate__seed = 1; \
} while (0)
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_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_MATH
+#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)
+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", g_progname, g_lineno, message);
}
{
unsigned idx = 0;
- while (*name) idx = *name++ + (idx << 6) - idx;
+ while (*name)
+ idx = *name++ + (idx << 6) - idx;
return idx;
}
{
xhash *newhash;
- newhash = xzalloc(sizeof(xhash));
+ newhash = xzalloc(sizeof(*newhash));
newhash->csize = FIRST_PRIME;
- newhash->items = xzalloc(newhash->csize * sizeof(hash_item *));
+ newhash->items = xzalloc(FIRST_PRIME * sizeof(newhash->items[0]));
return newhash;
}
{
hash_item *hi;
- hi = hash->items [ hashidx(name) % hash->csize ];
+ hi = hash->items[hashidx(name) % hash->csize];
while (hi) {
if (strcmp(hi->name, name) == 0)
- return &(hi->data);
+ return &hi->data;
hi = hi->next;
}
return NULL;
return;
newsize = PRIMES[hash->nprime++];
- newitems = xzalloc(newsize * sizeof(hash_item *));
+ newitems = xzalloc(newsize * sizeof(newitems[0]));
for (i = 0; i < hash->csize; i++) {
hi = hash->items[i];
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];
hash->items[idx] = hi;
hash->glen += l;
}
- return &(hi->data);
+ return &hi->data;
}
#define findvar(hash, name) ((var*) hash_find((hash), (name)))
{
hash_item *hi, **phi;
- phi = &(hash->items[hashidx(name) % hash->csize]);
+ phi = &hash->items[hashidx(name) % hash->csize];
while (*phi) {
hi = *phi;
if (strcmp(hi->name, name) == 0) {
free(hi);
break;
}
- phi = &(hi->next);
+ phi = &hi->next;
}
}
/* ------ some useful functions ------ */
-static void skip_spaces(char **s)
+static char *skip_spaces(char *p)
{
- char *p = *s;
-
while (1) {
if (*p == '\\' && p[1] == '\n') {
p++;
}
p++;
}
- *s = p;
+ return p;
}
+/* returns old *s, advances *s past word and terminating NUL */
static char *nextword(char **s)
{
char *p = *s;
-
- while (*(*s)++) /* */;
-
+ while (*(*s)++ != '\0')
+ continue;
return p;
}
{
char c, *pps;
- c = *((*s)++);
+ c = *(*s)++;
pps = *s;
- if (c == '\\') c = bb_process_escape_sequence((const char**)s);
- if (c == '\\' && *s == pps) c = *((*s)++);
+ 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)
+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) -------- */
return setvar_p(v, (value && *value) ? xstrdup(value) : NULL);
}
-/* same as setvar_s but set USER flag */
+/* same as setvar_s but sets USER flag */
static var *setvar_u(var *v, const char *value)
{
- setvar_s(v, value);
+ v = 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)
{
- char sidx[sizeof(int)*3 + 1];
var *v;
- sprintf(sidx, "%d", idx);
- v = findvar(iamarray(a), sidx);
+ v = findvar(iamarray(a), itoa(idx));
setvar_u(v, s);
}
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);
+ s = skip_spaces(s);
if (*s != '\0')
v->type &= ~VF_USER;
}
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) + 1.0);
}
/* return true if v is number or numeric string */
static int istrue(var *v)
{
if (is_numeric(v))
- return (v->number == 0) ? 0 : 1;
- return (v->string && *(v->string)) ? 1 : 0;
+ return (v->number != 0);
+ return (v->string && v->string[0]);
}
/* temporary variables allocator. Last allocated should be first freed */
while (g_cb) {
pb = g_cb;
- if ((g_cb->pos - g_cb->nv) + n <= g_cb->size) break;
+ 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 = xzalloc(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;
+ /*g_cb->next = NULL; - xzalloc did it */
+ if (pb)
+ pb->next = g_cb;
}
v = r = g_cb->pos;
free(p->x.array->items);
free(p->x.array);
}
- if (p->type & VF_WALK)
- free(p->x.walker);
-
+ if (p->type & VF_WALK) {
+ walker_list *n;
+ walker_list *w = p->x.walker;
+ debug_printf_walker("nvfree: freeing walker @%p\n", &p->x.walker);
+ p->x.walker = NULL;
+ while (w) {
+ n = w->prev;
+ debug_printf_walker(" free(%p)\n", w);
+ free(w);
+ w = n;
+ }
+ }
clrvar(p);
}
/* Initialized to TC_OPTERM: */
#define ltclass (G.next_token__ltclass)
- char *p, *pp, *s;
+ char *p, *s;
const char *tl;
uint32_t tc;
const uint32_t *ti;
} else {
p = g_pos;
readnext:
- skip_spaces(&p);
+ p = skip_spaces(p);
g_lineno = t_lineno;
if (*p == '#')
while (*p != '\n' && *p != '\0')
/* it's a string */
t_string = s = ++p;
while (*p != '\"') {
+ char *pp = p;
if (*p == '\0' || *p == '\n')
syntax_error(EMSG_UNEXP_EOS);
- *(s++) = nextchar(&p);
+ *s++ = nextchar(&pp);
+ p = pp;
}
p++;
*s = '\0';
syntax_error(EMSG_UNEXP_EOS);
*s = *p++;
if (*s++ == '\\') {
- pp = p;
- *(s-1) = bb_process_escape_sequence((const char **)&p);
- if (*pp == '\\')
+ char *pp = p;
+ s[-1] = bb_process_escape_sequence((const char **)&pp);
+ if (*p == '\\')
*s++ = '\\';
- if (p == pp)
+ if (pp == p)
*s++ = *p++;
+ else
+ p = pp;
}
}
p++;
} else if (*p == '.' || isdigit(*p)) {
/* it's a number */
- t_double = strtod(p, &p);
- if (*p == '.')
+ char *pp = p;
+ t_double = my_strtod(&pp);
+ p = pp;
+ if (*pp == '.')
syntax_error(EMSG_UNEXP_TOKEN);
tc = TC_NUMBER;
tc = 0x00000001;
ti = tokeninfo;
while (*tl) {
- l = *(tl++);
+ l = *tl++;
if (l == NTCC) {
tc <<= 1;
continue;
syntax_error(EMSG_UNEXP_TOKEN);
t_string = --p;
- while (isalnum_(*(++p))) {
- *(p-1) = *p;
+ while (isalnum_(*++p)) {
+ p[-1] = *p;
}
- *(p-1) = '\0';
+ p[-1] = '\0';
tc = TC_VARIABLE;
/* also consume whitespace between functionname and bracket */
- if (!(expected & TC_VARIABLE))
- skip_spaces(&p);
+ if (!(expected & TC_VARIABLE) || (expected & TC_ARRAY))
+ p = skip_spaces(p);
if (*p == '(') {
tc = TC_FUNCTION;
} else {
return n;
}
-static node *mk_re_node(const char *s, node *n, regex_t *re)
+static void mk_re_node(const char *s, node *n, regex_t *re)
{
n->info = OC_REGEXP;
n->l.re = re;
n->r.ire = re + 1;
xregcomp(re, s, REG_EXTENDED);
xregcomp(re + 1, s, REG_EXTENDED | REG_ICASE);
-
- return n;
}
static node *condition(void)
/* 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)) )
+ 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);
v = hash_search(ahash, t_string);
if (v != NULL) {
cn->info = OC_FNARG;
- cn->l.i = v->x.aidx;
+ cn->l.aidx = v->x.aidx;
} else {
cn->l.v = newvar(t_string);
}
if (seq->programname != g_progname) {
seq->programname = g_progname;
n = chain_node(OC_NEWSOURCE);
- n->l.s = xstrdup(g_progname);
+ n->l.new_progname = xstrdup(g_progname);
}
n = seq->last;
if (c & TC_GRPSTART) {
while (next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) {
- if (t_tclass & TC_NEWLINE) continue;
+ if (t_tclass & TC_NEWLINE)
+ continue;
rollback_token();
chain_group();
}
f->nargs = 0;
while (next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) {
v = findvar(ahash, t_string);
- v->x.aidx = (f->nargs)++;
+ v->x.aidx = f->nargs++;
if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM)
break;
}
- seq = &(f->body);
+ seq = &f->body;
chain_group();
clear_array(ahash);
*/
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)
+/* gradually increasing buffer.
+ * note that we reallocate even if n == old_size,
+ * and thus there is at least one extra allocated byte.
+ */
+static char* 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);
+ }
+ return b;
}
/* resize field storage space */
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;
} while (*s);
c[0] = toupper(c[0]);
c[1] = tolower(c[1]);
}
- if (*s1) n++;
+ if (*s1)
+ n++;
while ((s1 = strpbrk(s1, c))) {
*s1++ = '\0';
n++;
/* space split */
while (*s) {
s = skip_whitespace(s);
- if (!*s) break;
+ if (!*s)
+ break;
n++;
while (*s && !isspace(*s))
*s1++ = *s++;
memcpy(b+len, sep, sl);
len += sl;
}
- qrealloc(&b, len+l+sl, &bsize);
+ b = qrealloc(b, len+l+sl, &bsize);
memcpy(b+len, s, l);
len += l;
}
static void hashwalk_init(var *v, xhash *array)
{
- char **w;
hash_item *hi;
- int i;
+ unsigned i;
+ walker_list *w;
+ walker_list *prev_walker;
- if (v->type & VF_WALK)
- free(v->x.walker);
+ if (v->type & VF_WALK) {
+ prev_walker = v->x.walker;
+ } else {
+ v->type |= VF_WALK;
+ prev_walker = NULL;
+ }
+ debug_printf_walker("hashwalk_init: prev_walker:%p\n", prev_walker);
- v->type |= VF_WALK;
- w = v->x.walker = xzalloc(2 + 2*sizeof(char *) + array->glen);
- w[0] = w[1] = (char *)(w + 2);
+ w = v->x.walker = xzalloc(sizeof(*w) + array->glen + 1); /* why + 1? */
+ debug_printf_walker(" walker@%p=%p\n", &v->x.walker, w);
+ w->cur = w->end = w->wbuf;
+ w->prev = prev_walker;
for (i = 0; i < array->csize; i++) {
hi = array->items[i];
while (hi) {
- strcpy(*w, hi->name);
- nextword(w);
+ strcpy(w->end, hi->name);
+ nextword(&w->end);
hi = hi->next;
}
}
static int hashwalk_next(var *v)
{
- char **w;
+ walker_list *w = v->x.walker;
+
+ if (w->cur >= w->end) {
+ walker_list *prev_walker = w->prev;
- w = v->x.walker;
- if (w[1] == w[0])
+ debug_printf_walker("end of iteration, free(walker@%p:%p), prev_walker:%p\n", &v->x.walker, w, prev_walker);
+ free(w);
+ v->x.walker = prev_walker;
return FALSE;
+ }
- setvar_s(v, nextword(w+1));
+ setvar_s(v, nextword(&w->cur));
return TRUE;
}
{
char *b;
regmatch_t pmatch[2];
- int a, p, pp=0, size;
+ int size, a, p, pp = 0;
int fd, so, eo, r, rp;
char c, *m, *s;
c = (char) rsplitter.n.info;
rp = 0;
- if (!m) qrealloc(&m, 256, &size);
+ if (!m)
+ 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++;
s = strstr(b+rp, "\n\n");
if (s) {
so = eo = s-b;
- while (b[eo] == '\n') eo++;
+ while (b[eo] == '\n')
+ eo++;
if (b[eo] != '\0')
break;
}
}
if (a > 0) {
- memmove(m, (const void *)(m+a), p+1);
+ memmove(m, m+a, p+1);
b = m;
a = 0;
}
- qrealloc(&m, a+p+128, &size);
+ m = qrealloc(m, a+p+128, &size);
b = m + a;
pp = p;
p += safe_read(fd, b+p, size-p-1);
return r;
}
-
/* formatted output into an allocated buffer, return ptr to buffer */
static char *awk_printf(node *n)
{
i = 0;
while (*f) {
s = f;
- while (*f && (*f != '%' || *(++f) == '%'))
+ while (*f && (*f != '%' || *++f == '%'))
f++;
while (*f && !isalpha(*f)) {
if (*f == '*')
}
incr = (f - s) + MAXVARFMT;
- qrealloc(&b, incr + i, &bsize);
+ b = qrealloc(b, incr + i, &bsize);
c = *f;
- if (c != '\0') f++;
+ if (c != '\0')
+ f++;
c1 = *f;
*f = '\0';
arg = evaluate(nextarg(&n), v);
(char)getvar_i(arg) : *getvar_s(arg));
} else if (c == 's') {
s1 = getvar_s(arg);
- qrealloc(&b, incr+i+strlen(s1), &bsize);
+ b = 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;
+ if (i < j)
+ i = j;
}
- b = xrealloc(b, i + 1);
free(fmt);
nvfree(v);
+ b = xrealloc(b, i + 1);
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)
+/* Common substitution routine.
+ * Replace (nm)'th substring of (src) that matches (rn) 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 subexp != 0, enable subexpression matching (\1-\9).
*/
-static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int ex)
+static int awk_sub(node *rn, const char *repl, int nm, var *src, var *dest, int subexp)
{
- char *ds = NULL;
- const char *s;
+ char *resbuf;
const char *sp;
- int c, i, j, di, rl, so, eo, nbs, n, dssize;
+ int match_no, residx, replen, resbufsize;
+ int regexec_flags;
regmatch_t pmatch[10];
- regex_t sreg, *re;
+ regex_t sreg, *regex;
+
+ resbuf = NULL;
+ residx = 0;
+ match_no = 0;
+ regexec_flags = 0;
+ regex = as_regex(rn, &sreg);
+ sp = getvar_s(src ? src : intvar[F0]);
+ replen = strlen(repl);
+ while (regexec(regex, sp, 10, pmatch, regexec_flags) == 0) {
+ int so = pmatch[0].rm_so;
+ int eo = pmatch[0].rm_eo;
+
+ //bb_error_msg("match %u: [%u,%u] '%s'%p", match_no+1, so, eo, sp,sp);
+ resbuf = qrealloc(resbuf, residx + eo + replen, &resbufsize);
+ memcpy(resbuf + residx, sp, eo);
+ residx += eo;
+ if (++match_no >= nm) {
+ const char *s;
+ int nbs;
- re = as_regex(rn, &sreg);
- if (!src) src = intvar[F0];
- if (!dest) dest = intvar[F0];
-
- i = di = 0;
- sp = getvar_s(src);
- rl = strlen(repl);
- 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);
+ residx -= (eo - so);
nbs = 0;
for (s = repl; *s; s++) {
- ds[di++] = c = *s;
+ char c = resbuf[residx++] = *s;
if (c == '\\') {
nbs++;
continue;
}
- if (c == '&' || (ex && c >= '0' && c <= '9')) {
- di -= ((nbs + 3) >> 1);
+ if (c == '&' || (subexp && c >= '0' && c <= '9')) {
+ int j;
+ residx -= ((nbs + 3) >> 1);
j = 0;
if (c != '&') {
j = c - '0';
nbs++;
}
if (nbs % 2) {
- ds[di++] = c;
+ resbuf[residx++] = 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;
+ int n = pmatch[j].rm_eo - pmatch[j].rm_so;
+ resbuf = qrealloc(resbuf, residx + replen + n, &resbufsize);
+ memcpy(resbuf + residx, sp + pmatch[j].rm_so, n);
+ residx += n;
}
}
nbs = 0;
}
}
+ regexec_flags = REG_NOTBOL;
sp += eo;
- if (i == nm) break;
+ if (match_no == nm)
+ break;
if (eo == so) {
- ds[di] = *sp++;
- if (!ds[di++]) break;
+ /* Empty match (e.g. "b*" will match anywhere).
+ * Advance by one char. */
+//BUG (bug 1333):
+//gsub(/\<b*/,"") on "abc" will reach this point, advance to "bc"
+//... and will erroneously match "b" even though it is NOT at the word start.
+//we need REG_NOTBOW but it does not exist...
+//TODO: if EXTRA_COMPAT=y, use GNU matching and re_search,
+//it should be able to do it correctly.
+ /* Subtle: this is safe only because
+ * qrealloc allocated at least one extra byte */
+ resbuf[residx] = *sp;
+ if (*sp == '\0')
+ goto ret;
+ sp++;
+ residx++;
}
}
- qrealloc(&ds, di + strlen(sp), &dssize);
- strcpy(ds + di, sp);
- setvar_p(dest, ds);
- if (re == &sreg) regfree(re);
- return i;
+ resbuf = qrealloc(resbuf, residx + strlen(sp), &resbufsize);
+ strcpy(resbuf + residx, sp);
+ ret:
+ //bb_error_msg("end sp:'%s'%p", sp,sp);
+ setvar_p(dest ? dest : intvar[F0], resbuf);
+ if (regex == &sreg)
+ regfree(regex);
+ return match_no;
}
-static var *exec_builtin(node *op, var *res)
+static NOINLINE int do_mktime(const char *ds)
+{
+ struct tm then;
+ int count;
+
+ /*memset(&then, 0, sizeof(then)); - not needed */
+ then.tm_isdst = -1; /* default is unknown */
+
+ /* manpage of mktime says these fields are ints,
+ * so we can sscanf stuff directly into them */
+ count = sscanf(ds, "%u %u %u %u %u %u %d",
+ &then.tm_year, &then.tm_mon, &then.tm_mday,
+ &then.tm_hour, &then.tm_min, &then.tm_sec,
+ &then.tm_isdst);
+
+ if (count < 6
+ || (unsigned)then.tm_mon < 1
+ || (unsigned)then.tm_year < 1900
+ ) {
+ return -1;
+ }
+
+ then.tm_mon -= 1;
+ then.tm_year -= 1900;
+
+ return mktime(&then);
+}
+
+static NOINLINE var *exec_builtin(node *op, var *res)
{
#define tspl (G.exec_builtin__tspl)
- int (*to_xxx)(int);
var *tv;
node *an[4];
var *av[4];
uint32_t isr, info;
int nargs;
time_t tt;
- char *s, *s1;
int i, l, ll, n;
tv = nvalloc(4);
av[2] = av[3] = NULL;
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]);
+ 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))
+ if ((uint32_t)nargs < (info >> 30))
syntax_error(EMSG_TOO_FEW_ARGS);
- switch (info & OPNMASK) {
+ info &= OPNMASK;
+ switch (info) {
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
syntax_error(EMSG_NO_MATH);
#endif
break;
- case B_sp:
+ case B_sp: {
+ char *s, *s1;
+
if (nargs > 2) {
spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ?
an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl);
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);
+ for (i = 1; i <= n; i++)
+ setari_u(av[1], i, nextword(&s));
+ free(s1);
setvar_i(res, n);
break;
+ }
+
+ case B_ss: {
+ char *s;
- case B_ss:
l = strlen(as[0]);
i = getvar_i(av[1]) - 1;
- if (i > l) i = l;
- if (i < 0) i = 0;
+ if (i > l)
+ i = l;
+ 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';
+ if (n < 0)
+ 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:
- to_xxx = tolower;
- goto lo_cont;
-
- case B_up:
- to_xxx = toupper;
- lo_cont:
+ case B_up: {
+ char *s, *s1;
s1 = s = xstrdup(as[0]);
while (*s1) {
- *s1 = (*to_xxx)(*s1);
+ //*s1 = (info == B_up) ? toupper(*s1) : tolower(*s1);
+ if ((unsigned char)((*s1 | 0x20) - 'a') <= ('z' - 'a'))
+ *s1 = (info == B_up) ? (*s1 & 0xdf) : (*s1 | 0x20);
s1++;
}
setvar_p(res, s);
break;
+ }
case B_ix:
n = 0;
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;
+ char *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++) {
+ for (i = 0; i <= l; i++) {
if (strncasecmp(as[0]+i, as[1], ll) == 0) {
n = i+1;
break;
setvar_s(res, g_buf);
break;
+ case B_mt:
+ setvar_i(res, do_mktime(as[0]));
+ break;
+
case B_ma:
re = as_regex(an[1], &sreg);
n = regexec(re, as[0], 1, pmatch, 0);
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);
+ if (re == &sreg)
+ regfree(re);
break;
case B_ge:
#define seed (G.evaluate__seed)
#define sreg (G.evaluate__sreg)
- 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;
if (!op)
return setvar_s(res, NULL);
v1 = nvalloc(2);
while (op) {
+ struct {
+ var *v;
+ const char *s;
+ } L = L; /* for compiler */
+ struct {
+ var *v;
+ const char *s;
+ } R = R;
+ double L_d = L_d;
+ uint32_t opinfo;
+ int opn;
+ node *op1;
+
opinfo = op->info;
opn = (opinfo & OPNMASK);
g_lineno = op->lineno;
+ op1 = op->l.n;
/* 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);
+ if (opinfo & OF_RES1)
+ 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)) {
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;
+ op = ptest(op1) ? op->a.n : op->r.n;
}
break;
break;
case XC( OC_PRINT ):
- case XC( OC_PRINTF ):
- X.F = stdout;
+ case XC( OC_PRINTF ): {
+ FILE *F = stdout;
+
if (op->r.n) {
- X.rsm = newfile(R.s);
- if (!X.rsm->F) {
+ rstream *rsm = newfile(R.s);
+ if (!rsm->F) {
if (opn == '|') {
- X.rsm->F = popen(R.s, "w");
- if (X.rsm->F == NULL)
+ rsm->F = popen(R.s, "w");
+ if (rsm->F == NULL)
bb_perror_msg_and_die("popen");
- X.rsm->is_pipe = 1;
+ rsm->is_pipe = 1;
} else {
- X.rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
+ rsm->F = xfopen(R.s, opn=='w' ? "w" : "a");
}
}
- X.F = X.rsm->F;
+ F = rsm->F;
}
if ((opinfo & OPCLSMASK) == OC_PRINT) {
if (!op1) {
- fputs(getvar_s(intvar[F0]), X.F);
+ fputs(getvar_s(intvar[F0]), F);
} else {
while (op1) {
- L.v = evaluate(nextarg(&op1), v1);
- if (L.v->type & VF_NUMBER) {
+ var *v = evaluate(nextarg(&op1), v1);
+ if (v->type & VF_NUMBER) {
fmt_num(g_buf, MAXVARFMT, getvar_s(intvar[OFMT]),
- getvar_i(L.v), TRUE);
- fputs(g_buf, X.F);
+ getvar_i(v), TRUE);
+ fputs(g_buf, F);
} else {
- fputs(getvar_s(L.v), X.F);
+ fputs(getvar_s(v), F);
}
- if (op1) fputs(getvar_s(intvar[OFS]), X.F);
+ if (op1)
+ fputs(getvar_s(intvar[OFS]), F);
}
}
- fputs(getvar_s(intvar[ORS]), X.F);
+ fputs(getvar_s(intvar[ORS]), F);
} else { /* OC_PRINTF */
- L.s = awk_printf(op1);
- fputs(L.s, X.F);
- free((char*)L.s);
+ char *s = awk_printf(op1);
+ fputs(s, F);
+ free(s);
}
- fflush(X.F);
+ fflush(F);
break;
+ }
- case XC( OC_DELETE ):
- 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];
+ case XC( OC_DELETE ): {
+ uint32_t info = op1->info & OPCLSMASK;
+ var *v;
+
+ if (info == OC_VAR) {
+ v = op1->l.v;
+ } else if (info == OC_FNARG) {
+ v = &fnargs[op1->l.aidx];
} else {
syntax_error(EMSG_NOT_ARRAY);
}
if (op1->r.n) {
+ const char *s;
clrvar(L.v);
- L.s = getvar_s(evaluate(op1->r.n, v1));
- hash_remove(iamarray(R.v), L.s);
+ s = getvar_s(evaluate(op1->r.n, v1));
+ hash_remove(iamarray(v), s);
} else {
- clear_array(iamarray(R.v));
+ clear_array(iamarray(v));
}
break;
+ }
case XC( OC_NEWSOURCE ):
- g_progname = op->l.s;
+ g_progname = op->l.new_progname;
break;
case XC( OC_RETURN ):
break;
case XC( OC_EXIT ):
- awk_exit(L.d);
+ awk_exit(L_d);
/* -- recursive node type -- */
goto v_cont;
case XC( OC_FNARG ):
- L.v = &fnargs[op->l.i];
+ L.v = &fnargs[op->l.aidx];
v_cont:
res = op->r.n ? findvar(iamarray(L.v), R.s) : L.v;
break;
case XC( OC_MATCH ):
op1 = op->r.n;
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));
+ {
+ regex_t *re = as_regex(op1, &sreg);
+ int i = regexec(re, L.s, 0, NULL, 0);
+ if (re == &sreg)
+ regfree(re);
+ setvar_i(res, (i == 0) ^ (opn == '!'));
+ }
break;
case XC( OC_MOVE ):
/* 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 {
+//Disabled: if R.v is numeric but happens to have cached R.v->string,
+//then L.v ends up being a string, which is wrong
+// if (R.v == v1+1 && R.v->string) {
+// res = setvar_p(L.v, R.v->string);
+// R.v->string = NULL;
+// } else {
res = copyvar(L.v, R.v);
- }
+// }
break;
case XC( OC_TERNARY ):
res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res);
break;
- case XC( OC_FUNC ):
+ case XC( OC_FUNC ): {
+ var *vbeg, *v;
+ const char *sv_progname;
+
if (!op->r.f->body.first)
syntax_error(EMSG_UNDEF_FUNC);
- X.v = R.v = nvalloc(op->r.f->nargs+1);
+ vbeg = 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)
+ var *arg = evaluate(nextarg(&op1), v1);
+ copyvar(v, arg);
+ v->type |= VF_CHILD;
+ v->x.parent = arg;
+ if (++v - vbeg >= op->r.f->nargs)
break;
}
- R.v = fnargs;
- fnargs = X.v;
+ v = fnargs;
+ fnargs = vbeg;
+ sv_progname = g_progname;
- L.s = g_progname;
res = evaluate(op->r.f->body.first, res);
- g_progname = L.s;
+ g_progname = sv_progname;
nvfree(fnargs);
- fnargs = R.v;
+ fnargs = v;
+
break;
+ }
case XC( OC_GETLINE ):
- case XC( OC_PGETLINE ):
+ case XC( OC_PGETLINE ): {
+ rstream *rsm;
+ int i;
+
if (op1) {
- X.rsm = newfile(L.s);
- if (!X.rsm->F) {
+ rsm = newfile(L.s);
+ if (!rsm->F) {
if ((opinfo & OPCLSMASK) == OC_PGETLINE) {
- X.rsm->F = popen(L.s, "r");
- X.rsm->is_pipe = TRUE;
+ rsm->F = popen(L.s, "r");
+ rsm->is_pipe = TRUE;
} else {
- X.rsm->F = fopen(L.s, "r"); /* not xfopen! */
+ rsm->F = fopen_for_read(L.s); /* not xfopen! */
}
}
} else {
- if (!iF) iF = next_input_file();
- X.rsm = iF;
+ if (!iF)
+ iF = next_input_file();
+ rsm = iF;
}
- if (!X.rsm->F) {
+ if (!rsm->F) {
setvar_i(intvar[ERRNO], errno);
setvar_i(res, -1);
break;
if (!op->r.n)
R.v = intvar[F0];
- L.i = awk_getline(X.rsm, R.v);
- if (L.i > 0) {
- if (!op1) {
- incvar(intvar[FNR]);
- incvar(intvar[NR]);
- }
+ i = awk_getline(rsm, R.v);
+ if (i > 0 && !op1) {
+ incvar(intvar[FNR]);
+ incvar(intvar[NR]);
}
- setvar_i(res, L.i);
+ setvar_i(res, i);
break;
+ }
/* simple builtins */
- case XC( OC_FBLTIN ):
- switch (opn) {
+ case XC( OC_FBLTIN ): {
+ int i;
+ rstream *rsm;
+ double R_d = R_d; /* for compiler */
+ switch (opn) {
case F_in:
- R.d = (int)L.d;
+ R_d = (int)L_d;
break;
case F_rn:
- R.d = (double)rand() / (double)RAND_MAX;
+ 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);
+ R_d = cos(L_d);
break;
case F_ex:
- R.d = exp(L.d);
+ R_d = exp(L_d);
break;
case F_lg:
- R.d = log(L.d);
+ R_d = log(L_d);
break;
case F_si:
- R.d = sin(L.d);
+ R_d = sin(L_d);
break;
case F_sq:
- R.d = sqrt(L.d);
+ R_d = sqrt(L_d);
break;
#else
case F_co:
break;
#endif
case F_sr:
- R.d = (double)seed;
- seed = op1 ? (unsigned)L.d : (unsigned)time(NULL);
+ R_d = (double)seed;
+ seed = op1 ? (unsigned)L_d : (unsigned)time(NULL);
srand(seed);
break;
case F_ti:
- R.d = time(NULL);
+ R_d = time(NULL);
break;
case F_le:
if (!op1)
L.s = getvar_s(intvar[F0]);
- R.d = strlen(L.s);
+ R_d = strlen(L.s);
break;
case F_sy:
- fflush(NULL);
- R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
+ fflush_all();
+ R_d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
? (system(L.s) >> 8) : 0;
break;
case F_ff:
- if (!op1)
+ if (!op1) {
fflush(stdout);
- else {
- if (L.s && *L.s) {
- X.rsm = newfile(L.s);
- fflush(X.rsm->F);
- } else {
- fflush(NULL);
- }
+ } else if (L.s && *L.s) {
+ rsm = newfile(L.s);
+ fflush(rsm->F);
+ } else {
+ fflush_all();
}
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);
- free(X.rsm->buffer);
+ i = 0;
+ rsm = (rstream *)hash_search(fdhash, L.s);
+ if (rsm) {
+ i = rsm->is_pipe ? pclose(rsm->F) : fclose(rsm->F);
+ free(rsm->buffer);
hash_remove(fdhash, L.s);
}
- if (R.i != 0)
+ if (i != 0)
setvar_i(intvar[ERRNO], errno);
- R.d = (double)R.i;
+ R_d = (double)i;
break;
}
- setvar_i(res, R.d);
+ setvar_i(res, R_d);
break;
+ }
case XC( OC_BUILTIN ):
res = exec_builtin(op, res);
setvar_p(res, awk_printf(op1));
break;
- case XC( OC_UNARY ):
- X.v = R.v;
- L.d = R.d = getvar_i(R.v);
+ case XC( OC_UNARY ): {
+ double Ld, R_d;
+
+ Ld = R_d = getvar_i(R.v);
switch (opn) {
case 'P':
- L.d = ++R.d;
+ Ld = ++R_d;
goto r_op_change;
case 'p':
- R.d++;
+ R_d++;
goto r_op_change;
case 'M':
- L.d = --R.d;
+ Ld = --R_d;
goto r_op_change;
case 'm':
- R.d--;
- goto r_op_change;
+ R_d--;
+ r_op_change:
+ setvar_i(R.v, R_d);
+ break;
case '!':
- L.d = istrue(X.v) ? 0 : 1;
+ Ld = !istrue(R.v);
break;
case '-':
- L.d = -R.d;
+ Ld = -R_d;
break;
- r_op_change:
- setvar_i(X.v, R.d);
}
- setvar_i(res, L.d);
+ setvar_i(res, Ld);
break;
+ }
- case XC( OC_FIELD ):
- R.i = (int)getvar_i(R.v);
- if (R.i == 0) {
+ case XC( OC_FIELD ): {
+ int i = (int)getvar_i(R.v);
+ if (i == 0) {
res = intvar[F0];
} else {
split_f0();
- if (R.i > nfields)
- fsrealloc(R.i);
- res = &Fields[R.i - 1];
+ if (i > nfields)
+ fsrealloc(i);
+ res = &Fields[i - 1];
}
break;
+ }
/* concatenation (" ") and index joining (",") */
case XC( OC_CONCAT ):
- case XC( OC_COMMA ):
- opn = strlen(L.s) + strlen(R.s) + 2;
- X.s = xmalloc(opn);
- strcpy(X.s, L.s);
- if ((opinfo & OPCLSMASK) == OC_COMMA) {
- 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);
+ case XC( OC_COMMA ): {
+ const char *sep = "";
+ if ((opinfo & OPCLSMASK) == OC_COMMA)
+ sep = getvar_s(intvar[SUBSEP]);
+ setvar_p(res, xasprintf("%s%s%s", L.s, sep, R.s));
break;
+ }
case XC( OC_LAND ):
setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0);
break;
case XC( OC_BINARY ):
- case XC( OC_REPLACE ):
- R.d = getvar_i(R.v);
+ case XC( OC_REPLACE ): {
+ double R_d = getvar_i(R.v);
switch (opn) {
case '+':
- L.d += R.d;
+ L_d += R_d;
break;
case '-':
- L.d -= R.d;
+ L_d -= R_d;
break;
case '*':
- L.d *= R.d;
+ L_d *= R_d;
break;
case '/':
- if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
- L.d /= R.d;
+ if (R_d == 0)
+ syntax_error(EMSG_DIV_BY_ZERO);
+ L_d /= R_d;
break;
case '&':
-#if ENABLE_FEATURE_AWK_MATH
- L.d = pow(L.d, R.d);
+#if ENABLE_FEATURE_AWK_LIBM
+ 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);
- L.d -= (int)(L.d / R.d) * R.d;
+ if (R_d == 0)
+ syntax_error(EMSG_DIV_BY_ZERO);
+ L_d -= (int)(L_d / R_d) * R_d;
break;
}
- res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : X.v, L.d);
+ res = setvar_i(((opinfo & OPCLSMASK) == OC_BINARY) ? res : L.v, L_d);
break;
+ }
+
+ case XC( OC_COMPARE ): {
+ int i = i; /* for compiler */
+ double Ld;
- case XC( OC_COMPARE ):
if (is_numeric(L.v) && is_numeric(R.v)) {
- L.d = getvar_i(L.v) - getvar_i(R.v);
+ Ld = 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);
+ const char *l = getvar_s(L.v);
+ const char *r = getvar_s(R.v);
+ Ld = icase ? strcasecmp(l, r) : strcmp(l, r);
}
switch (opn & 0xfe) {
case 0:
- R.i = (L.d > 0);
+ i = (Ld > 0);
break;
case 2:
- R.i = (L.d >= 0);
+ i = (Ld >= 0);
break;
case 4:
- R.i = (L.d == 0);
+ i = (Ld == 0);
break;
}
- setvar_i(res, (opn & 0x1 ? R.i : !R.i) ? 1 : 0);
+ setvar_i(res, (i == 0) ^ (opn & 1));
break;
+ }
default:
syntax_error(EMSG_POSSIBLE_ERROR);
break;
if (nextrec)
break;
- }
+ } /* while (op) */
+
nvfree(v1);
return res;
#undef fnargs
return FALSE;
}
- *(s++) = '\0';
+ *s++ = '\0';
s0 = s1 = s;
while (*s)
- *(s1++) = nextchar(&s);
+ *s1++ = nextchar(&s);
*s1 = '\0';
setvar_u(newvar(exprc), s0);
FILE *F = NULL;
const char *fname, *ind;
- if (rsm.F) fclose(rsm.F);
+ if (rsm.F)
+ fclose(rsm.F);
rsm.F = NULL;
rsm.pos = rsm.adv = 0;
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);
#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;
*s1 = '=';
}
}
- opt_complementary = "v::";
- opt = getopt32(argc, argv, "F:v:f:W:", &opt_F, &opt_v, &g_progname, &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(g_progname, "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();
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)
awk_exit(EXIT_SUCCESS);
/* input file could already be opened in BEGIN block */
- if (!iF) iF = next_input_file();
+ if (!iF)
+ iF = next_input_file();
/* passing through input files */
while (iF) {