/[magellan-source]/trunk/mkinitrd-magellan/busybox/editors/awk.c

Diff of /trunk/mkinitrd-magellan/busybox/editors/awk.c

Parent Directory | Revision Log | View Patch Patch

-revision 983 by niro,
Fri Apr 24 18:33:46 2009 UTC
+revision 984 by niro,
Sun May 30 11:32:42 2010 UTC
 Line 114 
 typedef struct nvblock_s {
  	var *pos;
  	struct nvblock_s *prev;
  	struct nvblock_s *next;
- 	var nv[0];
+ 	var nv[];
  } nvblock;
  typedef struct tsplitter_s {
 Line 250 
 enum {
  /* 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,
  };
 Line 299 
 static const char tokenlist[] ALIGN1 =
  	"\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
 Line 353 
 static const uint32_t tokeninfo[] = {
  	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,
 Line 366 
 static const uint32_t tokeninfo[] = {
  enum {
  	CONVFMT,    OFMT,       FS,         OFS,
  	ORS,        RS,         RT,         FILENAME,
- 	SUBSEP,     ARGIND,     ARGC,       ARGV,
+ 	SUBSEP,     F0,         ARGIND,     ARGC,
- 	ERRNO,      FNR,
+ 	ARGV,       ERRNO,      FNR,        NR,
- 	NR,         NF,         IGNORECASE,
+ 	NF,         IGNORECASE,	ENVIRON,    NUM_INTERNAL_VARS
- 	ENVIRON,    F0,         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"
+ 	"SUBSEP\0"  "$\0*"      "ARGIND\0"  "ARGC\0"
- 	"ERRNO\0"   "FNR\0"
+ 	"ARGV\0"    "ERRNO\0"   "FNR\0"     "NR\0"
- 	"NR\0"      "NF\0*"     "IGNORECASE\0*"
+ 	"NF\0*"     "IGNORECASE\0*" "ENVIRON\0" "\0";
- 	"ENVIRON\0" "$\0*"      "\0";
  static const char vValues[] ALIGN1 =
  	"%.6g\0"    "%.6g\0"    " \0"       " \0"
  	"\n\0"      "\n\0"      "\0"        "\0"
- 	"\034\0"
+ 	"\034\0"    "\0"        "\377";
- 	"\377";
  /* hash size may grow to these values */
  #define FIRST_PRIME 61
-Line 489 
 struct globals2 {
+Line 486 
 struct globals2 {
  #define fsplitter    (G.fsplitter   )
  #define rsplitter    (G.rsplitter   )
  #define INIT_G() do { \
- 	SET_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)
-Line 519 
 static const char EMSG_UNDEF_FUNC[] ALIG
+Line 516 
 static const char EMSG_UNDEF_FUNC[] ALIG
  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) NORETURN;
+ static void syntax_error(const char *message) NORETURN;
- static void syntax_error(const char *const message)
+ static void syntax_error(const char *message)
  {
  	bb_error_msg_and_die("%s:%i: %s", g_progname, g_lineno, message);
  }
-Line 536 
 static unsigned hashidx(const char *name
+Line 533 
 static unsigned hashidx(const char *name
  {
  	unsigned idx = 0;
- 	while (*name) idx = *name++ + (idx << 6) - idx;
+ 	while (*name)
+ 		idx = *name++ + (idx << 6) - idx;
  	return idx;
  }
-Line 545 
 static xhash *hash_init(void)
+Line 543 
 static xhash *hash_init(void)
  {
  	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;
  }
-Line 557 
 static void *hash_search(xhash *hash, co
+Line 555 
 static void *hash_search(xhash *hash, co
  {
  	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);
-Line 576 
 static void hash_rebuild(xhash *hash)
+Line 574 
 static void hash_rebuild(xhash *hash)
  		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];
-Line 607 
 static void *hash_find(xhash *hash, cons
+Line 605 
 static void *hash_find(xhash *hash, cons
  			hash_rebuild(hash);
  		l = strlen(name) + 1;
- 		hi = xzalloc(sizeof(hash_item) + l);
+ 		hi = xzalloc(sizeof(*hi) + l);
- 		memcpy(hi->name, name, l);
+ 		strcpy(hi->name, name);
  		idx = hashidx(name) % hash->csize;
  		hi->next = hash->items[idx];
-Line 662 
 static void skip_spaces(char **s)
+Line 660 
 static void skip_spaces(char **s)
  static char *nextword(char **s)
  {
  	char *p = *s;
+ 	while (*(*s)++)
- 	while (*(*s)++) /* */;
+ 		continue;
  	return p;
  }
-Line 674 
 static char nextchar(char **s)
+Line 671 
 static char nextchar(char **s)
  	c = *((*s)++);
  	pps = *s;
- 	if (c == '\\') c = bb_process_escape_sequence((const char**)s);
+ 	if (c == '\\')
- 	if (c == '\\' && *s == pps) c = *((*s)++);
+ 		c = bb_process_escape_sequence((const char**)s);
+ 	if (c == '\\' && *s == pps)
+ 		c = *((*s)++);
  	return c;
  }
-Line 757 
 static var *setvar_s(var *v, const char
+Line 756 
 static var *setvar_s(var *v, const char
  	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;
  }
-Line 768 
 static var *setvar_u(var *v, const char
+Line 767 
 static var *setvar_u(var *v, const char
  /* 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), itoa(idx));
- 	v = findvar(iamarray(a), sidx);
  	setvar_u(v, s);
  }
-Line 847 
 static var *copyvar(var *dest, const var
+Line 844 
 static var *copyvar(var *dest, const var
  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 */
-Line 861 
 static int is_numeric(var *v)
+Line 858 
 static int is_numeric(var *v)
  static int istrue(var *v)
  {
  	if (is_numeric(v))
- 		return (v->number == 0) ? 0 : 1;
+ 		return (v->number != 0);
- 	return (v->string && *(v->string)) ? 1 : 0;
+ 	return (v->string && v->string[0]);
  }
  /* temporary variables allocator. Last allocated should be first freed */
-Line 874 
 static var *nvalloc(int n)
+Line 871 
 static var *nvalloc(int n)
  	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;
  	}
-Line 885 
 static var *nvalloc(int n)
+Line 883 
 static var *nvalloc(int n)
  		g_cb->pos = g_cb->nv;
  		g_cb->prev = pb;
  		/*g_cb->next = NULL; - xzalloc did it */
- 		if (pb) pb->next = g_cb;
+ 		if (pb)
+ 			pb->next = g_cb;
  	}
  	v = r = g_cb->pos;
-Line 1148 
 static node *parse_expr(uint32_t iexp)
+Line 1147 
 static node *parse_expr(uint32_t iexp)
  			/* 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))
+ 			while (((t_info & PRIMASK) > (vn->a.n->info & PRIMASK2))
- 			 || ((t_info == vn->info) && ((t_info & OPCLSMASK) == OC_COLON)) )
+ 			    || ((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);
-Line 1485 
 static node *mk_splitter(const char *s,
+Line 1486 
 static node *mk_splitter(const char *s,
   */
  static regex_t *as_regex(node *op, regex_t *preg)
  {
+ 	int cflags;
  	var *v;
  	const char *s;
-Line 1493 
 static regex_t *as_regex(node *op, regex
+Line 1495 
 static regex_t *as_regex(node *op, regex
  	}
  	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;
  }
-Line 1563 
 static int awk_split(const char *s, node
+Line 1575 
 static int awk_split(const char *s, node
  				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);
-Line 1859 
 static int fmt_num(char *b, int size, co
+Line 1875 
 static int fmt_num(char *b, int size, co
  	return r;
  }
  /* formatted output into an allocated buffer, return ptr to buffer */
  static char *awk_printf(node *n)
  {
-Line 1976 
 static int awk_sub(node *rn, const char
+Line 1991 
 static int awk_sub(node *rn, const char
  		}
  		sp += eo;
- 		if (i == nm) break;
+ 		if (i == nm)
+ 			break;
  		if (eo == so) {
  			ds[di] = *sp++;
- 			if (!ds[di++]) break;
+ 			if (!ds[di++])
+ 				break;
  		}
  	}
  	qrealloc(&ds, di + strlen(sp), &dssize);
  	strcpy(ds + di, sp);
  	setvar_p(dest, ds);
- 	if (re == &sreg) regfree(re);
+ 	if (re == &sreg)
+ 		regfree(re);
  	return i;
  }
- 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];
-Line 2024 
 static var *exec_builtin(node *op, var *
+Line 2069 
 static var *exec_builtin(node *op, var *
  	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_LIBM
-Line 2089 
 static var *exec_builtin(node *op, var *
+Line 2135 
 static var *exec_builtin(node *op, var *
  		break;
  	case B_lo:
- 		to_xxx = tolower;
- 		goto lo_cont;
  	case B_up:
- 		to_xxx = toupper;
-  lo_cont:
  		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);
-Line 2139 
 static var *exec_builtin(node *op, var *
+Line 2182 
 static var *exec_builtin(node *op, var *
  		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);
-Line 2377 
 static var *evaluate(node *op, var *res)
+Line 2424 
 static var *evaluate(node *op, var *res)
  			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));
+ 			setvar_i(res, (R.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) {
+ //Disabled: if R.v is numeric but happens to have cached R.v->string,
- 				res = setvar_p(L.v, R.v->string);
+ //then L.v ends up being a string, which is wrong
- 				R.v->string = NULL;
+ //			if (R.v == v1+1 && R.v->string) {
- 			} else {
+ //				res = setvar_p(L.v, R.v->string);
+ //				R.v->string = NULL;
+ //			} else {
  				res = copyvar(L.v, R.v);
- 			}
+ //			}
  			break;
  		case XC( OC_TERNARY ):
-Line 2400 
 static var *evaluate(node *op, var *res)
+Line 2449 
 static var *evaluate(node *op, var *res)
  			if (!op->r.f->body.first)
  				syntax_error(EMSG_UNDEF_FUNC);
- 			X.v = R.v = nvalloc(op->r.f->nargs+1);
+ 			X.v = R.v = nvalloc(op->r.f->nargs + 1);
  			while (op1) {
  				L.v = evaluate(nextarg(&op1), v1);
  				copyvar(R.v, L.v);
-Line 2514 
 static var *evaluate(node *op, var *res)
+Line 2563 
 static var *evaluate(node *op, var *res)
  				break;
  			case F_sy:
- 				fflush(NULL);
+ 				fflush_all();
  				R.d = (ENABLE_FEATURE_ALLOW_EXEC && L.s && *L.s)
  						? (system(L.s) >> 8) : 0;
  				break;
-Line 2527 
 static var *evaluate(node *op, var *res)
+Line 2576 
 static var *evaluate(node *op, var *res)
  						X.rsm = newfile(L.s);
  						fflush(X.rsm->F);
  					} else {
- 						fflush(NULL);
+ 						fflush_all();
  					}
  				}
  				break;
-Line 2572 
 static var *evaluate(node *op, var *res)
+Line 2621 
 static var *evaluate(node *op, var *res)
  				R.d--;
  				goto r_op_change;
  			case '!':
- 				L.d = istrue(X.v) ? 0 : 1;
+ 				L.d = !istrue(X.v);
  				break;
  			case '-':
  				L.d = -R.d;
-Line 2632 
 static var *evaluate(node *op, var *res)
+Line 2681 
 static var *evaluate(node *op, var *res)
  				L.d *= R.d;
  				break;
  			case '/':
- 				if (R.d == 0) syntax_error(EMSG_DIV_BY_ZERO);
+ 				if (R.d == 0)
+ 					syntax_error(EMSG_DIV_BY_ZERO);
  				L.d /= R.d;
  				break;
  			case '&':
-Line 2643 
 static var *evaluate(node *op, var *res)
+Line 2693 
 static var *evaluate(node *op, var *res)
  #endif
  				break;
  			case '%':
- 				if (R.d == 0) syntax_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;
  			}
-Line 2669 
 static var *evaluate(node *op, var *res)
+Line 2720 
 static var *evaluate(node *op, var *res)
  				R.i = (L.d == 0);
  				break;
  			}
- 			setvar_i(res, (opn & 0x1 ? R.i : !R.i) ? 1 : 0);
+ 			setvar_i(res, (opn & 1 ? R.i : !R.i) ? 1 : 0);
  			break;
  		default:
-Line 2751 
 static rstream *next_input_file(void)
+Line 2802 
 static rstream *next_input_file(void)
  	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;
-Line 2869 
 int awk_main(int argc, char **argv)
+Line 2921 
 int awk_main(int argc, char **argv)
  			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)
-Line 2891 
 int awk_main(int argc, char **argv)
+Line 2943 
 int awk_main(int argc, char **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) {

 Legend:



Removed from v.983
 


changed lines


 
Added in v.984
 Legend:



Removed from v.983
 


changed lines


 
Added in v.984
-Removed from v.983
+Added in v.984