summaryrefslogtreecommitdiff
path: root/string.c
diff options
context:
space:
mode:
Diffstat (limited to 'string.c')
-rw-r--r--string.c2015
1 files changed, 1117 insertions, 898 deletions
diff --git a/string.c b/string.c
index 3c3d24f..e5f0dd2 100644
--- a/string.c
+++ b/string.c
@@ -6,7 +6,7 @@
$Date$
created at: Mon Aug 9 17:12:58 JST 1993
- Copyright (C) 1993-1996 Yukihiro Matsumoto
+ Copyright (C) 1993-1998 Yukihiro Matsumoto
************************************************/
@@ -16,32 +16,28 @@
#define BEG(no) regs->beg[no]
#define END(no) regs->end[no]
-#include <stdio.h>
#include <ctype.h>
+
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
-VALUE cString;
-
-#define STRLEN(s) RSTRING(s)->len
+VALUE rb_cString;
#define STR_FREEZE FL_USER1
-#define STR_TAINT FL_USER2
-void reg_prepare_re _((VALUE));
-void kcode_reset_option _((void));
+#define STR_NO_ORIG FL_USER3
+
+extern VALUE rb_rs;
VALUE
-str_new(ptr, len)
- UCHAR *ptr;
- UINT len;
+rb_str_new(ptr, len)
+ char *ptr;
+ size_t len;
{
NEWOBJ(str, struct RString);
- OBJSETUP(str, cString, T_STRING);
+ OBJSETUP(str, rb_cString, T_STRING);
- if (rb_safe_level() >= 3) {
- FL_SET(str, STR_TAINT);
- }
+ str->ptr = 0;
str->len = len;
str->orig = 0;
str->ptr = ALLOC_N(char,len+1);
@@ -53,57 +49,98 @@ str_new(ptr, len)
}
VALUE
-str_new2(ptr)
- UCHAR *ptr;
+rb_str_new2(ptr)
+ char *ptr;
+{
+ return rb_str_new(ptr, strlen(ptr));
+}
+
+VALUE
+rb_tainted_str_new(ptr, len)
+ char *ptr;
+ size_t len;
{
- return str_new(ptr, strlen(ptr));
+ return rb_obj_taint(rb_str_new(ptr, len));
}
VALUE
-str_new3(str)
+rb_tainted_str_new2(ptr)
+ char *ptr;
+{
+ return rb_obj_taint(rb_str_new2(ptr));
+}
+
+VALUE
+rb_str_new3(str)
VALUE str;
{
NEWOBJ(str2, struct RString);
- OBJSETUP(str2, cString, T_STRING);
+ OBJSETUP(str2, rb_cString, T_STRING);
str2->len = RSTRING(str)->len;
str2->ptr = RSTRING(str)->ptr;
str2->orig = str;
- if (rb_safe_level() >= 3) {
- FL_SET(str2, STR_TAINT);
- }
-
return (VALUE)str2;
}
VALUE
-str_new4(orig)
+rb_str_new4(orig)
VALUE orig;
{
- NEWOBJ(str, struct RString);
- OBJSETUP(str, cString, T_STRING);
-
- str->len = RSTRING(orig)->len;
- str->ptr = RSTRING(orig)->ptr;
- if (RSTRING(orig)->orig) {
- str->orig = RSTRING(orig)->orig;
+ if (FL_TEST(orig, STR_FREEZE)) {
+ return orig;
+ }
+ else if (RSTRING(orig)->orig && !FL_TEST(orig, STR_NO_ORIG)) {
+ return rb_str_freeze(RSTRING(orig)->orig);
}
else {
+ NEWOBJ(str, struct RString);
+ OBJSETUP(str, rb_cString, T_STRING);
+
+ str->len = RSTRING(orig)->len;
+ str->ptr = RSTRING(orig)->ptr;
RSTRING(orig)->orig = (VALUE)str;
str->orig = 0;
+ if (rb_safe_level() >= 3) {
+ FL_SET(str, FL_TAINT);
+ }
+ return (VALUE)str;
}
- if (rb_safe_level() >= 3) {
- FL_SET(str, STR_TAINT);
- }
+}
- return (VALUE)str;
+VALUE
+rb_str_to_str(str)
+ VALUE str;
+{
+ return rb_convert_type(str, T_STRING, "String", "to_str");
+}
+
+static void
+rb_str_assign(str, str2)
+ VALUE str, str2;
+{
+ if (str == str2) return;
+ if (NIL_P(str2)) {
+ RSTRING(str)->ptr = 0;
+ RSTRING(str)->len = 0;
+ RSTRING(str)->orig = 0;
+ return;
+ }
+ if ((!RSTRING(str)->orig||FL_TEST(str, STR_NO_ORIG))&&RSTRING(str)->ptr)
+ free(RSTRING(str)->ptr);
+ RSTRING(str)->ptr = RSTRING(str2)->ptr;
+ RSTRING(str)->len = RSTRING(str2)->len;
+ RSTRING(str)->orig = RSTRING(str2)->orig;
+ RSTRING(str2)->ptr = 0; /* abandon str2 */
+ RSTRING(str2)->len = 0;
+ if (OBJ_TAINTED(str2)) OBJ_TAINT(str);
}
-static ID pr_str;
+static ID to_str;
VALUE
-obj_as_string(obj)
+rb_obj_as_string(obj)
VALUE obj;
{
VALUE str;
@@ -111,111 +148,130 @@ obj_as_string(obj)
if (TYPE(obj) == T_STRING) {
return obj;
}
- str = rb_funcall(obj, pr_str, 0);
+ str = rb_funcall(obj, to_str, 0);
if (TYPE(str) != T_STRING)
- return any_to_s(obj);
+ return rb_any_to_s(obj);
+ if (OBJ_TAINTED(obj)) OBJ_TAINT(str);
return str;
}
static VALUE
-str_clone(orig)
+rb_str_clone(orig)
VALUE orig;
{
VALUE str;
- if (RSTRING(orig)->orig)
- str = str_new3(RSTRING(orig)->orig);
+ if (RSTRING(orig)->orig && !FL_TEST(orig, STR_NO_ORIG))
+ str = rb_str_new3(RSTRING(orig)->orig);
else
- str = str_new(RSTRING(orig)->ptr, RSTRING(orig)->len);
+ str = rb_str_new(RSTRING(orig)->ptr, RSTRING(orig)->len);
+ if (RSTRING(orig)->orig && FL_TEST(orig, STR_NO_ORIG))
+ RSTRING(str)->orig = RSTRING(orig)->orig;
CLONESETUP(str, orig);
return str;
}
VALUE
-str_dup(str)
+rb_str_dup(str)
VALUE str;
{
- VALUE s = str_new(RSTRING(str)->ptr, RSTRING(str)->len);
- if (str_tainted(str)) s = str_taint(s);
+ VALUE s;
+
+ if (TYPE(str) != T_STRING) str = rb_str_to_str(str);
+ s = rb_str_new(RSTRING(str)->ptr, RSTRING(str)->len);
+ if (OBJ_TAINTED(str)) OBJ_TAINT(s);
+
return s;
}
static VALUE
-str_s_new(class, orig)
- VALUE class;
+rb_str_s_new(klass, orig)
+ VALUE klass;
VALUE orig;
{
NEWOBJ(str, struct RString);
- OBJSETUP(str, class, T_STRING);
+ OBJSETUP(str, klass, T_STRING);
- orig = obj_as_string(orig);
+ str->orig = 0;
+ orig = rb_obj_as_string(orig);
str->len = RSTRING(orig)->len;
- str->ptr = ALLOC_N(char, RSTRING(orig)->len+1);
- if (str->ptr) {
+ if (RSTRING(orig)->ptr) {
+ str->ptr = ALLOC_N(char, RSTRING(orig)->len+1);
memcpy(str->ptr, RSTRING(orig)->ptr, RSTRING(orig)->len);
+ str->ptr[RSTRING(orig)->len] = '\0';
}
- str->ptr[RSTRING(orig)->len] = '\0';
- str->orig = 0;
if (rb_safe_level() >= 3) {
- FL_SET(str, STR_TAINT);
+ FL_SET(str, FL_TAINT);
}
+ rb_obj_call_init((VALUE)str);
return (VALUE)str;
}
static VALUE
-str_length(str)
+rb_str_length(str)
VALUE str;
{
return INT2FIX(RSTRING(str)->len);
}
+static VALUE
+rb_str_empty(str)
+ VALUE str;
+{
+ if (RSTRING(str)->len == 0)
+ return Qtrue;
+ return Qfalse;
+}
+
VALUE
-str_plus(str1, str2)
+rb_str_plus(str1, str2)
VALUE str1, str2;
{
VALUE str3;
- str2 = obj_as_string(str2);
- str3 = str_new(0, RSTRING(str1)->len+RSTRING(str2)->len);
+ if (TYPE(str2) != T_STRING) str2 = rb_str_to_str(str2);
+ str3 = rb_str_new(0, RSTRING(str1)->len+RSTRING(str2)->len);
memcpy(RSTRING(str3)->ptr, RSTRING(str1)->ptr, RSTRING(str1)->len);
- memcpy(RSTRING(str3)->ptr+RSTRING(str1)->len, RSTRING(str2)->ptr, RSTRING(str2)->len);
+ memcpy(RSTRING(str3)->ptr + RSTRING(str1)->len,
+ RSTRING(str2)->ptr, RSTRING(str2)->len);
RSTRING(str3)->ptr[RSTRING(str3)->len] = '\0';
- if (str_tainted(str1) || str_tainted(str2))
- return str_taint(str3);
- return (VALUE)str3;
+ if (OBJ_TAINTED(str1) || OBJ_TAINTED(str2))
+ OBJ_TAINT(str3);
+ return str3;
}
VALUE
-str_times(str, times)
+rb_str_times(str, times)
VALUE str;
VALUE times;
{
VALUE str2;
- int i, len;
+ size_t i, len;
len = NUM2INT(times);
if (len < 0) {
- ArgError("negative argument");
+ rb_raise(rb_eArgError, "negative argument");
}
- str2 = str_new(0, RSTRING(str)->len*len);
+ str2 = rb_str_new(0, RSTRING(str)->len*len);
for (i=0; i<len; i++) {
- memcpy(RSTRING(str2)->ptr+(i*RSTRING(str)->len), RSTRING(str)->ptr, RSTRING(str)->len);
+ memcpy(RSTRING(str2)->ptr+(i*RSTRING(str)->len),
+ RSTRING(str)->ptr, RSTRING(str)->len);
}
RSTRING(str2)->ptr[RSTRING(str2)->len] = '\0';
- if (str_tainted(str)) {
- return str_taint((VALUE)str2);
+ if (OBJ_TAINTED(str)) {
+ OBJ_TAINT(str2);
}
return str2;
}
-VALUE
-str_format(str, arg)
+static VALUE
+rb_str_format(str, arg)
VALUE str, arg;
{
VALUE *argv;
@@ -224,44 +280,47 @@ str_format(str, arg)
argv = ALLOCA_N(VALUE, RARRAY(arg)->len + 1);
argv[0] = str;
MEMCPY(argv+1, RARRAY(arg)->ptr, VALUE, RARRAY(arg)->len);
- return f_sprintf(RARRAY(arg)->len+1, argv);
+ return rb_f_sprintf(RARRAY(arg)->len+1, argv);
}
argv = ALLOCA_N(VALUE, 2);
argv[0] = str;
argv[1] = arg;
- return f_sprintf(2, argv);
+ return rb_f_sprintf(2, argv);
}
VALUE
-str_substr(str, start, len)
+rb_str_substr(str, start, len)
VALUE str;
- int start, len;
+ size_t start, len;
{
- struct RString *str2;
+ VALUE str2;
if (start < 0) {
start = RSTRING(str)->len + start;
}
if (RSTRING(str)->len <= start || len < 0) {
- return str_new(0,0);
+ return rb_str_new(0,0);
}
if (RSTRING(str)->len < start + len) {
len = RSTRING(str)->len - start;
}
- return str_new(RSTRING(str)->ptr+start, len);
+ str2 = rb_str_new(RSTRING(str)->ptr+start, len);
+ if (OBJ_TAINTED(str)) OBJ_TAINT(str2);
+
+ return str2;
}
static VALUE
-str_subseq(str, beg, end)
+rb_str_subseq(str, beg, end)
VALUE str;
- int beg, end;
+ size_t beg, end;
{
- int len;
+ size_t len;
if ((beg > 0 && end > 0 || beg < 0 && end < 0) && beg > end) {
- IndexError("end smaller than beg [%d..%d]", beg, end);
+ rb_raise(rb_eIndexError, "end smaller than beg [%d..%d]", beg, end);
}
if (beg < 0) {
@@ -277,7 +336,7 @@ str_subseq(str, beg, end)
}
if (beg >= RSTRING(str)->len) {
- return str_new(0, 0);
+ return rb_str_new(0, 0);
}
len = end - beg + 1;
@@ -285,24 +344,20 @@ str_subseq(str, beg, end)
len = 0;
}
- return str_substr(str, beg, len);
+ return rb_str_substr(str, beg, len);
}
-extern VALUE ignorecase;
-
void
-str_modify(str)
+rb_str_modify(str)
VALUE str;
{
- UCHAR *ptr;
+ char *ptr;
- if (rb_safe_level() >= 5) {
- extern VALUE eSecurityError;
- Raise(eSecurityError, "cannot change string status");
- }
if (FL_TEST(str, STR_FREEZE))
- TypeError("can't modify frozen string");
- if (!RSTRING(str)->orig) return;
+ rb_raise(rb_eTypeError, "can't modify frozen string");
+ if (rb_safe_level() >= 4 && !FL_TEST(str, FL_TAINT))
+ rb_raise(rb_eSecurityError, "Insecure: can't modify string");
+ if (!RSTRING(str)->orig || FL_TEST(str, STR_NO_ORIG)) return;
ptr = RSTRING(str)->ptr;
RSTRING(str)->ptr = ALLOC_N(char, RSTRING(str)->len+1);
if (RSTRING(str)->ptr) {
@@ -313,7 +368,7 @@ str_modify(str)
}
VALUE
-str_freeze(str)
+rb_str_freeze(str)
VALUE str;
{
FL_SET(str, STR_FREEZE);
@@ -321,48 +376,32 @@ str_freeze(str)
}
static VALUE
-str_frozen_p(str)
+rb_str_frozen_p(str)
VALUE str;
{
if (FL_TEST(str, STR_FREEZE))
- return TRUE;
- return FALSE;
-}
-
-VALUE
-str_dup_freezed(str)
- VALUE str;
-{
- str = str_dup(str);
- str_freeze(str);
- return str;
+ return Qtrue;
+ return Qfalse;
}
VALUE
-str_taint(str)
+rb_str_dup_frozen(str)
VALUE str;
{
- if (TYPE(str) == T_STRING) {
- FL_SET(str, STR_TAINT);
+ if (RSTRING(str)->orig && !FL_TEST(str, STR_NO_ORIG)) {
+ return rb_str_freeze(RSTRING(str)->orig);
}
- return str;
-}
-
-VALUE
-str_tainted(str)
- VALUE str;
-{
- if (FL_TEST(str, STR_TAINT))
- return TRUE;
- return FALSE;
+ if (FL_TEST(str, STR_FREEZE))
+ return str;
+ return rb_str_freeze(rb_str_dup(str));
}
VALUE
-str_resize(str, len)
+rb_str_resize(str, len)
VALUE str;
- int len;
+ size_t len;
{
- str_modify(str);
+ rb_str_modify(str);
if (len >= 0) {
if (RSTRING(str)->len < len || RSTRING(str)->len - len > 1024) {
@@ -371,17 +410,17 @@ str_resize(str, len)
RSTRING(str)->len = len;
RSTRING(str)->ptr[len] = '\0'; /* sentinel */
}
- return (VALUE)str;
+ return str;
}
VALUE
-str_cat(str, ptr, len)
+rb_str_cat(str, ptr, len)
VALUE str;
- UCHAR *ptr;
- UINT len;
+ char *ptr;
+ size_t len;
{
if (len > 0) {
- str_modify(str);
+ rb_str_modify(str);
REALLOC_N(RSTRING(str)->ptr, char, RSTRING(str)->len + len + 1);
if (ptr)
memcpy(RSTRING(str)->ptr + RSTRING(str)->len, ptr, len);
@@ -391,24 +430,30 @@ str_cat(str, ptr, len)
return str;
}
-static VALUE
-str_concat(str1, str2)
+VALUE
+rb_str_concat(str1, str2)
VALUE str1, str2;
{
- str2 = obj_as_string(str2);
- str_cat(str1, RSTRING(str2)->ptr, RSTRING(str2)->len);
- return str1;
+ if (FIXNUM_P(str2)) {
+ int i = FIX2INT(str2);
+ if (0 <= i && i <= 0xff) { /* byte */
+ char c = i;
+ return rb_str_cat(str1, &c, 1);
+ }
+ }
+ if (TYPE(str2) != T_STRING) str2 = rb_str_to_str(str2);
+ return rb_str_cat(str1, RSTRING(str2)->ptr, RSTRING(str2)->len);
}
int
-str_hash(str)
+rb_str_hash(str)
VALUE str;
{
- register int len = RSTRING(str)->len;
- register UCHAR *p = RSTRING(str)->ptr;
+ register size_t len = RSTRING(str)->len;
+ register char *p = RSTRING(str)->ptr;
register int key = 0;
- if (RTEST(ignorecase)) {
+ if (ruby_ignorecase) {
while (len--) {
key = key*65599 + toupper(*p);
p++;
@@ -424,75 +469,79 @@ str_hash(str)
}
static VALUE
-str_hash_method(str)
+rb_str_hash_method(str)
VALUE str;
{
- int key = str_hash(str);
+ int key = rb_str_hash(str);
return INT2FIX(key);
}
-#define min(a,b) (((a)>(b))?(b):(a))
+#define lesser(a,b) (((a)>(b))?(b):(a))
int
-str_cmp(str1, str2)
+rb_str_cmp(str1, str2)
VALUE str1, str2;
{
- UINT len;
+ size_t len;
int retval;
- if (RTEST(ignorecase)) {
- return str_cicmp(str1, str2);
+ if (ruby_ignorecase) {
+ return rb_str_cicmp(str1, str2);
}
- len = min(RSTRING(str1)->len, RSTRING(str2)->len);
+ len = lesser(RSTRING(str1)->len, RSTRING(str2)->len);
retval = memcmp(RSTRING(str1)->ptr, RSTRING(str2)->ptr, len);
if (retval == 0) {
- return RSTRING(str1)->ptr[len] - RSTRING(str2)->ptr[len];
+ if (RSTRING(str1)->len == RSTRING(str2)->len) return 0;
+ if (RSTRING(str1)->len > RSTRING(str2)->len) return 1;
+ return -1;
}
- return retval;
+ if (retval == 0) return 0;
+ if (retval > 0) return 1;
+ return -1;
}
static VALUE
-str_equal(str1, str2)
+rb_str_equal(str1, str2)
VALUE str1, str2;
{
if (TYPE(str2) != T_STRING)
- return FALSE;
+ return Qfalse;
if (RSTRING(str1)->len == RSTRING(str2)->len
- && str_cmp(str1, str2) == 0) {
- return TRUE;
+ && rb_str_cmp(str1, str2) == 0) {
+ return Qtrue;
}
- return FALSE;
+ return Qfalse;
}
static VALUE
-str_cmp_method(str1, str2)
+rb_str_cmp_method(str1, str2)
VALUE str1, str2;
{
int result;
- str2 = obj_as_string(str2);
- result = str_cmp(str1, str2);
+ if (TYPE(str2) != T_STRING) str2 = rb_str_to_str(str2);
+ result = rb_str_cmp(str1, str2);
return INT2FIX(result);
}
static VALUE
-str_match(x, y)
+rb_str_match(x, y)
VALUE x, y;
{
VALUE reg;
- int start;
+ size_t start;
switch (TYPE(y)) {
case T_REGEXP:
- return reg_match(y, x);
+ return rb_reg_match(y, x);
case T_STRING:
- reg = reg_regcomp(y);
- start = reg_search(reg, x, 0, 0);
+ reg = rb_reg_regcomp(y);
+ start = rb_reg_search(reg, x, 0, 0);
if (start == -1) {
- return FALSE;
+ return Qfalse;
}
return INT2FIX(start);
@@ -502,19 +551,19 @@ str_match(x, y)
}
static VALUE
-str_match2(str)
+rb_str_match2(str)
VALUE str;
{
- return reg_match2(reg_regcomp(str));
+ return rb_reg_match2(rb_reg_regcomp(str));
}
-static int
-str_index(str, sub, offset)
+static size_t
+rb_str_index(str, sub, offset)
VALUE str, sub;
- int offset;
+ size_t offset;
{
- UCHAR *s, *e, *p;
- int len;
+ char *s, *e, *p;
+ size_t len;
if (RSTRING(str)->len - offset < RSTRING(sub)->len) return -1;
s = RSTRING(str)->ptr+offset;
@@ -527,21 +576,21 @@ str_index(str, sub, offset)
}
s++;
}
- return -1;
+ return (size_t)-1;
}
static VALUE
-str_index_method(argc, argv, str)
+rb_str_index_method(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
VALUE sub;
VALUE initpos;
- int pos;
+ size_t pos;
if (rb_scan_args(argc, argv, "11", &sub, &initpos) == 2) {
- pos = NUM2INT(initpos);
+ pos = NUM2UINT(initpos);
}
else {
pos = 0;
@@ -549,17 +598,17 @@ str_index_method(argc, argv, str)
switch (TYPE(sub)) {
case T_REGEXP:
- pos = reg_search(sub, str, pos, (struct re_registers *)-1);
+ pos = rb_reg_search(sub, str, pos, 0);
break;
case T_STRING:
- pos = str_index(str, sub, pos);
+ pos = rb_str_index(str, sub, pos);
break;
case T_FIXNUM:
{
int c = FIX2INT(sub);
- int len = RSTRING(str)->len;
+ size_t len = RSTRING(str)->len;
char *p = RSTRING(str)->ptr;
for (;pos<len;pos++) {
@@ -569,7 +618,8 @@ str_index_method(argc, argv, str)
}
default:
- TypeError("Type mismatch: %s given", rb_class2name(CLASS_OF(sub)));
+ rb_raise(rb_eTypeError, "Type mismatch: %s given",
+ rb_class2name(CLASS_OF(sub)));
}
if (pos == -1) return Qnil;
@@ -577,18 +627,18 @@ str_index_method(argc, argv, str)
}
static VALUE
-str_rindex(argc, argv, str)
+rb_str_rindex(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
VALUE sub;
VALUE initpos;
- int pos, len;
- UCHAR *s, *sbeg, *t;
+ size_t pos, len;
+ char *s, *sbeg, *t;
if (rb_scan_args(argc, argv, "11", &sub, &initpos) == 2) {
- pos = NUM2INT(initpos);
+ pos = NUM2UINT(initpos);
if (pos >= RSTRING(str)->len) pos = RSTRING(str)->len;
}
else {
@@ -597,11 +647,7 @@ str_rindex(argc, argv, str)
switch (TYPE(sub)) {
case T_REGEXP:
- reg_prepare_re(sub);
- pos = re_search(RREGEXP(sub)->ptr,
- RSTRING(str)->ptr, RSTRING(str)->len,
- pos, -pos, 0);
- kcode_reset_option();
+ pos = rb_reg_search(sub, str, pos, 1);
if (pos >= 0) return INT2FIX(pos);
break;
@@ -631,14 +677,15 @@ str_rindex(argc, argv, str)
}
default:
- TypeError("Type mismatch: %s given", rb_class2name(CLASS_OF(sub)));
+ rb_raise(rb_eTypeError, "Type mismatch: %s given",
+ rb_class2name(CLASS_OF(sub)));
}
return Qnil;
}
-static UCHAR
+static char
succ_char(s)
- UCHAR *s;
+ char *s;
{
char c = *s;
@@ -662,19 +709,19 @@ succ_char(s)
}
static VALUE
-str_succ(orig)
+rb_str_succ(orig)
VALUE orig;
{
VALUE str, str2;
- UCHAR *sbeg, *s;
+ char *sbeg, *s;
char c = -1;
- str = str_new(RSTRING(orig)->ptr, RSTRING(orig)->len);
+ str = rb_str_new(RSTRING(orig)->ptr, RSTRING(orig)->len);
sbeg = RSTRING(str)->ptr; s = sbeg + RSTRING(str)->len - 1;
while (sbeg <= s) {
- if (isalnum(*s) && (c = succ_char(s)) == 0) break;
+ if (ISALNUM(*s) && (c = succ_char(s)) == 0) break;
s--;
}
if (s < sbeg) {
@@ -682,35 +729,45 @@ str_succ(orig)
RSTRING(str)->ptr[RSTRING(str)->len-1] += 1;
}
else {
- str2 = str_new(0, RSTRING(str)->len+1);
+ str2 = rb_str_new(0, RSTRING(str)->len+1);
RSTRING(str2)->ptr[0] = c;
memcpy(RSTRING(str2)->ptr+1, RSTRING(str)->ptr, RSTRING(str)->len);
str = str2;
}
}
- if (str_tainted(orig)) {
- return str_taint(str);
+ if (OBJ_TAINTED(orig)) {
+ OBJ_TAINT(str);
}
return str;
}
+static VALUE
+rb_str_succ_bang(str)
+ VALUE str;
+{
+ rb_str_modify(str);
+ rb_str_assign(str, rb_str_succ(str));
+
+ return str;
+}
+
VALUE
-str_upto(beg, end)
+rb_str_upto(beg, end)
VALUE beg, end;
{
VALUE current;
- Check_Type(end, T_STRING);
+ if (TYPE(end) != T_STRING) end = rb_str_to_str(end);
if (RTEST(rb_funcall(beg, '>', 1, end)))
return Qnil;
current = beg;
for (;;) {
rb_yield(current);
- if (str_equal(current, end)) break;
- current = str_succ(current);
+ if (rb_str_equal(current, end)) break;
+ current = rb_str_succ(current);
if (RSTRING(current)->len > RSTRING(end)->len)
break;
}
@@ -719,11 +776,11 @@ str_upto(beg, end)
}
static VALUE
-str_aref(str, indx)
+rb_str_aref(str, indx)
VALUE str;
VALUE indx;
{
- int idx;
+ size_t idx;
switch (TYPE(indx)) {
case T_FIXNUM:
@@ -735,31 +792,32 @@ str_aref(str, indx)
if (idx < 0 || RSTRING(str)->len <= idx) {
return Qnil;
}
- return (VALUE)INT2FIX(RSTRING(str)->ptr[idx] & 0xff);
+ return INT2FIX(RSTRING(str)->ptr[idx] & 0xff);
case T_REGEXP:
- if (str_match(str, indx))
- return reg_last_match(0);
+ if (rb_reg_search(indx, str, 0, 0) >= 0)
+ return rb_reg_last_match(rb_backref_get());
return Qnil;
case T_STRING:
- if (str_index(str, indx, 0) != -1) return indx;
+ if (rb_str_index(str, indx, 0) != (size_t)-1) return indx;
return Qnil;
default:
/* check if indx is Range */
{
- int beg, end;
- if (range_beg_end(indx, &beg, &end)) {
- return str_subseq(str, beg, end);
+ size_t beg, end;
+ if (rb_range_beg_end(indx, &beg, &end)) {
+ return rb_str_subseq(str, beg, end);
}
}
- IndexError("Invalid index for string");
+ rb_raise(rb_eIndexError, "Invalid index for string");
}
+ return Qnil; /* not reached */
}
static VALUE
-str_aref_method(argc, argv, str)
+rb_str_aref_method(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
@@ -767,15 +825,15 @@ str_aref_method(argc, argv, str)
VALUE arg1, arg2;
if (rb_scan_args(argc, argv, "11", &arg1, &arg2) == 2) {
- return str_substr(str, NUM2INT(arg1), NUM2INT(arg2));
+ return rb_str_substr(str, NUM2INT(arg1), NUM2INT(arg2));
}
- return str_aref(str, arg1);
+ return rb_str_aref(str, arg1);
}
static void
-str_replace(str, beg, len, val)
+rb_str_replace(str, beg, len, val)
VALUE str, val;
- int beg, len;
+ size_t beg, len;
{
if (len < RSTRING(val)->len) {
/* expand string */
@@ -783,25 +841,28 @@ str_replace(str, beg, len, val)
}
if (len != RSTRING(val)->len) {
- memmove(RSTRING(str)->ptr+beg+RSTRING(val)->len,
- RSTRING(str)->ptr+beg+len,
- RSTRING(str)->len-(beg+len));
+ memmove(RSTRING(str)->ptr + beg + RSTRING(val)->len,
+ RSTRING(str)->ptr + beg + len,
+ RSTRING(str)->len - (beg + len));
+ }
+ if (RSTRING(str)->len < beg && len < 0) {
+ MEMZERO(RSTRING(str)->ptr + RSTRING(str)->len, char, -len);
}
memcpy(RSTRING(str)->ptr+beg, RSTRING(val)->ptr, RSTRING(val)->len);
RSTRING(str)->len += RSTRING(val)->len - len;
RSTRING(str)->ptr[RSTRING(str)->len] = '\0';
}
-/* str_replace2() understands negatice offset */
+/* rb_str_replace2() understands negatice offset */
static void
-str_replace2(str, beg, end, val)
- VALUE str, *val;
- int beg, end;
+rb_str_replace2(str, beg, end, val)
+ VALUE str, val;
+ size_t beg, end;
{
- int len;
+ size_t len;
if ((beg > 0 && end > 0 || beg < 0 && end < 0) && beg > end) {
- IndexError("end smaller than beg [%d..%d]", beg, end);
+ rb_raise(rb_eIndexError, "end smaller than beg [%d..%d]", beg, end);
}
if (beg < 0) {
@@ -827,307 +888,295 @@ str_replace2(str, beg, end, val)
len = 0;
}
- str_replace(str, beg, len, val);
+ rb_str_replace(str, beg, len, val);
}
-static VALUE
-str_sub_s(str, pat, val, once)
- VALUE str, pat, val;
- int once;
-{
- VALUE result, repl;
- int beg, offset, n;
- struct re_registers *regs;
-
- switch (TYPE(pat)) {
- case T_REGEXP:
- break;
-
- case T_STRING:
- pat = reg_regcomp(pat);
- break;
-
- default:
- /* type failed */
- Check_Type(pat, T_REGEXP);
- }
-
- val = obj_as_string(val);
- result = str_new(0,0);
- offset=0; n=0;
- while ((beg=reg_search(pat, str, offset, 0)) >= 0) {
- n++;
-
- regs = RMATCH(backref_get())->regs;
- str_cat(result, RSTRING(str)->ptr+offset, beg-offset);
-
- repl = reg_regsub(val, str, regs);
- str_cat(result, RSTRING(repl)->ptr, RSTRING(repl)->len);
- if (END(0) == offset) {
- /*
- * Always consume at least one character of the input string
- * in order to prevent infinite loops.
- */
- if (RSTRING(str)->len > 0) {
- str_cat(result, RSTRING(str)->ptr+END(0), 1);
- }
- offset = END(0)+1;
- }
- else {
- offset = END(0);
- }
-
- if (once) break;
- if (offset >= STRLEN(str)) break;
- }
- if (n == 0) return Qnil;
- if (RSTRING(str)->len > offset) {
- str_cat(result, RSTRING(str)->ptr+offset, RSTRING(str)->len-offset);
- }
-
- if (str_tainted(val)) str_taint(result);
- return result;
-}
+static VALUE rb_str_sub_bang _((int, VALUE*, VALUE));
static VALUE
-str_sub_f(str, pat, val, once)
- VALUE str;
- VALUE pat;
- VALUE val;
- int once;
-{
- VALUE result;
-
- str_modify(str);
- result = str_sub_s(str, pat, val, once);
-
- if (NIL_P(result)) return Qnil;
- str_resize(str, RSTRING(result)->len);
- memcpy(RSTRING(str)->ptr, RSTRING(result)->ptr, RSTRING(result)->len);
- if (str_tainted(result)) str_taint(str);
-
- return (VALUE)str;
-}
-
-static VALUE
-str_sub_iter_s(str, pat, once)
- VALUE str;
- VALUE pat;
- int once;
-{
- VALUE val, result;
- int beg, offset, n, null;
- struct re_registers *regs;
-
- if (!iterator_p()) {
- ArgError("Wrong # of arguments(1 for 2)");
- }
-
- switch (TYPE(pat)) {
- case T_REGEXP:
- break;
-
- case T_STRING:
- pat = reg_regcomp(pat);
- break;
-
- default:
- /* type failed */
- Check_Type(pat, T_REGEXP);
- }
-
- result = str_new(0,0);
- n = 0; offset = 0;
- while ((beg=reg_search(pat, str, offset, 0)) >= 0) {
- n++;
-
- null = 0;
- str_cat(result, RSTRING(str)->ptr+offset, beg-offset);
-
- regs = RMATCH(backref_get())->regs;
- if (END(0) == offset) {
- null = 1;
- offset = END(0)+1;
- }
- else {
- offset = END(0);
- }
-
- val = rb_yield(reg_nth_match(0, backref_get()));
- val = obj_as_string(val);
- str_cat(result, RSTRING(val)->ptr, RSTRING(val)->len);
- if (null && RSTRING(str)->len) {
- str_cat(result, RSTRING(str)->ptr+offset-1, 1);
- }
-
- if (once) break;
- if (offset >= STRLEN(str)) break;
- }
- if (n == 0) return Qnil;
- if (RSTRING(str)->len > offset) {
- str_cat(result, RSTRING(str)->ptr+offset, RSTRING(str)->len-offset);
- }
-
- return result;
-}
-
-static VALUE
-str_sub_iter_f(str, pat, once)
- VALUE str;
- VALUE pat;
- int once;
-{
- VALUE result;
-
- str_modify(str);
- result = str_sub_iter_s(str, pat, once);
-
- if (NIL_P(result)) return Qnil;
- str_resize(str, RSTRING(result)->len);
- memcpy(RSTRING(str)->ptr, RSTRING(result)->ptr, RSTRING(result)->len);
-
- return (VALUE)str;
-}
-
-static VALUE
-str_aset(str, indx, val)
+rb_str_aset(str, indx, val)
VALUE str;
VALUE indx, val;
{
- int idx, beg, end, offset;
+ size_t idx, beg, end;
switch (TYPE(indx)) {
case T_FIXNUM:
- idx = NUM2INT(indx);
+ idx = NUM2UINT(indx);
if (idx < 0) {
idx = RSTRING(str)->len + idx;
}
if (idx < 0 || RSTRING(str)->len <= idx) {
- IndexError("index %d out of range [0..%d]", idx, RSTRING(str)->len-1);
+ rb_raise(rb_eIndexError, "index %d out of range [0..%d]", idx,
+ RSTRING(str)->len - 1);
+ }
+ if (TYPE(val) == T_STRING) {
+ rb_str_replace(str, idx, 1, val);
+ }
+ else {
+ RSTRING(str)->ptr[idx] = NUM2INT(val) & 0xff;
}
- RSTRING(str)->ptr[idx] = FIX2INT(val) & 0xff;
return val;
case T_REGEXP:
- str_sub_f(str, indx, val, 0);
+ {
+ VALUE args[2];
+ args[0] = indx;
+ args[1] = val;
+ rb_str_sub_bang(2, args, str);
+ }
return val;
case T_STRING:
- for (offset=0;
- (beg=str_index(str, indx, offset)) >= 0;
- offset=beg+STRLEN(val)) {
- end = beg + STRLEN(indx) - 1;
- str_replace2(str, beg, end, val);
+ beg = rb_str_index(str, indx, 0);
+ if (beg != (size_t)-1) {
+ end = beg + RSTRING(indx)->len - 1;
+ rb_str_replace2(str, beg, end, val);
}
- if (offset == 0) return Qnil;
return val;
default:
/* check if indx is Range */
{
- int beg, end;
- if (range_beg_end(indx, &beg, &end)) {
- str_replace2(str, beg, end, val);
+ size_t beg, end;
+ if (rb_range_beg_end(indx, &beg, &end)) {
+ if (TYPE(val) != T_STRING) val = rb_str_to_str(val);
+ rb_str_replace2(str, beg, end, val);
return val;
}
}
- IndexError("Invalid index for string");
+ rb_raise(rb_eIndexError, "Invalid index for string");
}
}
static VALUE
-str_aset_method(argc, argv, str)
+rb_str_aset_method(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
VALUE arg1, arg2, arg3;
- str_modify(str);
+ rb_str_modify(str);
if (rb_scan_args(argc, argv, "21", &arg1, &arg2, &arg3) == 3) {
- int beg, len;
-
- Check_Type(arg3, T_STRING);
+ size_t beg, len;
- beg = NUM2INT(arg1);
+ if (TYPE(arg3) != T_STRING) arg3 = rb_str_to_str(arg3);
+ beg = NUM2UINT(arg1);
if (beg < 0) {
beg = RSTRING(str)->len + beg;
if (beg < 0) beg = 0;
}
- len = NUM2INT(arg2);
- if (len < 0) IndexError("negative length %d", len);
+ len = NUM2UINT(arg2);
+ if (len < 0) rb_raise(rb_eIndexError, "negative length %d", len);
if (beg + len > RSTRING(str)->len) {
len = RSTRING(str)->len - beg;
}
- str_replace(str, beg, len, arg3);
+ rb_str_replace(str, beg, len, arg3);
return arg3;
}
- return str_aset(str, arg1, arg2);
+ return rb_str_aset(str, arg1, arg2);
}
static VALUE
-str_sub_bang(argc, argv, str)
- int argc;
- VALUE *argv;
- VALUE str;
+get_pat(pat)
+ VALUE pat;
{
- VALUE pat, val;
+ switch (TYPE(pat)) {
+ case T_REGEXP:
+ break;
+
+ case T_STRING:
+ pat = rb_reg_regcomp(pat);
+ break;
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- return str_sub_iter_f(str, pat, 1);
+ default:
+ /* type failed */
+ Check_Type(pat, T_REGEXP);
}
- return str_sub_f(str, pat, val, 1);
+ return pat;
}
static VALUE
-str_sub(argc, argv, str)
+rb_str_sub_bang(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
- VALUE pat, val, v;
+ VALUE pat, repl, match;
+ struct re_registers *regs;
+ int iter = 0;
+ size_t plen;
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- v = str_sub_iter_s(str, pat, 1);
+ if (argc == 1 && rb_iterator_p()) {
+ iter = 1;
+ }
+ else if (argc == 2) {
+ repl = rb_obj_as_string(argv[1]);;
}
else {
- v = str_sub_s(str, pat, val, 1);
+ rb_raise(rb_eArgError, "Wrong # of arguments(%d for 2)", argc);
}
- if (NIL_P(v)) return str_dup(str);
- return v;
+
+ pat = get_pat(argv[0]);
+ if (rb_reg_search(pat, str, 0, 0) >= 0) {
+ rb_str_modify(str);
+ match = rb_backref_get();
+ regs = RMATCH(match)->regs;
+
+ if (iter) {
+ repl = rb_obj_as_string(rb_yield(rb_reg_nth_match(0, match)));
+ }
+ else {
+ repl = rb_reg_regsub(repl, str, regs);
+ }
+ plen = END(0) - BEG(0);
+ if (RSTRING(repl)->len > plen) {
+ REALLOC_N(RSTRING(str)->ptr, char,
+ RSTRING(str)->len + RSTRING(repl)->len - plen + 1);
+ }
+ if (RSTRING(repl)->len != plen) {
+ memmove(RSTRING(str)->ptr + BEG(0) + RSTRING(repl)->len,
+ RSTRING(str)->ptr + BEG(0) + plen,
+ RSTRING(str)->len - BEG(0) - plen);
+ }
+ memcpy(RSTRING(str)->ptr + BEG(0),
+ RSTRING(repl)->ptr, RSTRING(repl)->len);
+ RSTRING(str)->len += RSTRING(repl)->len - plen;
+ RSTRING(str)->ptr[RSTRING(str)->len] = '\0';
+ return str;
+ }
+ return Qnil;
}
static VALUE
-str_gsub_bang(argc, argv, str)
+rb_str_sub(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
- VALUE pat, val;
+ VALUE val = rb_str_sub_bang(argc, argv, str = rb_str_dup(str));
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- return str_sub_iter_f(str, pat, 0);
- }
- return str_sub_f(str, pat, val, 0);
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-str_gsub(argc, argv, str)
+rb_str_gsub_bang(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
- VALUE pat, val, v;
+ VALUE pat, val, repl, match;
+ struct re_registers *regs;
+ int beg, offset, n;
+ int iter = 0;
+ char *buf, *bp, *cp;
+ size_t blen, len;
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- v = str_sub_iter_s(str, pat, 0);
+ if (argc == 1 && rb_iterator_p()) {
+ iter = 1;
+ }
+ else if (argc == 2) {
+ repl = rb_obj_as_string(argv[1]);;
}
else {
- v = str_sub_s(str, pat, val, 0);
+ rb_raise(rb_eArgError, "Wrong # of arguments(%d for 2)", argc);
+ }
+
+ pat = get_pat(argv[0]);
+ offset=0; n=0;
+ beg = rb_reg_search(pat, str, 0, 0);
+ if (beg < 0) return Qnil; /* no match, no substitution */
+
+ blen = RSTRING(str)->len + 30; /* len + margin */
+ buf = ALLOC_N(char, blen);
+ bp = buf;
+ cp = RSTRING(str)->ptr;
+
+ while (beg >= 0) {
+ n++;
+ match = rb_backref_get();
+ regs = RMATCH(match)->regs;
+ if (iter) {
+ val = rb_obj_as_string(rb_yield(rb_reg_nth_match(0, match)));
+ }
+ else {
+ val = rb_reg_regsub(repl, str, regs);
+ }
+ len = (bp - buf) + (beg - offset) + RSTRING(val)->len + 3;
+ if (blen < len) {
+ while (blen < len) blen *= 2;
+ len = bp - buf;
+ REALLOC_N(buf, char, blen);
+ bp = buf + len;
+ }
+ len = beg - offset; /* copy pre-match substr */
+ memcpy(bp, cp, len);
+ bp += len;
+ memcpy(bp, RSTRING(val)->ptr, RSTRING(val)->len);
+ bp += RSTRING(val)->len;
+ if (BEG(0) == END(0)) {
+ /*
+ * Always consume at least one character of the input string
+ * in order to prevent infinite loops.
+ */
+ len = mbclen(RSTRING(str)->ptr[END(0)]);
+ if (RSTRING(str)->len > END(0)) {
+ memcpy(bp, RSTRING(str)->ptr+END(0), len);
+ bp += len;
+ }
+ offset = END(0) + len;
+ }
+ else {
+ offset = END(0);
+ }
+ cp = RSTRING(str)->ptr + offset;
+ if (offset > RSTRING(str)->len) break;
+ beg = rb_reg_search(pat, str, offset, 0);
+ }
+ if (RSTRING(str)->len > offset) {
+ len = bp - buf;
+ if (blen - len < RSTRING(str)->len - offset) {
+ REALLOC_N(buf, char, len + RSTRING(str)->len - offset + 1);
+ bp = buf + len;
+ }
+ memcpy(bp, cp, RSTRING(str)->len - offset);
+ bp += RSTRING(str)->len - offset;
}
- if (NIL_P(v)) return str_dup(str);
- return v;
+ rb_str_modify(str);
+ free(RSTRING(str)->ptr);
+ RSTRING(str)->ptr = buf;
+ RSTRING(str)->len = len = bp - buf;
+ RSTRING(str)->ptr[len] = '\0';
+
+ return str;
+}
+
+static VALUE
+rb_str_gsub(argc, argv, str)
+ int argc;
+ VALUE *argv;
+ VALUE str;
+{
+ VALUE val = rb_str_gsub_bang(argc, argv, str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
+}
+
+static VALUE
+rb_str_replace_method(str, str2)
+ VALUE str, str2;
+{
+ if (TYPE(str2) != T_STRING) str2 = rb_str_to_str(str2);
+ rb_str_modify(str);
+ rb_str_resize(str, RSTRING(str2)->len);
+ memcpy(RSTRING(str)->ptr, RSTRING(str2)->ptr, RSTRING(str2)->len);
+ if (OBJ_TAINTED(str2)) OBJ_TAINT(str);
+
+ return str;
}
static VALUE
@@ -1135,108 +1184,86 @@ uscore_get()
{
VALUE line;
- line = lastline_get();
+ line = rb_lastline_get();
if (TYPE(line) != T_STRING) {
- TypeError("$_ value need to be String (%s given)",
- rb_class2name(CLASS_OF(line)));
+ rb_raise(rb_eTypeError, "$_ value need to be String (%s given)",
+ NIL_P(line)?"nil":rb_class2name(CLASS_OF(line)));
}
return line;
}
static VALUE
-f_sub_bang(argc, argv)
+rb_f_sub_bang(argc, argv)
int argc;
VALUE *argv;
{
- VALUE pat, val, line;
-
- line = uscore_get();
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- return str_sub_iter_f(line, pat, 1);
- }
- return str_sub_f(line, pat, val, 1);
+ return rb_str_sub_bang(argc, argv, uscore_get());
}
static VALUE
-f_sub(argc, argv)
+rb_f_sub(argc, argv)
int argc;
VALUE *argv;
{
- VALUE pat, val, line, v;
+ VALUE line, v;
- line = uscore_get();
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- v = str_sub_iter_s(line, pat, 1);
- }
- else {
- v = str_sub_s(line, pat, val, 1);
- }
- if (!NIL_P(v)) {
- lastline_set(v);
- return v;
- }
- return line;
+ line = rb_str_dup(uscore_get());
+ v = rb_str_sub_bang(argc, argv, line);
+ if (NIL_P(v)) return line;
+ rb_lastline_set(v);
+ return v;
}
static VALUE
-f_gsub_bang(argc, argv)
+rb_f_gsub_bang(argc, argv)
int argc;
VALUE *argv;
{
- VALUE pat, val, line;
-
- line = uscore_get();
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- return str_sub_iter_f(line, pat, 0);
- }
- return str_sub_f(line, pat, val, 0);
+ return rb_str_gsub_bang(argc, argv, uscore_get());
}
static VALUE
-f_gsub(argc, argv)
+rb_f_gsub(argc, argv)
int argc;
VALUE *argv;
{
- VALUE pat, val, line, v;
-
- line = uscore_get();
- if (rb_scan_args(argc, argv, "11", &pat, &val) == 1) {
- v = str_sub_iter_s(line, pat, 0);
- }
- else {
- v = str_sub_s(line, pat, val, 0);
- }
- if (NIL_P(v)) v = str_dup(line);
- lastline_set(v);
+ VALUE line, v;
+ line = rb_str_dup(uscore_get());
+ v = rb_str_gsub_bang(argc, argv, line);
+ if (NIL_P(v)) return line;
+ rb_lastline_set(v);
return v;
}
static VALUE
-str_reverse_bang(str)
+rb_str_reverse_bang(str)
VALUE str;
{
- UCHAR *s, *e, *p;
+ char *s, *e, *p, *q;
s = RSTRING(str)->ptr;
e = s + RSTRING(str)->len - 1;
- p = ALLOCA_N(char, RSTRING(str)->len);
+ p = q = ALLOCA_N(char, RSTRING(str)->len);
while (e >= s) {
*p++ = *e--;
}
- MEMCPY(RSTRING(str)->ptr, p, char, RSTRING(str)->len);
+ MEMCPY(RSTRING(str)->ptr, q, char, RSTRING(str)->len);
- return (VALUE)str;
+ return str;
}
static VALUE
-str_reverse(str)
+rb_str_reverse(str)
VALUE str;
{
- VALUE obj = str_new(0, RSTRING(str)->len);
- UCHAR *s, *e, *p;
+ VALUE obj;
+ char *s, *e, *p;
+
+ if (RSTRING(str)->len <= 1) return str;
+ obj = rb_str_new(0, RSTRING(str)->len);
s = RSTRING(str)->ptr; e = s + RSTRING(str)->len - 1;
p = RSTRING(obj)->ptr;
@@ -1248,14 +1275,14 @@ str_reverse(str)
}
static VALUE
-str_include(str, arg)
+rb_str_include(str, arg)
VALUE str, arg;
{
- int i;
+ size_t i;
if (FIXNUM_P(arg)) {
int c = FIX2INT(arg);
- int len = RSTRING(str)->len;
+ size_t len = RSTRING(str)->len;
char *p = RSTRING(str)->ptr;
for (i=0; i<len; i++) {
@@ -1263,51 +1290,47 @@ str_include(str, arg)
return INT2FIX(i);
}
}
- return FALSE;
+ return Qfalse;
}
- Check_Type(arg, T_STRING);
- i = str_index(str, arg, 0);
+ if (TYPE(arg) != T_STRING) arg = rb_str_to_str(arg);
+ i = rb_str_index(str, arg, 0);
- if (i == -1) return FALSE;
+ if (i == (size_t)-1) return Qfalse;
return INT2FIX(i);
}
static VALUE
-str_to_i(str)
+rb_str_to_i(str)
VALUE str;
{
- return str2inum(RSTRING(str)->ptr, 10);
+ return rb_str2inum(RSTRING(str)->ptr, 10);
}
-#ifndef atof
-double atof();
-#endif
-
static VALUE
-str_to_f(str)
+rb_str_to_f(str)
VALUE str;
{
double f = atof(RSTRING(str)->ptr);
- return float_new(f);
+ return rb_float_new(f);
}
static VALUE
-str_to_s(str)
+rb_str_to_s(str)
VALUE str;
{
return str;
}
VALUE
-str_inspect(str)
+rb_str_inspect(str)
VALUE str;
{
#define STRMAX 80
- UCHAR buf[STRMAX];
- UCHAR *p, *pend;
- UCHAR *b;
+ char buf[STRMAX];
+ char *p, *pend;
+ char *b;
p = RSTRING(str)->ptr; pend = p + RSTRING(str)->len;
b = buf;
@@ -1322,23 +1345,26 @@ str_inspect(str)
}
while (p < pend) {
- UCHAR c = *p++;
+ char c = *p++;
if (ismbchar(c) && p < pend) {
- CHECK(2);
+ int len = mbclen(c)-1;
+
+ CHECK(len);
*b++ = c;
- *b++ = *p++;
+ while (len--) {
+ *b++ = *p++;
+ }
}
- else if (c == '"') {
- CHECK(2);
- *b++ = '\\';
- *b++ = '"';
+ else if ((c & 0x80) && rb_kcode() != MBCTYPE_EUC) {
+ CHECK(1);
+ *b++ = c;
}
- else if (c == '\\') {
+ else if (c == '"'|| c == '\\') {
CHECK(2);
*b++ = '\\';
- *b++ = '\\';
+ *b++ = c;
}
- else if (isprint(c)) {
+ else if (ISPRINT(c)) {
CHECK(1);
*b++ = c;
}
@@ -1362,7 +1388,7 @@ str_inspect(str)
*b++ = '\\';
*b++ = 'f';
}
- else if (c == '\13') {
+ else if (c == '\013') {
CHECK(2);
*b++ = '\\';
*b++ = 'v';
@@ -1380,123 +1406,248 @@ str_inspect(str)
else {
CHECK(4);
*b++ = '\\';
- sprintf(b, "%03o", c);
+ sprintf(b, "%03o", c & 0377);
b += 3;
}
}
*b++ = '"';
- return str_new(buf, b - buf);
+ return rb_str_new(buf, b - buf);
+}
+
+static VALUE
+rb_str_dump(str)
+ VALUE str;
+{
+ size_t len;
+ char *p, *pend;
+ char *q, *qend;
+ VALUE result;
+
+ len = 2; /* "" */
+ p = RSTRING(str)->ptr; pend = p + RSTRING(str)->len;
+ while (p < pend) {
+ char c = *p++;
+ switch (c) {
+ case '"': case '\\':
+ case '\n': case '\r':
+ case '\t': case '\f':
+ case '\013': case '\007': case '\033':
+ len += 2;
+ break;
+
+ default:
+ if (ISPRINT(c)) {
+ len++;
+ }
+ else {
+ len += 4; /* \nnn */
+ }
+ break;
+ }
+ }
+
+ result = rb_str_new(0, len);
+ p = RSTRING(str)->ptr; pend = p + RSTRING(str)->len;
+ q = RSTRING(result)->ptr; qend = q + len;
+
+ *q++ = '"';
+ while (p < pend) {
+ char c = *p++;
+
+ if (c == '"' || c == '\\') {
+ *q++ = '\\';
+ *q++ = c;
+ }
+ else if (ISPRINT(c)) {
+ *q++ = c;
+ }
+ else if (c == '\n') {
+ *q++ = '\\';
+ *q++ = 'n';
+ }
+ else if (c == '\r') {
+ *q++ = '\\';
+ *q++ = 'r';
+ }
+ else if (c == '\t') {
+ *q++ = '\\';
+ *q++ = 't';
+ }
+ else if (c == '\f') {
+ *q++ = '\\';
+ *q++ = 'f';
+ }
+ else if (c == '\013') {
+ *q++ = '\\';
+ *q++ = 'v';
+ }
+ else if (c == '\007') {
+ *q++ = '\\';
+ *q++ = 'a';
+ }
+ else if (c == '\033') {
+ *q++ = '\\';
+ *q++ = 'e';
+ }
+ else {
+ *q++ = '\\';
+ sprintf(q, "%03o", c&0xff);
+ q += 3;
+ }
+ }
+ *q++ = '"';
+
+ return result;
}
static VALUE
-str_upcase_bang(str)
+rb_str_upcase_bang(str)
VALUE str;
{
- UCHAR *s, *send;
+ char *s, *send;
+ int modify = 0;
- str_modify(str);
+ rb_str_modify(str);
s = RSTRING(str)->ptr; send = s + RSTRING(str)->len;
while (s < send) {
- if (islower(*s)) {
+ if (ismbchar(*s)) {
+ s+=mbclen(*s);
+ }
+ else if (islower(*s)) {
*s = toupper(*s);
+ modify = 1;
}
s++;
}
- return (VALUE)str;
+ if (modify) return str;
+ return Qnil;
}
static VALUE
-str_upcase(str)
+rb_str_upcase(str)
VALUE str;
{
- return str_upcase_bang(str_dup(str));
+ VALUE val = rb_str_upcase_bang(str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-str_downcase_bang(str)
+rb_str_downcase_bang(str)
VALUE str;
{
- UCHAR *s, *send;
+ char *s, *send;
+ int modify = 0;
- str_modify(str);
+ rb_str_modify(str);
s = RSTRING(str)->ptr; send = s + RSTRING(str)->len;
while (s < send) {
- if (isupper(*s)) {
+ if (ismbchar(*s)) {
+ s+=mbclen(*s);
+ }
+ else if (ISUPPER(*s)) {
*s = tolower(*s);
+ modify = 1;
}
s++;
}
- return (VALUE)str;
+ if (modify) return str;
+ return Qnil;
}
static VALUE
-str_downcase(str)
+rb_str_downcase(str)
VALUE str;
{
- return str_downcase_bang(str_dup(str));
+ VALUE val = rb_str_downcase_bang(str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-str_capitalize_bang(str)
+rb_str_capitalize_bang(str)
VALUE str;
{
- UCHAR *s, *send;
+ char *s, *send;
+ int modify = 0;
- str_modify(str);
+ rb_str_modify(str);
s = RSTRING(str)->ptr; send = s + RSTRING(str)->len;
- if (islower(*s))
+ if (ISLOWER(*s)) {
*s = toupper(*s);
+ modify = 1;
+ }
while (++s < send) {
- if (isupper(*s)) {
+ if (ismbchar(*s)) {
+ s+=mbclen(*s);
+ }
+ else if (ISUPPER(*s)) {
*s = tolower(*s);
+ modify = 1;
}
}
- return (VALUE)str;
+ if (modify) return str;
+ return Qnil;
}
static VALUE
-str_capitalize(str)
+rb_str_capitalize(str)
VALUE str;
{
- return str_capitalize_bang(str_dup(str));
+ VALUE val = rb_str_capitalize_bang(str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-str_swapcase_bang(str)
+rb_str_swapcase_bang(str)
VALUE str;
{
- UCHAR *s, *send;
+ char *s, *send;
+ int modify = 0;
- str_modify(str);
+ rb_str_modify(str);
s = RSTRING(str)->ptr; send = s + RSTRING(str)->len;
while (s < send) {
- if (isupper(*s)) {
+ if (ismbchar(*s)) {
+ s+=mbclen(*s);
+ }
+ else if (ISUPPER(*s)) {
*s = tolower(*s);
+ modify = 1;
}
- else if (islower(*s)) {
+ else if (ISLOWER(*s)) {
*s = toupper(*s);
+ modify = 1;
}
s++;
}
- return (VALUE)str;
+ if (modify) return str;
+ return Qnil;
}
static VALUE
-str_swapcase(str)
+rb_str_swapcase(str)
VALUE str;
{
- return str_swapcase_bang(str_dup(str));
+ VALUE val = rb_str_swapcase_bang(str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
-typedef UCHAR *USTR;
+typedef unsigned char *USTR;
-static struct tr {
+struct tr {
int gen, now, max;
- UCHAR *p, *pend;
-} trsrc, trrepl;
+ char *p, *pend;
+};
static int
trnext(t)
@@ -1505,7 +1656,7 @@ trnext(t)
for (;;) {
if (!t->gen) {
if (t->p == t->pend) return -1;
- t->now = *t->p++;
+ t->now = *(USTR)t->p++;
if (t->p < t->pend && *t->p == '-') {
t->p++;
if (t->p < t->pend) {
@@ -1529,7 +1680,7 @@ trnext(t)
}
}
-static VALUE str_delete_bang();
+static VALUE rb_str_delete_bang _((VALUE,VALUE));
static VALUE
tr_trans(str, src, repl, sflag)
@@ -1538,19 +1689,21 @@ tr_trans(str, src, repl, sflag)
{
struct tr trsrc, trrepl;
int cflag = 0;
- UCHAR trans[256];
- int i, c, c0;
- UCHAR *s, *send, *t;
+ char trans[256];
+ int i, c, c0, modify = 0;
+ char *s, *send;
- Check_Type(src, T_STRING);
+ rb_str_modify(str);
+ if (TYPE(src) != T_STRING) src = rb_str_to_str(src);
trsrc.p = RSTRING(src)->ptr; trsrc.pend = trsrc.p + RSTRING(src)->len;
if (RSTRING(src)->len > 2 && RSTRING(src)->ptr[0] == '^') {
cflag++;
trsrc.p++;
}
- Check_Type(repl, T_STRING);
- if (RSTRING(repl)->len == 0) return str_delete_bang(str, src);
- trrepl.p = RSTRING(repl)->ptr; trrepl.pend = trrepl.p + RSTRING(repl)->len;
+ if (TYPE(repl) != T_STRING) repl = rb_str_to_str(repl);
+ if (RSTRING(repl)->len == 0) return rb_str_delete_bang(str, src);
+ trrepl.p = RSTRING(repl)->ptr;
+ trrepl.pend = trrepl.p + RSTRING(repl)->len;
trsrc.gen = trrepl.gen = 0;
trsrc.now = trrepl.now = 0;
trsrc.max = trrepl.max = 0;
@@ -1590,48 +1743,64 @@ tr_trans(str, src, repl, sflag)
}
}
- str_modify(str);
- t = s = RSTRING(str)->ptr; send = s + RSTRING(str)->len;
+ s = RSTRING(str)->ptr; send = s + RSTRING(str)->len;
c0 = -1;
if (sflag) {
+ char *t = s;
+
while (s < send) {
c = trans[*s++ & 0xff] & 0xff;
if (s[-1] == c || c != c0) {
c0 = (s[-1] == c)?-1:c;
+ if (*t != c) {
+ *t = c;
+ modify = 1;
+ }
*t++ = c;
}
}
+ if (RSTRING(str)->len > (t - RSTRING(str)->ptr)) {
+ RSTRING(str)->len = (t - RSTRING(str)->ptr);
+ modify = 1;
+ *t = '\0';
+ }
}
else {
while (s < send) {
- c = trans[*s++ & 0xff] & 0xff;
- *t++ = c;
+ c = trans[*s & 0xff] & 0xff;
+ if (*s != c) {
+ *s = c;
+ modify = 1;
+ }
+ s++;
}
}
- *t = '\0';
- if (sflag) RSTRING(str)->len = (t - RSTRING(str)->ptr);
- return (VALUE)str;
+ if (modify) return str;
+ return Qnil;
}
static VALUE
-str_tr_bang(str, src, repl)
+rb_str_tr_bang(str, src, repl)
VALUE str, src, repl;
{
return tr_trans(str, src, repl, 0);
}
static VALUE
-str_tr(str, src, repl)
+rb_str_tr(str, src, repl)
VALUE str, src, repl;
{
- return tr_trans(str_dup(str), src, repl, 0);
+ VALUE val = tr_trans(str = rb_str_dup(str), src, repl, 0);
+
+ if (NIL_P(val)) return str;
+ return val;
}
static void
tr_setup_table(str, table)
VALUE str;
- UCHAR table[256];
+ char table[256];
{
struct tr tr;
int i, cflag = 0;
@@ -1653,45 +1822,51 @@ tr_setup_table(str, table)
}
static VALUE
-str_delete_bang(str1, str2)
- VALUE str1, *str2;
+rb_str_delete_bang(str1, str2)
+ VALUE str1, str2;
{
- UCHAR *s, *send, *t;
- UCHAR squeez[256];
+ char *s, *send, *t;
+ char squeez[256];
+ int modify = 0;
- Check_Type(str2, T_STRING);
+ if (TYPE(str2) != T_STRING) str2 = rb_str_to_str(str2);
tr_setup_table(str2, squeez);
- str_modify(str1);
+ rb_str_modify(str1);
s = t = RSTRING(str1)->ptr;
send = s + RSTRING(str1)->len;
while (s < send) {
- if (!squeez[*s & 0xff]) {
+ if (squeez[*s & 0xff])
+ modify = 1;
+ else
*t++ = *s;
- }
s++;
}
*t = '\0';
RSTRING(str1)->len = t - RSTRING(str1)->ptr;
- return (VALUE)str1;
+ if (modify) return str1;
+ return Qnil;
}
static VALUE
-str_delete(str1, str2)
- VALUE str1, *str2;
+rb_str_delete(str1, str2)
+ VALUE str1, str2;
{
- return str_delete_bang(str_dup(str1), str2);
+ VALUE val = rb_str_delete_bang(str1 = rb_str_dup(str1), str2);
+
+ if (NIL_P(val)) return str1;
+ return val;
}
static VALUE
tr_squeeze(str1, str2)
VALUE str1, str2;
{
- UCHAR squeez[256];
- UCHAR *s, *send, *t;
- char c, save;
+ char squeez[256];
+ char *s, *send, *t;
+ char c, save, modify = 0;
if (!NIL_P(str2)) {
tr_setup_table(str2, squeez);
@@ -1704,7 +1879,7 @@ tr_squeeze(str1, str2)
}
}
- str_modify(str1);
+ rb_str_modify(str1);
s = t = RSTRING(str1)->ptr;
send = s + RSTRING(str1)->len;
@@ -1713,78 +1888,82 @@ tr_squeeze(str1, str2)
c = *s++ & 0xff;
if (c != save || !squeez[c & 0xff]) {
*t++ = save = c;
+ modify = 1;
}
}
*t = '\0';
RSTRING(str1)->len = t - RSTRING(str1)->ptr;
- return (VALUE)str1;
+ if (modify) return str1;
+ return Qnil;
}
static VALUE
-str_squeeze_bang(argc, argv, str1)
+rb_str_squeeze_bang(argc, argv, str1)
int argc;
VALUE *argv;
VALUE str1;
{
VALUE str2;
- if (rb_scan_args(argc, argv, "01", &str2) == 1) {
- Check_Type(str2, T_STRING);
+ if (rb_scan_args(argc, argv, "01", &str2) == 1 && TYPE(str2) != T_STRING) {
+ str2 = rb_str_to_str(str2);
}
return tr_squeeze(str1, str2);
}
static VALUE
-str_squeeze(argc, argv, str)
+rb_str_squeeze(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
- return str_squeeze_bang(argc, argv, str_dup(str));
+ VALUE val = rb_str_squeeze_bang(argc, argv, str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-str_tr_s_bang(str, src, repl)
+rb_str_tr_s_bang(str, src, repl)
VALUE str, src, repl;
{
- Check_Type(src, T_STRING);
- Check_Type(repl, T_STRING);
-
return tr_trans(str, src, repl, 1);
}
static VALUE
-str_tr_s(str, src, repl)
+rb_str_tr_s(str, src, repl)
VALUE str, src, repl;
{
- return str_tr_s_bang(str_dup(str), src, repl);
+ VALUE val = tr_trans(str = rb_str_dup(str), src, repl, 1);
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-str_split_method(argc, argv, str)
+rb_str_split_method(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
- extern VALUE FS;
VALUE spat;
VALUE limit;
- char char_sep = 0;
- int beg, end, lim, i;
+ int char_sep = -1;
+ int beg, end, i;
+ int lim = 0;
VALUE result, tmp;
- rb_scan_args(argc, argv, "02", &spat, &limit);
- if (!NIL_P(limit)) {
+ if (rb_scan_args(argc, argv, "02", &spat, &limit) == 2) {
lim = NUM2INT(limit);
- if (lim == 0) limit = Qnil;
- else if (lim == 1) return ary_new3(1, str);
+ if (lim <= 0) limit = Qnil;
+ else if (lim == 1) return rb_ary_new3(1, str);
i = 1;
}
- if (NIL_P(spat)) {
- if (!NIL_P(FS)) {
- spat = FS;
+ if (argc == 0) {
+ if (!NIL_P(rb_fs)) {
+ spat = rb_fs;
goto fs_set;
}
char_sep = ' ';
@@ -1793,33 +1972,33 @@ str_split_method(argc, argv, str)
switch (TYPE(spat)) {
case T_STRING:
fs_set:
- if (STRLEN(spat) == 1) {
- char_sep = RSTRING(spat)->ptr[0];
+ if (RSTRING(spat)->len == 1) {
+ char_sep = (unsigned char)RSTRING(spat)->ptr[0];
}
else {
- spat = reg_regcomp(spat);
+ spat = rb_reg_regcomp(spat);
}
break;
case T_REGEXP:
break;
default:
- ArgError("split(): bad separator");
+ rb_raise(rb_eArgError, "bad separator");
}
}
- result = ary_new();
+ result = rb_ary_new();
beg = 0;
- if (char_sep != 0) {
- UCHAR *ptr = RSTRING(str)->ptr;
- int len = RSTRING(str)->len;
- UCHAR *eptr = ptr + len;
+ if (char_sep >= 0) {
+ char *ptr = RSTRING(str)->ptr;
+ size_t len = RSTRING(str)->len;
+ char *eptr = ptr + len;
if (char_sep == ' ') { /* AWK emulation */
int skip = 1;
for (end = beg = 0; ptr<eptr; ptr++) {
if (skip) {
- if (isspace(*ptr)) {
+ if (ISSPACE(*ptr)) {
beg++;
}
else {
@@ -1828,8 +2007,8 @@ str_split_method(argc, argv, str)
}
}
else {
- if (isspace(*ptr)) {
- ary_push(result, str_substr(str, beg, end-beg));
+ if (ISSPACE(*ptr)) {
+ rb_ary_push(result, rb_str_substr(str, beg, end-beg));
skip = 1;
beg = end + 1;
if (!NIL_P(limit) && lim <= ++i) break;
@@ -1842,8 +2021,8 @@ str_split_method(argc, argv, str)
}
else {
for (end = beg = 0; ptr<eptr; ptr++) {
- if (*ptr == char_sep) {
- ary_push(result, str_substr(str, beg, end-beg));
+ if (*ptr == (char)char_sep) {
+ rb_ary_push(result, rb_str_substr(str, beg, end-beg));
beg = end + 1;
if (!NIL_P(limit) && lim <= ++i) break;
}
@@ -1852,29 +2031,26 @@ str_split_method(argc, argv, str)
}
}
else {
- int start = beg;
+ size_t start = beg;
int last_null = 0;
- int idx;
+ size_t idx;
struct re_registers *regs;
- while ((end = reg_search(spat, str, start, 0)) >= 0) {
- regs = RMATCH(backref_get())->regs;
+ while ((end = rb_reg_search(spat, str, start, 0)) >= 0) {
+ regs = RMATCH(rb_backref_get())->regs;
if (start == end && BEG(0) == END(0)) {
if (last_null == 1) {
- if (ismbchar(RSTRING(str)->ptr[beg]))
- ary_push(result, str_substr(str, beg, 2));
- else
- ary_push(result, str_substr(str, beg, 1));
+ rb_ary_push(result, rb_str_substr(str, beg, mbclen(RSTRING(str)->ptr[beg])));
beg = start;
}
else {
- start += ismbchar(RSTRING(str)->ptr[start])?2:1;
+ start += mbclen(RSTRING(str)->ptr[start]);
last_null = 1;
continue;
}
}
else {
- ary_push(result, str_substr(str, beg, end-beg));
+ rb_ary_push(result, rb_str_substr(str, beg, end-beg));
beg = start = END(0);
}
last_null = 0;
@@ -1882,67 +2058,69 @@ str_split_method(argc, argv, str)
for (idx=1; idx < regs->num_regs; idx++) {
if (BEG(idx) == -1) continue;
if (BEG(idx) == END(idx))
- tmp = str_new(0, 0);
+ tmp = rb_str_new(0, 0);
else
- tmp = str_subseq(str, BEG(idx), END(idx)-1);
- ary_push(result, tmp);
+ tmp = rb_str_subseq(str, BEG(idx), END(idx)-1);
+ rb_ary_push(result, tmp);
}
if (!NIL_P(limit) && lim <= ++i) break;
}
}
- if (RSTRING(str)->len > beg) {
- ary_push(result, str_subseq(str, beg, -1));
+ if (!NIL_P(limit) || RSTRING(str)->len > beg || lim < 0) {
+ rb_ary_push(result, rb_str_subseq(str, beg, -1));
+ }
+ if (NIL_P(limit) && lim == 0) {
+ while (RARRAY(result)->len > 0 &&
+ RSTRING(RARRAY(result)->ptr[RARRAY(result)->len-1])->len == 0)
+ rb_ary_pop(result);
}
return result;
}
VALUE
-str_split(str, sep0)
+rb_str_split(str, sep0)
VALUE str;
char *sep0;
{
VALUE sep;
- Check_Type(str, T_STRING);
- sep = str_new2(sep0);
- return str_split_method(1, &sep, str);
+ if (TYPE(str) != T_STRING) str = rb_str_to_str(str);
+ sep = rb_str_new2(sep0);
+ return rb_str_split_method(1, &sep, str);
}
static VALUE
-f_split(argc, argv)
+rb_f_split(argc, argv)
int argc;
VALUE *argv;
{
- return str_split_method(argc, argv, uscore_get());
+ return rb_str_split_method(argc, argv, uscore_get());
}
static VALUE
-str_each_line(argc, argv, str)
+rb_str_each_line(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
- extern VALUE RS;
VALUE rs;
int newline;
int rslen;
- UCHAR *p = RSTRING(str)->ptr, *pend = p + RSTRING(str)->len, *s;
- UCHAR *ptr = p;
- int len = RSTRING(str)->len;
+ char *p = RSTRING(str)->ptr, *pend = p + RSTRING(str)->len, *s;
+ char *ptr = p;
+ size_t len = RSTRING(str)->len;
VALUE line;
- if (rb_scan_args(argc, argv, "01", &rs) == 1) {
- if (!NIL_P(rs)) Check_Type(rs, T_STRING);
- }
- else {
- rs = RS;
+ if (rb_scan_args(argc, argv, "01", &rs) == 0) {
+ rs = rb_rs;
}
if (NIL_P(rs)) {
rb_yield(str);
return Qnil;
}
+ if (TYPE(rs) != T_STRING) rs = rb_str_to_str(rs);
rslen = RSTRING(rs)->len;
if (rslen == 0) {
@@ -1954,25 +2132,22 @@ str_each_line(argc, argv, str)
for (s = p, p += rslen; p < pend; p++) {
if (rslen == 0 && *p == '\n') {
- if (*(p+1) != '\n') continue;
+ if (p[1] != '\n') continue;
while (*p == '\n') p++;
- p--;
}
- if (*p == newline &&
+ if (p[-1] == newline &&
(rslen <= 1 ||
- memcmp(RSTRING(rs)->ptr, p-rslen+1, rslen) == 0)) {
- line = str_new(s, p - s + 1);
- lastline_set(line);
+ memcmp(RSTRING(rs)->ptr, p-rslen, rslen) == 0)) {
+ line = rb_str_new(s, p - s);
rb_yield(line);
if (RSTRING(str)->ptr != ptr || RSTRING(str)->len != len)
- Fail("string modified");
- s = p + 1;
+ rb_raise(rb_eArgError, "string modified");
+ s = p;
}
}
if (s != pend) {
- line = str_new(s, p - s);
- lastline_set(line);
+ line = rb_str_new(s, p - s);
rb_yield(line);
}
@@ -1980,10 +2155,10 @@ str_each_line(argc, argv, str)
}
static VALUE
-str_each_byte(str)
+rb_str_each_byte(str)
struct RString* str;
{
- int i;
+ size_t i;
for (i=0; i<RSTRING(str)->len; i++) {
rb_yield(INT2FIX(RSTRING(str)->ptr[i] & 0xff));
@@ -1992,12 +2167,11 @@ str_each_byte(str)
}
static VALUE
-str_chop_bang(str)
+rb_str_chop_bang(str)
VALUE str;
{
- str_modify(str);
-
if (RSTRING(str)->len > 0) {
+ rb_str_modify(str);
RSTRING(str)->len--;
if (RSTRING(str)->ptr[RSTRING(str)->len] == '\n') {
if (RSTRING(str)->len > 0 &&
@@ -2006,90 +2180,132 @@ str_chop_bang(str)
}
}
RSTRING(str)->ptr[RSTRING(str)->len] = '\0';
+ return str;
}
-
- return str;
+ return Qnil;
}
static VALUE
-str_chop(str)
+rb_str_chop(str)
VALUE str;
{
- return str_chop_bang(str_dup(str));
+ VALUE val = rb_str_chop_bang(str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-f_chop_bang(str)
+rb_f_chop_bang(str)
VALUE str;
{
- return str_chop_bang(uscore_get());
+ return rb_str_chop_bang(uscore_get());
}
static VALUE
-f_chop()
+rb_f_chop()
{
- return str_chop_bang(str_dup(uscore_get()));
+ VALUE str = rb_str_dup(uscore_get());
+ VALUE val = rb_str_chop_bang(str);
+
+ if (NIL_P(str)) return str;
+ rb_lastline_set(val);
+ return val;
}
static VALUE
-str_chomp_bang(str)
+rb_str_chomp_bang(argc, argv, str)
+ int argc;
+ VALUE *argv;
VALUE str;
{
- str_modify(str);
+ VALUE rs;
+ int newline;
+ int rslen;
+ char *p = RSTRING(str)->ptr;
+ size_t len = RSTRING(str)->len;
- if (RSTRING(str)->len > 0 &&
- RSTRING(str)->ptr[RSTRING(str)->len-1] == '\n') {
- RSTRING(str)->len--;
- if (RSTRING(str)->len > 0 &&
- RSTRING(str)->ptr[RSTRING(str)->len] == '\r') {
- RSTRING(str)->len--;
+ if (rb_scan_args(argc, argv, "01", &rs) == 0) {
+ rs = rb_rs;
+ }
+ if (NIL_P(rs)) return Qnil;
+
+ if (TYPE(rs) != T_STRING) rs = rb_str_to_str(rs);
+ rslen = RSTRING(rs)->len;
+ if (rslen == 0) {
+ while (len>0 && p[len-1] == '\n') {
+ len--;
+ }
+ if (len < RSTRING(str)->len) {
+ RSTRING(str)->len = len;
+ RSTRING(str)->ptr[len] = '\0';
+ return str;
}
+ return Qnil;
+ }
+ if (rslen > len) return Qnil;
+ newline = RSTRING(rs)->ptr[rslen-1];
+
+ if (p[len-1] == newline &&
+ (rslen <= 1 ||
+ memcmp(RSTRING(rs)->ptr, p+len-rslen, rslen) == 0)) {
+ RSTRING(str)->len -= rslen;
RSTRING(str)->ptr[RSTRING(str)->len] = '\0';
+ return str;
}
- return str;
+ return Qnil;
}
static VALUE
-str_chomp(str)
+rb_str_chomp(argc, argv, str)
+ int argc;
+ VALUE *argv;
VALUE str;
{
- return str_chomp_bang(str_dup(str));
+ VALUE val = rb_str_chomp_bang(argc, argv, str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
-f_chomp_bang(str)
- VALUE str;
+rb_f_chomp_bang(argc, argv)
+ int argc;
+ VALUE *argv;
{
- return str_chomp_bang(uscore_get());
+ return rb_str_chomp_bang(argc, argv, uscore_get());
}
static VALUE
-f_chomp()
+rb_f_chomp(argc, argv)
+ int argc;
+ VALUE *argv;
{
- return str_chomp_bang(str_dup(uscore_get()));
+ VALUE str = rb_str_dup(uscore_get());
+ VALUE val = rb_str_chomp_bang(argc, argv, str);
+ return val;
}
static VALUE
-str_strip_bang(str)
+rb_str_strip_bang(str)
VALUE str;
{
- UCHAR *s, *t, *e;
-
- str_modify(str);
+ char *s, *t, *e;
+ rb_str_modify(str);
s = RSTRING(str)->ptr;
e = t = s + RSTRING(str)->len;
/* remove spaces at head */
- while (s < t && isspace(*s)) s++;
+ while (s < t && ISSPACE(*s)) s++;
/* remove trailing spaces */
t--;
- while (s <= t && isspace(*t)) t--;
+ while (s <= t && ISSPACE(*t)) t--;
t++;
RSTRING(str)->len = t-s;
if (s > RSTRING(str)->ptr) {
- UCHAR *p = RSTRING(str)->ptr;
+ char *p = RSTRING(str)->ptr;
RSTRING(str)->ptr = ALLOC_N(char, RSTRING(str)->len+1);
memcpy(RSTRING(str)->ptr, s, RSTRING(str)->len);
@@ -2099,38 +2315,50 @@ str_strip_bang(str)
else if (t < e) {
RSTRING(str)->ptr[RSTRING(str)->len] = '\0';
}
+ else {
+ return Qnil;
+ }
- return (VALUE)str;
+ return str;
+}
+
+static VALUE
+rb_str_strip(str)
+ VALUE str;
+{
+ VALUE val = rb_str_strip_bang(str = rb_str_dup(str));
+
+ if (NIL_P(val)) return str;
+ return val;
}
static VALUE
scan_once(str, pat, start)
VALUE str, pat;
- int *start;
+ size_t *start;
{
- VALUE result;
+ VALUE result, match;
struct re_registers *regs;
- int i;
+ size_t i;
- if (reg_search(pat, str, *start, 0) >= 0) {
- regs = RMATCH(backref_get())->regs;
- if (END(0) == *start) {
- *start = END(0)+1;
+ if (rb_reg_search(pat, str, *start, 0) >= 0) {
+ match = rb_backref_get();
+ regs = RMATCH(match)->regs;
+ if (BEG(0) == END(0)) {
+ /*
+ * Always consume at least one character of the input string
+ */
+ *start = END(0)+mbclen(RSTRING(str)->ptr[END(0)]);
}
else {
*start = END(0);
}
if (regs->num_regs == 1) {
- return str_substr(str, BEG(0), END(0)-BEG(0));
+ return rb_reg_nth_match(0, match);
}
- result = ary_new2(regs->num_regs);
+ result = rb_ary_new2(regs->num_regs);
for (i=1; i < regs->num_regs; i++) {
- if (BEG(i) == -1) {
- ary_push(result, Qnil);
- }
- else {
- ary_push(result, str_substr(str, BEG(i), END(i)-BEG(i)));
- }
+ rb_ary_push(result, rb_reg_nth_match(i, match));
}
return result;
@@ -2139,27 +2367,18 @@ scan_once(str, pat, start)
}
static VALUE
-str_scan(str, pat)
+rb_str_scan(str, pat)
VALUE str, pat;
{
VALUE result;
- int start = 0;
-
- switch (TYPE(pat)) {
- case T_STRING:
- pat = reg_regcomp(pat);
- break;
- case T_REGEXP:
- break;
- default:
- Check_Type(pat, T_REGEXP);
- }
+ size_t start = 0;
- if (!iterator_p()) {
- VALUE ary = ary_new();
+ pat = get_pat(pat);
+ if (!rb_iterator_p()) {
+ VALUE ary = rb_ary_new();
while (!NIL_P(result = scan_once(str, pat, &start))) {
- ary_push(ary, result);
+ rb_ary_push(ary, result);
}
return ary;
}
@@ -2171,65 +2390,66 @@ str_scan(str, pat)
}
static VALUE
-str_strip(str)
+rb_str_hex(str)
VALUE str;
{
- return str_strip_bang(str_dup(str));
+ return rb_str2inum(RSTRING(str)->ptr, 16);
}
static VALUE
-str_hex(str)
+rb_str_oct(str)
VALUE str;
{
- return str2inum(RSTRING(str)->ptr, 16);
-}
+ int base = 8;
-static VALUE
-str_oct(str)
- VALUE str;
-{
- return str2inum(RSTRING(str)->ptr, 8);
+ if (RSTRING(str)->len > 2 && RSTRING(str)->ptr[0] == '0' &&
+ (RSTRING(str)->ptr[1] == 'x' || RSTRING(str)->ptr[1] == 'X')) {
+ base = 16;
+ }
+ return rb_str2inum(RSTRING(str)->ptr, base);
}
static VALUE
-str_crypt(str, salt)
+rb_str_crypt(str, salt)
VALUE str, salt;
{
extern char *crypt();
- salt = obj_as_string(salt);
+
+ if (TYPE(salt) != T_STRING) salt = rb_str_to_str(salt);
if (RSTRING(salt)->len < 2)
- ArgError("salt too short(need >2 bytes)");
- return str_new2(crypt(RSTRING(str)->ptr, RSTRING(salt)->ptr));
+ rb_raise(rb_eArgError, "salt too short(need >2 bytes)");
+ return rb_str_new2(crypt(RSTRING(str)->ptr, RSTRING(salt)->ptr));
}
static VALUE
-str_intern(str)
+rb_str_intern(str)
VALUE str;
{
ID id;
if (strlen(RSTRING(str)->ptr) != RSTRING(str)->len)
- ArgError("string contains `\0'");
+ rb_raise(rb_eArgError, "string contains `\\0'");
id = rb_intern(RSTRING(str)->ptr);
return INT2FIX(id);
}
static VALUE
-str_sum(argc, argv, str)
+rb_str_sum(argc, argv, str)
int argc;
VALUE *argv;
VALUE str;
{
VALUE vbits;
int bits;
- UCHAR *p, *pend;
+ char *p, *pend;
- rb_scan_args(argc, argv, "01", &vbits);
- if (NIL_P(vbits)) bits = 16;
+ if (rb_scan_args(argc, argv, "01", &vbits) == 0) {
+ bits = 16;
+ }
else bits = NUM2INT(vbits);
p = RSTRING(str)->ptr; pend = p + RSTRING(str)->len;
- if (bits > 32) {
+ if (bits > sizeof(long)*CHAR_BIT) {
VALUE res = INT2FIX(0);
VALUE mod;
@@ -2237,36 +2457,39 @@ str_sum(argc, argv, str)
mod = rb_funcall(mod, '-', 1, INT2FIX(1));
while (p < pend) {
- res = rb_funcall(res, '+', 1, INT2FIX((UINT)*p));
- res = rb_funcall(res, '%', 1, mod);
+ res = rb_funcall(res, '+', 1, INT2FIX((unsigned int)*p));
p++;
}
+ res = rb_funcall(res, '&', 1, mod);
return res;
}
else {
- UINT res = 0;
- UINT mod = (1<<bits)-1;
+ unsigned int res = 0;
+ unsigned int mod = (1<<bits)-1;
+ if (mod == 0) {
+ mod = -1;
+ }
while (p < pend) {
- res += (UINT)*p;
- res %= mod;
+ res += (unsigned int)*p;
p++;
}
- return int2inum(res);
+ res &= mod;
+ return rb_int2inum(res);
}
}
static VALUE
-str_ljust(str, w)
+rb_str_ljust(str, w)
VALUE str;
VALUE w;
{
- int width = NUM2INT(w);
+ size_t width = NUM2UINT(w);
VALUE res;
- UCHAR *p, *pend;
+ char *p, *pend;
- if (RSTRING(str)->len >= width) return (VALUE)str;
- res = str_new(0, width);
+ if (width < 0 || RSTRING(str)->len >= width) return str;
+ res = rb_str_new(0, width);
memcpy(RSTRING(res)->ptr, RSTRING(str)->ptr, RSTRING(str)->len);
p = RSTRING(res)->ptr + RSTRING(str)->len; pend = RSTRING(res)->ptr + width;
while (p < pend) {
@@ -2276,16 +2499,16 @@ str_ljust(str, w)
}
static VALUE
-str_rjust(str, w)
+rb_str_rjust(str, w)
VALUE str;
VALUE w;
{
- int width = NUM2INT(w);
+ size_t width = NUM2UINT(w);
VALUE res;
- UCHAR *p, *pend;
+ char *p, *pend;
- if (RSTRING(str)->len >= width) return (VALUE)str;
- res = str_new(0, width);
+ if (width < 0 || RSTRING(str)->len >= width) return str;
+ res = rb_str_new(0, width);
p = RSTRING(res)->ptr; pend = p + width - RSTRING(str)->len;
while (p < pend) {
*p++ = ' ';
@@ -2295,17 +2518,17 @@ str_rjust(str, w)
}
static VALUE
-str_center(str, w)
+rb_str_center(str, w)
VALUE str;
VALUE w;
{
- int width = NUM2INT(w);
+ size_t width = NUM2UINT(w);
VALUE res;
- UCHAR *p, *pend;
- int n;
+ char *p, *pend;
+ size_t n;
- if (RSTRING(str)->len >= width) return (VALUE)str;
- res = str_new(0, width);
+ if (width < 0 || RSTRING(str)->len >= width) return str;
+ res = rb_str_new(0, width);
n = (width - RSTRING(str)->len)/2;
p = RSTRING(res)->ptr; pend = p + n;
while (p < pend) {
@@ -2319,122 +2542,118 @@ str_center(str, w)
return res;
}
-extern VALUE mKernel;
-extern VALUE mComparable;
-extern VALUE mEnumerable;
-extern VALUE eGlobalExit;
-
void
Init_String()
{
- cString = rb_define_class("String", cObject);
- rb_include_module(cString, mComparable);
- rb_include_module(cString, mEnumerable);
- rb_define_singleton_method(cString, "new", str_s_new, 1);
- rb_define_method(cString, "clone", str_clone, 0);
- rb_define_method(cString, "dup", str_dup, 0);
- rb_define_method(cString, "<=>", str_cmp_method, 1);
- rb_define_method(cString, "==", str_equal, 1);
- rb_define_method(cString, "===", str_equal, 1);
- rb_define_method(cString, "eql?", str_equal, 1);
- rb_define_method(cString, "hash", str_hash_method, 0);
- rb_define_method(cString, "+", str_plus, 1);
- rb_define_method(cString, "*", str_times, 1);
- rb_define_method(cString, "%", str_format, 1);
- rb_define_method(cString, "[]", str_aref_method, -1);
- rb_define_method(cString, "[]=", str_aset_method, -1);
- rb_define_method(cString, "length", str_length, 0);
- rb_define_alias(cString, "size", "length");
- rb_define_method(cString, "=~", str_match, 1);
- rb_define_method(cString, "~", str_match2, 0);
- rb_define_method(cString, "succ", str_succ, 0);
- rb_define_method(cString, "upto", str_upto, 1);
- rb_define_method(cString, "index", str_index_method, -1);
- rb_define_method(cString, "rindex", str_rindex, -1);
-
- rb_define_method(cString, "freeze", str_freeze, 0);
- rb_define_method(cString, "frozen?", str_frozen_p, 0);
-
- rb_define_method(cString, "taint", str_taint, 0);
- rb_define_method(cString, "tainted?", str_tainted, 0);
-
- rb_define_method(cString, "to_i", str_to_i, 0);
- rb_define_method(cString, "to_f", str_to_f, 0);
- rb_define_method(cString, "to_s", str_to_s, 0);
- rb_define_method(cString, "inspect", str_inspect, 0);
-
- rb_define_method(cString, "upcase", str_upcase, 0);
- rb_define_method(cString, "downcase", str_downcase, 0);
- rb_define_method(cString, "capitalize", str_capitalize, 0);
- rb_define_method(cString, "swapcase", str_swapcase, 0);
-
- rb_define_method(cString, "upcase!", str_upcase_bang, 0);
- rb_define_method(cString, "downcase!", str_downcase_bang, 0);
- rb_define_method(cString, "capitalize!", str_capitalize_bang, 0);
- rb_define_method(cString, "swapcase!", str_swapcase_bang, 0);
-
- rb_define_method(cString, "hex", str_hex, 0);
- rb_define_method(cString, "oct", str_oct, 0);
- rb_define_method(cString, "split", str_split_method, -1);
- rb_define_method(cString, "reverse", str_reverse, 0);
- rb_define_method(cString, "reverse!", str_reverse_bang, 0);
- rb_define_method(cString, "concat", str_concat, 1);
- rb_define_method(cString, "<<", str_concat, 1);
- rb_define_method(cString, "crypt", str_crypt, 1);
- rb_define_method(cString, "intern", str_intern, 0);
-
- rb_define_method(cString, "include?", str_include, 1);
-
- rb_define_method(cString, "scan", str_scan, 1);
-
- rb_define_method(cString, "ljust", str_ljust, 1);
- rb_define_method(cString, "rjust", str_rjust, 1);
- rb_define_method(cString, "center", str_center, 1);
-
- rb_define_method(cString, "sub", str_sub, -1);
- rb_define_method(cString, "gsub", str_gsub, -1);
- rb_define_method(cString, "chop", str_chop, 0);
- rb_define_method(cString, "chomp", str_chomp, 0);
- rb_define_method(cString, "strip", str_strip, 0);
-
- rb_define_method(cString, "sub!", str_sub_bang, -1);
- rb_define_method(cString, "gsub!", str_gsub_bang, -1);
- rb_define_method(cString, "strip!", str_strip_bang, 0);
- rb_define_method(cString, "chop!", str_chop_bang, 0);
- rb_define_method(cString, "chomp!", str_chomp_bang, 0);
-
- rb_define_method(cString, "tr", str_tr, 2);
- rb_define_method(cString, "tr_s", str_tr_s, 2);
- rb_define_method(cString, "delete", str_delete, 1);
- rb_define_method(cString, "squeeze", str_squeeze, -1);
-
- rb_define_method(cString, "tr!", str_tr_bang, 2);
- rb_define_method(cString, "tr_s!", str_tr_s_bang, 2);
- rb_define_method(cString, "delete!", str_delete_bang, 1);
- rb_define_method(cString, "squeeze!", str_squeeze_bang, -1);
-
- rb_define_method(cString, "each_line", str_each_line, -1);
- rb_define_method(cString, "each", str_each_line, -1);
- rb_define_method(cString, "each_byte", str_each_byte, 0);
-
- rb_define_method(cString, "sum", str_sum, -1);
-
- rb_define_global_function("sub", f_sub, -1);
- rb_define_global_function("gsub", f_gsub, -1);
-
- rb_define_global_function("sub!", f_sub_bang, -1);
- rb_define_global_function("gsub!", f_gsub_bang, -1);
-
- rb_define_global_function("chop", f_chop, 0);
- rb_define_global_function("chop!", f_chop_bang, 0);
-
- rb_define_global_function("chomp", f_chomp, 0);
- rb_define_global_function("chomp!", f_chomp_bang, 0);
-
- rb_define_global_function("split", f_split, -1);
-
- pr_str = rb_intern("to_s");
-
- /* Fix-up initialize ordering */
- RCLASS(eGlobalExit)->super = cString;
+ rb_cString = rb_define_class("String", rb_cObject);
+ rb_include_module(rb_cString, rb_mComparable);
+ rb_include_module(rb_cString, rb_mEnumerable);
+ rb_define_singleton_method(rb_cString, "new", rb_str_s_new, 1);
+ rb_define_method(rb_cString, "clone", rb_str_clone, 0);
+ rb_define_method(rb_cString, "dup", rb_str_dup, 0);
+ rb_define_method(rb_cString, "<=>", rb_str_cmp_method, 1);
+ rb_define_method(rb_cString, "==", rb_str_equal, 1);
+ rb_define_method(rb_cString, "===", rb_str_equal, 1);
+ rb_define_method(rb_cString, "eql?", rb_str_equal, 1);
+ rb_define_method(rb_cString, "hash", rb_str_hash_method, 0);
+ rb_define_method(rb_cString, "+", rb_str_plus, 1);
+ rb_define_method(rb_cString, "*", rb_str_times, 1);
+ rb_define_method(rb_cString, "%", rb_str_format, 1);
+ rb_define_method(rb_cString, "[]", rb_str_aref_method, -1);
+ rb_define_method(rb_cString, "[]=", rb_str_aset_method, -1);
+ rb_define_method(rb_cString, "length", rb_str_length, 0);
+ rb_define_alias(rb_cString, "size", "length");
+ rb_define_method(rb_cString, "empty?", rb_str_empty, 0);
+ rb_define_method(rb_cString, "=~", rb_str_match, 1);
+ rb_define_method(rb_cString, "~", rb_str_match2, 0);
+ rb_define_method(rb_cString, "succ", rb_str_succ, 0);
+ rb_define_method(rb_cString, "succ!", rb_str_succ_bang, 0);
+ rb_define_method(rb_cString, "next", rb_str_succ, 0);
+ rb_define_method(rb_cString, "next!", rb_str_succ_bang, 0);
+ rb_define_method(rb_cString, "upto", rb_str_upto, 1);
+ rb_define_method(rb_cString, "index", rb_str_index_method, -1);
+ rb_define_method(rb_cString, "rindex", rb_str_rindex, -1);
+ rb_define_method(rb_cString, "replace", rb_str_replace_method, 1);
+
+ rb_define_method(rb_cString, "freeze", rb_str_freeze, 0);
+ rb_define_method(rb_cString, "frozen?", rb_str_frozen_p, 0);
+
+ rb_define_method(rb_cString, "to_i", rb_str_to_i, 0);
+ rb_define_method(rb_cString, "to_f", rb_str_to_f, 0);
+ rb_define_method(rb_cString, "to_s", rb_str_to_s, 0);
+ rb_define_method(rb_cString, "to_str", rb_str_to_s, 0);
+ rb_define_method(rb_cString, "inspect", rb_str_inspect, 0);
+ rb_define_method(rb_cString, "dump", rb_str_dump, 0);
+
+ rb_define_method(rb_cString, "upcase", rb_str_upcase, 0);
+ rb_define_method(rb_cString, "downcase", rb_str_downcase, 0);
+ rb_define_method(rb_cString, "capitalize", rb_str_capitalize, 0);
+ rb_define_method(rb_cString, "swapcase", rb_str_swapcase, 0);
+
+ rb_define_method(rb_cString, "upcase!", rb_str_upcase_bang, 0);
+ rb_define_method(rb_cString, "downcase!", rb_str_downcase_bang, 0);
+ rb_define_method(rb_cString, "capitalize!", rb_str_capitalize_bang, 0);
+ rb_define_method(rb_cString, "swapcase!", rb_str_swapcase_bang, 0);
+
+ rb_define_method(rb_cString, "hex", rb_str_hex, 0);
+ rb_define_method(rb_cString, "oct", rb_str_oct, 0);
+ rb_define_method(rb_cString, "split", rb_str_split_method, -1);
+ rb_define_method(rb_cString, "reverse", rb_str_reverse, 0);
+ rb_define_method(rb_cString, "reverse!", rb_str_reverse_bang, 0);
+ rb_define_method(rb_cString, "concat", rb_str_concat, 1);
+ rb_define_method(rb_cString, "<<", rb_str_concat, 1);
+ rb_define_method(rb_cString, "crypt", rb_str_crypt, 1);
+ rb_define_method(rb_cString, "intern", rb_str_intern, 0);
+
+ rb_define_method(rb_cString, "include?", rb_str_include, 1);
+
+ rb_define_method(rb_cString, "scan", rb_str_scan, 1);
+
+ rb_define_method(rb_cString, "ljust", rb_str_ljust, 1);
+ rb_define_method(rb_cString, "rjust", rb_str_rjust, 1);
+ rb_define_method(rb_cString, "center", rb_str_center, 1);
+
+ rb_define_method(rb_cString, "sub", rb_str_sub, -1);
+ rb_define_method(rb_cString, "gsub", rb_str_gsub, -1);
+ rb_define_method(rb_cString, "chop", rb_str_chop, 0);
+ rb_define_method(rb_cString, "chomp", rb_str_chomp, -1);
+ rb_define_method(rb_cString, "strip", rb_str_strip, 0);
+
+ rb_define_method(rb_cString, "sub!", rb_str_sub_bang, -1);
+ rb_define_method(rb_cString, "gsub!", rb_str_gsub_bang, -1);
+ rb_define_method(rb_cString, "strip!", rb_str_strip_bang, 0);
+ rb_define_method(rb_cString, "chop!", rb_str_chop_bang, 0);
+ rb_define_method(rb_cString, "chomp!", rb_str_chomp_bang, -1);
+
+ rb_define_method(rb_cString, "tr", rb_str_tr, 2);
+ rb_define_method(rb_cString, "tr_s", rb_str_tr_s, 2);
+ rb_define_method(rb_cString, "delete", rb_str_delete, 1);
+ rb_define_method(rb_cString, "squeeze", rb_str_squeeze, -1);
+
+ rb_define_method(rb_cString, "tr!", rb_str_tr_bang, 2);
+ rb_define_method(rb_cString, "tr_s!", rb_str_tr_s_bang, 2);
+ rb_define_method(rb_cString, "delete!", rb_str_delete_bang, 1);
+ rb_define_method(rb_cString, "squeeze!", rb_str_squeeze_bang, -1);
+
+ rb_define_method(rb_cString, "each_line", rb_str_each_line, -1);
+ rb_define_method(rb_cString, "each", rb_str_each_line, -1);
+ rb_define_method(rb_cString, "each_byte", rb_str_each_byte, 0);
+
+ rb_define_method(rb_cString, "sum", rb_str_sum, -1);
+
+ rb_define_global_function("sub", rb_f_sub, -1);
+ rb_define_global_function("gsub", rb_f_gsub, -1);
+
+ rb_define_global_function("sub!", rb_f_sub_bang, -1);
+ rb_define_global_function("gsub!", rb_f_gsub_bang, -1);
+
+ rb_define_global_function("chop", rb_f_chop, 0);
+ rb_define_global_function("chop!", rb_f_chop_bang, 0);
+
+ rb_define_global_function("chomp", rb_f_chomp, -1);
+ rb_define_global_function("chomp!", rb_f_chomp_bang, -1);
+
+ rb_define_global_function("split", rb_f_split, -1);
+
+ to_str = rb_intern("to_s");
}