* include/ruby/ruby.h (ExportStringValue): new macro to convert

string in internal encoding to external to export.

* string.c (rb_str_export): new function to do conversion to
  external encoding.

* ext/sdbm/init.c: encoding conversion support.

* ext/dbm/dbm.c: ditto.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19855 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,15 @@
+Mon Oct 20 16:48:43 2008  Yukihiro Matsumoto  <matz@ruby-lang.org>
+
+	* include/ruby/ruby.h (ExportStringValue): new macro to convert
+	  string in internal encoding to external to export.
+
+	* string.c (rb_str_export): new function to do conversion to
+	  external encoding.
+
+	* ext/sdbm/init.c: encoding conversion support.
+
+	* ext/dbm/dbm.c: ditto.
+
 Mon Oct 20 15:42:02 2008  Yukihiro Matsumoto  <matz@ruby-lang.org>
 
 	* string.c (rb_locale_str_new): new function to convert string

ext/dbm/dbm.c

@@ -109,7 +109,7 @@ fdbm_initialize(int argc, VALUE *argv, VALUE obj)
     if (!NIL_P(vflags))
         flags = NUM2INT(vflags);
 
-    SafeStringValue(file);
+    FilePathValue(file);
 
     if (flags & RUBY_DBM_RW_BIT) {
         flags &= ~RUBY_DBM_RW_BIT;
@@ -164,7 +164,7 @@ fdbm_fetch(VALUE obj, VALUE keystr, VALUE ifnone)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(keystr);
+    ExportStringValue(keystr);
     key.dptr = RSTRING_PTR(keystr);
     key.dsize = RSTRING_LEN(keystr);
 
@@ -204,7 +204,7 @@ fdbm_index(VALUE obj, VALUE valstr)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(valstr);
+    ExportStringValue(valstr);
     val.dptr = RSTRING_PTR(valstr);
     val.dsize = RSTRING_LEN(valstr);
 
@@ -272,7 +272,7 @@ fdbm_delete(VALUE obj, VALUE keystr)
     VALUE valstr;
 
     fdbm_modify(obj);
-    StringValue(keystr);
+    ExportStringValue(keystr);
     key.dptr = RSTRING_PTR(keystr);
     key.dsize = RSTRING_LEN(keystr);
 
@@ -346,7 +346,7 @@ fdbm_delete_if(VALUE obj)
 
     for (i = 0; i < RARRAY_LEN(ary); i++) {
 	keystr = RARRAY_PTR(ary)[i];
-	StringValue(keystr);
+	ExportStringValue(keystr);
 	key.dptr = RSTRING_PTR(keystr);
 	key.dsize = RSTRING_LEN(keystr);
 	if (dbm_delete(dbm, key)) {
@@ -599,7 +599,7 @@ fdbm_has_key(VALUE obj, VALUE keystr)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(keystr);
+    ExportStringValue(keystr);
     key.dptr = RSTRING_PTR(keystr);
     key.dsize = RSTRING_LEN(keystr);
 
@@ -616,7 +616,7 @@ fdbm_has_value(VALUE obj, VALUE valstr)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(valstr);
+    ExportStringValue(valstr);
     val.dptr = RSTRING_PTR(valstr);
     val.dsize = RSTRING_LEN(valstr);
 

ext/sdbm/init.c

@@ -96,7 +96,7 @@ fsdbm_initialize(int argc, VALUE *argv, VALUE obj)
     else {
 	mode = NUM2INT(vmode);
     }
-    SafeStringValue(file);
+    FilePathValue(file);
 
     dbm = 0;
     if (mode >= 0)
@@ -142,7 +142,7 @@ fsdbm_fetch(VALUE obj, VALUE keystr, VALUE ifnone)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(keystr);
+    ExportStringValue(keystr);
     key.dptr = RSTRING_PTR(keystr);
     key.dsize = RSTRING_LEN(keystr);
 
@@ -150,10 +150,10 @@ fsdbm_fetch(VALUE obj, VALUE keystr, VALUE ifnone)
     value = sdbm_fetch(dbm, key);
     if (value.dptr == 0) {
 	if (ifnone == Qnil && rb_block_given_p())
-	    return rb_yield(rb_tainted_str_new(key.dptr, key.dsize));
+	    return rb_yield(rb_external_str_new(key.dptr, key.dsize));
 	return ifnone;
     }
-    return rb_tainted_str_new(value.dptr, value.dsize);
+    return rb_external_str_new(value.dptr, value.dsize);
 }
 
 static VALUE
@@ -182,7 +182,7 @@ fsdbm_index(VALUE obj, VALUE valstr)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(valstr);
+    ExportStringValue(valstr);
     val.dptr = RSTRING_PTR(valstr);
     val.dsize = RSTRING_LEN(valstr);
 
@@ -191,7 +191,7 @@ fsdbm_index(VALUE obj, VALUE valstr)
 	val = sdbm_fetch(dbm, key);
 	if (val.dsize == RSTRING_LEN(valstr) &&
 	    memcmp(val.dptr, RSTRING_PTR(valstr), val.dsize) == 0)
-	    return rb_tainted_str_new(key.dptr, key.dsize);
+	    return rb_external_str_new(key.dptr, key.dsize);
     }
     return Qnil;
 }
@@ -208,8 +208,8 @@ fsdbm_select(VALUE obj)
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	VALUE assoc, v;
 	val = sdbm_fetch(dbm, key);
-	assoc = rb_assoc_new(rb_tainted_str_new(key.dptr, key.dsize),
+	assoc = rb_assoc_new(rb_external_str_new(key.dptr, key.dsize),
-			     rb_tainted_str_new(val.dptr, val.dsize));
+			     rb_external_str_new(val.dptr, val.dsize));
 	v = rb_yield(assoc);
 	if (RTEST(v)) {
 	    rb_ary_push(new, assoc);
@@ -249,7 +249,7 @@ fsdbm_delete(VALUE obj, VALUE keystr)
     VALUE valstr;
 
     fdbm_modify(obj);
-    StringValue(keystr);
+    ExportStringValue(keystr);
     key.dptr = RSTRING_PTR(keystr);
     key.dsize = RSTRING_LEN(keystr);
 
@@ -263,7 +263,7 @@ fsdbm_delete(VALUE obj, VALUE keystr)
     }
 
     /* need to save value before sdbm_delete() */
-    valstr = rb_tainted_str_new(value.dptr, value.dsize);
+    valstr = rb_external_str_new(value.dptr, value.dsize);
 
     if (sdbm_delete(dbm, key)) {
 	dbmp->di_size = -1;
@@ -288,8 +288,8 @@ fsdbm_shift(VALUE obj)
     key = sdbm_firstkey(dbm); 
     if (!key.dptr) return Qnil;
     val = sdbm_fetch(dbm, key);
-    keystr = rb_tainted_str_new(key.dptr, key.dsize);
+    keystr = rb_external_str_new(key.dptr, key.dsize);
-    valstr = rb_tainted_str_new(val.dptr, val.dsize);
+    valstr = rb_external_str_new(val.dptr, val.dsize);
     sdbm_delete(dbm, key);
     if (dbmp->di_size >= 0) {
 	dbmp->di_size--;
@@ -314,8 +314,8 @@ fsdbm_delete_if(VALUE obj)
     dbmp->di_size = -1;
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	val = sdbm_fetch(dbm, key);
-	keystr = rb_tainted_str_new(key.dptr, key.dsize);
+	keystr = rb_external_str_new(key.dptr, key.dsize);
-	valstr = rb_tainted_str_new(val.dptr, val.dsize);
+	valstr = rb_external_str_new(val.dptr, val.dsize);
         ret = rb_protect(rb_yield, rb_assoc_new(rb_str_dup(keystr), valstr), &status);
         if (status != 0) break;
 	if (RTEST(ret)) rb_ary_push(ary, keystr);
@@ -324,7 +324,7 @@ fsdbm_delete_if(VALUE obj)
 
     for (i = 0; i < RARRAY_LEN(ary); i++) {
 	keystr = RARRAY_PTR(ary)[i];
-	StringValue(keystr);
+	ExportStringValue(keystr);
 	key.dptr = RSTRING_PTR(keystr);
 	key.dsize = RSTRING_LEN(keystr);
 	if (sdbm_delete(dbm, key)) {
@@ -369,8 +369,8 @@ fsdbm_invert(VALUE obj)
     GetDBM2(obj, dbmp, dbm);
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	val = sdbm_fetch(dbm, key);
-	keystr = rb_tainted_str_new(key.dptr, key.dsize);
+	keystr = rb_external_str_new(key.dptr, key.dsize);
-	valstr = rb_tainted_str_new(val.dptr, val.dsize);
+	valstr = rb_external_str_new(val.dptr, val.dsize);
 	rb_hash_aset(hash, valstr, keystr);
     }
     return hash;
@@ -389,8 +389,8 @@ fsdbm_store(VALUE obj, VALUE keystr, VALUE valstr)
     }
 
     fdbm_modify(obj);
-    StringValue(keystr);
+    ExportStringValue(keystr);
-    StringValue(valstr);
+    ExportStringValue(valstr);
 
     key.dptr = RSTRING_PTR(keystr);
     key.dsize = RSTRING_LEN(keystr);
@@ -491,7 +491,7 @@ fsdbm_each_value(VALUE obj)
     GetDBM2(obj, dbmp, dbm);
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	val = sdbm_fetch(dbm, key);
-	rb_yield(rb_tainted_str_new(val.dptr, val.dsize));
+	rb_yield(rb_external_str_new(val.dptr, val.dsize));
 	GetDBM2(obj, dbmp, dbm);
     }
     return obj;
@@ -508,7 +508,7 @@ fsdbm_each_key(VALUE obj)
 
     GetDBM2(obj, dbmp, dbm);
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
-	rb_yield(rb_tainted_str_new(key.dptr, key.dsize));
+	rb_yield(rb_external_str_new(key.dptr, key.dsize));
 	GetDBM2(obj, dbmp, dbm);
     }
     return obj;
@@ -527,8 +527,8 @@ fsdbm_each_pair(VALUE obj)
     GetDBM2(obj, dbmp, dbm);
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	val = sdbm_fetch(dbm, key);
-	keystr = rb_tainted_str_new(key.dptr, key.dsize);
+	keystr = rb_external_str_new(key.dptr, key.dsize);
-	valstr = rb_tainted_str_new(val.dptr, val.dsize);
+	valstr = rb_external_str_new(val.dptr, val.dsize);
 	rb_yield(rb_assoc_new(keystr, valstr));
 	GetDBM2(obj, dbmp, dbm);
     }
@@ -547,7 +547,7 @@ fsdbm_keys(VALUE obj)
     GetDBM2(obj, dbmp, dbm);
     ary = rb_ary_new();
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
-	rb_ary_push(ary, rb_tainted_str_new(key.dptr, key.dsize));
+	rb_ary_push(ary, rb_external_str_new(key.dptr, key.dsize));
     }
 
     return ary;
@@ -565,7 +565,7 @@ fsdbm_values(VALUE obj)
     ary = rb_ary_new();
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	val = sdbm_fetch(dbm, key);
-	rb_ary_push(ary, rb_tainted_str_new(val.dptr, val.dsize));
+	rb_ary_push(ary, rb_external_str_new(val.dptr, val.dsize));
     }
 
     return ary;
@@ -578,7 +578,7 @@ fsdbm_has_key(VALUE obj, VALUE keystr)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(keystr);
+    ExportStringValue(keystr);
     key.dptr = RSTRING_PTR(keystr);
     key.dsize = RSTRING_LEN(keystr);
 
@@ -595,7 +595,7 @@ fsdbm_has_value(VALUE obj, VALUE valstr)
     struct dbmdata *dbmp;
     DBM *dbm;
 
-    StringValue(valstr);
+    ExportStringValue(valstr);
     val.dptr = RSTRING_PTR(valstr);
     val.dsize = RSTRING_LEN(valstr);
 
@@ -621,8 +621,8 @@ fsdbm_to_a(VALUE obj)
     ary = rb_ary_new();
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	val = sdbm_fetch(dbm, key);
-	rb_ary_push(ary, rb_assoc_new(rb_tainted_str_new(key.dptr, key.dsize),
+	rb_ary_push(ary, rb_assoc_new(rb_external_str_new(key.dptr, key.dsize),
-				      rb_tainted_str_new(val.dptr, val.dsize)));
+				      rb_external_str_new(val.dptr, val.dsize)));
     }
 
     return ary;
@@ -640,8 +640,8 @@ fsdbm_to_hash(VALUE obj)
     hash = rb_hash_new();
     for (key = sdbm_firstkey(dbm); key.dptr; key = sdbm_nextkey(dbm)) {
 	val = sdbm_fetch(dbm, key);
-	rb_hash_aset(hash, rb_tainted_str_new(key.dptr, key.dsize),
+	rb_hash_aset(hash, rb_external_str_new(key.dptr, key.dsize),
-		           rb_tainted_str_new(val.dptr, val.dsize));
+		           rb_external_str_new(val.dptr, val.dsize));
     }
 
     return hash;

include/ruby/encoding.h

@@ -92,7 +92,8 @@ char* rb_enc_nth(const char*, const char*, int, rb_encoding*);
 VALUE rb_obj_encoding(VALUE);
 VALUE rb_enc_str_buf_cat(VALUE str, const char *ptr, long len, rb_encoding *enc);
 
-VALUE rb_external_str_new_with_enc(const char *ptr, long len, rb_encoding *enc);
+VALUE rb_external_str_new_with_enc(const char *ptr, long len, rb_encoding *);
+VALUE rb_str_export_to_enc(VALUE, rb_encoding *);
 
 /* index -> rb_encoding */
 rb_encoding* rb_enc_from_index(int idx);

include/ruby/ruby.h

@@ -373,6 +373,12 @@ void rb_check_safe_str(VALUE);
 /* obsolete macro - use SafeStringValue(v) */
 #define Check_SafeStr(v) rb_check_safe_str((VALUE)(v))
 
+VALUE rb_str_export(VALUE);
+#define ExportStringValue(v) do {\
+    SafeStringValue(v);\
+   (v) = rb_str_export(v);\
+} while (0)
+
 VALUE rb_get_path(VALUE);
 #define FilePathValue(v) ((v) = rb_get_path(v))
 

string.c

@@ -472,52 +472,61 @@ rb_tainted_str_new_cstr(const char *ptr)
 RUBY_ALIAS_FUNCTION(rb_tainted_str_new2(const char *ptr), rb_tainted_str_new_cstr, (ptr))
 #define rb_tainted_str_new2 rb_tainted_str_new_cstr
 
+static VALUE
+str_conv_enc(VALUE str, rb_encoding *from, rb_encoding *to)
+{
+    rb_econv_t *ec;
+    rb_econv_result_t ret;
+    long len;
+    VALUE newstr;
+    const unsigned char *sp;
+    unsigned char *dp;
+
+    if (!to) return str;
+    if (from == to) return str;
+    if (rb_enc_asciicompat(to) && ENC_CODERANGE(str) == ENC_CODERANGE_7BIT)
+	return str;
+
+    len = RSTRING_LEN(str);
+    newstr = rb_str_new(0, len);
+
+  retry:
+    ec = rb_econv_open_opts(from->name, to->name, 0, Qnil);
+    if (!ec) return str;
+
+    sp = (unsigned char*)RSTRING_PTR(str);
+    dp = (unsigned char*)RSTRING_PTR(newstr);
+    ret = rb_econv_convert(ec, &sp, (unsigned char*)RSTRING_END(str),
+			   &dp, (unsigned char*)RSTRING_END(newstr), 0);
+    rb_econv_close(ec);
+    switch (ret) {
+      case econv_destination_buffer_full:
+	/* destination buffer short */
+	len *= 2;
+	rb_str_resize(newstr, len);
+	goto retry;
+
+      case econv_finished:
+	len = dp - (unsigned char*)RSTRING_PTR(newstr);
+	rb_str_set_len(newstr, len);
+	rb_enc_associate(newstr, to);
+	return newstr;
+
+      default:
+	/* some error, return original */
+	return str;
+    }
+}
+
 VALUE
 rb_external_str_new_with_enc(const char *ptr, long len, rb_encoding *eenc)
 {
     VALUE str;
-    rb_encoding *ienc;
 
     if (len == 0 && !ptr) len = strlen(ptr);
     str = rb_tainted_str_new(ptr, len);
     rb_enc_associate(str, eenc);
-    ienc = rb_default_internal_encoding();
+    return str_conv_enc(str, eenc, rb_default_internal_encoding());
-    if (ienc) {
-	rb_econv_t *ec;
-	rb_econv_result_t ret;
-	VALUE newstr = rb_str_new(0, len);
-	long nlen = len;
-	const unsigned char *sp;
-	unsigned char *dp;
-
-      retry:
-	ec = rb_econv_open_opts(eenc->name, ienc->name, 0, Qnil);
-	if (!ec) return str;
-
-	sp = (unsigned char*)RSTRING_PTR(str);
-	dp = (unsigned char*)RSTRING_PTR(newstr);
-	ret = rb_econv_convert(ec, &sp, (unsigned char*)RSTRING_END(str),
-			       &dp, (unsigned char*)RSTRING_END(newstr), 0);
-	rb_econv_close(ec);
-        switch (ret) {
-          case econv_destination_buffer_full:
-	    /* destination buffer short */
-	    nlen *= 2;
-	    rb_str_resize(newstr, nlen);
-	    goto retry;
-
-          case econv_finished:
-	    nlen = dp - (unsigned char*)RSTRING_PTR(newstr);
-	    rb_str_set_len(newstr, nlen);
-	    rb_enc_associate(newstr, ienc);
-	    return newstr;
-
-	  default:
-	    /* some error, return original */
-	    return str;
-	}
-    }
-    return str;
 }
 
 VALUE
@@ -532,6 +541,18 @@ rb_locale_str_new(const char *ptr, long len)
     return rb_external_str_new_with_enc(ptr, len, rb_locale_encoding());
 }
 
+VALUE
+rb_str_export(VALUE str)
+{
+    return str_conv_enc(str, STR_ENC_GET(str), rb_default_external_encoding());
+}
+
+VALUE
+rb_str_export_to_enc(VALUE str, rb_encoding *enc)
+{
+    return str_conv_enc(str, STR_ENC_GET(str), enc);
+}
+
 static VALUE
 str_replace_shared(VALUE str2, VALUE str)
 {