[ruby-dev:38957] Re: String#valid_encoding?にオプションが欲しい

xibbarこと藤岡です。

>> force_encodingにかんしては破壊的なイメージがあるんですよねえ。
>> 勝手な思いこみか、forceという単語の持つ印象か。
>>
>> |あとはpresume_encoding!とかありますけれども……。
>>
>> presume_encodingも用意するのであれば。
>>
> ということは、
> presume_encoding(encoding)
> presume_encoding!(encoding)
> の２つを新設すると折り合いがつくということでしょうか？
> 
実装してみました。

irb(main):001:0> str="あ".force_encoding("EUC-JP")
=> "\xE3\x81\x82"
irb(main):004:0> str.presume_encoding("EUC-JP")
=> nil
irb(main):005:0> str.presume_encoding("UTF-8")
=> "あ"
irb(main):007:0> str
=> "\xE3\x81\x82"
irb(main):008:0> str.presume_encoding!("UTF-8")
=> "あ"
irb(main):009:0> str
=> "あ"
irb(main):010:0> str="あ".force_encoding("EUC-JP")
=> "\xE3\x81\x82"
irb(main):011:0> str.presume_encoding!("SHIFT_JIS")
=> nil
irb(main):012:0> str
=> "\xE3\x81\x82"
irb(main):013:0> str.presume_encoding!("EUC-JP")
=> nil
irb(main):014:0> str
=> "\xE3\x81\x82"

以下パッチ。

Index: string.c
===================================================================
--- string.c	(revision 24163)
+++ string.c	(working copy)
@@ -7090,6 +7090,51 @@

 /*
  *  call-seq:
+ *     str.presume_encoding(encoding)   => new_str or nil
+ *
+ *  If a string can be encoded correctly in the given encoding,
+ *  returns a copy of str with changing the encoding to +encoding+ ;
+ *  otherwise, returns nil.
+ */
+
+static VALUE
+rb_str_presume_encoding(VALUE str, VALUE encoding)
+{
+    rb_encoding *enc = rb_to_encoding(encoding);
+    int cr = coderange_scan(RSTRING_PTR(str), RSTRING_LEN(str), enc);
+
+    if (cr != ENC_CODERANGE_BROKEN) {
+        return
rb_enc_str_new(RSTRING_PTR(str),strlen(RSTRING_PTR(str)),enc);
+    } else {
+        return Qnil;
+    }
+}
+
+/*
+ *  call-seq:
+ *     str.presume_encoding_bang(encoding)   => str or nil
+ *
+ *  If a string can be encoded correctly in the given encoding,
+ *  changes the encoding to +encoding+ and returns self;
+ *  otherwise, returns nil.
+ */
+
+static VALUE
+rb_str_presume_encoding_bang(VALUE str, VALUE encoding)
+{
+    rb_encoding *enc = rb_to_encoding(encoding);
+    int cr = coderange_scan(RSTRING_PTR(str), RSTRING_LEN(str), enc);
+
+    if (cr != ENC_CODERANGE_BROKEN) {
+	ENCODING_CODERANGE_SET(str, rb_enc_to_index(enc), cr);
+        return str;
+    } else {
+        return Qnil;
+    }
+}
+
+/*
+ *  call-seq:
  *     str.ascii_only?  => true or false
  *
  *  Returns true for a string which has only ASCII characters.
@@ -7531,6 +7576,8 @@
     rb_define_method(rb_cString, "encoding", rb_obj_encoding, 0); /* in
encoding.c */
     rb_define_method(rb_cString, "force_encoding",
rb_str_force_encoding, 1);
     rb_define_method(rb_cString, "valid_encoding?",
rb_str_valid_encoding_p, 0);
+    rb_define_method(rb_cString, "presume_encoding",
rb_str_presume_encoding, 1);
+    rb_define_method(rb_cString, "presume_encoding!",
rb_str_presume_encoding_bang, 1);
     rb_define_method(rb_cString, "ascii_only?", rb_str_is_ascii_only_p, 0);

     id_to_s = rb_intern("to_s");