From 0f01c81cfb4e31fa60c42f873d2e7e734e0f6866 Mon Sep 17 00:00:00 2001
From: Sascha Matzke <sascha.matzke@didolo.org>
Date: Fri, 18 Jun 2010 12:51:35 +0200
Subject: [PATCH] fixed utf8 handling in bson

---
 extra/bson/bson-tests.factor    |  4 ++--
 extra/bson/reader/reader.factor |  6 ++++--
 extra/bson/writer/writer.factor | 12 ++++++++++--
 3 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/extra/bson/bson-tests.factor b/extra/bson/bson-tests.factor
index 7353a9a831..5540cb2ef5 100644
--- a/extra/bson/bson-tests.factor
+++ b/extra/bson/bson-tests.factor
@@ -8,8 +8,8 @@ IN: bson.tests
 
 [ H{ { "a" "a string" } } ] [ H{ { "a" "a string" } } turnaround ] unit-test
 
-[ H{ { "a" "a string" } { "b" H{ { "a" "a string" } } } } ]
-[ H{ { "a" "a string" } { "b" H{ { "a" "a string" } } } } turnaround ] unit-test
+[ H{ { "a" "a string" } { "b" H{ { "a" "アップルからの最新のニュースや情報を読む" } } } } ]
+[ H{ { "a" "a string" } { "b" H{ { "a" "アップルからの最新のニュースや情報を読む" } } } } turnaround ] unit-test
 
 [ H{ { "a list" { 1 2.234 "hello world" } } } ]
 [ H{ { "a list" { 1 2.234 "hello world" } } } turnaround ] unit-test
diff --git a/extra/bson/reader/reader.factor b/extra/bson/reader/reader.factor
index a007431e4a..f1f3ab8508 100644
--- a/extra/bson/reader/reader.factor
+++ b/extra/bson/reader/reader.factor
@@ -2,6 +2,7 @@
 ! See http://factorcode.org/license.txt for BSD license.
 USING: accessors assocs bson.constants calendar combinators
 combinators.short-circuit io io.binary kernel math locals
+io.encodings.utf8 io.encodings
 namespaces sequences serialize strings vectors byte-arrays ;
 
 FROM: io.encodings.binary => binary ;
@@ -34,10 +35,11 @@ DEFER: read-elements
     read-byte-raw first ; inline
 
 : read-cstring ( -- string )
-    "\0" read-until drop >string ; inline
+    input-stream get utf8 <decoder>
+    "\0" swap stream-read-until drop ; inline
 
 : read-sized-string ( length -- string )
-    read 1 head-slice* >string ; inline
+    read binary [ read-cstring ] with-byte-reader ; inline
 
 : read-timestamp ( -- timestamp )
     8 read [ 4 head signed-le> ] [ 4 tail signed-le> ] bi <mongo-timestamp> ;
diff --git a/extra/bson/writer/writer.factor b/extra/bson/writer/writer.factor
index c711451634..e02b2c6da2 100644
--- a/extra/bson/writer/writer.factor
+++ b/extra/bson/writer/writer.factor
@@ -2,6 +2,7 @@
 ! See http://factorcode.org/license.txt for BSD license.
 USING: accessors arrays assocs bson.constants byte-arrays
 calendar combinators.short-circuit fry hashtables io io.binary
+io.encodings.utf8 io.encodings io.streams.byte-array
 kernel linked-assocs literals math math.parser namespaces byte-vectors
 quotations sequences serialize strings vectors dlists alien.accessors ;
 FROM: words => word? word ;
@@ -42,8 +43,11 @@ TYPED: write-int32 ( int: integer -- ) INT32-SIZE (>le) ; inline
 
 TYPED: write-double ( real: float -- ) double>bits INT64-SIZE (>le) ; inline
 
+TYPED: write-utf8-string ( string: string -- )
+    output-stream get utf8 <encoder> stream-write ; inline
+
 TYPED: write-cstring ( string: string -- )
-    get-output [ length ] [  ] bi copy 0 write1 ; inline
+    write-utf8-string 0 write1 ; inline
 
 : write-longlong ( object -- ) INT64-SIZE (>le) ; inline
 
@@ -94,8 +98,12 @@ TYPED: (serialize-code) ( code: code -- )
   [ length write-int32 ]
   [ T_Binary_Custom write1 write ] bi ; inline
 
+: write-string-length ( string -- )
+    [ length>> 1 + ] 
+    [ aux>> [ length ] [ 0 ] if* ] bi + write-int32 ; inline
+
 TYPED: write-string ( string: string -- )
-    '[ _ write-cstring ] with-length-prefix-excl ; inline
+    dup write-string-length write-cstring ; inline
 
 TYPED: write-boolean ( bool: boolean -- )
     [ 1 write1 ] [ 0 write1 ] if ; inline