Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
openanolis
dragonwell8_jdk
提交
349cbe55
D
dragonwell8_jdk
项目概览
openanolis
/
dragonwell8_jdk
通知
4
Star
2
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
D
dragonwell8_jdk
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
提交
349cbe55
编写于
11月 26, 2008
作者:
C
chegar
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
6720866: Slow performance using HttpURLConnection for upload
Reviewed-by: michaelm
上级
37f4867b
变更
1
显示空白变更内容
内联
并排
Showing
1 changed file
with
164 addition
and
121 deletion
+164
-121
src/share/classes/sun/net/www/http/ChunkedOutputStream.java
src/share/classes/sun/net/www/http/ChunkedOutputStream.java
+164
-121
未找到文件。
src/share/classes/sun/net/www/http/ChunkedOutputStream.java
浏览文件 @
349cbe55
...
...
@@ -29,32 +29,58 @@ import java.io.*;
/**
* OutputStream that sends the output to the underlying stream using chunked
* encoding as specified in RFC 2068.
*
* @author Alan Bateman
*/
public
class
ChunkedOutputStream
extends
PrintStream
{
/* Default chunk size (including chunk header) if not specified */
static
final
int
DEFAULT_CHUNK_SIZE
=
4096
;
private
static
final
byte
[]
CRLF
=
{
'\r'
,
'\n'
};
private
static
final
int
CRLF_SIZE
=
CRLF
.
length
;
private
static
final
byte
[]
FOOTER
=
CRLF
;
private
static
final
int
FOOTER_SIZE
=
CRLF_SIZE
;
private
static
final
byte
[]
EMPTY_CHUNK_HEADER
=
getHeader
(
0
);
private
static
final
int
EMPTY_CHUNK_HEADER_SIZE
=
getHeaderSize
(
0
);
/* internal buffer */
private
byte
buf
[];
/* size of data (excluding footers and headers) already stored in buf */
private
int
size
;
/* current index in buf (i.e. buf[count] */
private
int
count
;
/* number of bytes to be filled up to complete a data chunk
* currently being built */
private
int
spaceInCurrentChunk
;
/* underlying stream */
private
PrintStream
out
;
/* the chunk size we use */
private
int
preferredChunkSize
;
/* if the users write buffer is bigger than this size, we
* write direct from the users buffer instead of copying
*/
static
final
int
MAX_BUF_SIZE
=
10
*
1024
;
private
int
preferredChunkDataSize
;
private
int
preferedHeaderSize
;
private
int
preferredChunkGrossSize
;
/* header for a complete Chunk */
private
byte
[]
completeHeader
;
/* return the size of the header for a particular chunk size */
private
int
headerSize
(
int
size
)
{
return
2
+
(
Integer
.
toHexString
(
size
)).
length
();
private
static
int
getHeaderSize
(
int
size
)
{
return
(
Integer
.
toHexString
(
size
)).
length
()
+
CRLF_SIZE
;
}
/* return a header for a particular chunk size */
private
static
byte
[]
getHeader
(
int
size
){
try
{
String
hexStr
=
Integer
.
toHexString
(
size
);
byte
[]
hexBytes
=
hexStr
.
getBytes
(
"US-ASCII"
);
byte
[]
header
=
new
byte
[
getHeaderSize
(
size
)];
for
(
int
i
=
0
;
i
<
hexBytes
.
length
;
i
++)
header
[
i
]
=
hexBytes
[
i
];
header
[
hexBytes
.
length
]
=
CRLF
[
0
];
header
[
hexBytes
.
length
+
1
]
=
CRLF
[
1
];
return
header
;
}
catch
(
java
.
io
.
UnsupportedEncodingException
e
)
{
/* This should never happen */
throw
new
InternalError
(
e
.
getMessage
());
}
}
public
ChunkedOutputStream
(
PrintStream
o
)
{
...
...
@@ -63,111 +89,114 @@ public class ChunkedOutputStream extends PrintStream {
public
ChunkedOutputStream
(
PrintStream
o
,
int
size
)
{
super
(
o
);
out
=
o
;
if
(
size
<=
0
)
{
size
=
DEFAULT_CHUNK_SIZE
;
}
/* Adjust the size to cater for the chunk header - eg: if the
* preferred chunk size is 1k this means the chunk size should
* be 1019 bytes (differs by 5 from preferred size because of
* 3 bytes for chunk size in hex and CRLF).
* be 1017 bytes (differs by 7 from preferred size because of
* 3 bytes for chunk size in hex and CRLF (header) and CRLF (footer)).
*
* If headerSize(adjusted_size) is shorter then headerSize(size)
* then try to use the extra byte unless headerSize(adjusted_size+1)
* increases back to headerSize(size)
*/
if
(
size
>
0
)
{
int
adjusted_size
=
size
-
headerSize
(
size
)
;
if
(
adjusted_size
+
headerSize
(
adjusted_size
)
<
size
)
{
int
adjusted_size
=
size
-
getHeaderSize
(
size
)
-
FOOTER_SIZE
;
if
(
getHeaderSize
(
adjusted_size
+
1
)
<
getHeaderSize
(
size
))
{
adjusted_size
++;
}
size
=
adjusted_size
;
}
if
(
size
>
0
)
{
preferredChunkSize
=
size
;
preferredChunk
Data
Size
=
size
;
}
else
{
preferredChunkSize
=
DEFAULT_CHUNK_SIZE
-
headerSize
(
DEFAULT_CHUNK_SIZE
);
preferredChunkDataSize
=
DEFAULT_CHUNK_SIZE
-
getHeaderSize
(
DEFAULT_CHUNK_SIZE
)
-
FOOTER_SIZE
;
}
preferedHeaderSize
=
getHeaderSize
(
preferredChunkDataSize
);
preferredChunkGrossSize
=
preferedHeaderSize
+
preferredChunkDataSize
+
FOOTER_SIZE
;
completeHeader
=
getHeader
(
preferredChunkDataSize
);
/* start with an initial buffer */
buf
=
new
byte
[
preferredChunkSize
+
32
];
buf
=
new
byte
[
preferredChunkDataSize
+
32
];
reset
();
}
/*
* If flushAll is true, then all data is flushed in one chunk.
*
* If false and the size of the buffer data exceeds the preferred
* chunk size then chunks are flushed to the output stream.
* If there isn't enough data to make up a complete chunk,
* then the method returns.
* Flush a buffered, completed chunk to an underlying stream. If the data in
* the buffer is insufficient to build up a chunk of "preferredChunkSize"
* then the data do not get flushed unless flushAll is true. If flushAll is
* true then the remaining data builds up a last chunk which size is smaller
* than preferredChunkSize, and then the last chunk gets flushed to
* underlying stream. If flushAll is true and there is no data in a buffer
* at all then an empty chunk (containing a header only) gets flushed to
* underlying stream.
*/
private
void
flush
(
byte
[]
buf
,
boolean
flushAll
)
{
flush
(
buf
,
flushAll
,
0
);
}
private
void
flush
(
byte
[]
buf
,
boolean
flushAll
,
int
offset
)
{
int
chunkSize
;
do
{
if
(
count
<
preferredChunkSize
)
{
if
(!
flushAll
)
{
break
;
}
chunkSize
=
count
;
private
void
flush
(
boolean
flushAll
)
{
if
(
spaceInCurrentChunk
==
0
)
{
/* flush a completed chunk to underlying stream */
out
.
write
(
buf
,
0
,
preferredChunkGrossSize
);
out
.
flush
();
reset
();
}
else
if
(
flushAll
){
/* complete the last chunk and flush it to underlying stream */
if
(
size
>
0
){
/* adjust a header start index in case the header of the last
* chunk is shorter then preferedHeaderSize */
int
adjustedHeaderStartIndex
=
preferedHeaderSize
-
getHeaderSize
(
size
);
/* write header */
System
.
arraycopy
(
getHeader
(
size
),
0
,
buf
,
adjustedHeaderStartIndex
,
getHeaderSize
(
size
));
/* write footer */
buf
[
count
++]
=
FOOTER
[
0
];
buf
[
count
++]
=
FOOTER
[
1
];
//send the last chunk to underlying stream
out
.
write
(
buf
,
adjustedHeaderStartIndex
,
count
-
adjustedHeaderStartIndex
);
}
else
{
chunkSize
=
preferredChunkSize
;
}
byte
[]
bytes
=
null
;
try
{
bytes
=
(
Integer
.
toHexString
(
chunkSize
)).
getBytes
(
"US-ASCII"
);
}
catch
(
java
.
io
.
UnsupportedEncodingException
e
)
{
//This should never happen.
throw
new
InternalError
(
e
.
getMessage
());
//send an empty chunk (containing just a header) to underlying stream
out
.
write
(
EMPTY_CHUNK_HEADER
,
0
,
EMPTY_CHUNK_HEADER_SIZE
);
}
out
.
write
(
bytes
,
0
,
bytes
.
length
);
out
.
write
((
byte
)
'\r'
);
out
.
write
((
byte
)
'\n'
);
if
(
chunkSize
>
0
)
{
out
.
write
(
buf
,
offset
,
chunkSize
);
out
.
write
((
byte
)
'\r'
);
out
.
write
((
byte
)
'\n'
);
}
out
.
flush
();
if
(
checkError
())
{
break
;
}
if
(
chunkSize
>
0
)
{
count
-=
chunkSize
;
offset
+=
chunkSize
;
}
}
while
(
count
>
0
);
if
(!
checkError
()
&&
count
>
0
)
{
System
.
arraycopy
(
buf
,
offset
,
this
.
buf
,
0
,
count
);
reset
();
}
}
@Override
public
boolean
checkError
()
{
return
out
.
checkError
();
}
/*
* Check if we have enough data for a chunk and if so flush to the
* underlying output stream.
*/
private
void
checkFlush
()
{
if
(
count
>=
preferredChunkSize
)
{
flush
(
buf
,
false
);
}
}
/* Check that the output stream is still open */
private
void
ensureOpen
()
{
if
(
out
==
null
)
setError
();
}
/*
* Writes data from b[] to an internal buffer and stores the data as data
* chunks of a following format: {Data length in Hex}{CRLF}{data}{CRLF}
* The size of the data is preferredChunkSize. As soon as a completed chunk
* is read from b[] a process of reading from b[] suspends, the chunk gets
* flushed to the underlying stream and then the reading process from b[]
* continues. When there is no more sufficient data in b[] to build up a
* chunk of preferredChunkSize size the data get stored as an incomplete
* chunk of a following format: {space for data length}{CRLF}{data}
* The size of the data is of course smaller than preferredChunkSize.
*/
@Override
public
synchronized
void
write
(
byte
b
[],
int
off
,
int
len
)
{
ensureOpen
();
if
((
off
<
0
)
||
(
off
>
b
.
length
)
||
(
len
<
0
)
||
...
...
@@ -177,81 +206,95 @@ public class ChunkedOutputStream extends PrintStream {
return
;
}
int
l
=
preferredChunkSize
-
count
;
/* if b[] contains enough data then one loop cycle creates one complete
* data chunk with a header, body and a footer, and then flushes the
* chunk to the underlying stream. Otherwise, the last loop cycle
* creates incomplete data chunk with empty header and with no footer
* and stores this incomplete chunk in an internal buffer buf[]
*/
int
bytesToWrite
=
len
;
int
inputIndex
=
off
;
/* the index of the byte[] currently being written */
if
((
len
>
MAX_BUF_SIZE
)
&&
(
len
>
l
))
{
/* current chunk is empty just write the data */
if
(
count
==
0
)
{
count
=
len
;
flush
(
b
,
false
,
off
);
return
;
do
{
/* enough data to complete a chunk */
if
(
bytesToWrite
>=
spaceInCurrentChunk
)
{
/* header */
for
(
int
i
=
0
;
i
<
completeHeader
.
length
;
i
++)
buf
[
i
]
=
completeHeader
[
i
];
/* data */
System
.
arraycopy
(
b
,
inputIndex
,
buf
,
count
,
spaceInCurrentChunk
);
inputIndex
+=
spaceInCurrentChunk
;
bytesToWrite
-=
spaceInCurrentChunk
;
count
+=
spaceInCurrentChunk
;
/* footer */
buf
[
count
++]
=
FOOTER
[
0
];
buf
[
count
++]
=
FOOTER
[
1
];
spaceInCurrentChunk
=
0
;
//chunk is complete
flush
(
false
);
if
(
checkError
()){
break
;
}
/* first finish the current chunk */
if
(
l
>
0
)
{
System
.
arraycopy
(
b
,
off
,
buf
,
count
,
l
);
count
=
preferredChunkSize
;
flush
(
buf
,
false
);
}
count
=
len
-
l
;
/* Now write the rest of the data */
flush
(
b
,
false
,
l
+
off
);
}
else
{
int
newcount
=
count
+
len
;
/* not enough data to build a chunk */
else
{
/* header */
/* do not write header if not enough bytes to build a chunk yet */
if
(
newcount
>
buf
.
length
)
{
byte
newbuf
[]
=
new
byte
[
Math
.
max
(
buf
.
length
<<
1
,
newcount
)];
System
.
arraycopy
(
buf
,
0
,
newbuf
,
0
,
count
);
buf
=
newbuf
;
}
System
.
arraycopy
(
b
,
off
,
buf
,
count
,
len
);
count
=
newcount
;
checkFlush
();
/* data */
System
.
arraycopy
(
b
,
inputIndex
,
buf
,
count
,
bytesToWrite
);
count
+=
bytesToWrite
;
size
+=
bytesToWrite
;
spaceInCurrentChunk
-=
bytesToWrite
;
bytesToWrite
=
0
;
/* footer */
/* do not write header if not enough bytes to build a chunk yet */
}
}
while
(
bytesToWrite
>
0
);
}
public
synchronized
void
write
(
int
b
)
{
ensureOpen
();
int
newcount
=
count
+
1
;
if
(
newcount
>
buf
.
length
)
{
byte
newbuf
[]
=
new
byte
[
Math
.
max
(
buf
.
length
<<
1
,
newcount
)];
System
.
arraycopy
(
buf
,
0
,
newbuf
,
0
,
count
);
buf
=
newbuf
;
}
buf
[
count
]
=
(
byte
)
b
;
count
=
newcount
;
checkFlush
();
@Override
public
synchronized
void
write
(
int
_b
)
{
byte
b
[]
=
{(
byte
)
_b
};
write
(
b
,
0
,
1
);
}
public
synchronized
void
reset
()
{
count
=
0
;
count
=
preferedHeaderSize
;
size
=
0
;
spaceInCurrentChunk
=
preferredChunkDataSize
;
}
public
int
size
()
{
return
count
;
return
size
;
}
@Override
public
synchronized
void
close
()
{
ensureOpen
();
/* if we have buffer a chunked send it */
if
(
count
>
0
)
{
flush
(
buf
,
true
);
if
(
size
>
0
)
{
flush
(
true
);
}
/* send a zero length chunk */
flush
(
buf
,
true
);
flush
(
true
);
/* don't close the underlying stream */
out
=
null
;
}
@Override
public
synchronized
void
flush
()
{
ensureOpen
();
if
(
count
>
0
)
{
flush
(
buf
,
true
);
if
(
size
>
0
)
{
flush
(
true
);
}
}
}
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录