-
Notifications
You must be signed in to change notification settings - Fork 110
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add managed writer for Ion 1.1 with basic round-trip testing (#830)
- Loading branch information
Showing
23 changed files
with
1,164 additions
and
125 deletions.
There are no files selected for viewing
181 changes: 181 additions & 0 deletions
181
src/main/java/com/amazon/ion/impl/BlockBufferingOutputStreamFastAppendable.kt
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,181 @@ | ||
| // Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved. | ||
| // SPDX-License-Identifier: Apache-2.0 | ||
| package com.amazon.ion.impl | ||
|
|
||
| import com.amazon.ion.impl.bin.* | ||
| import com.amazon.ion.util.* | ||
| import java.io.OutputStream | ||
|
|
||
| /** | ||
| * A [_Private_FastAppendable] that buffers data to blocks of memory which are managed by a [BlockAllocator]. Only when | ||
| * [flush] is called are the blocks written to the wrapped [OutputStream]. | ||
| * | ||
| * This is necessary for cases where an [IonManagedWriter_1_1] over Ion text needs to emit encoding directives that are | ||
| * not known in advance. The [OutputStreamFastAppendable] class only buffers a fixed amount of data, so it is not safe | ||
| * to use if there are system values to be written. For a sufficiently large user value, an [OutputStreamFastAppendable] | ||
| * can end up flushing partial or whole user values flushing to the [OutputStream] before the [IonManagedWriter_1_1] can | ||
| * write the system value that it depends on. | ||
| * | ||
| * Once [IonManagedWriter_1_1] supports an auto-flush feature, then this class will have very little practical | ||
| * difference from [OutputStreamFastAppendable] for the case where no system values are needed. | ||
| * | ||
| * TODO: | ||
| * - Add proper tests | ||
| */ | ||
| internal class BlockBufferingOutputStreamFastAppendable( | ||
| private val out: OutputStream, | ||
| private val allocator: BlockAllocator, | ||
| /** | ||
| * The minimum utilization of a block before a longer value | ||
| * can skip the end of a block and just start a new block. | ||
| */ | ||
| minBlockUtilization: Double = 1.0, | ||
| ) : OutputStream(), _Private_FastAppendable { | ||
|
|
||
| init { | ||
| // 0.0 would have the possibility of wasting entire blocks. | ||
| // 0.5 is somewhat arbitrary, but at least sensible that you should use at least | ||
| // half of a block before moving on to the next block. | ||
| require(minBlockUtilization in 0.5..1.0) { "minBlockUtilization must be between 0.5 and 1" } | ||
| require(allocator.blockSize > 10) | ||
| } | ||
|
|
||
| private val maxBlockWaste: Int = (allocator.blockSize * (1.0 - minBlockUtilization)).toInt() | ||
|
|
||
| private var index = -1 | ||
| private val blocks = mutableListOf<Block>() | ||
| private var current: Block = nextBlock() | ||
|
|
||
| private fun nextBlock(): Block { | ||
| index++ | ||
| if (index < 0) throw IllegalStateException("This output stream is closed.") | ||
| if (index >= blocks.size) blocks.add(allocator.allocateBlock()) | ||
| return blocks[index] | ||
| } | ||
|
|
||
| override fun close() { | ||
| flush() | ||
| blocks.onEach { it.close() }.clear() | ||
| index = Int.MIN_VALUE | ||
| } | ||
|
|
||
| override fun flush() { | ||
| blocks.forEach { block -> | ||
| out.write(block.data, 0, block.limit) | ||
| block.reset() | ||
| } | ||
| index = 0 | ||
| current = blocks[index] | ||
| } | ||
|
|
||
| override fun write(b: Int) { | ||
| if (current.remaining() < 1) current = nextBlock() | ||
| val block = current | ||
| block.data[block.limit] = b.toByte() | ||
| block.limit++ | ||
| } | ||
|
|
||
| override fun write(b: ByteArray, off: Int, len: Int) { | ||
| if (len > current.remaining()) { | ||
| if (current.remaining() < maxBlockWaste && len < allocator.blockSize) { | ||
| current = nextBlock() | ||
| } else { | ||
| writeBytesSlow(b, off, len) | ||
| return | ||
| } | ||
| } | ||
| val block = current | ||
| System.arraycopy(b, off, block.data, block.limit, len) | ||
| block.limit += len | ||
| } | ||
|
|
||
| // slow in the sense that we do all kind of block boundary checking | ||
| private fun writeBytesSlow(bytes: ByteArray, _off: Int, _len: Int) { | ||
| var off = _off | ||
| var len = _len | ||
| while (len > 0) { | ||
| val block = current | ||
| val amount = Math.min(len, block.remaining()) | ||
| System.arraycopy(bytes, off, block.data, block.limit, amount) | ||
| block.limit += amount | ||
| off += amount | ||
| len -= amount | ||
| if (block.remaining() == 0) { | ||
| current = nextBlock() | ||
| } | ||
| } | ||
| } | ||
|
|
||
| override fun append(c: Char): Appendable = apply { if (c.code < 0x80) appendAscii(c) else appendUtf16(c) } | ||
|
|
||
| override fun append(csq: CharSequence): Appendable = apply { append(csq, 0, csq.length) } | ||
|
|
||
| override fun append(csq: CharSequence, start: Int, end: Int): Appendable { | ||
| for (i in start until end) { | ||
| append(csq[i]) | ||
| } | ||
| return this | ||
| } | ||
|
|
||
| override fun appendAscii(c: Char) { | ||
| assert(c.code < 0x80) | ||
| write(c.code) | ||
| } | ||
|
|
||
| override fun appendAscii(csq: CharSequence) = appendAscii(csq, 0, csq.length) | ||
|
|
||
| override fun appendAscii(csq: CharSequence, start: Int, end: Int) { | ||
| if (csq is String) { | ||
| // Using deprecated String.getBytes intentionally, since it is | ||
| // correct behavior in this case, and much faster. | ||
| var pos = start | ||
| val len = end - start | ||
| if (len > current.remaining() && current.remaining() < maxBlockWaste && len < allocator.blockSize) { | ||
| current = nextBlock() | ||
| } | ||
| while (true) { | ||
| val copyAmount = minOf(current.remaining(), end - pos) | ||
| csq.copyBytes(pos, pos + copyAmount, current.data, current.limit) | ||
| current.limit += copyAmount | ||
| pos += copyAmount | ||
| if (pos >= end) return | ||
| current = nextBlock() | ||
| } | ||
| } else { | ||
| append(csq, start, end) | ||
| } | ||
| } | ||
|
|
||
| override fun appendUtf16(c: Char) { | ||
| assert(c.code >= 0x80) | ||
| if (current.remaining() < 3) { | ||
| current = nextBlock() | ||
| } | ||
| if (c.code < 0x800) { | ||
| current.data[current.limit++] = (0xff and (0xC0 or (c.code shr 6))).toByte() | ||
| current.data[current.limit++] = (0xff and (0x80 or (c.code and 0x3F))).toByte() | ||
| } else if (c.code < 0x10000) { | ||
| current.data[current.limit++] = (0xff and (0xE0 or (c.code shr 12))).toByte() | ||
| current.data[current.limit++] = (0xff and (0x80 or (c.code shr 6 and 0x3F))).toByte() | ||
| current.data[current.limit++] = (0xff and (0x80 or (c.code and 0x3F))).toByte() | ||
| } | ||
| } | ||
|
|
||
| override fun appendUtf16Surrogate(leadSurrogate: Char, trailSurrogate: Char) { | ||
| // Here we must convert a UTF-16 surrogate pair to UTF-8 bytes. | ||
| val c = _Private_IonConstants.makeUnicodeScalar(leadSurrogate.code, trailSurrogate.code) | ||
| assert(c >= 0x10000) | ||
| if (current.remaining() < 4) { | ||
| current = nextBlock() | ||
| } | ||
| current.data[current.limit++] = (0xff and (0xF0 or (c shr 18))).toByte() | ||
| current.data[current.limit++] = (0xff and (0x80 or (c shr 12 and 0x3F))).toByte() | ||
| current.data[current.limit++] = (0xff and (0x80 or (c shr 6 and 0x3F))).toByte() | ||
| current.data[current.limit++] = (0xff and (0x80 or (c and 0x3F))).toByte() | ||
| } | ||
|
|
||
| /** Helper function to wrap [java.lang.String.getBytes]. */ | ||
| private fun String.copyBytes(srcBegin: Int, srcEnd: Int, dst: ByteArray, dstBegin: Int) { | ||
| (this as java.lang.String).getBytes(srcBegin, srcEnd, dst, dstBegin) | ||
| } | ||
| } |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.