liblzma: Add lzma_block_uncomp_encode().
This also adds a new internal function
lzma_block_buffer_bound64() which is similar to
lzma_block_buffer_bound() but uses uint64_t instead
of size_t.
diff --git a/src/liblzma/api/lzma/block.h b/src/liblzma/api/lzma/block.h
index 0d77f32..e450ad6 100644
--- a/src/liblzma/api/lzma/block.h
+++ b/src/liblzma/api/lzma/block.h
@@ -497,6 +497,24 @@
/**
+ * \brief Single-call uncompress .xz Block encoder
+ *
+ * This is like lzma_block_buffer_encode() except this doesn't try to
+ * compress the data and instead encodes the data using LZMA2 uncompressed
+ * chunks. The required output buffer size can be determined with
+ * lzma_block_buffer_bound().
+ *
+ * Since the data won't be compressed, this function ignores block->filters.
+ * This function doesn't take lzma_allocator because this function doesn't
+ * allocate any memory from the heap.
+ */
+extern LZMA_API(lzma_ret) lzma_block_uncomp_encode(lzma_block *block,
+ const uint8_t *in, size_t in_size,
+ uint8_t *out, size_t *out_pos, size_t out_size)
+ lzma_nothrow lzma_attr_warn_unused_result;
+
+
+/**
* \brief Single-call .xz Block decoder
*
* This is single-call equivalent of lzma_block_decoder(), and requires that
diff --git a/src/liblzma/common/block_buffer_encoder.c b/src/liblzma/common/block_buffer_encoder.c
index e678b30..dcd574e 100644
--- a/src/liblzma/common/block_buffer_encoder.c
+++ b/src/liblzma/common/block_buffer_encoder.c
@@ -10,6 +10,7 @@
//
///////////////////////////////////////////////////////////////////////////////
+#include "block_buffer_encoder.h"
#include "block_encoder.h"
#include "filter_encoder.h"
#include "lzma2_encoder.h"
@@ -28,8 +29,8 @@
+ LZMA_CHECK_SIZE_MAX + 3) & ~3)
-static lzma_vli
-lzma2_bound(lzma_vli uncompressed_size)
+static uint64_t
+lzma2_bound(uint64_t uncompressed_size)
{
// Prevent integer overflow in overhead calculation.
if (uncompressed_size > COMPRESSED_SIZE_MAX)
@@ -39,7 +40,7 @@
// uncompressed_size up to the next multiple of LZMA2_CHUNK_MAX,
// multiply by the size of per-chunk header, and add one byte for
// the end marker.
- const lzma_vli overhead = ((uncompressed_size + LZMA2_CHUNK_MAX - 1)
+ const uint64_t overhead = ((uncompressed_size + LZMA2_CHUNK_MAX - 1)
/ LZMA2_CHUNK_MAX)
* LZMA2_HEADER_UNCOMPRESSED + 1;
@@ -51,30 +52,36 @@
}
-extern LZMA_API(size_t)
-lzma_block_buffer_bound(size_t uncompressed_size)
+extern uint64_t
+lzma_block_buffer_bound64(uint64_t uncompressed_size)
{
- // For now, if the data doesn't compress, we always use uncompressed
- // chunks of LZMA2. In future we may use Subblock filter too, but
- // but for simplicity we probably will still use the same bound
- // calculation even though Subblock filter would have slightly less
- // overhead.
- lzma_vli lzma2_size = lzma2_bound(uncompressed_size);
+ // If the data doesn't compress, we always use uncompressed
+ // LZMA2 chunks.
+ uint64_t lzma2_size = lzma2_bound(uncompressed_size);
if (lzma2_size == 0)
return 0;
// Take Block Padding into account.
- lzma2_size = (lzma2_size + 3) & ~LZMA_VLI_C(3);
+ lzma2_size = (lzma2_size + 3) & ~UINT64_C(3);
-#if SIZE_MAX < LZMA_VLI_MAX
- // Catch the possible integer overflow on 32-bit systems. There's no
- // overflow on 64-bit systems, because lzma2_bound() already takes
+ // No risk of integer overflow because lzma2_bound() already takes
// into account the size of the headers in the Block.
- if (SIZE_MAX - HEADERS_BOUND < lzma2_size)
+ return HEADERS_BOUND + lzma2_size;
+}
+
+
+extern LZMA_API(size_t)
+lzma_block_buffer_bound(size_t uncompressed_size)
+{
+ uint64_t ret = lzma_block_buffer_bound64(uncompressed_size);
+
+#if SIZE_MAX < UINT64_MAX
+ // Catch the possible integer overflow on 32-bit systems.
+ if (ret > SIZE_MAX)
return 0;
#endif
- return HEADERS_BOUND + lzma2_size;
+ return ret;
}
@@ -82,9 +89,6 @@
block_encode_uncompressed(lzma_block *block, const uint8_t *in, size_t in_size,
uint8_t *out, size_t *out_pos, size_t out_size)
{
- // TODO: Figure out if the last filter is LZMA2 or Subblock and use
- // that filter to encode the uncompressed chunks.
-
// Use LZMA2 uncompressed chunks. We wouldn't need a dictionary at
// all, but LZMA2 always requires a dictionary, so use the minimum
// value to minimize memory usage of the decoder.
@@ -165,11 +169,6 @@
uint8_t *out, size_t *out_pos, size_t out_size)
{
// Find out the size of the Block Header.
- block->compressed_size = lzma2_bound(in_size);
- if (block->compressed_size == 0)
- return LZMA_DATA_ERROR;
-
- block->uncompressed_size = in_size;
return_if_error(lzma_block_header_size(block));
// Reserve space for the Block Header and skip it for now.
@@ -221,10 +220,11 @@
}
-extern LZMA_API(lzma_ret)
-lzma_block_buffer_encode(lzma_block *block, const lzma_allocator *allocator,
+static lzma_ret
+block_buffer_encode(lzma_block *block, const lzma_allocator *allocator,
const uint8_t *in, size_t in_size,
- uint8_t *out, size_t *out_pos, size_t out_size)
+ uint8_t *out, size_t *out_pos, size_t out_size,
+ bool try_to_compress)
{
// Validate the arguments.
if (block == NULL || (in == NULL && in_size != 0) || out == NULL
@@ -237,7 +237,7 @@
return LZMA_OPTIONS_ERROR;
if ((unsigned int)(block->check) > LZMA_CHECK_ID_MAX
- || block->filters == NULL)
+ || (try_to_compress && block->filters == NULL))
return LZMA_PROG_ERROR;
if (!lzma_check_is_supported(block->check))
@@ -258,9 +258,19 @@
out_size -= check_size;
+ // Initialize block->uncompressed_size and calculate the worst-case
+ // value for block->compressed_size.
+ block->uncompressed_size = in_size;
+ block->compressed_size = lzma2_bound(in_size);
+ if (block->compressed_size == 0)
+ return LZMA_DATA_ERROR;
+
// Do the actual compression.
- const lzma_ret ret = block_encode_normal(block, allocator,
- in, in_size, out, out_pos, out_size);
+ lzma_ret ret = LZMA_BUF_ERROR;
+ if (try_to_compress)
+ ret = block_encode_normal(block, allocator,
+ in, in_size, out, out_pos, out_size);
+
if (ret != LZMA_OK) {
// If the error was something else than output buffer
// becoming full, return the error now.
@@ -303,3 +313,25 @@
return LZMA_OK;
}
+
+
+extern LZMA_API(lzma_ret)
+lzma_block_buffer_encode(lzma_block *block, const lzma_allocator *allocator,
+ const uint8_t *in, size_t in_size,
+ uint8_t *out, size_t *out_pos, size_t out_size)
+{
+ return block_buffer_encode(block, allocator,
+ in, in_size, out, out_pos, out_size, true);
+}
+
+
+extern LZMA_API(lzma_ret)
+lzma_block_uncomp_encode(lzma_block *block,
+ const uint8_t *in, size_t in_size,
+ uint8_t *out, size_t *out_pos, size_t out_size)
+{
+ // It won't allocate any memory from heap so no need
+ // for lzma_allocator.
+ return block_buffer_encode(block, NULL,
+ in, in_size, out, out_pos, out_size, false);
+}
diff --git a/src/liblzma/common/block_buffer_encoder.h b/src/liblzma/common/block_buffer_encoder.h
new file mode 100644
index 0000000..653207f
--- /dev/null
+++ b/src/liblzma/common/block_buffer_encoder.h
@@ -0,0 +1,24 @@
+///////////////////////////////////////////////////////////////////////////////
+//
+/// \file block_buffer_encoder.h
+/// \brief Single-call .xz Block encoder
+//
+// Author: Lasse Collin
+//
+// This file has been put into the public domain.
+// You can do whatever you want with this file.
+//
+///////////////////////////////////////////////////////////////////////////////
+
+#ifndef LZMA_BLOCK_BUFFER_ENCODER_H
+#define LZMA_BLOCK_BUFFER_ENCODER_H
+
+#include "common.h"
+
+
+/// uint64_t version of lzma_block_buffer_bound(). It is used by
+/// stream_encoder_mt.c. Probably the original lzma_block_buffer_bound()
+/// should have been 64-bit, but fixing it would break the ABI.
+extern uint64_t lzma_block_buffer_bound64(uint64_t uncompressed_size);
+
+#endif
diff --git a/src/liblzma/liblzma.map b/src/liblzma/liblzma.map
index 93a40f4..85d9d77 100644
--- a/src/liblzma/liblzma.map
+++ b/src/liblzma/liblzma.map
@@ -97,6 +97,7 @@
XZ_5.1.2alpha {
global:
+ lzma_block_uncomp_encode;
lzma_get_progress;
lzma_stream_encoder_mt;
lzma_stream_encoder_mt_memusage;
