1 files changed, 306 insertions, 0 deletions

diff --git a/src/liblzma/simple/simple_coder.c b/src/liblzma/simple/simple_coder.c
new file mode 100644
index 00000000..f60dff34
--- /dev/null
+++ b/src/liblzma/simple/simple_coder.c
@@ -0,0 +1,306 @@
+///////////////////////////////////////////////////////////////////////////////
+//
+/// \file       simple_coder.c
+/// \brief      Wrapper for simple filters
+///
+/// Simple filters don't change the size of the data i.e. number of bytes
+/// in equals the number of bytes out.
+//
+//  Copyright (C) 2007 Lasse Collin
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+///////////////////////////////////////////////////////////////////////////////
+
+#include "simple_private.h"
+
+
+/// Copied or encodes/decodes more data to out[]. Checks and updates
+/// uncompressed_size when we are the last coder in the chain.
+/// If we aren't the last filter in the chain, we don't need to care about
+/// uncompressed size, since we don't change it; the next filter in the
+/// chain will check it anyway.
+static lzma_ret
+copy_or_code(lzma_coder *coder, lzma_allocator *allocator,
+		const uint8_t *restrict in, size_t *restrict in_pos,
+		size_t in_size, uint8_t *restrict out,
+		size_t *restrict out_pos, size_t out_size, lzma_action action)
+{
+	assert(!coder->end_was_reached);
+
+	if (coder->next.code == NULL) {
+		const size_t in_avail = in_size - *in_pos;
+
+		if (coder->is_encoder) {
+			if (action == LZMA_FINISH) {
+				// If uncompressed size is known and the
+				// amount of available input doesn't match
+				// the uncompressed size, return an error.
+				if (coder->uncompressed_size
+						!= LZMA_VLI_VALUE_UNKNOWN
+						&& coder->uncompressed_size
+							!= in_avail)
+					return LZMA_DATA_ERROR;
+
+			} else if (coder->uncompressed_size
+					< (lzma_vli)(in_avail)) {
+				// There is too much input available.
+				return LZMA_DATA_ERROR;
+			}
+		} else {
+			// Limit in_size so that we don't copy too much.
+			if ((lzma_vli)(in_avail) > coder->uncompressed_size)
+				in_size = *in_pos + (size_t)(
+						coder->uncompressed_size);
+		}
+
+		// Store the old position so we can update uncompressed_size.
+		const size_t out_start = *out_pos;
+
+		// Copy the data
+		bufcpy(in, in_pos, in_size, out, out_pos, out_size);
+
+		// Update uncompressed_size.
+		if (coder->uncompressed_size != LZMA_VLI_VALUE_UNKNOWN)
+			coder->uncompressed_size -= *out_pos - out_start;
+
+		// Check if end of stream was reached.
+		if (coder->is_encoder) {
+			if (action == LZMA_FINISH && *in_pos == in_size)
+				coder->end_was_reached = true;
+		} else {
+			if (coder->uncompressed_size == 0)
+				coder->end_was_reached = true;
+		}
+
+	} else {
+		// Call the next coder in the chain to provide us some data.
+		// We don't care about uncompressed_size here, because
+		// the next filter in the chain will do it for us (since
+		// we don't change the size of the data).
+		const lzma_ret ret = coder->next.code(
+				coder->next.coder, allocator,
+				in, in_pos, in_size,
+				out, out_pos, out_size, action);
+
+		if (ret == LZMA_STREAM_END) {
+			assert(!coder->is_encoder
+					|| action == LZMA_FINISH);
+			coder->end_was_reached = true;
+
+		} else if (ret != LZMA_OK) {
+			return ret;
+		}
+	}
+
+	return LZMA_OK;
+}
+
+
+static size_t
+call_filter(lzma_coder *coder, uint8_t *buffer, size_t size)
+{
+	const size_t filtered = coder->filter(coder->simple,
+			coder->now_pos, coder->is_encoder,
+			buffer, size);
+	coder->now_pos += filtered;
+	return filtered;
+}
+
+
+static lzma_ret
+simple_code(lzma_coder *coder, lzma_allocator *allocator,
+		const uint8_t *restrict in, size_t *restrict in_pos,
+		size_t in_size, uint8_t *restrict out,
+		size_t *restrict out_pos, size_t out_size, lzma_action action)
+{
+	// Flush already filtered data from coder->buffer[] to out[].
+	if (coder->pos < coder->filtered) {
+		bufcpy(coder->buffer, &coder->pos, coder->filtered,
+				out, out_pos, out_size);
+
+		// If we couldn't flush all the filtered data, return to
+		// application immediatelly.
+		if (coder->pos < coder->filtered)
+			return LZMA_OK;
+
+		if (coder->end_was_reached) {
+			assert(coder->filtered == coder->size);
+			return LZMA_STREAM_END;
+		}
+	}
+
+	// If we get here, there is no filtered data left in the buffer.
+	coder->filtered = 0;
+
+	assert(!coder->end_was_reached);
+
+	// If there is more output space left than there is unfiltered data
+	// in coder->buffer[], flush coder->buffer[] to out[], and copy/code
+	// more data to out[] hopefully filling it completely. Then filter
+	// the data in out[]. This step is where most of the data gets
+	// filtered if the buffer sizes used by the application are reasonable.
+	const size_t out_avail = out_size - *out_pos;
+	const size_t buf_avail = coder->size - coder->pos;
+	if (out_avail > buf_avail) {
+		// Store the old position so that we know from which byte
+		// to start filtering.
+		const size_t out_start = *out_pos;
+
+		// Flush data from coder->buffer[] to out[], but don't reset
+		// coder->pos and coder->size yet. This way the coder can be
+		// restarted if the next filter in the chain returns e.g.
+		// LZMA_MEM_ERROR.
+		memcpy(out + *out_pos, coder->buffer + coder->pos, buf_avail);
+		*out_pos += buf_avail;
+
+		// Copy/Encode/Decode more data to out[].
+		{
+			const lzma_ret ret = copy_or_code(coder, allocator,
+					in, in_pos, in_size,
+					out, out_pos, out_size, action);
+			assert(ret != LZMA_STREAM_END);
+			if (ret != LZMA_OK)
+				return ret;
+		}
+
+		// Filter out[].
+		const size_t size = *out_pos - out_start;
+		const size_t filtered = call_filter(
+				coder, out + out_start, size);
+
+		const size_t unfiltered = size - filtered;
+		assert(unfiltered <= coder->allocated / 2);
+
+		// Now we can update coder->pos and coder->size, because
+		// the next coder in the chain (if any) was successful.
+		coder->pos = 0;
+		coder->size = unfiltered;
+
+		if (coder->end_was_reached) {
+			// The last byte has been copied to out[] already.
+			// They are left as is.
+			coder->size = 0;
+
+		} else if (unfiltered > 0) {
+			// There is unfiltered data left in out[]. Copy it to
+			// coder->buffer[] and rewind *out_pos appropriately.
+			*out_pos -= unfiltered;
+			memcpy(coder->buffer, out + *out_pos, unfiltered);
+		}
+	} else if (coder->pos > 0) {
+		memmove(coder->buffer, coder->buffer + coder->pos, buf_avail);
+		coder->size -= coder->pos;
+		coder->pos = 0;
+	}
+
+	assert(coder->pos == 0);
+
+	// If coder->buffer[] isn't empty, try to fill it by copying/decoding
+	// more data. Then filter coder->buffer[] and copy the successfully
+	// filtered data to out[]. It is probable, that some filtered and
+	// unfiltered data will be left to coder->buffer[].
+	if (coder->size > 0) {
+		{
+			const lzma_ret ret = copy_or_code(coder, allocator,
+					in, in_pos, in_size,
+					coder->buffer, &coder->size,
+					coder->allocated, action);
+			assert(ret != LZMA_STREAM_END);
+			if (ret != LZMA_OK)
+				return ret;
+		}
+
+		coder->filtered = call_filter(
+				coder, coder->buffer, coder->size);
+
+		// Everything is considered to be filtered if coder->buffer[]
+		// contains the last bytes of the data.
+		if (coder->end_was_reached)
+			coder->filtered = coder->size;
+
+		// Flush as much as possible.
+		bufcpy(coder->buffer, &coder->pos, coder->filtered,
+				out, out_pos, out_size);
+	}
+
+	// Check if we got everything done.
+	if (coder->end_was_reached && coder->pos == coder->size)
+		return LZMA_STREAM_END;
+
+	return LZMA_OK;
+}
+
+
+static void
+simple_coder_end(lzma_coder *coder, lzma_allocator *allocator)
+{
+	lzma_next_coder_end(&coder->next, allocator);
+	lzma_free(coder->simple, allocator);
+	lzma_free(coder, allocator);
+	return;
+}
+
+
+extern lzma_ret
+lzma_simple_coder_init(lzma_next_coder *next, lzma_allocator *allocator,
+		const lzma_filter_info *filters,
+		size_t (*filter)(lzma_simple *simple, uint32_t now_pos,
+			bool is_encoder, uint8_t *buffer, size_t size),
+		size_t simple_size, size_t unfiltered_max, bool is_encoder)
+{
+	// Allocate memory for the lzma_coder structure if needed.
+	if (next->coder == NULL) {
+		// Here we allocate space also for the temporary buffer. We
+		// need twice the size of unfiltered_max, because then it
+		// is always possible to filter at least unfiltered_max bytes
+		// more data in coder->buffer[] if it can be filled completely.
+		next->coder = lzma_alloc(sizeof(lzma_coder)
+				+ 2 * unfiltered_max, allocator);
+		if (next->coder == NULL)
+			return LZMA_MEM_ERROR;
+
+		next->code = &simple_code;
+		next->end = &simple_coder_end;
+
+		next->coder->next = LZMA_NEXT_CODER_INIT;
+		next->coder->filter = filter;
+		next->coder->allocated = 2 * unfiltered_max;
+
+		// Allocate memory for filter-specific data structure.
+		if (simple_size > 0) {
+			next->coder->simple = lzma_alloc(
+					simple_size, allocator);
+			if (next->coder->simple == NULL)
+				return LZMA_MEM_ERROR;
+		} else {
+			next->coder->simple = NULL;
+		}
+	}
+
+	if (filters[0].options != NULL) {
+		const lzma_options_simple *simple = filters[0].options;
+		next->coder->now_pos = simple->start_offset;
+	} else {
+		next->coder->now_pos = 0;
+	}
+
+	// Reset variables.
+	next->coder->is_encoder = is_encoder;
+	next->coder->end_was_reached = false;
+	next->coder->uncompressed_size = filters[0].uncompressed_size;
+	next->coder->pos = 0;
+	next->coder->filtered = 0;
+	next->coder->size = 0;
+
+	return lzma_next_filter_init(
+			&next->coder->next, allocator, filters + 1);
+}