xref: /minix3/external/bsd/libarchive/dist/libarchive/archive_write_set_compression_xz.c (revision 543adbed3a3a783ed36434adafbc258b6bde442d)
1 /*-
2  * Copyright (c) 2009 Michihiro NAKAJIMA
3  * Copyright (c) 2003-2007 Tim Kientzle
4  * All rights reserved.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice, this list of conditions and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
16  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
17  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
18  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
19  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
20  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
21  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
22  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
23  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
24  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25  */
26 
27 #include "archive_platform.h"
28 
29 __FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_xz.c 201108 2009-12-28 03:28:21Z kientzle $");
30 
31 #ifdef HAVE_ERRNO_H
32 #include <errno.h>
33 #endif
34 #ifdef HAVE_STDLIB_H
35 #include <stdlib.h>
36 #endif
37 #ifdef HAVE_STRING_H
38 #include <string.h>
39 #endif
40 #include <time.h>
41 #ifdef HAVE_LZMA_H
42 #include <lzma.h>
43 #endif
44 
45 #include "archive.h"
46 #include "archive_private.h"
47 #include "archive_write_private.h"
48 
49 #ifndef HAVE_LZMA_H
50 int
archive_write_set_compression_xz(struct archive * a)51 archive_write_set_compression_xz(struct archive *a)
52 {
53 	archive_set_error(a, ARCHIVE_ERRNO_MISC,
54 	    "xz compression not supported on this platform");
55 	return (ARCHIVE_FATAL);
56 }
57 
58 int
archive_write_set_compression_lzma(struct archive * a)59 archive_write_set_compression_lzma(struct archive *a)
60 {
61 	archive_set_error(a, ARCHIVE_ERRNO_MISC,
62 	    "lzma compression not supported on this platform");
63 	return (ARCHIVE_FATAL);
64 }
65 #else
66 /* Don't compile this if we don't have liblzma. */
67 
68 struct private_data {
69 	lzma_stream	 stream;
70 	lzma_filter	 lzmafilters[2];
71 	lzma_options_lzma lzma_opt;
72 	int64_t		 total_in;
73 	unsigned char	*compressed;
74 	size_t		 compressed_buffer_size;
75 };
76 
77 struct private_config {
78 	int		 compression_level;
79 };
80 
81 static int	archive_compressor_xz_init(struct archive_write *);
82 static int	archive_compressor_xz_options(struct archive_write *,
83 		    const char *, const char *);
84 static int	archive_compressor_xz_finish(struct archive_write *);
85 static int	archive_compressor_xz_write(struct archive_write *,
86 		    const void *, size_t);
87 static int	drive_compressor(struct archive_write *, struct private_data *,
88 		    int finishing);
89 
90 
91 /*
92  * Allocate, initialize and return a archive object.
93  */
94 int
archive_write_set_compression_xz(struct archive * _a)95 archive_write_set_compression_xz(struct archive *_a)
96 {
97 	struct private_config *config;
98 	struct archive_write *a = (struct archive_write *)_a;
99 	__archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC,
100 	    ARCHIVE_STATE_NEW, "archive_write_set_compression_xz");
101 	config = calloc(1, sizeof(*config));
102 	if (config == NULL) {
103 		archive_set_error(&a->archive, ENOMEM, "Out of memory");
104 		return (ARCHIVE_FATAL);
105 	}
106 	a->compressor.config = config;
107 	a->compressor.finish = archive_compressor_xz_finish;
108 	config->compression_level = LZMA_PRESET_DEFAULT;
109 	a->compressor.init = &archive_compressor_xz_init;
110 	a->compressor.options = &archive_compressor_xz_options;
111 	a->archive.compression_code = ARCHIVE_COMPRESSION_XZ;
112 	a->archive.compression_name = "xz";
113 	return (ARCHIVE_OK);
114 }
115 
116 /* LZMA is handled identically, we just need a different compression
117  * code set.  (The liblzma setup looks at the code to determine
118  * the one place that XZ and LZMA require different handling.) */
119 int
archive_write_set_compression_lzma(struct archive * _a)120 archive_write_set_compression_lzma(struct archive *_a)
121 {
122 	struct archive_write *a = (struct archive_write *)_a;
123 	int r = archive_write_set_compression_xz(_a);
124 	if (r != ARCHIVE_OK)
125 		return (r);
126 	a->archive.compression_code = ARCHIVE_COMPRESSION_LZMA;
127 	a->archive.compression_name = "lzma";
128 	return (ARCHIVE_OK);
129 }
130 
131 static int
archive_compressor_xz_init_stream(struct archive_write * a,struct private_data * state)132 archive_compressor_xz_init_stream(struct archive_write *a,
133     struct private_data *state)
134 {
135 	int ret;
136 
137 	state->stream = (lzma_stream)LZMA_STREAM_INIT;
138 	state->stream.next_out = state->compressed;
139 	state->stream.avail_out = state->compressed_buffer_size;
140 	if (a->archive.compression_code == ARCHIVE_COMPRESSION_XZ)
141 		ret = lzma_stream_encoder(&(state->stream),
142 		    state->lzmafilters, LZMA_CHECK_CRC64);
143 	else
144 		ret = lzma_alone_encoder(&(state->stream), &state->lzma_opt);
145 	if (ret == LZMA_OK)
146 		return (ARCHIVE_OK);
147 
148 	switch (ret) {
149 	case LZMA_MEM_ERROR:
150 		archive_set_error(&a->archive, ENOMEM,
151 		    "Internal error initializing compression library: "
152 		    "Cannot allocate memory");
153 		break;
154 	default:
155 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
156 		    "Internal error initializing compression library: "
157 		    "It's a bug in liblzma");
158 		break;
159 	}
160 	return (ARCHIVE_FATAL);
161 }
162 
163 /*
164  * Setup callback.
165  */
166 static int
archive_compressor_xz_init(struct archive_write * a)167 archive_compressor_xz_init(struct archive_write *a)
168 {
169 	int ret;
170 	struct private_data *state;
171 	struct private_config *config;
172 
173 	if (a->client_opener != NULL) {
174 		ret = (a->client_opener)(&a->archive, a->client_data);
175 		if (ret != ARCHIVE_OK)
176 			return (ret);
177 	}
178 
179 	state = (struct private_data *)malloc(sizeof(*state));
180 	if (state == NULL) {
181 		archive_set_error(&a->archive, ENOMEM,
182 		    "Can't allocate data for compression");
183 		return (ARCHIVE_FATAL);
184 	}
185 	memset(state, 0, sizeof(*state));
186 	config = a->compressor.config;
187 
188 	/*
189 	 * See comment above.  We should set compressed_buffer_size to
190 	 * max(bytes_per_block, 65536), but the code can't handle that yet.
191 	 */
192 	state->compressed_buffer_size = a->bytes_per_block;
193 	state->compressed = (unsigned char *)malloc(state->compressed_buffer_size);
194 	if (state->compressed == NULL) {
195 		archive_set_error(&a->archive, ENOMEM,
196 		    "Can't allocate data for compression buffer");
197 		free(state);
198 		return (ARCHIVE_FATAL);
199 	}
200 	a->compressor.write = archive_compressor_xz_write;
201 
202 	/* Initialize compression library. */
203 	if (lzma_lzma_preset(&state->lzma_opt, config->compression_level)) {
204 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
205 		    "Internal error initializing compression library");
206 		free(state->compressed);
207 		free(state);
208 	}
209 	state->lzmafilters[0].id = LZMA_FILTER_LZMA2;
210 	state->lzmafilters[0].options = &state->lzma_opt;
211 	state->lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */
212 	ret = archive_compressor_xz_init_stream(a, state);
213 	if (ret == LZMA_OK) {
214 		a->compressor.data = state;
215 		return (0);
216 	}
217 	/* Library setup failed: clean up. */
218 	free(state->compressed);
219 	free(state);
220 
221 	return (ARCHIVE_FATAL);
222 }
223 
224 /*
225  * Set write options.
226  */
227 static int
archive_compressor_xz_options(struct archive_write * a,const char * key,const char * value)228 archive_compressor_xz_options(struct archive_write *a, const char *key,
229     const char *value)
230 {
231 	struct private_config *config;
232 
233 	config = (struct private_config *)a->compressor.config;
234 	if (strcmp(key, "compression-level") == 0) {
235 		if (value == NULL || !(value[0] >= '0' && value[0] <= '9') ||
236 		    value[1] != '\0')
237 			return (ARCHIVE_WARN);
238 		config->compression_level = value[0] - '0';
239 		if (config->compression_level > 6)
240 			config->compression_level = 6;
241 		return (ARCHIVE_OK);
242 	}
243 
244 	return (ARCHIVE_WARN);
245 }
246 
247 /*
248  * Write data to the compressed stream.
249  */
250 static int
archive_compressor_xz_write(struct archive_write * a,const void * buff,size_t length)251 archive_compressor_xz_write(struct archive_write *a, const void *buff,
252     size_t length)
253 {
254 	struct private_data *state;
255 	int ret;
256 
257 	state = (struct private_data *)a->compressor.data;
258 	if (a->client_writer == NULL) {
259 		archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
260 		    "No write callback is registered?  "
261 		    "This is probably an internal programming error.");
262 		return (ARCHIVE_FATAL);
263 	}
264 
265 	/* Update statistics */
266 	state->total_in += length;
267 
268 	/* Compress input data to output buffer */
269 	state->stream.next_in = buff;
270 	state->stream.avail_in = length;
271 	if ((ret = drive_compressor(a, state, 0)) != ARCHIVE_OK)
272 		return (ret);
273 
274 	a->archive.file_position += length;
275 	return (ARCHIVE_OK);
276 }
277 
278 
279 /*
280  * Finish the compression...
281  */
282 static int
archive_compressor_xz_finish(struct archive_write * a)283 archive_compressor_xz_finish(struct archive_write *a)
284 {
285 	ssize_t block_length, target_block_length, bytes_written;
286 	int ret;
287 	struct private_data *state;
288 	unsigned tocopy;
289 
290 	ret = ARCHIVE_OK;
291 	state = (struct private_data *)a->compressor.data;
292 	if (state != NULL) {
293 		if (a->client_writer == NULL) {
294 			archive_set_error(&a->archive,
295 			    ARCHIVE_ERRNO_PROGRAMMER,
296 			    "No write callback is registered?  "
297 			    "This is probably an internal programming error.");
298 			ret = ARCHIVE_FATAL;
299 			goto cleanup;
300 		}
301 
302 		/* By default, always pad the uncompressed data. */
303 		if (a->pad_uncompressed) {
304 			tocopy = a->bytes_per_block -
305 			    (state->total_in % a->bytes_per_block);
306 			while (tocopy > 0 && tocopy < (unsigned)a->bytes_per_block) {
307 				state->stream.next_in = a->nulls;
308 				state->stream.avail_in = tocopy < a->null_length ?
309 				    tocopy : a->null_length;
310 				state->total_in += state->stream.avail_in;
311 				tocopy -= state->stream.avail_in;
312 				ret = drive_compressor(a, state, 0);
313 				if (ret != ARCHIVE_OK)
314 					goto cleanup;
315 			}
316 		}
317 
318 		/* Finish compression cycle */
319 		if (((ret = drive_compressor(a, state, 1))) != ARCHIVE_OK)
320 			goto cleanup;
321 
322 		/* Optionally, pad the final compressed block. */
323 		block_length = state->stream.next_out - state->compressed;
324 
325 		/* Tricky calculation to determine size of last block. */
326 		if (a->bytes_in_last_block <= 0)
327 			/* Default or Zero: pad to full block */
328 			target_block_length = a->bytes_per_block;
329 		else
330 			/* Round length to next multiple of bytes_in_last_block. */
331 			target_block_length = a->bytes_in_last_block *
332 			    ( (block_length + a->bytes_in_last_block - 1) /
333 				a->bytes_in_last_block);
334 		if (target_block_length > a->bytes_per_block)
335 			target_block_length = a->bytes_per_block;
336 		if (block_length < target_block_length) {
337 			memset(state->stream.next_out, 0,
338 			    target_block_length - block_length);
339 			block_length = target_block_length;
340 		}
341 
342 		/* Write the last block */
343 		bytes_written = (a->client_writer)(&a->archive, a->client_data,
344 		    state->compressed, block_length);
345 		if (bytes_written <= 0) {
346 			ret = ARCHIVE_FATAL;
347 			goto cleanup;
348 		}
349 		a->archive.raw_position += bytes_written;
350 
351 		/* Cleanup: shut down compressor, release memory, etc. */
352 	cleanup:
353 		lzma_end(&(state->stream));
354 		free(state->compressed);
355 		free(state);
356 	}
357 	free(a->compressor.config);
358 	a->compressor.config = NULL;
359 	return (ret);
360 }
361 
362 /*
363  * Utility function to push input data through compressor,
364  * writing full output blocks as necessary.
365  *
366  * Note that this handles both the regular write case (finishing ==
367  * false) and the end-of-archive case (finishing == true).
368  */
369 static int
drive_compressor(struct archive_write * a,struct private_data * state,int finishing)370 drive_compressor(struct archive_write *a, struct private_data *state, int finishing)
371 {
372 	ssize_t bytes_written;
373 	int ret;
374 
375 	for (;;) {
376 		if (state->stream.avail_out == 0) {
377 			bytes_written = (a->client_writer)(&a->archive,
378 			    a->client_data, state->compressed,
379 			    state->compressed_buffer_size);
380 			if (bytes_written <= 0) {
381 				/* TODO: Handle this write failure */
382 				return (ARCHIVE_FATAL);
383 			} else if ((size_t)bytes_written < state->compressed_buffer_size) {
384 				/* Short write: Move remaining to
385 				 * front of block and keep filling */
386 				memmove(state->compressed,
387 				    state->compressed + bytes_written,
388 				    state->compressed_buffer_size - bytes_written);
389 			}
390 			a->archive.raw_position += bytes_written;
391 			state->stream.next_out
392 			    = state->compressed +
393 			    state->compressed_buffer_size - bytes_written;
394 			state->stream.avail_out = bytes_written;
395 		}
396 
397 		/* If there's nothing to do, we're done. */
398 		if (!finishing && state->stream.avail_in == 0)
399 			return (ARCHIVE_OK);
400 
401 		ret = lzma_code(&(state->stream),
402 		    finishing ? LZMA_FINISH : LZMA_RUN );
403 
404 		switch (ret) {
405 		case LZMA_OK:
406 			/* In non-finishing case, check if compressor
407 			 * consumed everything */
408 			if (!finishing && state->stream.avail_in == 0)
409 				return (ARCHIVE_OK);
410 			/* In finishing case, this return always means
411 			 * there's more work */
412 			break;
413 		case LZMA_STREAM_END:
414 			/* This return can only occur in finishing case. */
415 			if (finishing)
416 				return (ARCHIVE_OK);
417 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
418 			    "lzma compression data error");
419 			return (ARCHIVE_FATAL);
420 		case LZMA_MEMLIMIT_ERROR:
421 			archive_set_error(&a->archive, ENOMEM,
422 			    "lzma compression error: "
423 			    "%ju MiB would have been needed",
424 			    (lzma_memusage(&(state->stream)) + 1024 * 1024 -1)
425 			    / (1024 * 1024));
426 			return (ARCHIVE_FATAL);
427 		default:
428 			/* Any other return value indicates an error. */
429 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
430 			    "lzma compression failed:"
431 			    " lzma_code() call returned status %d",
432 			    ret);
433 			return (ARCHIVE_FATAL);
434 		}
435 	}
436 }
437 
438 #endif /* HAVE_LZMA_H */
439