archive_write_set_compression_gzip.c revision 267654
1169689Skan/*-
2169689Skan * Copyright (c) 2003-2007 Tim Kientzle
3169689Skan * All rights reserved.
4169689Skan *
5169689Skan * Redistribution and use in source and binary forms, with or without
6169689Skan * modification, are permitted provided that the following conditions
7169689Skan * are met:
8169689Skan * 1. Redistributions of source code must retain the above copyright
9169689Skan *    notice, this list of conditions and the following disclaimer.
10169689Skan * 2. Redistributions in binary form must reproduce the above copyright
11169689Skan *    notice, this list of conditions and the following disclaimer in the
12169689Skan *    documentation and/or other materials provided with the distribution.
13169689Skan *
14169689Skan * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
15169689Skan * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
16169689Skan * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
17169689Skan * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
18169689Skan * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
19169689Skan * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
20169689Skan * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
21169689Skan * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22169689Skan * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
23169689Skan * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
24169689Skan */
25169689Skan
26169689Skan#include "archive_platform.h"
27169689Skan
28169689Skan__FBSDID("$FreeBSD: releng/9.3/contrib/libarchive/libarchive/archive_write_set_compression_gzip.c 229592 2012-01-05 12:06:54Z mm $");
29169689Skan
30169689Skan#ifdef HAVE_ERRNO_H
31169689Skan#include <errno.h>
32169689Skan#endif
33169689Skan#ifdef HAVE_STDLIB_H
34169689Skan#include <stdlib.h>
35169689Skan#endif
36169689Skan#ifdef HAVE_STRING_H
37169689Skan#include <string.h>
38169689Skan#endif
39169689Skan#include <time.h>
40169689Skan#ifdef HAVE_ZLIB_H
41169689Skan#include <zlib.h>
42169689Skan#endif
43169689Skan
44169689Skan#include "archive.h"
45169689Skan#include "archive_private.h"
46169689Skan#include "archive_write_private.h"
47169689Skan
48169689Skan#ifndef HAVE_ZLIB_H
49169689Skanint
50169689Skanarchive_write_set_compression_gzip(struct archive *a)
51169689Skan{
52169689Skan	archive_set_error(a, ARCHIVE_ERRNO_MISC,
53169689Skan	    "gzip compression not supported on this platform");
54169689Skan	return (ARCHIVE_FATAL);
55169689Skan}
56169689Skan#else
57169689Skan/* Don't compile this if we don't have zlib. */
58169689Skan
59169689Skanstruct private_data {
60169689Skan	z_stream	 stream;
61169689Skan	int64_t		 total_in;
62169689Skan	unsigned char	*compressed;
63169689Skan	size_t		 compressed_buffer_size;
64169689Skan	unsigned long	 crc;
65169689Skan};
66169689Skan
67169689Skanstruct private_config {
68169689Skan	int		 compression_level;
69169689Skan};
70169689Skan
71169689Skan
72169689Skan/*
73169689Skan * Yuck.  zlib.h is not const-correct, so I need this one bit
74169689Skan * of ugly hackery to convert a const * pointer to a non-const pointer.
75169689Skan */
76169689Skan#define	SET_NEXT_IN(st,src)					\
77169689Skan	(st)->stream.next_in = (Bytef *)(uintptr_t)(const void *)(src)
78169689Skan
79169689Skanstatic int	archive_compressor_gzip_finish(struct archive_write *);
80169689Skanstatic int	archive_compressor_gzip_init(struct archive_write *);
81169689Skanstatic int	archive_compressor_gzip_options(struct archive_write *,
82169689Skan		    const char *, const char *);
83169689Skanstatic int	archive_compressor_gzip_write(struct archive_write *,
84169689Skan		    const void *, size_t);
85169689Skanstatic int	drive_compressor(struct archive_write *, struct private_data *,
86169689Skan		    int finishing);
87169689Skan
88169689Skan
89169689Skan/*
90169689Skan * Allocate, initialize and return a archive object.
91169689Skan */
92169689Skanint
93169689Skanarchive_write_set_compression_gzip(struct archive *_a)
94169689Skan{
95169689Skan	struct archive_write *a = (struct archive_write *)_a;
96169689Skan	struct private_config *config;
97169689Skan	__archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC,
98169689Skan	    ARCHIVE_STATE_NEW, "archive_write_set_compression_gzip");
99169689Skan	config = malloc(sizeof(*config));
100169689Skan	if (config == NULL) {
101169689Skan		archive_set_error(&a->archive, ENOMEM, "Out of memory");
102169689Skan		return (ARCHIVE_FATAL);
103169689Skan	}
104169689Skan	a->compressor.config = config;
105169689Skan	a->compressor.finish = &archive_compressor_gzip_finish;
106169689Skan	config->compression_level = Z_DEFAULT_COMPRESSION;
107169689Skan	a->compressor.init = &archive_compressor_gzip_init;
108169689Skan	a->compressor.options = &archive_compressor_gzip_options;
109169689Skan	a->archive.compression_code = ARCHIVE_COMPRESSION_GZIP;
110169689Skan	a->archive.compression_name = "gzip";
111169689Skan	return (ARCHIVE_OK);
112169689Skan}
113169689Skan
114169689Skan/*
115169689Skan * Setup callback.
116169689Skan */
117169689Skanstatic int
118169689Skanarchive_compressor_gzip_init(struct archive_write *a)
119169689Skan{
120169689Skan	int ret;
121169689Skan	struct private_data *state;
122169689Skan	struct private_config *config;
123169689Skan	time_t t;
124169689Skan
125169689Skan	config = (struct private_config *)a->compressor.config;
126169689Skan
127169689Skan	if (a->client_opener != NULL) {
128169689Skan		ret = (a->client_opener)(&a->archive, a->client_data);
129169689Skan		if (ret != ARCHIVE_OK)
130169689Skan			return (ret);
131169689Skan	}
132169689Skan
133169689Skan	/*
134169689Skan	 * The next check is a temporary workaround until the gzip
135169689Skan	 * code can be overhauled some.  The code should not require
136169689Skan	 * that compressed_buffer_size == bytes_per_block.  Removing
137169689Skan	 * this assumption will allow us to compress larger chunks at
138169689Skan	 * a time, which should improve overall performance
139169689Skan	 * marginally.  As a minor side-effect, such a cleanup would
140169689Skan	 * allow us to support truly arbitrary block sizes.
141169689Skan	 */
142169689Skan	if (a->bytes_per_block < 10) {
143169689Skan		archive_set_error(&a->archive, EINVAL,
144169689Skan		    "GZip compressor requires a minimum 10 byte block size");
145169689Skan		return (ARCHIVE_FATAL);
146169689Skan	}
147169689Skan
148169689Skan	state = (struct private_data *)malloc(sizeof(*state));
149169689Skan	if (state == NULL) {
150169689Skan		archive_set_error(&a->archive, ENOMEM,
151169689Skan		    "Can't allocate data for compression");
152169689Skan		return (ARCHIVE_FATAL);
153169689Skan	}
154169689Skan	memset(state, 0, sizeof(*state));
155169689Skan
156169689Skan	/*
157169689Skan	 * See comment above.  We should set compressed_buffer_size to
158169689Skan	 * max(bytes_per_block, 65536), but the code can't handle that yet.
159169689Skan	 */
160169689Skan	state->compressed_buffer_size = a->bytes_per_block;
161169689Skan	state->compressed = (unsigned char *)malloc(state->compressed_buffer_size);
162169689Skan	state->crc = crc32(0L, NULL, 0);
163169689Skan
164169689Skan	if (state->compressed == NULL) {
165169689Skan		archive_set_error(&a->archive, ENOMEM,
166169689Skan		    "Can't allocate data for compression buffer");
167169689Skan		free(state);
168169689Skan		return (ARCHIVE_FATAL);
169169689Skan	}
170169689Skan
171169689Skan	state->stream.next_out = state->compressed;
172169689Skan	state->stream.avail_out = state->compressed_buffer_size;
173169689Skan
174169689Skan	/* Prime output buffer with a gzip header. */
175169689Skan	t = time(NULL);
176169689Skan	state->compressed[0] = 0x1f; /* GZip signature bytes */
177169689Skan	state->compressed[1] = 0x8b;
178169689Skan	state->compressed[2] = 0x08; /* "Deflate" compression */
179169689Skan	state->compressed[3] = 0; /* No options */
180169689Skan	state->compressed[4] = (t)&0xff;  /* Timestamp */
181169689Skan	state->compressed[5] = (t>>8)&0xff;
182169689Skan	state->compressed[6] = (t>>16)&0xff;
183169689Skan	state->compressed[7] = (t>>24)&0xff;
184169689Skan	state->compressed[8] = 0; /* No deflate options */
185169689Skan	state->compressed[9] = 3; /* OS=Unix */
186169689Skan	state->stream.next_out += 10;
187169689Skan	state->stream.avail_out -= 10;
188169689Skan
189169689Skan	a->compressor.write = archive_compressor_gzip_write;
190169689Skan
191169689Skan	/* Initialize compression library. */
192169689Skan	ret = deflateInit2(&(state->stream),
193169689Skan	    config->compression_level,
194169689Skan	    Z_DEFLATED,
195169689Skan	    -15 /* < 0 to suppress zlib header */,
196169689Skan	    8,
197169689Skan	    Z_DEFAULT_STRATEGY);
198169689Skan
199169689Skan	if (ret == Z_OK) {
200169689Skan		a->compressor.data = state;
201169689Skan		return (0);
202169689Skan	}
203169689Skan
204169689Skan	/* Library setup failed: clean up. */
205169689Skan	archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal error "
206169689Skan	    "initializing compression library");
207169689Skan	free(state->compressed);
208169689Skan	free(state);
209169689Skan
210169689Skan	/* Override the error message if we know what really went wrong. */
211169689Skan	switch (ret) {
212169689Skan	case Z_STREAM_ERROR:
213169689Skan		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
214169689Skan		    "Internal error initializing "
215169689Skan		    "compression library: invalid setup parameter");
216169689Skan		break;
217169689Skan	case Z_MEM_ERROR:
218169689Skan		archive_set_error(&a->archive, ENOMEM, "Internal error initializing "
219169689Skan		    "compression library");
220169689Skan		break;
221169689Skan	case Z_VERSION_ERROR:
222169689Skan		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
223169689Skan		    "Internal error initializing "
224169689Skan		    "compression library: invalid library version");
225169689Skan		break;
226169689Skan	}
227169689Skan
228169689Skan	return (ARCHIVE_FATAL);
229169689Skan}
230169689Skan
231169689Skan/*
232169689Skan * Set write options.
233169689Skan */
234169689Skanstatic int
235169689Skanarchive_compressor_gzip_options(struct archive_write *a, const char *key,
236169689Skan    const char *value)
237169689Skan{
238169689Skan	struct private_config *config;
239169689Skan
240169689Skan	config = (struct private_config *)a->compressor.config;
241169689Skan	if (strcmp(key, "compression-level") == 0) {
242169689Skan		if (value == NULL || !(value[0] >= '0' && value[0] <= '9') ||
243169689Skan		    value[1] != '\0')
244169689Skan			return (ARCHIVE_WARN);
245169689Skan		config->compression_level = value[0] - '0';
246169689Skan		return (ARCHIVE_OK);
247169689Skan	}
248169689Skan
249169689Skan	return (ARCHIVE_WARN);
250169689Skan}
251169689Skan
252169689Skan/*
253169689Skan * Write data to the compressed stream.
254169689Skan */
255169689Skanstatic int
256169689Skanarchive_compressor_gzip_write(struct archive_write *a, const void *buff,
257169689Skan    size_t length)
258169689Skan{
259169689Skan	struct private_data *state;
260169689Skan	int ret;
261169689Skan
262169689Skan	state = (struct private_data *)a->compressor.data;
263169689Skan	if (a->client_writer == NULL) {
264169689Skan		archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
265169689Skan		    "No write callback is registered?  "
266169689Skan		    "This is probably an internal programming error.");
267169689Skan		return (ARCHIVE_FATAL);
268169689Skan	}
269169689Skan
270169689Skan	/* Update statistics */
271169689Skan	state->crc = crc32(state->crc, (const Bytef *)buff, length);
272169689Skan	state->total_in += length;
273169689Skan
274169689Skan	/* Compress input data to output buffer */
275169689Skan	SET_NEXT_IN(state, buff);
276169689Skan	state->stream.avail_in = length;
277169689Skan	if ((ret = drive_compressor(a, state, 0)) != ARCHIVE_OK)
278169689Skan		return (ret);
279169689Skan
280169689Skan	a->archive.file_position += length;
281169689Skan	return (ARCHIVE_OK);
282169689Skan}
283169689Skan
284169689Skan/*
285169689Skan * Finish the compression...
286169689Skan */
287169689Skanstatic int
288169689Skanarchive_compressor_gzip_finish(struct archive_write *a)
289169689Skan{
290169689Skan	ssize_t block_length, target_block_length, bytes_written;
291169689Skan	int ret;
292169689Skan	struct private_data *state;
293169689Skan	unsigned tocopy;
294169689Skan	unsigned char trailer[8];
295169689Skan
296169689Skan	state = (struct private_data *)a->compressor.data;
297169689Skan	ret = 0;
298169689Skan	if (state != NULL) {
299169689Skan		if (a->client_writer == NULL) {
300169689Skan			archive_set_error(&a->archive,
301169689Skan			    ARCHIVE_ERRNO_PROGRAMMER,
302169689Skan			    "No write callback is registered?  "
303169689Skan			    "This is probably an internal programming error.");
304169689Skan			ret = ARCHIVE_FATAL;
305169689Skan			goto cleanup;
306169689Skan		}
307169689Skan
308169689Skan		/* By default, always pad the uncompressed data. */
309169689Skan		if (a->pad_uncompressed) {
310169689Skan			tocopy = a->bytes_per_block -
311169689Skan			    (state->total_in % a->bytes_per_block);
312169689Skan			while (tocopy > 0 && tocopy < (unsigned)a->bytes_per_block) {
313169689Skan				SET_NEXT_IN(state, a->nulls);
314169689Skan				state->stream.avail_in = tocopy < a->null_length ?
315169689Skan				    tocopy : a->null_length;
316169689Skan				state->crc = crc32(state->crc, a->nulls,
317169689Skan				    state->stream.avail_in);
318169689Skan				state->total_in += state->stream.avail_in;
319169689Skan				tocopy -= state->stream.avail_in;
320169689Skan				ret = drive_compressor(a, state, 0);
321169689Skan				if (ret != ARCHIVE_OK)
322169689Skan					goto cleanup;
323169689Skan			}
324169689Skan		}
325169689Skan
326169689Skan		/* Finish compression cycle */
327169689Skan		if (((ret = drive_compressor(a, state, 1))) != ARCHIVE_OK)
328169689Skan			goto cleanup;
329169689Skan
330169689Skan		/* Build trailer: 4-byte CRC and 4-byte length. */
331169689Skan		trailer[0] = (state->crc)&0xff;
332169689Skan		trailer[1] = (state->crc >> 8)&0xff;
333169689Skan		trailer[2] = (state->crc >> 16)&0xff;
334169689Skan		trailer[3] = (state->crc >> 24)&0xff;
335169689Skan		trailer[4] = (state->total_in)&0xff;
336169689Skan		trailer[5] = (state->total_in >> 8)&0xff;
337169689Skan		trailer[6] = (state->total_in >> 16)&0xff;
338169689Skan		trailer[7] = (state->total_in >> 24)&0xff;
339169689Skan
340169689Skan		/* Add trailer to current block. */
341169689Skan		tocopy = 8;
342169689Skan		if (tocopy > state->stream.avail_out)
343169689Skan			tocopy = state->stream.avail_out;
344169689Skan		memcpy(state->stream.next_out, trailer, tocopy);
345169689Skan		state->stream.next_out += tocopy;
346169689Skan		state->stream.avail_out -= tocopy;
347169689Skan
348169689Skan		/* If it overflowed, flush and start a new block. */
349169689Skan		if (tocopy < 8) {
350169689Skan			bytes_written = (a->client_writer)(&a->archive, a->client_data,
351169689Skan			    state->compressed, state->compressed_buffer_size);
352169689Skan			if (bytes_written <= 0) {
353169689Skan				ret = ARCHIVE_FATAL;
354169689Skan				goto cleanup;
355169689Skan			}
356169689Skan			a->archive.raw_position += bytes_written;
357169689Skan			state->stream.next_out = state->compressed;
358169689Skan			state->stream.avail_out = state->compressed_buffer_size;
359169689Skan			memcpy(state->stream.next_out, trailer + tocopy, 8-tocopy);
360169689Skan			state->stream.next_out += 8-tocopy;
361169689Skan			state->stream.avail_out -= 8-tocopy;
362169689Skan		}
363169689Skan
364169689Skan		/* Optionally, pad the final compressed block. */
365169689Skan		block_length = state->stream.next_out - state->compressed;
366169689Skan
367169689Skan		/* Tricky calculation to determine size of last block. */
368169689Skan		if (a->bytes_in_last_block <= 0)
369169689Skan			/* Default or Zero: pad to full block */
370169689Skan			target_block_length = a->bytes_per_block;
371169689Skan		else
372169689Skan			/* Round length to next multiple of bytes_in_last_block. */
373169689Skan			target_block_length = a->bytes_in_last_block *
374169689Skan			    ( (block_length + a->bytes_in_last_block - 1) /
375169689Skan				a->bytes_in_last_block);
376169689Skan		if (target_block_length > a->bytes_per_block)
377169689Skan			target_block_length = a->bytes_per_block;
378169689Skan		if (block_length < target_block_length) {
379169689Skan			memset(state->stream.next_out, 0,
380169689Skan			    target_block_length - block_length);
381169689Skan			block_length = target_block_length;
382169689Skan		}
383169689Skan
384169689Skan		/* Write the last block */
385169689Skan		bytes_written = (a->client_writer)(&a->archive, a->client_data,
386169689Skan		    state->compressed, block_length);
387169689Skan		if (bytes_written <= 0) {
388169689Skan			ret = ARCHIVE_FATAL;
389169689Skan			goto cleanup;
390169689Skan		}
391169689Skan		a->archive.raw_position += bytes_written;
392169689Skan
393169689Skan		/* Cleanup: shut down compressor, release memory, etc. */
394169689Skan	cleanup:
395169689Skan		switch (deflateEnd(&(state->stream))) {
396169689Skan		case Z_OK:
397169689Skan			break;
398169689Skan		default:
399169689Skan			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
400169689Skan			    "Failed to clean up compressor");
401169689Skan			ret = ARCHIVE_FATAL;
402169689Skan		}
403169689Skan		free(state->compressed);
404169689Skan		free(state);
405169689Skan	}
406169689Skan	/* Clean up config area even if we never initialized. */
407169689Skan	free(a->compressor.config);
408169689Skan	a->compressor.config = NULL;
409169689Skan	return (ret);
410169689Skan}
411169689Skan
412169689Skan/*
413169689Skan * Utility function to push input data through compressor,
414169689Skan * writing full output blocks as necessary.
415169689Skan *
416169689Skan * Note that this handles both the regular write case (finishing ==
417169689Skan * false) and the end-of-archive case (finishing == true).
418169689Skan */
419169689Skanstatic int
420169689Skandrive_compressor(struct archive_write *a, struct private_data *state, int finishing)
421169689Skan{
422169689Skan	ssize_t bytes_written;
423169689Skan	int ret;
424169689Skan
425169689Skan	for (;;) {
426169689Skan		if (state->stream.avail_out == 0) {
427169689Skan			bytes_written = (a->client_writer)(&a->archive,
428169689Skan			    a->client_data, state->compressed,
429169689Skan			    state->compressed_buffer_size);
430169689Skan			if (bytes_written <= 0) {
431169689Skan				/* TODO: Handle this write failure */
432169689Skan				return (ARCHIVE_FATAL);
433169689Skan			} else if ((size_t)bytes_written < state->compressed_buffer_size) {
434169689Skan				/* Short write: Move remaining to
435169689Skan				 * front of block and keep filling */
436169689Skan				memmove(state->compressed,
437169689Skan				    state->compressed + bytes_written,
438169689Skan				    state->compressed_buffer_size - bytes_written);
439169689Skan			}
440169689Skan			a->archive.raw_position += bytes_written;
441169689Skan			state->stream.next_out
442169689Skan			    = state->compressed +
443169689Skan			    state->compressed_buffer_size - bytes_written;
444169689Skan			state->stream.avail_out = bytes_written;
445169689Skan		}
446169689Skan
447169689Skan		/* If there's nothing to do, we're done. */
448169689Skan		if (!finishing && state->stream.avail_in == 0)
449169689Skan			return (ARCHIVE_OK);
450169689Skan
451169689Skan		ret = deflate(&(state->stream),
452169689Skan		    finishing ? Z_FINISH : Z_NO_FLUSH );
453169689Skan
454169689Skan		switch (ret) {
455169689Skan		case Z_OK:
456169689Skan			/* In non-finishing case, check if compressor
457169689Skan			 * consumed everything */
458169689Skan			if (!finishing && state->stream.avail_in == 0)
459169689Skan				return (ARCHIVE_OK);
460169689Skan			/* In finishing case, this return always means
461169689Skan			 * there's more work */
462169689Skan			break;
463169689Skan		case Z_STREAM_END:
464169689Skan			/* This return can only occur in finishing case. */
465169689Skan			return (ARCHIVE_OK);
466169689Skan		default:
467169689Skan			/* Any other return value indicates an error. */
468169689Skan			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
469169689Skan			    "GZip compression failed:"
470169689Skan			    " deflate() call returned status %d",
471169689Skan			    ret);
472169689Skan			return (ARCHIVE_FATAL);
473169689Skan		}
474169689Skan	}
475169689Skan}
476169689Skan
477169689Skan#endif /* HAVE_ZLIB_H */
478169689Skan