1228753Smm/*-
2228753Smm * Copyright (c) 2003-2007 Tim Kientzle
3228753Smm * All rights reserved.
4228753Smm *
5228753Smm * Redistribution and use in source and binary forms, with or without
6228753Smm * modification, are permitted provided that the following conditions
7228753Smm * are met:
8228753Smm * 1. Redistributions of source code must retain the above copyright
9228753Smm *    notice, this list of conditions and the following disclaimer.
10228753Smm * 2. Redistributions in binary form must reproduce the above copyright
11228753Smm *    notice, this list of conditions and the following disclaimer in the
12228753Smm *    documentation and/or other materials provided with the distribution.
13228753Smm *
14228753Smm * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
15228753Smm * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
16228753Smm * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
17228753Smm * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
18228753Smm * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
19228753Smm * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
20228753Smm * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
21228753Smm * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22228753Smm * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
23228753Smm * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
24228753Smm */
25228753Smm
26228753Smm#include "archive_platform.h"
27228753Smm
28229592Smm__FBSDID("$FreeBSD$");
29228753Smm
30228753Smm#ifdef HAVE_ERRNO_H
31228753Smm#include <errno.h>
32228753Smm#endif
33228753Smm#include <stdio.h>
34228753Smm#ifdef HAVE_STDLIB_H
35228753Smm#include <stdlib.h>
36228753Smm#endif
37228753Smm#ifdef HAVE_STRING_H
38228753Smm#include <string.h>
39228753Smm#endif
40228753Smm#ifdef HAVE_BZLIB_H
41228753Smm#include <bzlib.h>
42228753Smm#endif
43228753Smm
44228753Smm#include "archive.h"
45228753Smm#include "archive_private.h"
46228753Smm#include "archive_write_private.h"
47228753Smm
48228753Smm#if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR)
49228753Smmint
50228753Smmarchive_write_set_compression_bzip2(struct archive *a)
51228753Smm{
52228753Smm	archive_set_error(a, ARCHIVE_ERRNO_MISC,
53228753Smm	    "bzip2 compression not supported on this platform");
54228753Smm	return (ARCHIVE_FATAL);
55228753Smm}
56228753Smm#else
57228753Smm/* Don't compile this if we don't have bzlib. */
58228753Smm
59228753Smmstruct private_data {
60228753Smm	bz_stream	 stream;
61228753Smm	int64_t		 total_in;
62228753Smm	char		*compressed;
63228753Smm	size_t		 compressed_buffer_size;
64228753Smm};
65228753Smm
66228753Smmstruct private_config {
67228753Smm	int		 compression_level;
68228753Smm};
69228753Smm
70228753Smm/*
71228753Smm * Yuck.  bzlib.h is not const-correct, so I need this one bit
72228753Smm * of ugly hackery to convert a const * pointer to a non-const pointer.
73228753Smm */
74228753Smm#define	SET_NEXT_IN(st,src)					\
75228753Smm	(st)->stream.next_in = (char *)(uintptr_t)(const void *)(src)
76228753Smm
77228753Smmstatic int	archive_compressor_bzip2_finish(struct archive_write *);
78228753Smmstatic int	archive_compressor_bzip2_init(struct archive_write *);
79228753Smmstatic int	archive_compressor_bzip2_options(struct archive_write *,
80228753Smm		    const char *, const char *);
81228753Smmstatic int	archive_compressor_bzip2_write(struct archive_write *,
82228753Smm		    const void *, size_t);
83228753Smmstatic int	drive_compressor(struct archive_write *, struct private_data *,
84228753Smm		    int finishing);
85228753Smm
86228753Smm/*
87228753Smm * Allocate, initialize and return an archive object.
88228753Smm */
89228753Smmint
90228753Smmarchive_write_set_compression_bzip2(struct archive *_a)
91228753Smm{
92228753Smm	struct archive_write *a = (struct archive_write *)_a;
93228753Smm	struct private_config *config;
94228753Smm	__archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC,
95228753Smm	    ARCHIVE_STATE_NEW, "archive_write_set_compression_bzip2");
96228753Smm	config = malloc(sizeof(*config));
97228753Smm	if (config == NULL) {
98228753Smm		archive_set_error(&a->archive, ENOMEM, "Out of memory");
99228753Smm		return (ARCHIVE_FATAL);
100228753Smm	}
101228753Smm	a->compressor.config = config;
102228753Smm	a->compressor.finish = archive_compressor_bzip2_finish;
103228753Smm	config->compression_level = 9; /* default */
104228753Smm	a->compressor.init = &archive_compressor_bzip2_init;
105228753Smm	a->compressor.options = &archive_compressor_bzip2_options;
106228753Smm	a->archive.compression_code = ARCHIVE_COMPRESSION_BZIP2;
107228753Smm	a->archive.compression_name = "bzip2";
108228753Smm	return (ARCHIVE_OK);
109228753Smm}
110228753Smm
111228753Smm/*
112228753Smm * Setup callback.
113228753Smm */
114228753Smmstatic int
115228753Smmarchive_compressor_bzip2_init(struct archive_write *a)
116228753Smm{
117228753Smm	int ret;
118228753Smm	struct private_data *state;
119228753Smm	struct private_config *config;
120228753Smm
121228753Smm	config = (struct private_config *)a->compressor.config;
122228753Smm	if (a->client_opener != NULL) {
123228753Smm		ret = (a->client_opener)(&a->archive, a->client_data);
124228753Smm		if (ret != 0)
125228753Smm			return (ret);
126228753Smm	}
127228753Smm
128228753Smm	state = (struct private_data *)malloc(sizeof(*state));
129228753Smm	if (state == NULL) {
130228753Smm		archive_set_error(&a->archive, ENOMEM,
131228753Smm		    "Can't allocate data for compression");
132228753Smm		return (ARCHIVE_FATAL);
133228753Smm	}
134228753Smm	memset(state, 0, sizeof(*state));
135228753Smm
136228753Smm	state->compressed_buffer_size = a->bytes_per_block;
137228753Smm	state->compressed = (char *)malloc(state->compressed_buffer_size);
138228753Smm
139228753Smm	if (state->compressed == NULL) {
140228753Smm		archive_set_error(&a->archive, ENOMEM,
141228753Smm		    "Can't allocate data for compression buffer");
142228753Smm		free(state);
143228753Smm		return (ARCHIVE_FATAL);
144228753Smm	}
145228753Smm
146228753Smm	state->stream.next_out = state->compressed;
147228753Smm	state->stream.avail_out = state->compressed_buffer_size;
148228753Smm	a->compressor.write = archive_compressor_bzip2_write;
149228753Smm
150228753Smm	/* Initialize compression library */
151228753Smm	ret = BZ2_bzCompressInit(&(state->stream),
152228753Smm	    config->compression_level, 0, 30);
153228753Smm	if (ret == BZ_OK) {
154228753Smm		a->compressor.data = state;
155228753Smm		return (ARCHIVE_OK);
156228753Smm	}
157228753Smm
158228753Smm	/* Library setup failed: clean up. */
159228753Smm	archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
160228753Smm	    "Internal error initializing compression library");
161228753Smm	free(state->compressed);
162228753Smm	free(state);
163228753Smm
164228753Smm	/* Override the error message if we know what really went wrong. */
165228753Smm	switch (ret) {
166228753Smm	case BZ_PARAM_ERROR:
167228753Smm		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
168228753Smm		    "Internal error initializing compression library: "
169228753Smm		    "invalid setup parameter");
170228753Smm		break;
171228753Smm	case BZ_MEM_ERROR:
172228753Smm		archive_set_error(&a->archive, ENOMEM,
173228753Smm		    "Internal error initializing compression library: "
174228753Smm		    "out of memory");
175228753Smm		break;
176228753Smm	case BZ_CONFIG_ERROR:
177228753Smm		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
178228753Smm		    "Internal error initializing compression library: "
179228753Smm		    "mis-compiled library");
180228753Smm		break;
181228753Smm	}
182228753Smm
183228753Smm	return (ARCHIVE_FATAL);
184228753Smm
185228753Smm}
186228753Smm
187228753Smm/*
188228753Smm * Set write options.
189228753Smm */
190228753Smmstatic int
191228753Smmarchive_compressor_bzip2_options(struct archive_write *a, const char *key,
192228753Smm    const char *value)
193228753Smm{
194228753Smm	struct private_config *config;
195228753Smm
196228753Smm	config = (struct private_config *)a->compressor.config;
197228753Smm	if (strcmp(key, "compression-level") == 0) {
198228753Smm		if (value == NULL || !(value[0] >= '0' && value[0] <= '9') ||
199228753Smm		    value[1] != '\0')
200228753Smm			return (ARCHIVE_WARN);
201228753Smm		config->compression_level = value[0] - '0';
202228753Smm		/* Make '0' be a synonym for '1'. */
203228753Smm		/* This way, bzip2 compressor supports the same 0..9
204228753Smm		 * range of levels as gzip. */
205228753Smm		if (config->compression_level < 1)
206228753Smm			config->compression_level = 1;
207228753Smm		return (ARCHIVE_OK);
208228753Smm	}
209228753Smm
210228753Smm	return (ARCHIVE_WARN);
211228753Smm}
212228753Smm
213228753Smm/*
214228753Smm * Write data to the compressed stream.
215228753Smm *
216228753Smm * Returns ARCHIVE_OK if all data written, error otherwise.
217228753Smm */
218228753Smmstatic int
219228753Smmarchive_compressor_bzip2_write(struct archive_write *a, const void *buff,
220228753Smm    size_t length)
221228753Smm{
222228753Smm	struct private_data *state;
223228753Smm
224228753Smm	state = (struct private_data *)a->compressor.data;
225228753Smm	if (a->client_writer == NULL) {
226228753Smm		archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
227228753Smm		    "No write callback is registered?  "
228228753Smm		    "This is probably an internal programming error.");
229228753Smm		return (ARCHIVE_FATAL);
230228753Smm	}
231228753Smm
232228753Smm	/* Update statistics */
233228753Smm	state->total_in += length;
234228753Smm
235228753Smm	/* Compress input data to output buffer */
236228753Smm	SET_NEXT_IN(state, buff);
237228753Smm	state->stream.avail_in = length;
238228753Smm	if (drive_compressor(a, state, 0))
239228753Smm		return (ARCHIVE_FATAL);
240228753Smm	a->archive.file_position += length;
241228753Smm	return (ARCHIVE_OK);
242228753Smm}
243228753Smm
244228753Smm
245228753Smm/*
246228753Smm * Finish the compression.
247228753Smm */
248228753Smmstatic int
249228753Smmarchive_compressor_bzip2_finish(struct archive_write *a)
250228753Smm{
251228753Smm	ssize_t block_length;
252228753Smm	int ret;
253228753Smm	struct private_data *state;
254228753Smm	ssize_t target_block_length;
255228753Smm	ssize_t bytes_written;
256228753Smm	unsigned tocopy;
257228753Smm
258228753Smm	ret = ARCHIVE_OK;
259228753Smm	state = (struct private_data *)a->compressor.data;
260228753Smm	if (state != NULL) {
261228753Smm		if (a->client_writer == NULL) {
262228753Smm			archive_set_error(&a->archive,
263228753Smm			    ARCHIVE_ERRNO_PROGRAMMER,
264228753Smm			    "No write callback is registered?\n"
265228753Smm			    "This is probably an internal programming error.");
266228753Smm			ret = ARCHIVE_FATAL;
267228753Smm			goto cleanup;
268228753Smm		}
269228753Smm
270228753Smm		/* By default, always pad the uncompressed data. */
271228753Smm		if (a->pad_uncompressed) {
272228753Smm			tocopy = a->bytes_per_block -
273228753Smm			    (state->total_in % a->bytes_per_block);
274228753Smm			while (tocopy > 0 && tocopy < (unsigned)a->bytes_per_block) {
275228753Smm				SET_NEXT_IN(state, a->nulls);
276228753Smm				state->stream.avail_in = tocopy < a->null_length ?
277228753Smm				    tocopy : a->null_length;
278228753Smm				state->total_in += state->stream.avail_in;
279228753Smm				tocopy -= state->stream.avail_in;
280228753Smm				ret = drive_compressor(a, state, 0);
281228753Smm				if (ret != ARCHIVE_OK)
282228753Smm					goto cleanup;
283228753Smm			}
284228753Smm		}
285228753Smm
286228753Smm		/* Finish compression cycle. */
287228753Smm		if ((ret = drive_compressor(a, state, 1)))
288228753Smm			goto cleanup;
289228753Smm
290228753Smm		/* Optionally, pad the final compressed block. */
291228753Smm		block_length = state->stream.next_out - state->compressed;
292228753Smm
293228753Smm		/* Tricky calculation to determine size of last block. */
294228753Smm		if (a->bytes_in_last_block <= 0)
295228753Smm			/* Default or Zero: pad to full block */
296228753Smm			target_block_length = a->bytes_per_block;
297228753Smm		else
298228753Smm			/* Round length to next multiple of bytes_in_last_block. */
299228753Smm			target_block_length = a->bytes_in_last_block *
300228753Smm			    ( (block_length + a->bytes_in_last_block - 1) /
301228753Smm				a->bytes_in_last_block);
302228753Smm		if (target_block_length > a->bytes_per_block)
303228753Smm			target_block_length = a->bytes_per_block;
304228753Smm		if (block_length < target_block_length) {
305228753Smm			memset(state->stream.next_out, 0,
306228753Smm			    target_block_length - block_length);
307228753Smm			block_length = target_block_length;
308228753Smm		}
309228753Smm
310228753Smm		/* Write the last block */
311228753Smm		bytes_written = (a->client_writer)(&a->archive, a->client_data,
312228753Smm		    state->compressed, block_length);
313228753Smm
314228753Smm		/* TODO: Handle short write of final block. */
315228753Smm		if (bytes_written <= 0)
316228753Smm			ret = ARCHIVE_FATAL;
317228753Smm		else {
318228753Smm			a->archive.raw_position += ret;
319228753Smm			ret = ARCHIVE_OK;
320228753Smm		}
321228753Smm
322228753Smm		/* Cleanup: shut down compressor, release memory, etc. */
323228753Smmcleanup:
324228753Smm		switch (BZ2_bzCompressEnd(&(state->stream))) {
325228753Smm		case BZ_OK:
326228753Smm			break;
327228753Smm		default:
328228753Smm			archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
329228753Smm			    "Failed to clean up compressor");
330228753Smm			ret = ARCHIVE_FATAL;
331228753Smm		}
332228753Smm
333228753Smm		free(state->compressed);
334228753Smm		free(state);
335228753Smm	}
336228753Smm	/* Free configuration data even if we were never fully initialized. */
337228753Smm	free(a->compressor.config);
338228753Smm	a->compressor.config = NULL;
339228753Smm	return (ret);
340228753Smm}
341228753Smm
342228753Smm/*
343228753Smm * Utility function to push input data through compressor, writing
344228753Smm * full output blocks as necessary.
345228753Smm *
346228753Smm * Note that this handles both the regular write case (finishing ==
347228753Smm * false) and the end-of-archive case (finishing == true).
348228753Smm */
349228753Smmstatic int
350228753Smmdrive_compressor(struct archive_write *a, struct private_data *state, int finishing)
351228753Smm{
352228753Smm	ssize_t	bytes_written;
353228753Smm	int ret;
354228753Smm
355228753Smm	for (;;) {
356228753Smm		if (state->stream.avail_out == 0) {
357228753Smm			bytes_written = (a->client_writer)(&a->archive,
358228753Smm			    a->client_data, state->compressed,
359228753Smm			    state->compressed_buffer_size);
360228753Smm			if (bytes_written <= 0) {
361228753Smm				/* TODO: Handle this write failure */
362228753Smm				return (ARCHIVE_FATAL);
363228753Smm			} else if ((size_t)bytes_written < state->compressed_buffer_size) {
364228753Smm				/* Short write: Move remainder to
365228753Smm				 * front and keep filling */
366228753Smm				memmove(state->compressed,
367228753Smm				    state->compressed + bytes_written,
368228753Smm				    state->compressed_buffer_size - bytes_written);
369228753Smm			}
370228753Smm
371228753Smm			a->archive.raw_position += bytes_written;
372228753Smm			state->stream.next_out = state->compressed +
373228753Smm			    state->compressed_buffer_size - bytes_written;
374228753Smm			state->stream.avail_out = bytes_written;
375228753Smm		}
376228753Smm
377228753Smm		/* If there's nothing to do, we're done. */
378228753Smm		if (!finishing && state->stream.avail_in == 0)
379228753Smm			return (ARCHIVE_OK);
380228753Smm
381228753Smm		ret = BZ2_bzCompress(&(state->stream),
382228753Smm		    finishing ? BZ_FINISH : BZ_RUN);
383228753Smm
384228753Smm		switch (ret) {
385228753Smm		case BZ_RUN_OK:
386228753Smm			/* In non-finishing case, did compressor
387228753Smm			 * consume everything? */
388228753Smm			if (!finishing && state->stream.avail_in == 0)
389228753Smm				return (ARCHIVE_OK);
390228753Smm			break;
391228753Smm		case BZ_FINISH_OK:  /* Finishing: There's more work to do */
392228753Smm			break;
393228753Smm		case BZ_STREAM_END: /* Finishing: all done */
394228753Smm			/* Only occurs in finishing case */
395228753Smm			return (ARCHIVE_OK);
396228753Smm		default:
397228753Smm			/* Any other return value indicates an error */
398228753Smm			archive_set_error(&a->archive,
399228753Smm			    ARCHIVE_ERRNO_PROGRAMMER,
400228753Smm			    "Bzip2 compression failed;"
401228753Smm			    " BZ2_bzCompress() returned %d",
402228753Smm			    ret);
403228753Smm			return (ARCHIVE_FATAL);
404228753Smm		}
405228753Smm	}
406228753Smm}
407228753Smm
408228753Smm#endif /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */
409