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
28228753Smm__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_gzip.c 201081 2009-12-28 02:04:42Z kientzle $");
29228753Smm
30228753Smm#ifdef HAVE_ERRNO_H
31228753Smm#include <errno.h>
32228753Smm#endif
33228753Smm#ifdef HAVE_STDLIB_H
34228753Smm#include <stdlib.h>
35228753Smm#endif
36228753Smm#ifdef HAVE_STRING_H
37228753Smm#include <string.h>
38228753Smm#endif
39228753Smm#include <time.h>
40228753Smm#ifdef HAVE_ZLIB_H
41228753Smm#include <zlib.h>
42228753Smm#endif
43228753Smm
44228753Smm#include "archive.h"
45228753Smm#include "archive_private.h"
46248616Smm#include "archive_string.h"
47228753Smm#include "archive_write_private.h"
48228753Smm
49231200Smm#if ARCHIVE_VERSION_NUMBER < 4000000
50228753Smmint
51228753Smmarchive_write_set_compression_gzip(struct archive *a)
52228753Smm{
53231200Smm	__archive_write_filters_free(a);
54231200Smm	return (archive_write_add_filter_gzip(a));
55231200Smm}
56231200Smm#endif
57231200Smm
58228753Smm/* Don't compile this if we don't have zlib. */
59228753Smm
60228753Smmstruct private_data {
61231200Smm	int		 compression_level;
62248616Smm	int		 timestamp;
63248616Smm#ifdef HAVE_ZLIB_H
64228753Smm	z_stream	 stream;
65228753Smm	int64_t		 total_in;
66228753Smm	unsigned char	*compressed;
67228753Smm	size_t		 compressed_buffer_size;
68228753Smm	unsigned long	 crc;
69248616Smm#else
70248616Smm	struct archive_write_program_data *pdata;
71248616Smm#endif
72228753Smm};
73228753Smm
74228753Smm/*
75228753Smm * Yuck.  zlib.h is not const-correct, so I need this one bit
76228753Smm * of ugly hackery to convert a const * pointer to a non-const pointer.
77228753Smm */
78228753Smm#define	SET_NEXT_IN(st,src)					\
79228753Smm	(st)->stream.next_in = (Bytef *)(uintptr_t)(const void *)(src)
80228753Smm
81231200Smmstatic int archive_compressor_gzip_options(struct archive_write_filter *,
82228753Smm		    const char *, const char *);
83231200Smmstatic int archive_compressor_gzip_open(struct archive_write_filter *);
84231200Smmstatic int archive_compressor_gzip_write(struct archive_write_filter *,
85228753Smm		    const void *, size_t);
86231200Smmstatic int archive_compressor_gzip_close(struct archive_write_filter *);
87231200Smmstatic int archive_compressor_gzip_free(struct archive_write_filter *);
88248616Smm#ifdef HAVE_ZLIB_H
89231200Smmstatic int drive_compressor(struct archive_write_filter *,
90231200Smm		    struct private_data *, int finishing);
91248616Smm#endif
92228753Smm
93228753Smm
94228753Smm/*
95231200Smm * Add a gzip compression filter to this write handle.
96228753Smm */
97228753Smmint
98231200Smmarchive_write_add_filter_gzip(struct archive *_a)
99228753Smm{
100228753Smm	struct archive_write *a = (struct archive_write *)_a;
101231200Smm	struct archive_write_filter *f = __archive_write_allocate_filter(_a);
102231200Smm	struct private_data *data;
103231200Smm	archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC,
104231200Smm	    ARCHIVE_STATE_NEW, "archive_write_add_filter_gzip");
105231200Smm
106231200Smm	data = calloc(1, sizeof(*data));
107231200Smm	if (data == NULL) {
108228753Smm		archive_set_error(&a->archive, ENOMEM, "Out of memory");
109228753Smm		return (ARCHIVE_FATAL);
110228753Smm	}
111231200Smm	f->data = data;
112231200Smm	f->open = &archive_compressor_gzip_open;
113231200Smm	f->options = &archive_compressor_gzip_options;
114231200Smm	f->close = &archive_compressor_gzip_close;
115231200Smm	f->free = &archive_compressor_gzip_free;
116248616Smm	f->code = ARCHIVE_FILTER_GZIP;
117231200Smm	f->name = "gzip";
118248616Smm#ifdef HAVE_ZLIB_H
119248616Smm	data->compression_level = Z_DEFAULT_COMPRESSION;
120228753Smm	return (ARCHIVE_OK);
121248616Smm#else
122248616Smm	data->pdata = __archive_write_program_allocate();
123248616Smm	if (data->pdata == NULL) {
124248616Smm		free(data);
125248616Smm		archive_set_error(&a->archive, ENOMEM, "Out of memory");
126248616Smm		return (ARCHIVE_FATAL);
127248616Smm	}
128248616Smm	data->compression_level = 0;
129248616Smm	archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
130248616Smm	    "Using external gzip program");
131248616Smm	return (ARCHIVE_WARN);
132248616Smm#endif
133228753Smm}
134228753Smm
135248616Smmstatic int
136248616Smmarchive_compressor_gzip_free(struct archive_write_filter *f)
137248616Smm{
138248616Smm	struct private_data *data = (struct private_data *)f->data;
139248616Smm
140248616Smm#ifdef HAVE_ZLIB_H
141248616Smm	free(data->compressed);
142248616Smm#else
143248616Smm	__archive_write_program_free(data->pdata);
144248616Smm#endif
145248616Smm	free(data);
146248616Smm	f->data = NULL;
147248616Smm	return (ARCHIVE_OK);
148248616Smm}
149248616Smm
150228753Smm/*
151248616Smm * Set write options.
152248616Smm */
153248616Smmstatic int
154248616Smmarchive_compressor_gzip_options(struct archive_write_filter *f, const char *key,
155248616Smm    const char *value)
156248616Smm{
157248616Smm	struct private_data *data = (struct private_data *)f->data;
158248616Smm
159248616Smm	if (strcmp(key, "compression-level") == 0) {
160248616Smm		if (value == NULL || !(value[0] >= '0' && value[0] <= '9') ||
161248616Smm		    value[1] != '\0')
162248616Smm			return (ARCHIVE_WARN);
163248616Smm		data->compression_level = value[0] - '0';
164248616Smm		return (ARCHIVE_OK);
165248616Smm	}
166248616Smm	if (strcmp(key, "timestamp") == 0) {
167248616Smm		data->timestamp = (value == NULL)?-1:1;
168248616Smm		return (ARCHIVE_OK);
169248616Smm	}
170248616Smm
171248616Smm	/* Note: The "warn" return is just to inform the options
172248616Smm	 * supervisor that we didn't handle it.  It will generate
173248616Smm	 * a suitable error if no one used this option. */
174248616Smm	return (ARCHIVE_WARN);
175248616Smm}
176248616Smm
177248616Smm#ifdef HAVE_ZLIB_H
178248616Smm/*
179228753Smm * Setup callback.
180228753Smm */
181228753Smmstatic int
182231200Smmarchive_compressor_gzip_open(struct archive_write_filter *f)
183228753Smm{
184231200Smm	struct private_data *data = (struct private_data *)f->data;
185228753Smm	int ret;
186228753Smm
187231200Smm	ret = __archive_write_open_filter(f->next_filter);
188231200Smm	if (ret != ARCHIVE_OK)
189231200Smm		return (ret);
190228753Smm
191231200Smm	if (data->compressed == NULL) {
192238856Smm		size_t bs = 65536, bpb;
193238856Smm		if (f->archive->magic == ARCHIVE_WRITE_MAGIC) {
194248616Smm			/* Buffer size should be a multiple number of
195248616Smm			 * the of bytes per block for performance. */
196238856Smm			bpb = archive_write_get_bytes_per_block(f->archive);
197238856Smm			if (bpb > bs)
198238856Smm				bs = bpb;
199238856Smm			else if (bpb != 0)
200238856Smm				bs -= bs % bpb;
201238856Smm		}
202238856Smm		data->compressed_buffer_size = bs;
203231200Smm		data->compressed
204231200Smm		    = (unsigned char *)malloc(data->compressed_buffer_size);
205231200Smm		if (data->compressed == NULL) {
206231200Smm			archive_set_error(f->archive, ENOMEM,
207231200Smm			    "Can't allocate data for compression buffer");
208231200Smm			return (ARCHIVE_FATAL);
209231200Smm		}
210228753Smm	}
211228753Smm
212231200Smm	data->crc = crc32(0L, NULL, 0);
213231200Smm	data->stream.next_out = data->compressed;
214248616Smm	data->stream.avail_out = (uInt)data->compressed_buffer_size;
215228753Smm
216228753Smm	/* Prime output buffer with a gzip header. */
217231200Smm	data->compressed[0] = 0x1f; /* GZip signature bytes */
218231200Smm	data->compressed[1] = 0x8b;
219231200Smm	data->compressed[2] = 0x08; /* "Deflate" compression */
220231200Smm	data->compressed[3] = 0; /* No options */
221248616Smm	if (data->timestamp >= 0) {
222248616Smm		time_t t = time(NULL);
223248616Smm		data->compressed[4] = (uint8_t)(t)&0xff;  /* Timestamp */
224248616Smm		data->compressed[5] = (uint8_t)(t>>8)&0xff;
225248616Smm		data->compressed[6] = (uint8_t)(t>>16)&0xff;
226248616Smm		data->compressed[7] = (uint8_t)(t>>24)&0xff;
227248616Smm	} else
228248616Smm		memset(&data->compressed[4], 0, 4);
229231200Smm	data->compressed[8] = 0; /* No deflate options */
230231200Smm	data->compressed[9] = 3; /* OS=Unix */
231231200Smm	data->stream.next_out += 10;
232231200Smm	data->stream.avail_out -= 10;
233228753Smm
234231200Smm	f->write = archive_compressor_gzip_write;
235228753Smm
236228753Smm	/* Initialize compression library. */
237231200Smm	ret = deflateInit2(&(data->stream),
238231200Smm	    data->compression_level,
239228753Smm	    Z_DEFLATED,
240228753Smm	    -15 /* < 0 to suppress zlib header */,
241228753Smm	    8,
242228753Smm	    Z_DEFAULT_STRATEGY);
243228753Smm
244228753Smm	if (ret == Z_OK) {
245231200Smm		f->data = data;
246231200Smm		return (ARCHIVE_OK);
247228753Smm	}
248228753Smm
249228753Smm	/* Library setup failed: clean up. */
250231200Smm	archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error "
251228753Smm	    "initializing compression library");
252228753Smm
253228753Smm	/* Override the error message if we know what really went wrong. */
254228753Smm	switch (ret) {
255228753Smm	case Z_STREAM_ERROR:
256231200Smm		archive_set_error(f->archive, ARCHIVE_ERRNO_MISC,
257228753Smm		    "Internal error initializing "
258228753Smm		    "compression library: invalid setup parameter");
259228753Smm		break;
260228753Smm	case Z_MEM_ERROR:
261248616Smm		archive_set_error(f->archive, ENOMEM,
262248616Smm		    "Internal error initializing compression library");
263228753Smm		break;
264228753Smm	case Z_VERSION_ERROR:
265231200Smm		archive_set_error(f->archive, ARCHIVE_ERRNO_MISC,
266228753Smm		    "Internal error initializing "
267228753Smm		    "compression library: invalid library version");
268228753Smm		break;
269228753Smm	}
270228753Smm
271228753Smm	return (ARCHIVE_FATAL);
272228753Smm}
273228753Smm
274228753Smm/*
275228753Smm * Write data to the compressed stream.
276228753Smm */
277228753Smmstatic int
278231200Smmarchive_compressor_gzip_write(struct archive_write_filter *f, const void *buff,
279228753Smm    size_t length)
280228753Smm{
281231200Smm	struct private_data *data = (struct private_data *)f->data;
282228753Smm	int ret;
283228753Smm
284228753Smm	/* Update statistics */
285248616Smm	data->crc = crc32(data->crc, (const Bytef *)buff, (uInt)length);
286231200Smm	data->total_in += length;
287228753Smm
288228753Smm	/* Compress input data to output buffer */
289231200Smm	SET_NEXT_IN(data, buff);
290248616Smm	data->stream.avail_in = (uInt)length;
291231200Smm	if ((ret = drive_compressor(f, data, 0)) != ARCHIVE_OK)
292228753Smm		return (ret);
293228753Smm
294228753Smm	return (ARCHIVE_OK);
295228753Smm}
296228753Smm
297228753Smm/*
298228753Smm * Finish the compression...
299228753Smm */
300228753Smmstatic int
301231200Smmarchive_compressor_gzip_close(struct archive_write_filter *f)
302228753Smm{
303228753Smm	unsigned char trailer[8];
304231200Smm	struct private_data *data = (struct private_data *)f->data;
305231200Smm	int ret, r1;
306228753Smm
307231200Smm	/* Finish compression cycle */
308231200Smm	ret = drive_compressor(f, data, 1);
309231200Smm	if (ret == ARCHIVE_OK) {
310231200Smm		/* Write the last compressed data. */
311231200Smm		ret = __archive_write_filter(f->next_filter,
312231200Smm		    data->compressed,
313231200Smm		    data->compressed_buffer_size - data->stream.avail_out);
314231200Smm	}
315231200Smm	if (ret == ARCHIVE_OK) {
316231200Smm		/* Build and write out 8-byte trailer. */
317238856Smm		trailer[0] = (uint8_t)(data->crc)&0xff;
318238856Smm		trailer[1] = (uint8_t)(data->crc >> 8)&0xff;
319238856Smm		trailer[2] = (uint8_t)(data->crc >> 16)&0xff;
320238856Smm		trailer[3] = (uint8_t)(data->crc >> 24)&0xff;
321238856Smm		trailer[4] = (uint8_t)(data->total_in)&0xff;
322238856Smm		trailer[5] = (uint8_t)(data->total_in >> 8)&0xff;
323238856Smm		trailer[6] = (uint8_t)(data->total_in >> 16)&0xff;
324238856Smm		trailer[7] = (uint8_t)(data->total_in >> 24)&0xff;
325231200Smm		ret = __archive_write_filter(f->next_filter, trailer, 8);
326231200Smm	}
327228753Smm
328231200Smm	switch (deflateEnd(&(data->stream))) {
329231200Smm	case Z_OK:
330231200Smm		break;
331231200Smm	default:
332231200Smm		archive_set_error(f->archive, ARCHIVE_ERRNO_MISC,
333231200Smm		    "Failed to clean up compressor");
334231200Smm		ret = ARCHIVE_FATAL;
335228753Smm	}
336231200Smm	r1 = __archive_write_close_filter(f->next_filter);
337231200Smm	return (r1 < ret ? r1 : ret);
338228753Smm}
339228753Smm
340228753Smm/*
341228753Smm * Utility function to push input data through compressor,
342228753Smm * writing full output blocks as necessary.
343228753Smm *
344228753Smm * Note that this handles both the regular write case (finishing ==
345228753Smm * false) and the end-of-archive case (finishing == true).
346228753Smm */
347228753Smmstatic int
348231200Smmdrive_compressor(struct archive_write_filter *f,
349231200Smm    struct private_data *data, int finishing)
350228753Smm{
351228753Smm	int ret;
352228753Smm
353228753Smm	for (;;) {
354231200Smm		if (data->stream.avail_out == 0) {
355231200Smm			ret = __archive_write_filter(f->next_filter,
356231200Smm			    data->compressed,
357231200Smm			    data->compressed_buffer_size);
358231200Smm			if (ret != ARCHIVE_OK)
359228753Smm				return (ARCHIVE_FATAL);
360231200Smm			data->stream.next_out = data->compressed;
361248616Smm			data->stream.avail_out =
362248616Smm			    (uInt)data->compressed_buffer_size;
363228753Smm		}
364228753Smm
365228753Smm		/* If there's nothing to do, we're done. */
366231200Smm		if (!finishing && data->stream.avail_in == 0)
367228753Smm			return (ARCHIVE_OK);
368228753Smm
369231200Smm		ret = deflate(&(data->stream),
370228753Smm		    finishing ? Z_FINISH : Z_NO_FLUSH );
371228753Smm
372228753Smm		switch (ret) {
373228753Smm		case Z_OK:
374228753Smm			/* In non-finishing case, check if compressor
375228753Smm			 * consumed everything */
376231200Smm			if (!finishing && data->stream.avail_in == 0)
377228753Smm				return (ARCHIVE_OK);
378228753Smm			/* In finishing case, this return always means
379228753Smm			 * there's more work */
380228753Smm			break;
381228753Smm		case Z_STREAM_END:
382228753Smm			/* This return can only occur in finishing case. */
383228753Smm			return (ARCHIVE_OK);
384228753Smm		default:
385228753Smm			/* Any other return value indicates an error. */
386231200Smm			archive_set_error(f->archive, ARCHIVE_ERRNO_MISC,
387228753Smm			    "GZip compression failed:"
388228753Smm			    " deflate() call returned status %d",
389228753Smm			    ret);
390228753Smm			return (ARCHIVE_FATAL);
391228753Smm		}
392228753Smm	}
393228753Smm}
394228753Smm
395248616Smm#else /* HAVE_ZLIB_H */
396248616Smm
397248616Smmstatic int
398248616Smmarchive_compressor_gzip_open(struct archive_write_filter *f)
399248616Smm{
400248616Smm	struct private_data *data = (struct private_data *)f->data;
401248616Smm	struct archive_string as;
402248616Smm	int r;
403248616Smm
404248616Smm	archive_string_init(&as);
405248616Smm	archive_strcpy(&as, "gzip");
406248616Smm
407248616Smm	/* Specify compression level. */
408248616Smm	if (data->compression_level > 0) {
409248616Smm		archive_strcat(&as, " -");
410248616Smm		archive_strappend_char(&as, '0' + data->compression_level);
411248616Smm	}
412248616Smm	if (data->timestamp < 0)
413248616Smm		/* Do not save timestamp. */
414248616Smm		archive_strcat(&as, " -n");
415248616Smm	else if (data->timestamp > 0)
416248616Smm		/* Save timestamp. */
417248616Smm		archive_strcat(&as, " -N");
418248616Smm
419248616Smm	f->write = archive_compressor_gzip_write;
420248616Smm	r = __archive_write_program_open(f, data->pdata, as.s);
421248616Smm	archive_string_free(&as);
422248616Smm	return (r);
423248616Smm}
424248616Smm
425248616Smmstatic int
426248616Smmarchive_compressor_gzip_write(struct archive_write_filter *f, const void *buff,
427248616Smm    size_t length)
428248616Smm{
429248616Smm	struct private_data *data = (struct private_data *)f->data;
430248616Smm
431248616Smm	return __archive_write_program_write(f, data->pdata, buff, length);
432248616Smm}
433248616Smm
434248616Smmstatic int
435248616Smmarchive_compressor_gzip_close(struct archive_write_filter *f)
436248616Smm{
437248616Smm	struct private_data *data = (struct private_data *)f->data;
438248616Smm
439248616Smm	return __archive_write_program_close(f, data->pdata);
440248616Smm}
441248616Smm
442228753Smm#endif /* HAVE_ZLIB_H */
443