1 2181e0c8 2019-03-19 stsp /*
  2 2181e0c8 2019-03-19 stsp  * Copyright (c) 2019 Stefan Sperling <stsp@openbsd.org>
  3 2181e0c8 2019-03-19 stsp  *
  4 2181e0c8 2019-03-19 stsp  * Permission to use, copy, modify, and distribute this software for any
  5 2181e0c8 2019-03-19 stsp  * purpose with or without fee is hereby granted, provided that the above
  6 2181e0c8 2019-03-19 stsp  * copyright notice and this permission notice appear in all copies.
  7 2181e0c8 2019-03-19 stsp  *
  8 2181e0c8 2019-03-19 stsp  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  9 2181e0c8 2019-03-19 stsp  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 10 2181e0c8 2019-03-19 stsp  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 11 2181e0c8 2019-03-19 stsp  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 12 2181e0c8 2019-03-19 stsp  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 13 2181e0c8 2019-03-19 stsp  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 14 2181e0c8 2019-03-19 stsp  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 15 2181e0c8 2019-03-19 stsp  */
 16 2181e0c8 2019-03-19 stsp 
 17 2181e0c8 2019-03-19 stsp 
 18 2181e0c8 2019-03-19 stsp #include <errno.h>
 19 2181e0c8 2019-03-19 stsp #include <stdio.h>
 20 2181e0c8 2019-03-19 stsp #include <stdlib.h>
 21 2181e0c8 2019-03-19 stsp #include <string.h>
 22 2181e0c8 2019-03-19 stsp #include <zlib.h>
 23 2181e0c8 2019-03-19 stsp #include <time.h>
 24 2181e0c8 2019-03-19 stsp 
 25 2181e0c8 2019-03-19 stsp #include "got_error.h"
 26 2181e0c8 2019-03-19 stsp #include "got_object.h"
 27 324d37e7 2019-05-11 stsp #include "got_path.h"
 28 2181e0c8 2019-03-19 stsp 
 29 2181e0c8 2019-03-19 stsp #include "got_lib_deflate.h"
 30 2181e0c8 2019-03-19 stsp 
 31 2181e0c8 2019-03-19 stsp #ifndef MIN
 32 2181e0c8 2019-03-19 stsp #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
 33 2181e0c8 2019-03-19 stsp #endif
 34 2181e0c8 2019-03-19 stsp 
 35 2181e0c8 2019-03-19 stsp const struct got_error *
 36 3b9e6fcf 2021-06-05 stsp got_deflate_init(struct got_deflate_buf *zb, uint8_t *outbuf, size_t bufsize)
 37 2181e0c8 2019-03-19 stsp {
 38 2181e0c8 2019-03-19 stsp 	const struct got_error *err = NULL;
 39 2181e0c8 2019-03-19 stsp 	int zerr;
 40 2181e0c8 2019-03-19 stsp 
 41 2181e0c8 2019-03-19 stsp 	memset(&zb->z, 0, sizeof(zb->z));
 42 2181e0c8 2019-03-19 stsp 
 43 2181e0c8 2019-03-19 stsp 	zb->z.zalloc = Z_NULL;
 44 2181e0c8 2019-03-19 stsp 	zb->z.zfree = Z_NULL;
 45 2181e0c8 2019-03-19 stsp 	zerr = deflateInit(&zb->z, Z_DEFAULT_COMPRESSION);
 46 2181e0c8 2019-03-19 stsp 	if (zerr != Z_OK) {
 47 2181e0c8 2019-03-19 stsp 		if  (zerr == Z_ERRNO)
 48 638f9024 2019-05-13 stsp 			return got_error_from_errno("deflateInit");
 49 2181e0c8 2019-03-19 stsp 		if  (zerr == Z_MEM_ERROR) {
 50 2181e0c8 2019-03-19 stsp 			errno = ENOMEM;
 51 638f9024 2019-05-13 stsp 			return got_error_from_errno("deflateInit");
 52 2181e0c8 2019-03-19 stsp 		}
 53 2181e0c8 2019-03-19 stsp 		return got_error(GOT_ERR_COMPRESSION);
 54 2181e0c8 2019-03-19 stsp 	}
 55 2181e0c8 2019-03-19 stsp 
 56 2181e0c8 2019-03-19 stsp 	zb->inlen = zb->outlen = bufsize;
 57 2181e0c8 2019-03-19 stsp 
 58 2181e0c8 2019-03-19 stsp 	zb->inbuf = calloc(1, zb->inlen);
 59 2181e0c8 2019-03-19 stsp 	if (zb->inbuf == NULL) {
 60 638f9024 2019-05-13 stsp 		err = got_error_from_errno("calloc");
 61 2181e0c8 2019-03-19 stsp 		goto done;
 62 2181e0c8 2019-03-19 stsp 	}
 63 2181e0c8 2019-03-19 stsp 
 64 2181e0c8 2019-03-19 stsp 	zb->flags = 0;
 65 2181e0c8 2019-03-19 stsp 	if (outbuf == NULL) {
 66 2181e0c8 2019-03-19 stsp 		zb->outbuf = calloc(1, zb->outlen);
 67 2181e0c8 2019-03-19 stsp 		if (zb->outbuf == NULL) {
 68 638f9024 2019-05-13 stsp 			err = got_error_from_errno("calloc");
 69 2181e0c8 2019-03-19 stsp 			goto done;
 70 2181e0c8 2019-03-19 stsp 		}
 71 2181e0c8 2019-03-19 stsp 		zb->flags |= GOT_DEFLATE_F_OWN_OUTBUF;
 72 2181e0c8 2019-03-19 stsp 	} else
 73 2181e0c8 2019-03-19 stsp 		zb->outbuf = outbuf;
 74 2181e0c8 2019-03-19 stsp done:
 75 2181e0c8 2019-03-19 stsp 	if (err)
 76 2181e0c8 2019-03-19 stsp 		got_deflate_end(zb);
 77 2181e0c8 2019-03-19 stsp 	return err;
 78 2181e0c8 2019-03-19 stsp }
 79 2181e0c8 2019-03-19 stsp 
 80 91b40e30 2021-05-21 stsp static void
 81 dbaa2362 2021-09-28 thomas csum_output(struct got_deflate_checksum *csum, const uint8_t *buf, size_t len)
 82 91b40e30 2021-05-21 stsp {
 83 91b40e30 2021-05-21 stsp 	if (csum->output_crc)
 84 91b40e30 2021-05-21 stsp 		*csum->output_crc = crc32(*csum->output_crc, buf, len);
 85 91b40e30 2021-05-21 stsp 
 86 91b40e30 2021-05-21 stsp 	if (csum->output_sha1)
 87 91b40e30 2021-05-21 stsp 		SHA1Update(csum->output_sha1, buf, len);
 88 91b40e30 2021-05-21 stsp }
 89 91b40e30 2021-05-21 stsp 
 90 2181e0c8 2019-03-19 stsp const struct got_error *
 91 e8f02263 2022-01-23 thomas got_deflate_read(struct got_deflate_buf *zb, FILE *f, off_t len,
 92 e8f02263 2022-01-23 thomas     size_t *outlenp, off_t *consumed)
 93 2181e0c8 2019-03-19 stsp {
 94 2181e0c8 2019-03-19 stsp 	size_t last_total_out = zb->z.total_out;
 95 2181e0c8 2019-03-19 stsp 	z_stream *z = &zb->z;
 96 2181e0c8 2019-03-19 stsp 	int ret = Z_ERRNO;
 97 2181e0c8 2019-03-19 stsp 
 98 2181e0c8 2019-03-19 stsp 	z->next_out = zb->outbuf;
 99 2181e0c8 2019-03-19 stsp 	z->avail_out = zb->outlen;
100 2181e0c8 2019-03-19 stsp 
101 2181e0c8 2019-03-19 stsp 	*outlenp = 0;
102 e8f02263 2022-01-23 thomas 	*consumed = 0;
103 2181e0c8 2019-03-19 stsp 	do {
104 e8f02263 2022-01-23 thomas 		size_t last_total_in = z->total_in;
105 2181e0c8 2019-03-19 stsp 		if (z->avail_in == 0) {
106 e8f02263 2022-01-23 thomas 			size_t n = 0;
107 e8f02263 2022-01-23 thomas 			if (*consumed < len) {
108 e8f02263 2022-01-23 thomas 				n = fread(zb->inbuf, 1,
109 e8f02263 2022-01-23 thomas 				    MIN(zb->inlen, len - *consumed), f);
110 e8f02263 2022-01-23 thomas 			}
111 2181e0c8 2019-03-19 stsp 			if (n == 0) {
112 2181e0c8 2019-03-19 stsp 				if (ferror(f))
113 2181e0c8 2019-03-19 stsp 					return got_ferror(f, GOT_ERR_IO);
114 2181e0c8 2019-03-19 stsp 				/* EOF */
115 4e4a7005 2019-04-13 stsp 				ret = deflate(z, Z_FINISH);
116 2181e0c8 2019-03-19 stsp 				break;
117 2181e0c8 2019-03-19 stsp 			}
118 2181e0c8 2019-03-19 stsp 			z->next_in = zb->inbuf;
119 2181e0c8 2019-03-19 stsp 			z->avail_in = n;
120 2181e0c8 2019-03-19 stsp 		}
121 2181e0c8 2019-03-19 stsp 		ret = deflate(z, Z_NO_FLUSH);
122 e8f02263 2022-01-23 thomas 		*consumed += z->total_in - last_total_in;
123 2181e0c8 2019-03-19 stsp 	} while (ret == Z_OK && z->avail_out > 0);
124 2181e0c8 2019-03-19 stsp 
125 2181e0c8 2019-03-19 stsp 	if (ret == Z_OK) {
126 2181e0c8 2019-03-19 stsp 		zb->flags |= GOT_DEFLATE_F_HAVE_MORE;
127 2181e0c8 2019-03-19 stsp 	} else {
128 2181e0c8 2019-03-19 stsp 		if (ret != Z_STREAM_END)
129 2181e0c8 2019-03-19 stsp 			return got_error(GOT_ERR_COMPRESSION);
130 2181e0c8 2019-03-19 stsp 		zb->flags &= ~GOT_DEFLATE_F_HAVE_MORE;
131 2181e0c8 2019-03-19 stsp 	}
132 2181e0c8 2019-03-19 stsp 
133 2181e0c8 2019-03-19 stsp 	*outlenp = z->total_out - last_total_out;
134 2181e0c8 2019-03-19 stsp 	return NULL;
135 2181e0c8 2019-03-19 stsp }
136 2181e0c8 2019-03-19 stsp 
137 2b0ae357 2022-01-10 thomas const struct got_error *
138 2b0ae357 2022-01-10 thomas got_deflate_read_mmap(struct got_deflate_buf *zb, uint8_t *map, size_t offset,
139 2b0ae357 2022-01-10 thomas     size_t len, size_t *outlenp, size_t *consumed)
140 2b0ae357 2022-01-10 thomas {
141 2b0ae357 2022-01-10 thomas 	z_stream *z = &zb->z;
142 2b0ae357 2022-01-10 thomas 	size_t last_total_out = z->total_out;
143 2b0ae357 2022-01-10 thomas 	int ret = Z_ERRNO;
144 2b0ae357 2022-01-10 thomas 
145 2b0ae357 2022-01-10 thomas 	z->next_out = zb->outbuf;
146 2b0ae357 2022-01-10 thomas 	z->avail_out = zb->outlen;
147 2b0ae357 2022-01-10 thomas 
148 2b0ae357 2022-01-10 thomas 	*outlenp = 0;
149 2b0ae357 2022-01-10 thomas 	*consumed = 0;
150 2b0ae357 2022-01-10 thomas 	do {
151 2b0ae357 2022-01-10 thomas 		size_t last_total_in = z->total_in;
152 2b0ae357 2022-01-10 thomas 		if (z->avail_in == 0) {
153 2b0ae357 2022-01-10 thomas 			z->next_in = map + offset + *consumed;
154 2b0ae357 2022-01-10 thomas 			z->avail_in = len - *consumed;
155 2b0ae357 2022-01-10 thomas 			if (z->avail_in == 0) {
156 2b0ae357 2022-01-10 thomas 				/* EOF */
157 2b0ae357 2022-01-10 thomas 				ret = deflate(z, Z_FINISH);
158 2b0ae357 2022-01-10 thomas 				break;
159 2b0ae357 2022-01-10 thomas 			}
160 2b0ae357 2022-01-10 thomas 		}
161 2b0ae357 2022-01-10 thomas 		ret = deflate(z, Z_NO_FLUSH);
162 2b0ae357 2022-01-10 thomas 		*consumed += z->total_in - last_total_in;
163 2b0ae357 2022-01-10 thomas 	} while (ret == Z_OK && z->avail_out > 0);
164 2b0ae357 2022-01-10 thomas 
165 2b0ae357 2022-01-10 thomas 	if (ret == Z_OK) {
166 2b0ae357 2022-01-10 thomas 		zb->flags |= GOT_DEFLATE_F_HAVE_MORE;
167 2b0ae357 2022-01-10 thomas 	} else {
168 2b0ae357 2022-01-10 thomas 		if (ret != Z_STREAM_END)
169 2b0ae357 2022-01-10 thomas 			return got_error(GOT_ERR_COMPRESSION);
170 2b0ae357 2022-01-10 thomas 		zb->flags &= ~GOT_DEFLATE_F_HAVE_MORE;
171 2b0ae357 2022-01-10 thomas 	}
172 2b0ae357 2022-01-10 thomas 
173 2b0ae357 2022-01-10 thomas 	*outlenp = z->total_out - last_total_out;
174 2b0ae357 2022-01-10 thomas 	return NULL;
175 2b0ae357 2022-01-10 thomas }
176 2b0ae357 2022-01-10 thomas 
177 2181e0c8 2019-03-19 stsp void
178 2181e0c8 2019-03-19 stsp got_deflate_end(struct got_deflate_buf *zb)
179 2181e0c8 2019-03-19 stsp {
180 2181e0c8 2019-03-19 stsp 	free(zb->inbuf);
181 2181e0c8 2019-03-19 stsp 	if (zb->flags & GOT_DEFLATE_F_OWN_OUTBUF)
182 2181e0c8 2019-03-19 stsp 		free(zb->outbuf);
183 2181e0c8 2019-03-19 stsp 	deflateEnd(&zb->z);
184 2181e0c8 2019-03-19 stsp }
185 2181e0c8 2019-03-19 stsp 
186 2181e0c8 2019-03-19 stsp const struct got_error *
187 e8f02263 2022-01-23 thomas got_deflate_to_file(off_t *outlen, FILE *infile, off_t len,
188 e8f02263 2022-01-23 thomas     FILE *outfile, struct got_deflate_checksum *csum)
189 2181e0c8 2019-03-19 stsp {
190 2181e0c8 2019-03-19 stsp 	const struct got_error *err;
191 2181e0c8 2019-03-19 stsp 	size_t avail;
192 e8f02263 2022-01-23 thomas 	off_t consumed;
193 2181e0c8 2019-03-19 stsp 	struct got_deflate_buf zb;
194 2181e0c8 2019-03-19 stsp 
195 3b9e6fcf 2021-06-05 stsp 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
196 2181e0c8 2019-03-19 stsp 	if (err)
197 2181e0c8 2019-03-19 stsp 		goto done;
198 2181e0c8 2019-03-19 stsp 
199 2181e0c8 2019-03-19 stsp 	*outlen = 0;
200 2181e0c8 2019-03-19 stsp 
201 2181e0c8 2019-03-19 stsp 	do {
202 e8f02263 2022-01-23 thomas 		err = got_deflate_read(&zb, infile, len, &avail, &consumed);
203 2181e0c8 2019-03-19 stsp 		if (err)
204 2181e0c8 2019-03-19 stsp 			goto done;
205 e8f02263 2022-01-23 thomas 		len -= consumed;
206 2181e0c8 2019-03-19 stsp 		if (avail > 0) {
207 2181e0c8 2019-03-19 stsp 			size_t n;
208 2181e0c8 2019-03-19 stsp 			n = fwrite(zb.outbuf, avail, 1, outfile);
209 2181e0c8 2019-03-19 stsp 			if (n != 1) {
210 2181e0c8 2019-03-19 stsp 				err = got_ferror(outfile, GOT_ERR_IO);
211 2181e0c8 2019-03-19 stsp 				goto done;
212 2181e0c8 2019-03-19 stsp 			}
213 3b9e6fcf 2021-06-05 stsp 			if (csum)
214 3b9e6fcf 2021-06-05 stsp 				csum_output(csum, zb.outbuf, avail);
215 2181e0c8 2019-03-19 stsp 			*outlen += avail;
216 2181e0c8 2019-03-19 stsp 		}
217 2181e0c8 2019-03-19 stsp 	} while (zb.flags & GOT_DEFLATE_F_HAVE_MORE);
218 2181e0c8 2019-03-19 stsp 
219 2181e0c8 2019-03-19 stsp done:
220 2181e0c8 2019-03-19 stsp 	got_deflate_end(&zb);
221 2181e0c8 2019-03-19 stsp 	return err;
222 2181e0c8 2019-03-19 stsp }
223 2b0ae357 2022-01-10 thomas 
224 2b0ae357 2022-01-10 thomas const struct got_error *
225 e8f02263 2022-01-23 thomas got_deflate_to_file_mmap(off_t *outlen, uint8_t *map, size_t offset,
226 2b0ae357 2022-01-10 thomas     size_t len, FILE *outfile, struct got_deflate_checksum *csum)
227 2b0ae357 2022-01-10 thomas {
228 2b0ae357 2022-01-10 thomas 	const struct got_error *err;
229 2b0ae357 2022-01-10 thomas 	size_t avail, consumed;
230 2b0ae357 2022-01-10 thomas 	struct got_deflate_buf zb;
231 2b0ae357 2022-01-10 thomas 
232 2b0ae357 2022-01-10 thomas 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
233 2b0ae357 2022-01-10 thomas 	if (err)
234 2b0ae357 2022-01-10 thomas 		goto done;
235 2b0ae357 2022-01-10 thomas 
236 2b0ae357 2022-01-10 thomas 	*outlen = 0;
237 2b0ae357 2022-01-10 thomas 	do {
238 2b0ae357 2022-01-10 thomas 		err = got_deflate_read_mmap(&zb, map, offset, len, &avail,
239 2b0ae357 2022-01-10 thomas 		    &consumed);
240 2b0ae357 2022-01-10 thomas 		if (err)
241 2b0ae357 2022-01-10 thomas 			goto done;
242 2b0ae357 2022-01-10 thomas 		offset += consumed;
243 2b0ae357 2022-01-10 thomas 		len -= consumed;
244 2b0ae357 2022-01-10 thomas 		if (avail > 0) {
245 2b0ae357 2022-01-10 thomas 			size_t n;
246 2b0ae357 2022-01-10 thomas 			n = fwrite(zb.outbuf, avail, 1, outfile);
247 2b0ae357 2022-01-10 thomas 			if (n != 1) {
248 2b0ae357 2022-01-10 thomas 				err = got_ferror(outfile, GOT_ERR_IO);
249 2b0ae357 2022-01-10 thomas 				goto done;
250 2b0ae357 2022-01-10 thomas 			}
251 2b0ae357 2022-01-10 thomas 			if (csum)
252 2b0ae357 2022-01-10 thomas 				csum_output(csum, zb.outbuf, avail);
253 2b0ae357 2022-01-10 thomas 			*outlen += avail;
254 2b0ae357 2022-01-10 thomas 		}
255 2b0ae357 2022-01-10 thomas 	} while (zb.flags & GOT_DEFLATE_F_HAVE_MORE);
256 2b0ae357 2022-01-10 thomas 
257 2b0ae357 2022-01-10 thomas done:
258 2b0ae357 2022-01-10 thomas 	got_deflate_end(&zb);
259 2b0ae357 2022-01-10 thomas 	return err;
260 2b0ae357 2022-01-10 thomas }