1 /*
2 * Copyright (C) 2017 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17 #include <errno.h>
18 #include <fcntl.h>
19 #include <fnmatch.h>
20 #include <getopt.h>
21 #include <inttypes.h>
22 #include <libgen.h>
23 #include <stdarg.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <sys/stat.h>
27 #include <sys/types.h>
28 #include <time.h>
29 #include <unistd.h>
30
31 #include <set>
32 #include <string>
33
34 #include <android-base/file.h>
35 #include <android-base/strings.h>
36 #include <ziparchive/zip_archive.h>
37
38 using android::base::EndsWith;
39 using android::base::StartsWith;
40
41 enum OverwriteMode {
42 kAlways,
43 kNever,
44 kPrompt,
45 };
46
47 enum Role {
48 kUnzip,
49 kZipinfo,
50 };
51
52 static Role role;
53 static OverwriteMode overwrite_mode = kPrompt;
54 static bool flag_1 = false;
55 static std::string flag_d;
56 static bool flag_l = false;
57 static bool flag_p = false;
58 static bool flag_q = false;
59 static bool flag_v = false;
60 static bool flag_x = false;
61 static const char* archive_name = nullptr;
62 static std::set<std::string> includes;
63 static std::set<std::string> excludes;
64 static uint64_t total_uncompressed_length = 0;
65 static uint64_t total_compressed_length = 0;
66 static size_t file_count = 0;
67
68 static const char* g_progname;
69
die(int error,const char * fmt,...)70 static void die(int error, const char* fmt, ...) {
71 va_list ap;
72
73 va_start(ap, fmt);
74 fprintf(stderr, "%s: ", g_progname);
75 vfprintf(stderr, fmt, ap);
76 if (error != 0) fprintf(stderr, ": %s", strerror(error));
77 fprintf(stderr, "\n");
78 va_end(ap);
79 exit(1);
80 }
81
ShouldInclude(const std::string & name)82 static bool ShouldInclude(const std::string& name) {
83 // Explicitly excluded?
84 if (!excludes.empty()) {
85 for (const auto& exclude : excludes) {
86 if (!fnmatch(exclude.c_str(), name.c_str(), 0)) return false;
87 }
88 }
89
90 // Implicitly included?
91 if (includes.empty()) return true;
92
93 // Explicitly included?
94 for (const auto& include : includes) {
95 if (!fnmatch(include.c_str(), name.c_str(), 0)) return true;
96 }
97 return false;
98 }
99
MakeDirectoryHierarchy(const std::string & path)100 static bool MakeDirectoryHierarchy(const std::string& path) {
101 // stat rather than lstat because a symbolic link to a directory is fine too.
102 struct stat sb;
103 if (stat(path.c_str(), &sb) != -1 && S_ISDIR(sb.st_mode)) return true;
104
105 // Ensure the parent directories exist first.
106 if (!MakeDirectoryHierarchy(android::base::Dirname(path))) return false;
107
108 // Then try to create this directory.
109 return (mkdir(path.c_str(), 0777) != -1);
110 }
111
CompressionRatio(int64_t uncompressed,int64_t compressed)112 static float CompressionRatio(int64_t uncompressed, int64_t compressed) {
113 if (uncompressed == 0) return 0;
114 return static_cast<float>(100LL * (uncompressed - compressed)) /
115 static_cast<float>(uncompressed);
116 }
117
MaybeShowHeader(ZipArchiveHandle zah)118 static void MaybeShowHeader(ZipArchiveHandle zah) {
119 if (role == kUnzip) {
120 // unzip has three formats.
121 if (!flag_q) printf("Archive: %s\n", archive_name);
122 if (flag_v) {
123 printf(
124 " Length Method Size Cmpr Date Time CRC-32 Name\n"
125 "-------- ------ ------- ---- ---------- ----- -------- ----\n");
126 } else if (flag_l) {
127 printf(
128 " Length Date Time Name\n"
129 "--------- ---------- ----- ----\n");
130 }
131 } else {
132 // zipinfo.
133 if (!flag_1 && includes.empty() && excludes.empty()) {
134 ZipArchiveInfo info{GetArchiveInfo(zah)};
135 printf("Archive: %s\n", archive_name);
136 printf("Zip file size: %" PRId64 " bytes, number of entries: %zu\n", info.archive_size,
137 info.entry_count);
138 }
139 }
140 }
141
MaybeShowFooter()142 static void MaybeShowFooter() {
143 if (role == kUnzip) {
144 if (flag_v) {
145 printf(
146 "-------- ------- --- -------\n"
147 "%8" PRId64 " %8" PRId64 " %3.0f%% %zu file%s\n",
148 total_uncompressed_length, total_compressed_length,
149 CompressionRatio(total_uncompressed_length, total_compressed_length), file_count,
150 (file_count == 1) ? "" : "s");
151 } else if (flag_l) {
152 printf(
153 "--------- -------\n"
154 "%9" PRId64 " %zu file%s\n",
155 total_uncompressed_length, file_count, (file_count == 1) ? "" : "s");
156 }
157 } else {
158 if (!flag_1 && includes.empty() && excludes.empty()) {
159 printf("%zu files, %" PRId64 " bytes uncompressed, %" PRId64 " bytes compressed: %.1f%%\n",
160 file_count, total_uncompressed_length, total_compressed_length,
161 CompressionRatio(total_uncompressed_length, total_compressed_length));
162 }
163 }
164 }
165
PromptOverwrite(const std::string & dst)166 static bool PromptOverwrite(const std::string& dst) {
167 // TODO: [r]ename not implemented because it doesn't seem useful.
168 printf("replace %s? [y]es, [n]o, [A]ll, [N]one: ", dst.c_str());
169 fflush(stdout);
170 while (true) {
171 char* line = nullptr;
172 size_t n;
173 if (getline(&line, &n, stdin) == -1) {
174 die(0, "(EOF/read error; assuming [N]one...)");
175 overwrite_mode = kNever;
176 return false;
177 }
178 if (n == 0) continue;
179 char cmd = line[0];
180 free(line);
181 switch (cmd) {
182 case 'y':
183 return true;
184 case 'n':
185 return false;
186 case 'A':
187 overwrite_mode = kAlways;
188 return true;
189 case 'N':
190 overwrite_mode = kNever;
191 return false;
192 }
193 }
194 }
195
ExtractToPipe(ZipArchiveHandle zah,ZipEntry & entry,const std::string & name)196 static void ExtractToPipe(ZipArchiveHandle zah, ZipEntry& entry, const std::string& name) {
197 // We need to extract to memory because ExtractEntryToFile insists on
198 // being able to seek and truncate, and you can't do that with stdout.
199 uint8_t* buffer = new uint8_t[entry.uncompressed_length];
200 int err = ExtractToMemory(zah, &entry, buffer, entry.uncompressed_length);
201 if (err < 0) {
202 die(0, "failed to extract %s: %s", name.c_str(), ErrorCodeString(err));
203 }
204 if (!android::base::WriteFully(1, buffer, entry.uncompressed_length)) {
205 die(errno, "failed to write %s to stdout", name.c_str());
206 }
207 delete[] buffer;
208 }
209
ExtractOne(ZipArchiveHandle zah,ZipEntry & entry,const std::string & name)210 static void ExtractOne(ZipArchiveHandle zah, ZipEntry& entry, const std::string& name) {
211 // Bad filename?
212 if (StartsWith(name, "/") || StartsWith(name, "../") || name.find("/../") != std::string::npos) {
213 die(0, "bad filename %s", name.c_str());
214 }
215
216 // Where are we actually extracting to (for human-readable output)?
217 // flag_d is the empty string if -d wasn't used, or has a trailing '/'
218 // otherwise.
219 std::string dst = flag_d + name;
220
221 // Ensure the directory hierarchy exists.
222 if (!MakeDirectoryHierarchy(android::base::Dirname(name))) {
223 die(errno, "couldn't create directory hierarchy for %s", dst.c_str());
224 }
225
226 // An entry in a zip file can just be a directory itself.
227 if (EndsWith(name, "/")) {
228 if (mkdir(name.c_str(), entry.unix_mode) == -1) {
229 // If the directory already exists, that's fine.
230 if (errno == EEXIST) {
231 struct stat sb;
232 if (stat(name.c_str(), &sb) != -1 && S_ISDIR(sb.st_mode)) return;
233 }
234 die(errno, "couldn't extract directory %s", dst.c_str());
235 }
236 return;
237 }
238
239 // Create the file.
240 int fd = open(name.c_str(), O_CREAT | O_WRONLY | O_CLOEXEC | O_EXCL, entry.unix_mode);
241 if (fd == -1 && errno == EEXIST) {
242 if (overwrite_mode == kNever) return;
243 if (overwrite_mode == kPrompt && !PromptOverwrite(dst)) return;
244 // Either overwrite_mode is kAlways or the user consented to this specific case.
245 fd = open(name.c_str(), O_WRONLY | O_CREAT | O_CLOEXEC | O_TRUNC, entry.unix_mode);
246 }
247 if (fd == -1) die(errno, "couldn't create file %s", dst.c_str());
248
249 // Actually extract into the file.
250 if (!flag_q) printf(" inflating: %s\n", dst.c_str());
251 int err = ExtractEntryToFile(zah, &entry, fd);
252 if (err < 0) die(0, "failed to extract %s: %s", dst.c_str(), ErrorCodeString(err));
253 close(fd);
254 }
255
ListOne(const ZipEntry & entry,const std::string & name)256 static void ListOne(const ZipEntry& entry, const std::string& name) {
257 tm t = entry.GetModificationTime();
258 char time[32];
259 snprintf(time, sizeof(time), "%04d-%02d-%02d %02d:%02d", t.tm_year + 1900, t.tm_mon + 1,
260 t.tm_mday, t.tm_hour, t.tm_min);
261 if (flag_v) {
262 printf("%8d %s %7d %3.0f%% %s %08x %s\n", entry.uncompressed_length,
263 (entry.method == kCompressStored) ? "Stored" : "Defl:N", entry.compressed_length,
264 CompressionRatio(entry.uncompressed_length, entry.compressed_length), time, entry.crc32,
265 name.c_str());
266 } else {
267 printf("%9d %s %s\n", entry.uncompressed_length, time, name.c_str());
268 }
269 }
270
InfoOne(const ZipEntry & entry,const std::string & name)271 static void InfoOne(const ZipEntry& entry, const std::string& name) {
272 if (flag_1) {
273 // "android-ndk-r19b/sources/android/NOTICE"
274 printf("%s\n", name.c_str());
275 return;
276 }
277
278 int version = entry.version_made_by & 0xff;
279 int os = (entry.version_made_by >> 8) & 0xff;
280
281 // TODO: Support suid/sgid? Non-Unix/non-FAT host file system attributes?
282 const char* src_fs = "???";
283 char mode[] = "??? ";
284 if (os == 0) {
285 src_fs = "fat";
286 // https://docs.microsoft.com/en-us/windows/win32/fileio/file-attribute-constants
287 int attrs = entry.external_file_attributes & 0xff;
288 mode[0] = (attrs & 0x10) ? 'd' : '-';
289 mode[1] = 'r';
290 mode[2] = (attrs & 0x01) ? '-' : 'w';
291 // The man page also mentions ".btm", but that seems to be obsolete?
292 mode[3] = EndsWith(name, ".exe") || EndsWith(name, ".com") || EndsWith(name, ".bat") ||
293 EndsWith(name, ".cmd")
294 ? 'x'
295 : '-';
296 mode[4] = (attrs & 0x20) ? 'a' : '-';
297 mode[5] = (attrs & 0x02) ? 'h' : '-';
298 mode[6] = (attrs & 0x04) ? 's' : '-';
299 } else if (os == 3) {
300 src_fs = "unx";
301 mode[0] = S_ISDIR(entry.unix_mode) ? 'd' : (S_ISREG(entry.unix_mode) ? '-' : '?');
302 mode[1] = entry.unix_mode & S_IRUSR ? 'r' : '-';
303 mode[2] = entry.unix_mode & S_IWUSR ? 'w' : '-';
304 mode[3] = entry.unix_mode & S_IXUSR ? 'x' : '-';
305 mode[4] = entry.unix_mode & S_IRGRP ? 'r' : '-';
306 mode[5] = entry.unix_mode & S_IWGRP ? 'w' : '-';
307 mode[6] = entry.unix_mode & S_IXGRP ? 'x' : '-';
308 mode[7] = entry.unix_mode & S_IROTH ? 'r' : '-';
309 mode[8] = entry.unix_mode & S_IWOTH ? 'w' : '-';
310 mode[9] = entry.unix_mode & S_IXOTH ? 'x' : '-';
311 }
312
313 char method[5] = "stor";
314 if (entry.method == kCompressDeflated) {
315 snprintf(method, sizeof(method), "def%c", "NXFS"[(entry.gpbf >> 1) & 0x3]);
316 }
317
318 // TODO: zipinfo (unlike unzip) sometimes uses time zone?
319 // TODO: this uses 4-digit years because we're not barbarians unless interoperability forces it.
320 tm t = entry.GetModificationTime();
321 char time[32];
322 snprintf(time, sizeof(time), "%04d-%02d-%02d %02d:%02d", t.tm_year + 1900, t.tm_mon + 1,
323 t.tm_mday, t.tm_hour, t.tm_min);
324
325 // "-rw-r--r-- 3.0 unx 577 t- defX 19-Feb-12 16:09 android-ndk-r19b/sources/android/NOTICE"
326 printf("%s %2d.%d %s %8d %c%c %s %s %s\n", mode, version / 10, version % 10, src_fs,
327 entry.uncompressed_length, entry.is_text ? 't' : 'b',
328 entry.has_data_descriptor ? 'X' : 'x', method, time, name.c_str());
329 }
330
ProcessOne(ZipArchiveHandle zah,ZipEntry & entry,const std::string & name)331 static void ProcessOne(ZipArchiveHandle zah, ZipEntry& entry, const std::string& name) {
332 if (role == kUnzip) {
333 if (flag_l || flag_v) {
334 // -l or -lv or -lq or -v.
335 ListOne(entry, name);
336 } else {
337 // Actually extract.
338 if (flag_p) {
339 ExtractToPipe(zah, entry, name);
340 } else {
341 ExtractOne(zah, entry, name);
342 }
343 }
344 } else {
345 // zipinfo or zipinfo -1.
346 InfoOne(entry, name);
347 }
348 total_uncompressed_length += entry.uncompressed_length;
349 total_compressed_length += entry.compressed_length;
350 ++file_count;
351 }
352
ProcessAll(ZipArchiveHandle zah)353 static void ProcessAll(ZipArchiveHandle zah) {
354 MaybeShowHeader(zah);
355
356 // libziparchive iteration order doesn't match the central directory.
357 // We could sort, but that would cost extra and wouldn't match either.
358 void* cookie;
359 int err = StartIteration(zah, &cookie);
360 if (err != 0) {
361 die(0, "couldn't iterate %s: %s", archive_name, ErrorCodeString(err));
362 }
363
364 ZipEntry entry;
365 std::string name;
366 while ((err = Next(cookie, &entry, &name)) >= 0) {
367 if (ShouldInclude(name)) ProcessOne(zah, entry, name);
368 }
369
370 if (err < -1) die(0, "failed iterating %s: %s", archive_name, ErrorCodeString(err));
371 EndIteration(cookie);
372
373 MaybeShowFooter();
374 }
375
ShowHelp(bool full)376 static void ShowHelp(bool full) {
377 if (role == kUnzip) {
378 fprintf(full ? stdout : stderr, "usage: unzip [-d DIR] [-lnopqv] ZIP [FILE...] [-x FILE...]\n");
379 if (!full) exit(EXIT_FAILURE);
380
381 printf(
382 "\n"
383 "Extract FILEs from ZIP archive. Default is all files. Both the include and\n"
384 "exclude (-x) lists use shell glob patterns.\n"
385 "\n"
386 "-d DIR Extract into DIR\n"
387 "-l List contents (-lq excludes archive name, -lv is verbose)\n"
388 "-n Never overwrite files (default: prompt)\n"
389 "-o Always overwrite files\n"
390 "-p Pipe to stdout\n"
391 "-q Quiet\n"
392 "-v List contents verbosely\n"
393 "-x FILE Exclude files\n");
394 } else {
395 fprintf(full ? stdout : stderr, "usage: zipinfo [-1] ZIP [FILE...] [-x FILE...]\n");
396 if (!full) exit(EXIT_FAILURE);
397
398 printf(
399 "\n"
400 "Show information about FILEs from ZIP archive. Default is all files.\n"
401 "Both the include and exclude (-x) lists use shell glob patterns.\n"
402 "\n"
403 "-1 Show filenames only, one per line\n"
404 "-x FILE Exclude files\n");
405 }
406 exit(EXIT_SUCCESS);
407 }
408
HandleCommonOption(int opt)409 static void HandleCommonOption(int opt) {
410 switch (opt) {
411 case 'h':
412 ShowHelp(true);
413 break;
414 case 'x':
415 flag_x = true;
416 break;
417 case 1:
418 // -x swallows all following arguments, so we use '-' in the getopt
419 // string and collect files here.
420 if (!archive_name) {
421 archive_name = optarg;
422 } else if (flag_x) {
423 excludes.insert(optarg);
424 } else {
425 includes.insert(optarg);
426 }
427 break;
428 default:
429 ShowHelp(false);
430 break;
431 }
432 }
433
main(int argc,char * argv[])434 int main(int argc, char* argv[]) {
435 // Who am I, and what am I doing?
436 g_progname = basename(argv[0]);
437 if (!strcmp(g_progname, "ziptool") && argc > 1) return main(argc - 1, argv + 1);
438 if (!strcmp(g_progname, "unzip")) {
439 role = kUnzip;
440 } else if (!strcmp(g_progname, "zipinfo")) {
441 role = kZipinfo;
442 } else {
443 die(0, "run as ziptool with unzip or zipinfo as the first argument, or symlink");
444 }
445
446 static const struct option opts[] = {
447 {"help", no_argument, 0, 'h'},
448 {},
449 };
450
451 if (role == kUnzip) {
452 // `unzip -Z` is "zipinfo mode", so in that case just restart...
453 if (argc > 1 && !strcmp(argv[1], "-Z")) {
454 argv[1] = const_cast<char*>("zipinfo");
455 return main(argc - 1, argv + 1);
456 }
457
458 int opt;
459 while ((opt = getopt_long(argc, argv, "-d:hlnopqvx", opts, nullptr)) != -1) {
460 switch (opt) {
461 case 'd':
462 flag_d = optarg;
463 if (!EndsWith(flag_d, "/")) flag_d += '/';
464 break;
465 case 'l':
466 flag_l = true;
467 break;
468 case 'n':
469 overwrite_mode = kNever;
470 break;
471 case 'o':
472 overwrite_mode = kAlways;
473 break;
474 case 'p':
475 flag_p = flag_q = true;
476 break;
477 case 'q':
478 flag_q = true;
479 break;
480 case 'v':
481 flag_v = true;
482 break;
483 default:
484 HandleCommonOption(opt);
485 break;
486 }
487 }
488 } else {
489 int opt;
490 while ((opt = getopt_long(argc, argv, "-1hx", opts, nullptr)) != -1) {
491 switch (opt) {
492 case '1':
493 flag_1 = true;
494 break;
495 default:
496 HandleCommonOption(opt);
497 break;
498 }
499 }
500 }
501
502 if (!archive_name) die(0, "missing archive filename");
503
504 // We can't support "-" to unzip from stdin because libziparchive relies on mmap.
505 ZipArchiveHandle zah;
506 int32_t err;
507 if ((err = OpenArchive(archive_name, &zah)) != 0) {
508 die(0, "couldn't open %s: %s", archive_name, ErrorCodeString(err));
509 }
510
511 // Implement -d by changing into that directory.
512 // We'll create implicit directories based on paths in the zip file, and we'll create
513 // the -d directory itself, but we require that *parents* of the -d directory already exists.
514 // This is pretty arbitrary, but it's the behavior of the original unzip.
515 if (!flag_d.empty()) {
516 if (mkdir(flag_d.c_str(), 0777) == -1 && errno != EEXIST) {
517 die(errno, "couldn't created %s", flag_d.c_str());
518 }
519 if (chdir(flag_d.c_str()) == -1) {
520 die(errno, "couldn't chdir to %s", flag_d.c_str());
521 }
522 }
523
524 ProcessAll(zah);
525
526 CloseArchive(zah);
527 return 0;
528 }
529