| Line |
Branch |
Exec |
Source |
| 1 |
|
|
/** |
| 2 |
|
|
* This file is part of the CernVM File System. |
| 3 |
|
|
* |
| 4 |
|
|
* Process a set of input files and create appropriate graft files. |
| 5 |
|
|
*/ |
| 6 |
|
|
|
| 7 |
|
|
#include "swissknife_graft.h" |
| 8 |
|
|
#include "cvmfs_config.h" |
| 9 |
|
|
|
| 10 |
|
|
#include <fcntl.h> |
| 11 |
|
|
#include <unistd.h> |
| 12 |
|
|
|
| 13 |
|
|
#include <cstdio> |
| 14 |
|
|
#include <vector> |
| 15 |
|
|
|
| 16 |
|
|
#include "crypto/hash.h" |
| 17 |
|
|
#include "util/fs_traversal.h" |
| 18 |
|
|
#include "util/platform.h" |
| 19 |
|
|
#include "util/posix.h" |
| 20 |
|
|
|
| 21 |
|
✗ |
bool swissknife::CommandGraft::ChecksumFdWithChunks( |
| 22 |
|
|
int fd, zlib::Compressor *compressor, uint64_t *file_size, |
| 23 |
|
|
shash::Any *file_hash, std::vector<uint64_t> *chunk_offsets, |
| 24 |
|
|
std::vector<shash::Any> *chunk_checksums) { |
| 25 |
|
✗ |
if (!compressor || !file_size || !file_hash) { |
| 26 |
|
✗ |
return false; |
| 27 |
|
|
} |
| 28 |
|
✗ |
*file_size = 0; |
| 29 |
|
✗ |
shash::Any chunk_hash(hash_alg_); |
| 30 |
|
|
ssize_t bytes_read; |
| 31 |
|
|
unsigned char in_buf[zlib::kZChunk]; |
| 32 |
|
✗ |
unsigned char *cur_in_buf = in_buf; |
| 33 |
|
✗ |
size_t in_buf_size = zlib::kZChunk; |
| 34 |
|
|
unsigned char out_buf[zlib::kZChunk]; |
| 35 |
|
✗ |
size_t avail_in = 0; |
| 36 |
|
|
|
| 37 |
|
|
// Initialize the file and per-chunk checksums |
| 38 |
|
✗ |
shash::ContextPtr file_hash_context(hash_alg_); |
| 39 |
|
✗ |
file_hash_context.buffer = alloca(file_hash_context.size); |
| 40 |
|
✗ |
shash::Init(file_hash_context); |
| 41 |
|
|
|
| 42 |
|
✗ |
shash::ContextPtr chunk_hash_context(hash_alg_); |
| 43 |
|
✗ |
chunk_hash_context.buffer = alloca(chunk_hash_context.size); |
| 44 |
|
✗ |
shash::Init(chunk_hash_context); |
| 45 |
|
|
|
| 46 |
|
✗ |
bool do_chunk = chunk_size_ > 0; |
| 47 |
|
✗ |
if (do_chunk) { |
| 48 |
|
✗ |
if (!chunk_offsets || !chunk_checksums) { |
| 49 |
|
✗ |
return false; |
| 50 |
|
|
} |
| 51 |
|
✗ |
chunk_offsets->push_back(0); |
| 52 |
|
|
} |
| 53 |
|
|
|
| 54 |
|
✗ |
bool flush = 0; |
| 55 |
|
|
do { |
| 56 |
|
✗ |
bytes_read = read(fd, cur_in_buf + avail_in, in_buf_size); |
| 57 |
|
✗ |
if (-1 == bytes_read) { |
| 58 |
|
✗ |
if (errno == EINTR) { |
| 59 |
|
✗ |
continue; |
| 60 |
|
|
} |
| 61 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "Failure when reading file: %s", |
| 62 |
|
|
strerror(errno)); |
| 63 |
|
✗ |
return false; |
| 64 |
|
|
} |
| 65 |
|
✗ |
*file_size += bytes_read; |
| 66 |
|
✗ |
avail_in += bytes_read; |
| 67 |
|
|
|
| 68 |
|
✗ |
flush = (static_cast<size_t>(bytes_read) < in_buf_size); |
| 69 |
|
|
|
| 70 |
|
|
// If possible, make progress on deflate. |
| 71 |
|
✗ |
unsigned char *cur_out_buf = out_buf; |
| 72 |
|
✗ |
size_t avail_out = zlib::kZChunk; |
| 73 |
|
✗ |
compressor->Deflate(flush, &cur_in_buf, &avail_in, &cur_out_buf, |
| 74 |
|
|
&avail_out); |
| 75 |
|
✗ |
if (do_chunk) { |
| 76 |
|
✗ |
shash::Update(out_buf, avail_out, chunk_hash_context); |
| 77 |
|
✗ |
if (generate_bulk_hash_) |
| 78 |
|
✗ |
shash::Update(out_buf, avail_out, file_hash_context); |
| 79 |
|
|
} else { |
| 80 |
|
✗ |
shash::Update(out_buf, avail_out, file_hash_context); |
| 81 |
|
|
} |
| 82 |
|
|
|
| 83 |
|
✗ |
if (!avail_in) { |
| 84 |
|
|
// All bytes are consumed; set the buffer back to the beginning. |
| 85 |
|
✗ |
cur_in_buf = in_buf; |
| 86 |
|
✗ |
in_buf_size = zlib::kZChunk; |
| 87 |
|
|
} else { |
| 88 |
|
✗ |
in_buf_size = zlib::kZChunk - (cur_in_buf - in_buf) - avail_in; |
| 89 |
|
|
} |
| 90 |
|
|
|
| 91 |
|
|
// Start a new hash if current one is above threshold |
| 92 |
|
✗ |
if (do_chunk && (*file_size - chunk_offsets->back() >= chunk_size_)) { |
| 93 |
|
✗ |
shash::Final(chunk_hash_context, &chunk_hash); |
| 94 |
|
✗ |
chunk_offsets->push_back(*file_size); |
| 95 |
|
✗ |
chunk_checksums->push_back(chunk_hash); |
| 96 |
|
✗ |
shash::Init(chunk_hash_context); |
| 97 |
|
|
} |
| 98 |
|
✗ |
} while (!flush); |
| 99 |
|
|
|
| 100 |
|
✗ |
shash::Final(file_hash_context, file_hash); |
| 101 |
|
✗ |
if (do_chunk) { |
| 102 |
|
✗ |
shash::Final(chunk_hash_context, &chunk_hash); |
| 103 |
|
✗ |
chunk_checksums->push_back(chunk_hash); |
| 104 |
|
|
} |
| 105 |
|
|
|
| 106 |
|
|
// Zero-size chunks are not allowed; except if there is only one chunk |
| 107 |
|
✗ |
if (do_chunk && (chunk_offsets->back() == *file_size) && |
| 108 |
|
✗ |
(chunk_offsets->size() > 1)) |
| 109 |
|
|
{ |
| 110 |
|
✗ |
chunk_offsets->pop_back(); |
| 111 |
|
✗ |
chunk_checksums->pop_back(); |
| 112 |
|
|
} |
| 113 |
|
|
|
| 114 |
|
✗ |
if (do_chunk && !generate_bulk_hash_) |
| 115 |
|
✗ |
file_hash->SetNull(); |
| 116 |
|
|
|
| 117 |
|
|
// Do not chunk a file if it is under threshold. |
| 118 |
|
✗ |
if (do_chunk && (chunk_offsets->size() == 1)) { |
| 119 |
|
✗ |
*file_hash = (*chunk_checksums)[0]; |
| 120 |
|
✗ |
chunk_offsets->clear(); |
| 121 |
|
✗ |
chunk_checksums->clear(); |
| 122 |
|
|
} |
| 123 |
|
✗ |
return true; |
| 124 |
|
|
} |
| 125 |
|
|
|
| 126 |
|
✗ |
bool swissknife::CommandGraft::DirCallback(const std::string &relative_path, |
| 127 |
|
|
const std::string &dir_name) { |
| 128 |
|
✗ |
if (!output_file_.size()) { |
| 129 |
|
✗ |
return true; |
| 130 |
|
|
} |
| 131 |
|
✗ |
std::string full_output_path = output_file_ + "/" + |
| 132 |
|
✗ |
(relative_path.size() ? relative_path : ".") + |
| 133 |
|
✗ |
"/" + dir_name; |
| 134 |
|
✗ |
std::string full_input_path = input_file_ + "/" + |
| 135 |
|
✗ |
(relative_path.size() ? relative_path : ".") + |
| 136 |
|
✗ |
"/" + dir_name; |
| 137 |
|
|
platform_stat64 sbuf; |
| 138 |
|
✗ |
if (-1 == platform_stat(full_input_path.c_str(), &sbuf)) { |
| 139 |
|
✗ |
return false; |
| 140 |
|
|
} |
| 141 |
|
✗ |
return MkdirDeep(full_output_path, sbuf.st_mode); |
| 142 |
|
|
} |
| 143 |
|
|
|
| 144 |
|
✗ |
void swissknife::CommandGraft::FileCallback(const std::string &relative_path, |
| 145 |
|
|
const std::string &file_name) { |
| 146 |
|
✗ |
std::string full_input_path = input_file_ + "/" + |
| 147 |
|
✗ |
(relative_path.size() ? relative_path : ".") + |
| 148 |
|
✗ |
"/" + file_name; |
| 149 |
|
✗ |
std::string full_output_path; |
| 150 |
|
✗ |
if (output_file_.size()) { |
| 151 |
|
✗ |
full_output_path = output_file_ + "/" + |
| 152 |
|
✗ |
(relative_path.size() ? relative_path : ".") + "/" + |
| 153 |
|
✗ |
file_name; |
| 154 |
|
|
} |
| 155 |
|
✗ |
Publish(full_input_path, full_output_path, false, false); |
| 156 |
|
|
} |
| 157 |
|
|
|
| 158 |
|
✗ |
int swissknife::CommandGraft::Main(const swissknife::ArgumentList &args) { |
| 159 |
|
✗ |
const std::string &input_file = *args.find('i')->second; |
| 160 |
|
|
const std::string output_file = |
| 161 |
|
✗ |
(args.find('o') == args.end()) ? "" : *(args.find('o')->second); |
| 162 |
|
✗ |
verbose_ = args.find('v') != args.end(); |
| 163 |
|
✗ |
generate_bulk_hash_ = args.find('b') != args.end(); |
| 164 |
|
✗ |
hash_alg_ = (args.find('a') == args.end()) |
| 165 |
|
✗ |
? shash::kSha1 |
| 166 |
|
✗ |
: shash::ParseHashAlgorithm(*args.find('a')->second); |
| 167 |
|
✗ |
compression_alg_ = |
| 168 |
|
✗ |
(args.find('Z') == args.end()) |
| 169 |
|
✗ |
? zlib::kNoCompression |
| 170 |
|
✗ |
: zlib::ParseCompressionAlgorithm(*args.find('Z')->second); |
| 171 |
|
|
|
| 172 |
|
✗ |
if (args.find('c') == args.end()) { |
| 173 |
|
✗ |
chunk_size_ = kDefaultChunkSize; |
| 174 |
|
|
} else { |
| 175 |
|
✗ |
std::string chunk_size = *args.find('c')->second; |
| 176 |
|
✗ |
if (!String2Uint64Parse(chunk_size, &chunk_size_)) { |
| 177 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "Unable to parse chunk size: %s", |
| 178 |
|
|
chunk_size.c_str()); |
| 179 |
|
✗ |
return 1; |
| 180 |
|
|
} |
| 181 |
|
|
} |
| 182 |
|
✗ |
chunk_size_ *= 1024 * 1024; // Convert to MB. |
| 183 |
|
|
|
| 184 |
|
|
platform_stat64 sbuf; |
| 185 |
|
✗ |
bool output_file_is_dir = output_file.size() && |
| 186 |
|
✗ |
(0 == platform_stat(output_file.c_str(), &sbuf)) && |
| 187 |
|
✗ |
S_ISDIR(sbuf.st_mode); |
| 188 |
|
✗ |
if (output_file_is_dir && (input_file == "-")) { |
| 189 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "Output file (%s): Is a directory\n", |
| 190 |
|
|
output_file.c_str()); |
| 191 |
|
✗ |
return 1; |
| 192 |
|
|
} |
| 193 |
|
|
|
| 194 |
|
✗ |
if (input_file != "-") { |
| 195 |
|
✗ |
bool input_file_is_dir = (0 == platform_stat(input_file.c_str(), &sbuf)) && |
| 196 |
|
✗ |
S_ISDIR(sbuf.st_mode); |
| 197 |
|
✗ |
if (input_file_is_dir) { |
| 198 |
|
✗ |
if (!output_file_is_dir && output_file.size()) { |
| 199 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 200 |
|
|
"Input (%s) is a directory but output" |
| 201 |
|
|
" (%s) is not\n", |
| 202 |
|
|
input_file.c_str(), output_file.c_str()); |
| 203 |
|
✗ |
return 1; |
| 204 |
|
|
} |
| 205 |
|
✗ |
if (verbose_) { |
| 206 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "Recursing into directory %s\n", |
| 207 |
|
|
input_file.c_str()); |
| 208 |
|
|
} |
| 209 |
|
✗ |
return Recurse(input_file, output_file); |
| 210 |
|
|
} else { |
| 211 |
|
✗ |
return Publish(input_file, output_file, output_file_is_dir, false); |
| 212 |
|
|
} |
| 213 |
|
|
} |
| 214 |
|
✗ |
return Publish(input_file, output_file, output_file_is_dir, true); |
| 215 |
|
|
} |
| 216 |
|
|
|
| 217 |
|
✗ |
int swissknife::CommandGraft::Publish(const std::string &input_file, |
| 218 |
|
|
const std::string &output_file, |
| 219 |
|
|
bool output_file_is_dir, |
| 220 |
|
|
bool input_file_is_stdin) { |
| 221 |
|
✗ |
if (output_file.size() && verbose_) { |
| 222 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, "Grafting %s to %s", input_file.c_str(), |
| 223 |
|
|
output_file.c_str()); |
| 224 |
|
✗ |
} else if (!output_file.size()) { |
| 225 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, "Grafting %s", input_file.c_str()); |
| 226 |
|
|
} |
| 227 |
|
|
int fd; |
| 228 |
|
✗ |
if (input_file_is_stdin) { |
| 229 |
|
✗ |
fd = 0; |
| 230 |
|
|
} else { |
| 231 |
|
✗ |
fd = open(input_file.c_str(), O_RDONLY); |
| 232 |
|
✗ |
if (fd < 0) { |
| 233 |
|
✗ |
std::string errmsg = "Unable to open input file (" + input_file + ")"; |
| 234 |
|
✗ |
perror(errmsg.c_str()); |
| 235 |
|
✗ |
return 1; |
| 236 |
|
|
} |
| 237 |
|
|
} |
| 238 |
|
|
|
| 239 |
|
|
// Get input file mode; output file will be set identically. |
| 240 |
|
|
platform_stat64 sbuf; |
| 241 |
|
✗ |
if (-1 == platform_fstat(fd, &sbuf)) { |
| 242 |
|
✗ |
std::string errmsg = "Unable to stat input file (" + input_file + ")"; |
| 243 |
|
✗ |
perror(errmsg.c_str()); |
| 244 |
|
|
} |
| 245 |
|
✗ |
mode_t input_file_mode = input_file_is_stdin ? 0644 : sbuf.st_mode; |
| 246 |
|
|
|
| 247 |
|
✗ |
shash::Any file_hash(hash_alg_); |
| 248 |
|
|
uint64_t processed_size; |
| 249 |
|
✗ |
std::vector<uint64_t> chunk_offsets; |
| 250 |
|
✗ |
std::vector<shash::Any> chunk_checksums; |
| 251 |
|
✗ |
zlib::Compressor *compressor = zlib::Compressor::Construct(compression_alg_); |
| 252 |
|
|
|
| 253 |
|
|
bool retval = |
| 254 |
|
✗ |
ChecksumFdWithChunks(fd, compressor, &processed_size, &file_hash, |
| 255 |
|
|
&chunk_offsets, &chunk_checksums); |
| 256 |
|
|
|
| 257 |
|
✗ |
if (!input_file_is_stdin) { |
| 258 |
|
✗ |
close(fd); |
| 259 |
|
|
} |
| 260 |
|
✗ |
if (!retval) { |
| 261 |
|
✗ |
std::string errmsg = "Unable to checksum input file (" + input_file + ")"; |
| 262 |
|
✗ |
perror(errmsg.c_str()); |
| 263 |
|
✗ |
return 1; |
| 264 |
|
|
} |
| 265 |
|
|
|
| 266 |
|
|
// Build the .cvmfsgraft-$filename |
| 267 |
|
✗ |
if (output_file.size()) { |
| 268 |
|
✗ |
std::string dirname, fname; |
| 269 |
|
✗ |
std::string graft_fname; |
| 270 |
|
✗ |
if (output_file_is_dir) { |
| 271 |
|
✗ |
SplitPath(input_file, &dirname, &fname); |
| 272 |
|
✗ |
graft_fname = output_file + "/.cvmfsgraft-" + fname; |
| 273 |
|
|
} else { |
| 274 |
|
✗ |
SplitPath(output_file, &dirname, &fname); |
| 275 |
|
✗ |
graft_fname = dirname + "/.cvmfsgraft-" + fname; |
| 276 |
|
|
} |
| 277 |
|
✗ |
fd = open(graft_fname.c_str(), O_CREAT | O_TRUNC | O_WRONLY, 0644); |
| 278 |
|
✗ |
if (fd < 0) { |
| 279 |
|
✗ |
std::string errmsg = "Unable to open graft file (" + graft_fname + ")"; |
| 280 |
|
✗ |
perror(errmsg.c_str()); |
| 281 |
|
✗ |
return 1; |
| 282 |
|
|
} |
| 283 |
|
✗ |
} else { |
| 284 |
|
✗ |
fd = 1; |
| 285 |
|
|
} |
| 286 |
|
✗ |
const bool with_suffix = true; |
| 287 |
|
|
std::string graft_contents = |
| 288 |
|
✗ |
"size=" + StringifyInt(processed_size) + "\n" + |
| 289 |
|
✗ |
"checksum=" + file_hash.ToString(with_suffix) + "\n" + |
| 290 |
|
✗ |
"compression=" + zlib::AlgorithmName(compression_alg_) + "\n"; |
| 291 |
|
✗ |
if (!chunk_offsets.empty()) { |
| 292 |
|
✗ |
std::vector<std::string> chunk_off_str; |
| 293 |
|
✗ |
chunk_off_str.reserve(chunk_offsets.size()); |
| 294 |
|
✗ |
std::vector<std::string> chunk_ck_str; |
| 295 |
|
✗ |
chunk_ck_str.reserve(chunk_offsets.size()); |
| 296 |
|
✗ |
for (unsigned idx = 0; idx < chunk_offsets.size(); idx++) { |
| 297 |
|
✗ |
chunk_off_str.push_back(StringifyInt(chunk_offsets[idx])); |
| 298 |
|
✗ |
chunk_ck_str.push_back(chunk_checksums[idx].ToStringWithSuffix()); |
| 299 |
|
|
} |
| 300 |
|
✗ |
graft_contents += "chunk_offsets=" + JoinStrings(chunk_off_str, ",") + "\n"; |
| 301 |
|
|
graft_contents += |
| 302 |
|
✗ |
"chunk_checksums=" + JoinStrings(chunk_ck_str, ",") + "\n"; |
| 303 |
|
|
} |
| 304 |
|
✗ |
size_t nbytes = graft_contents.size(); |
| 305 |
|
✗ |
const char *buf = graft_contents.c_str(); |
| 306 |
|
✗ |
retval = SafeWrite(fd, buf, nbytes); |
| 307 |
|
✗ |
if (!retval) { |
| 308 |
|
✗ |
perror("Failed writing to graft file"); |
| 309 |
|
✗ |
close(fd); |
| 310 |
|
✗ |
return 1; |
| 311 |
|
|
} |
| 312 |
|
✗ |
if (output_file.size()) { |
| 313 |
|
✗ |
close(fd); |
| 314 |
|
|
} else { |
| 315 |
|
✗ |
return 0; |
| 316 |
|
|
} |
| 317 |
|
|
|
| 318 |
|
|
// Create and truncate the output file. |
| 319 |
|
✗ |
std::string output_fname; |
| 320 |
|
✗ |
if (output_file_is_dir) { |
| 321 |
|
✗ |
output_fname = output_file + "/" + GetFileName(input_file); |
| 322 |
|
|
} else { |
| 323 |
|
✗ |
output_fname = output_file; |
| 324 |
|
|
} |
| 325 |
|
|
fd = |
| 326 |
|
✗ |
open(output_fname.c_str(), O_CREAT | O_TRUNC | O_WRONLY, input_file_mode); |
| 327 |
|
✗ |
if (fd < 0) { |
| 328 |
|
✗ |
std::string errmsg = "Unable to open output file (" + output_file + ")"; |
| 329 |
|
✗ |
perror(errmsg.c_str()); |
| 330 |
|
✗ |
return 1; |
| 331 |
|
|
} |
| 332 |
|
✗ |
close(fd); |
| 333 |
|
|
|
| 334 |
|
✗ |
return 0; |
| 335 |
|
|
} |
| 336 |
|
|
|
| 337 |
|
✗ |
int swissknife::CommandGraft::Recurse(const std::string &input_file, |
| 338 |
|
|
const std::string &output_file) { |
| 339 |
|
✗ |
output_file_ = output_file; |
| 340 |
|
✗ |
input_file_ = input_file; |
| 341 |
|
|
|
| 342 |
|
✗ |
FileSystemTraversal<CommandGraft> traverser(this, input_file, true); |
| 343 |
|
✗ |
traverser.fn_new_file = &CommandGraft::FileCallback; |
| 344 |
|
✗ |
traverser.fn_new_dir_prefix = &CommandGraft::DirCallback; |
| 345 |
|
✗ |
traverser.Recurse(input_file); |
| 346 |
|
✗ |
return 0; |
| 347 |
|
|
} |
| 348 |
|
|
|