Googler | 041f7ce | 2022-03-23 21:32:08 -0700 | [diff] [blame] | 1 | #!/usr/bin/python3 |
| 2 | |
| 3 | # Copyright 2015 The Bazel Authors. All rights reserved. |
| 4 | # |
| 5 | # Licensed under the Apache License, Version 2.0 (the "License"); |
| 6 | # you may not use this file except in compliance with the License. |
| 7 | # You may obtain a copy of the License at |
| 8 | # |
| 9 | # http://www.apache.org/licenses/LICENSE-2.0 |
| 10 | # |
| 11 | # Unless required by applicable law or agreed to in writing, software |
| 12 | # distributed under the License is distributed on an "AS IS" BASIS, |
| 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 14 | # See the License for the specific language governing permissions and |
| 15 | # limitations under the License. |
| 16 | |
| 17 | """A script for J2ObjC dead code removal in Blaze. |
| 18 | |
| 19 | This script removes unused J2ObjC-translated classes from compilation and |
| 20 | linking by: |
| 21 | 1. Build a class dependency tree among translated source files. |
| 22 | 2. Use user-provided Java class entry points to get a list of reachable |
| 23 | classes. |
| 24 | 3. Go through all translated source files and rewrite unreachable ones with |
| 25 | dummy content. |
| 26 | """ |
| 27 | |
| 28 | import argparse |
| 29 | import collections |
| 30 | import multiprocessing |
| 31 | import os |
| 32 | import queue |
| 33 | import re |
| 34 | import shlex |
| 35 | import shutil |
| 36 | import subprocess |
| 37 | import threading |
| 38 | |
| 39 | |
| 40 | PRUNED_SRC_CONTENT = 'static int DUMMY_unused __attribute__((unused,used)) = 0;' |
| 41 | |
| 42 | |
| 43 | def BuildReachabilityTree(dependency_mapping_files, file_open=open): |
| 44 | """Builds a reachability tree using entries from dependency mapping files. |
| 45 | |
| 46 | Args: |
| 47 | dependency_mapping_files: A comma separated list of J2ObjC-generated |
| 48 | dependency mapping files. |
| 49 | file_open: Reference to the builtin open function so it may be |
| 50 | overridden for testing. |
| 51 | Returns: |
| 52 | A dict mapping J2ObjC-generated source files to the corresponding direct |
| 53 | dependent source files. |
| 54 | """ |
| 55 | return BuildArtifactSourceTree(dependency_mapping_files, file_open) |
| 56 | |
| 57 | |
| 58 | def BuildHeaderMapping(header_mapping_files, file_open=open): |
| 59 | """Builds a mapping between Java classes and J2ObjC-translated header files. |
| 60 | |
| 61 | Args: |
| 62 | header_mapping_files: A comma separated list of J2ObjC-generated |
| 63 | header mapping files. |
| 64 | file_open: Reference to the builtin open function so it may be |
| 65 | overridden for testing. |
| 66 | Returns: |
| 67 | An ordered dict mapping Java class names to corresponding J2ObjC-translated |
| 68 | source files. |
| 69 | """ |
| 70 | header_mapping = collections.OrderedDict() |
| 71 | for header_mapping_file in header_mapping_files.split(','): |
| 72 | with file_open(header_mapping_file, 'r') as f: |
| 73 | for line in f: |
| 74 | java_class_name = line.strip().split('=')[0] |
| 75 | transpiled_file_name = os.path.splitext(line.strip().split('=')[1])[0] |
| 76 | header_mapping[java_class_name] = transpiled_file_name |
| 77 | return header_mapping |
| 78 | |
| 79 | |
| 80 | def BuildReachableFileSet(entry_classes, reachability_tree, header_mapping, |
| 81 | archive_source_file_mapping=None): |
| 82 | """Builds a set of reachable translated files from entry Java classes. |
| 83 | |
| 84 | Args: |
| 85 | entry_classes: A comma separated list of Java entry classes. |
| 86 | reachability_tree: A dict mapping translated files to their direct |
| 87 | dependencies. |
| 88 | header_mapping: A dict mapping Java class names to translated source files. |
| 89 | archive_source_file_mapping: A dict mapping source files to the associated |
| 90 | archive file that contains them. |
| 91 | Returns: |
| 92 | A set of reachable translated files from the given list of entry classes. |
| 93 | Raises: |
| 94 | Exception: If there is an entry class that is not being transpiled in this |
| 95 | j2objc_library. |
| 96 | """ |
| 97 | transpiled_entry_files = [] |
| 98 | for entry_class in entry_classes.split(','): |
| 99 | if entry_class not in header_mapping: |
| 100 | raise Exception( |
| 101 | entry_class + |
| 102 | ' is not in the transitive Java deps of included ' + |
| 103 | 'j2objc_library rules.') |
| 104 | transpiled_entry_files.append(header_mapping[entry_class]) |
| 105 | |
| 106 | # Translated files going into the same static library archive with duplicated |
| 107 | # base names also need to be added to the set of entry files. |
| 108 | # |
| 109 | # This edge case is ignored because we currently cannot correctly perform |
| 110 | # dead code removal in this case. The object file entries in static library |
| 111 | # archives are named by the base names of the original source files. If two |
| 112 | # source files (e.g., foo/bar.m, bar/bar.m) go into the same archive and |
| 113 | # share the same base name (bar.m), their object file entries inside the |
| 114 | # archive will have the same name (bar.o). We cannot correctly handle this |
| 115 | # case because current archive tools (ar, ranlib, etc.) do not handle this |
| 116 | # case very well. |
| 117 | if archive_source_file_mapping: |
| 118 | transpiled_entry_files.extend(_DuplicatedFiles(archive_source_file_mapping)) |
| 119 | |
| 120 | # Translated files from package-info.java are also added to the entry files |
| 121 | # because they are needed to resolve ObjC class names with prefixes and these |
| 122 | # files may also have dependencies. |
| 123 | for transpiled_file in reachability_tree: |
| 124 | if transpiled_file.endswith('package-info'): |
| 125 | transpiled_entry_files.append(transpiled_file) |
| 126 | |
| 127 | reachable_files = set() |
| 128 | for transpiled_entry_file in transpiled_entry_files: |
| 129 | reachable_files.add(transpiled_entry_file) |
| 130 | current_level_deps = [] |
| 131 | # We need to check if the transpiled file is in the reachability tree |
| 132 | # because J2ObjC protos are not analyzed for dead code stripping and |
| 133 | # therefore are not in the reachability tree at all. |
| 134 | if transpiled_entry_file in reachability_tree: |
| 135 | current_level_deps = reachability_tree[transpiled_entry_file] |
| 136 | while current_level_deps: |
| 137 | next_level_deps = [] |
| 138 | for dep in current_level_deps: |
| 139 | if dep not in reachable_files: |
| 140 | reachable_files.add(dep) |
| 141 | if dep in reachability_tree: |
| 142 | next_level_deps.extend(reachability_tree[dep]) |
| 143 | current_level_deps = next_level_deps |
| 144 | return reachable_files |
| 145 | |
| 146 | |
| 147 | def PruneFiles(input_files, output_files, objc_file_path, reachable_files, |
| 148 | file_open=open, file_shutil=shutil): |
| 149 | """Copies over translated files and remove the contents of unreachable files. |
| 150 | |
| 151 | Args: |
| 152 | input_files: A comma separated list of input source files to prune. It has |
| 153 | a one-on-one pair mapping with the output_file list. |
| 154 | output_files: A comma separated list of output source files to write pruned |
| 155 | source files to. It has a one-on-one pair mapping with the input_file |
| 156 | list. |
| 157 | objc_file_path: The file path which represents a directory where the |
| 158 | generated ObjC files reside. |
| 159 | reachable_files: A set of reachable source files. |
| 160 | file_open: Reference to the builtin open function so it may be |
| 161 | overridden for testing. |
| 162 | file_shutil: Reference to the builtin shutil module so it may be |
| 163 | overridden for testing. |
| 164 | Returns: |
| 165 | None. |
| 166 | """ |
| 167 | file_queue = queue.queue() |
| 168 | for input_file, output_file in zip( |
| 169 | input_files.split(','), |
| 170 | output_files.split(',')): |
| 171 | file_queue.put((input_file, output_file)) |
| 172 | |
| 173 | for _ in range(multiprocessing.cpu_count()): |
| 174 | t = threading.Thread(target=_PruneFile, args=(file_queue, |
| 175 | reachable_files, |
| 176 | objc_file_path, |
| 177 | file_open, |
| 178 | file_shutil)) |
| 179 | t.start() |
| 180 | |
| 181 | file_queue.join() |
| 182 | |
| 183 | |
| 184 | def _PruneFile(file_queue, reachable_files, objc_file_path, file_open=open, |
| 185 | file_shutil=shutil): |
| 186 | while True: |
| 187 | try: |
| 188 | input_file, output_file = file_queue.get_nowait() |
| 189 | except queue.Empty: |
| 190 | return |
| 191 | file_name = os.path.relpath(os.path.splitext(input_file)[0], |
| 192 | objc_file_path) |
| 193 | if file_name in reachable_files: |
| 194 | file_shutil.copy(input_file, output_file) |
| 195 | else: |
| 196 | with file_open(output_file, 'w') as f: |
| 197 | # Use a static variable scoped to the source file to suppress |
| 198 | # the "has no symbols" linker warning for empty object files. |
| 199 | f.write(PRUNED_SRC_CONTENT) |
| 200 | file_queue.task_done() |
| 201 | |
| 202 | |
| 203 | def _DuplicatedFiles(archive_source_file_mapping): |
| 204 | """Returns a list of file with duplicated base names in each archive file. |
| 205 | |
| 206 | Args: |
| 207 | archive_source_file_mapping: A dict mapping source files to the associated |
| 208 | archive file that contains them. |
| 209 | Returns: |
| 210 | A list containing files with duplicated base names. |
| 211 | """ |
| 212 | duplicated_files = [] |
| 213 | dict_with_duplicates = dict() |
| 214 | |
| 215 | for source_files in archive_source_file_mapping.values(): |
| 216 | for source_file in source_files: |
| 217 | file_basename = os.path.basename(source_file) |
| 218 | file_without_ext = os.path.splitext(source_file)[0] |
| 219 | if file_basename in dict_with_duplicates: |
| 220 | dict_with_duplicates[file_basename].append(file_without_ext) |
| 221 | else: |
| 222 | dict_with_duplicates[file_basename] = [file_without_ext] |
| 223 | for basename in dict_with_duplicates: |
| 224 | if len(dict_with_duplicates[basename]) > 1: |
| 225 | duplicated_files.extend(dict_with_duplicates[basename]) |
| 226 | dict_with_duplicates = dict() |
| 227 | |
| 228 | return duplicated_files |
| 229 | |
| 230 | |
| 231 | def BuildArchiveSourceFileMapping(archive_source_mapping_files, file_open): |
| 232 | """Builds a mapping between archive files and their associated source files. |
| 233 | |
| 234 | Args: |
| 235 | archive_source_mapping_files: A comma separated list of J2ObjC-generated |
| 236 | mapping between archive files and their associated source files. |
| 237 | file_open: Reference to the builtin open function so it may be |
| 238 | overridden for testing. |
| 239 | Returns: |
| 240 | A dict mapping between archive files and their associated source files. |
| 241 | """ |
| 242 | return BuildArtifactSourceTree(archive_source_mapping_files, file_open) |
| 243 | |
| 244 | |
| 245 | def PruneSourceFiles(input_files, output_files, dependency_mapping_files, |
| 246 | header_mapping_files, entry_classes, objc_file_path, |
| 247 | file_open=open, file_shutil=shutil): |
| 248 | """Copies over translated files and remove the contents of unreachable files. |
| 249 | |
| 250 | Args: |
| 251 | input_files: A comma separated list of input source files to prune. It has |
| 252 | a one-on-one pair mapping with the output_file list. |
| 253 | output_files: A comma separated list of output source files to write pruned |
| 254 | source files to. It has a one-on-one pair mapping with the input_file |
| 255 | list. |
| 256 | dependency_mapping_files: A comma separated list of J2ObjC-generated |
| 257 | dependency mapping files. |
| 258 | header_mapping_files: A comma separated list of J2ObjC-generated |
| 259 | header mapping files. |
| 260 | entry_classes: A comma separated list of Java entry classes. |
| 261 | objc_file_path: The file path which represents a directory where the |
| 262 | generated ObjC files reside. |
| 263 | file_open: Reference to the builtin open function so it may be |
| 264 | overridden for testing. |
| 265 | file_shutil: Reference to the builtin shutil module so it may be |
| 266 | overridden for testing. |
| 267 | """ |
| 268 | reachability_file_mapping = BuildReachabilityTree( |
| 269 | dependency_mapping_files, file_open) |
| 270 | header_map = BuildHeaderMapping(header_mapping_files, file_open) |
| 271 | reachable_files_set = BuildReachableFileSet(entry_classes, |
| 272 | reachability_file_mapping, |
| 273 | header_map) |
| 274 | PruneFiles(input_files, |
| 275 | output_files, |
| 276 | objc_file_path, |
| 277 | reachable_files_set, |
| 278 | file_open, |
| 279 | file_shutil) |
| 280 | |
| 281 | |
| 282 | def MatchObjectNamesInArchive(xcrunwrapper, archive, object_names): |
| 283 | """Returns object names matching their identity in an archive file. |
| 284 | |
| 285 | The linker that blaze uses appends an md5 hash to object file |
| 286 | names prior to inclusion in the archive file. Thus, object names |
| 287 | such as 'foo.o' need to be matched to their appropriate name in |
| 288 | the archive file, such as 'foo_<hash>.o'. |
| 289 | |
| 290 | Args: |
| 291 | xcrunwrapper: A wrapper script over xcrun. |
| 292 | archive: The location of the archive file. |
| 293 | object_names: The expected basenames of object files to match, |
| 294 | sans extension. For example 'foo' (not 'foo.o'). |
| 295 | Returns: |
| 296 | A list of basenames of matching members of the given archive |
| 297 | """ |
| 298 | ar_contents_cmd = [xcrunwrapper, 'ar', '-t', archive] |
| 299 | real_object_names_output = subprocess.check_output(ar_contents_cmd) |
| 300 | real_object_names = real_object_names_output.decode('utf-8') |
| 301 | expected_object_name_regex = r'^(?:%s)(?:_[0-9a-f]{32}(?:-[0-9]+)?)?\.o$' % ( |
| 302 | '|'.join([re.escape(name) for name in object_names])) |
| 303 | return re.findall( |
| 304 | expected_object_name_regex, |
| 305 | real_object_names, |
| 306 | flags=re.MULTILINE) |
| 307 | |
| 308 | |
| 309 | def PruneArchiveFile(input_archive, output_archive, dummy_archive, |
| 310 | dependency_mapping_files, header_mapping_files, |
| 311 | archive_source_mapping_files, entry_classes, xcrunwrapper, |
| 312 | file_open=open): |
| 313 | """Remove unreachable objects from archive file. |
| 314 | |
| 315 | Args: |
| 316 | input_archive: The source archive file to prune. |
| 317 | output_archive: The location of the pruned archive file. |
| 318 | dummy_archive: A dummy archive file that contains no object. |
| 319 | dependency_mapping_files: A comma separated list of J2ObjC-generated |
| 320 | dependency mapping files. |
| 321 | header_mapping_files: A comma separated list of J2ObjC-generated |
| 322 | header mapping files. |
| 323 | archive_source_mapping_files: A comma separated list of J2ObjC-generated |
| 324 | mapping between archive files and their associated source files. |
| 325 | entry_classes: A comma separated list of Java entry classes. |
| 326 | xcrunwrapper: A wrapper script over xcrun. |
| 327 | file_open: Reference to the builtin open function so it may be |
| 328 | overridden for testing. |
| 329 | """ |
| 330 | reachability_file_mapping = BuildReachabilityTree( |
| 331 | dependency_mapping_files, file_open) |
| 332 | header_map = BuildHeaderMapping(header_mapping_files, file_open) |
| 333 | archive_source_file_mapping = BuildArchiveSourceFileMapping( |
| 334 | archive_source_mapping_files, file_open) |
| 335 | reachable_files_set = BuildReachableFileSet(entry_classes, |
| 336 | reachability_file_mapping, |
| 337 | header_map, |
| 338 | archive_source_file_mapping) |
| 339 | |
| 340 | # Copy the current processes' environment, as xcrunwrapper depends on these |
| 341 | # variables. |
| 342 | cmd_env = dict(os.environ) |
| 343 | j2objc_cmd = '' |
| 344 | if input_archive in archive_source_file_mapping: |
| 345 | source_files = archive_source_file_mapping[input_archive] |
| 346 | unreachable_object_names = [] |
| 347 | |
| 348 | for source_file in source_files: |
| 349 | if os.path.splitext(source_file)[0] not in reachable_files_set: |
| 350 | unreachable_object_names.append( |
| 351 | os.path.basename(os.path.splitext(source_file)[0])) |
| 352 | |
| 353 | # There are unreachable objects in the archive to prune |
| 354 | if unreachable_object_names: |
| 355 | # If all objects in the archive are unreachable, just copy over a dummy |
| 356 | # archive that contains no object |
| 357 | if len(unreachable_object_names) == len(source_files): |
| 358 | j2objc_cmd = 'cp %s %s' % (shlex.quote(dummy_archive), |
| 359 | shlex.quote(output_archive)) |
| 360 | # Else we need to prune the archive of unreachable objects |
| 361 | else: |
| 362 | cmd_env['ZERO_AR_DATE'] = '1' |
| 363 | # Copy the input archive to the output location |
| 364 | j2objc_cmd += 'cp %s %s && ' % (shlex.quote(input_archive), |
| 365 | shlex.quote(output_archive)) |
| 366 | # Make the output archive editable |
| 367 | j2objc_cmd += 'chmod +w %s && ' % (shlex.quote(output_archive)) |
| 368 | # Remove the unreachable objects from the archive |
| 369 | unreachable_object_names = MatchObjectNamesInArchive( |
| 370 | xcrunwrapper, input_archive, unreachable_object_names) |
| 371 | j2objc_cmd += '%s ar -d -s %s %s && ' % ( |
| 372 | shlex.quote(xcrunwrapper), |
| 373 | shlex.quote(output_archive), |
| 374 | ' '.join(shlex.quote(uon) for uon in unreachable_object_names)) |
| 375 | # Update the table of content of the archive file |
| 376 | j2objc_cmd += '%s ranlib %s' % (shlex.quote(xcrunwrapper), |
| 377 | shlex.quote(output_archive)) |
| 378 | # There are no unreachable objects, we just copy over the original archive |
| 379 | else: |
| 380 | j2objc_cmd = 'cp %s %s' % (shlex.quote(input_archive), |
| 381 | shlex.quote(output_archive)) |
| 382 | # The archive cannot be pruned by J2ObjC dead code removal, just copy over |
| 383 | # the original archive |
| 384 | else: |
| 385 | j2objc_cmd = 'cp %s %s' % (shlex.quote(input_archive), |
| 386 | shlex.quote(output_archive)) |
| 387 | |
| 388 | try: |
| 389 | subprocess.check_output( |
| 390 | j2objc_cmd, stderr=subprocess.STDOUT, shell=True, env=cmd_env) |
| 391 | except OSError as e: |
| 392 | raise Exception( |
| 393 | 'executing command failed: %s (%s)' % (j2objc_cmd, e.strerror)) |
| 394 | |
| 395 | # "Touch" the output file. |
| 396 | # Prevents a pre-Xcode-8 bug in which passing zero-date archive files to ld |
| 397 | # would cause ld to error. |
| 398 | os.utime(output_archive, None) |
| 399 | |
| 400 | |
| 401 | def BuildArtifactSourceTree(files, file_open=open): |
| 402 | """Builds a dependency tree using from dependency mapping files. |
| 403 | |
| 404 | Args: |
| 405 | files: A comma separated list of dependency mapping files. |
| 406 | file_open: Reference to the builtin open function so it may be overridden for |
| 407 | testing. |
| 408 | |
| 409 | Returns: |
| 410 | A dict mapping build artifacts (possibly generated source files) to the |
| 411 | corresponding direct dependent source files. |
| 412 | """ |
| 413 | tree = dict() |
| 414 | if not files: |
| 415 | return tree |
| 416 | for filename in files.split(','): |
| 417 | with file_open(filename, 'r') as f: |
| 418 | for line in f: |
| 419 | entry = line.strip().split(':')[0] |
| 420 | dep = line.strip().split(':')[1] |
| 421 | if entry in tree: |
| 422 | tree[entry].append(dep) |
| 423 | else: |
| 424 | tree[entry] = [dep] |
| 425 | return tree |
| 426 | |
| 427 | |
| 428 | if __name__ == '__main__': |
| 429 | parser = argparse.ArgumentParser(fromfile_prefix_chars='@') |
| 430 | |
| 431 | # TODO(rduan): Remove these three flags once J2ObjC compile actions are fully |
| 432 | # moved to the edges. |
| 433 | parser.add_argument( |
| 434 | '--input_files', |
| 435 | help=('The comma-separated file paths of translated source files to ' |
| 436 | 'prune.')) |
| 437 | parser.add_argument( |
| 438 | '--output_files', |
| 439 | help='The comma-separated file paths of pruned source files to write to.') |
| 440 | parser.add_argument( |
| 441 | '--objc_file_path', |
| 442 | help='The file path which represents a directory where the generated ObjC' |
| 443 | ' files reside') |
| 444 | |
| 445 | parser.add_argument( |
| 446 | '--input_archive', |
| 447 | help=('The path of the translated archive to prune.')) |
| 448 | parser.add_argument( |
| 449 | '--output_archive', |
| 450 | help='The path of the pruned archive file to write to.') |
| 451 | parser.add_argument( |
| 452 | '--dummy_archive', |
| 453 | help='The dummy archive file that contains no symbol.') |
| 454 | parser.add_argument( |
| 455 | '--dependency_mapping_files', |
| 456 | help='The comma-separated file paths of dependency mapping files.') |
| 457 | parser.add_argument( |
| 458 | '--header_mapping_files', |
| 459 | help='The comma-separated file paths of header mapping files.') |
| 460 | parser.add_argument( |
| 461 | '--archive_source_mapping_files', |
| 462 | help='The comma-separated file paths of archive to source mapping files.' |
| 463 | 'These mapping files should contain mappings between the ' |
| 464 | 'translated source files and the archive file compiled from those ' |
| 465 | 'source files.') |
| 466 | parser.add_argument( |
| 467 | '--entry_classes', |
| 468 | help=('The comma-separated list of Java entry classes to be used as entry' |
| 469 | ' point of the dead code analysis.')) |
| 470 | parser.add_argument( |
| 471 | '--xcrunwrapper', |
| 472 | help=('The xcrun wrapper script.')) |
| 473 | |
| 474 | args = parser.parse_args() |
| 475 | |
| 476 | if not args.entry_classes: |
| 477 | raise Exception('J2objC dead code removal is on but no entry class is ', |
| 478 | 'specified in any j2objc_library targets in the transitive', |
| 479 | ' closure') |
| 480 | if args.input_archive and args.output_archive: |
| 481 | PruneArchiveFile( |
| 482 | args.input_archive, |
| 483 | args.output_archive, |
| 484 | args.dummy_archive, |
| 485 | args.dependency_mapping_files, |
| 486 | args.header_mapping_files, |
| 487 | args.archive_source_mapping_files, |
| 488 | args.entry_classes, |
| 489 | args.xcrunwrapper) |
| 490 | else: |
| 491 | # TODO(rduan): Remove once J2ObjC compile actions are fully moved to the |
| 492 | # edges. |
| 493 | PruneSourceFiles( |
| 494 | args.input_files, |
| 495 | args.output_files, |
| 496 | args.dependency_mapping_files, |
| 497 | args.header_mapping_files, |
| 498 | args.entry_classes, |
| 499 | args.objc_file_path) |