Pirama Arumuga Nainar | d285ad0 | 2022-02-08 09:26:56 -0800 | [diff] [blame] | 1 | """Module/script to byte-compile all .py files to .pyc files. |
| 2 | |
| 3 | When called as a script with arguments, this compiles the directories |
| 4 | given as arguments recursively; the -l option prevents it from |
| 5 | recursing into directories. |
| 6 | |
| 7 | Without arguments, if compiles all modules on sys.path, without |
| 8 | recursing into subdirectories. (Even though it should do so for |
| 9 | packages -- for now, you'll have to deal with packages separately.) |
| 10 | |
| 11 | See module py_compile for details of the actual byte-compilation. |
| 12 | """ |
| 13 | import os |
| 14 | import sys |
| 15 | import importlib.util |
| 16 | import py_compile |
| 17 | import struct |
| 18 | import filecmp |
| 19 | |
| 20 | from functools import partial |
| 21 | from pathlib import Path |
| 22 | |
| 23 | __all__ = ["compile_dir","compile_file","compile_path"] |
| 24 | |
| 25 | def _walk_dir(dir, maxlevels, quiet=0): |
| 26 | if quiet < 2 and isinstance(dir, os.PathLike): |
| 27 | dir = os.fspath(dir) |
| 28 | if not quiet: |
| 29 | print('Listing {!r}...'.format(dir)) |
| 30 | try: |
| 31 | names = os.listdir(dir) |
| 32 | except OSError: |
| 33 | if quiet < 2: |
| 34 | print("Can't list {!r}".format(dir)) |
| 35 | names = [] |
| 36 | names.sort() |
| 37 | for name in names: |
| 38 | if name == '__pycache__': |
| 39 | continue |
| 40 | fullname = os.path.join(dir, name) |
| 41 | if not os.path.isdir(fullname): |
| 42 | yield fullname |
| 43 | elif (maxlevels > 0 and name != os.curdir and name != os.pardir and |
| 44 | os.path.isdir(fullname) and not os.path.islink(fullname)): |
| 45 | yield from _walk_dir(fullname, maxlevels=maxlevels - 1, |
| 46 | quiet=quiet) |
| 47 | |
| 48 | def compile_dir(dir, maxlevels=None, ddir=None, force=False, |
| 49 | rx=None, quiet=0, legacy=False, optimize=-1, workers=1, |
| 50 | invalidation_mode=None, *, stripdir=None, |
| 51 | prependdir=None, limit_sl_dest=None, hardlink_dupes=False): |
| 52 | """Byte-compile all modules in the given directory tree. |
| 53 | |
| 54 | Arguments (only dir is required): |
| 55 | |
| 56 | dir: the directory to byte-compile |
| 57 | maxlevels: maximum recursion level (default `sys.getrecursionlimit()`) |
| 58 | ddir: the directory that will be prepended to the path to the |
| 59 | file as it is compiled into each byte-code file. |
| 60 | force: if True, force compilation, even if timestamps are up-to-date |
| 61 | quiet: full output with False or 0, errors only with 1, |
| 62 | no output with 2 |
| 63 | legacy: if True, produce legacy pyc paths instead of PEP 3147 paths |
| 64 | optimize: int or list of optimization levels or -1 for level of |
| 65 | the interpreter. Multiple levels leads to multiple compiled |
| 66 | files each with one optimization level. |
| 67 | workers: maximum number of parallel workers |
| 68 | invalidation_mode: how the up-to-dateness of the pyc will be checked |
| 69 | stripdir: part of path to left-strip from source file path |
| 70 | prependdir: path to prepend to beginning of original file path, applied |
| 71 | after stripdir |
| 72 | limit_sl_dest: ignore symlinks if they are pointing outside of |
| 73 | the defined path |
| 74 | hardlink_dupes: hardlink duplicated pyc files |
| 75 | """ |
| 76 | ProcessPoolExecutor = None |
| 77 | if ddir is not None and (stripdir is not None or prependdir is not None): |
| 78 | raise ValueError(("Destination dir (ddir) cannot be used " |
| 79 | "in combination with stripdir or prependdir")) |
| 80 | if ddir is not None: |
| 81 | stripdir = dir |
| 82 | prependdir = ddir |
| 83 | ddir = None |
| 84 | if workers < 0: |
| 85 | raise ValueError('workers must be greater or equal to 0') |
| 86 | if workers != 1: |
| 87 | try: |
| 88 | # Only import when needed, as low resource platforms may |
| 89 | # fail to import it |
| 90 | from concurrent.futures import ProcessPoolExecutor |
| 91 | except ImportError: |
| 92 | workers = 1 |
| 93 | if maxlevels is None: |
| 94 | maxlevels = sys.getrecursionlimit() |
| 95 | files = _walk_dir(dir, quiet=quiet, maxlevels=maxlevels) |
| 96 | success = True |
| 97 | if workers != 1 and ProcessPoolExecutor is not None: |
| 98 | # If workers == 0, let ProcessPoolExecutor choose |
| 99 | workers = workers or None |
| 100 | with ProcessPoolExecutor(max_workers=workers) as executor: |
| 101 | results = executor.map(partial(compile_file, |
| 102 | ddir=ddir, force=force, |
| 103 | rx=rx, quiet=quiet, |
| 104 | legacy=legacy, |
| 105 | optimize=optimize, |
| 106 | invalidation_mode=invalidation_mode, |
| 107 | stripdir=stripdir, |
| 108 | prependdir=prependdir, |
| 109 | limit_sl_dest=limit_sl_dest, |
| 110 | hardlink_dupes=hardlink_dupes), |
| 111 | files) |
| 112 | success = min(results, default=True) |
| 113 | else: |
| 114 | for file in files: |
| 115 | if not compile_file(file, ddir, force, rx, quiet, |
| 116 | legacy, optimize, invalidation_mode, |
| 117 | stripdir=stripdir, prependdir=prependdir, |
| 118 | limit_sl_dest=limit_sl_dest, |
| 119 | hardlink_dupes=hardlink_dupes): |
| 120 | success = False |
| 121 | return success |
| 122 | |
| 123 | def compile_file(fullname, ddir=None, force=False, rx=None, quiet=0, |
| 124 | legacy=False, optimize=-1, |
| 125 | invalidation_mode=None, *, stripdir=None, prependdir=None, |
| 126 | limit_sl_dest=None, hardlink_dupes=False): |
| 127 | """Byte-compile one file. |
| 128 | |
| 129 | Arguments (only fullname is required): |
| 130 | |
| 131 | fullname: the file to byte-compile |
| 132 | ddir: if given, the directory name compiled in to the |
| 133 | byte-code file. |
| 134 | force: if True, force compilation, even if timestamps are up-to-date |
| 135 | quiet: full output with False or 0, errors only with 1, |
| 136 | no output with 2 |
| 137 | legacy: if True, produce legacy pyc paths instead of PEP 3147 paths |
| 138 | optimize: int or list of optimization levels or -1 for level of |
| 139 | the interpreter. Multiple levels leads to multiple compiled |
| 140 | files each with one optimization level. |
| 141 | invalidation_mode: how the up-to-dateness of the pyc will be checked |
| 142 | stripdir: part of path to left-strip from source file path |
| 143 | prependdir: path to prepend to beginning of original file path, applied |
| 144 | after stripdir |
| 145 | limit_sl_dest: ignore symlinks if they are pointing outside of |
| 146 | the defined path. |
| 147 | hardlink_dupes: hardlink duplicated pyc files |
| 148 | """ |
| 149 | |
| 150 | if ddir is not None and (stripdir is not None or prependdir is not None): |
| 151 | raise ValueError(("Destination dir (ddir) cannot be used " |
| 152 | "in combination with stripdir or prependdir")) |
| 153 | |
| 154 | success = True |
| 155 | if quiet < 2 and isinstance(fullname, os.PathLike): |
| 156 | fullname = os.fspath(fullname) |
| 157 | name = os.path.basename(fullname) |
| 158 | |
| 159 | dfile = None |
| 160 | |
| 161 | if ddir is not None: |
| 162 | dfile = os.path.join(ddir, name) |
| 163 | |
| 164 | if stripdir is not None: |
| 165 | fullname_parts = fullname.split(os.path.sep) |
| 166 | stripdir_parts = stripdir.split(os.path.sep) |
| 167 | ddir_parts = list(fullname_parts) |
| 168 | |
| 169 | for spart, opart in zip(stripdir_parts, fullname_parts): |
| 170 | if spart == opart: |
| 171 | ddir_parts.remove(spart) |
| 172 | |
| 173 | dfile = os.path.join(*ddir_parts) |
| 174 | |
| 175 | if prependdir is not None: |
| 176 | if dfile is None: |
| 177 | dfile = os.path.join(prependdir, fullname) |
| 178 | else: |
| 179 | dfile = os.path.join(prependdir, dfile) |
| 180 | |
| 181 | if isinstance(optimize, int): |
| 182 | optimize = [optimize] |
| 183 | |
| 184 | # Use set() to remove duplicates. |
| 185 | # Use sorted() to create pyc files in a deterministic order. |
| 186 | optimize = sorted(set(optimize)) |
| 187 | |
| 188 | if hardlink_dupes and len(optimize) < 2: |
| 189 | raise ValueError("Hardlinking of duplicated bytecode makes sense " |
| 190 | "only for more than one optimization level") |
| 191 | |
| 192 | if rx is not None: |
| 193 | mo = rx.search(fullname) |
| 194 | if mo: |
| 195 | return success |
| 196 | |
| 197 | if limit_sl_dest is not None and os.path.islink(fullname): |
| 198 | if Path(limit_sl_dest).resolve() not in Path(fullname).resolve().parents: |
| 199 | return success |
| 200 | |
| 201 | opt_cfiles = {} |
| 202 | |
| 203 | if os.path.isfile(fullname): |
| 204 | for opt_level in optimize: |
| 205 | if legacy: |
| 206 | opt_cfiles[opt_level] = fullname + 'c' |
| 207 | else: |
| 208 | if opt_level >= 0: |
| 209 | opt = opt_level if opt_level >= 1 else '' |
| 210 | cfile = (importlib.util.cache_from_source( |
| 211 | fullname, optimization=opt)) |
| 212 | opt_cfiles[opt_level] = cfile |
| 213 | else: |
| 214 | cfile = importlib.util.cache_from_source(fullname) |
| 215 | opt_cfiles[opt_level] = cfile |
| 216 | |
| 217 | head, tail = name[:-3], name[-3:] |
| 218 | if tail == '.py': |
| 219 | if not force: |
| 220 | try: |
| 221 | mtime = int(os.stat(fullname).st_mtime) |
| 222 | expect = struct.pack('<4sll', importlib.util.MAGIC_NUMBER, |
| 223 | 0, mtime) |
| 224 | for cfile in opt_cfiles.values(): |
| 225 | with open(cfile, 'rb') as chandle: |
| 226 | actual = chandle.read(12) |
| 227 | if expect != actual: |
| 228 | break |
| 229 | else: |
| 230 | return success |
| 231 | except OSError: |
| 232 | pass |
| 233 | if not quiet: |
| 234 | print('Compiling {!r}...'.format(fullname)) |
| 235 | try: |
| 236 | for index, opt_level in enumerate(optimize): |
| 237 | cfile = opt_cfiles[opt_level] |
| 238 | ok = py_compile.compile(fullname, cfile, dfile, True, |
| 239 | optimize=opt_level, |
| 240 | invalidation_mode=invalidation_mode) |
| 241 | if index > 0 and hardlink_dupes: |
| 242 | previous_cfile = opt_cfiles[optimize[index - 1]] |
| 243 | if filecmp.cmp(cfile, previous_cfile, shallow=False): |
| 244 | os.unlink(cfile) |
| 245 | os.link(previous_cfile, cfile) |
| 246 | except py_compile.PyCompileError as err: |
| 247 | success = False |
| 248 | if quiet >= 2: |
| 249 | return success |
| 250 | elif quiet: |
| 251 | print('*** Error compiling {!r}...'.format(fullname)) |
| 252 | else: |
| 253 | print('*** ', end='') |
| 254 | # escape non-printable characters in msg |
| 255 | msg = err.msg.encode(sys.stdout.encoding, |
| 256 | errors='backslashreplace') |
| 257 | msg = msg.decode(sys.stdout.encoding) |
| 258 | print(msg) |
| 259 | except (SyntaxError, UnicodeError, OSError) as e: |
| 260 | success = False |
| 261 | if quiet >= 2: |
| 262 | return success |
| 263 | elif quiet: |
| 264 | print('*** Error compiling {!r}...'.format(fullname)) |
| 265 | else: |
| 266 | print('*** ', end='') |
| 267 | print(e.__class__.__name__ + ':', e) |
| 268 | else: |
| 269 | if ok == 0: |
| 270 | success = False |
| 271 | return success |
| 272 | |
| 273 | def compile_path(skip_curdir=1, maxlevels=0, force=False, quiet=0, |
| 274 | legacy=False, optimize=-1, |
| 275 | invalidation_mode=None): |
| 276 | """Byte-compile all module on sys.path. |
| 277 | |
| 278 | Arguments (all optional): |
| 279 | |
| 280 | skip_curdir: if true, skip current directory (default True) |
| 281 | maxlevels: max recursion level (default 0) |
| 282 | force: as for compile_dir() (default False) |
| 283 | quiet: as for compile_dir() (default 0) |
| 284 | legacy: as for compile_dir() (default False) |
| 285 | optimize: as for compile_dir() (default -1) |
| 286 | invalidation_mode: as for compiler_dir() |
| 287 | """ |
| 288 | success = True |
| 289 | for dir in sys.path: |
| 290 | if (not dir or dir == os.curdir) and skip_curdir: |
| 291 | if quiet < 2: |
| 292 | print('Skipping current directory') |
| 293 | else: |
| 294 | success = success and compile_dir( |
| 295 | dir, |
| 296 | maxlevels, |
| 297 | None, |
| 298 | force, |
| 299 | quiet=quiet, |
| 300 | legacy=legacy, |
| 301 | optimize=optimize, |
| 302 | invalidation_mode=invalidation_mode, |
| 303 | ) |
| 304 | return success |
| 305 | |
| 306 | |
| 307 | def main(): |
| 308 | """Script main program.""" |
| 309 | import argparse |
| 310 | |
| 311 | parser = argparse.ArgumentParser( |
| 312 | description='Utilities to support installing Python libraries.') |
| 313 | parser.add_argument('-l', action='store_const', const=0, |
| 314 | default=None, dest='maxlevels', |
| 315 | help="don't recurse into subdirectories") |
| 316 | parser.add_argument('-r', type=int, dest='recursion', |
| 317 | help=('control the maximum recursion level. ' |
| 318 | 'if `-l` and `-r` options are specified, ' |
| 319 | 'then `-r` takes precedence.')) |
| 320 | parser.add_argument('-f', action='store_true', dest='force', |
| 321 | help='force rebuild even if timestamps are up to date') |
| 322 | parser.add_argument('-q', action='count', dest='quiet', default=0, |
| 323 | help='output only error messages; -qq will suppress ' |
| 324 | 'the error messages as well.') |
| 325 | parser.add_argument('-b', action='store_true', dest='legacy', |
| 326 | help='use legacy (pre-PEP3147) compiled file locations') |
| 327 | parser.add_argument('-d', metavar='DESTDIR', dest='ddir', default=None, |
| 328 | help=('directory to prepend to file paths for use in ' |
| 329 | 'compile-time tracebacks and in runtime ' |
| 330 | 'tracebacks in cases where the source file is ' |
| 331 | 'unavailable')) |
| 332 | parser.add_argument('-s', metavar='STRIPDIR', dest='stripdir', |
| 333 | default=None, |
| 334 | help=('part of path to left-strip from path ' |
| 335 | 'to source file - for example buildroot. ' |
| 336 | '`-d` and `-s` options cannot be ' |
| 337 | 'specified together.')) |
| 338 | parser.add_argument('-p', metavar='PREPENDDIR', dest='prependdir', |
| 339 | default=None, |
| 340 | help=('path to add as prefix to path ' |
| 341 | 'to source file - for example / to make ' |
| 342 | 'it absolute when some part is removed ' |
| 343 | 'by `-s` option. ' |
| 344 | '`-d` and `-p` options cannot be ' |
| 345 | 'specified together.')) |
| 346 | parser.add_argument('-x', metavar='REGEXP', dest='rx', default=None, |
| 347 | help=('skip files matching the regular expression; ' |
| 348 | 'the regexp is searched for in the full path ' |
| 349 | 'of each file considered for compilation')) |
| 350 | parser.add_argument('-i', metavar='FILE', dest='flist', |
| 351 | help=('add all the files and directories listed in ' |
| 352 | 'FILE to the list considered for compilation; ' |
| 353 | 'if "-", names are read from stdin')) |
| 354 | parser.add_argument('compile_dest', metavar='FILE|DIR', nargs='*', |
| 355 | help=('zero or more file and directory names ' |
| 356 | 'to compile; if no arguments given, defaults ' |
| 357 | 'to the equivalent of -l sys.path')) |
| 358 | parser.add_argument('-j', '--workers', default=1, |
| 359 | type=int, help='Run compileall concurrently') |
| 360 | invalidation_modes = [mode.name.lower().replace('_', '-') |
| 361 | for mode in py_compile.PycInvalidationMode] |
| 362 | parser.add_argument('--invalidation-mode', |
| 363 | choices=sorted(invalidation_modes), |
| 364 | help=('set .pyc invalidation mode; defaults to ' |
| 365 | '"checked-hash" if the SOURCE_DATE_EPOCH ' |
| 366 | 'environment variable is set, and ' |
| 367 | '"timestamp" otherwise.')) |
| 368 | parser.add_argument('-o', action='append', type=int, dest='opt_levels', |
| 369 | help=('Optimization levels to run compilation with.' |
| 370 | 'Default is -1 which uses optimization level of' |
| 371 | 'Python interpreter itself (specified by -O).')) |
| 372 | parser.add_argument('-e', metavar='DIR', dest='limit_sl_dest', |
| 373 | help='Ignore symlinks pointing outsite of the DIR') |
| 374 | parser.add_argument('--hardlink-dupes', action='store_true', |
| 375 | dest='hardlink_dupes', |
| 376 | help='Hardlink duplicated pyc files') |
| 377 | |
| 378 | args = parser.parse_args() |
| 379 | compile_dests = args.compile_dest |
| 380 | |
| 381 | if args.rx: |
| 382 | import re |
| 383 | args.rx = re.compile(args.rx) |
| 384 | |
| 385 | if args.limit_sl_dest == "": |
| 386 | args.limit_sl_dest = None |
| 387 | |
| 388 | if args.recursion is not None: |
| 389 | maxlevels = args.recursion |
| 390 | else: |
| 391 | maxlevels = args.maxlevels |
| 392 | |
| 393 | if args.opt_levels is None: |
| 394 | args.opt_levels = [-1] |
| 395 | |
| 396 | if len(args.opt_levels) == 1 and args.hardlink_dupes: |
| 397 | parser.error(("Hardlinking of duplicated bytecode makes sense " |
| 398 | "only for more than one optimization level.")) |
| 399 | |
| 400 | if args.ddir is not None and ( |
| 401 | args.stripdir is not None or args.prependdir is not None |
| 402 | ): |
| 403 | parser.error("-d cannot be used in combination with -s or -p") |
| 404 | |
| 405 | # if flist is provided then load it |
| 406 | if args.flist: |
| 407 | try: |
| 408 | with (sys.stdin if args.flist=='-' else open(args.flist)) as f: |
| 409 | for line in f: |
| 410 | compile_dests.append(line.strip()) |
| 411 | except OSError: |
| 412 | if args.quiet < 2: |
| 413 | print("Error reading file list {}".format(args.flist)) |
| 414 | return False |
| 415 | |
| 416 | if args.invalidation_mode: |
| 417 | ivl_mode = args.invalidation_mode.replace('-', '_').upper() |
| 418 | invalidation_mode = py_compile.PycInvalidationMode[ivl_mode] |
| 419 | else: |
| 420 | invalidation_mode = None |
| 421 | |
| 422 | success = True |
| 423 | try: |
| 424 | if compile_dests: |
| 425 | for dest in compile_dests: |
| 426 | if os.path.isfile(dest): |
| 427 | if not compile_file(dest, args.ddir, args.force, args.rx, |
| 428 | args.quiet, args.legacy, |
| 429 | invalidation_mode=invalidation_mode, |
| 430 | stripdir=args.stripdir, |
| 431 | prependdir=args.prependdir, |
| 432 | optimize=args.opt_levels, |
| 433 | limit_sl_dest=args.limit_sl_dest, |
| 434 | hardlink_dupes=args.hardlink_dupes): |
| 435 | success = False |
| 436 | else: |
| 437 | if not compile_dir(dest, maxlevels, args.ddir, |
| 438 | args.force, args.rx, args.quiet, |
| 439 | args.legacy, workers=args.workers, |
| 440 | invalidation_mode=invalidation_mode, |
| 441 | stripdir=args.stripdir, |
| 442 | prependdir=args.prependdir, |
| 443 | optimize=args.opt_levels, |
| 444 | limit_sl_dest=args.limit_sl_dest, |
| 445 | hardlink_dupes=args.hardlink_dupes): |
| 446 | success = False |
| 447 | return success |
| 448 | else: |
| 449 | return compile_path(legacy=args.legacy, force=args.force, |
| 450 | quiet=args.quiet, |
| 451 | invalidation_mode=invalidation_mode) |
| 452 | except KeyboardInterrupt: |
| 453 | if args.quiet < 2: |
| 454 | print("\n[interrupted]") |
| 455 | return False |
| 456 | return True |
| 457 | |
| 458 | |
| 459 | if __name__ == '__main__': |
| 460 | exit_status = int(not main()) |
| 461 | sys.exit(exit_status) |