bpo-34726: Fix handling of hash-based pycs in zipimport. (GH-10327) · python/cpython@a6e956b (original) (raw)

`@@ -578,33 +578,53 @@ def _eq_mtime(t1, t2):

578

`# dostime only stores even seconds, so be lenient

579

`return abs(t1 - t2) <= 1

580

581

+

581

582

`# Given the contents of a .py[co] file, unmarshal the data

582

583

`# and return the code object. Return None if it the magic word doesn't

583

match (we do this instead of raising an exception as we fall back

584

match, or if the recorded .py[co] metadata does not match the source,

585

(we do this instead of raising an exception as we fall back

584

586

`# to .py if available and we don't want to mask other errors).

585

def _unmarshal_code(pathname, data, mtime):

586

if len(data) < 16:

587

raise ZipImportError('bad pyc data')

588

-

589

if data[:4] != _bootstrap_external.MAGIC_NUMBER:

590

_bootstrap._verbose_message('{!r} has bad magic', pathname)

591

return None # signal caller to try alternative

592

-

593

flags = _unpack_uint32(data[4:8])

594

if flags != 0:

595

Hash-based pyc. We currently refuse to handle checked hash-based

596

pycs. We could validate hash-based pycs against the source, but it

597

seems likely that most people putting hash-based pycs in a zipfile

598

will use unchecked ones.

587

def _unmarshal_code(self, pathname, fullpath, fullname, data):

588

exc_details = {

589

'name': fullname,

590

'path': fullpath,

591

}

592

+

593

try:

594

flags = _bootstrap_external._classify_pyc(data, fullname, exc_details)

595

except ImportError:

596

return None

597

+

598

hash_based = flags & 0b1 != 0

599

if hash_based:

600

check_source = flags & 0b10 != 0

599

601

`if (_imp.check_hash_based_pycs != 'never' and

600

(flags != 0x1 or _imp.check_hash_based_pycs == 'always')):

601

return None

602

elif mtime != 0 and not _eq_mtime(_unpack_uint32(data[8:12]), mtime):

603

_bootstrap._verbose_message('{!r} has bad mtime', pathname)

604

return None # signal caller to try alternative

602

(check_source or _imp.check_hash_based_pycs == 'always')):

603

source_bytes = _get_pyc_source(self, fullpath)

604

if source_bytes is not None:

605

source_hash = _imp.source_hash(

606

_bootstrap_external._RAW_MAGIC_NUMBER,

607

source_bytes,

608

)

609

+

610

try:

611

_boostrap_external._validate_hash_pyc(

612

data, source_hash, fullname, exc_details)

613

except ImportError:

614

return None

615

else:

616

source_mtime, source_size = \

617

_get_mtime_and_size_of_source(self, fullpath)

618

+

619

if source_mtime:

620

We don't use _bootstrap_external._validate_timestamp_pyc

621

to allow for a more lenient timestamp check.

622

if (not _eq_mtime(_unpack_uint32(data[8:12]), source_mtime) or

623

_unpack_uint32(data[12:16]) != source_size):

624

_bootstrap._verbose_message(

625

f'bytecode is stale for {fullname!r}')

626

return None

605

627

606

XXX the pyc's size field is ignored; timestamp collisions are probably

607

unimportant with zip files.

608

628

`code = marshal.loads(data[16:])

609

629

`if not isinstance(code, _code_type):

610

630

`raise TypeError(f'compiled module {pathname!r} is not a code object')

`@@ -639,9 +659,9 @@ def _parse_dostime(d, t):

639

659

`-1, -1, -1))

640

660

641

661

`# Given a path to a .pyc file in the archive, return the

642

modification time of the matching .py file, or 0 if no source

643

is available.

644

def _get_mtime_of_source(self, path):

662

modification time of the matching .py file and its size,

663

or (0, 0) if no source is available.

664

def _get_mtime_and_size_of_source(self, path):

645

665

`try:

646

666

`# strip 'c' or 'o' from *.py[co]

647

667

`assert path[-1:] in ('c', 'o')

`@@ -651,9 +671,27 @@ def _get_mtime_of_source(self, path):

651

671

`# with an embedded pyc time stamp

652

672

`time = toc_entry[5]

653

673

`date = toc_entry[6]

654

return _parse_dostime(date, time)

674

uncompressed_size = toc_entry[3]

675

return _parse_dostime(date, time), uncompressed_size

655

676

`except (KeyError, IndexError, TypeError):

656

return 0

677

return 0, 0

678

+

679

+

680

Given a path to a .pyc file in the archive, return the

681

contents of the matching .py file, or None if no source

682

is available.

683

def _get_pyc_source(self, path):

684

strip 'c' or 'o' from *.py[co]

685

assert path[-1:] in ('c', 'o')

686

path = path[:-1]

687

+

688

try:

689

toc_entry = self._files[path]

690

except KeyError:

691

return None

692

else:

693

return _get_data(self.archive, toc_entry)

694

+

657

695

658

696

`# Get the code object associated with the module specified by

659

697

`# 'fullname'.

`@@ -670,8 +708,7 @@ def _get_module_code(self, fullname):

670

708

`modpath = toc_entry[0]

671

709

`data = _get_data(self.archive, toc_entry)

672

710

`if isbytecode:

673

mtime = _get_mtime_of_source(self, fullpath)

674

code = _unmarshal_code(modpath, data, mtime)

711

code = _unmarshal_code(self, modpath, fullpath, fullname, data)

675

712

`else:

676

713

`code = _compile_source(modpath, data)

677

714

`if code is None: