view tests/test-doctest.py @ 46607:e9901d01d135

revlog: add a mechanism to verify expected file position before appending

If someone uses `hg debuglocks`, or some non-hg process writes to the .hg
directory without respecting the locks, or if the repo's on a networked
filesystem, it's possible for the revlog code to write out corrupted data.

The form of this corruption can vary depending on what data was written and how
that happened. We are in the "networked filesystem" case (though I've had users
also do this to themselves with the "`hg debuglocks`" scenario), and most often
see this with the changelog. What ends up happening is we produce two items
(let's call them rev1 and rev2) in the .i file that have the same linkrev,
baserev, and offset into the .d file, while the data in the .d file is appended
properly. rev2's compressed_size is accurate for rev2, but when we go to
decompress the data in the .d file, we use the offset that's recorded in the
index file, which is the same as rev1, and attempt to decompress
rev2.compressed_size bytes of rev1's data. This usually does not succeed. :)

When using inline data, this also fails, though I haven't investigated why too
closely. This shows up as a "patch decode" error. I believe what's happening
there is that we're basically ignoring the offset field, getting the data
properly, but since baserev != rev, it thinks this is a delta based on rev
(instead of a full text) and can't actually apply it as such.

For now, I'm going to make this an optional component and default it to entirely
off. I may increase the default severity of this in the future, once I've
enabled it for my users and we gain more experience with it. Luckily, most of my
users have a versioned filesystem and can roll back to before the corruption has
been written, it's just a hassle to do so and not everyone knows how (so it's a
support burden). Users on other filesystems will not have that luxury, and this
can cause them to have a corrupted repository that they are unlikely to know how
to resolve, and they'll see this as a data-loss event. Refusing to create the
corruption is a much better user experience.

This mechanism is not perfect. There may be false-negatives (racy writes that
are not detected). There should not be any false-positives (non-racy writes that
are detected as such). This is not a mechanism that makes putting a repo on a
networked filesystem "safe" or "supported", just *less* likely to cause
corruption.

Differential Revision: https://phab.mercurial-scm.org/D9952

# this is hack to make sure no escape characters are inserted into the output

from __future__ import absolute_import
from __future__ import print_function

import doctest
import os
import re
import subprocess
import sys

ispy3 = sys.version_info[0] >= 3

if 'TERM' in os.environ:
    del os.environ['TERM']


class py3docchecker(doctest.OutputChecker):
    def check_output(self, want, got, optionflags):
        want2 = re.sub(r'''\bu(['"])(.*?)\1''', r'\1\2\1', want)  # py2: u''
        got2 = re.sub(r'''\bb(['"])(.*?)\1''', r'\1\2\1', got)  # py3: b''
        # py3: <exc.name>: b'<msg>' -> <name>: <msg>
        #      <exc.name>: <others> -> <name>: <others>
        got2 = re.sub(
            r'''^mercurial\.\w+\.(\w+): (['"])(.*?)\2''',
            r'\1: \3',
            got2,
            re.MULTILINE,
        )
        got2 = re.sub(r'^mercurial\.\w+\.(\w+): ', r'\1: ', got2, re.MULTILINE)
        return any(
            doctest.OutputChecker.check_output(self, w, g, optionflags)
            for w, g in [(want, got), (want2, got2)]
        )


def testmod(name, optionflags=0, testtarget=None):
    __import__(name)
    mod = sys.modules[name]
    if testtarget is not None:
        mod = getattr(mod, testtarget)

    # minimal copy of doctest.testmod()
    finder = doctest.DocTestFinder()
    checker = None
    if ispy3:
        checker = py3docchecker()
    runner = doctest.DocTestRunner(checker=checker, optionflags=optionflags)
    for test in finder.find(mod, name):
        runner.run(test)
    runner.summarize()


DONT_RUN = []

# Exceptions to the defaults for a given detected module. The value for each
# module name is a list of dicts that specify the kwargs to pass to testmod.
# testmod is called once per item in the list, so an empty list will cause the
# module to not be tested.
testmod_arg_overrides = {
    'i18n.check-translation': DONT_RUN,  # may require extra installation
    'mercurial.dagparser': [{'optionflags': doctest.NORMALIZE_WHITESPACE}],
    'mercurial.keepalive': DONT_RUN,  # >>> is an example, not a doctest
    'mercurial.posix': DONT_RUN,  # run by mercurial.platform
    'mercurial.statprof': DONT_RUN,  # >>> is an example, not a doctest
    'mercurial.util': [{}, {'testtarget': 'platform'}],  # run twice!
    'mercurial.windows': DONT_RUN,  # run by mercurial.platform
    'tests.test-url': [{'optionflags': doctest.NORMALIZE_WHITESPACE}],
}

fileset = 'set:(**.py)'

cwd = os.path.dirname(os.environ["TESTDIR"])

if not os.path.isdir(os.path.join(cwd, ".hg")):
    sys.exit(0)

files = subprocess.check_output(
    "hg files --print0 \"%s\"" % fileset,
    shell=True,
    cwd=cwd,
).split(b'\0')

if sys.version_info[0] >= 3:
    cwd = os.fsencode(cwd)

mods_tested = set()
for f in files:
    if not f:
        continue

    with open(os.path.join(cwd, f), "rb") as fh:
        if not re.search(br'\n\s*>>>', fh.read()):
            continue

    if ispy3:
        f = f.decode()

    modname = f.replace('.py', '').replace('\\', '.').replace('/', '.')

    # Third-party modules aren't our responsibility to test, and the modules in
    # contrib generally do not have doctests in a good state, plus they're hard
    # to import if this test is running with py2, so we just skip both for now.
    if modname.startswith('mercurial.thirdparty.') or modname.startswith(
        'contrib.'
    ):
        continue

    for kwargs in testmod_arg_overrides.get(modname, [{}]):
        mods_tested.add((modname, '%r' % (kwargs,)))
        if modname.startswith('tests.'):
            # On py2, we can't import from tests.foo, but it works on both py2
            # and py3 with the way that PYTHONPATH is setup to import without
            # the 'tests.' prefix, so we do that.
            modname = modname[len('tests.') :]

        testmod(modname, **kwargs)

# Meta-test: let's make sure that we actually ran what we expected to, above.
# Each item in the set is a 2-tuple of module name and stringified kwargs passed
# to testmod.
expected_mods_tested = set(
    [
        ('hgext.convert.convcmd', '{}'),
        ('hgext.convert.cvsps', '{}'),
        ('hgext.convert.filemap', '{}'),
        ('hgext.convert.p4', '{}'),
        ('hgext.convert.subversion', '{}'),
        ('hgext.fix', '{}'),
        ('hgext.mq', '{}'),
        ('mercurial.changelog', '{}'),
        ('mercurial.cmdutil', '{}'),
        ('mercurial.color', '{}'),
        ('mercurial.config', '{}'),
        ('mercurial.dagparser', "{'optionflags': 4}"),
        ('mercurial.encoding', '{}'),
        ('mercurial.fancyopts', '{}'),
        ('mercurial.formatter', '{}'),
        ('mercurial.hg', '{}'),
        ('mercurial.hgweb.hgwebdir_mod', '{}'),
        ('mercurial.match', '{}'),
        ('mercurial.mdiff', '{}'),
        ('mercurial.minirst', '{}'),
        ('mercurial.parser', '{}'),
        ('mercurial.patch', '{}'),
        ('mercurial.pathutil', '{}'),
        ('mercurial.pycompat', '{}'),
        ('mercurial.revlogutils.deltas', '{}'),
        ('mercurial.revset', '{}'),
        ('mercurial.revsetlang', '{}'),
        ('mercurial.simplemerge', '{}'),
        ('mercurial.smartset', '{}'),
        ('mercurial.store', '{}'),
        ('mercurial.subrepo', '{}'),
        ('mercurial.templater', '{}'),
        ('mercurial.ui', '{}'),
        ('mercurial.util', "{'testtarget': 'platform'}"),
        ('mercurial.util', '{}'),
        ('mercurial.utils.dateutil', '{}'),
        ('mercurial.utils.stringutil', '{}'),
        ('tests.drawdag', '{}'),
        ('tests.test-run-tests', '{}'),
        ('tests.test-url', "{'optionflags': 4}"),
    ]
)

unexpectedly_run = mods_tested.difference(expected_mods_tested)
not_run = expected_mods_tested.difference(mods_tested)

if unexpectedly_run:
    print('Unexpectedly ran (probably need to add to list):')
    for r in sorted(unexpectedly_run):
        print('  %r' % (r,))
if not_run:
    print('Expected to run, but was not run (doctest removed?):')
    for r in sorted(not_run):
        print('  %r' % (r,))