view tests/test-url.py @ 40326:fed697fa1734

sqlitestore: file storage backend using SQLite

This commit provides an extension which uses SQLite to store file
data (as opposed to revlogs).

As the inline documentation describes, there are still several
aspects to the extension that are incomplete. But it's a start.
The extension does support basic clone, checkout, and commit
workflows, which makes it suitable for simple use cases.

One notable missing feature is support for "bundlerepos." This is
probably responsible for the most test failures when the extension
is activated as part of the test suite.

All revision data is stored in SQLite. Data is stored as zstd
compressed chunks (default if zstd is available), zlib compressed
chunks (default if zstd is not available), or raw chunks (if
configured or if a compressed delta is not smaller than the raw
delta). This makes things very similar to revlogs.

Unlike revlogs, the extension doesn't yet enforce a limit on delta
chain length. This is an obvious limitation and should be addressed.
This is somewhat mitigated by the use of zstd, which is much faster
than zlib to decompress.

There is a dedicated table for storing deltas. Deltas are stored
by the SHA-1 hash of their uncompressed content. The "fileindex" table
has columns that reference the delta for each revision and the base
delta that delta should be applied against. A recursive SQL query
is used to resolve the delta chain along with the delta data.

By storing deltas by hash, we are able to de-duplicate delta storage!
With revlogs, the same deltas in different revlogs would result in
duplicate storage of that delta. In this scheme, inserting the
duplicate delta is a no-op and delta chains simply reference the
existing delta.

When initially implementing this extension, I did not have
content-indexed deltas and deltas could be duplicated across files
(just like revlogs). When I implemented content-indexed deltas, the
size of the SQLite database for a full clone of mozilla-unified
dropped:

before: 2,554,261,504 bytes
after:  2,488,754,176 bytes

Surprisingly, this is still larger than the bytes size of revlog
files:

revlog files: 2,104,861,230 bytes
du -b:        2,254,381,614

I would have expected storage to be smaller since we're not limiting
delta chain length and since we're using zstd instead of zlib. I
suspect the SQLite indexes and per-column overhead account for the
bulk of the differences. (Keep in mind that revlog uses a 64-byte
packed struct for revision index data and deltas are stored without
padding. Aside from the 12 unused bytes in the 32 byte node field,
revlogs are pretty efficient.) Another source of overhead is file
name storage. With revlogs, file names are stored in the filesystem.
But with SQLite, we need to store file names in the database. This is
roughly equivalent to the size of the fncache file, which for the
mozilla-unified repository is ~34MB.

Since the SQLite database isn't append-only and since delta chains
can reference any delta, this opens some interesting possibilities.
For example, we could store deltas in reverse, such that fulltexts
are stored for newer revisions and deltas are applied to reconstruct
older revisions. This is likely a more optimal storage strategy for
version control, as new data tends to be more frequently accessed
than old data. We would obviously need wire protocol support for
transferring revision data from newest to oldest. And we would
probably need some kind of mechanism for "re-encoding" stores. But
it should be doable.

This extension is very much experimental quality. There are a handful
of features that don't work. It probably isn't suitable for day-to-day
use. But it could be used in limited cases (e.g. read-only checkouts
like in CI). And it is also a good proving ground for alternate
storage backends. As we continue to define interfaces for all things
storage, it will be useful to have a viable alternate storage backend
to see how things shake out in practice.

test-storage.py passes on Python 2 and introduces no new test failures on
Python 3. Having the storage-level unit tests has proved to be insanely
useful when developing this extension. Those tests caught numerous bugs
during development and I'm convinced this style of testing is the way
forward for ensuring alternate storage backends work as intended. Of
course, test coverage isn't close to what it needs to be. But it is
a start. And what coverage we have gives me confidence that basic store
functionality is implemented properly.

Differential Revision: https://phab.mercurial-scm.org/D4928

# coding=utf-8
from __future__ import absolute_import, print_function

import doctest
import os

def check(a, b):
    if a != b:
        print((a, b))

def cert(cn):
    return {'subject': ((('commonName', cn),),)}

from mercurial import (
    sslutil,
)

_verifycert = sslutil._verifycert
# Test non-wildcard certificates
check(_verifycert(cert('example.com'), 'example.com'),
      None)
check(_verifycert(cert('example.com'), 'www.example.com'),
      b'certificate is for example.com')
check(_verifycert(cert('www.example.com'), 'example.com'),
      b'certificate is for www.example.com')

# Test wildcard certificates
check(_verifycert(cert('*.example.com'), 'www.example.com'),
      None)
check(_verifycert(cert('*.example.com'), 'example.com'),
      b'certificate is for *.example.com')
check(_verifycert(cert('*.example.com'), 'w.w.example.com'),
      b'certificate is for *.example.com')

# Test subjectAltName
san_cert = {'subject': ((('commonName', 'example.com'),),),
            'subjectAltName': (('DNS', '*.example.net'),
                               ('DNS', 'example.net'))}
check(_verifycert(san_cert, 'example.net'),
      None)
check(_verifycert(san_cert, 'foo.example.net'),
      None)
# no fallback to subject commonName when subjectAltName has DNS
check(_verifycert(san_cert, 'example.com'),
      b'certificate is for *.example.net, example.net')
# fallback to subject commonName when no DNS in subjectAltName
san_cert = {'subject': ((('commonName', 'example.com'),),),
            'subjectAltName': (('IP Address', '8.8.8.8'),)}
check(_verifycert(san_cert, 'example.com'), None)

# Avoid some pitfalls
check(_verifycert(cert('*.foo'), 'foo'),
      b'certificate is for *.foo')
check(_verifycert(cert('*o'), 'foo'), None)

check(_verifycert({'subject': ()},
                  'example.com'),
      b'no commonName or subjectAltName found in certificate')
check(_verifycert(None, 'example.com'),
      b'no certificate received')

# Unicode (IDN) certname isn't supported
check(_verifycert(cert(u'\u4f8b.jp'), 'example.jp'),
      b'IDN in certificate not supported')

# The following tests are from CPython's test_ssl.py.
check(_verifycert(cert('example.com'), 'example.com'), None)
check(_verifycert(cert('example.com'), 'ExAmple.cOm'), None)
check(_verifycert(cert('example.com'), 'www.example.com'),
      b'certificate is for example.com')
check(_verifycert(cert('example.com'), '.example.com'),
      b'certificate is for example.com')
check(_verifycert(cert('example.com'), 'example.org'),
      b'certificate is for example.com')
check(_verifycert(cert('example.com'), 'exampleXcom'),
      b'certificate is for example.com')
check(_verifycert(cert('*.a.com'), 'foo.a.com'), None)
check(_verifycert(cert('*.a.com'), 'bar.foo.a.com'),
      b'certificate is for *.a.com')
check(_verifycert(cert('*.a.com'), 'a.com'),
      b'certificate is for *.a.com')
check(_verifycert(cert('*.a.com'), 'Xa.com'),
      b'certificate is for *.a.com')
check(_verifycert(cert('*.a.com'), '.a.com'),
      b'certificate is for *.a.com')

# only match one left-most wildcard
check(_verifycert(cert('f*.com'), 'foo.com'), None)
check(_verifycert(cert('f*.com'), 'f.com'), None)
check(_verifycert(cert('f*.com'), 'bar.com'),
      b'certificate is for f*.com')
check(_verifycert(cert('f*.com'), 'foo.a.com'),
      b'certificate is for f*.com')
check(_verifycert(cert('f*.com'), 'bar.foo.com'),
      b'certificate is for f*.com')

# NULL bytes are bad, CVE-2013-4073
check(_verifycert(cert('null.python.org\x00example.org'),
                  'null.python.org\x00example.org'), None)
check(_verifycert(cert('null.python.org\x00example.org'),
                  'example.org'),
      b'certificate is for null.python.org\x00example.org')
check(_verifycert(cert('null.python.org\x00example.org'),
                  'null.python.org'),
      b'certificate is for null.python.org\x00example.org')

# error cases with wildcards
check(_verifycert(cert('*.*.a.com'), 'bar.foo.a.com'),
      b'certificate is for *.*.a.com')
check(_verifycert(cert('*.*.a.com'), 'a.com'),
      b'certificate is for *.*.a.com')
check(_verifycert(cert('*.*.a.com'), 'Xa.com'),
      b'certificate is for *.*.a.com')
check(_verifycert(cert('*.*.a.com'), '.a.com'),
      b'certificate is for *.*.a.com')

check(_verifycert(cert('a.*.com'), 'a.foo.com'),
      b'certificate is for a.*.com')
check(_verifycert(cert('a.*.com'), 'a..com'),
      b'certificate is for a.*.com')
check(_verifycert(cert('a.*.com'), 'a.com'),
      b'certificate is for a.*.com')

# wildcard doesn't match IDNA prefix 'xn--'
idna = u'püthon.python.org'.encode('idna').decode('ascii')
check(_verifycert(cert(idna), idna), None)
check(_verifycert(cert('x*.python.org'), idna),
      b'certificate is for x*.python.org')
check(_verifycert(cert('xn--p*.python.org'), idna),
      b'certificate is for xn--p*.python.org')

# wildcard in first fragment and  IDNA A-labels in sequent fragments
# are supported.
idna = u'www*.pythön.org'.encode('idna').decode('ascii')
check(_verifycert(cert(idna),
                  u'www.pythön.org'.encode('idna').decode('ascii')),
      None)
check(_verifycert(cert(idna),
                  u'www1.pythön.org'.encode('idna').decode('ascii')),
      None)
check(_verifycert(cert(idna),
                  u'ftp.pythön.org'.encode('idna').decode('ascii')),
      b'certificate is for www*.xn--pythn-mua.org')
check(_verifycert(cert(idna),
                  u'pythön.org'.encode('idna').decode('ascii')),
      b'certificate is for www*.xn--pythn-mua.org')

c = {
    'notAfter': 'Jun 26 21:41:46 2011 GMT',
    'subject': (((u'commonName', u'linuxfrz.org'),),),
    'subjectAltName': (
        ('DNS', 'linuxfr.org'),
        ('DNS', 'linuxfr.com'),
        ('othername', '<unsupported>'),
    )
}
check(_verifycert(c, 'linuxfr.org'), None)
check(_verifycert(c, 'linuxfr.com'), None)
# Not a "DNS" entry
check(_verifycert(c, '<unsupported>'),
      b'certificate is for linuxfr.org, linuxfr.com')
# When there is a subjectAltName, commonName isn't used
check(_verifycert(c, 'linuxfrz.org'),
      b'certificate is for linuxfr.org, linuxfr.com')

# A pristine real-world example
c = {
    'notAfter': 'Dec 18 23:59:59 2011 GMT',
    'subject': (
        ((u'countryName', u'US'),),
        ((u'stateOrProvinceName', u'California'),),
        ((u'localityName', u'Mountain View'),),
        ((u'organizationName', u'Google Inc'),),
        ((u'commonName', u'mail.google.com'),),
    ),
}
check(_verifycert(c, 'mail.google.com'), None)
check(_verifycert(c, 'gmail.com'), b'certificate is for mail.google.com')

# Only commonName is considered
check(_verifycert(c, 'California'), b'certificate is for mail.google.com')

# Neither commonName nor subjectAltName
c = {
    'notAfter': 'Dec 18 23:59:59 2011 GMT',
    'subject': (
        ((u'countryName', u'US'),),
        ((u'stateOrProvinceName', u'California'),),
        ((u'localityName', u'Mountain View'),),
        ((u'organizationName', u'Google Inc'),),
    ),
}
check(_verifycert(c, 'mail.google.com'),
      b'no commonName or subjectAltName found in certificate')

# No DNS entry in subjectAltName but a commonName
c = {
    'notAfter': 'Dec 18 23:59:59 2099 GMT',
    'subject': (
        ((u'countryName', u'US'),),
        ((u'stateOrProvinceName', u'California'),),
        ((u'localityName', u'Mountain View'),),
        ((u'commonName', u'mail.google.com'),),
    ),
    'subjectAltName': (('othername', 'blabla'),),
}
check(_verifycert(c, 'mail.google.com'), None)

# No DNS entry subjectAltName and no commonName
c = {
    'notAfter': 'Dec 18 23:59:59 2099 GMT',
    'subject': (
        ((u'countryName', u'US'),),
        ((u'stateOrProvinceName', u'California'),),
        ((u'localityName', u'Mountain View'),),
        ((u'organizationName', u'Google Inc'),),
    ),
    'subjectAltName': (('othername', 'blabla'),),
}
check(_verifycert(c, 'google.com'),
      b'no commonName or subjectAltName found in certificate')

# Empty cert / no cert
check(_verifycert(None, 'example.com'), b'no certificate received')
check(_verifycert({}, 'example.com'), b'no certificate received')

# avoid denials of service by refusing more than one
# wildcard per fragment.
check(_verifycert({'subject': (((u'commonName', u'a*b.com'),),)},
                  'axxb.com'), None)
check(_verifycert({'subject': (((u'commonName', u'a*b.co*'),),)},
                  'axxb.com'), b'certificate is for a*b.co*')
check(_verifycert({'subject': (((u'commonName', u'a*b*.com'),),)},
                  'axxbxxc.com'),
      b'too many wildcards in certificate DNS name: a*b*.com')

def test_url():
    """
    >>> from mercurial import error, pycompat
    >>> from mercurial.util import url
    >>> from mercurial.utils.stringutil import forcebytestr

    This tests for edge cases in url.URL's parsing algorithm. Most of
    these aren't useful for documentation purposes, so they aren't
    part of the class's doc tests.

    Query strings and fragments:

    >>> url(b'http://host/a?b#c')
    <url scheme: 'http', host: 'host', path: 'a', query: 'b', fragment: 'c'>
    >>> url(b'http://host/a?')
    <url scheme: 'http', host: 'host', path: 'a'>
    >>> url(b'http://host/a#b#c')
    <url scheme: 'http', host: 'host', path: 'a', fragment: 'b#c'>
    >>> url(b'http://host/a#b?c')
    <url scheme: 'http', host: 'host', path: 'a', fragment: 'b?c'>
    >>> url(b'http://host/?a#b')
    <url scheme: 'http', host: 'host', path: '', query: 'a', fragment: 'b'>
    >>> url(b'http://host/?a#b', parsequery=False)
    <url scheme: 'http', host: 'host', path: '?a', fragment: 'b'>
    >>> url(b'http://host/?a#b', parsefragment=False)
    <url scheme: 'http', host: 'host', path: '', query: 'a#b'>
    >>> url(b'http://host/?a#b', parsequery=False, parsefragment=False)
    <url scheme: 'http', host: 'host', path: '?a#b'>

    IPv6 addresses:

    >>> url(b'ldap://[2001:db8::7]/c=GB?objectClass?one')
    <url scheme: 'ldap', host: '[2001:db8::7]', path: 'c=GB',
         query: 'objectClass?one'>
    >>> url(b'ldap://joe:xxx@[2001:db8::7]:80/c=GB?objectClass?one')
    <url scheme: 'ldap', user: 'joe', passwd: 'xxx', host: '[2001:db8::7]',
         port: '80', path: 'c=GB', query: 'objectClass?one'>

    Missing scheme, host, etc.:

    >>> url(b'://192.0.2.16:80/')
    <url path: '://192.0.2.16:80/'>
    >>> url(b'https://mercurial-scm.org')
    <url scheme: 'https', host: 'mercurial-scm.org'>
    >>> url(b'/foo')
    <url path: '/foo'>
    >>> url(b'bundle:/foo')
    <url scheme: 'bundle', path: '/foo'>
    >>> url(b'a?b#c')
    <url path: 'a?b', fragment: 'c'>
    >>> url(b'http://x.com?arg=/foo')
    <url scheme: 'http', host: 'x.com', query: 'arg=/foo'>
    >>> url(b'http://joe:xxx@/foo')
    <url scheme: 'http', user: 'joe', passwd: 'xxx', path: 'foo'>

    Just a scheme and a path:

    >>> url(b'mailto:[e-mail]')
    <url scheme: 'mailto', path: '[e-mail]'>
    >>> url(b'a:b:c:d')
    <url path: 'a:b:c:d'>
    >>> url(b'aa:bb:cc:dd')
    <url scheme: 'aa', path: 'bb:cc:dd'>

    SSH examples:

    >>> url(b'ssh://joe@host//home/joe')
    <url scheme: 'ssh', user: 'joe', host: 'host', path: '/home/joe'>
    >>> url(b'ssh://joe:xxx@host/src')
    <url scheme: 'ssh', user: 'joe', passwd: 'xxx', host: 'host', path: 'src'>
    >>> url(b'ssh://joe:xxx@host')
    <url scheme: 'ssh', user: 'joe', passwd: 'xxx', host: 'host'>
    >>> url(b'ssh://joe@host')
    <url scheme: 'ssh', user: 'joe', host: 'host'>
    >>> url(b'ssh://host')
    <url scheme: 'ssh', host: 'host'>
    >>> url(b'ssh://')
    <url scheme: 'ssh'>
    >>> url(b'ssh:')
    <url scheme: 'ssh'>

    Non-numeric port:

    >>> url(b'http://example.com:dd')
    <url scheme: 'http', host: 'example.com', port: 'dd'>
    >>> url(b'ssh://joe:xxx@host:ssh/foo')
    <url scheme: 'ssh', user: 'joe', passwd: 'xxx', host: 'host', port: 'ssh',
         path: 'foo'>

    Bad authentication credentials:

    >>> url(b'http://joe@joeville:123@4:@host/a?b#c')
    <url scheme: 'http', user: 'joe@joeville', passwd: '123@4:',
         host: 'host', path: 'a', query: 'b', fragment: 'c'>
    >>> url(b'http://!*#?/@!*#?/:@host/a?b#c')
    <url scheme: 'http', host: '!*', fragment: '?/@!*#?/:@host/a?b#c'>
    >>> url(b'http://!*#?@!*#?:@host/a?b#c')
    <url scheme: 'http', host: '!*', fragment: '?@!*#?:@host/a?b#c'>
    >>> url(b'http://!*@:!*@@host/a?b#c')
    <url scheme: 'http', user: '!*@', passwd: '!*@', host: 'host',
         path: 'a', query: 'b', fragment: 'c'>

    File paths:

    >>> url(b'a/b/c/d.g.f')
    <url path: 'a/b/c/d.g.f'>
    >>> url(b'/x///z/y/')
    <url path: '/x///z/y/'>
    >>> url(b'/foo:bar')
    <url path: '/foo:bar'>
    >>> url(b'\\\\foo:bar')
    <url path: '\\\\foo:bar'>
    >>> url(b'./foo:bar')
    <url path: './foo:bar'>

    Non-localhost file URL:

    >>> try:
    ...   u = url(b'file://mercurial-scm.org/foo')
    ... except error.Abort as e:
    ...   forcebytestr(e)
    'file:// URLs can only refer to localhost'

    Empty URL:

    >>> u = url(b'')
    >>> u
    <url path: ''>
    >>> str(u)
    ''

    Empty path with query string:

    >>> str(url(b'http://foo/?bar'))
    'http://foo/?bar'

    Invalid path:

    >>> u = url(b'http://foo/bar')
    >>> u.path = b'bar'
    >>> str(u)
    'http://foo/bar'

    >>> u = url(b'file:/foo/bar/baz')
    >>> u
    <url scheme: 'file', path: '/foo/bar/baz'>
    >>> str(u)
    'file:///foo/bar/baz'
    >>> pycompat.bytestr(u.localpath())
    '/foo/bar/baz'

    >>> u = url(b'file:///foo/bar/baz')
    >>> u
    <url scheme: 'file', path: '/foo/bar/baz'>
    >>> str(u)
    'file:///foo/bar/baz'
    >>> pycompat.bytestr(u.localpath())
    '/foo/bar/baz'

    >>> u = url(b'file:///f:oo/bar/baz')
    >>> u
    <url scheme: 'file', path: 'f:oo/bar/baz'>
    >>> str(u)
    'file:///f:oo/bar/baz'
    >>> pycompat.bytestr(u.localpath())
    'f:oo/bar/baz'

    >>> u = url(b'file://localhost/f:oo/bar/baz')
    >>> u
    <url scheme: 'file', host: 'localhost', path: 'f:oo/bar/baz'>
    >>> str(u)
    'file://localhost/f:oo/bar/baz'
    >>> pycompat.bytestr(u.localpath())
    'f:oo/bar/baz'

    >>> u = url(b'file:foo/bar/baz')
    >>> u
    <url scheme: 'file', path: 'foo/bar/baz'>
    >>> str(u)
    'file:foo/bar/baz'
    >>> pycompat.bytestr(u.localpath())
    'foo/bar/baz'
    """

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

doctest.testmod(optionflags=doctest.NORMALIZE_WHITESPACE)