~bzr-pqm/bzr/bzr.dev : contents of bzrlib/tests/test

~bzr-pqm/bzr/bzr.dev : (revision 5397)
# Copyright (C) 2006-2010 Canonical Ltd
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

"""Tests for Knit data structure"""

from cStringIO import StringIO
import sys

from bzrlib import (
    errors,
    knit,
    multiparent,
    osutils,
    pack,
    tests,
    transport,
    tuned_gzip,
    )
from bzrlib.errors import (
    KnitHeaderError,
    NoSuchFile,
    )
from bzrlib.index import *
from bzrlib.knit import (
    AnnotatedKnitContent,
    KnitContent,
    KnitVersionedFiles,
    PlainKnitContent,
    _VFContentMapGenerator,
    _DirectPackAccess,
    _KndxIndex,
    _KnitGraphIndex,
    _KnitKeyAccess,
    make_file_factory,
    )
from bzrlib.patiencediff import PatienceSequenceMatcher
from bzrlib.repofmt import pack_repo
from bzrlib.tests import (
    TestCase,
    TestCaseWithMemoryTransport,
    TestCaseWithTransport,
    TestNotApplicable,
    )
from bzrlib.versionedfile import (
    AbsentContentFactory,
    ConstantMapper,
    network_bytes_to_kind_and_offset,
    RecordingVersionedFilesDecorator,
    )


compiled_knit_feature = tests.ModuleAvailableFeature(
                            'bzrlib._knit_load_data_pyx')


class KnitContentTestsMixin(object):

    def test_constructor(self):
        content = self._make_content([])

    def test_text(self):
        content = self._make_content([])
        self.assertEqual(content.text(), [])

        content = self._make_content([("origin1", "text1"), ("origin2", "text2")])
        self.assertEqual(content.text(), ["text1", "text2"])

    def test_copy(self):
        content = self._make_content([("origin1", "text1"), ("origin2", "text2")])
        copy = content.copy()
        self.assertIsInstance(copy, content.__class__)
        self.assertEqual(copy.annotate(), content.annotate())

    def assertDerivedBlocksEqual(self, source, target, noeol=False):
        """Assert that the derived matching blocks match real output"""
        source_lines = source.splitlines(True)
        target_lines = target.splitlines(True)
        def nl(line):
            if noeol and not line.endswith('\n'):
                return line + '\n'
            else:
                return line
        source_content = self._make_content([(None, nl(l)) for l in source_lines])
        target_content = self._make_content([(None, nl(l)) for l in target_lines])
        line_delta = source_content.line_delta(target_content)
        delta_blocks = list(KnitContent.get_line_delta_blocks(line_delta,
            source_lines, target_lines))
        matcher = PatienceSequenceMatcher(None, source_lines, target_lines)
        matcher_blocks = list(matcher.get_matching_blocks())
        self.assertEqual(matcher_blocks, delta_blocks)

    def test_get_line_delta_blocks(self):
        self.assertDerivedBlocksEqual('a\nb\nc\n', 'q\nc\n')
        self.assertDerivedBlocksEqual(TEXT_1, TEXT_1)
        self.assertDerivedBlocksEqual(TEXT_1, TEXT_1A)
        self.assertDerivedBlocksEqual(TEXT_1, TEXT_1B)
        self.assertDerivedBlocksEqual(TEXT_1B, TEXT_1A)
        self.assertDerivedBlocksEqual(TEXT_1A, TEXT_1B)
        self.assertDerivedBlocksEqual(TEXT_1A, '')
        self.assertDerivedBlocksEqual('', TEXT_1A)
        self.assertDerivedBlocksEqual('', '')
        self.assertDerivedBlocksEqual('a\nb\nc', 'a\nb\nc\nd')

    def test_get_line_delta_blocks_noeol(self):
        """Handle historical knit deltas safely

        Some existing knit deltas don't consider the last line to differ
        when the only difference whether it has a final newline.

        New knit deltas appear to always consider the last line to differ
        in this case.
        """
        self.assertDerivedBlocksEqual('a\nb\nc', 'a\nb\nc\nd\n', noeol=True)
        self.assertDerivedBlocksEqual('a\nb\nc\nd\n', 'a\nb\nc', noeol=True)
        self.assertDerivedBlocksEqual('a\nb\nc\n', 'a\nb\nc', noeol=True)
        self.assertDerivedBlocksEqual('a\nb\nc', 'a\nb\nc\n', noeol=True)


TEXT_1 = """\
Banana cup cakes:

- bananas
- eggs
- broken tea cups
"""

TEXT_1A = """\
Banana cup cake recipe
(serves 6)

- bananas
- eggs
- broken tea cups
- self-raising flour
"""

TEXT_1B = """\
Banana cup cake recipe

- bananas (do not use plantains!!!)
- broken tea cups
- flour
"""

delta_1_1a = """\
0,1,2
Banana cup cake recipe
(serves 6)
5,5,1
- self-raising flour
"""

TEXT_2 = """\
Boeuf bourguignon

- beef
- red wine
- small onions
- carrot
- mushrooms
"""


class TestPlainKnitContent(TestCase, KnitContentTestsMixin):

    def _make_content(self, lines):
        annotated_content = AnnotatedKnitContent(lines)
        return PlainKnitContent(annotated_content.text(), 'bogus')

    def test_annotate(self):
        content = self._make_content([])
        self.assertEqual(content.annotate(), [])

        content = self._make_content([("origin1", "text1"), ("origin2", "text2")])
        self.assertEqual(content.annotate(),
            [("bogus", "text1"), ("bogus", "text2")])

    def test_line_delta(self):
        content1 = self._make_content([("", "a"), ("", "b")])
        content2 = self._make_content([("", "a"), ("", "a"), ("", "c")])
        self.assertEqual(content1.line_delta(content2),
            [(1, 2, 2, ["a", "c"])])

    def test_line_delta_iter(self):
        content1 = self._make_content([("", "a"), ("", "b")])
        content2 = self._make_content([("", "a"), ("", "a"), ("", "c")])
        it = content1.line_delta_iter(content2)
        self.assertEqual(it.next(), (1, 2, 2, ["a", "c"]))
        self.assertRaises(StopIteration, it.next)


class TestAnnotatedKnitContent(TestCase, KnitContentTestsMixin):

    def _make_content(self, lines):
        return AnnotatedKnitContent(lines)

    def test_annotate(self):
        content = self._make_content([])
        self.assertEqual(content.annotate(), [])

        content = self._make_content([("origin1", "text1"), ("origin2", "text2")])
        self.assertEqual(content.annotate(),
            [("origin1", "text1"), ("origin2", "text2")])

    def test_line_delta(self):
        content1 = self._make_content([("", "a"), ("", "b")])
        content2 = self._make_content([("", "a"), ("", "a"), ("", "c")])
        self.assertEqual(content1.line_delta(content2),
            [(1, 2, 2, [("", "a"), ("", "c")])])

    def test_line_delta_iter(self):
        content1 = self._make_content([("", "a"), ("", "b")])
        content2 = self._make_content([("", "a"), ("", "a"), ("", "c")])
        it = content1.line_delta_iter(content2)
        self.assertEqual(it.next(), (1, 2, 2, [("", "a"), ("", "c")]))
        self.assertRaises(StopIteration, it.next)


class MockTransport(object):

    def __init__(self, file_lines=None):
        self.file_lines = file_lines
        self.calls = []
        # We have no base directory for the MockTransport
        self.base = ''

    def get(self, filename):
        if self.file_lines is None:
            raise NoSuchFile(filename)
        else:
            return StringIO("\n".join(self.file_lines))

    def readv(self, relpath, offsets):
        fp = self.get(relpath)
        for offset, size in offsets:
            fp.seek(offset)
            yield offset, fp.read(size)

    def __getattr__(self, name):
        def queue_call(*args, **kwargs):
            self.calls.append((name, args, kwargs))
        return queue_call


class MockReadvFailingTransport(MockTransport):
    """Fail in the middle of a readv() result.

    This Transport will successfully yield the first two requested hunks, but
    raise NoSuchFile for the rest.
    """

    def readv(self, relpath, offsets):
        count = 0
        for result in MockTransport.readv(self, relpath, offsets):
            count += 1
            # we use 2 because the first offset is the pack header, the second
            # is the first actual content requset
            if count > 2:
                raise errors.NoSuchFile(relpath)
            yield result


class KnitRecordAccessTestsMixin(object):
    """Tests for getting and putting knit records."""

    def test_add_raw_records(self):
        """Add_raw_records adds records retrievable later."""
        access = self.get_access()
        memos = access.add_raw_records([('key', 10)], '1234567890')
        self.assertEqual(['1234567890'], list(access.get_raw_records(memos)))

    def test_add_several_raw_records(self):
        """add_raw_records with many records and read some back."""
        access = self.get_access()
        memos = access.add_raw_records([('key', 10), ('key2', 2), ('key3', 5)],
            '12345678901234567')
        self.assertEqual(['1234567890', '12', '34567'],
            list(access.get_raw_records(memos)))
        self.assertEqual(['1234567890'],
            list(access.get_raw_records(memos[0:1])))
        self.assertEqual(['12'],
            list(access.get_raw_records(memos[1:2])))
        self.assertEqual(['34567'],
            list(access.get_raw_records(memos[2:3])))
        self.assertEqual(['1234567890', '34567'],
            list(access.get_raw_records(memos[0:1] + memos[2:3])))


class TestKnitKnitAccess(TestCaseWithMemoryTransport, KnitRecordAccessTestsMixin):
    """Tests for the .kndx implementation."""

    def get_access(self):
        """Get a .knit style access instance."""
        mapper = ConstantMapper("foo")
        access = _KnitKeyAccess(self.get_transport(), mapper)
        return access


class _TestException(Exception):
    """Just an exception for local tests to use."""


class TestPackKnitAccess(TestCaseWithMemoryTransport, KnitRecordAccessTestsMixin):
    """Tests for the pack based access."""

    def get_access(self):
        return self._get_access()[0]

    def _get_access(self, packname='packfile', index='FOO'):
        transport = self.get_transport()
        def write_data(bytes):
            transport.append_bytes(packname, bytes)
        writer = pack.ContainerWriter(write_data)
        writer.begin()
        access = _DirectPackAccess({})
        access.set_writer(writer, index, (transport, packname))
        return access, writer

    def make_pack_file(self):
        """Create a pack file with 2 records."""
        access, writer = self._get_access(packname='packname', index='foo')
        memos = []
        memos.extend(access.add_raw_records([('key1', 10)], '1234567890'))
        memos.extend(access.add_raw_records([('key2', 5)], '12345'))
        writer.end()
        return memos

    def make_vf_for_retrying(self):
        """Create 3 packs and a reload function.

        Originally, 2 pack files will have the data, but one will be missing.
        And then the third will be used in place of the first two if reload()
        is called.

        :return: (versioned_file, reload_counter)
            versioned_file  a KnitVersionedFiles using the packs for access
        """
        builder = self.make_branch_builder('.', format="1.9")
        builder.start_series()
        builder.build_snapshot('rev-1', None, [
            ('add', ('', 'root-id', 'directory', None)),
            ('add', ('file', 'file-id', 'file', 'content\nrev 1\n')),
            ])
        builder.build_snapshot('rev-2', ['rev-1'], [
            ('modify', ('file-id', 'content\nrev 2\n')),
            ])
        builder.build_snapshot('rev-3', ['rev-2'], [
            ('modify', ('file-id', 'content\nrev 3\n')),
            ])
        builder.finish_series()
        b = builder.get_branch()
        b.lock_write()
        self.addCleanup(b.unlock)
        # Pack these three revisions into another pack file, but don't remove
        # the originals
        repo = b.repository
        collection = repo._pack_collection
        collection.ensure_loaded()
        orig_packs = collection.packs
        packer = pack_repo.Packer(collection, orig_packs, '.testpack')
        new_pack = packer.pack()
        # forget about the new pack
        collection.reset()
        repo.refresh_data()
        vf = repo.revisions
        # Set up a reload() function that switches to using the new pack file
        new_index = new_pack.revision_index
        access_tuple = new_pack.access_tuple()
        reload_counter = [0, 0, 0]
        def reload():
            reload_counter[0] += 1
            if reload_counter[1] > 0:
                # We already reloaded, nothing more to do
                reload_counter[2] += 1
                return False
            reload_counter[1] += 1
            vf._index._graph_index._indices[:] = [new_index]
            vf._access._indices.clear()
            vf._access._indices[new_index] = access_tuple
            return True
        # Delete one of the pack files so the data will need to be reloaded. We
        # will delete the file with 'rev-2' in it
        trans, name = orig_packs[1].access_tuple()
        trans.delete(name)
        # We don't have the index trigger reloading because we want to test
        # that we reload when the .pack disappears
        vf._access._reload_func = reload
        return vf, reload_counter

    def make_reload_func(self, return_val=True):
        reload_called = [0]
        def reload():
            reload_called[0] += 1
            return return_val
        return reload_called, reload

    def make_retry_exception(self):
        # We raise a real exception so that sys.exc_info() is properly
        # populated
        try:
            raise _TestException('foobar')
        except _TestException, e:
            retry_exc = errors.RetryWithNewPacks(None, reload_occurred=False,
                                                 exc_info=sys.exc_info())
        return retry_exc

    def test_read_from_several_packs(self):
        access, writer = self._get_access()
        memos = []
        memos.extend(access.add_raw_records([('key', 10)], '1234567890'))
        writer.end()
        access, writer = self._get_access('pack2', 'FOOBAR')
        memos.extend(access.add_raw_records([('key', 5)], '12345'))
        writer.end()
        access, writer = self._get_access('pack3', 'BAZ')
        memos.extend(access.add_raw_records([('key', 5)], 'alpha'))
        writer.end()
        transport = self.get_transport()
        access = _DirectPackAccess({"FOO":(transport, 'packfile'),
            "FOOBAR":(transport, 'pack2'),
            "BAZ":(transport, 'pack3')})
        self.assertEqual(['1234567890', '12345', 'alpha'],
            list(access.get_raw_records(memos)))
        self.assertEqual(['1234567890'],
            list(access.get_raw_records(memos[0:1])))
        self.assertEqual(['12345'],
            list(access.get_raw_records(memos[1:2])))
        self.assertEqual(['alpha'],
            list(access.get_raw_records(memos[2:3])))
        self.assertEqual(['1234567890', 'alpha'],
            list(access.get_raw_records(memos[0:1] + memos[2:3])))

    def test_set_writer(self):
        """The writer should be settable post construction."""
        access = _DirectPackAccess({})
        transport = self.get_transport()
        packname = 'packfile'
        index = 'foo'
        def write_data(bytes):
            transport.append_bytes(packname, bytes)
        writer = pack.ContainerWriter(write_data)
        writer.begin()
        access.set_writer(writer, index, (transport, packname))
        memos = access.add_raw_records([('key', 10)], '1234567890')
        writer.end()
        self.assertEqual(['1234567890'], list(access.get_raw_records(memos)))

    def test_missing_index_raises_retry(self):
        memos = self.make_pack_file()
        transport = self.get_transport()
        reload_called, reload_func = self.make_reload_func()
        # Note that the index key has changed from 'foo' to 'bar'
        access = _DirectPackAccess({'bar':(transport, 'packname')},
                                   reload_func=reload_func)
        e = self.assertListRaises(errors.RetryWithNewPacks,
                                  access.get_raw_records, memos)
        # Because a key was passed in which does not match our index list, we
        # assume that the listing was already reloaded
        self.assertTrue(e.reload_occurred)
        self.assertIsInstance(e.exc_info, tuple)
        self.assertIs(e.exc_info[0], KeyError)
        self.assertIsInstance(e.exc_info[1], KeyError)

    def test_missing_index_raises_key_error_with_no_reload(self):
        memos = self.make_pack_file()
        transport = self.get_transport()
        # Note that the index key has changed from 'foo' to 'bar'
        access = _DirectPackAccess({'bar':(transport, 'packname')})
        e = self.assertListRaises(KeyError, access.get_raw_records, memos)

    def test_missing_file_raises_retry(self):
        memos = self.make_pack_file()
        transport = self.get_transport()
        reload_called, reload_func = self.make_reload_func()
        # Note that the 'filename' has been changed to 'different-packname'
        access = _DirectPackAccess({'foo':(transport, 'different-packname')},
                                   reload_func=reload_func)
        e = self.assertListRaises(errors.RetryWithNewPacks,
                                  access.get_raw_records, memos)
        # The file has gone missing, so we assume we need to reload
        self.assertFalse(e.reload_occurred)
        self.assertIsInstance(e.exc_info, tuple)
        self.assertIs(e.exc_info[0], errors.NoSuchFile)
        self.assertIsInstance(e.exc_info[1], errors.NoSuchFile)
        self.assertEqual('different-packname', e.exc_info[1].path)

    def test_missing_file_raises_no_such_file_with_no_reload(self):
        memos = self.make_pack_file()
        transport = self.get_transport()
        # Note that the 'filename' has been changed to 'different-packname'
        access = _DirectPackAccess({'foo':(transport, 'different-packname')})
        e = self.assertListRaises(errors.NoSuchFile,
                                  access.get_raw_records, memos)

    def test_failing_readv_raises_retry(self):
        memos = self.make_pack_file()
        transport = self.get_transport()
        failing_transport = MockReadvFailingTransport(
                                [transport.get_bytes('packname')])
        reload_called, reload_func = self.make_reload_func()
        access = _DirectPackAccess({'foo':(failing_transport, 'packname')},
                                   reload_func=reload_func)
        # Asking for a single record will not trigger the Mock failure
        self.assertEqual(['1234567890'],
            list(access.get_raw_records(memos[:1])))
        self.assertEqual(['12345'],
            list(access.get_raw_records(memos[1:2])))
        # A multiple offset readv() will fail mid-way through
        e = self.assertListRaises(errors.RetryWithNewPacks,
                                  access.get_raw_records, memos)
        # The file has gone missing, so we assume we need to reload
        self.assertFalse(e.reload_occurred)
        self.assertIsInstance(e.exc_info, tuple)
        self.assertIs(e.exc_info[0], errors.NoSuchFile)
        self.assertIsInstance(e.exc_info[1], errors.NoSuchFile)
        self.assertEqual('packname', e.exc_info[1].path)

    def test_failing_readv_raises_no_such_file_with_no_reload(self):
        memos = self.make_pack_file()
        transport = self.get_transport()
        failing_transport = MockReadvFailingTransport(
                                [transport.get_bytes('packname')])
        reload_called, reload_func = self.make_reload_func()
        access = _DirectPackAccess({'foo':(failing_transport, 'packname')})
        # Asking for a single record will not trigger the Mock failure
        self.assertEqual(['1234567890'],
            list(access.get_raw_records(memos[:1])))
        self.assertEqual(['12345'],
            list(access.get_raw_records(memos[1:2])))
        # A multiple offset readv() will fail mid-way through
        e = self.assertListRaises(errors.NoSuchFile,
                                  access.get_raw_records, memos)

    def test_reload_or_raise_no_reload(self):
        access = _DirectPackAccess({}, reload_func=None)
        retry_exc = self.make_retry_exception()
        # Without a reload_func, we will just re-raise the original exception
        self.assertRaises(_TestException, access.reload_or_raise, retry_exc)

    def test_reload_or_raise_reload_changed(self):
        reload_called, reload_func = self.make_reload_func(return_val=True)
        access = _DirectPackAccess({}, reload_func=reload_func)
        retry_exc = self.make_retry_exception()
        access.reload_or_raise(retry_exc)
        self.assertEqual([1], reload_called)
        retry_exc.reload_occurred=True
        access.reload_or_raise(retry_exc)
        self.assertEqual([2], reload_called)

    def test_reload_or_raise_reload_no_change(self):
        reload_called, reload_func = self.make_reload_func(return_val=False)
        access = _DirectPackAccess({}, reload_func=reload_func)
        retry_exc = self.make_retry_exception()
        # If reload_occurred is False, then we consider it an error to have
        # reload_func() return False (no changes).
        self.assertRaises(_TestException, access.reload_or_raise, retry_exc)
        self.assertEqual([1], reload_called)
        retry_exc.reload_occurred=True
        # If reload_occurred is True, then we assume nothing changed because
        # it had changed earlier, but didn't change again
        access.reload_or_raise(retry_exc)
        self.assertEqual([2], reload_called)

    def test_annotate_retries(self):
        vf, reload_counter = self.make_vf_for_retrying()
        # It is a little bit bogus to annotate the Revision VF, but it works,
        # as we have ancestry stored there
        key = ('rev-3',)
        reload_lines = vf.annotate(key)
        self.assertEqual([1, 1, 0], reload_counter)
        plain_lines = vf.annotate(key)
        self.assertEqual([1, 1, 0], reload_counter) # No extra reloading
        if reload_lines != plain_lines:
            self.fail('Annotation was not identical with reloading.')
        # Now delete the packs-in-use, which should trigger another reload, but
        # this time we just raise an exception because we can't recover
        for trans, name in vf._access._indices.itervalues():
            trans.delete(name)
        self.assertRaises(errors.NoSuchFile, vf.annotate, key)
        self.assertEqual([2, 1, 1], reload_counter)

    def test__get_record_map_retries(self):
        vf, reload_counter = self.make_vf_for_retrying()
        keys = [('rev-1',), ('rev-2',), ('rev-3',)]
        records = vf._get_record_map(keys)
        self.assertEqual(keys, sorted(records.keys()))
        self.assertEqual([1, 1, 0], reload_counter)
        # Now delete the packs-in-use, which should trigger another reload, but
        # this time we just raise an exception because we can't recover
        for trans, name in vf._access._indices.itervalues():
            trans.delete(name)
        self.assertRaises(errors.NoSuchFile, vf._get_record_map, keys)
        self.assertEqual([2, 1, 1], reload_counter)

    def test_get_record_stream_retries(self):
        vf, reload_counter = self.make_vf_for_retrying()
        keys = [('rev-1',), ('rev-2',), ('rev-3',)]
        record_stream = vf.get_record_stream(keys, 'topological', False)
        record = record_stream.next()
        self.assertEqual(('rev-1',), record.key)
        self.assertEqual([0, 0, 0], reload_counter)
        record = record_stream.next()
        self.assertEqual(('rev-2',), record.key)
        self.assertEqual([1, 1, 0], reload_counter)
        record = record_stream.next()
        self.assertEqual(('rev-3',), record.key)
        self.assertEqual([1, 1, 0], reload_counter)
        # Now delete all pack files, and see that we raise the right error
        for trans, name in vf._access._indices.itervalues():
            trans.delete(name)
        self.assertListRaises(errors.NoSuchFile,
            vf.get_record_stream, keys, 'topological', False)

    def test_iter_lines_added_or_present_in_keys_retries(self):
        vf, reload_counter = self.make_vf_for_retrying()
        keys = [('rev-1',), ('rev-2',), ('rev-3',)]
        # Unfortunately, iter_lines_added_or_present_in_keys iterates the
        # result in random order (determined by the iteration order from a
        # set()), so we don't have any solid way to trigger whether data is
        # read before or after. However we tried to delete the middle node to
        # exercise the code well.
        # What we care about is that all lines are always yielded, but not
        # duplicated
        count = 0
        reload_lines = sorted(vf.iter_lines_added_or_present_in_keys(keys))
        self.assertEqual([1, 1, 0], reload_counter)
        # Now do it again, to make sure the result is equivalent
        plain_lines = sorted(vf.iter_lines_added_or_present_in_keys(keys))
        self.assertEqual([1, 1, 0], reload_counter) # No extra reloading
        self.assertEqual(plain_lines, reload_lines)
        self.assertEqual(21, len(plain_lines))
        # Now delete all pack files, and see that we raise the right error
        for trans, name in vf._access._indices.itervalues():
            trans.delete(name)
        self.assertListRaises(errors.NoSuchFile,
            vf.iter_lines_added_or_present_in_keys, keys)
        self.assertEqual([2, 1, 1], reload_counter)

    def test_get_record_stream_yields_disk_sorted_order(self):
        # if we get 'unordered' pick a semi-optimal order for reading. The
        # order should be grouped by pack file, and then by position in file
        repo = self.make_repository('test', format='pack-0.92')
        repo.lock_write()
        self.addCleanup(repo.unlock)
        repo.start_write_group()
        vf = repo.texts
        vf.add_lines(('f-id', 'rev-5'), [('f-id', 'rev-4')], ['lines\n'])
        vf.add_lines(('f-id', 'rev-1'), [], ['lines\n'])
        vf.add_lines(('f-id', 'rev-2'), [('f-id', 'rev-1')], ['lines\n'])
        repo.commit_write_group()
        # We inserted them as rev-5, rev-1, rev-2, we should get them back in
        # the same order
        stream = vf.get_record_stream([('f-id', 'rev-1'), ('f-id', 'rev-5'),
                                       ('f-id', 'rev-2')], 'unordered', False)
        keys = [r.key for r in stream]
        self.assertEqual([('f-id', 'rev-5'), ('f-id', 'rev-1'),
                          ('f-id', 'rev-2')], keys)
        repo.start_write_group()
        vf.add_lines(('f-id', 'rev-4'), [('f-id', 'rev-3')], ['lines\n'])
        vf.add_lines(('f-id', 'rev-3'), [('f-id', 'rev-2')], ['lines\n'])
        vf.add_lines(('f-id', 'rev-6'), [('f-id', 'rev-5')], ['lines\n'])
        repo.commit_write_group()
        # Request in random order, to make sure the output order isn't based on
        # the request
        request_keys = set(('f-id', 'rev-%d' % i) for i in range(1, 7))
        stream = vf.get_record_stream(request_keys, 'unordered', False)
        keys = [r.key for r in stream]
        # We want to get the keys back in disk order, but it doesn't matter
        # which pack we read from first. So this can come back in 2 orders
        alt1 = [('f-id', 'rev-%d' % i) for i in [4, 3, 6, 5, 1, 2]]
        alt2 = [('f-id', 'rev-%d' % i) for i in [5, 1, 2, 4, 3, 6]]
        if keys != alt1 and keys != alt2:
            self.fail('Returned key order did not match either expected order.'
                      ' expected %s or %s, not %s'
                      % (alt1, alt2, keys))


class LowLevelKnitDataTests(TestCase):

    def create_gz_content(self, text):
        sio = StringIO()
        gz_file = tuned_gzip.GzipFile(mode='wb', fileobj=sio)
        gz_file.write(text)
        gz_file.close()
        return sio.getvalue()

    def make_multiple_records(self):
        """Create the content for multiple records."""
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
        total_txt = []
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
                                        'foo\n'
                                        'bar\n'
                                        'end rev-id-1\n'
                                        % (sha1sum,))
        record_1 = (0, len(gz_txt), sha1sum)
        total_txt.append(gz_txt)
        sha1sum = osutils.sha('baz\n').hexdigest()
        gz_txt = self.create_gz_content('version rev-id-2 1 %s\n'
                                        'baz\n'
                                        'end rev-id-2\n'
                                        % (sha1sum,))
        record_2 = (record_1[1], len(gz_txt), sha1sum)
        total_txt.append(gz_txt)
        return total_txt, record_1, record_2

    def test_valid_knit_data(self):
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
                                        'foo\n'
                                        'bar\n'
                                        'end rev-id-1\n'
                                        % (sha1sum,))
        transport = MockTransport([gz_txt])
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
        knit = KnitVersionedFiles(None, access)
        records = [(('rev-id-1',), (('rev-id-1',), 0, len(gz_txt)))]

        contents = list(knit._read_records_iter(records))
        self.assertEqual([(('rev-id-1',), ['foo\n', 'bar\n'],
            '4e48e2c9a3d2ca8a708cb0cc545700544efb5021')], contents)

        raw_contents = list(knit._read_records_iter_raw(records))
        self.assertEqual([(('rev-id-1',), gz_txt, sha1sum)], raw_contents)

    def test_multiple_records_valid(self):
        total_txt, record_1, record_2 = self.make_multiple_records()
        transport = MockTransport([''.join(total_txt)])
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
        knit = KnitVersionedFiles(None, access)
        records = [(('rev-id-1',), (('rev-id-1',), record_1[0], record_1[1])),
                   (('rev-id-2',), (('rev-id-2',), record_2[0], record_2[1]))]

        contents = list(knit._read_records_iter(records))
        self.assertEqual([(('rev-id-1',), ['foo\n', 'bar\n'], record_1[2]),
                          (('rev-id-2',), ['baz\n'], record_2[2])],
                         contents)

        raw_contents = list(knit._read_records_iter_raw(records))
        self.assertEqual([(('rev-id-1',), total_txt[0], record_1[2]),
                          (('rev-id-2',), total_txt[1], record_2[2])],
                         raw_contents)

    def test_not_enough_lines(self):
        sha1sum = osutils.sha('foo\n').hexdigest()
        # record says 2 lines data says 1
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
                                        'foo\n'
                                        'end rev-id-1\n'
                                        % (sha1sum,))
        transport = MockTransport([gz_txt])
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
        knit = KnitVersionedFiles(None, access)
        records = [(('rev-id-1',), (('rev-id-1',), 0, len(gz_txt)))]
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter(records))

        # read_records_iter_raw won't detect that sort of mismatch/corruption
        raw_contents = list(knit._read_records_iter_raw(records))
        self.assertEqual([(('rev-id-1',),  gz_txt, sha1sum)], raw_contents)

    def test_too_many_lines(self):
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
        # record says 1 lines data says 2
        gz_txt = self.create_gz_content('version rev-id-1 1 %s\n'
                                        'foo\n'
                                        'bar\n'
                                        'end rev-id-1\n'
                                        % (sha1sum,))
        transport = MockTransport([gz_txt])
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
        knit = KnitVersionedFiles(None, access)
        records = [(('rev-id-1',), (('rev-id-1',), 0, len(gz_txt)))]
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter(records))

        # read_records_iter_raw won't detect that sort of mismatch/corruption
        raw_contents = list(knit._read_records_iter_raw(records))
        self.assertEqual([(('rev-id-1',), gz_txt, sha1sum)], raw_contents)

    def test_mismatched_version_id(self):
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
                                        'foo\n'
                                        'bar\n'
                                        'end rev-id-1\n'
                                        % (sha1sum,))
        transport = MockTransport([gz_txt])
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
        knit = KnitVersionedFiles(None, access)
        # We are asking for rev-id-2, but the data is rev-id-1
        records = [(('rev-id-2',), (('rev-id-2',), 0, len(gz_txt)))]
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter(records))

        # read_records_iter_raw detects mismatches in the header
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter_raw(records))

    def test_uncompressed_data(self):
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
        txt = ('version rev-id-1 2 %s\n'
               'foo\n'
               'bar\n'
               'end rev-id-1\n'
               % (sha1sum,))
        transport = MockTransport([txt])
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
        knit = KnitVersionedFiles(None, access)
        records = [(('rev-id-1',), (('rev-id-1',), 0, len(txt)))]

        # We don't have valid gzip data ==> corrupt
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter(records))

        # read_records_iter_raw will notice the bad data
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter_raw(records))

    def test_corrupted_data(self):
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
                                        'foo\n'
                                        'bar\n'
                                        'end rev-id-1\n'
                                        % (sha1sum,))
        # Change 2 bytes in the middle to \xff
        gz_txt = gz_txt[:10] + '\xff\xff' + gz_txt[12:]
        transport = MockTransport([gz_txt])
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
        knit = KnitVersionedFiles(None, access)
        records = [(('rev-id-1',), (('rev-id-1',), 0, len(gz_txt)))]
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter(records))
        # read_records_iter_raw will barf on bad gz data
        self.assertRaises(errors.KnitCorrupt, list,
            knit._read_records_iter_raw(records))


class LowLevelKnitIndexTests(TestCase):

    def get_knit_index(self, transport, name, mode):
        mapper = ConstantMapper(name)
        from bzrlib._knit_load_data_py import _load_data_py
        self.overrideAttr(knit, '_load_data', _load_data_py)
        allow_writes = lambda: 'w' in mode
        return _KndxIndex(transport, mapper, lambda:None, allow_writes, lambda:True)

    def test_create_file(self):
        transport = MockTransport()
        index = self.get_knit_index(transport, "filename", "w")
        index.keys()
        call = transport.calls.pop(0)
        # call[1][1] is a StringIO - we can't test it by simple equality.
        self.assertEqual('put_file_non_atomic', call[0])
        self.assertEqual('filename.kndx', call[1][0])
        # With no history, _KndxIndex writes a new index:
        self.assertEqual(_KndxIndex.HEADER,
            call[1][1].getvalue())
        self.assertEqual({'create_parent_dir': True}, call[2])

    def test_read_utf8_version_id(self):
        unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
        utf8_revision_id = unicode_revision_id.encode('utf-8')
        transport = MockTransport([
            _KndxIndex.HEADER,
            '%s option 0 1 :' % (utf8_revision_id,)
            ])
        index = self.get_knit_index(transport, "filename", "r")
        # _KndxIndex is a private class, and deals in utf8 revision_ids, not
        # Unicode revision_ids.
        self.assertEqual({(utf8_revision_id,):()},
            index.get_parent_map(index.keys()))
        self.assertFalse((unicode_revision_id,) in index.keys())

    def test_read_utf8_parents(self):
        unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
        utf8_revision_id = unicode_revision_id.encode('utf-8')
        transport = MockTransport([
            _KndxIndex.HEADER,
            "version option 0 1 .%s :" % (utf8_revision_id,)
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertEqual({("version",):((utf8_revision_id,),)},
            index.get_parent_map(index.keys()))

    def test_read_ignore_corrupted_lines(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "corrupted",
            "corrupted options 0 1 .b .c ",
            "version options 0 1 :"
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertEqual(1, len(index.keys()))
        self.assertEqual(set([("version",)]), index.keys())

    def test_read_corrupted_header(self):
        transport = MockTransport(['not a bzr knit index header\n'])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertRaises(KnitHeaderError, index.keys)

    def test_read_duplicate_entries(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "parent options 0 1 :",
            "version options1 0 1 0 :",
            "version options2 1 2 .other :",
            "version options3 3 4 0 .other :"
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertEqual(2, len(index.keys()))
        # check that the index used is the first one written. (Specific
        # to KnitIndex style indices.
        self.assertEqual("1", index._dictionary_compress([("version",)]))
        self.assertEqual((("version",), 3, 4), index.get_position(("version",)))
        self.assertEqual(["options3"], index.get_options(("version",)))
        self.assertEqual({("version",):(("parent",), ("other",))},
            index.get_parent_map([("version",)]))

    def test_read_compressed_parents(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 1 :",
            "b option 0 1 0 :",
            "c option 0 1 1 0 :",
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertEqual({("b",):(("a",),), ("c",):(("b",), ("a",))},
            index.get_parent_map([("b",), ("c",)]))

    def test_write_utf8_version_id(self):
        unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
        utf8_revision_id = unicode_revision_id.encode('utf-8')
        transport = MockTransport([
            _KndxIndex.HEADER
            ])
        index = self.get_knit_index(transport, "filename", "r")
        index.add_records([
            ((utf8_revision_id,), ["option"], ((utf8_revision_id,), 0, 1), [])])
        call = transport.calls.pop(0)
        # call[1][1] is a StringIO - we can't test it by simple equality.
        self.assertEqual('put_file_non_atomic', call[0])
        self.assertEqual('filename.kndx', call[1][0])
        # With no history, _KndxIndex writes a new index:
        self.assertEqual(_KndxIndex.HEADER +
            "\n%s option 0 1  :" % (utf8_revision_id,),
            call[1][1].getvalue())
        self.assertEqual({'create_parent_dir': True}, call[2])

    def test_write_utf8_parents(self):
        unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
        utf8_revision_id = unicode_revision_id.encode('utf-8')
        transport = MockTransport([
            _KndxIndex.HEADER
            ])
        index = self.get_knit_index(transport, "filename", "r")
        index.add_records([
            (("version",), ["option"], (("version",), 0, 1), [(utf8_revision_id,)])])
        call = transport.calls.pop(0)
        # call[1][1] is a StringIO - we can't test it by simple equality.
        self.assertEqual('put_file_non_atomic', call[0])
        self.assertEqual('filename.kndx', call[1][0])
        # With no history, _KndxIndex writes a new index:
        self.assertEqual(_KndxIndex.HEADER +
            "\nversion option 0 1 .%s :" % (utf8_revision_id,),
            call[1][1].getvalue())
        self.assertEqual({'create_parent_dir': True}, call[2])

    def test_keys(self):
        transport = MockTransport([
            _KndxIndex.HEADER
            ])
        index = self.get_knit_index(transport, "filename", "r")

        self.assertEqual(set(), index.keys())

        index.add_records([(("a",), ["option"], (("a",), 0, 1), [])])
        self.assertEqual(set([("a",)]), index.keys())

        index.add_records([(("a",), ["option"], (("a",), 0, 1), [])])
        self.assertEqual(set([("a",)]), index.keys())

        index.add_records([(("b",), ["option"], (("b",), 0, 1), [])])
        self.assertEqual(set([("a",), ("b",)]), index.keys())

    def add_a_b(self, index, random_id=None):
        kwargs = {}
        if random_id is not None:
            kwargs["random_id"] = random_id
        index.add_records([
            (("a",), ["option"], (("a",), 0, 1), [("b",)]),
            (("a",), ["opt"], (("a",), 1, 2), [("c",)]),
            (("b",), ["option"], (("b",), 2, 3), [("a",)])
            ], **kwargs)

    def assertIndexIsAB(self, index):
        self.assertEqual({
            ('a',): (('c',),),
            ('b',): (('a',),),
            },
            index.get_parent_map(index.keys()))
        self.assertEqual((("a",), 1, 2), index.get_position(("a",)))
        self.assertEqual((("b",), 2, 3), index.get_position(("b",)))
        self.assertEqual(["opt"], index.get_options(("a",)))

    def test_add_versions(self):
        transport = MockTransport([
            _KndxIndex.HEADER
            ])
        index = self.get_knit_index(transport, "filename", "r")

        self.add_a_b(index)
        call = transport.calls.pop(0)
        # call[1][1] is a StringIO - we can't test it by simple equality.
        self.assertEqual('put_file_non_atomic', call[0])
        self.assertEqual('filename.kndx', call[1][0])
        # With no history, _KndxIndex writes a new index:
        self.assertEqual(
            _KndxIndex.HEADER +
            "\na option 0 1 .b :"
            "\na opt 1 2 .c :"
            "\nb option 2 3 0 :",
            call[1][1].getvalue())
        self.assertEqual({'create_parent_dir': True}, call[2])
        self.assertIndexIsAB(index)

    def test_add_versions_random_id_is_accepted(self):
        transport = MockTransport([
            _KndxIndex.HEADER
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.add_a_b(index, random_id=True)

    def test_delay_create_and_add_versions(self):
        transport = MockTransport()

        index = self.get_knit_index(transport, "filename", "w")
        # dir_mode=0777)
        self.assertEqual([], transport.calls)
        self.add_a_b(index)
        #self.assertEqual(
        #[    {"dir_mode": 0777, "create_parent_dir": True, "mode": "wb"},
        #    kwargs)
        # Two calls: one during which we load the existing index (and when its
        # missing create it), then a second where we write the contents out.
        self.assertEqual(2, len(transport.calls))
        call = transport.calls.pop(0)
        self.assertEqual('put_file_non_atomic', call[0])
        self.assertEqual('filename.kndx', call[1][0])
        # With no history, _KndxIndex writes a new index:
        self.assertEqual(_KndxIndex.HEADER, call[1][1].getvalue())
        self.assertEqual({'create_parent_dir': True}, call[2])
        call = transport.calls.pop(0)
        # call[1][1] is a StringIO - we can't test it by simple equality.
        self.assertEqual('put_file_non_atomic', call[0])
        self.assertEqual('filename.kndx', call[1][0])
        # With no history, _KndxIndex writes a new index:
        self.assertEqual(
            _KndxIndex.HEADER +
            "\na option 0 1 .b :"
            "\na opt 1 2 .c :"
            "\nb option 2 3 0 :",
            call[1][1].getvalue())
        self.assertEqual({'create_parent_dir': True}, call[2])

    def assertTotalBuildSize(self, size, keys, positions):
        self.assertEqual(size,
                         knit._get_total_build_size(None, keys, positions))

    def test__get_total_build_size(self):
        positions = {
            ('a',): (('fulltext', False), (('a',), 0, 100), None),
            ('b',): (('line-delta', False), (('b',), 100, 21), ('a',)),
            ('c',): (('line-delta', False), (('c',), 121, 35), ('b',)),
            ('d',): (('line-delta', False), (('d',), 156, 12), ('b',)),
            }
        self.assertTotalBuildSize(100, [('a',)], positions)
        self.assertTotalBuildSize(121, [('b',)], positions)
        # c needs both a & b
        self.assertTotalBuildSize(156, [('c',)], positions)
        # we shouldn't count 'b' twice
        self.assertTotalBuildSize(156, [('b',), ('c',)], positions)
        self.assertTotalBuildSize(133, [('d',)], positions)
        self.assertTotalBuildSize(168, [('c',), ('d',)], positions)

    def test_get_position(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 1 :",
            "b option 1 2 :"
            ])
        index = self.get_knit_index(transport, "filename", "r")

        self.assertEqual((("a",), 0, 1), index.get_position(("a",)))
        self.assertEqual((("b",), 1, 2), index.get_position(("b",)))

    def test_get_method(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a fulltext,unknown 0 1 :",
            "b unknown,line-delta 1 2 :",
            "c bad 3 4 :"
            ])
        index = self.get_knit_index(transport, "filename", "r")

        self.assertEqual("fulltext", index.get_method("a"))
        self.assertEqual("line-delta", index.get_method("b"))
        self.assertRaises(errors.KnitIndexUnknownMethod, index.get_method, "c")

    def test_get_options(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a opt1 0 1 :",
            "b opt2,opt3 1 2 :"
            ])
        index = self.get_knit_index(transport, "filename", "r")

        self.assertEqual(["opt1"], index.get_options("a"))
        self.assertEqual(["opt2", "opt3"], index.get_options("b"))

    def test_get_parent_map(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 1 :",
            "b option 1 2 0 .c :",
            "c option 1 2 1 0 .e :"
            ])
        index = self.get_knit_index(transport, "filename", "r")

        self.assertEqual({
            ("a",):(),
            ("b",):(("a",), ("c",)),
            ("c",):(("b",), ("a",), ("e",)),
            }, index.get_parent_map(index.keys()))

    def test_impossible_parent(self):
        """Test we get KnitCorrupt if the parent couldn't possibly exist."""
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 1 :",
            "b option 0 1 4 :"  # We don't have a 4th record
            ])
        index = self.get_knit_index(transport, 'filename', 'r')
        try:
            self.assertRaises(errors.KnitCorrupt, index.keys)
        except TypeError, e:
            if (str(e) == ('exceptions must be strings, classes, or instances,'
                           ' not exceptions.IndexError')
                and sys.version_info[0:2] >= (2,5)):
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
                                  ' raising new style exceptions with python'
                                  ' >=2.5')
            else:
                raise

    def test_corrupted_parent(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 1 :",
            "b option 0 1 :",
            "c option 0 1 1v :", # Can't have a parent of '1v'
            ])
        index = self.get_knit_index(transport, 'filename', 'r')
        try:
            self.assertRaises(errors.KnitCorrupt, index.keys)
        except TypeError, e:
            if (str(e) == ('exceptions must be strings, classes, or instances,'
                           ' not exceptions.ValueError')
                and sys.version_info[0:2] >= (2,5)):
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
                                  ' raising new style exceptions with python'
                                  ' >=2.5')
            else:
                raise

    def test_corrupted_parent_in_list(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 1 :",
            "b option 0 1 :",
            "c option 0 1 1 v :", # Can't have a parent of 'v'
            ])
        index = self.get_knit_index(transport, 'filename', 'r')
        try:
            self.assertRaises(errors.KnitCorrupt, index.keys)
        except TypeError, e:
            if (str(e) == ('exceptions must be strings, classes, or instances,'
                           ' not exceptions.ValueError')
                and sys.version_info[0:2] >= (2,5)):
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
                                  ' raising new style exceptions with python'
                                  ' >=2.5')
            else:
                raise

    def test_invalid_position(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 1v 1 :",
            ])
        index = self.get_knit_index(transport, 'filename', 'r')
        try:
            self.assertRaises(errors.KnitCorrupt, index.keys)
        except TypeError, e:
            if (str(e) == ('exceptions must be strings, classes, or instances,'
                           ' not exceptions.ValueError')
                and sys.version_info[0:2] >= (2,5)):
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
                                  ' raising new style exceptions with python'
                                  ' >=2.5')
            else:
                raise

    def test_invalid_size(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 1 1v :",
            ])
        index = self.get_knit_index(transport, 'filename', 'r')
        try:
            self.assertRaises(errors.KnitCorrupt, index.keys)
        except TypeError, e:
            if (str(e) == ('exceptions must be strings, classes, or instances,'
                           ' not exceptions.ValueError')
                and sys.version_info[0:2] >= (2,5)):
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
                                  ' raising new style exceptions with python'
                                  ' >=2.5')
            else:
                raise

    def test_scan_unvalidated_index_not_implemented(self):
        transport = MockTransport()
        index = self.get_knit_index(transport, 'filename', 'r')
        self.assertRaises(
            NotImplementedError, index.scan_unvalidated_index,
            'dummy graph_index')
        self.assertRaises(
            NotImplementedError, index.get_missing_compression_parents)

    def test_short_line(self):
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 10  :",
            "b option 10 10 0", # This line isn't terminated, ignored
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertEqual(set([('a',)]), index.keys())

    def test_skip_incomplete_record(self):
        # A line with bogus data should just be skipped
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 10  :",
            "b option 10 10 0", # This line isn't terminated, ignored
            "c option 20 10 0 :", # Properly terminated, and starts with '\n'
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertEqual(set([('a',), ('c',)]), index.keys())

    def test_trailing_characters(self):
        # A line with bogus data should just be skipped
        transport = MockTransport([
            _KndxIndex.HEADER,
            "a option 0 10  :",
            "b option 10 10 0 :a", # This line has extra trailing characters
            "c option 20 10 0 :", # Properly terminated, and starts with '\n'
            ])
        index = self.get_knit_index(transport, "filename", "r")
        self.assertEqual(set([('a',), ('c',)]), index.keys())


class LowLevelKnitIndexTests_c(LowLevelKnitIndexTests):

    _test_needs_features = [compiled_knit_feature]

    def get_knit_index(self, transport, name, mode):
        mapper = ConstantMapper(name)
        from bzrlib._knit_load_data_pyx import _load_data_c
        self.overrideAttr(knit, '_load_data', _load_data_c)
        allow_writes = lambda: mode == 'w'
        return _KndxIndex(transport, mapper, lambda:None,
                          allow_writes, lambda:True)


class Test_KnitAnnotator(TestCaseWithMemoryTransport):

    def make_annotator(self):
        factory = knit.make_pack_factory(True, True, 1)
        vf = factory(self.get_transport())
        return knit._KnitAnnotator(vf)

    def test__expand_fulltext(self):
        ann = self.make_annotator()
        rev_key = ('rev-id',)
        ann._num_compression_children[rev_key] = 1
        res = ann._expand_record(rev_key, (('parent-id',),), None,
                           ['line1\n', 'line2\n'], ('fulltext', True))
        # The content object and text lines should be cached appropriately
        self.assertEqual(['line1\n', 'line2'], res)
        content_obj = ann._content_objects[rev_key]
        self.assertEqual(['line1\n', 'line2\n'], content_obj._lines)
        self.assertEqual(res, content_obj.text())
        self.assertEqual(res, ann._text_cache[rev_key])

    def test__expand_delta_comp_parent_not_available(self):
        # Parent isn't available yet, so we return nothing, but queue up this
        # node for later processing
        ann = self.make_annotator()
        rev_key = ('rev-id',)
        parent_key = ('parent-id',)
        record = ['0,1,1\n', 'new-line\n']
        details = ('line-delta', False)
        res = ann._expand_record(rev_key, (parent_key,), parent_key,
                                 record, details)
        self.assertEqual(None, res)
        self.assertTrue(parent_key in ann._pending_deltas)
        pending = ann._pending_deltas[parent_key]
        self.assertEqual(1, len(pending))
        self.assertEqual((rev_key, (parent_key,), record, details), pending[0])

    def test__expand_record_tracks_num_children(self):
        ann = self.make_annotator()
        rev_key = ('rev-id',)
        rev2_key = ('rev2-id',)
        parent_key = ('parent-id',)
        record = ['0,1,1\n', 'new-line\n']
        details = ('line-delta', False)
        ann._num_compression_children[parent_key] = 2
        ann._expand_record(parent_key, (), None, ['line1\n', 'line2\n'],
                           ('fulltext', False))
        res = ann._expand_record(rev_key, (parent_key,), parent_key,
                                 record, details)
        self.assertEqual({parent_key: 1}, ann._num_compression_children)
        # Expanding the second child should remove the content object, and the
        # num_compression_children entry
        res = ann._expand_record(rev2_key, (parent_key,), parent_key,
                                 record, details)
        self.assertFalse(parent_key in ann._content_objects)
        self.assertEqual({}, ann._num_compression_children)
        # We should not cache the content_objects for rev2 and rev, because
        # they do not have compression children of their own.
        self.assertEqual({}, ann._content_objects)

    def test__expand_delta_records_blocks(self):
        ann = self.make_annotator()
        rev_key = ('rev-id',)
        parent_key = ('parent-id',)
        record = ['0,1,1\n', 'new-line\n']
        details = ('line-delta', True)
        ann._num_compression_children[parent_key] = 2
        ann._expand_record(parent_key, (), None,
                           ['line1\n', 'line2\n', 'line3\n'],
                           ('fulltext', False))
        ann._expand_record(rev_key, (parent_key,), parent_key, record, details)
        self.assertEqual({(rev_key, parent_key): [(1, 1, 1), (3, 3, 0)]},
                         ann._matching_blocks)
        rev2_key = ('rev2-id',)
        record = ['0,1,1\n', 'new-line\n']
        details = ('line-delta', False)
        ann._expand_record(rev2_key, (parent_key,), parent_key, record, details)
        self.assertEqual([(1, 1, 2), (3, 3, 0)],
                         ann._matching_blocks[(rev2_key, parent_key)])

    def test__get_parent_ann_uses_matching_blocks(self):
        ann = self.make_annotator()
        rev_key = ('rev-id',)
        parent_key = ('parent-id',)
        parent_ann = [(parent_key,)]*3
        block_key = (rev_key, parent_key)
        ann._annotations_cache[parent_key] = parent_ann
        ann._matching_blocks[block_key] = [(0, 1, 1), (3, 3, 0)]
        # We should not try to access any parent_lines content, because we know
        # we already have the matching blocks
        par_ann, blocks = ann._get_parent_annotations_and_matches(rev_key,
                                        ['1\n', '2\n', '3\n'], parent_key)
        self.assertEqual(parent_ann, par_ann)
        self.assertEqual([(0, 1, 1), (3, 3, 0)], blocks)
        self.assertEqual({}, ann._matching_blocks)

    def test__process_pending(self):
        ann = self.make_annotator()
        rev_key = ('rev-id',)
        p1_key = ('p1-id',)
        p2_key = ('p2-id',)
        record = ['0,1,1\n', 'new-line\n']
        details = ('line-delta', False)
        p1_record = ['line1\n', 'line2\n']
        ann._num_compression_children[p1_key] = 1
        res = ann._expand_record(rev_key, (p1_key,p2_key), p1_key,
                                 record, details)
        self.assertEqual(None, res)
        # self.assertTrue(p1_key in ann._pending_deltas)
        self.assertEqual({}, ann._pending_annotation)
        # Now insert p1, and we should be able to expand the delta
        res = ann._expand_record(p1_key, (), None, p1_record,
                                 ('fulltext', False))
        self.assertEqual(p1_record, res)
        ann._annotations_cache[p1_key] = [(p1_key,)]*2
        res = ann._process_pending(p1_key)
        self.assertEqual([], res)
        self.assertFalse(p1_key in ann._pending_deltas)
        self.assertTrue(p2_key in ann._pending_annotation)
        self.assertEqual({p2_key: [(rev_key, (p1_key, p2_key))]},
                         ann._pending_annotation)
        # Now fill in parent 2, and pending annotation should be satisfied
        res = ann._expand_record(p2_key, (), None, [], ('fulltext', False))
        ann._annotations_cache[p2_key] = []
        res = ann._process_pending(p2_key)
        self.assertEqual([rev_key], res)
        self.assertEqual({}, ann._pending_annotation)
        self.assertEqual({}, ann._pending_deltas)

    def test_record_delta_removes_basis(self):
        ann = self.make_annotator()
        ann._expand_record(('parent-id',), (), None,
                           ['line1\n', 'line2\n'], ('fulltext', False))
        ann._num_compression_children['parent-id'] = 2

    def test_annotate_special_text(self):
        ann = self.make_annotator()
        vf = ann._vf
        rev1_key = ('rev-1',)
        rev2_key = ('rev-2',)
        rev3_key = ('rev-3',)
        spec_key = ('special:',)
        vf.add_lines(rev1_key, [], ['initial content\n'])
        vf.add_lines(rev2_key, [rev1_key], ['initial content\n',
                                            'common content\n',
                                            'content in 2\n'])
        vf.add_lines(rev3_key, [rev1_key], ['initial content\n',
                                            'common content\n',
                                            'content in 3\n'])
        spec_text = ('initial content\n'
                     'common content\n'
                     'content in 2\n'
                     'content in 3\n')
        ann.add_special_text(spec_key, [rev2_key, rev3_key], spec_text)
        anns, lines = ann.annotate(spec_key)
        self.assertEqual([(rev1_key,),
                          (rev2_key, rev3_key),
                          (rev2_key,),
                          (rev3_key,),
                         ], anns)
        self.assertEqualDiff(spec_text, ''.join(lines))


class KnitTests(TestCaseWithTransport):
    """Class containing knit test helper routines."""

    def make_test_knit(self, annotate=False, name='test'):
        mapper = ConstantMapper(name)
        return make_file_factory(annotate, mapper)(self.get_transport())


class TestBadShaError(KnitTests):
    """Tests for handling of sha errors."""

    def test_sha_exception_has_text(self):
        # having the failed text included in the error allows for recovery.
        source = self.make_test_knit()
        target = self.make_test_knit(name="target")
        if not source._max_delta_chain:
            raise TestNotApplicable(
                "cannot get delta-caused sha failures without deltas.")
        # create a basis
        basis = ('basis',)
        broken = ('broken',)
        source.add_lines(basis, (), ['foo\n'])
        source.add_lines(broken, (basis,), ['foo\n', 'bar\n'])
        # Seed target with a bad basis text
        target.add_lines(basis, (), ['gam\n'])
        target.insert_record_stream(
            source.get_record_stream([broken], 'unordered', False))
        err = self.assertRaises(errors.KnitCorrupt,
            target.get_record_stream([broken], 'unordered', True
            ).next().get_bytes_as, 'chunked')
        self.assertEqual(['gam\n', 'bar\n'], err.content)
        # Test for formatting with live data
        self.assertStartsWith(str(err), "Knit ")


class TestKnitIndex(KnitTests):

    def test_add_versions_dictionary_compresses(self):
        """Adding versions to the index should update the lookup dict"""
        knit = self.make_test_knit()
        idx = knit._index
        idx.add_records([(('a-1',), ['fulltext'], (('a-1',), 0, 0), [])])
        self.check_file_contents('test.kndx',
            '# bzr knit index 8\n'
            '\n'
            'a-1 fulltext 0 0  :'
            )
        idx.add_records([
            (('a-2',), ['fulltext'], (('a-2',), 0, 0), [('a-1',)]),
            (('a-3',), ['fulltext'], (('a-3',), 0, 0), [('a-2',)]),
            ])
        self.check_file_contents('test.kndx',
            '# bzr knit index 8\n'
            '\n'
            'a-1 fulltext 0 0  :\n'
            'a-2 fulltext 0 0 0 :\n'
            'a-3 fulltext 0 0 1 :'
            )
        self.assertEqual(set([('a-3',), ('a-1',), ('a-2',)]), idx.keys())
        self.assertEqual({
            ('a-1',): ((('a-1',), 0, 0), None, (), ('fulltext', False)),
            ('a-2',): ((('a-2',), 0, 0), None, (('a-1',),), ('fulltext', False)),
            ('a-3',): ((('a-3',), 0, 0), None, (('a-2',),), ('fulltext', False)),
            }, idx.get_build_details(idx.keys()))
        self.assertEqual({('a-1',):(),
            ('a-2',):(('a-1',),),
            ('a-3',):(('a-2',),),},
            idx.get_parent_map(idx.keys()))

    def test_add_versions_fails_clean(self):
        """If add_versions fails in the middle, it restores a pristine state.

        Any modifications that are made to the index are reset if all versions
        cannot be added.
        """
        # This cheats a little bit by passing in a generator which will
        # raise an exception before the processing finishes
        # Other possibilities would be to have an version with the wrong number
        # of entries, or to make the backing transport unable to write any
        # files.

        knit = self.make_test_knit()
        idx = knit._index
        idx.add_records([(('a-1',), ['fulltext'], (('a-1',), 0, 0), [])])

        class StopEarly(Exception):
            pass

        def generate_failure():
            """Add some entries and then raise an exception"""
            yield (('a-2',), ['fulltext'], (None, 0, 0), ('a-1',))
            yield (('a-3',), ['fulltext'], (None, 0, 0), ('a-2',))
            raise StopEarly()

        # Assert the pre-condition
        def assertA1Only():
            self.assertEqual(set([('a-1',)]), set(idx.keys()))
            self.assertEqual(
                {('a-1',): ((('a-1',), 0, 0), None, (), ('fulltext', False))},
                idx.get_build_details([('a-1',)]))
            self.assertEqual({('a-1',):()}, idx.get_parent_map(idx.keys()))

        assertA1Only()
        self.assertRaises(StopEarly, idx.add_records, generate_failure())
        # And it shouldn't be modified
        assertA1Only()

    def test_knit_index_ignores_empty_files(self):
        # There was a race condition in older bzr, where a ^C at the right time
        # could leave an empty .kndx file, which bzr would later claim was a
        # corrupted file since the header was not present. In reality, the file
        # just wasn't created, so it should be ignored.
        t = transport.get_transport('.')
        t.put_bytes('test.kndx', '')

        knit = self.make_test_knit()

    def test_knit_index_checks_header(self):
        t = transport.get_transport('.')
        t.put_bytes('test.kndx', '# not really a knit header\n\n')
        k = self.make_test_knit()
        self.assertRaises(KnitHeaderError, k.keys)


class TestGraphIndexKnit(KnitTests):
    """Tests for knits using a GraphIndex rather than a KnitIndex."""

    def make_g_index(self, name, ref_lists=0, nodes=[]):
        builder = GraphIndexBuilder(ref_lists)
        for node, references, value in nodes:
            builder.add_node(node, references, value)
        stream = builder.finish()
        trans = self.get_transport()
        size = trans.put_file(name, stream)
        return GraphIndex(trans, name, size)

    def two_graph_index(self, deltas=False, catch_adds=False):
        """Build a two-graph index.

        :param deltas: If true, use underlying indices with two node-ref
            lists and 'parent' set to a delta-compressed against tail.
        """
        # build a complex graph across several indices.
        if deltas:
            # delta compression inn the index
            index1 = self.make_g_index('1', 2, [
                (('tip', ), 'N0 100', ([('parent', )], [], )),
                (('tail', ), '', ([], []))])
            index2 = self.make_g_index('2', 2, [
                (('parent', ), ' 100 78', ([('tail', ), ('ghost', )], [('tail', )])),
                (('separate', ), '', ([], []))])
        else:
            # just blob location and graph in the index.
            index1 = self.make_g_index('1', 1, [
                (('tip', ), 'N0 100', ([('parent', )], )),
                (('tail', ), '', ([], ))])
            index2 = self.make_g_index('2', 1, [
                (('parent', ), ' 100 78', ([('tail', ), ('ghost', )], )),
                (('separate', ), '', ([], ))])
        combined_index = CombinedGraphIndex([index1, index2])
        if catch_adds:
            self.combined_index = combined_index
            self.caught_entries = []
            add_callback = self.catch_add
        else:
            add_callback = None
        return _KnitGraphIndex(combined_index, lambda:True, deltas=deltas,
            add_callback=add_callback)

    def test_keys(self):
        index = self.two_graph_index()
        self.assertEqual(set([('tail',), ('tip',), ('parent',), ('separate',)]),
            set(index.keys()))

    def test_get_position(self):
        index = self.two_graph_index()
        self.assertEqual((index._graph_index._indices[0], 0, 100), index.get_position(('tip',)))
        self.assertEqual((index._graph_index._indices[1], 100, 78), index.get_position(('parent',)))

    def test_get_method_deltas(self):
        index = self.two_graph_index(deltas=True)
        self.assertEqual('fulltext', index.get_method(('tip',)))
        self.assertEqual('line-delta', index.get_method(('parent',)))

    def test_get_method_no_deltas(self):
        # check that the parent-history lookup is ignored with deltas=False.
        index = self.two_graph_index(deltas=False)
        self.assertEqual('fulltext', index.get_method(('tip',)))
        self.assertEqual('fulltext', index.get_method(('parent',)))

    def test_get_options_deltas(self):
        index = self.two_graph_index(deltas=True)
        self.assertEqual(['fulltext', 'no-eol'], index.get_options(('tip',)))
        self.assertEqual(['line-delta'], index.get_options(('parent',)))

    def test_get_options_no_deltas(self):
        # check that the parent-history lookup is ignored with deltas=False.
        index = self.two_graph_index(deltas=False)
        self.assertEqual(['fulltext', 'no-eol'], index.get_options(('tip',)))
        self.assertEqual(['fulltext'], index.get_options(('parent',)))

    def test_get_parent_map(self):
        index = self.two_graph_index()
        self.assertEqual({('parent',):(('tail',), ('ghost',))},
            index.get_parent_map([('parent',), ('ghost',)]))

    def catch_add(self, entries):
        self.caught_entries.append(entries)

    def test_add_no_callback_errors(self):
        index = self.two_graph_index()
        self.assertRaises(errors.ReadOnlyError, index.add_records,
            [(('new',), 'fulltext,no-eol', (None, 50, 60), ['separate'])])

    def test_add_version_smoke(self):
        index = self.two_graph_index(catch_adds=True)
        index.add_records([(('new',), 'fulltext,no-eol', (None, 50, 60),
            [('separate',)])])
        self.assertEqual([[(('new', ), 'N50 60', ((('separate',),),))]],
            self.caught_entries)

    def test_add_version_delta_not_delta_index(self):
        index = self.two_graph_index(catch_adds=True)
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('new',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
        self.assertEqual([], self.caught_entries)

    def test_add_version_same_dup(self):
        index = self.two_graph_index(catch_adds=True)
        # options can be spelt two different ways
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)])])
        index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100), [('parent',)])])
        # position/length are ignored (because each pack could have fulltext or
        # delta, and be at a different position.
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100),
            [('parent',)])])
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000),
            [('parent',)])])
        # but neither should have added data:
        self.assertEqual([[], [], [], []], self.caught_entries)

    def test_add_version_different_dup(self):
        index = self.two_graph_index(deltas=True, catch_adds=True)
        # change options
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext', (None, 0, 100), [('parent',)])])
        # parents
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
        self.assertEqual([], self.caught_entries)

    def test_add_versions_nodeltas(self):
        index = self.two_graph_index(catch_adds=True)
        index.add_records([
                (('new',), 'fulltext,no-eol', (None, 50, 60), [('separate',)]),
                (('new2',), 'fulltext', (None, 0, 6), [('new',)]),
                ])
        self.assertEqual([(('new', ), 'N50 60', ((('separate',),),)),
            (('new2', ), ' 0 6', ((('new',),),))],
            sorted(self.caught_entries[0]))
        self.assertEqual(1, len(self.caught_entries))

    def test_add_versions_deltas(self):
        index = self.two_graph_index(deltas=True, catch_adds=True)
        index.add_records([
                (('new',), 'fulltext,no-eol', (None, 50, 60), [('separate',)]),
                (('new2',), 'line-delta', (None, 0, 6), [('new',)]),
                ])
        self.assertEqual([(('new', ), 'N50 60', ((('separate',),), ())),
            (('new2', ), ' 0 6', ((('new',),), (('new',),), ))],
            sorted(self.caught_entries[0]))
        self.assertEqual(1, len(self.caught_entries))

    def test_add_versions_delta_not_delta_index(self):
        index = self.two_graph_index(catch_adds=True)
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('new',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
        self.assertEqual([], self.caught_entries)

    def test_add_versions_random_id_accepted(self):
        index = self.two_graph_index(catch_adds=True)
        index.add_records([], random_id=True)

    def test_add_versions_same_dup(self):
        index = self.two_graph_index(catch_adds=True)
        # options can be spelt two different ways
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100),
            [('parent',)])])
        index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100),
            [('parent',)])])
        # position/length are ignored (because each pack could have fulltext or
        # delta, and be at a different position.
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100),
            [('parent',)])])
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000),
            [('parent',)])])
        # but neither should have added data.
        self.assertEqual([[], [], [], []], self.caught_entries)

    def test_add_versions_different_dup(self):
        index = self.two_graph_index(deltas=True, catch_adds=True)
        # change options
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext', (None, 0, 100), [('parent',)])])
        # parents
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
        # change options in the second record
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)]),
             (('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
        self.assertEqual([], self.caught_entries)

    def make_g_index_missing_compression_parent(self):
        graph_index = self.make_g_index('missing_comp', 2,
            [(('tip', ), ' 100 78',
              ([('missing-parent', ), ('ghost', )], [('missing-parent', )]))])
        return graph_index

    def make_g_index_missing_parent(self):
        graph_index = self.make_g_index('missing_parent', 2,
            [(('parent', ), ' 100 78', ([], [])),
             (('tip', ), ' 100 78',
              ([('parent', ), ('missing-parent', )], [('parent', )])),
              ])
        return graph_index

    def make_g_index_no_external_refs(self):
        graph_index = self.make_g_index('no_external_refs', 2,
            [(('rev', ), ' 100 78',
              ([('parent', ), ('ghost', )], []))])
        return graph_index

    def test_add_good_unvalidated_index(self):
        unvalidated = self.make_g_index_no_external_refs()
        combined = CombinedGraphIndex([unvalidated])
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
        index.scan_unvalidated_index(unvalidated)
        self.assertEqual(frozenset(), index.get_missing_compression_parents())

    def test_add_missing_compression_parent_unvalidated_index(self):
        unvalidated = self.make_g_index_missing_compression_parent()
        combined = CombinedGraphIndex([unvalidated])
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
        index.scan_unvalidated_index(unvalidated)
        # This also checks that its only the compression parent that is
        # examined, otherwise 'ghost' would also be reported as a missing
        # parent.
        self.assertEqual(
            frozenset([('missing-parent',)]),
            index.get_missing_compression_parents())

    def test_add_missing_noncompression_parent_unvalidated_index(self):
        unvalidated = self.make_g_index_missing_parent()
        combined = CombinedGraphIndex([unvalidated])
        index = _KnitGraphIndex(combined, lambda: True, deltas=True,
            track_external_parent_refs=True)
        index.scan_unvalidated_index(unvalidated)
        self.assertEqual(
            frozenset([('missing-parent',)]), index.get_missing_parents())

    def test_track_external_parent_refs(self):
        g_index = self.make_g_index('empty', 2, [])
        combined = CombinedGraphIndex([g_index])
        index = _KnitGraphIndex(combined, lambda: True, deltas=True,
            add_callback=self.catch_add, track_external_parent_refs=True)
        self.caught_entries = []
        index.add_records([
            (('new-key',), 'fulltext,no-eol', (None, 50, 60),
             [('parent-1',), ('parent-2',)])])
        self.assertEqual(
            frozenset([('parent-1',), ('parent-2',)]),
            index.get_missing_parents())

    def test_add_unvalidated_index_with_present_external_references(self):
        index = self.two_graph_index(deltas=True)
        # Ugly hack to get at one of the underlying GraphIndex objects that
        # two_graph_index built.
        unvalidated = index._graph_index._indices[1]
        # 'parent' is an external ref of _indices[1] (unvalidated), but is
        # present in _indices[0].
        index.scan_unvalidated_index(unvalidated)
        self.assertEqual(frozenset(), index.get_missing_compression_parents())

    def make_new_missing_parent_g_index(self, name):
        missing_parent = name + '-missing-parent'
        graph_index = self.make_g_index(name, 2,
            [((name + 'tip', ), ' 100 78',
              ([(missing_parent, ), ('ghost', )], [(missing_parent, )]))])
        return graph_index

    def test_add_mulitiple_unvalidated_indices_with_missing_parents(self):
        g_index_1 = self.make_new_missing_parent_g_index('one')
        g_index_2 = self.make_new_missing_parent_g_index('two')
        combined = CombinedGraphIndex([g_index_1, g_index_2])
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
        index.scan_unvalidated_index(g_index_1)
        index.scan_unvalidated_index(g_index_2)
        self.assertEqual(
            frozenset([('one-missing-parent',), ('two-missing-parent',)]),
            index.get_missing_compression_parents())

    def test_add_mulitiple_unvalidated_indices_with_mutual_dependencies(self):
        graph_index_a = self.make_g_index('one', 2,
            [(('parent-one', ), ' 100 78', ([('non-compression-parent',)], [])),
             (('child-of-two', ), ' 100 78',
              ([('parent-two',)], [('parent-two',)]))])
        graph_index_b = self.make_g_index('two', 2,
            [(('parent-two', ), ' 100 78', ([('non-compression-parent',)], [])),
             (('child-of-one', ), ' 100 78',
              ([('parent-one',)], [('parent-one',)]))])
        combined = CombinedGraphIndex([graph_index_a, graph_index_b])
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
        index.scan_unvalidated_index(graph_index_a)
        index.scan_unvalidated_index(graph_index_b)
        self.assertEqual(
            frozenset([]), index.get_missing_compression_parents())


class TestNoParentsGraphIndexKnit(KnitTests):
    """Tests for knits using _KnitGraphIndex with no parents."""

    def make_g_index(self, name, ref_lists=0, nodes=[]):
        builder = GraphIndexBuilder(ref_lists)
        for node, references in nodes:
            builder.add_node(node, references)
        stream = builder.finish()
        trans = self.get_transport()
        size = trans.put_file(name, stream)
        return GraphIndex(trans, name, size)

    def test_add_good_unvalidated_index(self):
        unvalidated = self.make_g_index('unvalidated')
        combined = CombinedGraphIndex([unvalidated])
        index = _KnitGraphIndex(combined, lambda: True, parents=False)
        index.scan_unvalidated_index(unvalidated)
        self.assertEqual(frozenset(),
            index.get_missing_compression_parents())

    def test_parents_deltas_incompatible(self):
        index = CombinedGraphIndex([])
        self.assertRaises(errors.KnitError, _KnitGraphIndex, lambda:True,
            index, deltas=True, parents=False)

    def two_graph_index(self, catch_adds=False):
        """Build a two-graph index.

        :param deltas: If true, use underlying indices with two node-ref
            lists and 'parent' set to a delta-compressed against tail.
        """
        # put several versions in the index.
        index1 = self.make_g_index('1', 0, [
            (('tip', ), 'N0 100'),
            (('tail', ), '')])
        index2 = self.make_g_index('2', 0, [
            (('parent', ), ' 100 78'),
            (('separate', ), '')])
        combined_index = CombinedGraphIndex([index1, index2])
        if catch_adds:
            self.combined_index = combined_index
            self.caught_entries = []
            add_callback = self.catch_add
        else:
            add_callback = None
        return _KnitGraphIndex(combined_index, lambda:True, parents=False,
            add_callback=add_callback)

    def test_keys(self):
        index = self.two_graph_index()
        self.assertEqual(set([('tail',), ('tip',), ('parent',), ('separate',)]),
            set(index.keys()))

    def test_get_position(self):
        index = self.two_graph_index()
        self.assertEqual((index._graph_index._indices[0], 0, 100),
            index.get_position(('tip',)))
        self.assertEqual((index._graph_index._indices[1], 100, 78),
            index.get_position(('parent',)))

    def test_get_method(self):
        index = self.two_graph_index()
        self.assertEqual('fulltext', index.get_method(('tip',)))
        self.assertEqual(['fulltext'], index.get_options(('parent',)))

    def test_get_options(self):
        index = self.two_graph_index()
        self.assertEqual(['fulltext', 'no-eol'], index.get_options(('tip',)))
        self.assertEqual(['fulltext'], index.get_options(('parent',)))

    def test_get_parent_map(self):
        index = self.two_graph_index()
        self.assertEqual({('parent',):None},
            index.get_parent_map([('parent',), ('ghost',)]))

    def catch_add(self, entries):
        self.caught_entries.append(entries)

    def test_add_no_callback_errors(self):
        index = self.two_graph_index()
        self.assertRaises(errors.ReadOnlyError, index.add_records,
            [(('new',), 'fulltext,no-eol', (None, 50, 60), [('separate',)])])

    def test_add_version_smoke(self):
        index = self.two_graph_index(catch_adds=True)
        index.add_records([(('new',), 'fulltext,no-eol', (None, 50, 60), [])])
        self.assertEqual([[(('new', ), 'N50 60')]],
            self.caught_entries)

    def test_add_version_delta_not_delta_index(self):
        index = self.two_graph_index(catch_adds=True)
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('new',), 'no-eol,line-delta', (None, 0, 100), [])])
        self.assertEqual([], self.caught_entries)

    def test_add_version_same_dup(self):
        index = self.two_graph_index(catch_adds=True)
        # options can be spelt two different ways
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
        index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100), [])])
        # position/length are ignored (because each pack could have fulltext or
        # delta, and be at a different position.
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100), [])])
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000), [])])
        # but neither should have added data.
        self.assertEqual([[], [], [], []], self.caught_entries)

    def test_add_version_different_dup(self):
        index = self.two_graph_index(catch_adds=True)
        # change options
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'no-eol,line-delta', (None, 0, 100), [])])
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'line-delta,no-eol', (None, 0, 100), [])])
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext', (None, 0, 100), [])])
        # parents
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)])])
        self.assertEqual([], self.caught_entries)

    def test_add_versions(self):
        index = self.two_graph_index(catch_adds=True)
        index.add_records([
                (('new',), 'fulltext,no-eol', (None, 50, 60), []),
                (('new2',), 'fulltext', (None, 0, 6), []),
                ])
        self.assertEqual([(('new', ), 'N50 60'), (('new2', ), ' 0 6')],
            sorted(self.caught_entries[0]))
        self.assertEqual(1, len(self.caught_entries))

    def test_add_versions_delta_not_delta_index(self):
        index = self.two_graph_index(catch_adds=True)
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('new',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
        self.assertEqual([], self.caught_entries)

    def test_add_versions_parents_not_parents_index(self):
        index = self.two_graph_index(catch_adds=True)
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('new',), 'no-eol,fulltext', (None, 0, 100), [('parent',)])])
        self.assertEqual([], self.caught_entries)

    def test_add_versions_random_id_accepted(self):
        index = self.two_graph_index(catch_adds=True)
        index.add_records([], random_id=True)

    def test_add_versions_same_dup(self):
        index = self.two_graph_index(catch_adds=True)
        # options can be spelt two different ways
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
        index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100), [])])
        # position/length are ignored (because each pack could have fulltext or
        # delta, and be at a different position.
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100), [])])
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000), [])])
        # but neither should have added data.
        self.assertEqual([[], [], [], []], self.caught_entries)

    def test_add_versions_different_dup(self):
        index = self.two_graph_index(catch_adds=True)
        # change options
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'no-eol,line-delta', (None, 0, 100), [])])
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'line-delta,no-eol', (None, 0, 100), [])])
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext', (None, 0, 100), [])])
        # parents
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)])])
        # change options in the second record
        self.assertRaises(errors.KnitCorrupt, index.add_records,
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), []),
             (('tip',), 'no-eol,line-delta', (None, 0, 100), [])])
        self.assertEqual([], self.caught_entries)


class TestKnitVersionedFiles(KnitTests):

    def assertGroupKeysForIo(self, exp_groups, keys, non_local_keys,
                             positions, _min_buffer_size=None):
        kvf = self.make_test_knit()
        if _min_buffer_size is None:
            _min_buffer_size = knit._STREAM_MIN_BUFFER_SIZE
        self.assertEqual(exp_groups, kvf._group_keys_for_io(keys,
                                        non_local_keys, positions,
                                        _min_buffer_size=_min_buffer_size))

    def assertSplitByPrefix(self, expected_map, expected_prefix_order,
                            keys):
        split, prefix_order = KnitVersionedFiles._split_by_prefix(keys)
        self.assertEqual(expected_map, split)
        self.assertEqual(expected_prefix_order, prefix_order)

    def test__group_keys_for_io(self):
        ft_detail = ('fulltext', False)
        ld_detail = ('line-delta', False)
        f_a = ('f', 'a')
        f_b = ('f', 'b')
        f_c = ('f', 'c')
        g_a = ('g', 'a')
        g_b = ('g', 'b')
        g_c = ('g', 'c')
        positions = {
            f_a: (ft_detail, (f_a, 0, 100), None),
            f_b: (ld_detail, (f_b, 100, 21), f_a),
            f_c: (ld_detail, (f_c, 180, 15), f_b),
            g_a: (ft_detail, (g_a, 121, 35), None),
            g_b: (ld_detail, (g_b, 156, 12), g_a),
            g_c: (ld_detail, (g_c, 195, 13), g_a),
            }
        self.assertGroupKeysForIo([([f_a], set())],
                                  [f_a], [], positions)
        self.assertGroupKeysForIo([([f_a], set([f_a]))],
                                  [f_a], [f_a], positions)
        self.assertGroupKeysForIo([([f_a, f_b], set([]))],
                                  [f_a, f_b], [], positions)
        self.assertGroupKeysForIo([([f_a, f_b], set([f_b]))],
                                  [f_a, f_b], [f_b], positions)
        self.assertGroupKeysForIo([([f_a, f_b, g_a, g_b], set())],
                                  [f_a, g_a, f_b, g_b], [], positions)
        self.assertGroupKeysForIo([([f_a, f_b, g_a, g_b], set())],
                                  [f_a, g_a, f_b, g_b], [], positions,
                                  _min_buffer_size=150)
        self.assertGroupKeysForIo([([f_a, f_b], set()), ([g_a, g_b], set())],
                                  [f_a, g_a, f_b, g_b], [], positions,
                                  _min_buffer_size=100)
        self.assertGroupKeysForIo([([f_c], set()), ([g_b], set())],
                                  [f_c, g_b], [], positions,
                                  _min_buffer_size=125)
        self.assertGroupKeysForIo([([g_b, f_c], set())],
                                  [g_b, f_c], [], positions,
                                  _min_buffer_size=125)

    def test__split_by_prefix(self):
        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
                                  'g': [('g', 'b'), ('g', 'a')],
                                 }, ['f', 'g'],
                                 [('f', 'a'), ('g', 'b'),
                                  ('g', 'a'), ('f', 'b')])

        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
                                  'g': [('g', 'b'), ('g', 'a')],
                                 }, ['f', 'g'],
                                 [('f', 'a'), ('f', 'b'),
                                  ('g', 'b'), ('g', 'a')])

        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
                                  'g': [('g', 'b'), ('g', 'a')],
                                 }, ['f', 'g'],
                                 [('f', 'a'), ('f', 'b'),
                                  ('g', 'b'), ('g', 'a')])

        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
                                  'g': [('g', 'b'), ('g', 'a')],
                                  '': [('a',), ('b',)]
                                 }, ['f', 'g', ''],
                                 [('f', 'a'), ('g', 'b'),
                                  ('a',), ('b',),
                                  ('g', 'a'), ('f', 'b')])


class TestStacking(KnitTests):

    def get_basis_and_test_knit(self):
        basis = self.make_test_knit(name='basis')
        basis = RecordingVersionedFilesDecorator(basis)
        test = self.make_test_knit(name='test')
        test.add_fallback_versioned_files(basis)
        return basis, test

    def test_add_fallback_versioned_files(self):
        basis = self.make_test_knit(name='basis')
        test = self.make_test_knit(name='test')
        # It must not error; other tests test that the fallback is referred to
        # when accessing data.
        test.add_fallback_versioned_files(basis)

    def test_add_lines(self):
        # lines added to the test are not added to the basis
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_cross_border = ('quux',)
        key_delta = ('zaphod',)
        test.add_lines(key, (), ['foo\n'])
        self.assertEqual({}, basis.get_parent_map([key]))
        # lines added to the test that reference across the stack do a
        # fulltext.
        basis.add_lines(key_basis, (), ['foo\n'])
        basis.calls = []
        test.add_lines(key_cross_border, (key_basis,), ['foo\n'])
        self.assertEqual('fulltext', test._index.get_method(key_cross_border))
        # we don't even need to look at the basis to see that this should be
        # stored as a fulltext
        self.assertEqual([], basis.calls)
        # Subsequent adds do delta.
        basis.calls = []
        test.add_lines(key_delta, (key_cross_border,), ['foo\n'])
        self.assertEqual('line-delta', test._index.get_method(key_delta))
        self.assertEqual([], basis.calls)

    def test_annotate(self):
        # annotations from the test knit are answered without asking the basis
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_missing = ('missing',)
        test.add_lines(key, (), ['foo\n'])
        details = test.annotate(key)
        self.assertEqual([(key, 'foo\n')], details)
        self.assertEqual([], basis.calls)
        # But texts that are not in the test knit are looked for in the basis
        # directly.
        basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
        basis.calls = []
        details = test.annotate(key_basis)
        self.assertEqual([(key_basis, 'foo\n'), (key_basis, 'bar\n')], details)
        # Not optimised to date:
        # self.assertEqual([("annotate", key_basis)], basis.calls)
        self.assertEqual([('get_parent_map', set([key_basis])),
            ('get_parent_map', set([key_basis])),
            ('get_record_stream', [key_basis], 'topological', True)],
            basis.calls)

    def test_check(self):
        # At the moment checking a stacked knit does implicitly check the
        # fallback files.
        basis, test = self.get_basis_and_test_knit()
        test.check()

    def test_get_parent_map(self):
        # parents in the test knit are answered without asking the basis
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_missing = ('missing',)
        test.add_lines(key, (), [])
        parent_map = test.get_parent_map([key])
        self.assertEqual({key: ()}, parent_map)
        self.assertEqual([], basis.calls)
        # But parents that are not in the test knit are looked for in the basis
        basis.add_lines(key_basis, (), [])
        basis.calls = []
        parent_map = test.get_parent_map([key, key_basis, key_missing])
        self.assertEqual({key: (),
            key_basis: ()}, parent_map)
        self.assertEqual([("get_parent_map", set([key_basis, key_missing]))],
            basis.calls)

    def test_get_record_stream_unordered_fulltexts(self):
        # records from the test knit are answered without asking the basis:
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_missing = ('missing',)
        test.add_lines(key, (), ['foo\n'])
        records = list(test.get_record_stream([key], 'unordered', True))
        self.assertEqual(1, len(records))
        self.assertEqual([], basis.calls)
        # Missing (from test knit) objects are retrieved from the basis:
        basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
        basis.calls = []
        records = list(test.get_record_stream([key_basis, key_missing],
            'unordered', True))
        self.assertEqual(2, len(records))
        calls = list(basis.calls)
        for record in records:
            self.assertSubset([record.key], (key_basis, key_missing))
            if record.key == key_missing:
                self.assertIsInstance(record, AbsentContentFactory)
            else:
                reference = list(basis.get_record_stream([key_basis],
                    'unordered', True))[0]
                self.assertEqual(reference.key, record.key)
                self.assertEqual(reference.sha1, record.sha1)
                self.assertEqual(reference.storage_kind, record.storage_kind)
                self.assertEqual(reference.get_bytes_as(reference.storage_kind),
                    record.get_bytes_as(record.storage_kind))
                self.assertEqual(reference.get_bytes_as('fulltext'),
                    record.get_bytes_as('fulltext'))
        # It's not strictly minimal, but it seems reasonable for now for it to
        # ask which fallbacks have which parents.
        self.assertEqual([
            ("get_parent_map", set([key_basis, key_missing])),
            ("get_record_stream", [key_basis], 'unordered', True)],
            calls)

    def test_get_record_stream_ordered_fulltexts(self):
        # ordering is preserved down into the fallback store.
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_basis_2 = ('quux',)
        key_missing = ('missing',)
        test.add_lines(key, (key_basis,), ['foo\n'])
        # Missing (from test knit) objects are retrieved from the basis:
        basis.add_lines(key_basis, (key_basis_2,), ['foo\n', 'bar\n'])
        basis.add_lines(key_basis_2, (), ['quux\n'])
        basis.calls = []
        # ask for in non-topological order
        records = list(test.get_record_stream(
            [key, key_basis, key_missing, key_basis_2], 'topological', True))
        self.assertEqual(4, len(records))
        results = []
        for record in records:
            self.assertSubset([record.key],
                (key_basis, key_missing, key_basis_2, key))
            if record.key == key_missing:
                self.assertIsInstance(record, AbsentContentFactory)
            else:
                results.append((record.key, record.sha1, record.storage_kind,
                    record.get_bytes_as('fulltext')))
        calls = list(basis.calls)
        order = [record[0] for record in results]
        self.assertEqual([key_basis_2, key_basis, key], order)
        for result in results:
            if result[0] == key:
                source = test
            else:
                source = basis
            record = source.get_record_stream([result[0]], 'unordered',
                True).next()
            self.assertEqual(record.key, result[0])
            self.assertEqual(record.sha1, result[1])
            # We used to check that the storage kind matched, but actually it
            # depends on whether it was sourced from the basis, or in a single
            # group, because asking for full texts returns proxy objects to a
            # _ContentMapGenerator object; so checking the kind is unneeded.
            self.assertEqual(record.get_bytes_as('fulltext'), result[3])
        # It's not strictly minimal, but it seems reasonable for now for it to
        # ask which fallbacks have which parents.
        self.assertEqual([
            ("get_parent_map", set([key_basis, key_basis_2, key_missing])),
            # topological is requested from the fallback, because that is what
            # was requested at the top level.
            ("get_record_stream", [key_basis_2, key_basis], 'topological', True)],
            calls)

    def test_get_record_stream_unordered_deltas(self):
        # records from the test knit are answered without asking the basis:
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_missing = ('missing',)
        test.add_lines(key, (), ['foo\n'])
        records = list(test.get_record_stream([key], 'unordered', False))
        self.assertEqual(1, len(records))
        self.assertEqual([], basis.calls)
        # Missing (from test knit) objects are retrieved from the basis:
        basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
        basis.calls = []
        records = list(test.get_record_stream([key_basis, key_missing],
            'unordered', False))
        self.assertEqual(2, len(records))
        calls = list(basis.calls)
        for record in records:
            self.assertSubset([record.key], (key_basis, key_missing))
            if record.key == key_missing:
                self.assertIsInstance(record, AbsentContentFactory)
            else:
                reference = list(basis.get_record_stream([key_basis],
                    'unordered', False))[0]
                self.assertEqual(reference.key, record.key)
                self.assertEqual(reference.sha1, record.sha1)
                self.assertEqual(reference.storage_kind, record.storage_kind)
                self.assertEqual(reference.get_bytes_as(reference.storage_kind),
                    record.get_bytes_as(record.storage_kind))
        # It's not strictly minimal, but it seems reasonable for now for it to
        # ask which fallbacks have which parents.
        self.assertEqual([
            ("get_parent_map", set([key_basis, key_missing])),
            ("get_record_stream", [key_basis], 'unordered', False)],
            calls)

    def test_get_record_stream_ordered_deltas(self):
        # ordering is preserved down into the fallback store.
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_basis_2 = ('quux',)
        key_missing = ('missing',)
        test.add_lines(key, (key_basis,), ['foo\n'])
        # Missing (from test knit) objects are retrieved from the basis:
        basis.add_lines(key_basis, (key_basis_2,), ['foo\n', 'bar\n'])
        basis.add_lines(key_basis_2, (), ['quux\n'])
        basis.calls = []
        # ask for in non-topological order
        records = list(test.get_record_stream(
            [key, key_basis, key_missing, key_basis_2], 'topological', False))
        self.assertEqual(4, len(records))
        results = []
        for record in records:
            self.assertSubset([record.key],
                (key_basis, key_missing, key_basis_2, key))
            if record.key == key_missing:
                self.assertIsInstance(record, AbsentContentFactory)
            else:
                results.append((record.key, record.sha1, record.storage_kind,
                    record.get_bytes_as(record.storage_kind)))
        calls = list(basis.calls)
        order = [record[0] for record in results]
        self.assertEqual([key_basis_2, key_basis, key], order)
        for result in results:
            if result[0] == key:
                source = test
            else:
                source = basis
            record = source.get_record_stream([result[0]], 'unordered',
                False).next()
            self.assertEqual(record.key, result[0])
            self.assertEqual(record.sha1, result[1])
            self.assertEqual(record.storage_kind, result[2])
            self.assertEqual(record.get_bytes_as(record.storage_kind), result[3])
        # It's not strictly minimal, but it seems reasonable for now for it to
        # ask which fallbacks have which parents.
        self.assertEqual([
            ("get_parent_map", set([key_basis, key_basis_2, key_missing])),
            ("get_record_stream", [key_basis_2, key_basis], 'topological', False)],
            calls)

    def test_get_sha1s(self):
        # sha1's in the test knit are answered without asking the basis
        basis, test = self.get_basis_and_test_knit()
        key = ('foo',)
        key_basis = ('bar',)
        key_missing = ('missing',)
        test.add_lines(key, (), ['foo\n'])
        key_sha1sum = osutils.sha('foo\n').hexdigest()
        sha1s = test.get_sha1s([key])
        self.assertEqual({key: key_sha1sum}, sha1s)
        self.assertEqual([], basis.calls)
        # But texts that are not in the test knit are looked for in the basis
        # directly (rather than via text reconstruction) so that remote servers
        # etc don't have to answer with full content.
        basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
        basis_sha1sum = osutils.sha('foo\nbar\n').hexdigest()
        basis.calls = []
        sha1s = test.get_sha1s([key, key_missing, key_basis])
        self.assertEqual({key: key_sha1sum,
            key_basis: basis_sha1sum}, sha1s)
        self.assertEqual([("get_sha1s", set([key_basis, key_missing]))],
            basis.calls)

    def test_insert_record_stream(self):
        # records are inserted as normal; insert_record_stream builds on
        # add_lines, so a smoke test should be all that's needed:
        key = ('foo',)
        key_basis = ('bar',)
        key_delta = ('zaphod',)
        basis, test = self.get_basis_and_test_knit()
        source = self.make_test_knit(name='source')
        basis.add_lines(key_basis, (), ['foo\n'])
        basis.calls = []
        source.add_lines(key_basis, (), ['foo\n'])
        source.add_lines(key_delta, (key_basis,), ['bar\n'])
        stream = source.get_record_stream([key_delta], 'unordered', False)
        test.insert_record_stream(stream)
        # XXX: this does somewhat too many calls in making sure of whether it
        # has to recreate the full text.
        self.assertEqual([("get_parent_map", set([key_basis])),
             ('get_parent_map', set([key_basis])),
             ('get_record_stream', [key_basis], 'unordered', True)],
            basis.calls)
        self.assertEqual({key_delta:(key_basis,)},
            test.get_parent_map([key_delta]))
        self.assertEqual('bar\n', test.get_record_stream([key_delta],
            'unordered', True).next().get_bytes_as('fulltext'))

    def test_iter_lines_added_or_present_in_keys(self):
        # Lines from the basis are returned, and lines for a given key are only
        # returned once.
        key1 = ('foo1',)
        key2 = ('foo2',)
        # all sources are asked for keys:
        basis, test = self.get_basis_and_test_knit()
        basis.add_lines(key1, (), ["foo"])
        basis.calls = []
        lines = list(test.iter_lines_added_or_present_in_keys([key1]))
        self.assertEqual([("foo\n", key1)], lines)
        self.assertEqual([("iter_lines_added_or_present_in_keys", set([key1]))],
            basis.calls)
        # keys in both are not duplicated:
        test.add_lines(key2, (), ["bar\n"])
        basis.add_lines(key2, (), ["bar\n"])
        basis.calls = []
        lines = list(test.iter_lines_added_or_present_in_keys([key2]))
        self.assertEqual([("bar\n", key2)], lines)
        self.assertEqual([], basis.calls)

    def test_keys(self):
        key1 = ('foo1',)
        key2 = ('foo2',)
        # all sources are asked for keys:
        basis, test = self.get_basis_and_test_knit()
        keys = test.keys()
        self.assertEqual(set(), set(keys))
        self.assertEqual([("keys",)], basis.calls)
        # keys from a basis are returned:
        basis.add_lines(key1, (), [])
        basis.calls = []
        keys = test.keys()
        self.assertEqual(set([key1]), set(keys))
        self.assertEqual([("keys",)], basis.calls)
        # keys in both are not duplicated:
        test.add_lines(key2, (), [])
        basis.add_lines(key2, (), [])
        basis.calls = []
        keys = test.keys()
        self.assertEqual(2, len(keys))
        self.assertEqual(set([key1, key2]), set(keys))
        self.assertEqual([("keys",)], basis.calls)

    def test_add_mpdiffs(self):
        # records are inserted as normal; add_mpdiff builds on
        # add_lines, so a smoke test should be all that's needed:
        key = ('foo',)
        key_basis = ('bar',)
        key_delta = ('zaphod',)
        basis, test = self.get_basis_and_test_knit()
        source = self.make_test_knit(name='source')
        basis.add_lines(key_basis, (), ['foo\n'])
        basis.calls = []
        source.add_lines(key_basis, (), ['foo\n'])
        source.add_lines(key_delta, (key_basis,), ['bar\n'])
        diffs = source.make_mpdiffs([key_delta])
        test.add_mpdiffs([(key_delta, (key_basis,),
            source.get_sha1s([key_delta])[key_delta], diffs[0])])
        self.assertEqual([("get_parent_map", set([key_basis])),
            ('get_record_stream', [key_basis], 'unordered', True),],
            basis.calls)
        self.assertEqual({key_delta:(key_basis,)},
            test.get_parent_map([key_delta]))
        self.assertEqual('bar\n', test.get_record_stream([key_delta],
            'unordered', True).next().get_bytes_as('fulltext'))

    def test_make_mpdiffs(self):
        # Generating an mpdiff across a stacking boundary should detect parent
        # texts regions.
        key = ('foo',)
        key_left = ('bar',)
        key_right = ('zaphod',)
        basis, test = self.get_basis_and_test_knit()
        basis.add_lines(key_left, (), ['bar\n'])
        basis.add_lines(key_right, (), ['zaphod\n'])
        basis.calls = []
        test.add_lines(key, (key_left, key_right),
            ['bar\n', 'foo\n', 'zaphod\n'])
        diffs = test.make_mpdiffs([key])
        self.assertEqual([
            multiparent.MultiParent([multiparent.ParentText(0, 0, 0, 1),
                multiparent.NewText(['foo\n']),
                multiparent.ParentText(1, 0, 2, 1)])],
            diffs)
        self.assertEqual(3, len(basis.calls))
        self.assertEqual([
            ("get_parent_map", set([key_left, key_right])),
            ("get_parent_map", set([key_left, key_right])),
            ],
            basis.calls[:-1])
        last_call = basis.calls[-1]
        self.assertEqual('get_record_stream', last_call[0])
        self.assertEqual(set([key_left, key_right]), set(last_call[1]))
        self.assertEqual('topological', last_call[2])
        self.assertEqual(True, last_call[3])


class TestNetworkBehaviour(KnitTests):
    """Tests for getting data out of/into knits over the network."""

    def test_include_delta_closure_generates_a_knit_delta_closure(self):
        vf = self.make_test_knit(name='test')
        # put in three texts, giving ft, delta, delta
        vf.add_lines(('base',), (), ['base\n', 'content\n'])
        vf.add_lines(('d1',), (('base',),), ['d1\n'])
        vf.add_lines(('d2',), (('d1',),), ['d2\n'])
        # But heuristics could interfere, so check what happened:
        self.assertEqual(['knit-ft-gz', 'knit-delta-gz', 'knit-delta-gz'],
            [record.storage_kind for record in
             vf.get_record_stream([('base',), ('d1',), ('d2',)],
                'topological', False)])
        # generate a stream of just the deltas include_delta_closure=True,
        # serialise to the network, and check that we get a delta closure on the wire.
        stream = vf.get_record_stream([('d1',), ('d2',)], 'topological', True)
        netb = [record.get_bytes_as(record.storage_kind) for record in stream]
        # The first bytes should be a memo from _ContentMapGenerator, and the
        # second bytes should be empty (because its a API proxy not something
        # for wire serialisation.
        self.assertEqual('', netb[1])
        bytes = netb[0]
        kind, line_end = network_bytes_to_kind_and_offset(bytes)
        self.assertEqual('knit-delta-closure', kind)


class TestContentMapGenerator(KnitTests):
    """Tests for ContentMapGenerator"""

    def test_get_record_stream_gives_records(self):
        vf = self.make_test_knit(name='test')
        # put in three texts, giving ft, delta, delta
        vf.add_lines(('base',), (), ['base\n', 'content\n'])
        vf.add_lines(('d1',), (('base',),), ['d1\n'])
        vf.add_lines(('d2',), (('d1',),), ['d2\n'])
        keys = [('d1',), ('d2',)]
        generator = _VFContentMapGenerator(vf, keys,
            global_map=vf.get_parent_map(keys))
        for record in generator.get_record_stream():
            if record.key == ('d1',):
                self.assertEqual('d1\n', record.get_bytes_as('fulltext'))
            else:
                self.assertEqual('d2\n', record.get_bytes_as('fulltext'))

    def test_get_record_stream_kinds_are_raw(self):
        vf = self.make_test_knit(name='test')
        # put in three texts, giving ft, delta, delta
        vf.add_lines(('base',), (), ['base\n', 'content\n'])
        vf.add_lines(('d1',), (('base',),), ['d1\n'])
        vf.add_lines(('d2',), (('d1',),), ['d2\n'])
        keys = [('base',), ('d1',), ('d2',)]
        generator = _VFContentMapGenerator(vf, keys,
            global_map=vf.get_parent_map(keys))
        kinds = {('base',): 'knit-delta-closure',
            ('d1',): 'knit-delta-closure-ref',
            ('d2',): 'knit-delta-closure-ref',
            }
        for record in generator.get_record_stream():
            self.assertEqual(kinds[record.key], record.storage_kind)