~bzr-pqm/bzr/bzr.dev

# Copyright (C) 2005, 2007 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

import os

import re

import sys

import bzrlib

from bzrlib import (

    bzrdir,

    errors,

    osutils,

    merge,

    repository,

    versionedfile,

    )

from bzrlib.branch import Branch

from bzrlib.bzrdir import BzrDir

from bzrlib.repofmt import knitrepo

from bzrlib.tests import TestCaseWithTransport

from bzrlib.tests.http_utils import TestCaseWithWebserver

from bzrlib.tests.test_revision import make_branches

from bzrlib.trace import mutter

from bzrlib.upgrade import Convert

from bzrlib.workingtree import WorkingTree

# These tests are a bit old; please instead add new tests into

# interrepository_implementations/ so they'll run on all relevant

# combinations.

def has_revision(branch, revision_id):

    return branch.repository.has_revision(revision_id)

def fetch_steps(self, br_a, br_b, writable_a):

    """A foreign test method for testing fetch locally and remotely."""

    # TODO RBC 20060201 make this a repository test.

    repo_b = br_b.repository

    self.assertFalse(repo_b.has_revision(br_a.revision_history()[3]))

    self.assertTrue(repo_b.has_revision(br_a.revision_history()[2]))

    self.assertEquals(len(br_b.revision_history()), 7)

    br_b.fetch(br_a, br_a.revision_history()[2])

    # branch.fetch is not supposed to alter the revision history

    self.assertEquals(len(br_b.revision_history()), 7)

    self.assertFalse(repo_b.has_revision(br_a.revision_history()[3]))

    # fetching the next revision up in sample data copies one revision

    br_b.fetch(br_a, br_a.revision_history()[3])

    self.assertTrue(repo_b.has_revision(br_a.revision_history()[3]))

    self.assertFalse(has_revision(br_a, br_b.revision_history()[6]))

    self.assertTrue(br_a.repository.has_revision(br_b.revision_history()[5]))

    # When a non-branch ancestor is missing, it should be unlisted...

    # as its not reference from the inventory weave.

    br_b4 = self.make_branch('br_4')

    br_b4.fetch(br_b)

    writable_a.fetch(br_b)

    self.assertTrue(has_revision(br_a, br_b.revision_history()[3]))

    self.assertTrue(has_revision(br_a, br_b.revision_history()[4]))

    br_b2 = self.make_branch('br_b2')

    br_b2.fetch(br_b)

    self.assertTrue(has_revision(br_b2, br_b.revision_history()[4]))

    self.assertTrue(has_revision(br_b2, br_a.revision_history()[2]))

    self.assertFalse(has_revision(br_b2, br_a.revision_history()[3]))

    br_a2 = self.make_branch('br_a2')

    br_a2.fetch(br_a)

    self.assertTrue(has_revision(br_a2, br_b.revision_history()[4]))

    self.assertTrue(has_revision(br_a2, br_a.revision_history()[3]))

    self.assertTrue(has_revision(br_a2, br_a.revision_history()[2]))

    br_a3 = self.make_branch('br_a3')

    # pulling a branch with no revisions grabs nothing, regardless of

    # whats in the inventory.

    br_a3.fetch(br_a2)

    for revno in range(4):

        self.assertFalse(

            br_a3.repository.has_revision(br_a.revision_history()[revno]))

    br_a3.fetch(br_a2, br_a.revision_history()[2])

    # pull the 3 revisions introduced by a@u-0-3

    br_a3.fetch(br_a2, br_a.revision_history()[3])

    # NoSuchRevision should be raised if the branch is missing the revision

    # that was requested.

    self.assertRaises(errors.NoSuchRevision, br_a3.fetch, br_a2, 'pizza')

    # TODO: Test trying to fetch from a branch that points to a revision not

    # actually present in its repository.  Not every branch format allows you

    # to directly point to such revisions, so it's a bit complicated to

    # construct.  One way would be to uncommit and gc the revision, but not

    # every branch supports that.  -- mbp 20070814

    #TODO: test that fetch correctly does reweaving when needed. RBC 20051008

    # Note that this means - updating the weave when ghosts are filled in to

    # add the right parents.

class TestFetch(TestCaseWithTransport):

    def test_fetch(self):

        #highest indices a: 5, b: 7

        br_a, br_b = make_branches(self, format='dirstate-tags')

        fetch_steps(self, br_a, br_b, br_a)

    def test_fetch_self(self):

        wt = self.make_branch_and_tree('br')

        wt.branch.fetch(wt.branch)

    def test_fetch_root_knit(self):

        """Ensure that knit2.fetch() updates the root knit

        This tests the case where the root has a new revision, but there are no

        corresponding filename, parent, contents or other changes.

        """

        knit1_format = bzrdir.BzrDirMetaFormat1()

        knit1_format.repository_format = knitrepo.RepositoryFormatKnit1()

        knit2_format = bzrdir.BzrDirMetaFormat1()

        knit2_format.repository_format = knitrepo.RepositoryFormatKnit3()

        # we start with a knit1 repository because that causes the

        # root revision to change for each commit, even though the content,

        # parent, name, and other attributes are unchanged.

        tree = self.make_branch_and_tree('tree', knit1_format)

        tree.set_root_id('tree-root')

        tree.commit('rev1', rev_id='rev1')

        tree.commit('rev2', rev_id='rev2')

        # Now we convert it to a knit2 repository so that it has a root knit

        Convert(tree.basedir, knit2_format)

        tree = WorkingTree.open(tree.basedir)

        branch = self.make_branch('branch', format=knit2_format)

        branch.pull(tree.branch, stop_revision='rev1')

        repo = branch.repository

        repo.lock_read()

        try:

            # Make sure fetch retrieved only what we requested

            self.assertEqual({('tree-root', 'rev1'):()},

                repo.texts.get_parent_map(

                    [('tree-root', 'rev1'), ('tree-root', 'rev2')]))

        finally:

            repo.unlock()

        branch.pull(tree.branch)

        # Make sure that the next revision in the root knit was retrieved,

        # even though the text, name, parent_id, etc., were unchanged.

        repo.lock_read()

        try:

            # Make sure fetch retrieved only what we requested

            self.assertEqual({('tree-root', 'rev2'):(('tree-root', 'rev1'),)},

                repo.texts.get_parent_map([('tree-root', 'rev2')]))

        finally:

            repo.unlock()

    def test_fetch_incompatible(self):

        knit_tree = self.make_branch_and_tree('knit', format='knit')

        knit3_tree = self.make_branch_and_tree('knit3',

            format='dirstate-with-subtree')

        knit3_tree.commit('blah')

        e = self.assertRaises(errors.IncompatibleRepositories,

                              knit_tree.branch.fetch, knit3_tree.branch)

        self.assertContainsRe(str(e),

            r"(?m).*/knit.*\nis not compatible with\n.*/knit3/.*\n"

            r"different rich-root support")

class TestMergeFetch(TestCaseWithTransport):

    def test_merge_fetches_unrelated(self):

        """Merge brings across history from unrelated source"""

        wt1 = self.make_branch_and_tree('br1')

        br1 = wt1.branch

        wt1.commit(message='rev 1-1', rev_id='1-1')

        wt1.commit(message='rev 1-2', rev_id='1-2')

        wt2 = self.make_branch_and_tree('br2')

        br2 = wt2.branch

        wt2.commit(message='rev 2-1', rev_id='2-1')

        wt2.merge_from_branch(br1, from_revision='null:')

        self._check_revs_present(br2)

    def test_merge_fetches(self):

        """Merge brings across history from source"""

        wt1 = self.make_branch_and_tree('br1')

        br1 = wt1.branch

        wt1.commit(message='rev 1-1', rev_id='1-1')

        dir_2 = br1.bzrdir.sprout('br2')

        br2 = dir_2.open_branch()

        wt1.commit(message='rev 1-2', rev_id='1-2')

        wt2 = dir_2.open_workingtree()

        wt2.commit(message='rev 2-1', rev_id='2-1')

        wt2.merge_from_branch(br1)

        self._check_revs_present(br2)

    def _check_revs_present(self, br2):

        for rev_id in '1-1', '1-2', '2-1':

            self.assertTrue(br2.repository.has_revision(rev_id))

            rev = br2.repository.get_revision(rev_id)

            self.assertEqual(rev.revision_id, rev_id)

            self.assertTrue(br2.repository.get_inventory(rev_id))

class TestMergeFileHistory(TestCaseWithTransport):

    def setUp(self):

        super(TestMergeFileHistory, self).setUp()

        wt1 = self.make_branch_and_tree('br1')

        br1 = wt1.branch

        self.build_tree_contents([('br1/file', 'original contents\n')])

        wt1.add('file', 'this-file-id')

        wt1.commit(message='rev 1-1', rev_id='1-1')

        dir_2 = br1.bzrdir.sprout('br2')

        br2 = dir_2.open_branch()

        wt2 = dir_2.open_workingtree()

        self.build_tree_contents([('br1/file', 'original from 1\n')])

        wt1.commit(message='rev 1-2', rev_id='1-2')

        self.build_tree_contents([('br1/file', 'agreement\n')])

        wt1.commit(message='rev 1-3', rev_id='1-3')

        self.build_tree_contents([('br2/file', 'contents in 2\n')])

        wt2.commit(message='rev 2-1', rev_id='2-1')

        self.build_tree_contents([('br2/file', 'agreement\n')])

        wt2.commit(message='rev 2-2', rev_id='2-2')

    def test_merge_fetches_file_history(self):

        """Merge brings across file histories"""

        br2 = Branch.open('br2')

        br1 = Branch.open('br1')

        wt2 = WorkingTree.open('br2').merge_from_branch(br1)

        br2.lock_read()

        self.addCleanup(br2.unlock)

        for rev_id, text in [('1-2', 'original from 1\n'),

                             ('1-3', 'agreement\n'),

                             ('2-1', 'contents in 2\n'),

                             ('2-2', 'agreement\n')]:

            self.assertEqualDiff(

                br2.repository.revision_tree(

                    rev_id).get_file_text('this-file-id'), text)

class TestHttpFetch(TestCaseWithWebserver):

    # FIXME RBC 20060124 this really isn't web specific, perhaps an

    # instrumented readonly transport? Can we do an instrumented

    # adapter and use self.get_readonly_url ?

    def test_fetch(self):

        #highest indices a: 5, b: 7

        br_a, br_b = make_branches(self)

        br_rem_a = Branch.open(self.get_readonly_url('branch1'))

        fetch_steps(self, br_rem_a, br_b, br_a)

    def _count_log_matches(self, target, logs):

        """Count the number of times the target file pattern was fetched in an http log"""

        get_succeeds_re = re.compile(

            '.*"GET .*%s HTTP/1.1" 20[06] - "-" "bzr/%s' %

            (     target,                    bzrlib.__version__))

        c = 0

        for line in logs:

            if get_succeeds_re.match(line):

                c += 1

        return c

    def test_weaves_are_retrieved_once(self):

        self.build_tree(("source/", "source/file", "target/"))

        # This test depends on knit dasta storage.

        wt = self.make_branch_and_tree('source', format='dirstate-tags')

        branch = wt.branch

        wt.add(["file"], ["id"])

        wt.commit("added file")

        open("source/file", 'w').write("blah\n")

        wt.commit("changed file")

        target = BzrDir.create_branch_and_repo("target/")

        source = Branch.open(self.get_readonly_url("source/"))

        target.fetch(source)

        # this is the path to the literal file. As format changes

        # occur it needs to be updated. FIXME: ask the store for the

        # path.

        self.log("web server logs are:")

        http_logs = self.get_readonly_server().logs

        self.log('\n'.join(http_logs))

        # unfortunately this log entry is branch format specific. We could

        # factor out the 'what files does this format use' to a method on the

        # repository, which would let us to this generically. RBC 20060419

        # RBC 20080408: Or perhaps we can assert that no files are fully read

        # twice?

        self.assertEqual(1, self._count_log_matches('/ce/id.kndx', http_logs))

        self.assertEqual(1, self._count_log_matches('/ce/id.knit', http_logs))

        self.assertEqual(1, self._count_log_matches('inventory.kndx', http_logs))

        # this r-h check test will prevent regressions, but it currently already

        # passes, before the patch to cache-rh is applied :[

        self.assertTrue(1 >= self._count_log_matches('revision-history',

                                                     http_logs))

        self.assertTrue(1 >= self._count_log_matches('last-revision',

                                                     http_logs))

        # FIXME naughty poking in there.

        self.get_readonly_server().logs = []

        # check there is nothing more to fetch.  We take care to re-use the

        # existing transport so that the request logs we're about to examine

        # aren't cluttered with redundant probes for a smart server.

        # XXX: Perhaps this further parameterisation: test http with smart

        # server, and test http without smart server?

        source = Branch.open(

            self.get_readonly_url("source/"),

            possible_transports=[source.bzrdir.root_transport])

        target.fetch(source)

        # should make just two requests

        http_logs = self.get_readonly_server().logs

        self.log("web server logs are:")

        self.log('\n'.join(http_logs))

        self.assertEqual(1, self._count_log_matches('branch-format', http_logs))

        self.assertEqual(1, self._count_log_matches('branch/format', http_logs))

        self.assertEqual(1, self._count_log_matches('repository/format',

            http_logs))

        self.assertEqual(1, self._count_log_matches('revisions.kndx',

            http_logs))

        self.assertTrue(1 >= self._count_log_matches('revision-history',

                                                     http_logs))

        self.assertTrue(1 >= self._count_log_matches('last-revision',

                                                     http_logs))

        self.assertLength(5, http_logs)

class TestKnitToPackFetch(TestCaseWithTransport):

    def find_get_record_stream(self, calls, expected_count=1):

        """In a list of calls, find the last 'get_record_stream'.

        :param expected_count: The number of calls we should exepect to find.

            If a different number is found, an assertion is raised.

        """

        get_record_call = None

        call_count = 0

        for call in calls:

            if call[0] == 'get_record_stream':

                call_count += 1

                get_record_call = call

        self.assertEqual(expected_count, call_count)

        return get_record_call

    def test_fetch_with_deltas_no_delta_closure(self):

        tree = self.make_branch_and_tree('source', format='dirstate')

        target = self.make_repository('target', format='pack-0.92')

        self.build_tree(['source/file'])

        tree.set_root_id('root-id')

        tree.add('file', 'file-id')

        tree.commit('one', rev_id='rev-one')

        source = tree.branch.repository

        source.texts = versionedfile.RecordingVersionedFilesDecorator(

                        source.texts)

        source.signatures = versionedfile.RecordingVersionedFilesDecorator(

                        source.signatures)

        source.revisions = versionedfile.RecordingVersionedFilesDecorator(

                        source.revisions)

        source.inventories = versionedfile.RecordingVersionedFilesDecorator(

                        source.inventories)

        # precondition

        self.assertTrue(target._format._fetch_uses_deltas)

        target.fetch(source, revision_id='rev-one')

        self.assertEqual(('get_record_stream', [('file-id', 'rev-one')],

                          target._format._fetch_order, False),

                         self.find_get_record_stream(source.texts.calls))

        self.assertEqual(('get_record_stream', [('rev-one',)],

          target._format._fetch_order, False),

          self.find_get_record_stream(source.inventories.calls, 2))

        self.assertEqual(('get_record_stream', [('rev-one',)],

                          target._format._fetch_order, False),

                         self.find_get_record_stream(source.revisions.calls))

        # XXX: Signatures is special, and slightly broken. The

        # standard item_keys_introduced_by actually does a lookup for every

        # signature to see if it exists, rather than waiting to do them all at

        # once at the end. The fetch code then does an all-at-once and just

        # allows for some of them to be missing.

        # So we know there will be extra calls, but the *last* one is the one

        # we care about.

        signature_calls = source.signatures.calls[-1:]

        self.assertEqual(('get_record_stream', [('rev-one',)],

                          target._format._fetch_order, False),

                         self.find_get_record_stream(signature_calls))

    def test_fetch_no_deltas_with_delta_closure(self):

        tree = self.make_branch_and_tree('source', format='dirstate')

        target = self.make_repository('target', format='pack-0.92')

        self.build_tree(['source/file'])

        tree.set_root_id('root-id')

        tree.add('file', 'file-id')

        tree.commit('one', rev_id='rev-one')

        source = tree.branch.repository

        source.texts = versionedfile.RecordingVersionedFilesDecorator(

                        source.texts)

        source.signatures = versionedfile.RecordingVersionedFilesDecorator(

                        source.signatures)

        source.revisions = versionedfile.RecordingVersionedFilesDecorator(

                        source.revisions)

        source.inventories = versionedfile.RecordingVersionedFilesDecorator(

                        source.inventories)

        # XXX: This won't work in general, but for the dirstate format it does.

        old_fetch_uses_deltas_setting = target._format._fetch_uses_deltas

        def restore():

            target._format._fetch_uses_deltas = old_fetch_uses_deltas_setting

        self.addCleanup(restore)

        target._format._fetch_uses_deltas = False

        target.fetch(source, revision_id='rev-one')

        self.assertEqual(('get_record_stream', [('file-id', 'rev-one')],

                          target._format._fetch_order, True),

                         self.find_get_record_stream(source.texts.calls))

        self.assertEqual(('get_record_stream', [('rev-one',)],

            target._format._fetch_order, True),

            self.find_get_record_stream(source.inventories.calls, 2))

        self.assertEqual(('get_record_stream', [('rev-one',)],

                          target._format._fetch_order, True),

                         self.find_get_record_stream(source.revisions.calls))

        # XXX: Signatures is special, and slightly broken. The

        # standard item_keys_introduced_by actually does a lookup for every

        # signature to see if it exists, rather than waiting to do them all at

        # once at the end. The fetch code then does an all-at-once and just

        # allows for some of them to be missing.

        # So we know there will be extra calls, but the *last* one is the one

        # we care about.

        signature_calls = source.signatures.calls[-1:]

        self.assertEqual(('get_record_stream', [('rev-one',)],

                          target._format._fetch_order, True),

                         self.find_get_record_stream(signature_calls))

    def test_fetch_revisions_with_deltas_into_pack(self):

        # See BUG #261339, dev versions of bzr could accidentally create deltas

        # in revision texts in knit branches (when fetching from packs). So we

        # ensure that *if* a knit repository has a delta in revisions, that it

        # gets properly expanded back into a fulltext when stored in the pack

        # file.

        tree = self.make_branch_and_tree('source', format='dirstate')

        target = self.make_repository('target', format='pack-0.92')

        self.build_tree(['source/file'])

        tree.set_root_id('root-id')

        tree.add('file', 'file-id')

        tree.commit('one', rev_id='rev-one')

        # Hack the KVF for revisions so that it "accidentally" allows a delta

        tree.branch.repository.revisions._max_delta_chain = 200

        tree.commit('two', rev_id='rev-two')

        source = tree.branch.repository

        # Ensure that we stored a delta

        source.lock_read()

        self.addCleanup(source.unlock)

        record = source.revisions.get_record_stream([('rev-two',)],

            'unordered', False).next()

        self.assertEqual('knit-delta-gz', record.storage_kind)

        target.fetch(tree.branch.repository, revision_id='rev-two')

        # The record should get expanded back to a fulltext

        target.lock_read()

        self.addCleanup(target.unlock)

        record = target.revisions.get_record_stream([('rev-two',)],

            'unordered', False).next()

        self.assertEqual('knit-ft-gz', record.storage_kind)

    def test_fetch_with_fallback_and_merge(self):

        builder = self.make_branch_builder('source', format='pack-0.92')

        builder.start_series()

        # graph

        #   A

        #   |\

        #   B C

        #   | |

        #   | D

        #   | |

        #   | E

        #    \|

        #     F

        # A & B are present in the base (stacked-on) repository, A-E are

        # present in the source.

        # This reproduces bug #304841

        # We need a large enough inventory that total size of compressed deltas

        # is shorter than the size of a compressed fulltext. We have to use

        # random ids because otherwise the inventory fulltext compresses too

        # well and the deltas get bigger.

        to_add = [

            ('add', ('', 'TREE_ROOT', 'directory', None))]

        for i in xrange(10):

            fname = 'file%03d' % (i,)

            fileid = '%s-%s' % (fname, osutils.rand_chars(64))

            to_add.append(('add', (fname, fileid, 'file', 'content\n')))

        builder.build_snapshot('A', None, to_add)

        builder.build_snapshot('B', ['A'], [])

        builder.build_snapshot('C', ['A'], [])

        builder.build_snapshot('D', ['C'], [])

        builder.build_snapshot('E', ['D'], [])

        builder.build_snapshot('F', ['E', 'B'], [])

        builder.finish_series()

        source_branch = builder.get_branch()

        source_branch.bzrdir.sprout('base', revision_id='B')

        target_branch = self.make_branch('target', format='1.6')

        target_branch.set_stacked_on_url('../base')

        source = source_branch.repository

        source.lock_read()

        self.addCleanup(source.unlock)

        source.inventories = versionedfile.OrderingVersionedFilesDecorator(

                        source.inventories,

                        key_priority={('E',): 1, ('D',): 2, ('C',): 4,

                                      ('F',): 3})

        # Ensure that the content is yielded in the proper order, and given as

        # the expected kinds

        records = [(record.key, record.storage_kind)

                   for record in source.inventories.get_record_stream(

                        [('D',), ('C',), ('E',), ('F',)], 'unordered', False)]

        self.assertEqual([(('E',), 'knit-delta-gz'), (('D',), 'knit-delta-gz'),

                          (('F',), 'knit-delta-gz'), (('C',), 'knit-delta-gz')],

                          records)

        target_branch.lock_write()

        self.addCleanup(target_branch.unlock)

        target = target_branch.repository

        target.fetch(source, revision_id='F')

        # 'C' should be expanded to a fulltext, but D and E should still be

        # deltas

        stream = target.inventories.get_record_stream(

            [('C',), ('D',), ('E',), ('F',)],

            'unordered', False)

        kinds = dict((record.key, record.storage_kind) for record in stream)

        self.assertEqual({('C',): 'knit-ft-gz', ('D',): 'knit-delta-gz',

                          ('E',): 'knit-delta-gz', ('F',): 'knit-delta-gz'},

                         kinds)

class Test1To2Fetch(TestCaseWithTransport):

    """Tests for Model1To2 failure modes"""

    def make_tree_and_repo(self):

        self.tree = self.make_branch_and_tree('tree', format='pack-0.92')

        self.repo = self.make_repository('rich-repo', format='rich-root-pack')

        self.repo.lock_write()

        self.addCleanup(self.repo.unlock)

    def do_fetch_order_test(self, first, second):

        """Test that fetch works no matter what the set order of revision is.

        This test depends on the order of items in a set, which is

        implementation-dependant, so we test A, B and then B, A.

        """

        self.make_tree_and_repo()

        self.tree.commit('Commit 1', rev_id=first)

        self.tree.commit('Commit 2', rev_id=second)

        self.repo.fetch(self.tree.branch.repository, second)

    def test_fetch_order_AB(self):

        """See do_fetch_order_test"""

        self.do_fetch_order_test('A', 'B')

    def test_fetch_order_BA(self):

        """See do_fetch_order_test"""

        self.do_fetch_order_test('B', 'A')

    def get_parents(self, file_id, revision_id):

        self.repo.lock_read()

        try:

            parent_map = self.repo.texts.get_parent_map([(file_id, revision_id)])

            return parent_map[(file_id, revision_id)]

        finally:

            self.repo.unlock()

    def test_fetch_ghosts(self):

        self.make_tree_and_repo()

        self.tree.commit('first commit', rev_id='left-parent')

        self.tree.add_parent_tree_id('ghost-parent')

        fork = self.tree.bzrdir.sprout('fork', 'null:').open_workingtree()

        fork.commit('not a ghost', rev_id='not-ghost-parent')

        self.tree.branch.repository.fetch(fork.branch.repository,