← Back to branch summary

~abentley/bzrtools/bzrtools.dev

« back to all changes in this revision

Viewing changes to baz_import.py

  • Committer: Aaron Bentley
  • Date: 2008-04-11 00:03:51 UTC
  • Revision ID: aaron@aaronbentley.com-20080411000351-dmbvgmanygnphzgp
Add escaping to HTML output

Show diffs side-by-side

added added

removed removed

Lines of Context:
baz_import.py
1
 
# Copyright (C) 2005 by Aaron Bentley
 
1
# Copyright (C) 2005, 2006 by Aaron Bentley
2
2
 
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
16
 
 
17
import errno
 
18
 
 
19
from bzrlib.bzrdir import BzrDir
 
20
import bzrlib.bzrdir as bzrdir
 
21
from bzrlib.errors import (BzrError,
 
22
                           NotBranchError,
 
23
                           NoWorkingTree,
 
24
                           BzrCommandError,
 
25
                           NoSuchRevision,
 
26
                           NoRepositoryPresent,
 
27
                          )
16
28
from bzrlib.branch import Branch
 
29
from bzrlib.commit import Commit, NullCommitReporter
17
30
from bzrlib.commands import Command
 
31
from bzrlib.option import _global_option, Option
 
32
from bzrlib.merge import merge_inner
 
33
from bzrlib.revision import NULL_REVISION, is_null
 
34
import bzrlib.ui
 
35
import bzrlib.ui.text
 
36
from bzrlib.workingtree import WorkingTree
18
37
from errors import NoPyBaz
19
38
try:
20
39
    import pybaz
21
40
    import pybaz.errors
 
41
    from pybaz import NameParser as NameParser
22
42
    from pybaz.backends.baz import null_cmd
23
43
except ImportError:
24
44
    raise NoPyBaz
 
45
from baz_helper import iter_new_merges, direct_merges
25
46
import tempfile
26
47
import os
27
48
import os.path
28
49
import shutil
29
50
import bzrlib
30
 
from bzrlib.errors import BzrError
31
51
import bzrlib.trace
32
52
import bzrlib.merge
33
53
import bzrlib.inventory
36
56
import email.Utils
37
57
from progress import *
38
58
 
 
59
 
 
60
BAZ_IMPORT_ROOT = 'TREE_ROOT'
 
61
 
 
62
 
 
63
class ImportCommitReporter(NullCommitReporter):
 
64
 
 
65
    def escaped(self, escape_count, message):
 
66
        bzrlib.trace.warning("replaced %d control characters in message" %
 
67
                             escape_count)
 
68
 
39
69
def add_id(files, id=None):
40
70
    """Adds an explicit id to a list of files.
41
71
 
50
80
    args.extend(files)
51
81
    return null_cmd(args)
52
82
 
 
83
saved_dir = None
 
84
 
 
85
def make_archive(name, location):
 
86
    pb_location = pybaz.ArchiveLocation(location)
 
87
    pb_location.create_master(pybaz.Archive(name),
 
88
                              pybaz.ArchiveLocationParams())
 
89
 
53
90
def test_environ():
54
91
    """
55
92
    >>> q = test_environ()
61
98
    >>> os.path.exists(q)
62
99
    False
63
100
    """
 
101
    global saved_dir
 
102
    saved_dir = os.getcwdu()
64
103
    tdir = tempfile.mkdtemp(prefix="testdir-")
65
104
    os.environ["HOME"] = os.path.join(tdir, "home")
66
105
    os.mkdir(os.environ["HOME"])
67
106
    arch_dir = os.path.join(tdir, "archive_dir")
68
 
    pybaz.make_archive("test@example.com", arch_dir)
 
107
    make_archive("test@example.com", arch_dir)
69
108
    work_dir = os.path.join(tdir, "work_dir")
70
109
    os.mkdir(work_dir)
71
110
    os.chdir(work_dir)
104
143
    add_id([path], id)
105
144
 
106
145
def teardown_environ(tdir):
107
 
    os.chdir("/")
 
146
    os.chdir(saved_dir)
108
147
    shutil.rmtree(tdir)
109
148
 
110
149
def timport(tree, summary):
196
235
    """
197
236
    try:
198
237
        revision = version.iter_revisions(reverse=True).next()
 
238
    except StopIteration:
 
239
        return ()
199
240
    except:
 
241
        print version
200
242
        if not version.exists():
201
243
            raise NoSuchVersion(version)
202
244
        else:
206
248
    return ancestors
207
249
 
208
250
def get_last_revision(branch):
209
 
    last_patch = branch.last_patch()
 
251
    last_patch = branch.last_revision()
210
252
    try:
211
253
        return arch_revision(last_patch)
212
254
    except NotArchRevision:
213
255
        raise UserError(
214
256
            "Directory \"%s\" already exists, and the last revision is not"
215
 
            " an Arch revision (%s)" % (output_dir, last_patch))
216
 
 
217
 
 
218
 
def get_remaining_revisions(output_dir, version):
 
257
            " an Arch revision (%s)" % (branch.base, last_patch))
 
258
 
 
259
def do_branch(br_from, to_location, revision_id):
 
260
    """Derived from branch in builtins."""
 
261
    br_from.lock_read()
 
262
    try:
 
263
        try:
 
264
            os.mkdir(to_location)
 
265
        except OSError, e:
 
266
            if e.errno == errno.EEXIST:
 
267
                raise UserError('Target directory "%s" already'
 
268
                                      ' exists.' % to_location)
 
269
            if e.errno == errno.ENOENT:
 
270
                raise UserError('Parent of "%s" does not exist.' %
 
271
                                      to_location)
 
272
            else:
 
273
                raise
 
274
        try:
 
275
            br_from.bzrdir.clone(to_location, revision_id)
 
276
        except NoSuchRevision:
 
277
            rmtree(to_location)
 
278
            msg = "The branch %s has no revision %s." % (from_location,
 
279
                                                         revision_id)
 
280
            raise UserError(msg)
 
281
    finally:
 
282
        br_from.unlock()
 
283
 
 
284
def get_remaining_revisions(output_dir, version, encoding,
 
285
                            reuse_history_from=[]):
219
286
    last_patch = None
220
287
    old_revno = None
221
 
    if os.path.exists(output_dir):
 
288
    output_exists = os.path.exists(output_dir)
 
289
    if output_exists:
222
290
        # We are starting from an existing directory, figure out what
223
291
        # the current version is
224
 
        branch = find_branch(output_dir)
225
 
        last_patch = get_last_revision(branch)
226
 
        if version is None:
 
292
        branch = Branch.open(output_dir)
 
293
        last_patch, last_encoding = get_last_revision(branch)
 
294
        assert encoding == last_encoding
 
295
        if last_patch is None:
 
296
            if not is_null(branch.last_revision()):
 
297
                raise NotPreviousImport(branch.base)
 
298
        elif version is None:
227
299
            version = last_patch.version
228
300
    elif version is None:
229
301
        raise UserError("No version specified, and directory does not exist.")
230
302
 
231
303
    try:
232
304
        ancestors = version_ancestry(version)
 
305
        if not output_exists and reuse_history_from != []:
 
306
            for ancestor in reversed(ancestors):
 
307
                if last_patch is not None:
 
308
                    # found something to copy
 
309
                    break
 
310
                # try to grab a copy of ancestor
 
311
                # note that is not optimised: we could look for namespace
 
312
                # transitions and only look for the past after the
 
313
                # transition.
 
314
                for history_root in reuse_history_from:
 
315
                    possible_source = os.path.join(history_root,
 
316
                        map_namespace(ancestor.version))
 
317
                    try:
 
318
                        source = Branch.open(possible_source)
 
319
                        rev_id = revision_id(ancestor, encoding)
 
320
                        if rev_id in source.revision_history():
 
321
                            do_branch(source, output_dir, rev_id)
 
322
                            last_patch = ancestor
 
323
                            break
 
324
                    except NotBranchError:
 
325
                        pass
233
326
    except NoSuchVersion, e:
234
 
        raise UserError(e)
 
327
        raise UserError(str(e))
235
328
 
236
329
    if last_patch:
237
330
        for i in range(len(ancestors)):
239
332
                break
240
333
        else:
241
334
            raise UserError("Directory \"%s\" already exists, and the last "
242
 
                "revision (%s) is not in the ancestry of %s" % 
 
335
                "revision (%s) is not in the ancestry of %s" %
243
336
                (output_dir, last_patch, version))
244
337
        # Strip off all of the ancestors which are already present
245
338
        # And get a directory starting with the latest ancestor
246
339
        latest_ancestor = ancestors[i]
247
 
        old_revno = find_branch(output_dir).revno()
 
340
        old_revno = Branch.open(output_dir).revno()
248
341
        ancestors = ancestors[i+1:]
249
342
    return ancestors, old_revno
250
343
 
251
 
def import_version(output_dir, version, fancy=True, fast=False, verbose=False, 
252
 
                   dry_run=False, max_count=None, skip_symlinks=False):
 
344
 
 
345
###class Importer(object):
 
346
###    """An importer.
 
347
###
 
348
###    Currently this is used as a parameter object, though more behaviour is
 
349
###    possible later.
 
350
###    """
 
351
###
 
352
###    def __init__(self, output_dir, version, fast=False,
 
353
###                 verbose=False, dry_run=False, max_count=None,
 
354
###                   reuse_history_from=[]):
 
355
###        self.output_dir = output_dir
 
356
###        self.version = version
 
357
###        self.
 
358
 
 
359
 
 
360
def import_version(output_dir, version, encoding, fast=False,
 
361
                   verbose=False, dry_run=False, max_count=None,
 
362
                   reuse_history_from=[], standalone=True):
253
363
    """
254
364
    >>> q = test_environ()
 
365
 
 
366
    Progress bars output to stderr, but doctest does not capture that.
 
367
 
 
368
    >>> old_stderr = sys.stderr
 
369
    >>> sys.stderr = sys.stdout
 
370
 
255
371
    >>> result_path = os.path.join(q, "result")
256
372
    >>> commit_test_revisions()
257
373
    >>> version = pybaz.Version("test@example.com/test--test--0.1")
258
 
    >>> import_version('/', version, fancy=False, dry_run=True)
 
374
    >>> old_ui = bzrlib.ui.ui_factory
 
375
    >>> bzrlib.ui.ui_factory = bzrlib.ui.text.TextUIFactory(
 
376
    ...     bar_type=bzrlib.progress.DotsProgressBar)
 
377
 
 
378
    >>> import_version('/', version, None, dry_run=True)
259
379
    Traceback (most recent call last):
260
 
    UserError: / exists, but is not a bzr branch.
261
 
    >>> import_version(result_path, version, fancy=False, dry_run=True)
 
380
    NotPreviousImport: / is not the location of a previous import.
 
381
    >>> import_version(result_path, version, None, dry_run=True)
262
382
    Traceback (most recent call last):
263
383
    UserError: The version test@example.com/test--test--0.1 does not exist.
264
384
    >>> version = pybaz.Version("test@example.com/test--test--0")
265
 
    >>> import_version(result_path, version, fancy=False, dry_run=True)
266
 
    not fancy
267
 
    ....
 
385
    >>> import_version(result_path, version, None, dry_run=True) #doctest: +ELLIPSIS
 
386
    importing test@example.com/test--test--0 into ...
 
387
    ...
 
388
    revisions: ..........................................
268
389
    Dry run, not modifying output_dir
269
390
    Cleaning up
270
 
    >>> import_version(result_path, version, fancy=False)
271
 
    not fancy
272
 
    ....
 
391
    >>> import_version(result_path, version, None) #doctest: +ELLIPSIS
 
392
    importing test@example.com/test--test--0 into ...
 
393
    ...
 
394
    revisions: .....................................................................
273
395
    Cleaning up
274
396
    Import complete.
275
 
    >>> import_version(result_path, version, fancy=False)
 
397
    >>> import_version(result_path, version, None) #doctest: +ELLIPSIS
276
398
    Tree is up-to-date with test@example.com/test--test--0--patch-2
277
399
    >>> commit_more_test_revisions()
278
 
    >>> import_version(result_path, version, fancy=False)
279
 
    not fancy
280
 
    ..
 
400
    >>> import_version(result_path, version, None) #doctest: +ELLIPSIS
 
401
    importing test@example.com/test--test--0 into ...
 
402
    revisions: ....................................................
281
403
    Cleaning up
282
404
    Import complete.
 
405
    >>> bzrlib.ui.ui_factory = old_ui
 
406
    >>> sys.stderr = old_stderr
283
407
    >>> teardown_environ(q)
284
408
    """
285
 
    try:
286
 
        ancestors, old_revno = get_remaining_revisions(output_dir, version)
287
 
    except NotInABranch, e:
288
 
        raise UserError("%s exists, but is not a bzr branch." % e.path)
289
 
    if len(ancestors) == 0:
290
 
        last_revision = get_last_revision(find_branch(output_dir))
291
 
        print 'Tree is up-to-date with %s' % last_revision
292
 
        return
293
 
 
294
 
    progress_bar = ProgressBar()
295
 
    tempdir = tempfile.mkdtemp(prefix="baz2bzr-",
296
 
                               dir=os.path.dirname(output_dir))
297
 
    try:
298
 
        if not fancy:
299
 
            print "not fancy"
 
409
    progress_bar = bzrlib.ui.ui_factory.nested_progress_bar()
 
410
    try:
 
411
        try:
 
412
            ancestors, old_revno = get_remaining_revisions(output_dir, version,
 
413
                                                           encoding,
 
414
                                                           reuse_history_from)
 
415
        except NotBranchError, e:
 
416
            raise NotPreviousImport(e.path)
 
417
        if old_revno is None and len(ancestors) == 0:
 
418
            progress_bar.note('Version %s has no revisions.' % version)
 
419
            return
 
420
        if len(ancestors) == 0:
 
421
            last_revision, last_encoding = \
 
422
                get_last_revision(Branch.open(output_dir))
 
423
            progress_bar.note('Tree is up-to-date with %s' % last_revision)
 
424
            return
 
425
 
 
426
        progress_bar.note("importing %s into %s" % (version, output_dir))
 
427
 
 
428
        tempdir = tempfile.mkdtemp(prefix="baz2bzr-",
 
429
                                   dir=os.path.dirname(output_dir))
 
430
        try:
 
431
            wt = WorkingTree.open(output_dir)
 
432
        except (NotBranchError, NoWorkingTree):
 
433
            wt = None
300
434
        try:
301
435
            for result in iter_import_version(output_dir, ancestors, tempdir,
302
 
                    fast=fast, verbose=verbose, dry_run=dry_run, 
303
 
                    max_count=max_count, skip_symlinks=skip_symlinks):
304
 
                if fancy:
305
 
                    show_progress(progress_bar, result)
306
 
                else:
307
 
                    sys.stdout.write('.')
 
436
                    pb=progress_bar, encoding=encoding, fast=fast,
 
437
                    verbose=verbose, dry_run=dry_run, max_count=max_count,
 
438
                    standalone=standalone):
 
439
                show_progress(progress_bar, result)
 
440
            if dry_run:
 
441
                progress_bar.note('Dry run, not modifying output_dir')
 
442
                return
 
443
 
 
444
            # Update the working tree of the branch
 
445
            try:
 
446
                wt = WorkingTree.open(output_dir)
 
447
            except NoWorkingTree:
 
448
                wt = None
 
449
            if wt is not None:
 
450
                wt.set_last_revision(wt.branch.last_revision())
 
451
                wt.set_root_id(BAZ_IMPORT_ROOT)
 
452
                wt.revert()
 
453
 
308
454
        finally:
309
 
            if fancy:
310
 
                progress_bar.clear()
311
 
            else:
312
 
                sys.stdout.write('\n')
313
 
 
314
 
        if dry_run:
315
 
            print 'Dry run, not modifying output_dir'
316
 
            return
317
 
        if os.path.exists(output_dir):
318
 
            # Move the bzr control directory back, and update the working tree
319
 
            tmp_bzr_dir = os.path.join(tempdir, '.bzr')
320
 
            
321
 
            bzr_dir = os.path.join(output_dir, '.bzr')
322
 
            new_bzr_dir = os.path.join(tempdir, "rd", '.bzr')
323
 
 
324
 
            os.rename(bzr_dir, tmp_bzr_dir) # Move the original bzr out of the way
325
 
            os.rename(new_bzr_dir, bzr_dir)
326
 
            try:
327
 
                bzrlib.merge.merge((output_dir, -1), (output_dir, old_revno), 
328
 
                                   check_clean=False, this_dir=output_dir, 
329
 
                                   ignore_zero=True)
330
 
            except:
331
 
                # If something failed, move back the original bzr directory
332
 
                os.rename(bzr_dir, new_bzr_dir)
333
 
                os.rename(tmp_bzr_dir, bzr_dir)
334
 
                raise
335
 
        else:
336
 
            revdir = os.path.join(tempdir, "rd")
337
 
            os.rename(revdir, output_dir)
338
 
 
 
455
 
 
456
            progress_bar.note('Cleaning up')
 
457
            shutil.rmtree(tempdir)
 
458
        progress_bar.note("Import complete.")
339
459
    finally:
340
 
        print 'Cleaning up'
341
 
        shutil.rmtree(tempdir)
342
 
    print "Import complete."
343
 
            
344
 
class UserError(Exception):
 
460
        progress_bar.finished()
 
461
 
 
462
class UserError(BzrCommandError):
345
463
    def __init__(self, message):
346
464
        """Exception to throw when a user makes an impossible request
347
465
        :param message: The message to emit when printing this exception
348
466
        :type message: string
349
467
        """
350
 
        Exception.__init__(self, message)
351
 
 
352
 
def revision_id(arch_revision):
 
468
        BzrCommandError.__init__(self, message)
 
469
 
 
470
class NotPreviousImport(UserError):
 
471
    def __init__(self, path):
 
472
        UserError.__init__(self, "%s is not the location of a previous import."
 
473
                           % path)
 
474
 
 
475
 
 
476
def revision_id(arch_revision, encoding):
353
477
    """
354
478
    Generate a Bzr revision id from an Arch revision id.  'x' in the id
355
479
    designates a revision imported with an experimental algorithm.  A number
357
481
 
358
482
    :param arch_revision: The Arch revision to generate an ID for.
359
483
 
360
 
    >>> revision_id(pybaz.Revision("you@example.com/cat--br--0--base-0"))
361
 
    'Arch-x:you@example.com%cat--br--0--base-0'
 
484
    >>> revision_id(pybaz.Revision("you@example.com/cat--br--0--base-0"), None)
 
485
    'Arch-1:you@example.com%cat--br--0--base-0'
 
486
    >>> revision_id(pybaz.Revision("you@example.com/cat--br--0--base-0"), 'utf-8')
 
487
    'Arch-1-utf-8:you@example.com%cat--br--0--base-0'
362
488
    """
363
 
    return "Arch-x:%s" % str(arch_revision).replace('/', '%')
 
489
    if encoding is None:
 
490
        encoding = ''
 
491
    else:
 
492
        encoding = '-' + encoding
 
493
    return "Arch-1%s:%s" % (encoding, str(arch_revision).replace('/', '%'))
364
494
 
365
495
class NotArchRevision(Exception):
366
496
    def __init__(self, revision_id):
370
500
 
371
501
def arch_revision(revision_id):
372
502
    """
373
 
    >>> str(arch_revision("Arch-x:jrandom@example.com%test--test--0"))
374
 
    Traceback (most recent call last):
375
 
    NotArchRevision: The revision id Arch-x:jrandom@example.com%test--test--0 does not look like it came from Arch.
376
 
    >>> str(arch_revision("Arch-x:jrandom@example.com%test--test--0--base-5"))
377
 
    Traceback (most recent call last):
378
 
    NotArchRevision: The revision id Arch-x:jrandom@example.com%test--test--0--base-5 does not look like it came from Arch.
379
 
    >>> str(arch_revision("Arch-x:jrandom@example.com%test--test--0--patch-5"))
380
 
    'jrandom@example.com/test--test--0--patch-5'
 
503
    >>> str(arch_revision("Arch-1:jrandom@example.com%test--test--0"))
 
504
    Traceback (most recent call last):
 
505
    NotArchRevision: The revision id Arch-1:jrandom@example.com%test--test--0 does not look like it came from Arch.
 
506
    >>> str(arch_revision("Arch-1:jrandom@example.com%test--test--0--base-5"))
 
507
    Traceback (most recent call last):
 
508
    NotArchRevision: The revision id Arch-1:jrandom@example.com%test--test--0--base-5 does not look like it came from Arch.
 
509
    >>> str(arch_revision("Arch-1:jrandom@example.com%test--test--0--patch-5")[0])
 
510
    'jrandom@example.com/test--test--0--patch-5'
 
511
    >>> str(arch_revision("Arch-1:jrandom@example.com%test--test--0--patch-5")[0])
 
512
    'jrandom@example.com/test--test--0--patch-5'
 
513
    >>> str(arch_revision("Arch-1:jrandom@example.com%test--test--0--patch-5")[1])
 
514
    'None'
 
515
    >>> str(arch_revision("Arch-1-utf-8:jrandom@example.com%test--test--0--patch-5")[1])
 
516
    'utf-8'
 
517
    >>> str(arch_revision('null:'))
 
518
    '(None, None)'
381
519
    """
382
 
    if revision_id is None:
383
 
        return None
384
 
    if revision_id[:7] != 'Arch-x:':
 
520
    if is_null(revision_id):
 
521
        return None, None
 
522
    if revision_id[:7] not in ('Arch-1:', 'Arch-1-'):
385
523
        raise NotArchRevision(revision_id)
386
524
    else:
387
525
        try:
388
 
            return pybaz.Revision(revision_id[7:].replace('%', '/'))
 
526
            encoding, arch_name = revision_id[6:].split(':', 1)
 
527
            arch_name = arch_name.replace('%', '/')
 
528
            if encoding == '':
 
529
                encoding = None
 
530
            else:
 
531
                encoding = encoding[1:]
 
532
            return pybaz.Revision(arch_name), encoding
389
533
        except pybaz.errors.NamespaceError, e:
390
534
            raise NotArchRevision(revision_id)
391
 
            
392
 
def iter_import_version(output_dir, ancestors, tempdir, fast=False,
393
 
                        verbose=False, dry_run=False, max_count=None,
394
 
                        skip_symlinks=False):
 
535
 
 
536
 
 
537
def create_shared_repository(output_dir):
 
538
    bd = bzrdir.BzrDirMetaFormat1().initialize(output_dir)
 
539
    bd.create_repository(shared=True)
 
540
 
 
541
def create_branch(output_dir):
 
542
    os.mkdir(output_dir)
 
543
    bd = bzrdir.BzrDirMetaFormat1().initialize(output_dir)
 
544
    return bd.create_branch()
 
545
 
 
546
 
 
547
def create_checkout(source, to_location, revision_id=None):
 
548
    checkout = bzrdir.BzrDirMetaFormat1().initialize(to_location)
 
549
    bzrlib.branch.BranchReferenceFormat().initialize(checkout, source)
 
550
    return checkout.create_workingtree(revision_id)
 
551
 
 
552
 
 
553
def create_checkout_metadata(source, to_location, revision_id=None):
 
554
    if revision_id is None:
 
555
        revision_id = source.last_revision()
 
556
    wt = create_checkout(source, to_location, NULL_REVISION)
 
557
    wt.lock_write()
 
558
    try:
 
559
        wt.set_last_revision(revision_id)
 
560
        wt.flush()
 
561
        if revision_id not in (NULL_REVISION, None):
 
562
            basis = wt.basis_tree()
 
563
            basis.lock_read()
 
564
            try:
 
565
                wt._write_inventory(basis.inventory)
 
566
            finally:
 
567
                basis.unlock()
 
568
    finally:
 
569
        wt.unlock()
 
570
    return wt
 
571
 
 
572
 
 
573
def iter_import_version(output_dir, ancestors, tempdir, pb, encoding,
 
574
                        fast=False, verbose=False, dry_run=False,
 
575
                        max_count=None, standalone=False):
395
576
    revdir = None
 
577
    log_encoding = 'ascii'
 
578
    if encoding is not None:
 
579
        log_encoding = encoding
396
580
 
397
581
    # Uncomment this for testing, it basically just has baz2bzr only update
398
582
    # 5 patches at a time
409
593
    #        print '\t%s' % a
410
594
 
411
595
    previous_version=None
412
 
 
413
 
    for i in range(len(ancestors)):
414
 
        revision = ancestors[i]
 
596
    missing_ancestor = None
 
597
    if dry_run:
 
598
        dry_output_dir = os.path.join(tempdir, 'od')
 
599
        if os.path.exists(output_dir):
 
600
            shutil.copytree(output_dir, dry_output_dir)
 
601
        output_dir = dry_output_dir
 
602
 
 
603
    if os.path.exists(output_dir):
 
604
        target_branch = Branch.open(output_dir)
 
605
    else:
 
606
        if standalone:
 
607
            wt = BzrDir.create_standalone_workingtree(output_dir)
 
608
            target_branch = wt.branch
 
609
        else:
 
610
            target_branch = create_branch(output_dir)
 
611
 
 
612
    for i, revision in enumerate(ancestors):
 
613
        rev_id = revision_id(revision, encoding)
 
614
        direct_merges = []
415
615
        if verbose:
416
616
            version = str(revision.version)
417
617
            if version != previous_version:
418
 
                clear_progress_bar()
419
 
                print '\rOn version: %s' % version
 
618
                pb.note('On version: %s' % version)
420
619
            yield Progress(str(revision.patchlevel), i, len(ancestors))
421
620
            previous_version = version
422
621
        else:
423
622
            yield Progress("revisions", i, len(ancestors))
 
623
 
 
624
        if target_branch.repository.has_revision(rev_id):
 
625
            target_branch.set_last_revision_info(i+1, rev_id)
 
626
            continue
424
627
        if revdir is None:
425
628
            revdir = os.path.join(tempdir, "rd")
426
 
            baz_inv, log = get_revision(revdir, revision, 
427
 
                                        skip_symlinks=skip_symlinks)
428
 
            if os.path.exists(output_dir):
429
 
                bzr_dir = os.path.join(output_dir, '.bzr')
430
 
                new_bzr_dir = os.path.join(tempdir, "rd", '.bzr')
431
 
                # This would be much faster with a simple os.rename(), but if
432
 
                # we fail, we have corrupted the original .bzr directory.  Is
433
 
                # that a big problem, as we can just back out the last
434
 
                # revisions in .bzr/revision_history I don't really know
435
 
                shutil.copytree(bzr_dir, new_bzr_dir)
436
 
                # Now revdir should have a tree with the latest .bzr, and the
437
 
                # next revision of the baz tree
438
 
                branch = find_branch(revdir)
439
 
            else:
440
 
                branch = Branch(revdir, init=True)
 
629
            try:
 
630
                tree, baz_inv, log = get_revision(revdir, revision)
 
631
            except pybaz.errors.ExecProblem, e:
 
632
                if ("%s" % e.args).find('could not connect') == -1:
 
633
                    raise
 
634
                missing_ancestor = revision
 
635
                revdir = None
 
636
                pb.note("unable to access ancestor %s, making into a merge."
 
637
                       % missing_ancestor)
 
638
                continue
 
639
            target_tree = create_checkout_metadata(target_branch, revdir)
 
640
            branch = target_tree.branch
441
641
        else:
442
642
            old = os.path.join(revdir, ".bzr")
443
643
            new = os.path.join(tempdir, ".bzr")
444
644
            os.rename(old, new)
445
 
            baz_inv, log = apply_revision(revdir, revision, 
446
 
                                          skip_symlinks=skip_symlinks)
 
645
            baz_inv, log = apply_revision(tree, revision)
447
646
            os.rename(new, old)
448
 
            branch = find_branch(revdir)
449
 
        timestamp = email.Utils.mktime_tz(log.date + (0,))
450
 
        rev_id = revision_id(revision)
 
647
            target_tree = WorkingTree.open(revdir)
 
648
            branch = target_tree.branch
 
649
        # cached so we can delete the log
 
650
        log_date = log.date
 
651
        log_summary = log.summary
 
652
        log_description = log.description
 
653
        is_continuation = log.continuation_of is not None
 
654
        log_creator = log.creator
 
655
        direct_merges = get_direct_merges(revdir, revision, log)
 
656
 
 
657
        timestamp = email.Utils.mktime_tz(log_date + (0,))
 
658
        if log_summary is None:
 
659
            log_summary = ""
 
660
        # log_descriptions of None and "" are ignored.
 
661
        if not is_continuation and log_description:
 
662
            log_message = "\n".join((log_summary, log_description))
 
663
        else:
 
664
            log_message = log_summary
 
665
        target_tree.lock_write()
451
666
        branch.lock_write()
452
667
        try:
453
 
            branch.set_inventory(baz_inv)
454
 
            bzrlib.trace.silent = True
455
 
            branch.commit(log.summary, verbose=False, committer=log.creator,
456
 
                          timestamp=timestamp, timezone=0, rev_id=rev_id)
 
668
            if missing_ancestor:
 
669
                # if we want it to be in revision-history, do that here.
 
670
                target_tree.set_parent_ids(
 
671
                    [revision_id(missing_ancestor, encoding)],
 
672
                    allow_leftmost_as_ghost=True)
 
673
                missing_ancestor = None
 
674
            for merged_rev in direct_merges:
 
675
                target_tree.add_pending_merge(revision_id(merged_rev,
 
676
                                                          encoding))
 
677
            target_tree.set_root_id(BAZ_IMPORT_ROOT)
 
678
            target_tree.flush()
 
679
            target_tree.set_inventory(baz_inv)
 
680
            commitobj = Commit(reporter=ImportCommitReporter())
 
681
            commitobj.commit(working_tree=target_tree,
 
682
                message=log_message.decode(log_encoding, 'replace'),
 
683
                verbose=False, committer=log_creator, timestamp=timestamp,
 
684
                timezone=0, rev_id=rev_id, revprops={})
457
685
        finally:
458
 
            bzrlib.trace.silent = False   
 
686
            target_tree.unlock()
459
687
            branch.unlock()
460
688
    yield Progress("revisions", len(ancestors), len(ancestors))
461
 
    unlink_unversioned(branch, revdir)
462
 
 
463
 
def unlink_unversioned(branch, revdir):
464
 
    for unversioned in branch.working_tree().extras():
465
 
        path = os.path.join(revdir, unversioned)
 
689
 
 
690
def get_direct_merges(revdir, revision, log):
 
691
    continuation = log.continuation_of
 
692
    previous_version = revision.version
 
693
    if pybaz.WorkingTree(revdir).tree_version != previous_version:
 
694
        pybaz.WorkingTree(revdir).set_tree_version(previous_version)
 
695
    log_path = "%s/{arch}/%s/%s/%s/%s/patch-log/%s" % (revdir,
 
696
        revision.category.nonarch, revision.branch.nonarch,
 
697
        revision.version.nonarch, revision.archive, revision.patchlevel)
 
698
    temp_path = tempfile.mktemp(dir=os.path.dirname(revdir))
 
699
    os.rename(log_path, temp_path)
 
700
    merges = list(iter_new_merges(revdir, revision.version))
 
701
    direct = direct_merges(merges, [continuation])
 
702
    os.rename(temp_path, log_path)
 
703
    return direct
 
704
 
 
705
def unlink_unversioned(wt):
 
706
    for unversioned in wt.extras():
 
707
        path = wt.abspath(unversioned)
466
708
        if os.path.isdir(path):
467
709
            shutil.rmtree(path)
468
710
        else:
469
711
            os.unlink(path)
470
712
 
471
713
def get_log(tree, revision):
472
 
    log = tree.iter_logs(version=revision.version, reverse=True).next()
 
714
    log = pybaz.Patchlog(revision, tree=tree)
473
715
    assert str(log.revision) == str(revision), (log.revision, revision)
474
716
    return log
475
717
 
476
 
def get_revision(revdir, revision, skip_symlinks=False):
477
 
    revision.get(revdir)
478
 
    tree = pybaz.tree_root(revdir)
 
718
def get_revision(revdir, revision):
 
719
    tree = revision.get(revdir)
479
720
    log = get_log(tree, revision)
480
721
    try:
481
 
        return bzr_inventory_data(tree, skip_symlinks=skip_symlinks), log 
 
722
        return tree, bzr_inventory_data(tree), log
482
723
    except BadFileKind, e:
483
 
        raise UserError("Cannot convert %s because %s is a %s" % (revision,e.path, e.kind) )
484
 
 
485
 
 
486
 
def apply_revision(revdir, revision, skip_symlinks=False):
487
 
    tree = pybaz.tree_root(revdir)
 
724
        raise UserError("Cannot convert %s because %s is a %s" %
 
725
                        (revision,e.path, e.kind))
 
726
 
 
727
 
 
728
def apply_revision(tree, revision):
488
729
    revision.apply(tree)
489
730
    log = get_log(tree, revision)
490
731
    try:
491
 
        return bzr_inventory_data(tree, skip_symlinks=skip_symlinks), log
 
732
        return bzr_inventory_data(tree), log
492
733
    except BadFileKind, e:
493
 
        raise UserError("Cannot convert %s because %s is a %s" % (revision,e.path, e.kind) )
494
 
 
495
 
 
 
734
        raise UserError("Cannot convert %s because %s is a %s" %
 
735
                        (revision,e.path, e.kind))
496
736
 
497
737
 
498
738
class BadFileKind(Exception):
504
744
        Exception.__init__(self, "File %s is of forbidden type %s" %
505
745
                           (os.path.join(tree_root, path), kind))
506
746
 
507
 
def bzr_inventory_data(tree, skip_symlinks=False):
 
747
 
 
748
def bzr_inventory_data(tree):
508
749
    inv_iter = tree.iter_inventory_ids(source=True, both=True)
509
750
    inv_map = {}
510
751
    for arch_id, path in inv_iter:
511
 
        bzr_file_id = arch_id.replace('%', '%25').replace('/', '%2f')
512
 
        inv_map[path] = bzr_file_id 
 
752
        bzr_file_id = map_file_id(arch_id)
 
753
        inv_map[path] = bzr_file_id
513
754
 
514
755
    bzr_inv = []
515
756
    for path, file_id in inv_map.iteritems():
516
757
        full_path = os.path.join(tree, path)
517
758
        kind = bzrlib.osutils.file_kind(full_path)
518
 
        if skip_symlinks and kind == "symlink":
519
 
            continue
520
 
        if kind not in ("file", "directory"):
 
759
        if kind not in ("file", "directory", "symlink"):
521
760
            raise BadFileKind(tree, path, kind)
522
761
        parent_dir = os.path.dirname(path)
523
762
        if parent_dir != "":
528
767
    bzr_inv.sort()
529
768
    return bzr_inv
530
769
 
 
770
 
 
771
def baz_import_branch(to_location, from_branch, fast, max_count, verbose,
 
772
                      encoding, dry_run, reuse_history_list):
 
773
    to_location = os.path.realpath(str(to_location))
 
774
    if from_branch is not None:
 
775
        try:
 
776
            from_branch = pybaz.Version(from_branch)
 
777
        except pybaz.errors.NamespaceError:
 
778
            print "%s is not a valid Arch branch." % from_branch
 
779
            return 1
 
780
    if reuse_history_list is None:
 
781
        reuse_history_list = []
 
782
    import_version(to_location, from_branch, encoding, max_count=max_count,
 
783
                   reuse_history_from=reuse_history_list)
 
784
 
 
785
 
531
786
class NotInABranch(Exception):
532
787
    def __init__(self, path):
533
788
        Exception.__init__(self, "%s is not in a branch." % path)
534
789
        self.path = path
535
790
 
536
791
 
537
 
def find_branch(path):
538
 
    """
539
 
    >>> find_branch('/')
540
 
    Traceback (most recent call last):
541
 
    NotInABranch: / is not in a branch.
542
 
    >>> sb = bzrlib.ScratchBranch()
543
 
    >>> isinstance(find_branch(sb.base), Branch)
544
 
    True
545
 
    """
 
792
 
 
793
def baz_import(to_root_dir, from_archive, encoding, verbose=False,
 
794
               reuse_history_list=[], prefixes=None):
 
795
    if reuse_history_list is None:
 
796
        reuse_history_list = []
 
797
    to_root = str(os.path.realpath(to_root_dir))
 
798
    if not os.path.exists(to_root):
 
799
        os.mkdir(to_root)
 
800
    if prefixes is not None:
 
801
        prefixes = prefixes.split(':')
 
802
    import_archive(to_root, from_archive, verbose, encoding,
 
803
                   reuse_history_list, prefixes=prefixes)
 
804
 
 
805
 
 
806
def import_archive(to_root, from_archive, verbose,
 
807
                   encoding, reuse_history_from=[], standalone=False,
 
808
                   prefixes=None):
 
809
    def selected(version):
 
810
        if prefixes is None:
 
811
            return True
 
812
        else:
 
813
            for prefix in prefixes:
 
814
                if version.nonarch.startswith(prefix):
 
815
                    return True
 
816
            return False
 
817
    real_to = os.path.realpath(to_root)
 
818
    history_locations = [real_to] + reuse_history_from
 
819
    if standalone is False:
 
820
        try:
 
821
            bd = BzrDir.open(to_root)
 
822
            bd.find_repository()
 
823
        except NotBranchError:
 
824
            create_shared_repository(to_root)
 
825
        except NoRepositoryPresent:
 
826
            raise BzrCommandError("Can't create repository at existing branch.")
 
827
    versions = list(pybaz.Archive(str(from_archive)).iter_versions())
 
828
    progress_bar = bzrlib.ui.ui_factory.nested_progress_bar()
546
829
    try:
547
 
        return Branch(path)
548
 
    except BzrError, e:
549
 
        if e.args[0].endswith("' is not in a branch"):
550
 
            raise NotInABranch(path)
551
 
 
552
 
class cmd_baz_import(Command):
553
 
    """Import an Arch or Baz branch into a bzr branch"""
554
 
    takes_args = ['to_location', 'from_branch?']
555
 
    takes_options = ['verbose']
556
 
 
557
 
    def run(self, to_location, from_branch=None, skip_symlinks=False, 
558
 
            fast=False, max_count=None, verbose=False, dry_run=False):
559
 
        to_location = os.path.realpath(str(to_location))
560
 
        if from_branch is not None:
 
830
        for num, version in enumerate(versions):
 
831
            progress_bar.update("Branch", num, len(versions))
 
832
            if not selected(version):
 
833
                print "Skipping %s" % version
 
834
                continue
 
835
            target = os.path.join(to_root, map_namespace(version))
 
836
            if not os.path.exists(os.path.dirname(target)):
 
837
                os.makedirs(os.path.dirname(target))
561
838
            try:
562
 
                from_branch = pybaz.Version(from_branch)
563
 
            except pybaz.errors.NamespaceError:
564
 
                print "%s is not a valid Arch branch." % from_branch
565
 
                return 1
566
 
        import_version(to_location, from_branch)
 
839
                import_version(target, version, encoding,
 
840
                               reuse_history_from=reuse_history_from,
 
841
                               standalone=standalone)
 
842
            except pybaz.errors.ExecProblem,e:
 
843
                if str(e).find('The requested revision cannot be built.') != -1:
 
844
                    progress_bar.note(
 
845
                        "Skipping version %s as it cannot be built due"
 
846
                        " to a missing parent archive." % version)
 
847
                else:
 
848
                    raise
 
849
            except UserError, e:
 
850
                if str(e).find('already exists, and the last revision ') != -1:
 
851
                    progress_bar.note(
 
852
                        "Skipping version %s as it has had commits made"
 
853
                        " since it was converted to bzr." % version)
 
854
                else:
 
855
                    raise
 
856
    finally:
 
857
        progress_bar.finished()
 
858
 
 
859
 
 
860
def map_namespace(a_version):
 
861
    a_version = pybaz.Version("%s" % a_version)
 
862
    parser = NameParser(a_version)
 
863
    version = parser.get_version()
 
864
    branch = parser.get_branch()
 
865
    category = parser.get_category()
 
866
    if branch is None or branch == '':
 
867
        branch = "+trunk"
 
868
    if version == '0':
 
869
        return "%s/%s" % (category, branch)
 
870
    return "%s/%s/%s" % (category, version, branch)
 
871
 
 
872
 
 
873
def map_file_id(file_id):
 
874
    """Convert a baz file id to a bzr one."""
 
875
    return file_id.replace('%', '%25').replace('/', '%2f')