def create_from_data(self,
repository,
diff_file_name,
diff_file_contents,
parent_diff_file_name=None,
parent_diff_file_contents=None,
diffset_history=None,
basedir=None,
request=None,
base_commit_id=None,
check_existence=True,
validate_only=False,
**kwargs):
"""Create a DiffSet from raw diff data.
This parses a diff and optional parent diff covering one or more files,
validates, and constructs :py:class:`DiffSets
<reviewboard.diffviewer.models.DiffSet>` and :py:class:`FileDiffs
<reviewboard.diffviewer.models.FileDiff>` representing the diff.
This can optionally validate the diff without saving anything to the
database. In this case, no value will be returned. Instead, callers
should take any result as success.
Args:
repository (reviewboard.scmtools.models.Repository):
The repository the diff applies to.
diff_file_name (unicode):
The filename of the main diff file.
diff_file_contents (bytes):
The contents of the main diff file.
parent_diff_file_name (unicode, optional):
The filename of the parent diff, if one is provided.
parent_diff_file_contents (bytes, optional):
The contents of the parent diff, if one is provided.
diffset_history (reviewboard.diffviewer.models.DiffSetHistory, optional):
The history object to associate the DiffSet with. This is
not required if using ``validate_only=True``.
basedir (unicode, optional):
The base directory to prepend to all file paths in the diff.
request (django.http.HttpRequest, optional):
The current HTTP request, if any. This will result in better
logging.
base_commit_id (unicode, optional):
The ID of the commit that the diff is based upon. This is
needed by some SCMs or hosting services to properly look up
files, if the diffs represent blob IDs instead of commit IDs
and the service doesn't support those lookups.
check_existence (bool, optional):
Whether to check for file existence as part of the validation
process. This defaults to ``True``.
validate_only (bool, optional):
Whether to just validate and not save. If ``True``, then this
won't populate the database at all and will return ``None``
upon success. This defaults to ``False``.
Returns:
reviewboard.diffviewer.models.DiffSet:
The resulting DiffSet stored in the database, if processing
succeeded and ``validate_only=False``.
Raises:
reviewboard.diffviewer.errors.DiffParserError:
There was an error parsing the main diff or parent diff.
reviewboard.diffviewer.errors.EmptyDiffError:
The provided diff file did not contain any file changes.
reviewboard.scmtools.core.FileNotFoundError:
A file specified in the diff could not be found in the
repository.
reviewboard.scmtools.core.SCMError:
There was an error talking to the repository when validating
the existence of a file.
reviewboard.scmtools.git.ShortSHA1Error:
A SHA1 specified in the diff was in the short form, which
could not be used to look up the file. This is applicable only
to Git.
"""
from reviewboard.diffviewer.diffutils import convert_to_unicode
from reviewboard.diffviewer.models import FileDiff
if 'save' in kwargs:
warnings.warn(
'The save parameter to '
'DiffSet.objects.create_from_data is deprecated. '
'Please set validate_only instead.', DeprecationWarning)
validate_only = not kwargs['save']
tool = repository.get_scmtool()
parser = tool.get_parser(diff_file_contents)
files = list(
self._process_files(parser,
basedir,
repository,
base_commit_id,
request,
check_existence=check_existence
and not parent_diff_file_contents))
# Parse the diff
if len(files) == 0:
raise EmptyDiffError(_("The diff file is empty"))
# Sort the files so that header files come before implementation.
files.sort(cmp=self._compare_files, key=lambda f: f.origFile)
# Parse the parent diff
parent_files = {}
# This is used only for tools like Mercurial that use atomic changeset
# IDs to identify all file versions but not individual file version
# IDs.
parent_commit_id = None
if parent_diff_file_contents:
diff_filenames = set([f.origFile for f in files])
parent_parser = tool.get_parser(parent_diff_file_contents)
# If the user supplied a base diff, we need to parse it and
# later apply each of the files that are in the main diff
for f in self._process_files(parent_parser,
basedir,
repository,
base_commit_id,
request,
check_existence=check_existence,
limit_to=diff_filenames):
parent_files[f.newFile] = f
# This will return a non-None value only for tools that use
# commit IDs to identify file versions as opposed to file revision
# IDs.
parent_commit_id = parent_parser.get_orig_commit_id()
diffset = self.model(name=diff_file_name,
revision=0,
basedir=basedir,
history=diffset_history,
repository=repository,
diffcompat=DiffCompatVersion.DEFAULT,
base_commit_id=base_commit_id)
if not validate_only:
diffset.save()
encoding_list = repository.get_encoding_list()
filediffs = []
for f in files:
parent_file = None
orig_rev = None
parent_content = b''
if f.origFile in parent_files:
parent_file = parent_files[f.origFile]
parent_content = parent_file.data
orig_rev = parent_file.origInfo
# If there is a parent file there is not necessarily an original
# revision for the parent file in the case of a renamed file in
# git.
if not orig_rev:
if parent_commit_id and f.origInfo != PRE_CREATION:
orig_rev = parent_commit_id
else:
orig_rev = f.origInfo
enc, orig_file = convert_to_unicode(f.origFile, encoding_list)
enc, dest_file = convert_to_unicode(f.newFile, encoding_list)
if f.deleted:
status = FileDiff.DELETED
elif f.moved:
status = FileDiff.MOVED
elif f.copied:
status = FileDiff.COPIED
else:
status = FileDiff.MODIFIED
filediff = FileDiff(
diffset=diffset,
source_file=parser.normalize_diff_filename(orig_file),
dest_file=parser.normalize_diff_filename(dest_file),
source_revision=smart_unicode(orig_rev),
dest_detail=f.newInfo,
binary=f.binary,
status=status)
filediff.extra_data = {
'is_symlink': f.is_symlink,
}
if (parent_file and (parent_file.moved or parent_file.copied)
and parent_file.insert_count == 0
and parent_file.delete_count == 0):
filediff.extra_data['parent_moved'] = True
if not validate_only:
# This state all requires making modifications to the database.
# We only want to do this if we're saving.
filediff.diff = f.data
filediff.parent_diff = parent_content
filediff.set_line_counts(raw_insert_count=f.insert_count,
raw_delete_count=f.delete_count)
filediffs.append(filediff)
if validate_only:
return None
if filediffs:
FileDiff.objects.bulk_create(filediffs)
return diffset
def create_filediffs(diff_file_contents, parent_diff_file_contents,
repository, basedir, base_commit_id, diffset,
request=None, check_existence=True, get_file_exists=None,
diffcommit=None, validate_only=False):
"""Create FileDiffs from the given data.
Args:
diff_file_contents (bytes):
The contents of the diff file.
parent_diff_file_contents (bytes):
The contents of the parent diff file.
repository (reviewboard.scmtools.models.Repository):
The repository the diff is being posted against.
basedir (unicode):
The base directory to prepend to all file paths in the diff.
base_commit_id (unicode):
The ID of the commit that the diff is based upon. This is
needed by some SCMs or hosting services to properly look up
files, if the diffs represent blob IDs instead of commit IDs
and the service doesn't support those lookups.
diffset (reviewboard.diffviewer.models.diffset.DiffSet):
The DiffSet to attach the created FileDiffs to.
request (django.http.HttpRequest, optional):
The current HTTP request.
check_existence (bool, optional):
Whether or not existence checks should be performed against
the upstream repository.
This argument defaults to ``True``.
get_file_exists (callable, optional):
A callable that is used to determine if a file exists.
This must be provided if ``check_existence`` is ``True``.
diffcommit (reviewboard.diffviewer.models.diffcommit.DiffCommit,
optional):
The Diffcommit to attach the created FileDiffs to.
validate_only (bool, optional):
Whether to just validate and not save. If ``True``, then this
won't populate the database at all and will return ``None``
upon success. This defaults to ``False``.
Returns:
list of reviewboard.diffviewer.models.filediff.FileDiff:
The created FileDiffs.
If ``validate_only`` is ``True``, the returned list will be empty.
"""
from reviewboard.diffviewer.diffutils import convert_to_unicode
from reviewboard.diffviewer.models import FileDiff
files, parser, parent_commit_id, parent_files = _prepare_file_list(
diff_file_contents=diff_file_contents,
parent_diff_file_contents=parent_diff_file_contents,
repository=repository,
request=request,
basedir=basedir,
check_existence=check_existence,
get_file_exists=get_file_exists,
base_commit_id=base_commit_id)
encoding_list = repository.get_encoding_list()
filediffs = []
for f in files:
parent_file = None
orig_rev = None
parent_content = b''
if f.origFile in parent_files:
parent_file = parent_files[f.origFile]
parent_content = parent_file.data
orig_rev = parent_file.origInfo
# If there is a parent file there is not necessarily an original
# revision for the parent file in the case of a renamed file in
# git.
if not orig_rev:
if parent_commit_id and f.origInfo != PRE_CREATION:
orig_rev = parent_commit_id
else:
orig_rev = f.origInfo
orig_file = convert_to_unicode(f.origFile, encoding_list)[1]
dest_file = convert_to_unicode(f.newFile, encoding_list)[1]
if f.deleted:
status = FileDiff.DELETED
elif f.moved:
status = FileDiff.MOVED
elif f.copied:
status = FileDiff.COPIED
else:
status = FileDiff.MODIFIED
filediff = FileDiff(
diffset=diffset,
commit=diffcommit,
source_file=parser.normalize_diff_filename(orig_file),
dest_file=parser.normalize_diff_filename(dest_file),
source_revision=force_text(orig_rev),
dest_detail=f.newInfo,
binary=f.binary,
status=status)
filediff.extra_data = {
'is_symlink': f.is_symlink,
}
if parent_file:
if (parent_file.insert_count == 0 and
parent_file.delete_count == 0):
filediff.extra_data[FileDiff._IS_PARENT_EMPTY_KEY] = True
if parent_file.moved or parent_file.copied:
filediff.extra_data['parent_moved'] = True
else:
filediff.extra_data[FileDiff._IS_PARENT_EMPTY_KEY] = False
if not validate_only:
# This state all requires making modifications to the database.
# We only want to do this if we're saving.
filediff.diff = f.data
filediff.parent_diff = parent_content
filediff.set_line_counts(raw_insert_count=f.insert_count,
raw_delete_count=f.delete_count)
filediffs.append(filediff)
if not validate_only:
FileDiff.objects.bulk_create(filediffs)
num_filediffs = len(filediffs)
return filediffs
def create_filediffs(diff_file_contents, parent_diff_file_contents,
repository, basedir, base_commit_id, diffset,
request=None, check_existence=True, get_file_exists=None,
diffcommit=None, validate_only=False):
"""Create FileDiffs from the given data.
Args:
diff_file_contents (bytes):
The contents of the diff file.
parent_diff_file_contents (bytes):
The contents of the parent diff file.
repository (reviewboard.scmtools.models.Repository):
The repository the diff is being posted against.
basedir (unicode):
The base directory to prepend to all file paths in the diff.
base_commit_id (unicode):
The ID of the commit that the diff is based upon. This is
needed by some SCMs or hosting services to properly look up
files, if the diffs represent blob IDs instead of commit IDs
and the service doesn't support those lookups.
diffset (reviewboard.diffviewer.models.diffset.DiffSet):
The DiffSet to attach the created FileDiffs to.
request (django.http.HttpRequest, optional):
The current HTTP request.
check_existence (bool, optional):
Whether or not existence checks should be performed against
the upstream repository.
This argument defaults to ``True``.
get_file_exists (callable, optional):
A callable that is used to determine if a file exists.
This must be provided if ``check_existence`` is ``True``.
diffcommit (reviewboard.diffviewer.models.diffcommit.DiffCommit,
optional):
The Diffcommit to attach the created FileDiffs to.
validate_only (bool, optional):
Whether to just validate and not save. If ``True``, then this
won't populate the database at all and will return ``None``
upon success. This defaults to ``False``.
Returns:
list of reviewboard.diffviewer.models.filediff.FileDiff:
The created FileDiffs.
If ``validate_only`` is ``True``, the returned list will be empty.
"""
from reviewboard.diffviewer.diffutils import convert_to_unicode
from reviewboard.diffviewer.models import FileDiff
files, parser, parent_commit_id, parent_files = _prepare_file_list(
diff_file_contents=diff_file_contents,
parent_diff_file_contents=parent_diff_file_contents,
repository=repository,
request=request,
basedir=basedir,
check_existence=check_existence,
get_file_exists=get_file_exists,
base_commit_id=base_commit_id)
encoding_list = repository.get_encoding_list()
filediffs = []
for f in files:
parent_file = None
orig_rev = None
parent_content = b''
if f.origFile in parent_files:
parent_file = parent_files[f.origFile]
parent_content = parent_file.data
orig_rev = parent_file.origInfo
# If there is a parent file there is not necessarily an original
# revision for the parent file in the case of a renamed file in
# git.
if not orig_rev:
if parent_commit_id and f.origInfo != PRE_CREATION:
orig_rev = parent_commit_id
else:
orig_rev = f.origInfo
orig_file = convert_to_unicode(f.origFile, encoding_list)[1]
dest_file = convert_to_unicode(f.newFile, encoding_list)[1]
if f.deleted:
status = FileDiff.DELETED
elif f.moved:
status = FileDiff.MOVED
elif f.copied:
status = FileDiff.COPIED
else:
status = FileDiff.MODIFIED
filediff = FileDiff(
diffset=diffset,
commit=diffcommit,
source_file=parser.normalize_diff_filename(orig_file),
dest_file=parser.normalize_diff_filename(dest_file),
source_revision=smart_unicode(orig_rev),
dest_detail=f.newInfo,
binary=f.binary,
status=status)
filediff.extra_data = {
'is_symlink': f.is_symlink,
}
if parent_file:
if (parent_file.insert_count == 0 and
parent_file.delete_count == 0):
filediff.extra_data[FileDiff._IS_PARENT_EMPTY_KEY] = True
if parent_file.moved or parent_file.copied:
filediff.extra_data['parent_moved'] = True
else:
filediff.extra_data[FileDiff._IS_PARENT_EMPTY_KEY] = False
if not validate_only:
# This state all requires making modifications to the database.
# We only want to do this if we're saving.
filediff.diff = f.data
filediff.parent_diff = parent_content
filediff.set_line_counts(raw_insert_count=f.insert_count,
raw_delete_count=f.delete_count)
filediffs.append(filediff)
if not validate_only:
FileDiff.objects.bulk_create(filediffs)
num_filediffs = len(filediffs)
if diffset.file_count is None:
diffset.reinit_file_count()
else:
diffset.file_count += num_filediffs
diffset.save(update_fields=('file_count',))
if diffcommit is not None:
diffcommit.file_count = len(filediffs)
diffcommit.save(update_fields=('file_count',))
return filediffs
def create_filediffs(diff_file_contents,
parent_diff_file_contents,
repository,
basedir,
base_commit_id,
diffset,
request=None,
check_existence=True,
get_file_exists=None,
diffcommit=None,
validate_only=False):
"""Create FileDiffs from the given data.
Args:
diff_file_contents (bytes):
The contents of the diff file.
parent_diff_file_contents (bytes):
The contents of the parent diff file.
repository (reviewboard.scmtools.models.Repository):
The repository the diff is being posted against.
basedir (unicode):
The base directory to prepend to all file paths in the diff.
base_commit_id (unicode):
The ID of the commit that the diff is based upon. This is
needed by some SCMs or hosting services to properly look up
files, if the diffs represent blob IDs instead of commit IDs
and the service doesn't support those lookups.
diffset (reviewboard.diffviewer.models.diffset.DiffSet):
The DiffSet to attach the created FileDiffs to.
request (django.http.HttpRequest, optional):
The current HTTP request.
check_existence (bool, optional):
Whether or not existence checks should be performed against
the upstream repository.
This argument defaults to ``True``.
get_file_exists (callable, optional):
A callable that is used to determine if a file exists.
This must be provided if ``check_existence`` is ``True``.
diffcommit (reviewboard.diffviewer.models.diffcommit.DiffCommit,
optional):
The DiffCommit to attach the created FileDiffs to.
validate_only (bool, optional):
Whether to just validate and not save. If ``True``, then this
won't populate the database at all and will return ``None``
upon success. This defaults to ``False``.
Returns:
list of reviewboard.diffviewer.models.filediff.FileDiff:
The created FileDiffs.
If ``validate_only`` is ``True``, the returned list will be empty.
"""
from reviewboard.diffviewer.diffutils import convert_to_unicode
from reviewboard.diffviewer.models import FileDiff
diff_info = _prepare_diff_info(
diff_file_contents=diff_file_contents,
parent_diff_file_contents=parent_diff_file_contents,
repository=repository,
request=request,
basedir=basedir,
check_existence=check_existence,
get_file_exists=get_file_exists,
base_commit_id=base_commit_id)
parent_files = diff_info['parent_files']
parsed_diff = diff_info['parsed_diff']
parsed_parent_diff = diff_info['parsed_parent_diff']
parser = diff_info['parser']
encoding_list = repository.get_encoding_list()
# Copy over any extra_data for the DiffSet and DiffCommit, if any were
# set by the parser.
#
# We'll do this even if we're validating, to ensure the data can be
# copied over fine.
main_extra_data = deepcopy(parsed_diff.extra_data)
change_extra_data = deepcopy(parsed_diff.changes[0].extra_data)
if change_extra_data:
if diffcommit is not None:
# We've already checked in _parse_diff that there's only a single
# change in the diff, so we can assume that here.
diffcommit.extra_data.update(change_extra_data)
else:
main_extra_data['change_extra_data'] = change_extra_data
if main_extra_data:
diffset.extra_data.update(main_extra_data)
if parsed_parent_diff is not None:
parent_extra_data = deepcopy(parsed_parent_diff.extra_data)
parent_change_extra_data = deepcopy(
parsed_parent_diff.changes[0].extra_data)
if parent_change_extra_data:
if diffcommit is not None:
diffcommit.extra_data['parent_extra_data'] = \
parent_change_extra_data
else:
parent_extra_data['change_extra_data'] = \
parent_change_extra_data
if parent_extra_data:
diffset.extra_data['parent_extra_data'] = parent_extra_data
# Convert the list of parsed files into FileDiffs.
filediffs = []
for f in diff_info['files']:
parent_file = None
parent_content = b''
extra_data = f.extra_data.copy()
if parsed_parent_diff is not None:
parent_file = parent_files.get(f.orig_filename)
if parent_file is not None:
parent_content = parent_file.data
# Store the information on the parent's filename and revision.
# It's important we force these to text, since they may be
# byte strings and the revision may be a Revision instance.
parent_source_filename = parent_file.orig_filename
parent_source_revision = parent_file.orig_file_details
parent_is_empty = (parent_file.insert_count == 0
and parent_file.delete_count == 0)
if parent_file.moved or parent_file.copied:
extra_data['parent_moved'] = True
if parent_file.extra_data:
extra_data['parent_extra_data'] = \
parent_file.extra_data.copy()
else:
# We don't have an entry, but we still want to record the
# parent ID, so we have something in common for all the files
# when looking up the source revision to fetch from the
# repository.
parent_is_empty = True
parent_source_filename = f.orig_filename
parent_source_revision = f.orig_file_details
if (parent_source_revision != PRE_CREATION
and parsed_diff.uses_commit_ids_as_revisions):
# Since the file wasn't explicitly provided in the parent
# diff, but the ParsedDiff says that commit IDs are used
# as revisions, we can use its parent commit ID as the
# parent revision here.
parent_commit_id = \
parsed_parent_diff.changes[0].parent_commit_id
assert parent_commit_id
parent_source_revision = parent_commit_id
# Store the information on the parent's filename and revision.
# It's important we force these to text, since they may be
# byte strings and the revision may be a Revision instance.
#
# Starting in Review Board 4.0.5, we store this any time there's
# a parent diff, whether or not the file existed in the parent
# diff.
extra_data.update({
FileDiff._IS_PARENT_EMPTY_KEY:
parent_is_empty,
'parent_source_filename':
convert_to_unicode(parent_source_filename, encoding_list)[1],
'parent_source_revision':
convert_to_unicode(parent_source_revision, encoding_list)[1],
})
orig_file = convert_to_unicode(f.orig_filename, encoding_list)[1]
dest_file = convert_to_unicode(f.modified_filename, encoding_list)[1]
if f.deleted:
status = FileDiff.DELETED
elif f.moved:
status = FileDiff.MOVED
elif f.copied:
status = FileDiff.COPIED
else:
status = FileDiff.MODIFIED
filediff = FileDiff(
diffset=diffset,
commit=diffcommit,
source_file=parser.normalize_diff_filename(orig_file),
dest_file=parser.normalize_diff_filename(dest_file),
source_revision=force_text(f.orig_file_details),
dest_detail=force_text(f.modified_file_details),
binary=f.binary,
status=status,
extra_data=extra_data)
# Set this unconditionally, for backwards-compatibility purposes.
# Review Board 4.0.6 introduced attribute wrappers in FileDiff and
# introduced symlink targets. We ideally would not set this unless
# it's True, but we don't want to risk breaking any assumptions on
# its presence at this time.
filediff.is_symlink = f.is_symlink
if f.is_symlink:
if f.old_symlink_target:
filediff.old_symlink_target = \
convert_to_unicode(f.old_symlink_target, encoding_list)[1]
if f.new_symlink_target:
filediff.new_symlink_target = \
convert_to_unicode(f.new_symlink_target, encoding_list)[1]
filediff.old_unix_mode = f.old_unix_mode
filediff.new_unix_mode = f.new_unix_mode
if not validate_only:
# This state all requires making modifications to the database.
# We only want to do this if we're saving.
filediff.diff = f.data
filediff.parent_diff = parent_content
filediff.set_line_counts(raw_insert_count=f.insert_count,
raw_delete_count=f.delete_count)
filediffs.append(filediff)
if not validate_only:
FileDiff.objects.bulk_create(filediffs)
if diffset.extra_data:
diffset.save(update_fields=('extra_data', ))
if diffcommit is not None and diffcommit.extra_data:
diffcommit.save(update_fields=('extra_data', ))
return filediffs
