def is_same_line_of_development(rev1, rev2):
"""Return True if rev1 and rev2 are on the same line of
development (i.e., both on trunk, or both on the same branch);
return False otherwise. Either rev1 or rev2 can be None, in
which case automatically return False."""
if rev1 is None or rev2 is None:
return False
if is_trunk_revision(rev1) and is_trunk_revision(rev2):
# Trunk revisions have to be handled specially because the main
# trunk version number can be changed; e.g., from 1 to 2.
return True
if rev1[0:rev1.rfind('.')] == rev2[0:rev2.rfind('.')]:
return True
return False
def is_same_line_of_development(rev1, rev2):
"""Return True if rev1 and rev2 are on the same line of
development (i.e., both on trunk, or both on the same branch);
return False otherwise. Either rev1 or rev2 can be None, in
which case automatically return False."""
if rev1 is None or rev2 is None:
return False
if is_trunk_revision(rev1) and is_trunk_revision(rev2):
# Trunk revisions have to be handled specially because the main
# trunk version number can be changed; e.g., from 1 to 2.
return True
if rev1[0:rev1.rfind('.')] == rev2[0:rev2.rfind('.')]:
return True
return False
def set_revision_info(self, revision, log, text):
if revision in self.revisions_seen:
# One common form of CVS repository corruption is that the
# Deltatext block for revision 1.1 appears twice. CollectData
# has already warned about this problem; here we can just ignore
# it.
return
else:
self.revisions_seen.add(revision)
cvs_rev_id = self.cvs_file_items.original_ids[revision]
if is_trunk_revision(revision):
# On trunk, revisions are encountered in reverse order (1.<N>
# ... 1.1) and deltas are inverted. The first text that we see
# is the fulltext for the HEAD revision. After that, the text
# corresponding to revision 1.N is the delta (1.<N+1> ->
# 1.<N>)). We have to invert the deltas here so that we can
# read the revisions out in dependency order; that is, for
# revision 1.1 we want the fulltext, and for revision 1.<N> we
# want the delta (1.<N-1> -> 1.<N>). This means that we can't
# compute the delta for a revision until we see its logical
# parent. When we finally see revision 1.1 (which is recognized
# because it doesn't have a parent), we can record the diff (1.1
# -> 1.2) for revision 1.2, and also the fulltext for 1.1.
if revision == self.head_revision:
# This is HEAD, as fulltext. Initialize the RCSStream so
# that we can compute deltas backwards in time.
self._rcs_stream = RCSStream(text)
self._rcs_stream_revision = revision
else:
# Any other trunk revision is a backward delta. Apply the
# delta to the RCSStream to mutate it to the contents of this
# revision, and also to get the reverse delta, which we store
# as the forward delta of our child revision.
try:
text = self._rcs_stream.invert_diff(text)
except MalformedDeltaException, e:
logger.error(
'Malformed RCS delta in %s, revision %s: %s' %
(self.cvs_file_items.cvs_file.rcs_path, revision, e))
raise RuntimeError()
text_record = DeltaTextRecord(
self.cvs_file_items.original_ids[
self._rcs_stream_revision], cvs_rev_id)
self.revision_collector._writeout(text_record, text)
self._rcs_stream_revision = revision
if revision == self.revision_1_1:
# This is revision 1.1. Write its fulltext:
text_record = FullTextRecord(cvs_rev_id)
self.revision_collector._writeout(text_record,
self._rcs_stream.get_text())
# There will be no more trunk revisions delivered, so free the
# RCSStream.
del self._rcs_stream
del self._rcs_stream_revision
def define_revision(self, revision, timestamp, author, state, branches,
next):
"""This is a callback method declared in Sink."""
for branch in branches:
try:
branch_data = self.sdc.rev_to_branch_data(branch)
except KeyError:
# Normally we learn about the branches from the branch names
# and numbers parsed from the symbolic name header. But this
# must have been an unlabeled branch that slipped through the
# net. Generate a name for it and create a _BranchData record
# for it now.
branch_data = self.sdc._add_unlabeled_branch(
self.sdc.rev_to_branch_number(branch))
assert branch_data.child is None
branch_data.child = branch
if revision in self._rev_data:
# This revision has already been seen.
logger.error(
'File %r contains duplicate definitions of revision %s.' % (
self.cvs_file.rcs_path,
revision,
))
raise RuntimeError()
# Record basic information about the revision:
rev_data = _RevisionData(self.collect_data.item_key_generator.gen_id(),
revision, int(timestamp), author, state)
self._rev_data[revision] = rev_data
# When on trunk, the RCS 'next' revision number points to what
# humans might consider to be the 'previous' revision number. For
# example, 1.3's RCS 'next' is 1.2.
#
# However, on a branch, the RCS 'next' revision number really does
# point to what humans would consider to be the 'next' revision
# number. For example, 1.1.2.1's RCS 'next' would be 1.1.2.2.
#
# In other words, in RCS, 'next' always means "where to find the next
# deltatext that you need this revision to retrieve.
#
# That said, we don't *want* RCS's behavior here, so we determine
# whether we're on trunk or a branch and set the dependencies
# accordingly.
if next:
if is_trunk_revision(revision):
self._primary_dependencies.append((
next,
revision,
))
else:
self._primary_dependencies.append((
revision,
next,
))
def define_revision(self, revision, timestamp, author, state, branches,
next):
if next:
self.base_revisions[next] = revision
else:
if is_trunk_revision(revision):
self.revision_1_1 = revision
for branch in branches:
self.base_revisions[branch] = revision
def rev_to_lod(self, revision):
"""Return the line of development on which REVISION lies.
REVISION must be a revision number with an even number of
components. Raise KeyError iff REVISION is unknown."""
if is_trunk_revision(revision):
return self.pdc.trunk
else:
return self.rev_to_branch_data(revision).symbol
def rev_to_branch_data(self, revision):
"""Return the branch_data of the branch on which REVISION lies.
REVISION must be a branch revision number with an even number of
components; for example '1.7.2.1' (never '1.7.2' nor '1.7.0.2').
Raise KeyError iff REVISION is unknown."""
assert not is_trunk_revision(revision)
return self.branches_data[self.rev_to_branch_number(revision)]
def rev_to_lod(self, revision):
"""Return the line of development on which REVISION lies.
REVISION must be a revision number with an even number of
components. Raise KeyError iff REVISION is unknown."""
if is_trunk_revision(revision):
return self.pdc.trunk
else:
return self.rev_to_branch_data(revision).symbol
def rev_to_branch_data(self, revision):
"""Return the branch_data of the branch on which REVISION lies.
REVISION must be a branch revision number with an even number of
components; for example '1.7.2.1' (never '1.7.2' nor '1.7.0.2').
Raise KeyError iff REVISION is unknown."""
assert not is_trunk_revision(revision)
return self.branches_data[self.rev_to_branch_number(revision)]
def rev_to_branch_number(revision):
"""Return the branch_number of the branch on which REVISION lies.
REVISION is a branch revision number with an even number of
components; for example '1.7.2.1' (never '1.7.2' nor '1.7.0.2').
The return value is the branch number (for example, '1.7.2').
Return none iff REVISION is a trunk revision such as '1.2'."""
if is_trunk_revision(revision):
return None
return revision[:revision.rindex(".")]
def rev_to_branch_number(revision):
"""Return the branch_number of the branch on which REVISION lies.
REVISION is a branch revision number with an even number of
components; for example '1.7.2.1' (never '1.7.2' nor '1.7.0.2').
The return value is the branch number (for example, '1.7.2').
Return none iff REVISION is a trunk revision such as '1.2'."""
if is_trunk_revision(revision):
return None
return revision[:revision.rindex(".")]
def define_revision(
self, revision, timestamp, author, state, branches, next
):
if next:
self.base_revisions[next] = revision
else:
if is_trunk_revision(revision):
self.revision_1_1 = revision
for branch in branches:
self.base_revisions[branch] = revision
def define_revision(self, revision, timestamp, author, state,
branches, next):
"""This is a callback method declared in Sink."""
for branch in branches:
try:
branch_data = self.sdc.rev_to_branch_data(branch)
except KeyError:
# Normally we learn about the branches from the branch names
# and numbers parsed from the symbolic name header. But this
# must have been an unlabeled branch that slipped through the
# net. Generate a name for it and create a _BranchData record
# for it now.
branch_data = self.sdc._add_unlabeled_branch(
self.sdc.rev_to_branch_number(branch))
assert branch_data.child is None
branch_data.child = branch
if revision in self._rev_data:
# This revision has already been seen.
logger.error('File %r contains duplicate definitions of revision %s.'
% (self.cvs_file.rcs_path, revision,))
raise RuntimeError()
# Record basic information about the revision:
rev_data = _RevisionData(
self.collect_data.item_key_generator.gen_id(),
revision, int(timestamp), author, state)
self._rev_data[revision] = rev_data
# When on trunk, the RCS 'next' revision number points to what
# humans might consider to be the 'previous' revision number. For
# example, 1.3's RCS 'next' is 1.2.
#
# However, on a branch, the RCS 'next' revision number really does
# point to what humans would consider to be the 'next' revision
# number. For example, 1.1.2.1's RCS 'next' would be 1.1.2.2.
#
# In other words, in RCS, 'next' always means "where to find the next
# deltatext that you need this revision to retrieve.
#
# That said, we don't *want* RCS's behavior here, so we determine
# whether we're on trunk or a branch and set the dependencies
# accordingly.
if next:
if is_trunk_revision(revision):
self._primary_dependencies.append( (next, revision,) )
else:
self._primary_dependencies.append( (revision, next,) )
def set_revision_info(self, revision, log, text):
"""This is a callback method declared in Sink."""
rev_data = self._rev_data[revision]
cvs_rev = self._cvs_file_items[rev_data.cvs_rev_id]
if cvs_rev.metadata_id is not None:
# Users have reported problems with repositories in which the
# deltatext block for revision 1.1 appears twice. It is not
# known whether this results from a CVS/RCS bug, or from botched
# hand-editing of the repository. In any case, empirically, cvs
# and rcs both use the first version when checking out data, so
# that's what we will do. (For the record: "cvs log" fails on
# such a file; "rlog" prints the log message from the first
# block and ignores the second one.)
logger.warn(
"%s: in '%s':\n"
" Deltatext block for revision %s appeared twice;\n"
" ignoring the second occurrence.\n"
% (warning_prefix, self.cvs_file.rcs_path, revision,)
)
return
if is_trunk_revision(revision):
branch_name = None
else:
branch_name = self.sdc.rev_to_branch_data(revision).symbol.name
cvs_rev.metadata_id = self.collect_data.metadata_logger.store(
self.project, branch_name, rev_data.author, log
)
cvs_rev.deltatext_exists = bool(text)
# If this is revision 1.1, determine whether the file appears to
# have been created via 'cvs add' instead of 'cvs import'. The
# test is that the log message CVS uses for 1.1 in imports is
# "Initial revision\n" with no period. (This fact helps determine
# whether this file might have had a default branch in the past.)
if revision == '1.1':
self._file_imported = (log == 'Initial revision\n')
def set_revision_info(self, revision, log, text):
"""This is a callback method declared in Sink."""
rev_data = self._rev_data[revision]
cvs_rev = self._cvs_file_items[rev_data.cvs_rev_id]
if cvs_rev.metadata_id is not None:
# Users have reported problems with repositories in which the
# deltatext block for revision 1.1 appears twice. It is not
# known whether this results from a CVS/RCS bug, or from botched
# hand-editing of the repository. In any case, empirically, cvs
# and rcs both use the first version when checking out data, so
# that's what we will do. (For the record: "cvs log" fails on
# such a file; "rlog" prints the log message from the first
# block and ignores the second one.)
logger.warn("%s: in '%s':\n"
" Deltatext block for revision %s appeared twice;\n"
" ignoring the second occurrence.\n" % (
warning_prefix,
self.cvs_file.rcs_path,
revision,
))
return
if is_trunk_revision(revision):
branch_name = None
else:
branch_name = self.sdc.rev_to_branch_data(revision).symbol.name
cvs_rev.metadata_id = self.collect_data.metadata_logger.store(
self.project, branch_name, rev_data.author, log)
cvs_rev.deltatext_exists = bool(text)
# If this is revision 1.1, determine whether the file appears to
# have been created via 'cvs add' instead of 'cvs import'. The
# test is that the log message CVS uses for 1.1 in imports is
# "Initial revision\n" with no period. (This fact helps determine
# whether this file might have had a default branch in the past.)
if revision == '1.1':
self._file_imported = (log == 'Initial revision\n')
def set_revision_info(self, revision, log, text):
if revision in self.revisions_seen:
# One common form of CVS repository corruption is that the
# Deltatext block for revision 1.1 appears twice. CollectData
# has already warned about this problem; here we can just ignore
# it.
return
else:
self.revisions_seen.add(revision)
cvs_rev_id = self.cvs_file_items.original_ids[revision]
if is_trunk_revision(revision):
# On trunk, revisions are encountered in reverse order (1.<N>
# ... 1.1) and deltas are inverted. The first text that we see
# is the fulltext for the HEAD revision. After that, the text
# corresponding to revision 1.N is the delta (1.<N+1> ->
# 1.<N>)). We have to invert the deltas here so that we can
# read the revisions out in dependency order; that is, for
# revision 1.1 we want the fulltext, and for revision 1.<N> we
# want the delta (1.<N-1> -> 1.<N>). This means that we can't
# compute the delta for a revision until we see its logical
# parent. When we finally see revision 1.1 (which is recognized
# because it doesn't have a parent), we can record the diff (1.1
# -> 1.2) for revision 1.2, and also the fulltext for 1.1.
if revision == self.head_revision:
# This is HEAD, as fulltext. Initialize the RCSStream so
# that we can compute deltas backwards in time.
self._rcs_stream = RCSStream(text)
self._rcs_stream_revision = revision
else:
# Any other trunk revision is a backward delta. Apply the
# delta to the RCSStream to mutate it to the contents of this
# revision, and also to get the reverse delta, which we store
# as the forward delta of our child revision.
try:
text = self._rcs_stream.invert_diff(text)
except MalformedDeltaException, e:
logger.error(
'Malformed RCS delta in %s, revision %s: %s'
% (self.cvs_file_items.cvs_file.rcs_path, revision, e)
)
raise RuntimeError()
text_record = DeltaTextRecord(
self.cvs_file_items.original_ids[self._rcs_stream_revision],
cvs_rev_id
)
self.revision_collector._writeout(text_record, text)
self._rcs_stream_revision = revision
if revision == self.revision_1_1:
# This is revision 1.1. Write its fulltext:
text_record = FullTextRecord(cvs_rev_id)
self.revision_collector._writeout(
text_record, self._rcs_stream.get_text()
)
# There will be no more trunk revisions delivered, so free the
# RCSStream.
del self._rcs_stream
del self._rcs_stream_revision
