Changeset - 58905069da21
[Not reviewed]
beta
0 4 0
Marcin Kuzminski - 13 years ago 2013-03-07 13:46:24
marcin@python-works.com
Speed up of last_changeset extraction in VCS, in edge cases for git we can get 10x speed improvement by limiting the history extraction if we only need last changeset
3 files changed with 23 insertions and 13 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/vcs/backends/base.py
Show inline comments
 
@@ -11,24 +11,25 @@
 

	
 

	
 
from itertools import chain
 
from rhodecode.lib.vcs.utils import author_name, author_email
 
from rhodecode.lib.vcs.utils.lazy import LazyProperty
 
from rhodecode.lib.vcs.utils.helpers import get_dict_for_attrs
 
from rhodecode.lib.vcs.conf import settings
 

	
 
from rhodecode.lib.vcs.exceptions import ChangesetError, EmptyRepositoryError, \
 
    NodeAlreadyAddedError, NodeAlreadyChangedError, NodeAlreadyExistsError, \
 
    NodeAlreadyRemovedError, NodeDoesNotExistError, NodeNotChangedError, \
 
    RepositoryError
 
import datetime
 

	
 

	
 
class BaseRepository(object):
 
    """
 
    Base Repository for final backends
 

	
 
    **Attributes**
 

	
 
        ``DEFAULT_BRANCH_NAME``
 
            name of default branch (i.e. "trunk" for svn, "master" for git etc.
 

	
 
        ``scm``
 
@@ -971,30 +972,30 @@ class BaseInMemoryChangeset(object):
 
        :raises ``CommitError``: if any error occurs while committing
 
        """
 
        raise NotImplementedError
 

	
 

	
 
class EmptyChangeset(BaseChangeset):
 
    """
 
    An dummy empty changeset. It's possible to pass hash when creating
 
    an EmptyChangeset
 
    """
 

	
 
    def __init__(self, cs='0' * 40, repo=None, requested_revision=None,
 
                 alias=None, revision=-1, message='', author='', date=''):
 
                 alias=None, revision=-1, message='', author='', date=None):
 
        self._empty_cs = cs
 
        self.revision = revision
 
        self.message = message
 
        self.author = author
 
        self.date = date
 
        self.date = date or datetime.datetime.fromtimestamp(0)
 
        self.repository = repo
 
        self.requested_revision = requested_revision
 
        self.alias = alias
 

	
 
    @LazyProperty
 
    def raw_id(self):
 
        """
 
        Returns raw string identifying this changeset, useful for web
 
        representation.
 
        """
 

	
 
        return self._empty_cs
rhodecode/lib/vcs/backends/git/changeset.py
Show inline comments
 
@@ -8,24 +8,25 @@ from rhodecode.lib.vcs.exceptions import
 
from rhodecode.lib.vcs.exceptions import ChangesetError
 
from rhodecode.lib.vcs.exceptions import NodeDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import VCSError
 
from rhodecode.lib.vcs.exceptions import ChangesetDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import ImproperArchiveTypeError
 
from rhodecode.lib.vcs.backends.base import BaseChangeset, EmptyChangeset
 
from rhodecode.lib.vcs.nodes import FileNode, DirNode, NodeKind, RootNode, \
 
    RemovedFileNode, SubModuleNode, ChangedFileNodesGenerator,\
 
    AddedFileNodesGenerator, RemovedFileNodesGenerator
 
from rhodecode.lib.vcs.utils import safe_unicode
 
from rhodecode.lib.vcs.utils import date_fromtimestamp
 
from rhodecode.lib.vcs.utils.lazy import LazyProperty
 
from rhodecode.lib.utils2 import safe_int
 

	
 

	
 
class GitChangeset(BaseChangeset):
 
    """
 
    Represents state of the repository at single revision.
 
    """
 

	
 
    def __init__(self, repository, revision):
 
        self._stat_modes = {}
 
        self.repository = repository
 

	
 
        try:
 
@@ -266,38 +267,42 @@ class GitChangeset(BaseChangeset):
 
    def get_file_size(self, path):
 
        """
 
        Returns size of the file at given ``path``.
 
        """
 
        id = self._get_id_for_path(path)
 
        blob = self.repository._repo[id]
 
        return blob.raw_length()
 

	
 
    def get_file_changeset(self, path):
 
        """
 
        Returns last commit of the file at the given ``path``.
 
        """
 
        node = self.get_node(path)
 
        return node.history[0]
 
        return self.get_file_history(path, limit=1)[0]
 

	
 
    def get_file_history(self, path):
 
    def get_file_history(self, path, limit=None):
 
        """
 
        Returns history of file as reversed list of ``Changeset`` objects for
 
        which file at given ``path`` has been modified.
 

	
 
        TODO: This function now uses os underlying 'git' and 'grep' commands
 
        which is generally not good. Should be replaced with algorithm
 
        iterating commits.
 
        """
 

	
 
        self._get_filectx(path)
 

	
 
        if limit:
 
            cmd = 'log -n %s --pretty="format: %%H" -s -p %s -- "%s"' % (
 
                      safe_int(limit, 0), self.id, path
 
                   )
 
        else:
 
        cmd = 'log --pretty="format: %%H" -s -p %s -- "%s"' % (
 
                  self.id, path
 
               )
 
        so, se = self.repository.run_git_command(cmd)
 
        ids = re.findall(r'[0-9a-fA-F]{40}', so)
 
        return [self.repository.get_changeset(id) for id in ids]
 

	
 
    def get_file_history_2(self, path):
 
        """
 
        Returns history of file as reversed list of ``Changeset`` objects for
 
        which file at given ``path`` has been modified.
 

	
rhodecode/lib/vcs/backends/hg/changeset.py
Show inline comments
 
@@ -210,37 +210,41 @@ class MercurialChangeset(BaseChangeset):
 

	
 
    def get_file_size(self, path):
 
        """
 
        Returns size of the file at given ``path``.
 
        """
 
        fctx = self._get_filectx(path)
 
        return fctx.size()
 

	
 
    def get_file_changeset(self, path):
 
        """
 
        Returns last commit of the file at the given ``path``.
 
        """
 
        node = self.get_node(path)
 
        return node.history[0]
 
        return self.get_file_history(path, limit=1)[0]
 

	
 
    def get_file_history(self, path):
 
    def get_file_history(self, path, limit=None):
 
        """
 
        Returns history of file as reversed list of ``Changeset`` objects for
 
        which file at given ``path`` has been modified.
 
        """
 
        fctx = self._get_filectx(path)
 
        nodes = [fctx.filectx(x).node() for x in fctx.filelog()]
 
        changesets = [self.repository.get_changeset(hex(node))
 
            for node in reversed(nodes)]
 
        return changesets
 
        hist = []
 
        cnt = 0
 
        for cs in reversed([x for x in fctx.filelog()]):
 
            cnt += 1
 
            hist.append(hex(fctx.filectx(cs).node()))
 
            if limit and cnt == limit:
 
                break
 

	
 
        return [self.repository.get_changeset(node) for node in hist]
 

	
 
    def get_file_annotate(self, path):
 
        """
 
        Returns a generator of four element tuples with
 
            lineno, sha, changeset lazy loader and line
 
        """
 

	
 
        fctx = self._get_filectx(path)
 
        for i, annotate_data in enumerate(fctx.annotate()):
 
            ln_no = i + 1
 
            sha = hex(annotate_data[0].node())
 
            yield (ln_no, sha, lambda: self.repository.get_changeset(sha), annotate_data[1],)
0 comments (0 inline, 0 general)