Changeset - 63adb3d5233a
[Not reviewed]
beta
0 2 0
Marcin Kuzminski - 13 years ago 2013-04-26 14:14:46
marcin@python-works.com
fixed some unicode problems with git file path
2 files changed with 8 insertions and 5 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/vcs/backends/git/changeset.py
Show inline comments
 
import re
 
from itertools import chain
 
from dulwich import objects
 
from subprocess import Popen, PIPE
 
import rhodecode
 
from rhodecode.lib.vcs.conf import settings
 
from rhodecode.lib.vcs.exceptions import RepositoryError
 
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
 
from rhodecode.lib.utils2 import safe_int, safe_str
 

	
 

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

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

	
 
        try:
 
            commit = self.repository._repo.get_object(revision)
 
            if isinstance(commit, objects.Tag):
 
                revision = commit.object[1]
 
                commit = self.repository._repo.get_object(commit.object[1])
 
        except KeyError:
 
            raise RepositoryError("Cannot get object with id %s" % revision)
 
        self.raw_id = revision
 
        self.id = self.raw_id
 
        self.short_id = self.raw_id[:12]
 
        self._commit = commit
 

	
 
        self._tree_id = commit.tree
 
        self._committer_property = 'committer'
 
        self._author_property = 'author'
 
        self._date_property = 'commit_time'
 
        self._date_tz_property = 'commit_timezone'
 
        self.revision = repository.revisions.index(revision)
 

	
 
        self.message = safe_unicode(commit.message)
 

	
 
        self.nodes = {}
 
        self._paths = {}
 

	
 
    @LazyProperty
 
    def committer(self):
 
        return safe_unicode(getattr(self._commit, self._committer_property))
 

	
 
    @LazyProperty
 
    def author(self):
 
        return safe_unicode(getattr(self._commit, self._author_property))
 

	
 
    @LazyProperty
 
    def date(self):
 
        return date_fromtimestamp(getattr(self._commit, self._date_property),
 
                                  getattr(self._commit, self._date_tz_property))
 

	
 
    @LazyProperty
 
    def _timestamp(self):
 
        return getattr(self._commit, self._date_property)
 

	
 
    @LazyProperty
 
    def status(self):
 
        """
 
        Returns modified, added, removed, deleted files for current changeset
 
        """
 
        return self.changed, self.added, self.removed
 

	
 
    @LazyProperty
 
    def tags(self):
 
        _tags = []
 
        for tname, tsha in self.repository.tags.iteritems():
 
            if tsha == self.raw_id:
 
                _tags.append(tname)
 
        return _tags
 

	
 
    @LazyProperty
 
    def branch(self):
 

	
 
        heads = self.repository._heads(reverse=False)
 

	
 
        ref = heads.get(self.raw_id)
 
        if ref:
 
            return safe_unicode(ref)
 

	
 
    def _fix_path(self, path):
 
        """
 
        Paths are stored without trailing slash so we need to get rid off it if
 
        needed.
 
        """
 
        if path.endswith('/'):
 
            path = path.rstrip('/')
 
        return path
 

	
 
    def _get_id_for_path(self, path):
 

	
 
        # FIXME: Please, spare a couple of minutes and make those codes cleaner;
 
        if not path in self._paths:
 
            path = path.strip('/')
 
            # set root tree
 
            tree = self.repository._repo[self._tree_id]
 
            if path == '':
 
                self._paths[''] = tree.id
 
                return tree.id
 
            splitted = path.split('/')
 
@@ -196,201 +196,204 @@ class GitChangeset(BaseChangeset):
 
        children = []
 
        for l in so.splitlines():
 
            childs = l.split(' ')[1:]
 
            children.extend(childs)
 
        return [self.repository.get_changeset(cs) for cs in children]
 

	
 
    def next(self, branch=None):
 

	
 
        if branch and self.branch != branch:
 
            raise VCSError('Branch option used on changeset not belonging '
 
                           'to that branch')
 

	
 
        def _next(changeset, branch):
 
            try:
 
                next_ = changeset.revision + 1
 
                next_rev = changeset.repository.revisions[next_]
 
            except IndexError:
 
                raise ChangesetDoesNotExistError
 
            cs = changeset.repository.get_changeset(next_rev)
 

	
 
            if branch and branch != cs.branch:
 
                return _next(cs, branch)
 

	
 
            return cs
 

	
 
        return _next(self, branch)
 

	
 
    def prev(self, branch=None):
 
        if branch and self.branch != branch:
 
            raise VCSError('Branch option used on changeset not belonging '
 
                           'to that branch')
 

	
 
        def _prev(changeset, branch):
 
            try:
 
                prev_ = changeset.revision - 1
 
                if prev_ < 0:
 
                    raise IndexError
 
                prev_rev = changeset.repository.revisions[prev_]
 
            except IndexError:
 
                raise ChangesetDoesNotExistError
 

	
 
            cs = changeset.repository.get_changeset(prev_rev)
 

	
 
            if branch and branch != cs.branch:
 
                return _prev(cs, branch)
 

	
 
            return cs
 

	
 
        return _prev(self, branch)
 

	
 
    def diff(self, ignore_whitespace=True, context=3):
 
        rev1 = self.parents[0] if self.parents else self.repository.EMPTY_CHANGESET
 
        rev2 = self
 
        return ''.join(self.repository.get_diff(rev1, rev2,
 
                                    ignore_whitespace=ignore_whitespace,
 
                                    context=context))
 

	
 
    def get_file_mode(self, path):
 
        """
 
        Returns stat mode of the file at the given ``path``.
 
        """
 
        # ensure path is traversed
 
        self._get_id_for_path(path)
 
        return self._stat_modes[path]
 

	
 
    def get_file_content(self, path):
 
        """
 
        Returns content of the file at given ``path``.
 
        """
 
        id = self._get_id_for_path(path)
 
        blob = self.repository._repo[id]
 
        return blob.as_pretty_string()
 

	
 
    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``.
 
        """
 
        return self.get_file_history(path, limit=1)[0]
 

	
 
    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)
 
        cs_id = safe_str(self.id)
 
        f_path = safe_str(path)
 

	
 
        self._get_filectx(path)
 
        if limit:
 
            cmd = 'log -n %s --pretty="format: %%H" -s -p %s -- "%s"' % (
 
                      safe_int(limit, 0), self.id, path
 
                      safe_int(limit, 0), cs_id, f_path
 
                   )
 

	
 
        else:
 
            cmd = 'log --pretty="format: %%H" -s -p %s -- "%s"' % (
 
                      self.id, path
 
                      cs_id, f_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.
 

	
 
        """
 
        self._get_filectx(path)
 
        from dulwich.walk import Walker
 
        include = [self.id]
 
        walker = Walker(self.repository._repo.object_store, include,
 
                        paths=[path], max_entries=1)
 
        return [self.repository.get_changeset(sha)
 
                for sha in (x.commit.id for x in walker)]
 

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

	
 
        TODO: This function now uses os underlying 'git' command which is
 
        generally not good. Should be replaced with algorithm iterating
 
        commits.
 
        """
 
        cmd = 'blame -l --root -r %s -- "%s"' % (self.id, path)
 
        # -l     ==> outputs long shas (and we need all 40 characters)
 
        # --root ==> doesn't put '^' character for bounderies
 
        # -r sha ==> blames for the given revision
 
        so, se = self.repository.run_git_command(cmd)
 

	
 
        for i, blame_line in enumerate(so.split('\n')[:-1]):
 
            ln_no = i + 1
 
            sha, line = re.split(r' ', blame_line, 1)
 
            yield (ln_no, sha, lambda: self.repository.get_changeset(sha), line)
 

	
 
    def fill_archive(self, stream=None, kind='tgz', prefix=None,
 
                     subrepos=False):
 
        """
 
        Fills up given stream.
 

	
 
        :param stream: file like object.
 
        :param kind: one of following: ``zip``, ``tgz`` or ``tbz2``.
 
            Default: ``tgz``.
 
        :param prefix: name of root directory in archive.
 
            Default is repository name and changeset's raw_id joined with dash
 
            (``repo-tip.<KIND>``).
 
        :param subrepos: include subrepos in this archive.
 

	
 
        :raise ImproperArchiveTypeError: If given kind is wrong.
 
        :raise VcsError: If given stream is None
 

	
 
        """
 
        allowed_kinds = settings.ARCHIVE_SPECS.keys()
 
        if kind not in allowed_kinds:
 
            raise ImproperArchiveTypeError('Archive kind not supported use one'
 
                'of %s', allowed_kinds)
 

	
 
        if prefix is None:
 
            prefix = '%s-%s' % (self.repository.name, self.short_id)
 
        elif prefix.startswith('/'):
 
            raise VCSError("Prefix cannot start with leading slash")
 
        elif prefix.strip() == '':
 
            raise VCSError("Prefix cannot be empty")
 

	
 
        if kind == 'zip':
 
            frmt = 'zip'
 
        else:
 
            frmt = 'tar'
 
        _git_path = rhodecode.CONFIG.get('git_path', 'git')
 
        cmd = '%s archive --format=%s --prefix=%s/ %s' % (_git_path,
 
                                                frmt, prefix, self.raw_id)
 
        if kind == 'tgz':
 
            cmd += ' | gzip -9'
 
        elif kind == 'tbz2':
 
            cmd += ' | bzip2 -9'
 

	
 
        if stream is None:
 
            raise VCSError('You need to pass in a valid stream for filling'
 
                           ' with archival data')
 
        popen = Popen(cmd, stdout=PIPE, stderr=PIPE, shell=True,
 
            cwd=self.repository.path)
 

	
 
        buffer_size = 1024 * 8
 
        chunk = popen.stdout.read(buffer_size)
 
        while chunk:
 
            stream.write(chunk)
 
            chunk = popen.stdout.read(buffer_size)
 
        # Make sure all descriptors would be read
 
        popen.communicate()
 

	
 
    def get_nodes(self, path):
 
        if self._get_kind(path) != NodeKind.DIR:
rhodecode/templates/files/files.html
Show inline comments
 
<%inherit file="/base/base.html"/>
 

	
 
<%def name="title(*args)">
 
    ${_('%s Files') % c.repo_name}
 
    %if hasattr(c,'file'):
 
        &middot; ${c.file.path or '\\'}
 
        &middot; ${h.safe_unicode(c.file.path) or '\\'}
 
    %endif
 
    &middot; ${c.rhodecode_name}
 
</%def>
 

	
 
<%def name="breadcrumbs_links()">
 
    ${_('Files')}
 
    %if c.file:
 
        @ ${h.show_id(c.changeset)}
 
    %endif
 
</%def>
 

	
 
<%def name="page_nav()">
 
    ${self.menu('repositories')}
 
</%def>
 

	
 
<%def name="main()">
 
${self.context_bar('files')}
 
<div class="box">
 
    <!-- box / title -->
 
    <div class="title">
 
        ${self.breadcrumbs()}
 
        <ul class="links">
 
            <li>
 
              <span style="text-transform: uppercase;"><a href="#">${_('Branch')}: ${c.changeset.branch}</a></span>
 
            </li>
 
        </ul>
 
    </div>
 
    <div class="table">
 
        <div id="files_data">
 
            <%include file='files_ypjax.html'/>
 
        </div>
 
    </div>
 
</div>
 

	
 
<script type="text/javascript">
 
var CACHE = {};
 
var CACHE_EXPIRE = 5*60*1000; //cache for 5*60s
 
//used to construct links from the search list
 
var url_base = '${h.url("files_home",repo_name=c.repo_name,revision='__REV__',f_path='__FPATH__')}';
 
//send the nodelist request to this url
 
var node_list_url = '${h.url("files_nodelist_home",repo_name=c.repo_name,revision='__REV__',f_path='__FPATH__')}';
 
// send the node history requst to this url
 
var node_history_url = '${h.url("files_history_home",repo_name=c.repo_name,revision='__REV__',f_path='__FPATH__')}';
 

	
 
var ypjax_links = function(){
 
    YUE.on(YUQ('.ypjax-link'), 'click',function(e){
 

	
 
        //don't do ypjax on middle click
 
        if(e.which == 2 || !History.enabled){
 
            return true;
 
        }
 

	
 
        var el = e.currentTarget;
 
        var url = el.href;
 

	
 
        var _base_url = '${h.url("files_home",repo_name=c.repo_name,revision='',f_path='')}';
 
        _base_url = _base_url.replace('//','/')
 

	
 
        //extract rev and the f_path from url.
 
        parts = url.split(_base_url)
 
        if(parts.length != 2){
 
            return false;
 
        }
 

	
 
        var parts2 = parts[1].split('/');
 
        var rev = parts2.shift(); // pop the first element which is the revision
 
        var f_path = parts2.join('/');
 

	
 
        //page title make this consistent with title() mako function on top
 
        var title = "${_('%s Files') % c.repo_name}" + " &middot; " + (f_path || '\\') + " &middot; " + "${c.rhodecode_name}";
 

	
 
        var _node_list_url = node_list_url.replace('__REV__',rev).replace('__FPATH__', f_path);
 
        var _url_base = url_base.replace('__REV__',rev);
 

	
 
        // Change our States and save some data for handling events
 
        var data = {url:url,title:title, url_base:_url_base,
 
                    node_list_url:_node_list_url, rev:rev, f_path:f_path};
 
        History.pushState(data, title, url);
 

	
 
        //now we're sure that we can do ypjax things
 
        YUE.preventDefault(e);
 
        return false;
 
    });
 
}
 

	
 
var callbacks = function(State){
 
    ypjax_links();
 
    tooltip_activate();
 
    fileBrowserListeners(State.url, State.data.node_list_url, State.data.url_base);
 

	
 
    if(YUD.get('hlcode')){
 
        YUE.on('hlcode', 'mouseup', getSelectionLink);
 
    }
 
    //console.log(State);
 
    if(YUD.get('load_node_history')){
 
      //remove all listeners due to problems of history state
0 comments (0 inline, 0 general)