Changeset - 000aa9cb25f1
[Not reviewed]
default
0 2 0
Mads Kiilerich - 6 years ago 2020-01-02 02:04:57
mads@kiilerich.com
Grafted from: 3e5ab2718c2e
vcs: refactor and simplify _get_revision

This will be more thorough checking that revisions actually exist, and will in
some cases give better error messages.

But mainly, prepare for py3 where this was easier to make work correctly.
2 files changed with 24 insertions and 28 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/vcs/backends/git/repository.py
Show inline comments
 
@@ -164,238 +164,236 @@ class GitRepository(BaseRepository):
 
        if url.startswith('git://'):
 
            return True
 

	
 
        if '+' in url[:url.find('://')]:
 
            url = url[url.find('+') + 1:]
 

	
 
        handlers = []
 
        url_obj = hg_url(url)
 
        test_uri, authinfo = url_obj.authinfo()
 
        url_obj.passwd = '*****'
 
        cleaned_uri = str(url_obj)
 

	
 
        if not test_uri.endswith('info/refs'):
 
            test_uri = test_uri.rstrip('/') + '/info/refs'
 

	
 
        if authinfo:
 
            # create a password manager
 
            passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
 
            passmgr.add_password(*authinfo)
 

	
 
            handlers.extend((httpbasicauthhandler(passmgr),
 
                             httpdigestauthhandler(passmgr)))
 

	
 
        o = urllib2.build_opener(*handlers)
 
        o.addheaders = [('User-Agent', 'git/1.7.8.0')]  # fake some git
 

	
 
        q = {"service": 'git-upload-pack'}
 
        qs = '?%s' % urllib.urlencode(q)
 
        cu = "%s%s" % (test_uri, qs)
 
        req = urllib2.Request(cu, None, {})
 

	
 
        try:
 
            resp = o.open(req)
 
            if resp.code != 200:
 
                raise Exception('Return Code is not 200')
 
        except Exception as e:
 
            # means it cannot be cloned
 
            raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))
 

	
 
        # now detect if it's proper git repo
 
        gitdata = resp.read()
 
        if 'service=git-upload-pack' not in gitdata:
 
            raise urllib2.URLError(
 
                "url [%s] does not look like an git" % cleaned_uri)
 

	
 
        return True
 

	
 
    def _get_repo(self, create, src_url=None, update_after_clone=False,
 
                  bare=False):
 
        if create and os.path.exists(self.path):
 
            raise RepositoryError("Location already exist")
 
        if src_url and not create:
 
            raise RepositoryError("Create should be set to True if src_url is "
 
                                  "given (clone operation creates repository)")
 
        try:
 
            if create and src_url:
 
                GitRepository._check_url(src_url)
 
                self.clone(src_url, update_after_clone, bare)
 
                return Repo(self.path)
 
            elif create:
 
                os.makedirs(self.path)
 
                if bare:
 
                    return Repo.init_bare(self.path)
 
                else:
 
                    return Repo.init(self.path)
 
            else:
 
                return Repo(self.path)
 
        except (NotGitRepository, OSError) as err:
 
            raise RepositoryError(err)
 

	
 
    def _get_all_revisions(self):
 
        # we must check if this repo is not empty, since later command
 
        # fails if it is. And it's cheaper to ask than throw the subprocess
 
        # errors
 
        try:
 
            self._repo.head()
 
        except KeyError:
 
            return []
 

	
 
        rev_filter = settings.GIT_REV_FILTER
 
        cmd = ['rev-list', rev_filter, '--reverse', '--date-order']
 
        try:
 
            so = self.run_git_command(cmd)
 
        except RepositoryError:
 
            # Can be raised for empty repositories
 
            return []
 
        return so.splitlines()
 

	
 
    def _get_all_revisions2(self):
 
        # alternate implementation using dulwich
 
        includes = [x[1][0] for x in self._parsed_refs.iteritems()
 
                    if x[1][1] != 'T']
 
        return [c.commit.id for c in self._repo.get_walker(include=includes)]
 

	
 
    def _get_revision(self, revision):
 
        """
 
        For git backend we always return integer here. This way we ensure
 
        that changeset's revision attribute would become integer.
 
        Given any revision identifier, returns a 40 char string with revision hash.
 
        """
 

	
 
        is_null = lambda o: len(o) == revision.count('0')
 

	
 
        if self._empty:
 
            raise EmptyRepositoryError("There are no changesets yet")
 

	
 
        if revision in (None, '', 'tip', 'HEAD', 'head', -1):
 
            return self.revisions[-1]
 
            revision = -1
 

	
 
        is_bstr = isinstance(revision, (str, unicode))
 
        if ((is_bstr and revision.isdigit() and len(revision) < 12)
 
            or isinstance(revision, int) or is_null(revision)
 
        ):
 
        if isinstance(revision, int):
 
            try:
 
                revision = self.revisions[int(revision)]
 
                return self.revisions[revision]
 
            except IndexError:
 
                msg = ("Revision %s does not exist for %s" % (revision, self))
 
                raise ChangesetDoesNotExistError(msg)
 

	
 
        elif is_bstr:
 
        if isinstance(revision, (str, unicode)):
 
            if revision.isdigit() and (len(revision) < 12 or len(revision) == revision.count('0')):
 
                try:
 
                    return self.revisions[int(revision)]
 
                except IndexError:
 
                    msg = "Revision %r does not exist for %s" % (revision, self)
 
                    raise ChangesetDoesNotExistError(msg)
 

	
 
            # get by branch/tag name
 
            _ref_revision = self._parsed_refs.get(revision)
 
            if _ref_revision:  # and _ref_revision[1] in ['H', 'RH', 'T']:
 
                return _ref_revision[0]
 

	
 
            _tags_shas = self.tags.values()
 
            if revision in self.revisions:
 
                return revision
 

	
 
            # maybe it's a tag ? we don't have them in self.revisions
 
            if revision in _tags_shas:
 
                return _tags_shas[_tags_shas.index(revision)]
 
            if revision in self.tags.values():
 
                return revision
 

	
 
            elif not SHA_PATTERN.match(revision) or revision not in self.revisions:
 
            if SHA_PATTERN.match(revision):
 
                msg = ("Revision %s does not exist for %s" % (revision, self))
 
                raise ChangesetDoesNotExistError(msg)
 

	
 
        # Ensure we return full id
 
        if not SHA_PATTERN.match(str(revision)):
 
            raise ChangesetDoesNotExistError("Given revision %s not recognized"
 
                % revision)
 
        return revision
 
        raise ChangesetDoesNotExistError("Given revision %r not recognized" % revision)
 

	
 
    def get_ref_revision(self, ref_type, ref_name):
 
        """
 
        Returns ``MercurialChangeset`` object representing repository's
 
        Returns ``GitChangeset`` object representing repository's
 
        changeset at the given ``revision``.
 
        """
 
        return self._get_revision(ref_name)
 

	
 
    def _get_archives(self, archive_name='tip'):
 

	
 
        for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:
 
            yield {"type": i[0], "extension": i[1], "node": archive_name}
 

	
 
    def _get_url(self, url):
 
        """
 
        Returns normalized url. If schema is not given, would fall to
 
        filesystem (``file:///``) schema.
 
        """
 
        url = safe_str(url)
 
        if url != 'default' and '://' not in url:
 
            url = ':///'.join(('file', url))
 
        return url
 

	
 
    def get_hook_location(self):
 
        """
 
        returns absolute path to location where hooks are stored
 
        """
 
        loc = os.path.join(self.path, 'hooks')
 
        if not self.bare:
 
            loc = os.path.join(self.path, '.git', 'hooks')
 
        return loc
 

	
 
    @LazyProperty
 
    def name(self):
 
        return os.path.basename(self.path)
 

	
 
    @LazyProperty
 
    def last_change(self):
 
        """
 
        Returns last change made on this repository as datetime object
 
        """
 
        return date_fromtimestamp(self._get_mtime(), makedate()[1])
 

	
 
    def _get_mtime(self):
 
        try:
 
            return time.mktime(self.get_changeset().date.timetuple())
 
        except RepositoryError:
 
            idx_loc = '' if self.bare else '.git'
 
            # fallback to filesystem
 
            in_path = os.path.join(self.path, idx_loc, "index")
 
            he_path = os.path.join(self.path, idx_loc, "HEAD")
 
            if os.path.exists(in_path):
 
                return os.stat(in_path).st_mtime
 
            else:
 
                return os.stat(he_path).st_mtime
 

	
 
    @LazyProperty
 
    def description(self):
 
        undefined_description = u'unknown'
 
        _desc = self._repo.get_description()
 
        return safe_unicode(_desc or undefined_description)
 

	
 
    @LazyProperty
 
    def contact(self):
 
        undefined_contact = u'Unknown'
 
        return undefined_contact
 

	
 
    @property
 
    def branches(self):
 
        if not self.revisions:
 
            return {}
 
        sortkey = lambda ctx: ctx[0]
 
        _branches = [(x[0], x[1][0])
 
                     for x in self._parsed_refs.iteritems() if x[1][1] == 'H']
 
        return OrderedDict(sorted(_branches, key=sortkey, reverse=False))
 

	
 
    @LazyProperty
 
    def closed_branches(self):
 
        return {}
 

	
 
    @LazyProperty
 
    def tags(self):
 
        return self._get_tags()
 

	
 
    def _get_tags(self):
 
        if not self.revisions:
 
            return {}
 

	
 
        sortkey = lambda ctx: ctx[0]
 
        _tags = [(x[0], x[1][0])
 
                 for x in self._parsed_refs.iteritems() if x[1][1] == 'T']
 
        return OrderedDict(sorted(_tags, key=sortkey, reverse=True))
 

	
 
    def tag(self, name, user, revision=None, message=None, date=None,
 
            **kwargs):
 
        """
 
        Creates and returns a tag for the given ``revision``.
 

	
 
        :param name: name for new tag
 
        :param user: full username, i.e.: "Joe Doe <joe.doe@example.com>"
kallithea/lib/vcs/backends/hg/repository.py
Show inline comments
 
# -*- coding: utf-8 -*-
 
"""
 
    vcs.backends.hg.repository
 
    ~~~~~~~~~~~~~~~~~~~~~~~~~~
 

	
 
    Mercurial repository implementation.
 

	
 
    :created_on: Apr 8, 2010
 
    :copyright: (c) 2010-2011 by Marcin Kuzminski, Lukasz Balcerzak.
 
"""
 

	
 
import datetime
 
import logging
 
import os
 
import time
 
import urllib
 
import urllib2
 
from collections import OrderedDict
 

	
 
from kallithea.lib.vcs.backends.base import BaseRepository, CollectionGenerator
 
from kallithea.lib.vcs.exceptions import (
 
    BranchDoesNotExistError, ChangesetDoesNotExistError, EmptyRepositoryError, RepositoryError, TagAlreadyExistError, TagDoesNotExistError, VCSError)
 
from kallithea.lib.vcs.utils import author_email, author_name, date_fromtimestamp, makedate, safe_str, safe_unicode
 
from kallithea.lib.vcs.utils import author_email, author_name, date_fromtimestamp, makedate, safe_bytes, safe_str, safe_unicode
 
from kallithea.lib.vcs.utils.hgcompat import (
 
    Abort, RepoError, RepoLookupError, clone, diffopts, get_contact, hex, hg_url, httpbasicauthhandler, httpdigestauthhandler, httppeer, localrepo, match_exact, nullid, patch, peer, scmutil, sshpeer, tag, ui)
 
from kallithea.lib.vcs.utils.lazy import LazyProperty
 
from kallithea.lib.vcs.utils.paths import abspath
 

	
 
from .changeset import MercurialChangeset
 
from .inmemory import MercurialInMemoryChangeset
 
from .workdir import MercurialWorkdir
 

	
 

	
 
log = logging.getLogger(__name__)
 

	
 

	
 
class MercurialRepository(BaseRepository):
 
    """
 
    Mercurial repository backend
 
    """
 
    DEFAULT_BRANCH_NAME = 'default'
 
    scm = 'hg'
 

	
 
    def __init__(self, repo_path, create=False, baseui=None, src_url=None,
 
                 update_after_clone=False):
 
        """
 
        Raises RepositoryError if repository could not be find at the given
 
        ``repo_path``.
 

	
 
        :param repo_path: local path of the repository
 
        :param create=False: if set to True, would try to create repository if
 
           it does not exist rather than raising exception
 
        :param baseui=None: user data
 
        :param src_url=None: would try to clone repository from given location
 
        :param update_after_clone=False: sets update of working copy after
 
          making a clone
 
        """
 

	
 
        if not isinstance(repo_path, str):
 
            raise VCSError('Mercurial backend requires repository path to '
 
                           'be instance of <str> got %s instead' %
 
                           type(repo_path))
 

	
 
        self.path = abspath(repo_path)
 
        self.baseui = baseui or ui.ui()
 
        # We've set path and ui, now we can set _repo itself
 
        self._repo = self._get_repo(create, src_url, update_after_clone)
 

	
 
    @property
 
    def _empty(self):
 
        """
 
        Checks if repository is empty ie. without any changesets
 
        """
 
        # TODO: Following raises errors when using InMemoryChangeset...
 
        # return len(self._repo.changelog) == 0
 
        return len(self.revisions) == 0
 

	
 
    @LazyProperty
 
    def revisions(self):
 
        """
 
        Returns list of revisions' ids, in ascending order.  Being lazy
 
        attribute allows external tools to inject shas from cache.
 
        """
 
        return self._get_all_revisions()
 

	
 
    @LazyProperty
 
    def name(self):
 
        return os.path.basename(self.path)
 

	
 
    @LazyProperty
 
    def branches(self):
 
        return self._get_branches()
 

	
 
    @LazyProperty
 
    def closed_branches(self):
 
        return self._get_branches(normal=False, closed=True)
 

	
 
    @LazyProperty
 
    def allbranches(self):
 
        """
 
        List all branches, including closed branches.
 
        """
 
        return self._get_branches(closed=True)
 

	
 
    def _get_branches(self, normal=True, closed=False):
 
        """
 
        Gets branches for this repository
 
        Returns only not closed branches by default
 

	
 
        :param closed: return also closed branches for mercurial
 
        :param normal: return also normal branches
 
        """
 

	
 
        if self._empty:
 
            return {}
 

	
 
        bt = OrderedDict()
 
        for bn, _heads, tip, isclosed in sorted(self._repo.branchmap().iterbranches()):
 
            if isclosed:
 
@@ -312,205 +312,203 @@ class MercurialRepository(BaseRepository
 
                        ('Accept', 'application/mercurial-0.1')]
 

	
 
        q = {"cmd": 'between'}
 
        q.update({'pairs': "%s-%s" % ('0' * 40, '0' * 40)})
 
        qs = '?%s' % urllib.urlencode(q)
 
        cu = "%s%s" % (test_uri, qs)
 
        req = urllib2.Request(cu, None, {})
 

	
 
        try:
 
            resp = o.open(req)
 
            if resp.code != 200:
 
                raise Exception('Return Code is not 200')
 
        except Exception as e:
 
            # means it cannot be cloned
 
            raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))
 

	
 
        if not url_prefix: # skip svn+http://... (and git+... too)
 
            # now check if it's a proper hg repo
 
            try:
 
                httppeer.instance(repoui or ui.ui(), url, False).lookup('tip')
 
            except Exception as e:
 
                raise urllib2.URLError(
 
                    "url [%s] does not look like an hg repo org_exc: %s"
 
                    % (cleaned_uri, e))
 

	
 
        return True
 

	
 
    def _get_repo(self, create, src_url=None, update_after_clone=False):
 
        """
 
        Function will check for mercurial repository in given path and return
 
        a localrepo object. If there is no repository in that path it will
 
        raise an exception unless ``create`` parameter is set to True - in
 
        that case repository would be created and returned.
 
        If ``src_url`` is given, would try to clone repository from the
 
        location at given clone_point. Additionally it'll make update to
 
        working copy accordingly to ``update_after_clone`` flag
 
        """
 

	
 
        try:
 
            if src_url:
 
                url = safe_str(self._get_url(src_url))
 
                opts = {}
 
                if not update_after_clone:
 
                    opts.update({'noupdate': True})
 
                MercurialRepository._check_url(url, self.baseui)
 
                clone(self.baseui, url, self.path, **opts)
 

	
 
                # Don't try to create if we've already cloned repo
 
                create = False
 
            return localrepo.instance(self.baseui, self.path, create=create)
 
        except (Abort, RepoError) as err:
 
            if create:
 
                msg = "Cannot create repository at %s. Original error was %s" \
 
                    % (self.path, err)
 
            else:
 
                msg = "Not valid repository at %s. Original error was %s" \
 
                    % (self.path, err)
 
            raise RepositoryError(msg)
 

	
 
    @LazyProperty
 
    def in_memory_changeset(self):
 
        return MercurialInMemoryChangeset(self)
 

	
 
    @LazyProperty
 
    def description(self):
 
        undefined_description = u'unknown'
 
        _desc = self._repo.ui.config('web', 'description', None, untrusted=True)
 
        return safe_unicode(_desc or undefined_description)
 

	
 
    @LazyProperty
 
    def contact(self):
 
        undefined_contact = u'Unknown'
 
        return safe_unicode(get_contact(self._repo.ui.config)
 
                            or undefined_contact)
 

	
 
    @LazyProperty
 
    def last_change(self):
 
        """
 
        Returns last change made on this repository as datetime object
 
        """
 
        return date_fromtimestamp(self._get_mtime(), makedate()[1])
 

	
 
    def _get_mtime(self):
 
        try:
 
            return time.mktime(self.get_changeset().date.timetuple())
 
        except RepositoryError:
 
            # fallback to filesystem
 
            cl_path = os.path.join(self.path, '.hg', "00changelog.i")
 
            st_path = os.path.join(self.path, '.hg', "store")
 
            if os.path.exists(cl_path):
 
                return os.stat(cl_path).st_mtime
 
            else:
 
                return os.stat(st_path).st_mtime
 

	
 
    def _get_revision(self, revision):
 
        """
 
        Gets an ID revision given as str. This will always return a full
 
        40 char revision number
 
        Given any revision identifier, returns a 40 char string with revision hash.
 

	
 
        :param revision: str or int or None
 
        """
 
        if isinstance(revision, unicode):
 
            revision = safe_str(revision)
 

	
 
        if self._empty:
 
            raise EmptyRepositoryError("There are no changesets yet")
 

	
 
        if revision in [-1, None]:
 
            revision = 'tip'
 
        elif isinstance(revision, unicode):
 
            revision = safe_bytes(revision)
 

	
 
        try:
 
            if isinstance(revision, int):
 
                return self._repo[revision].hex()
 
            return scmutil.revsymbol(self._repo, revision).hex()
 
        except (IndexError, ValueError, RepoLookupError, TypeError):
 
            msg = ("Revision %s does not exist for %s" % (revision, self))
 
            raise ChangesetDoesNotExistError(msg)
 
        except (LookupError, ):
 
            msg = ("Ambiguous identifier `%s` for %s" % (revision, self))
 
            raise ChangesetDoesNotExistError(msg)
 

	
 
    def get_ref_revision(self, ref_type, ref_name):
 
        """
 
        Returns revision number for the given reference.
 
        """
 
        ref_name = safe_str(ref_name)
 
        if ref_type == 'rev' and not ref_name.strip('0'):
 
            return self.EMPTY_CHANGESET
 
        # lookup up the exact node id
 
        _revset_predicates = {
 
                'branch': 'branch',
 
                'book': 'bookmark',
 
                'tag': 'tag',
 
                'rev': 'id',
 
            }
 
        # avoid expensive branch(x) iteration over whole repo
 
        rev_spec = "%%s & %s(%%s)" % _revset_predicates[ref_type]
 
        try:
 
            revs = self._repo.revs(rev_spec, ref_name, ref_name)
 
        except LookupError:
 
            msg = ("Ambiguous identifier %s:%s for %s" % (ref_type, ref_name, self.name))
 
            raise ChangesetDoesNotExistError(msg)
 
        except RepoLookupError:
 
            msg = ("Revision %s:%s does not exist for %s" % (ref_type, ref_name, self.name))
 
            raise ChangesetDoesNotExistError(msg)
 
        if revs:
 
            revision = revs.last()
 
        else:
 
            # TODO: just report 'not found'?
 
            revision = ref_name
 

	
 
        return self._get_revision(revision)
 

	
 
    def _get_archives(self, archive_name='tip'):
 
        allowed = self.baseui.configlist("web", "allow_archive",
 
                                         untrusted=True)
 
        for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:
 
            if i[0] in allowed or self._repo.ui.configbool("web",
 
                                                           "allow" + i[0],
 
                                                           untrusted=True):
 
                yield {"type": i[0], "extension": i[1], "node": archive_name}
 

	
 
    def _get_url(self, url):
 
        """
 
        Returns normalized url. If schema is not given, would fall
 
        to filesystem
 
        (``file:///``) schema.
 
        """
 
        url = safe_str(url)
 
        if url != 'default' and '://' not in url:
 
            url = "file:" + urllib.pathname2url(url)
 
        return url
 

	
 
    def get_hook_location(self):
 
        """
 
        returns absolute path to location where hooks are stored
 
        """
 
        return os.path.join(self.path, '.hg', '.hgrc')
 

	
 
    def get_changeset(self, revision=None):
 
        """
 
        Returns ``MercurialChangeset`` object representing repository's
 
        changeset at the given ``revision``.
 
        """
 
        revision = self._get_revision(revision)
 
        changeset = MercurialChangeset(repository=self, revision=revision)
 
        return changeset
 

	
 
    def get_changesets(self, start=None, end=None, start_date=None,
 
                       end_date=None, branch_name=None, reverse=False, max_revisions=None):
 
        """
 
        Returns iterator of ``MercurialChangeset`` objects from start to end
 
        (both are inclusive)
 

	
 
        :param start: None, str, int or mercurial lookup format
 
        :param end:  None, str, int or mercurial lookup format
 
        :param start_date:
 
        :param end_date:
 
        :param branch_name:
 
        :param reversed: return changesets in reversed order
 
        """
 
        start_raw_id = self._get_revision(start)
 
        start_pos = self.revisions.index(start_raw_id) if start else None
 
        end_raw_id = self._get_revision(end)
 
        end_pos = self.revisions.index(end_raw_id) if end else None
0 comments (0 inline, 0 general)