X-Git-Url: https://git.mdrn.pl/redakcja.git/blobdiff_plain/c601bf6ac7304192cac242965217b7c089f37f8d..e3d6e94589c1356d3f8d1b7fc36d9f21d966e16a:/apps/dvcs/models.py?ds=sidebyside diff --git a/apps/dvcs/models.py b/apps/dvcs/models.py index 3991efc8..6ecb97ce 100644 --- a/apps/dvcs/models.py +++ b/apps/dvcs/models.py @@ -1,109 +1,72 @@ +# -*- coding: utf-8 -*- +# +# This file is part of MIL/PEER, licensed under GNU Affero GPLv3 or later. +# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information. +# +from __future__ import unicode_literals, print_function + from datetime import datetime -import os.path +import os +import re +from subprocess import PIPE, Popen +from tempfile import NamedTemporaryFile -from django.contrib.auth.models import User +from django.conf import settings from django.core.files.base import ContentFile -from django.core.files.storage import FileSystemStorage -from django.db import models, transaction -from django.db.models.base import ModelBase +from django.db import models +from django.utils.encoding import python_2_unicode_compatible from django.utils.translation import ugettext_lazy as _ -from mercurial import mdiff, simplemerge -from django.conf import settings -from dvcs.signals import post_commit, post_publishable +from dvcs.signals import post_commit, post_merge from dvcs.storage import GzipFileSystemStorage +# default repository path; make a setting for it +REPO_PATH = os.path.join(settings.MEDIA_ROOT, 'dvcs') +repo = GzipFileSystemStorage(location=REPO_PATH) -class Tag(models.Model): - """A tag (e.g. document stage) which can be applied to a Change.""" - name = models.CharField(_('name'), max_length=64) - slug = models.SlugField(_('slug'), unique=True, max_length=64, - null=True, blank=True) - ordering = models.IntegerField(_('ordering')) - - _object_cache = {} - - class Meta: - abstract = True - ordering = ['ordering'] - verbose_name = _("tag") - verbose_name_plural = _("tags") - - def __unicode__(self): - return self.name - - @classmethod - def get(cls, slug): - if slug in cls._object_cache: - return cls._object_cache[slug] - else: - obj = cls.objects.get(slug=slug) - cls._object_cache[slug] = obj - return obj - - @staticmethod - def listener_changed(sender, instance, **kwargs): - sender._object_cache = {} - def next(self): - """ - Returns the next tag - stage to work on. - Returns None for the last stage. - """ - try: - return type(self).objects.filter(ordering__gt=self.ordering)[0] - except IndexError: - return None - -models.signals.pre_save.connect(Tag.listener_changed, sender=Tag) +@python_2_unicode_compatible +class Revision(models.Model): + """ + A document revision. The "parent" + argument points to the version against which this change has been + recorded. Initial text will have a null parent. + Gzipped text of the document is stored in a file. + """ + author = models.ForeignKey(settings.AUTH_USER_MODEL, null=True, blank=True, verbose_name=_('author')) + author_name = models.CharField( + _('author name'), max_length=128, null=True, blank=True, help_text=_("Used if author is not set.")) + author_email = models.CharField( + _('author email'), max_length=128, null=True, blank=True, help_text=_("Used if author is not set.")) + # Any other author data? + # How do we identify an author? -def data_upload_to(instance, filename): - return "%d/%d" % (instance.tree.pk, instance.pk) + parent = models.ForeignKey( + 'self', null=True, blank=True, default=None, verbose_name=_('parent'), related_name="children") -class Change(models.Model): - """ - Single document change related to previous change. The "parent" - argument points to the version against which this change has been - recorded. Initial text will have a null parent. - - Data file contains a gzipped text of the document. - """ - author = models.ForeignKey(User, null=True, blank=True, verbose_name=_('author')) - author_name = models.CharField(_('author name'), max_length=128, - null=True, blank=True, - help_text=_("Used if author is not set.") - ) - author_email = models.CharField(_('author email'), max_length=128, - null=True, blank=True, - help_text=_("Used if author is not set.") - ) - revision = models.IntegerField(_('revision'), db_index=True) - - parent = models.ForeignKey('self', - null=True, blank=True, default=None, - verbose_name=_('parent'), - related_name="children") - - merge_parent = models.ForeignKey('self', - null=True, blank=True, default=None, - verbose_name=_('merge parent'), - related_name="merge_children") + merge_parent = models.ForeignKey( + 'self', null=True, blank=True, default=None, verbose_name=_('merge parent'), related_name="merge_children") description = models.TextField(_('description'), blank=True, default='') - created_at = models.DateTimeField(editable=False, db_index=True, - default=datetime.now) - publishable = models.BooleanField(_('publishable'), default=False) + created_at = models.DateTimeField(editable=False, db_index=True, default=datetime.now) class Meta: - abstract = True ordering = ('created_at',) - unique_together = ['tree', 'revision'] - verbose_name = _("change") - verbose_name_plural = _("changes") + verbose_name = _("revision") + verbose_name_plural = _("revisions") + + def __str__(self): + return "Id: %r, Parent %r, Data: %s" % (self.id, self.parent_id, self.get_text_path()) + + def get_text_path(self): + if self.pk: + return re.sub(r'([0-9a-f]{2})([^.])', r'\1/\2', '%x.gz' % self.pk) + else: + return None - def __unicode__(self): - return u"Id: %r, Tree %r, Parent %r, Data: %s" % (self.id, self.tree_id, self.parent_id, self.data) + def save_text(self, content): + return repo.save(self.get_text_path(), ContentFile(content.encode('utf-8'))) def author_str(self): if self.author: @@ -117,221 +80,182 @@ class Change(models.Model): self.author_email ) - - def save(self, *args, **kwargs): - """ - take the next available revision number if none yet - """ - if self.revision is None: - tree_rev = self.tree.revision() - if tree_rev is None: - self.revision = 1 - else: - self.revision = tree_rev + 1 - return super(Change, self).save(*args, **kwargs) + @classmethod + def create(cls, text, parent=None, merge_parent=None, author=None, author_name=None, author_email=None, + description=''): + + if text: + text = text.replace( + '', '').replace( + '
', '
') + + revision = cls.objects.create( + parent=parent, + merge_parent=merge_parent, + author=author, + author_name=author_name, + author_email=author_email, + description=description + ) + revision.save_text(text) + return revision def materialize(self): - f = self.data.storage.open(self.data) - text = f.read() + f = repo.open(self.get_text_path()) + text = f.read().decode('utf-8') f.close() - return unicode(text, 'utf-8') - - def merge_with(self, other, author=None, - author_name=None, author_email=None, - description=u"Automatic merge."): - """Performs an automatic merge after straying commits.""" - assert self.tree_id == other.tree_id # same tree - if other.parent_id == self.pk: - # immediate child - fast forward + if text: + text = text.replace( + '', '').replace( + '
', '
') + return text + + def is_descendant_of(self, other): + # Naive approach. + return ( + ( + self.parent is not None and ( + self.parent.pk == other.pk or + self.parent.is_descendant_of(other) + ) + ) or ( + self.merge_parent is not None and ( + self.merge_parent.pk == other.pk or + self.merge_parent.is_descendant_of(other) + ) + ) + ) + + def get_common_ancestor_with(self, other): + # VERY naive approach. + if self.pk == other.pk: + return self + if self.is_descendant_of(other): return other + if other.is_descendant_of(self): + return self - local = self.materialize().encode('utf-8') - base = other.parent.materialize().encode('utf-8') - remote = other.materialize().encode('utf-8') - - merge = simplemerge.Merge3Text(base, local, remote) - result = ''.join(merge.merge_lines()) - merge_node = self.children.create( - merge_parent=other, tree=self.tree, - author=author, - author_name=author_name, - author_email=author_email, - description=description) - merge_node.data.save('', ContentFile(result)) - return merge_node - - def revert(self, **kwargs): - """ commit this version of a doc as new head """ - self.tree.commit(text=self.materialize(), **kwargs) - - def set_publishable(self, publishable): - self.publishable = publishable - self.save() - post_publishable.send(sender=self, publishable=publishable) - - -def create_tag_model(model): - name = model.__name__ + 'Tag' - - class Meta(Tag.Meta): - app_label = model._meta.app_label - - attrs = { - '__module__': model.__module__, - 'Meta': Meta, - } - return type(name, (Tag,), attrs) - - -def create_change_model(model): - name = model.__name__ + 'Change' - repo = GzipFileSystemStorage(location=model.REPO_PATH) - - class Meta(Change.Meta): - app_label = model._meta.app_label - - attrs = { - '__module__': model.__module__, - 'tree': models.ForeignKey(model, related_name='change_set', verbose_name=_('document')), - 'tags': models.ManyToManyField(model.tag_model, verbose_name=_('tags'), related_name='change_set'), - 'data': models.FileField(_('data'), upload_to=data_upload_to, storage=repo), - 'Meta': Meta, - } - return type(name, (Change,), attrs) - - -class DocumentMeta(ModelBase): - "Metaclass for Document models." - def __new__(cls, name, bases, attrs): - - model = super(DocumentMeta, cls).__new__(cls, name, bases, attrs) - if not model._meta.abstract: - # create a real Tag object and `stage' fk - model.tag_model = create_tag_model(model) - models.ForeignKey(model.tag_model, verbose_name=_('stage'), - null=True, blank=True).contribute_to_class(model, 'stage') - - # create real Change model and `head' fk - model.change_model = create_change_model(model) - - models.ForeignKey(model.change_model, - null=True, blank=True, default=None, - verbose_name=_('head'), - help_text=_("This document's current head."), - editable=False).contribute_to_class(model, 'head') - - models.ForeignKey(User, null=True, blank=True, editable=False, - verbose_name=_('creator'), related_name="created_%s" % name.lower() - ).contribute_to_class(model, 'creator') - - return model - - -class Document(models.Model): - """File in repository. Subclass it to use version control in your app.""" - - __metaclass__ = DocumentMeta - - # default repository path - REPO_PATH = os.path.join(settings.MEDIA_ROOT, 'dvcs') - - user = models.ForeignKey(User, null=True, blank=True, - verbose_name=_('user'), help_text=_('Work assignment.')) - - class Meta: - abstract = True + if self.parent is not None: + parent_ca = self.parent.get_common_ancestor_with(other) + else: + parent_ca = None - def __unicode__(self): - return u"{0}, HEAD: {1}".format(self.id, self.head_id) + if self.merge_parent is not None: + merge_parent_ca = self.merge_parent.get_common_ancestor_with(other) + else: + return parent_ca + + if parent_ca is None or parent_ca.created_at < merge_parent_ca.created_at: + return merge_parent_ca + + return parent_ca + + def get_ancestors(self): + revs = set() + if self.parent is not None: + revs.add(self.parent) + revs.update(self.parent.get_ancestors()) + if self.merge_parent is not None: + revs.add(self.merge_parent) + revs.update(self.merge_parent.get_ancestors()) + return revs + + +@python_2_unicode_compatible +class Ref(models.Model): + """A reference pointing to a specific revision.""" + + revision = models.ForeignKey( + Revision, null=True, blank=True, default=None, verbose_name=_('revision'), + help_text=_("The document's revision."), editable=False) + + def __str__(self): + return "ref:{0}->rev:{1}".format(self.id, self.revision_id) + + def merge_text(self, base, local, remote): + """Override in subclass to have different kinds of merges.""" + files = [] + for f in local, base, remote: + temp = NamedTemporaryFile(delete=False) + temp.write(f) + temp.close() + files.append(temp.name) + p = Popen(['/usr/bin/diff3', '-mE', '-L', 'old', '-L', '', '-L', 'new'] + files, stdout=PIPE) + result, errs = p.communicate() + + for f in files: + os.unlink(f) + return result.decode('utf-8') + + def merge_with(self, revision, author=None, author_name=None, author_email=None, description="Automatic merge."): + """Merges a given revision into the ref.""" + if self.revision is None: + fast_forward = True + self.revision = revision + elif self.revision.pk == revision.pk or self.revision.is_descendant_of(revision): + # Already merged. + return + elif revision.is_descendant_of(self.revision): + # Fast forward. + fast_forward = True + self.revision = revision + else: + # Need to create a merge revision. + fast_forward = False + base = self.revision.get_common_ancestor_with(revision) + + local_text = self.materialize().encode('utf-8') + base_text = base.materialize().encode('utf-8') + other_text = revision.materialize().encode('utf-8') + + merge_text = self.merge_text(base_text, local_text, other_text) + + merge_revision = Revision.create( + text=merge_text, + parent=self.revision, + merge_parent=revision, + author=author, + author_name=author_name, + author_email=author_email, + description=description + ) + self.revision = merge_revision + self.save() + post_merge.send(sender=type(self), instance=self, fast_forward=fast_forward) - def materialize(self, change=None): - if self.head is None: - return u'' - if change is None: - change = self.head - elif not isinstance(change, Change): - change = self.change_set.get(pk=change) - return change.materialize() + def materialize(self): + return self.revision.materialize() if self.revision is not None else '' - def commit(self, text, author=None, author_name=None, author_email=None, - publishable=False, **kwargs): - """Commits a new revision. + def commit(self, text, parent=False, author=None, author_name=None, author_email=None, description=''): + """Creates a new revision and sets it as the ref. This will automatically merge the commit into the main branch, if parent is not document's head. :param unicode text: new version of the document - :param parent: parent revision (head, if not specified) - :type parent: Change or None :param User author: the commiter :param unicode author_name: commiter name (if ``author`` not specified) :param unicode author_email: commiter e-mail (if ``author`` not specified) - :param Tag[] tags: list of tags to apply to the new commit - :param bool publishable: set new commit as ready to publish :returns: new head """ - if 'parent' not in kwargs: - parent = self.head - else: - parent = kwargs['parent'] - if parent is not None and not isinstance(parent, Change): - parent = self.change_set.objects.get(pk=kwargs['parent']) - - tags = kwargs.get('tags', []) - if tags: - # set stage to next tag after the commited one - self.stage = max(tags, key=lambda t: t.ordering).next() - - change = self.change_set.create(author=author, - author_name=author_name, - author_email=author_email, - description=kwargs.get('description', ''), - publishable=publishable, - parent=parent) - - change.tags = tags - change.data.save('', ContentFile(text.encode('utf-8'))) - change.save() - - if self.head: - # merge new change as new head - self.head = self.head.merge_with(change, author=author, - author_name=author_name, - author_email=author_email) - else: - self.head = change - self.save() - - post_commit.send(sender=self.head) - - return self.head + if parent is False: + # If parent revision not set explicitly, use your head. + parent = self.revision + + # Warning: this will silently leave revs unreferenced. + rev = Revision.create( + text=text, + author=author, + author_name=author_name, + author_email=author_email, + description=description, + parent=parent + ) + self.merge_with(rev, author=author, author_name=author_name, author_email=author_email) + + post_commit.send(sender=type(self), instance=self) def history(self): - return self.change_set.filter(revision__gt=-1) - - def revision(self): - rev = self.change_set.aggregate( - models.Max('revision'))['revision__max'] - return rev - - def at_revision(self, rev): - """Returns a Change with given revision number.""" - return self.change_set.get(revision=rev) - - def publishable(self): - changes = self.change_set.filter(publishable=True) - if changes.exists(): - return changes.order_by('-created_at')[0] - else: - return None - - @transaction.commit_on_success - def prepend_history(self, other): - """Takes over the the other document's history and prepends to own.""" - - assert self != other - other_revs = other.change_set.all().count() - self.change_set.all().update(revision=models.F('revision') + other_revs) - other.change_set.all().update(tree=self) - assert not other.change_set.exists() - other.delete() + revs = self.revision.get_ancestors() + revs.add(self.revision) + return sorted(revs, key=lambda x: x.created_at)