~bzr-pqm/bzr/bzr.dev

# Copyright (C) 2005-2010 Canonical Ltd

#

# This program is free software; you can redistribute it and/or modify

# it under the terms of the GNU General Public License as published by

# the Free Software Foundation; either version 2 of the License, or

# (at your option) any later version.

#

# This program is distributed in the hope that it will be useful,

# but WITHOUT ANY WARRANTY; without even the implied warranty of

# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

# GNU General Public License for more details.

#

# You should have received a copy of the GNU General Public License

# along with this program; if not, write to the Free Software

# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

"""Tree classes, representing directory at point in time.

"""

import os

from collections import deque

import bzrlib

from bzrlib import (

    conflicts as _mod_conflicts,

    debug,

    delta,

    filters,

    osutils,

    revision as _mod_revision,

    rules,

    )

from bzrlib.decorators import needs_read_lock

from bzrlib.errors import BzrError, NoSuchId

from bzrlib import errors

from bzrlib.inventory import InventoryFile

from bzrlib.inter import InterObject

from bzrlib.osutils import fingerprint_file

from bzrlib.symbol_versioning import deprecated_function, deprecated_in

from bzrlib.trace import note

class Tree(object):

    """Abstract file tree.

    There are several subclasses:

    * `WorkingTree` exists as files on disk editable by the user.

    * `RevisionTree` is a tree as recorded at some point in the past.

    Trees contain an `Inventory` object, and also know how to retrieve

    file texts mentioned in the inventory, either from a working

    directory or from a store.

    It is possible for trees to contain files that are not described

    in their inventory or vice versa; for this use `filenames()`.

    Trees can be compared, etc, regardless of whether they are working

    trees or versioned trees.

    """

    def changes_from(self, other, want_unchanged=False, specific_files=None,

        extra_trees=None, require_versioned=False, include_root=False,

        want_unversioned=False):

        """Return a TreeDelta of the changes from other to this tree.

        :param other: A tree to compare with.

        :param specific_files: An optional list of file paths to restrict the

            comparison to. When mapping filenames to ids, all matches in all

            trees (including optional extra_trees) are used, and all children of

            matched directories are included.

        :param want_unchanged: An optional boolean requesting the inclusion of

            unchanged entries in the result.

        :param extra_trees: An optional list of additional trees to use when

            mapping the contents of specific_files (paths) to file_ids.

        :param require_versioned: An optional boolean (defaults to False). When

            supplied and True all the 'specific_files' must be versioned, or

            a PathsNotVersionedError will be thrown.

        :param want_unversioned: Scan for unversioned paths.

        The comparison will be performed by an InterTree object looked up on

        self and other.

        """

        # Martin observes that Tree.changes_from returns a TreeDelta and this

        # may confuse people, because the class name of the returned object is

        # a synonym of the object referenced in the method name.

        return InterTree.get(other, self).compare(

            want_unchanged=want_unchanged,

            specific_files=specific_files,

            extra_trees=extra_trees,

            require_versioned=require_versioned,

            include_root=include_root,

            want_unversioned=want_unversioned,

            )

    def iter_changes(self, from_tree, include_unchanged=False,

                     specific_files=None, pb=None, extra_trees=None,

                     require_versioned=True, want_unversioned=False):

        """See InterTree.iter_changes"""

        intertree = InterTree.get(from_tree, self)

        return intertree.iter_changes(include_unchanged, specific_files, pb,

            extra_trees, require_versioned, want_unversioned=want_unversioned)

    def conflicts(self):

        """Get a list of the conflicts in the tree.

        Each conflict is an instance of bzrlib.conflicts.Conflict.

        """

        return _mod_conflicts.ConflictList()

    def extras(self):

        """For trees that can have unversioned files, return all such paths."""

        return []

    def get_parent_ids(self):

        """Get the parent ids for this tree.

        :return: a list of parent ids. [] is returned to indicate

        a tree with no parents.

        :raises: BzrError if the parents are not known.

        """

        raise NotImplementedError(self.get_parent_ids)

    def has_filename(self, filename):

        """True if the tree has given filename."""

        raise NotImplementedError(self.has_filename)

    def has_id(self, file_id):

        return self.inventory.has_id(file_id)

    def __contains__(self, file_id):

        return self.has_id(file_id)

    def has_or_had_id(self, file_id):

        return self.inventory.has_id(file_id)

    def is_ignored(self, filename):

        """Check whether the filename is ignored by this tree.

        :param filename: The relative filename within the tree.

        :return: True if the filename is ignored.

        """

        return False

    def __iter__(self):

        return iter(self.inventory)

    def all_file_ids(self):

        """Iterate through all file ids, including ids for missing files."""

        return set(self.inventory)

    def id2path(self, file_id):

        """Return the path for a file id.

        :raises NoSuchId:

        """

        return self.inventory.id2path(file_id)

    @needs_read_lock

    def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):

        """Walk the tree in 'by_dir' order.

        This will yield each entry in the tree as a (path, entry) tuple.

        The order that they are yielded is:

        Directories are walked in a depth-first lexicographical order,

        however, whenever a directory is reached, all of its direct child

        nodes are yielded in  lexicographical order before yielding the

        grandchildren.

        For example, in the tree::

           a/

             b/

               c

             d/

               e

           f/

             g

        The yield order (ignoring root) would be::

          a, f, a/b, a/d, a/b/c, a/d/e, f/g

        :param yield_parents: If True, yield the parents from the root leading

            down to specific_file_ids that have been requested. This has no

            impact if specific_file_ids is None.

        """

        return self.inventory.iter_entries_by_dir(

            specific_file_ids=specific_file_ids, yield_parents=yield_parents)

    def iter_references(self):

        if self.supports_tree_reference():

            for path, entry in self.iter_entries_by_dir():

                if entry.kind == 'tree-reference':

                    yield path, entry.file_id

    def kind(self, file_id):

        raise NotImplementedError("Tree subclass %s must implement kind"

            % self.__class__.__name__)

    def stored_kind(self, file_id):

        """File kind stored for this file_id.

        May not match kind on disk for working trees.  Always available

        for versioned files, even when the file itself is missing.

        """

        return self.kind(file_id)

    def path_content_summary(self, path):

        """Get a summary of the information about path.

        All the attributes returned are for the canonical form, not the

        convenient form (if content filters are in use.)

        :param path: A relative path within the tree.

        :return: A tuple containing kind, size, exec, sha1-or-link.

            Kind is always present (see tree.kind()).

            size is present if kind is file and the size of the 

                canonical form can be cheaply determined, None otherwise.

            exec is None unless kind is file and the platform supports the 'x'

                bit.

            sha1-or-link is the link target if kind is symlink, or the sha1 if

                it can be obtained without reading the file.

        """

        raise NotImplementedError(self.path_content_summary)

    def get_reference_revision(self, file_id, path=None):

        raise NotImplementedError("Tree subclass %s must implement "

                                  "get_reference_revision"

            % self.__class__.__name__)

    def _comparison_data(self, entry, path):

        """Return a tuple of kind, executable, stat_value for a file.

        entry may be None if there is no inventory entry for the file, but

        path must always be supplied.

        kind is None if there is no file present (even if an inventory id is

        present).  executable is False for non-file entries.

        """

        raise NotImplementedError(self._comparison_data)

    def _file_size(self, entry, stat_value):

        raise NotImplementedError(self._file_size)

    def _get_inventory(self):

        return self._inventory

    def get_file(self, file_id, path=None):

        """Return a file object for the file file_id in the tree.

        If both file_id and path are defined, it is implementation defined as

        to which one is used.

        """

        raise NotImplementedError(self.get_file)

    def get_file_with_stat(self, file_id, path=None):

        """Get a file handle and stat object for file_id.

        The default implementation returns (self.get_file, None) for backwards

        compatibility.

        :param file_id: The file id to read.

        :param path: The path of the file, if it is known.

        :return: A tuple (file_handle, stat_value_or_None). If the tree has

            no stat facility, or need for a stat cache feedback during commit,

            it may return None for the second element of the tuple.

        """

        return (self.get_file(file_id, path), None)

    def get_file_text(self, file_id, path=None):

        """Return the byte content of a file.

        :param file_id: The file_id of the file.

        :param path: The path of the file.

        If both file_id and path are supplied, an implementation may use

        either one.

        """

        my_file = self.get_file(file_id, path)

        try:

            return my_file.read()

        finally:

            my_file.close()

    def get_file_lines(self, file_id, path=None):

        """Return the content of a file, as lines.

        :param file_id: The file_id of the file.

        :param path: The path of the file.

        If both file_id and path are supplied, an implementation may use

        either one.

        """

        return osutils.split_lines(self.get_file_text(file_id, path))

    def get_file_mtime(self, file_id, path=None):

        """Return the modification time for a file.

        :param file_id: The handle for this file.

        :param path: The path that this file can be found at.

            These must point to the same object.

        """

        raise NotImplementedError(self.get_file_mtime)

    def get_file_size(self, file_id):

        """Return the size of a file in bytes.

        This applies only to regular files.  If invoked on directories or

        symlinks, it will return None.

        :param file_id: The file-id of the file

        """

        raise NotImplementedError(self.get_file_size)

    def get_file_by_path(self, path):

        return self.get_file(self._inventory.path2id(path), path)

    def iter_files_bytes(self, desired_files):

        """Iterate through file contents.

        Files will not necessarily be returned in the order they occur in

        desired_files.  No specific order is guaranteed.

        Yields pairs of identifier, bytes_iterator.  identifier is an opaque

        value supplied by the caller as part of desired_files.  It should

        uniquely identify the file version in the caller's context.  (Examples:

        an index number or a TreeTransform trans_id.)

        bytes_iterator is an iterable of bytestrings for the file.  The

        kind of iterable and length of the bytestrings are unspecified, but for

        this implementation, it is a tuple containing a single bytestring with

        the complete text of the file.

        :param desired_files: a list of (file_id, identifier) pairs

        """

        for file_id, identifier in desired_files:

            # We wrap the string in a tuple so that we can return an iterable

            # of bytestrings.  (Technically, a bytestring is also an iterable

            # of bytestrings, but iterating through each character is not

            # performant.)

            cur_file = (self.get_file_text(file_id),)

            yield identifier, cur_file

    def get_symlink_target(self, file_id):

        """Get the target for a given file_id.

        It is assumed that the caller already knows that file_id is referencing

        a symlink.

        :param file_id: Handle for the symlink entry.

        :return: The path the symlink points to.

        """

        raise NotImplementedError(self.get_symlink_target)

    def get_canonical_inventory_paths(self, paths):

        """Like get_canonical_inventory_path() but works on multiple items.

        :param paths: A sequence of paths relative to the root of the tree.

        :return: A list of paths, with each item the corresponding input path

        adjusted to account for existing elements that match case

        insensitively.

        """

        return list(self._yield_canonical_inventory_paths(paths))

    def get_canonical_inventory_path(self, path):

        """Returns the first inventory item that case-insensitively matches path.

        If a path matches exactly, it is returned. If no path matches exactly

        but more than one path matches case-insensitively, it is implementation

        defined which is returned.

        If no path matches case-insensitively, the input path is returned, but

        with as many path entries that do exist changed to their canonical

        form.

        If you need to resolve many names from the same tree, you should

        use get_canonical_inventory_paths() to avoid O(N) behaviour.

        :param path: A paths relative to the root of the tree.

        :return: The input path adjusted to account for existing elements

        that match case insensitively.

        """

        return self._yield_canonical_inventory_paths([path]).next()

    def _yield_canonical_inventory_paths(self, paths):

        for path in paths:

            # First, if the path as specified exists exactly, just use it.

            if self.path2id(path) is not None:

                yield path

                continue

            # go walkin...

            cur_id = self.get_root_id()

            cur_path = ''

            bit_iter = iter(path.split("/"))

            for elt in bit_iter:

                lelt = elt.lower()

                new_path = None

                for child in self.iter_children(cur_id):

                    try:

                        # XXX: it seem like if the child is known to be in the

                        # tree, we shouldn't need to go from its id back to

                        # its path -- mbp 2010-02-11

                        #

                        # XXX: it seems like we could be more efficient

                        # by just directly looking up the original name and

                        # only then searching all children; also by not

                        # chopping paths so much. -- mbp 2010-02-11

                        child_base = os.path.basename(self.id2path(child))

                        if (child_base == elt):

                            # if we found an exact match, we can stop now; if

                            # we found an approximate match we need to keep

                            # searching because there might be an exact match

                            # later.  

                            cur_id = child

                            new_path = osutils.pathjoin(cur_path, child_base)

                            break

                        elif child_base.lower() == lelt:

                            cur_id = child

                            new_path = osutils.pathjoin(cur_path, child_base)

                    except NoSuchId:

                        # before a change is committed we can see this error...

                        continue

                if new_path:

                    cur_path = new_path

                else:

                    # got to the end of this directory and no entries matched.

                    # Return what matched so far, plus the rest as specified.

                    cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))

                    break

            yield cur_path

        # all done.

    def get_root_id(self):

        """Return the file_id for the root of this tree."""

        raise NotImplementedError(self.get_root_id)

    def annotate_iter(self, file_id,

                      default_revision=_mod_revision.CURRENT_REVISION):

        """Return an iterator of revision_id, line tuples.

        For working trees (and mutable trees in general), the special

        revision_id 'current:' will be used for lines that are new in this

        tree, e.g. uncommitted changes.

        :param file_id: The file to produce an annotated version from

        :param default_revision: For lines that don't match a basis, mark them

            with this revision id. Not all implementations will make use of

            this value.

        """

        raise NotImplementedError(self.annotate_iter)

    def _get_plan_merge_data(self, file_id, other, base):

        from bzrlib import versionedfile

        vf = versionedfile._PlanMergeVersionedFile(file_id)

        last_revision_a = self._get_file_revision(file_id, vf, 'this:')

        last_revision_b = other._get_file_revision(file_id, vf, 'other:')

        if base is None:

            last_revision_base = None

        else:

            last_revision_base = base._get_file_revision(file_id, vf, 'base:')

        return vf, last_revision_a, last_revision_b, last_revision_base

    def plan_file_merge(self, file_id, other, base=None):

        """Generate a merge plan based on annotations.

        If the file contains uncommitted changes in this tree, they will be

        attributed to the 'current:' pseudo-revision.  If the file contains

        uncommitted changes in the other tree, they will be assigned to the

        'other:' pseudo-revision.

        """

        data = self._get_plan_merge_data(file_id, other, base)

        vf, last_revision_a, last_revision_b, last_revision_base = data

        return vf.plan_merge(last_revision_a, last_revision_b,

                             last_revision_base)

    def plan_file_lca_merge(self, file_id, other, base=None):

        """Generate a merge plan based lca-newness.

        If the file contains uncommitted changes in this tree, they will be

        attributed to the 'current:' pseudo-revision.  If the file contains

        uncommitted changes in the other tree, they will be assigned to the

        'other:' pseudo-revision.

        """

        data = self._get_plan_merge_data(file_id, other, base)

        vf, last_revision_a, last_revision_b, last_revision_base = data

        return vf.plan_lca_merge(last_revision_a, last_revision_b,

                                 last_revision_base)

    def _iter_parent_trees(self):

        """Iterate through parent trees, defaulting to Tree.revision_tree."""

        for revision_id in self.get_parent_ids():

            try:

                yield self.revision_tree(revision_id)

            except errors.NoSuchRevisionInTree:

                yield self.repository.revision_tree(revision_id)

    @staticmethod

    def _file_revision(revision_tree, file_id):

        """Determine the revision associated with a file in a given tree."""

        revision_tree.lock_read()

        try:

            return revision_tree.inventory[file_id].revision

        finally:

            revision_tree.unlock()

    def _get_file_revision(self, file_id, vf, tree_revision):

        """Ensure that file_id, tree_revision is in vf to plan the merge."""

        if getattr(self, '_repository', None) is None:

            last_revision = tree_revision

            parent_keys = [(file_id, self._file_revision(t, file_id)) for t in

                self._iter_parent_trees()]

            vf.add_lines((file_id, last_revision), parent_keys,

                         self.get_file_lines(file_id))

            repo = self.branch.repository

            base_vf = repo.texts

        else:

            last_revision = self._file_revision(self, file_id)

            base_vf = self._repository.texts

        if base_vf not in vf.fallback_versionedfiles:

            vf.fallback_versionedfiles.append(base_vf)

        return last_revision

    inventory = property(_get_inventory,

                         doc="Inventory of this Tree")

    def _check_retrieved(self, ie, f):

        if not __debug__:

            return

        fp = fingerprint_file(f)

        f.seek(0)

        if ie.text_size is not None:

            if ie.text_size != fp['size']:

                raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),

                        ["inventory expects %d bytes" % ie.text_size,

                         "file is actually %d bytes" % fp['size'],

                         "store is probably damaged/corrupt"])

        if ie.text_sha1 != fp['sha1']:

            raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),

                    ["inventory expects %s" % ie.text_sha1,

                     "file is actually %s" % fp['sha1'],

                     "store is probably damaged/corrupt"])

    @needs_read_lock

    def path2id(self, path):

        """Return the id for path in this tree."""

        return self._inventory.path2id(path)

    def paths2ids(self, paths, trees=[], require_versioned=True):

        """Return all the ids that can be reached by walking from paths.

        Each path is looked up in this tree and any extras provided in

        trees, and this is repeated recursively: the children in an extra tree

        of a directory that has been renamed under a provided path in this tree

        are all returned, even if none exist under a provided path in this

        tree, and vice versa.

        :param paths: An iterable of paths to start converting to ids from.

            Alternatively, if paths is None, no ids should be calculated and None

            will be returned. This is offered to make calling the api unconditional

            for code that *might* take a list of files.

        :param trees: Additional trees to consider.

        :param require_versioned: If False, do not raise NotVersionedError if

            an element of paths is not versioned in this tree and all of trees.

        """

        return find_ids_across_trees(paths, [self] + list(trees), require_versioned)

    def iter_children(self, file_id):

        entry = self.iter_entries_by_dir([file_id]).next()[1]

        for child in getattr(entry, 'children', {}).itervalues():

            yield child.file_id

    def lock_read(self):

        """Lock this tree for multiple read only operations.

        :return: A bzrlib.lock.LogicalLockResult.

        """

        pass

    def revision_tree(self, revision_id):

        """Obtain a revision tree for the revision revision_id.

        The intention of this method is to allow access to possibly cached

        tree data. Implementors of this method should raise NoSuchRevision if

        the tree is not locally available, even if they could obtain the

        tree via a repository or some other means. Callers are responsible

        for finding the ultimate source for a revision tree.

        :param revision_id: The revision_id of the requested tree.

        :return: A Tree.

        :raises: NoSuchRevision if the tree cannot be obtained.

        """

        raise errors.NoSuchRevisionInTree(self, revision_id)

    def unknowns(self):

        """What files are present in this tree and unknown.

        :return: an iterator over the unknown files.

        """

        return iter([])

    def unlock(self):

        pass

    def filter_unversioned_files(self, paths):

        """Filter out paths that are versioned.

        :return: set of paths.

        """

        # NB: we specifically *don't* call self.has_filename, because for

        # WorkingTrees that can indicate files that exist on disk but that

        # are not versioned.

        pred = self.inventory.has_filename

        return set((p for p in paths if not pred(p)))

    def walkdirs(self, prefix=""):

        """Walk the contents of this tree from path down.

        This yields all the data about the contents of a directory at a time.

        After each directory has been yielded, if the caller has mutated the

        list to exclude some directories, they are then not descended into.

        The data yielded is of the form:

        ((directory-relpath, directory-path-from-root, directory-fileid),

        [(relpath, basename, kind, lstat, path_from_tree_root, file_id,

          versioned_kind), ...]),

         - directory-relpath is the containing dirs relpath from prefix

         - directory-path-from-root is the containing dirs path from /

         - directory-fileid is the id of the directory if it is versioned.

         - relpath is the relative path within the subtree being walked.

         - basename is the basename

         - kind is the kind of the file now. If unknonwn then the file is not

           present within the tree - but it may be recorded as versioned. See

           versioned_kind.

         - lstat is the stat data *if* the file was statted.

         - path_from_tree_root is the path from the root of the tree.

         - file_id is the file_id if the entry is versioned.

         - versioned_kind is the kind of the file as last recorded in the

           versioning system. If 'unknown' the file is not versioned.

        One of 'kind' and 'versioned_kind' must not be 'unknown'.

        :param prefix: Start walking from prefix within the tree rather than

        at the root. This allows one to walk a subtree but get paths that are

        relative to a tree rooted higher up.

        :return: an iterator over the directory data.

        """

        raise NotImplementedError(self.walkdirs)

    def supports_content_filtering(self):

        return False

    def _content_filter_stack(self, path=None, file_id=None):

        """The stack of content filters for a path if filtering is supported.

        Readers will be applied in first-to-last order.

        Writers will be applied in last-to-first order.

        Either the path or the file-id needs to be provided.

        :param path: path relative to the root of the tree

            or None if unknown

        :param file_id: file_id or None if unknown

        :return: the list of filters - [] if there are none

        """

        filter_pref_names = filters._get_registered_names()

        if len(filter_pref_names) == 0:

            return []

        if path is None:

            path = self.id2path(file_id)

        prefs = self.iter_search_rules([path], filter_pref_names).next()

        stk = filters._get_filter_stack_for(prefs)

        if 'filters' in debug.debug_flags:

            note("*** %s content-filter: %s => %r" % (path,prefs,stk))

        return stk

    def _content_filter_stack_provider(self):

        """A function that returns a stack of ContentFilters.

        The function takes a path (relative to the top of the tree) and a

        file-id as parameters.

        :return: None if content filtering is not supported by this tree.

        """

        if self.supports_content_filtering():

            return lambda path, file_id: \

                    self._content_filter_stack(path, file_id)

        else:

            return None

    def iter_search_rules(self, path_names, pref_names=None,

        _default_searcher=None):

        """Find the preferences for filenames in a tree.

        :param path_names: an iterable of paths to find attributes for.

          Paths are given relative to the root of the tree.

        :param pref_names: the list of preferences to lookup - None for all

        :param _default_searcher: private parameter to assist testing - don't use

        :return: an iterator of tuple sequences, one per path-name.

          See _RulesSearcher.get_items for details on the tuple sequence.

        """

        if _default_searcher is None:

            _default_searcher = rules._per_user_searcher

        searcher = self._get_rules_searcher(_default_searcher)

        if searcher is not None:

            if pref_names is not None:

                for path in path_names:

                    yield searcher.get_selected_items(path, pref_names)

            else:

                for path in path_names:

                    yield searcher.get_items(path)

    def _get_rules_searcher(self, default_searcher):

        """Get the RulesSearcher for this tree given the default one."""

        searcher = default_searcher

        return searcher

######################################################################

# diff

# TODO: Merge these two functions into a single one that can operate

# on either a whole tree or a set of files.

# TODO: Return the diff in order by filename, not by category or in

# random order.  Can probably be done by lock-stepping through the

# filenames from both trees.

def file_status(filename, old_tree, new_tree):

    """Return single-letter status, old and new names for a file.

    The complexity here is in deciding how to represent renames;

    many complex cases are possible.

    """

    old_inv = old_tree.inventory

    new_inv = new_tree.inventory