Integrating with Breezy¶
This document provides some general observations on integrating with Breezy and some recipes for typical tasks. It is intended to be useful to someone developing either a plugin or some other piece of software that integrates with brz. If you want to know about a topic that’s not covered here, just ask us.
Starting with breezy¶
When using breezy within the
brz program (for instance as a brz
plugin), breezy’s global state is already available for use.
From outside brz¶
To use breezy outside of
brz some global state needs to be setup.
breezy needs ways to handle user input, passwords, a place to emit
progress bars, logging setup appropriately for your program. The easiest
way to set all this up in the same fashion
brz does is to call
This returns a context manager within which breezy functions will work
correctly. See the pydoc for
breezy.initialize for more information.
(You can get away without entering the context manager, because the setup
work happens directly from
Running brz commands¶
To run command-line commands in-process:
from breezy.commands import get_command cmd = get_command('version') cmd.run()
This will send output through the current UIFactory; you can redirect this elsewhere through the parameters to breezy.initialize.
The code below have the prerequisite that you have imported either of
git module, or both:
import breezy.bzr import breezy.git
Depending on what the code is intended to do, the corresponding module or
modules need to be imported (
bzr for Bazaar, and
git for git), or some
of the code will not work, and will cause errors like
Manipulating the Working Tree¶
Most objects in Breezy are in files, named after the class they contain. To manipulate the Working Tree we need a valid WorkingTree object, which is loaded from the workingtree.py file, eg:
from breezy import workingtree wt = workingtree.WorkingTree.open('/home/jebw/brztest')
This gives us a WorkingTree object, which has various methods spread over itself, and its parent classes MutableTree and Tree - it’s worth having a look through these three files (workingtree.py, mutabletree.py and tree.py) to see which methods are available.
There are two methods for comparing trees:
iter_changes is more regular and precise, but it is
somewhat harder to work with. See the API documentation for more details.
changes_from creates a Delta object showing changes:
from breezy import delta changes = wt.changes_from(wt.basis_tree())
This gives us a Delta object, which has several lists of files for each type of change, eg changes.added is a list of added files, changes.removed is list of removed files, changes.modified is a list of modified files. The contents of the lists aren’t just filenames, but include other information as well. To grab just the filename we want the first value, eg:
print("list of newly added files") for filename in changes.added: print("%s has been added" % filename)
The exception to this is changes.renamed, where the list returned for each renamed files contains both the old and new names – one or both may interest you, depending on what you’re doing.
print("list of renamed files") for filename in changes.renamed: print("%s has been renamed to %s" % (filename, filename))
If you want to add files the same way
brz add does, you can use
MutableTree.smart_add. By default, this is recursive. Paths can either be
absolute or relative to the workingtree:
wt.smart_add(['dir1/filea.txt', 'fileb.txt', '/home/jebw/brztesttree/filec.txt'])
For more precise control over which files to add, use MutableTree.add:
wt.add(['dir1/filea.txt', 'fileb.txt', '/home/jebw/brztesttree/filec.txt'])
You can remove multiple files at once. The file paths need to be relative to the workingtree:
wt.remove(['filea.txt', 'fileb.txt', 'dir1'])
By default, the files are not deleted, just removed from the inventory. To delete them from the filesystem as well:
wt.remove(['filea.txt', 'fileb.txt', 'dir1'], keep_files=False)
Renaming a File¶
You can rename one file to a different name using WorkingTree.rename_one. You just provide the old and new names, eg:
You can move multiple files from one directory into another using WorkingTree.move:
More complicated renames/moves can be done with transform.TreeTransform, which is outside the scope of this document.
To commit _all_ the changes to our working tree we can just call the WorkingTree’s commit method, giving it a commit message, eg:
wt.commit('this is my commit message')
To commit only certain files, we need to provide a list of filenames which we want committing, eg:
wt.commit(message='this is my commit message', specific_files=['fileA.txt', 'dir2/fileB.txt', 'fileD.txt'])
Generating a Log for a File¶
Generating a log is, in itself, simple. Grab a branch (see below) and pass it to show_log together with a log formatter, eg:
from breezy import log from breezy import branch b = branch.Branch.open('/path/to/bazaar/branch') lf = log.LongLogFormatter(to_file=sys.stdout) log.show_log(b, lf)
Three log formatters are included with breezy: LongLogFormatter, ShortLogFormatter and LineLogFormatter. These provide long, short and single-line log output formats. It’s also possible to write your own in very little code.
Annotating a File¶
To annotate a file, we want to walk every line of a file, retrieving the revision which last modified/created that line and then retrieving the information for that revision.
First we get an annotation iterator for the file we are interested in:
tree, relpath = workingtree.WorkingTree.open_containing('/path/to/file.txt') fileid = tree.path2id(relpath) annotation = list(tree.annotate_iter(fileid))
To avoid repeatedly retrieving the same revisions we grab all revisions associated with the file at once and build up a map of id to revision information. We also build an map of revision numbers, again indexed by the revision id:
revision_ids = set(revision_id for revision_id, text in annotation) revisions = tree.branch.repository.get_revisions(revision_ids) revision_map = dict(izip(revision_ids, revisions)) revno_map = tree.branch.get_revision_id_to_revno_map()
Finally, we use our annotation iterator to walk the lines of the file, displaying the information from our revision maps as we go:
for revision_id, text in annotation : rev = revision_map[revision_id] revno = revno_map[revision_id] revno_string = '.'.join(str(i) for i in revno) print("%s, %s: %s" % (revno_string, rev.committer, text))
Working with branches¶
To work with a branch you need a branch object, created from your branch:
from breezy import branch b = branch.Branch.open('/home/jebw/brztest')
Branching from an existing branch¶
To branch you create a branch object representing the branch you are branching from, and supply a path/url to the new branch location. The following code clones the brz trunk branch (the latest copy of the Breezy source code) - be warned it has to download 60meg so takes a while to run with no feedback:
from breezy import branch b = branch.Branch.open('bzr+ssh://bazaar.launchpad.net/+branch/brz/') nb = b.controldir.sprout('/tmp/newBrzBzranch').open_branch()
This provides no feedback, since Breezy automatically uses the ‘silent’ UI.
Pushing and pulling branches¶
To push a branch you need to open the source and destination branches, then just call push with the other branch as a parameter:
from breezy import branch b1 = branch.Branch.open('file:///home/user/mybranch') b2 = branch.Branch.open('bzr+ssh://bazaar.launchpad.net/+branch/brz/') b1.push(b2)
Pulling is much the same:
If you have a working tree, as well as a branch, you should use WorkingTree.pull, not Branch.pull.
This won’t handle conflicts automatically though, so any conflicts will be left in the working tree for the user to resolve.
Checkout from an existing branch¶
This performs a Lightweight checkout from an existing Branch:
from breezy import bzrdir accelerator_tree, source = bzrdir.BzrDir.open_tree_or_branch('http:URL') source.create_checkout('/tmp/newBrzCheckout', None, True, accelerator_tree)
To make a heavyweight checkout, change the last line to:
source.create_checkout('/tmp/newBrzCheckout', None, False, accelerator_tree
Finding the last revision number or id¶
To get the last revision number and id of a branch use:
revision_number, revision_id = branch.last_revision_info()
If all you care about is the revision_id there is also the method:
revision_id = branch.last_revision()
Getting the list of revision ids that make up a branch¶
IMPORTANT: This should be avoided wherever possible, as it scales with the length of history:
revisions = branch.revision_history()
now revisions is the revision id of the first commit, and revs[-1] is the revision id of the most recent. Note that if all you want is the last revision then you should use branch.last_revision() as described above, as it is vastly more efficient.
Getting a Revision object from a revision id¶
The Revision object has attributes like “message” to get the information about the revision:
repo = branch.repository revision = repo.get_revision(rev_id)
Accessing the files from a revision¶
To get the file contents and tree shape for a specific revision you need a RevisionTree. These are supplied by the repository for a specific revision id:
revtree = repo.revision_tree(rev_id)
RevisionTrees, like all trees, can be compared as described in “Comparing Trees” above.
The most common way to list files in a tree is
The simplest way to get file content is
Tree.get_file(). The best way
to retrieve file content for large numbers of files Tree.iter_files_bytes()`