# Subversion 1.4/1.5 Python API backend

#

# Copyright(C) 2007 Daniel Holth et al

#

# Configuration options:

#

# convert.svn.trunk

#   Relative path to the trunk (default: "trunk")

# convert.svn.branches

#   Relative path to tree of branches (default: "branches")

#

# Set these in a hgrc, or on the command line as follows:

#

#   hg convert --config convert.svn.trunk=wackoname [...]

import locale

import os

import cPickle as pickle

from mercurial import util

# Subversion stuff. Works best with very recent Python SVN bindings

# e.g. SVN 1.5 or backports. Thanks to the bzr folks for enhancing

# these bindings.

from cStringIO import StringIO

from common import NoRepo, commit, converter_source

try:

    from svn.core import SubversionException, Pool

    import svn

    import svn.client

    import svn.core

    import svn.ra

    import svn.delta

    import transport

except ImportError:

    pass

def geturl(path):

    try:

        return svn.client.url_from_path(svn.core.svn_path_canonicalize(path))

    except SubversionException:

        pass

    if os.path.isdir(path):

        return 'file://%s' % os.path.normpath(os.path.abspath(path))

    return path

def optrev(number):

    optrev = svn.core.svn_opt_revision_t()

    optrev.kind = svn.core.svn_opt_revision_number

    optrev.value.number = number

    return optrev

class changedpath(object):

    def __init__(self, p):

        self.copyfrom_path = p.copyfrom_path

        self.copyfrom_rev = p.copyfrom_rev

        self.action = p.action

# SVN conversion code stolen from bzr-svn and tailor

class convert_svn(converter_source):

    def __init__(self, ui, url, rev=None):

        super(convert_svn, self).__init__(ui, url, rev=rev)

        try:

            SubversionException

        except NameError:

            msg = 'subversion python bindings could not be loaded\n'

            ui.warn(msg)

            raise NoRepo(msg)

        self.encoding = locale.getpreferredencoding()

        self.lastrevs = {}

        latest = None

        if rev:

            try:

                latest = int(rev)

            except ValueError:

                raise util.Abort('svn: revision %s is not an integer' % rev)

        try:

            # Support file://path@rev syntax. Useful e.g. to convert

            # deleted branches.

            at = url.rfind('@')

            if at >= 0:

                latest = int(url[at+1:])

                url = url[:at]

        except ValueError, e:

            pass

        self.url = geturl(url)

        self.encoding = 'UTF-8' # Subversion is always nominal UTF-8

        try:

            self.transport = transport.SvnRaTransport(url=self.url)

            self.ra = self.transport.ra

            self.ctx = self.transport.client

            self.base = svn.ra.get_repos_root(self.ra)

            self.module = self.url[len(self.base):]

            self.modulemap = {} # revision, module

            self.commits = {}

            self.files = {}

            self.uuid = svn.ra.get_uuid(self.ra).decode(self.encoding)

        except SubversionException, e:

            raise NoRepo("couldn't open SVN repo %s" % self.url)

        try:

            self.get_blacklist()

        except IOError, e:

            pass

        self.last_changed = self.latest(self.module, latest)

        self.head = self.revid(self.last_changed)

    def setrevmap(self, revmap):

        lastrevs = {}

        for revid in revmap.keys():

            uuid, module, revnum = self.revsplit(revid)

            lastrevnum = lastrevs.setdefault(module, revnum)

            if revnum > lastrevnum:

                lastrevs[module] = revnum

        self.lastrevs = lastrevs

    def exists(self, path, optrev):

        try:

            return svn.client.ls(self.url.rstrip('/') + '/' + path,

                                 optrev, False, self.ctx)

        except SubversionException, err:

            return []

    def getheads(self):

        # detect standard /branches, /tags, /trunk layout

        rev = optrev(self.last_changed)

        rpath = self.url.strip('/')

        cfgtrunk = self.ui.config('convert', 'svn.trunk')

        cfgbranches = self.ui.config('convert', 'svn.branches')

        trunk = (cfgtrunk or 'trunk').strip('/')

        branches = (cfgbranches or 'branches').strip('/')

        if self.exists(trunk, rev) and self.exists(branches, rev):

            self.ui.note('found trunk at %r and branches at %r\n' %

                         (trunk, branches))

            oldmodule = self.module

            self.module += '/' + trunk

            lt = self.latest(self.module, self.last_changed)

            self.head = self.revid(lt)

            self.heads = [self.head]

            branchnames = svn.client.ls(rpath + '/' + branches, rev, False,

                                        self.ctx)

            for branch in branchnames.keys():

                if oldmodule:

                    module = '/' + oldmodule + '/' + branches + '/' + branch

                else:

                    module = '/' + branches + '/' + branch

                brevnum = self.latest(module, self.last_changed)

                brev = self.revid(brevnum, module)

                self.ui.note('found branch %s at %d\n' % (branch, brevnum))

                self.heads.append(brev)

        elif cfgtrunk or cfgbranches:

            raise util.Abort(_('trunk/branch layout expected, '

                               'but not found'))

        else:

            self.ui.note('working with one branch\n')

            self.heads = [self.head]

        return self.heads

    def getfile(self, file, rev):

        data, mode = self._getfile(file, rev)

        self.modecache[(file, rev)] = mode

        return data

    def getmode(self, file, rev):

        return self.modecache[(file, rev)]

    def getchanges(self, rev):

        self.modecache = {}

        files = self.files[rev]

        cl = files

        cl.sort()

        # caller caches the result, so free it here to release memory

        del self.files[rev]

        return cl

    def getcommit(self, rev):

        if rev not in self.commits:

            uuid, module, revnum = self.revsplit(rev)

            self.module = module

            self.reparent(module)

            stop = self.lastrevs.get(module, 0)

            self._fetch_revisions(from_revnum=revnum, to_revnum=stop)

        commit = self.commits[rev]

        # caller caches the result, so free it here to release memory

        del self.commits[rev]

        return commit

    def get_log(self, paths, start, end, limit=0, discover_changed_paths=True,

                strict_node_history=False):

        '''wrapper for svn.ra.get_log.

        on a large repository, svn.ra.get_log pins huge amounts of

        memory that cannot be recovered.  work around it by forking

        and writing results over a pipe.'''

        def child(fp):

            protocol = -1

            def receiver(orig_paths, revnum, author, date, message, pool):

                if orig_paths is not None:

                    for k, v in orig_paths.iteritems():

                        orig_paths[k] = changedpath(v)

                pickle.dump((orig_paths, revnum, author, date, message),

                            fp, protocol)

            try:

                # Use an ra of our own so that our parent can consume

                # our results without confusing the server.

                t = transport.SvnRaTransport(url=self.url)

                svn.ra.get_log(t.ra, paths, start, end, limit,

                               discover_changed_paths,

                               strict_node_history,

                               receiver)

            except SubversionException, (_, num):

                self.ui.print_exc()

                pickle.dump(num, fp, protocol)

            else:

                pickle.dump(None, fp, protocol)

            fp.close()

        def parent(fp):

            while True:

                entry = pickle.load(fp)

                try:

                    orig_paths, revnum, author, date, message = entry

                except:

                    if entry is None:

                        break

                    raise SubversionException("child raised exception", entry)

                yield entry

        rfd, wfd = os.pipe()

        pid = os.fork()

        if pid:

            os.close(wfd)

            for p in parent(os.fdopen(rfd, 'rb')):

                yield p

            ret = os.waitpid(pid, 0)[1]

            if ret:

                raise util.Abort(_('get_log %s') % util.explain_exit(ret))

        else:

            os.close(rfd)

            child(os.fdopen(wfd, 'wb'))

            os._exit(0)

    def gettags(self):

        tags = {}

        start = self.revnum(self.head)

        try:

            for entry in self.get_log(['/tags'], 0, start):

                orig_paths, revnum, author, date, message = entry

                for path in orig_paths:

                    if not path.startswith('/tags/'):

                        continue

                    ent = orig_paths[path]

                    source = ent.copyfrom_path

                    rev = ent.copyfrom_rev

                    tag = path.split('/', 2)[2]

                    tags[tag] = self.revid(rev, module=source)

        except SubversionException, (_, num):

            self.ui.note('no tags found at revision %d\n' % start)

        return tags

    # -- helper functions --

    def revid(self, revnum, module=None):

        if not module:

            module = self.module

        return (u"svn:%s%s@%s" % (self.uuid, module, revnum)).decode(self.encoding)

    def revnum(self, rev):

        return int(rev.split('@')[-1])

    def revsplit(self, rev):

        url, revnum = rev.encode(self.encoding).split('@', 1)

        revnum = int(revnum)

        parts = url.split('/', 1)

        uuid = parts.pop(0)[4:]

        mod = ''

        if parts:

            mod = '/' + parts[0]

        return uuid, mod, revnum

    def latest(self, path, stop=0):

        'find the latest revision affecting path, up to stop'

        if not stop:

            stop = svn.ra.get_latest_revnum(self.ra)

        try:

            self.reparent('')

            dirent = svn.ra.stat(self.ra, path.strip('/'), stop)

            self.reparent(self.module)

        except SubversionException:

            dirent = None

        if not dirent:

            print self.base, path

            raise util.Abort('%s not found up to revision %d' % (path, stop))

        return dirent.created_rev

    def get_blacklist(self):

        """Avoid certain revision numbers.

        It is not uncommon for two nearby revisions to cancel each other

        out, e.g. 'I copied trunk into a subdirectory of itself instead

        of making a branch'. The converted repository is significantly

        smaller if we ignore such revisions."""

        self.blacklist = set()

        blacklist = self.blacklist

        for line in file("blacklist.txt", "r"):

            if not line.startswith("#"):

                try:

                    svn_rev = int(line.strip())

                    blacklist.add(svn_rev)

                except ValueError, e:

                    pass # not an integer or a comment

    def is_blacklisted(self, svn_rev):

        return svn_rev in self.blacklist

    def reparent(self, module):

        svn_url = self.base + module

        self.ui.debug("reparent to %s\n" % svn_url.encode(self.encoding))

        svn.ra.reparent(self.ra, svn_url.encode(self.encoding))

        def get_entry_from_path(path, module=self.module):

            # Given the repository url of this wc, say

            #   "http://server/plone/CMFPlone/branches/Plone-2_0-branch"

            # extract the "entry" portion (a relative path) from what

            # svn log --xml says, ie

            #   "/CMFPlone/branches/Plone-2_0-branch/tests/PloneTestCase.py"

            # that is to say "tests/PloneTestCase.py"

            if path.startswith(module):

                relative = path[len(module):]

                if relative.startswith('/'):

                    return relative[1:]

                else:

                    return relative

            # The path is outside our tracked tree...

            return None