# Subversion 1.4/1.5 Python API backend

#

# Copyright(C) 2007 Daniel Holth et al

import pprint

import locale

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.core

    import svn.ra

    import svn.delta

    import svn

    import transport

except ImportError:

    pass

class CompatibilityException(Exception): pass

# 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.

            url, latest = url.rsplit("@", 1)

            latest = int(latest)

        except ValueError, e:

            pass

        self.url = url

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

        try:

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

            self.ra = self.transport.ra

            self.ctx = svn.client.create_context()

            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" % 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 getheads(self):

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

        optrev = svn.core.svn_opt_revision_t()

        optrev.kind = svn.core.svn_opt_revision_number

        optrev.value.number = self.last_changed

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

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

            self.head = self.revid(lt)

            self.heads = [self.head]

                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)

        else:

            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 gettags(self):

        tags = {}

        def parselogentry(*arg, **args):

            orig_paths, revnum, author, date, message, pool = arg

            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)

        start = self.revnum(self.head)

        try:

            svn.ra.get_log(self.ra, ['/tags'], 0, start, 0, True, False,

                           parselogentry)

            return tags

        except SubversionException:

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

            return {}

    # -- 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:

        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 _fetch_revisions(self, from_revnum = 0, to_revnum = 347):

        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...

            self.ui.debug('Ignoring %r since it is not under %r\n' % (path, module))

            return None

        received = []

        # svn.ra.get_log requires no other calls to the ra until it completes,

        # so we just collect the log entries and parse them afterwards

        def receivelog(*arg, **args):

            received.append(arg)

        self.child_cset = None

        def parselogentry(*arg, **args):

            orig_paths, revnum, author, date, message, pool = arg

            if self.is_blacklisted(revnum):

                self.ui.note('skipping blacklisted revision %d\n' % revnum)

                return

            self.ui.debug("parsing revision %d\n" % revnum)

            if orig_paths is None:

                self.ui.debug('revision %d has no entries\n' % revnum)

                return

            if revnum in self.modulemap:

                new_module = self.modulemap[revnum]

                if new_module != self.module:

                    self.module = new_module

                    self.reparent(self.module)

            copyfrom = {} # Map of entrypath, revision for finding source of deleted revisions.

            copies = {}

            entries = []

            rev = self.revid(revnum)

            parents = []

            # branch log might return entries for a parent we already have

            if (rev in self.commits or

                (revnum < self.lastrevs.get(self.module, 0))):

                return

            try:

                branch = self.module.split("/")[-1]

                if branch == 'trunk':

                    branch = ''

            except IndexError:

                branch = None

            paths = orig_paths.keys()

            paths.sort()

            for path in paths:

                # self.ui.write("path %s\n" % path)

                if path == self.module: # Follow branching back in history

                    ent = orig_paths[path]

                    if ent:

                        if ent.copyfrom_path:

                            # ent.copyfrom_rev may not be the actual last revision

                            prev = self.latest(ent.copyfrom_path, ent.copyfrom_rev)

                            self.modulemap[prev] = ent.copyfrom_path

                            parents = [self.revid(prev, ent.copyfrom_path)]

                            self.ui.note('found parent of branch %s at %d: %s\n' % \

                                         (self.module, prev, ent.copyfrom_path))

                        else:

                            self.ui.debug("No copyfrom path, don't know what to do.\n")

                            # Maybe it was added and there is no more history.

                entrypath = get_entry_from_path(path, module=self.module)

                # self.ui.write("entrypath %s\n" % entrypath)

                if entrypath is None:

                    # Outside our area of interest

                    self.ui.debug("boring@%s: %s\n" % (revnum, path))

                    continue

                entry = entrypath.decode(self.encoding)

                ent = orig_paths[path]

                kind = svn.ra.check_path(self.ra, entrypath, revnum)

                if kind == svn.core.svn_node_file:

                    if ent.copyfrom_path:

                        copyfrom_path = get_entry_from_path(ent.copyfrom_path)

                        if copyfrom_path:

                            self.ui.debug("Copied to %s from %s@%s\n" % (entry, copyfrom_path, ent.copyfrom_rev))

                            # It's probably important for hg that the source

                            # exists in the revision's parent, not just the

                            # ent.copyfrom_rev

                            fromkind = svn.ra.check_path(self.ra, copyfrom_path, ent.copyfrom_rev)

                            if fromkind != 0:

                                copies[self.recode(entry)] = self.recode(copyfrom_path)

                    entries.append(self.recode(entry))

                elif kind == 0: # gone, but had better be a deleted *file*

                    self.ui.debug("gone from %s\n" % ent.copyfrom_rev)

                    # if a branch is created but entries are removed in the same

                    # changeset, get the right fromrev

                    if parents:

                        uuid, old_module, fromrev = self.revsplit(parents[0])

                    else:

                        fromrev = revnum - 1

                        # might always need to be revnum - 1 in these 3 lines?

                        old_module = self.modulemap.get(fromrev, self.module)

                    basepath = old_module + "/" + get_entry_from_path(path, module=self.module)

                    entrypath = old_module + "/" + get_entry_from_path(path, module=self.module)

                    def lookup_parts(p):

                        rc = None

                        parts = p.split("/")

                        for i in range(len(parts)):

                            part = "/".join(parts[:i])

                            info = part, copyfrom.get(part, None)

                            if info[1] is not None:

                                self.ui.debug("Found parent directory %s\n" % info[1])

                                rc = info

                        return rc

                    self.ui.debug("base, entry %s %s\n" % (basepath, entrypath))

                    frompath, froment = lookup_parts(entrypath) or (None, revnum - 1)

                    # need to remove fragment from lookup_parts and replace with copyfrom_path

                    if frompath is not None:

                        self.ui.debug("munge-o-matic\n")

                        self.ui.debug(entrypath + '\n')

                        self.ui.debug(entrypath[len(frompath):] + '\n')

                        entrypath = froment.copyfrom_path + entrypath[len(frompath):]

                        fromrev = froment.copyfrom_rev

                        self.ui.debug("Info: %s %s %s %s\n" % (frompath, froment, ent, entrypath))

                    fromkind = svn.ra.check_path(self.ra, entrypath, fromrev)

                    if fromkind == svn.core.svn_node_file:   # a deleted file

                        entries.append(self.recode(entry))

                    elif fromkind == svn.core.svn_node_dir:

                        # print "Deleted/moved non-file:", revnum, path, ent

                        # children = self._find_children(path, revnum - 1)

                        # print "find children %s@%d from %d action %s" % (path, revnum, ent.copyfrom_rev, ent.action)

                        # Sometimes this is tricky. For example: in

                        # The Subversion Repository revision 6940 a dir

                        # was copied and one of its files was deleted 

                        # from the new location in the same commit. This

                        # code can't deal with that yet.

                        if ent.action == 'C':

                            children = self._find_children(path, fromrev)

                        else:

                            oroot = entrypath.strip('/')

                            nroot = path.strip('/')

                            children = self._find_children(oroot, fromrev)

                            children = [s.replace(oroot,nroot) for s in children]

                        # Mark all [files, not directories] as deleted.

                        for child in children:

                            # Can we move a child directory and its

                            # parent in the same commit? (probably can). Could

                            # cause problems if instead of revnum -1, 

                            # we have to look in (copyfrom_path, revnum - 1)

                            entrypath = get_entry_from_path("/" + child, module=old_module)

                            if entrypath:

                                entry = self.recode(entrypath.decode(self.encoding))

                                if entry in copies:

                                    # deleted file within a copy

                                    del copies[entry]

                                else:

                                    entries.append(entry)

                    else:

                        self.ui.debug('unknown path in revision %d: %s\n' % \

                                      (revnum, path))

                elif kind == svn.core.svn_node_dir:

                    # Should probably synthesize normal file entries

                    # and handle as above to clean up copy/rename handling.

                    # If the directory just had a prop change,

                    # then we shouldn't need to look for its children.

                    # Also this could create duplicate entries. Not sure

                    # whether this will matter. Maybe should make entries a set.

                    # print "Changed directory", revnum, path, ent.action, ent.copyfrom_path, ent.copyfrom_rev

                    # This will fail if a directory was copied

                    # from another branch and then some of its files

                    # were deleted in the same transaction.

                    children = self._find_children(path, revnum)

                    children.sort()

                    for child in children:

                        # Can we move a child directory and its

                        # parent in the same commit? (probably can). Could

                        # cause problems if instead of revnum -1, 

                        # we have to look in (copyfrom_path, revnum - 1)

                        entrypath = get_entry_from_path("/" + child, module=self.module)

                        # print child, self.module, entrypath

                        if entrypath:

                            # Need to filter out directories here...

                            kind = svn.ra.check_path(self.ra, entrypath, revnum)

                            if kind != svn.core.svn_node_dir:

                                entries.append(self.recode(entrypath))

                    # Copies here (must copy all from source)

                    # Probably not a real problem for us if

                    # source does not exist

                    # Can do this with the copy command "hg copy"

                    # if ent.copyfrom_path:

                    #     copyfrom_entry = get_entry_from_path(ent.copyfrom_path.decode(self.encoding),

                    #             module=self.module)

                    #     copyto_entry = entrypath

                    #

                    #     print "copy directory", copyfrom_entry, 'to', copyto_entry

                    #

                    #     copies.append((copyfrom_entry, copyto_entry))

                    if ent.copyfrom_path:

                        copyfrom_path = ent.copyfrom_path.decode(self.encoding)

                        copyfrom_entry = get_entry_from_path(copyfrom_path, module=self.module)

                        if copyfrom_entry:

                            copyfrom[path] = ent

                            self.ui.debug("mark %s came from %s\n" % (path, copyfrom[path]))

                            # Good, /probably/ a regular copy. Really should check

                            # to see whether the parent revision actually contains

                            # the directory in question.

                            children = self._find_children(self.recode(copyfrom_path), ent.copyfrom_rev)

                            children.sort()

                            for child in children:

                                entrypath = get_entry_from_path("/" + child, module=self.module)

                                if entrypath:

                                    entry = entrypath.decode(self.encoding)

                                    # print "COPY COPY From", copyfrom_entry, entry

                                    copyto_path = path + entry[len(copyfrom_entry):]

                                    copyto_entry =  get_entry_from_path(copyto_path, module=self.module)

                                    # print "COPY", entry, "COPY To", copyto_entry

                                    copies[self.recode(copyto_entry)] = self.recode(entry)