Branch concatenation on complex repositories

do_test.py

@@ -48,6 +48,44 @@ def fail( msg , fd=os.sys.stdout ) :
 os.chdir('../..')
 
 
+os.chdir( 'tests/branchcount.git' )
+repo = git_workflow_quality.Repository()
+count = len(repo.branches)
+if count == 5 :
+    ret += ok( "branch count working" )
+else :
+    ret += fail( "counted %d branches" % count )
+os.chdir('../..')
+
+
+os.chdir( 'tests/complexnetwork.git' )
+repo = git_workflow_quality.Repository()
+count = len([ c for c in repo.values() if not c.child ])
+if count == 2 :
+    ret += ok( "open branches properly found" )
+else :
+    ret += fail( "found %d opened branches" % count )
+#
+count = repo.event_list()[0]['multimerged']
+if count == 1 :
+    ret += ok( "1 multimerged found" )
+else :
+    ret += fail( "%d multimerged found" % count )
+#
+count = repo.event_list()[0]['conflict']
+if count == 1 :
+    ret += ok( "1 conflict found" )
+else :
+    ret += fail( "%d conflict found" % count )
+#
+count = len(repo.branches)
+if count == 8 :
+    ret += ok( "complex branches properly concatenated" )
+else :
+    ret += fail( "complex concatenation produced %d branches" % count )
+os.chdir('../..')
+
+
 def test_event ( event ) :
     os.chdir( 'tests/%s.git' % event )
     repo = git_workflow_quality.Repository()

git_workflow_quality/__init__.py

@@ -44,6 +44,8 @@ def add_child ( self , child ) :
             self.child = child
         elif child not in self.forks :
             self.forks.append( child )
+        if not child.child and child.forks :
+            child.child = child.forks.pop(0)
 
     def get_parents ( self , full=True ) :
         parents = []
@@ -102,6 +104,9 @@ def merges ( self ) :
     def stats ( self ) :
         return len(self), len(self.commits()), len(self.merges())
 
+    def matches ( self , other ) :
+        return self.name.startswith( other.name ) or other.name.startswith( self.name )
+
     def begin ( self ) :
         begins = [ c for c in list.__iter__(self) if not c.parent ]
         if not begins :
@@ -113,7 +118,7 @@ def end ( self ) :
         ends = [ c for c in list.__iter__(self) if not c.child ]
         if not ends :
             ends = [ c for c in list.__iter__(self) if c.child.branch != self ]
-            if not ends :
+            if len(ends) != 1 :
                 # This just searches end on direct childs, and will not detect indirect merge-backs
                 ends = [ end for end in ends if not end.child or not end.child.branch.end().child or end.child.branch.end().child.branch != self ]
         assert len(ends) == 1
@@ -378,12 +383,32 @@ def get_branches () :
             fd.close()
     return branches
 
+class BranchList ( list ) :
+
+    def __init__ ( self ) :
+        self.names = []
+        list.__init__( self )
+
+    def get ( self , name ) :
+        match = [ b for b in self if b.name == name ]
+        if item.name in self.names :
+            assert len(match) == 1
+            return match[0]
+        assert not match
+        return None
+
+    def append ( self , item ) :
+        if item.name in self.names : assert False
+        self.names.append( item.name )
+        list.append( self , item )
+
 class Repository ( dict ) :
 
-  def __init__ ( self ) :
+  def __init__ ( self , concatenate=True ) :
 
     self.order = []
-    self.branches = []
+    self.branches = BranchList()
+    self.concatenated = concatenate
 
     cmd = subprocess.Popen( ['git', 'log', '--all', '--format="%H \"%ae\" \"%ce\" %s"'] , stdout=subprocess.PIPE )
     line = cmd.stdout.readline()
@@ -408,6 +433,9 @@ def __init__ ( self ) :
 
     self.set_childs()
 
+    if self.concatenated :
+        self.concatenated = self.concatenation()
+
   def set_params ( self , sha , line ) :
       self[sha].author_date = int(line[0])
       self[sha].committer_date = int(line[1])
@@ -419,8 +447,8 @@ def set_params ( self , sha , line ) :
                   raise Exception( "Octopus merges on %s from %s not handled" % ( self[sha].sha , ", ".join([c.sha for c in self[sha].parents]) ) )
 
   def new_branch ( self , branchname , orphan=False ) :
-      match = [ branch for branch in self.branches if branch.name == branchname ]
-      if match :
+      if branchname in self.branches.names :
+          match = [ branch for branch in self.branches if branch.name == branchname ]
           assert len(match) == 1
           if match[0].is_primary() :
               return match[0]
@@ -429,19 +457,34 @@ def new_branch ( self , branchname , orphan=False ) :
           self.branches.append( Branch(branchname, orphan) )
           return self.branches[-1]
 
+  def join_branch ( self , source , branch ) :
+        source.set_child( branch.begin() )
+        for commit in list(branch) :
+            source.branch.append( commit )
+        self.branches.remove(branch)
+        if len(source.branch.name) > len(branch.name) :
+            source.branch.name = branch.name
+
   def events( self , details=False) :
       output = ['']
       event_list , msgs = self.event_list( details )
 
       if msgs : output.extend( msgs )
 
       output.append( "Branch events" )
+      if self.concatenated and self.base_events : output[-1] += "   %+10d branches" % self.concatenated
       output.append( "  multitarget   %4d" % event_list['multitarget'] )
+      if self.concatenated and self.base_events : output[-1] += "  %4d" % self.base_events['multitarget']
       output.append( "  reutilized    %4d" % event_list['reutilized'] )
+      if self.concatenated and self.base_events : output[-1] += "  %4d" % self.base_events['reutilized']
       output.append( "  multimerged   %4d" % event_list['multimerged'] )
+      if self.concatenated and self.base_events : output[-1] += "  %4d" % self.base_events['multimerged']
       output.append( "  indirect      %4d" % event_list['indirect'] )
+      if self.concatenated and self.base_events : output[-1] += "  %4d" % self.base_events['indirect']
       output.append( "  multisource   %4d" % event_list['multisource'] )
+      if self.concatenated and self.base_events : output[-1] += "  %4d" % self.base_events['multisource']
       output.append( "  conflict      %4d" % event_list['conflict'] )
+      if self.concatenated and self.base_events : output[-1] += "  %4d" % self.base_events['conflict']
       output.append( "" )
 
       return "\n".join(output)
@@ -655,27 +698,53 @@ def set_childs ( self ) :
             branch.ancestry( c )
     self.order.reverse()
 
-    n , m = 0 , 0
+    m = 0
+    # Remove items within the loop is not safe
+    __branches = [ branch for branch in self.branches if len(branch) == 0 ]
+    for branch in __branches :
+        self.branches.remove(branch)
+        m += 1
+    if m :
+        print "ERROR : generated %d empty branches" % m
+
+  def concatenation ( self ) :
+
+    self.base_events , drop = self.event_list()
+
+    n = 0
     # Remove items within the loop is not safe
     __branches = [ branch for branch in self.branches ]
     for branch in __branches :
-        if len(branch) == 0 :
-            self.branches.remove(branch)
-            m += 1
+
+        begin = branch.begin()
+        if not begin.parent : continue
+
+        source = begin.parent
+        if not source.forks and not begin.parents :
+            # This is in fact a plain commit with a single edge. As ancestry assigns childs, it cannot be detected there
+            self.join_branch( source , branch )
+            n += 1
             continue
-        source = branch.begin().parent
-        if not source : continue
-        if ( source.child and source.child.branch == branch ) or \
-                branch.name.startswith( source.branch.name ) or source.branch.name.startswith( branch.name ) :
-            source.set_child( branch.begin() )
-            for commit in list(branch) :
-                source.branch.append( commit )
-            assert len(branch) == 0
-            self.branches.remove(branch)
-            source.branch.name = branch.name
+
+        if [ c for c in begin.get_childs() if c.branch == source.branch ] :
+            print "WARNING : %s merged back into %s" % ( begin.branch.pretty() , source.branch.pretty() )
+            continue
+
+        # Concatenation happens in two cases :
+        #   incoming merge : a parent commit whose single child is the first commit on branch
+        #   outgoing merge : the first commit has a single parent, and this is the only one of their childs with a single parent
+        candidates = [ c for c in begin.get_parents() if not c.forks ] + [ c.parent for c in begin.parent.get_childs() if not begin.parents and not c.parents ]
+        if len(candidates) != 1 : continue
+
+        source = candidates[0]
+        if source.branch.end() == source and branch.matches( source.branch ) :
+            self.join_branch( source , branch )
             n += 1
+
     if n :
         print "WARNING : %d branches removed by concatenation with parents" % n
-    if m :
-        print "ERROR : generated %d empty branches" % m
+    else :
+        self.base_events = []
+
+    return n
 

gwfqa

@@ -10,10 +10,11 @@ if __name__ == '__main__' :
     parser = argparse.ArgumentParser(description='Extract simple analytics from git repository')
     parser.add_argument('--detailed', action='store_true', help='Show per-branch statistics')
     parser.add_argument('--graph', choices=['topo', 'date', 'backwards'], help='Generate gitgraphjs file')
+    parser.add_argument('--no-concatenate', dest='concatenate', action='store_false', help='Express branch relations in dot language')
     parser.add_argument('--dot', help='Express branch relations in dot language')
     args = parser.parse_args()
 
-    repo = git_workflow_quality.Repository()
+    repo = git_workflow_quality.Repository(args.concatenate)
 
     print repo.report(args.detailed)
 

tests/branchcount.git/HEAD

@@ -0,0 +1 @@
+ref: refs/heads/branch3

tests/branchcount.git/config

@@ -0,0 +1,8 @@
+[core]
+	repositoryformatversion = 0
+	filemode = false
+	bare = false
+	logallrefupdates = true
+	symlinks = false
+	ignorecase = true
+	hideDotFiles = dotGitOnly

tests/branchcount.git/objects/0b/bb2de8671e2453f096d51cd4582aa197949835

@@ -0,0 +1,5 @@
+x
��Mj1@�}
+�A��A)]g�+hdO2e&\����3t���g��-����
�g���N�-I���T�
+
+Zʹ�1���v�5H�j�$�
+U�J�sc_+���g��q�;��wi.��-�||᱾���.����|$�<�#!����p��n�z��?�FF{

tests/branchcount.git/objects/84/fd538a36d65b0cf3ab7f6c503af812fce9cb1a

@@ -0,0 +1,2 @@
+x
��I
+1@Q�9E��T:��v�*�D[z����gp����u��Da�[)�Cn0�X��R���k�^��C��v�����A�L�f/���l0A��=U+1�G����w��.�K�+O,����c��{��<NY�hk(��f�=���6�7�W��"wT?��G�

tests/branchcount.git/refs/heads/branch1

@@ -0,0 +1 @@
+4bd1087aaf4d5b24fa17b6aaba80f8f4bc90502a

tests/branchcount.git/refs/heads/branch2

@@ -0,0 +1 @@
+84fd538a36d65b0cf3ab7f6c503af812fce9cb1a

tests/branchcount.git/refs/heads/branch3

@@ -0,0 +1 @@
+45a1efea7ca2d55fae24a8f62f29feeee4990313

tests/branchcount.git/refs/heads/branch4

@@ -0,0 +1 @@
+0bbb2de8671e2453f096d51cd4582aa197949835

tests/branchcount.git/refs/heads/master

@@ -0,0 +1 @@
+a6670da021af97b9421a1ae49cb1f9bd5b60a6ed

tests/complexnetwork.git/HEAD

@@ -0,0 +1 @@
+ref: refs/heads/master

tests/complexnetwork.git/config

@@ -0,0 +1,7 @@
+[core]
+	repositoryformatversion = 0
+	filemode = false
+	bare = true
+	symlinks = false
+	ignorecase = true
+	hideDotFiles = dotGitOnly

tests/complexnetwork.git/description

@@ -0,0 +1 @@
+Unnamed repository; edit this file 'description' to name the repository.

tests/complexnetwork.git/info/exclude

@@ -0,0 +1,6 @@
+# git ls-files --others --exclude-from=.git/info/exclude
+# Lines that start with '#' are comments.
+# For a project mostly in C, the following would be a good set of
+# exclude patterns (uncomment them if you want to use them):
+# *.[oa]
+# *~

tests/complexnetwork.git/objects/07/c277ea283e7d175e0e03aaf5e15a3377ed388c

@@ -0,0 +1,2 @@
+x
��Kj1��}
+��c�/!�zY��)��z�����_�u�8�>FWr��RT�h]��}�PSk�|LSiC�l�u 8Ҋ�`��|��b֪�d��$\j4���Í߳v��eޟp�9�c��~�</����<"��1��:k�Q��q�qS:or'����H

tests/complexnetwork.git/objects/09/3ce978b81acfd8a6cd592743f08b05ab60d997

@@ -0,0 +1,2 @@
+x
��MJCA�]�S�.Azz�A$kA�
+����&�x|��*꣪�x_׹�Л�
b�'$�9�E���'�IjɁ�:�5ݺ)�G�C��Scd,
�P�)r���O�-�lc��
��SE���2(}����+}���;-��~7��
n���Ҽ<���9���# �4�8�G�_��M�Yͩ��s��x0���_��o��_?

tests/complexnetwork.git/objects/0d/58300606997e8eacf8504197c65982d0cbb8eb

@@ -0,0 +1 @@
+x
��MJ1�a�}��f!HU*�T@ĵ x�JR���L�oz
�/�|P�ض�|�M���&���3+ ��>���$����4ݻQϕ=��D3ZF[x�JL!V\b��w_�D�J�X��#*��	위9x��_�f��k�f>d��w�|�~��7Y֧|l/=���<��FO���ŧwmg5�ɞ/��}L�̲��7���]l

tests/complexnetwork.git/objects/10/0d726fb9b103486c1b312a6d47b6046ad9835a

@@ -0,0 +1 @@
+x
+)JMU047a040031QH*J�K�0d�U���k���{?-zW'\R�E�CD�7˯��j�Ntޑ�[��1E�1a%����&��1쩸����W*����I)���r
#Fl

tests/complexnetwork.git/objects/25/21f9e788e9c1b8618263e494491dff64599616

@@ -0,0 +1 @@
+x
K��OR02c���,�L�QH����,�2,.�1
��p

tests/complexnetwork.git/objects/2d/a1c6f425e30cb49e7108e60c26ab7fd9288af1

@@ -0,0 +1 @@
+x
��MJ1�]���B�JU~Aĵ x�J�2��==�V<�
��|����}ۖ�b.w�P�9� �
�� ���넠Z!v���17z=l���G��T�p�(;��L�'��O�BŖri�UJ5�p��zjɗ9b���þ�עþ�ʲ���}����|�xYdߞ�D>g��0��S�����Wg�u�U.���t2?w�Z-

tests/complexnetwork.git/objects/35/b0b2d786ccd201db37b83ca8f21f7d363499af

@@ -0,0 +1,2 @@
+x
��Aj�0@Ѯu���hƒ%(%�r��hT��qP����g��Ã���-��Gof�B
+:d�����	�&�U�X�JY�]��:ĨT*K,�Y�Ԑ��g�LcF?��g��y-��*��?�����}��dY?u߾��!%NH���d?ܿ����tf@���G�

tests/complexnetwork.git/objects/42/27b26b07fd991512881a4ab630dd75311c10b8

@@ -0,0 +1,3 @@
+x
��A
+1E]��ҴӴ׮�B��:2�H������z�O�e�8�6���fJ�}h�B%�$}�0�K?go�\�Ѡ��űđ��ET1c�IHpL6�˄����V8�g�
+g�Y���{��{�.<�;Y�`�~H݌���Z�i�l��K7��Cnh~�
G�

tests/complexnetwork.git/objects/90/4ada2e4081e45aadaf66c29524da6676c7fa0e

@@ -0,0 +1,2 @@
+x
���
+!@��>�� ���"Z���8ZƘ�c��g����'�Rr�J�]o1r�g��v�sER B@k"�L��0Kl���Jg=�$�0�F7%�)��I�����6~�w��_i���Ə���s���!�r��h��<�%uL������Æ}
ۋGM

tests/complexnetwork.git/objects/91/bc0dca437a3377984c069bf36f4e6c6c10af4b

@@ -0,0 +1 @@
+x
��MJA�]�)��� I'��ׂ��ꪙ�g�i��� ^�Ճ��=xy[�iFz�[)��Ŵ��(".4�1�WW	�F�8Rw�V���W�TQ<�8�^#g�,EG�`�J�Z�y�W�*��B0�<h���A-����~����T|�ly���|=�m�~=/6�Oy[^��YbQ�{��ǩ����w諾x�5_���z�i�7X�~w?0#^8

tests/complexnetwork.git/objects/92/dd5a7ae1d649368f58a4a284c4aecde1e65ba8

@@ -0,0 +1,2 @@
+x
��K
+B!��Ʈ�̃�u}@D�Fm�疡^���m��l��
����D�)/U4ɯHƠ��.A�O\��)8$��qo.����cn3<��|���!�z�(���J��K�٬s:������s�qe:�

tests/complexnetwork.git/objects/b2/93501db7da737bcee19189b16c6c1d805e2961

@@ -0,0 +1,2 @@
+x
��M
+�0@a�9���L����kW^a2Mm�iK�����>��ɖ�\��jI	$I��E�ч��d�F4$l��4��#��KZ+�1t
5�hƈ~�Hl��5�6�iK�U���V�Ɵ9���2o/8?[ؗ���y^N��`oL�]�GMZ�V�dm�/�b�U&�~��G�

tests/complexnetwork.git/objects/c4/6a964ed35b3f794561c039ede9b9c3f34cabd7

@@ -0,0 +1,2 @@
+x
��K
+1EQ�YE�I�����cGn����[�G�����^ށ��e�*��@LL?Ff��s0΋8���I2J2ݠv.�Vp.�9�c�3E�<t�( i�Q��!��(~�q+p��$�<s����-����Xx�Oy[.������6Z�V����_\��k-h��G�

tests/complexnetwork.git/objects/f1/33084691e488fdf3dcd67f0dff4df2f55fb711

@@ -0,0 +1,3 @@
+x
��K
+1]�����̴ �ڕW�|tF�G����\��*^\�y��,�Zr%��w^�XE	m`&GD�R�U�fӒ�
+������x$O,B�4tS�%��A�����g�n:i���l���u�q��@�Z�2}{dD�h;Y���nB�%l~��Fj

tests/complexnetwork.git/objects/f4/608d27b543d97506d490187b094dc89f5e1131

@@ -0,0 +1 @@
+x
+)JMU040a040031QH*J�K�0d�U���k���{?-zW'\R�E�1CD�7˯��j�Ntޑ�[��1�$7��$���W�ߚ���'ٚ����5B�
g�*

tests/complexnetwork.git/objects/f7/f0860affbf5445ef0d8b5388cd9814a2063d57

@@ -0,0 +1,2 @@
+x
��M
+�0�a�9E�������v���T#�)i�o�����PrNM�]�̒��#z?+�pd��1����3b���&Qi�ݴ�kC}��@:���7�bA�v/U^蝸�+-R����a]>�[��B�'9�q�Հr�@)ѵ�l��+��_l�]H�

tests/complexnetwork.git/refs/heads/branch1

@@ -0,0 +1 @@
+f133084691e488fdf3dcd67f0dff4df2f55fb711

tests/complexnetwork.git/refs/heads/branch2

@@ -0,0 +1 @@
+fb6e7570645235ebe94889f95783bc0fd1f5bec6

tests/complexnetwork.git/refs/heads/branch3

@@ -0,0 +1 @@
+35b0b2d786ccd201db37b83ca8f21f7d363499af

tests/complexnetwork.git/refs/heads/branch4

@@ -0,0 +1 @@
+0d58300606997e8eacf8504197c65982d0cbb8eb

tests/complexnetwork.git/refs/heads/branch5

@@ -0,0 +1 @@
+fe613ab22f00f45c5041697089c756fda875ac8e

tests/complexnetwork.git/refs/heads/branch6

@@ -0,0 +1 @@
+74071922c169dd14300d31d67f2dfa3316c5dd09

tests/complexnetwork.git/refs/heads/branch7

@@ -0,0 +1 @@
+b20afdee9454ae329646b7b568b08c6065c90217

tests/complexnetwork.git/refs/heads/master

@@ -0,0 +1 @@
+f7f0860affbf5445ef0d8b5388cd9814a2063d57