Added submodule type and handling through Tree listing.

3 files changed, 145 insertions, 7 deletions

diff --git a/lib/git/commit.py b/lib/git/commit.py
index b074d3a8..9a9a77ad 100644
--- a/lib/git/commit.py
+++ b/lib/git/commit.py
@@ -71,7 +71,7 @@ class Commit(LazyMixin):
             if parents is not None:
                 self.parents = [Commit(repo, p) for p in parents]
             if tree is not None:
-                self.tree = Tree(repo, id=tree)
+                self.tree = Tree(repo, id=tree, commit_context = self.id)
 
     def __bake__(self):
         """
diff --git a/lib/git/submodule.py b/lib/git/submodule.py
new file mode 100755
index 00000000..c6c506f4
--- /dev/null
+++ b/lib/git/submodule.py
@@ -0,0 +1,103 @@
+# head.py
+# Copyright (C) 2008-2010 Michael Trier (mtrier@gmail.com) and contributors
+#
+# This module is part of GitPython and is released under
+# the BSD License: http://www.opensource.org/licenses/bsd-license.php
+
+import re
+
+class Submodule(object):
+    """
+    A Submodule is a named reference to a Commit on another Repo.
+    Every Submodule instance contains a name and local and remote paths.
+
+    Submodules are very close in behavior to HEAD pointer. It just sits on
+    top of a structure, in this case, at the end of folders tree, and says
+    something about that ending.
+
+    Examples::
+        >>> repo = Repo("/path/to/repo")
+        >>> s = repo.commit('master').tree['lib']['mysubmodule_folder']
+        >>> s.name
+        'mysubmodule_folder'
+        >>> s.path
+        '/lib/mysubmodule_folder'
+        >>> s.url
+        'http://example.com/path/to/repo.git'
+        >>> s.id
+        "1c09f116cbc2cb4100fb6935bb162daa4723f455"
+    """
+
+    def __init__(self, repo=None, id=None, mode=None, name='',
+                 commit_context=None, path=''):
+        """
+        Initialize a newly instanced Submodule
+
+        'repo'
+            Pointer to Repo object instance.
+        'id'
+            Is the Sha of the commit on a remote server. This object does NOT
+            (usually) exist on this, current repo.
+        'mode'
+            A black hole at this time. Trying to keep the input args
+            similar between Tree, Blob and Subprocess instantiation classes.
+        'name'
+            This is just the last segment in the submodule's full local path.
+            It's the name of the actual folder to which a submodule is tied.
+        'mode'
+            A black hole at this time. Trying to keep the input args
+            similar between Tree, Blob and Subprocess instantiation classes.
+        'commit_context'
+            A string with ID of the commit that was the root for the tree
+            structure that lead us to this folder (Tree object) that contains
+            this submodule reference.
+            See comments in Tree object code for background.
+        'path'
+            This is the "longer" version of "name" argument. It includes all
+            the parent folders we passed on the way from root of the commit to
+            this point in the folder tree.
+            Submodules in the .gitmodules are referenced by their full path
+            and contents of this argument are used to retrieve the URI of the
+            remote repo tied to this full local path.
+            Example: "/lib/vendor/vendors_repoA"
+        """
+        self.repo = repo
+        self.id = id
+        self.path = path
+        self.name = name
+        self._commit_context = commit_context
+        self._cached_URI = None
+
+    def getURI(self, commit_context = None):
+        '''Returns the remote repo URI for the submodule.
+        
+        This data is NOT stored in the blob or anywhere close. It's in a 
+        .gitmodules file in the root Tree of SOME commit.
+        
+        We need to know what commit to look into to look for .gitmodules.
+        
+        We try to retain the "origin" commit ID within the object when we
+        traverse the Tree chain if it started with a particular commit.
+        
+        When this does not work, or if you want to override the behavior,
+        pass the string with commit's ID to the commit_context argument.
+        '''
+        if not self._cached_URI and ( commit_context or self._commit_context ):
+            _b = self.repo.commit(commit_context or self._commit_context).tree.get('.gitmodules')
+            if _b:
+                _m = re.findall(
+                    r'\[submodule "[^\t]+?\s+path\s*=\s*([^\t]+)\s+url\s*=\s*([^\t]+)'
+                    ,'\t'.join(_b.data.splitlines())
+                )
+                for _e in _m:
+                    if _e[0] == self.path.strip('/'):
+                        self._cached_URI = _e[1].strip().strip('"').strip("'")
+                        break
+        return self._cached_URI
+
+    @property
+    def url(self):
+        return self.getURI()
+
+    def __repr__(self):
+        return '<git.Submodule "%s">' % self.id
diff --git a/lib/git/tree.py b/lib/git/tree.py
index 02266dce..45f5fdcb 100644
--- a/lib/git/tree.py
+++ b/lib/git/tree.py
@@ -7,14 +7,47 @@
 import os
 from lazy import LazyMixin
 import blob
+import submodule
 
 class Tree(LazyMixin):
-    def __init__(self, repo, id, mode=None, name=None):
+    def __init__(self, repo, id, mode=None, name=None, commit_context = None, path = ''):
         LazyMixin.__init__(self)
         self.repo = repo
         self.id = id
         self.mode = mode
         self.name = name
+        # commit_context (A string with ID of the commit) is a "crutch" that
+        # allows us to look up details for submodules, should we find any in
+        # this particular tree.
+        # Trees don't have a reference to parent (commit, other tree).
+        # They can have infinite amounts of parents.
+        # However, we need to know what commit got us to this particular
+        # tree if we want to know the URI of the submodule.
+        # The commit ID of the repo pointed out by submodule is here, in the tree.
+        # However, the only way to know what URI that submodule refers to is
+        # to read .gitmodules file that's in the top-most tree of SOME commit.
+        # Each commit can have a different version of .gitmodule, but through 
+        # tree chain lead to the same Tree instance where the submodule is rooted.
+        # 
+        # There is a short-cut. If submodule is placed in top-most Tree in a 
+        # commit (i.e. submodule's path value is "mysubmodule") the .gitmodules
+        # file will be in the same exact tree. YEY! we just read that and know
+        # the submodule's URI. Shortcut is gone when submodule is nested in the
+        # commit like so: "commonfolder/otherfolder/mysubmodule" In this case,
+        # commit's root tree will have "Tree 'commonfolder'" which will have 
+        # "Tree "otherfolder", which will have "Submodule 'mysubmodule'"
+        # By the time we get to "Tree 'otherfolder'" we don't know where to
+        # look for ".gitmodules". This is what commit_context is for.
+        # The only way you get a value here if you either set it by hand, or
+        # traverse the Tree chain that started with CommitInstance.tree, which
+        # populates the context upon Tree instantiation.
+        self.commit_context = commit_context
+        # path is the friend commit_context. since trees don't have links to
+        # parents, we have no clue what the "full local path" of a child
+        # submodule would be. Submodules are listed as "name" in trees and
+        # as "folder/folder/name" in .gitmodules. path helps us keep up with the
+        # the folder changes.
+        self.path = path
         self._contents = None
 
     def __bake__(self):
@@ -26,12 +59,12 @@ class Tree(LazyMixin):
         # Read the tree contents.
         self._contents = {}
         for line in self.repo.git.ls_tree(self.id).splitlines():
-            obj = self.content_from_string(self.repo, line)
+            obj = self.content_from_string(self.repo, line, commit_context = self.commit_context, path = self.path)
             if obj is not None:
                 self._contents[obj.name] = obj
 
     @staticmethod
-    def content_from_string(repo, text):
+    def content_from_string(repo, text, commit_context = None, path=''):
         """
         Parse a content item and create the appropriate object
 
@@ -50,11 +83,13 @@ class Tree(LazyMixin):
             return None
 
         if typ == "tree":
-            return Tree(repo, id=id, mode=mode, name=name)
+            return Tree(repo, id=id, mode=mode, name=name,
+                        commit_context = commit_context, path='/'.join([path,name]))
         elif typ == "blob":
             return blob.Blob(repo, id=id, mode=mode, name=name)
-        elif typ == "commit":
-            return None
+        elif typ == "commit" and mode == '160000':
+            return submodule.Submodule(repo, id=id, name=name,
+                        commit_context = commit_context, path='/'.join([path,name]))
         else:
           raise(TypeError, "Invalid type: %s" % typ)