include subdir & multiple files tests

3 files changed, 104 insertions, 1 deletions

diff --git a/helpers.py b/helpers.py
index b85f233..8630e86 100644
--- a/helpers.py
+++ b/helpers.py
@@ -48,6 +48,7 @@ def invert_dict(d):
         keys.append(k)
     return result
 
+
 def invert_dictset(d):
     """Invert a dictionary with keys matching a set of values, turned into lists."""
     # Based on recipe from ASPN
@@ -57,3 +58,37 @@ def invert_dictset(d):
             keys = result.setdefault(v, [])
             keys.append(k)
     return result
+
+
+def common_path(l1, l2, common=[]):
+    """Find the common bit of 2 paths."""
+    # From http://code.activestate.com/recipes/208993/
+    if len(l1) < 1: return (common, l1, l2)
+    if len(l2) < 1: return (common, l1, l2)
+    if l1[0] != l2[0]: return (common, l1, l2)
+    return common_path(l1[1:], l2[1:], common+[l1[0]])
+
+
+def common_directory(paths):
+    """Find the deepest common directory of a list of paths.
+    
+    :return: if no paths are provided, None is returned,
+      otherwise a directory with a trailing /.
+    """
+    from bzrlib import osutils
+    def get_dir_with_slash(path):
+        if path.endswith('/'):
+            return path
+        else:
+            dirname, basename = osutils.split(path)
+            return dirname + '/'
+
+    if not paths:
+        return None
+    elif len(paths) == 1:
+        return get_dir_with_slash(paths[0])
+    else:
+        common = common_path(paths[0], paths[1])[0]
+        for path in paths[2:]:
+            common = common_path(common, path)
+        return get_dir_with_slash(''.join(common))
diff --git a/processors/filter_processor.py b/processors/filter_processor.py
index d37585a..6d07088 100644
--- a/processors/filter_processor.py
+++ b/processors/filter_processor.py
@@ -23,6 +23,7 @@ from bzrlib.trace import (
     )
 from bzrlib.plugins.fastimport import (
     commands,
+    helpers,
     processor,
     )
 
@@ -50,7 +51,7 @@ class FilterProcessor(processor.ImportProcessor):
         self.includes = self.params.get('include_paths')
         self.excludes = self.params.get('exclude_paths')
         # What's the new root, if any
-        self.new_root = self._find_new_root(self.includes)
+        self.new_root = helpers.common_directory(self.includes)
         # Buffer of blobs until we know we need them: mark -> cmd
         self.blobs = {}
         # These are the commits we've output so far
diff --git a/tests/test_filter_processor.py b/tests/test_filter_processor.py
index 0d0576e..fd7db59 100644
--- a/tests/test_filter_processor.py
+++ b/tests/test_filter_processor.py
@@ -197,3 +197,70 @@ ing
 from :100
 M 644 :3 README.txt
 """)
+
+    def test_subdir(self):
+        params = {'include_paths': ['doc/']}
+        self.assertFiltering(_SAMPLE_WITH_DIR, params, \
+"""blob
+mark :1
+data 9
+Welcome!
+commit refs/heads/master
+mark :100
+committer a <b@c> 1234798653 +0000
+data 4
+test
+M 644 :1 README.txt
+blob
+mark :3
+data 19
+Welcome!
+my friend
+blob
+mark :4
+data 11
+== Docs ==
+commit refs/heads/master
+mark :102
+committer d <b@c> 1234798653 +0000
+data 8
+test
+ing
+from :100
+M 644 :3 README.txt
+M 644 :4 index.txt
+""")
+
+    def test_multiple_files_in_subdir(self):
+        # The new root should be the subdrectory
+        params = {'include_paths': ['doc/README.txt', 'doc/index.txt']}
+        self.assertFiltering(_SAMPLE_WITH_DIR, params, \
+"""blob
+mark :1
+data 9
+Welcome!
+commit refs/heads/master
+mark :100
+committer a <b@c> 1234798653 +0000
+data 4
+test
+M 644 :1 README.txt
+blob
+mark :3
+data 19
+Welcome!
+my friend
+blob
+mark :4
+data 11
+== Docs ==
+commit refs/heads/master
+mark :102
+committer d <b@c> 1234798653 +0000
+data 8
+test
+ing
+from :100
+M 644 :3 README.txt
+M 644 :4 index.txt
+""")