Add a class for normalizing paths.

1 files changed, 65 insertions, 0 deletions

diff --git a/summainlib.py b/summainlib.py
index e9d6c60..5dfd133 100644
--- a/summainlib.py
+++ b/summainlib.py
@@ -77,6 +77,71 @@ class NumberNormalizer(object):
         self.next_dev = 1
 
 
+class PathNormalizer(object):
+
+    '''Normalize a filesystem path.
+    
+    For every input path, a new output path is given. The same output
+    path is given every time the same input path is given. The output
+    path has the same number of elements as the input path, and if
+    /foo/bar results in /a/b, then /foo/yo results in /a/b, i.e., the
+    tree structure is the same.
+    
+    '''
+    
+    def __init__(self):
+        self._counter = 0
+        self._dict = dict()
+        self._dict[os.sep] = os.sep
+        self._dict['.'] = '.'
+        self._dict['..'] = '..'
+
+    def split(self, path):
+        if path == os.sep:
+            return [os.sep]
+        parts = path.split(os.sep)
+        if parts:
+            if parts[0] == '':
+                parts[0] = os.sep
+            if parts[-1] == '':
+                parts[-1] = os.sep
+        return parts
+        
+    def _base26(self, n):
+        if n == 0:
+            digits = [0]
+        else:
+            digits = []
+            while n > 0:
+                digits.append(n % 26)
+                n /= 26
+
+        letters = 'abcdefghijklmnopqrstuvwxyz'
+        assert len(letters) == 26
+        return ''.join(letters[x] for x in reversed(digits))
+        
+    def normalize_part(self, part):
+        if part not in self._dict:
+            self._dict[part] = self._base26(self._counter)
+            self._counter += 1
+        return self._dict[part]
+        
+    def normalize(self, path):
+        parts = self.split(path)
+        normalized = [self.normalize_part(x) for x in parts]
+        result = ''
+        for x in normalized:
+            if not result:
+                result = x
+            elif x == os.sep:
+                result += x
+            elif result.endswith(os.sep):
+                result += x
+            else:
+                result += os.sep + x
+        return result
+
+
 class FilesystemObject(object):
 
     '''An object in the file system.