posixpath.realpath() now detects symlink loops and returns the path just before

brettcannon · brettcannon · commit f50299c37877 · 2004-07-10T22:55:15.000Z
the loop starts.

Closes bug #930024.  Thanks AM Kuchling.
diff --git a/Lib/posixpath.py b/Lib/posixpath.py
@@ -405,13 +405,37 @@ def realpath(filename):
     bits = ['/'] + filename.split('/')[1:]
     for i in range(2, len(bits)+1):
         component = join(*bits[0:i])
-        if islink(component):
-            resolved = os.readlink(component)
-            (dir, file) = split(component)
-            resolved = normpath(join(dir, resolved))
-            newpath = join(*([resolved] + bits[i:]))
-            return realpath(newpath)
+        # Resolve symbolic links.
+	if islink(component):
+            resolved = _resolve_link(component)
+            if resolved is None:
+                # Infinite loop -- return original component + rest of the path
+                return join(*([component] + bits[i:]))
+            else:
+                newpath = join(*([resolved] + bits[i:]))
+                return realpath(newpath)
 
     return filename
 
+
+def _resolve_link(path):
+    """Internal helper function.  Takes a path and follows symlinks
+    until we either arrive at something that isn't a symlink, or 
+    encounter a path we've seen before (meaning that there's a loop).
+    """
+    paths_seen = []
+    while islink(path):
+	if path in paths_seen:
+            # Already seen this path, so we must have a symlink loop
+            return None
+	paths_seen.append(path)
+        # Resolve where the link points to
+	resolved = os.readlink(path)
+        if not abspath(resolved):
+            dir = dirname(path)
+            path = normpath(join(dir, resolved))
+        else:
+            path = normpath(resolved)
+    return path
+
 supports_unicode_filenames = False
diff --git a/Misc/NEWS b/Misc/NEWS
@@ -29,6 +29,10 @@ Extension modules
 Library
 -------
 
+- Bug #930024: posixpath.realpath() now handles infinite loops in symlinks by
+  returning the last point in the path that was not part of any loop.  Thanks
+  AM Kuchling.
+
 - Bug #980327: ntpath not handles compressing erroneous slashes between the
   drive letter and the rest of the path.  Also clearly handles UNC addresses now
   as well.  Thanks Paul Moore.
