Add options to ignore symbolic links; ignore both kinds by default

3 files changed, 30 insertions, 2 deletions

diff --git a/README b/README
index c2a12a1..463826e 100644
--- a/README
+++ b/README
@@ -102,6 +102,12 @@ Recommendations
 Since encoding eats a lot of CPU, you should start it at a low priority.
 The simplest way is to run ``nice -n19 brutha`` instead of only ``brutha``.
 
+Symbolic links
+~~~~~~~~~~~~~~
+brutha ignores symbolic links by default, but can follow links outside of the
+source path, or inside the source path, using options ``--outside`` and ``--inside``.
+
+This is modelled after mpd's ``follow_outside_symlinks`` and ``follow_inside_symlinks`` options.
 
 Changes
 -------
diff --git a/brutha/__main__.py b/brutha/__main__.py
index a279c53..b839a59 100644
--- a/brutha/__main__.py
+++ b/brutha/__main__.py
@@ -40,6 +40,10 @@ def main():
                         help="Execute the script instead of printing it")
     parser.add_argument('-j', '--jobs', type=int, default=cores,
                         help="Number of concurrent jobs")
+    parser.add_argument('-O', '--outside', action='store_true',
+                        help="Follow directory symbolic links outside of the source directory")
+    parser.add_argument('-I', '--inside', action='store_true',
+                        help="Follow directory symbolic links inside of the source directory")
     group = parser.add_mutually_exclusive_group()
     group.add_argument('-l', '--hardlink', action='store_true', help="Use hardlinks instead of a copy")
     group.add_argument('-r', '--reflink', action='store_true', help="Use reflinks instead of a copy")
@@ -52,7 +56,8 @@ def main():
                 {'quality': args.quality, 'gain': args.gain, 'delete': args.delete,
                  'maxrate': args.maxrate, 'maxbits': args.maxbits,
                  'lossycheck': args.lossycheck,
-                 'hardlink': args.hardlink, 'reflink': args.reflink},
+                 'hardlink': args.hardlink, 'reflink': args.reflink,
+                 'inside': args.inside, 'outside': args.outside},
                 log)
     if args.execute:
         stream = StringIO()
diff --git a/brutha/tree.py b/brutha/tree.py
index e6968e8..db4e4fc 100644
--- a/brutha/tree.py
+++ b/brutha/tree.py
@@ -15,11 +15,26 @@ class Tree(object):
         self.destpath = destpath
         self.options = {'quality': 8, 'gain': False, 'delete': False,
                         'maxrate': None, 'maxbits': None, 'lossycheck': True,
-                        'hardlink': False, 'reflink': False}
+                        'hardlink': False, 'reflink': False,
+                        'outside': False, 'inside': False}
         if options:
             self.options.update(options)
         self.log = log
 
+    def allowed(self, root, dirname):
+        path = os.path.join(root, dirname)
+        if not os.path.islink(path):
+            return True
+
+        dest = os.path.join(root, os.readlink(path))
+        outside = os.path.relpath(dest, self.path).startswith(os.path.pardir + os.path.sep)
+        if outside and self.options['outside']:
+            return True
+        elif not outside and self.options['inside']:
+            return True
+        else:
+            return False
+
     def commands(self):
         commands = []
         wanted = []
@@ -27,6 +42,8 @@ class Tree(object):
         if self.log:
             print >>self.log, "Walking source directory..."
         for root, dirs, files in os.walk(self.path, followlinks=True):
+            for dirname in [dirname for dirname in dirs if not self.allowed(root, dirname)]:
+                dirs.remove(dirname)
             num += 1
             if not num % 200:
                 self.progress(num)