keyfinder: Better output parsing (#1248)

We were being sloppy about bytes output from the process. Also, it seems like
the tools outputs the path also, so it's necessary to break on whitespace to
actually get the key name.

beetsplug/keyfinder.py

@@ -58,11 +58,18 @@ def find_key(self, items, write=False):
                 continue
 
             try:
-                key = util.command_output([bin, '-f', item.path])
+                output = util.command_output([bin, '-f', item.path])
             except (subprocess.CalledProcessError, OSError) as exc:
                 self._log.error(u'execution failed: {0}', exc)
                 continue
 
+            key_raw = output.rsplit(None, 1)[-1]
+            try:
+                key = key_raw.decode('utf8')
+            except UnicodeDecodeError:
+                self._log.error(u'output is invalid UTF-8')
+                continue
+
             item['initial_key'] = key
             self._log.info(u'added computed initial key {0} for {1}',
                            key, util.displayable_path(item.path))

docs/changelog.rst

@@ -39,6 +39,8 @@ Fixes:
   company since 2013, so we are assuming access will not be restored.
 * Incremental imports now (once again) show a "skipped N directories" message.
 * :doc:`/plugins/embedart`: Handle errors in ImageMagick's output. :bug:`1241`
+* :doc:`/plugins/keyfinder`: Parse the underlying tool's output more robustly.
+  :bug:`1248`
 
 For developers: