Translate UTF-8 in command output more robustly

nilmrun/server.py

@@ -60,12 +60,16 @@ class AppProcess(object):
         self.manager = manager
 
     def process_status(self, pid):
+        # We need to convert the log (which is bytes) to Unicode
+        # characters, in order to send it via JSON.  Treat it as UTF-8
+        # but replace invalid characters with markers.
+        log = self.manager[pid].log.decode('utf-8', errors='replace')
         return {
             "pid": pid,
             "alive": self.manager[pid].alive,
             "exitcode": self.manager[pid].exitcode,
             "start_time": self.manager[pid].start_time,
-            "log": self.manager[pid].log,
+            "log": log
             }
 
     # /process/status

tests/test_nilmrun.py

@@ -372,3 +372,21 @@ class TestClient(object):
         # kill all processes
         for pid in client.get("process/list"):
             client.post("process/remove", { "pid": pid })
+
+    def test_client_10_unicode(self):
+        client = HTTPClient(baseurl = testurl, post_json = True)
+        eq_(client.get("process/list"), [])
+        def verify(cmd, result):
+            pid = client.post("run/command", { "argv": [ "sh", "-c", cmd ] })
+            eq_(client.get("process/list"), [pid])
+            status = self.wait_end(client, pid)
+            eq_(result, status["log"])
+
+        # Unicode should work
+        verify("echo -n hello", "hello")
+        verify(u"echo -n ☠", u"☠")
+        verify("echo -ne \\\\xe2\\\\x98\\\\xa0", u"☠")
+
+        # Programs that spit out invalid UTF-8 should get replacement
+        # markers
+        verify("echo -ne \\\\xae", u"\ufffd")