Fix various string/bytes issues with Python 3

nilmdb/client/numpyclient.py

@@ -70,14 +70,14 @@ class NumpyClient(nilmdb.client.client.Client):
 
             # See if we have enough to make the requested Numpy array
             while total_len >= maxsize:
-                assembled = "".join(chunks)
+                assembled = b"".join(chunks)
                 total_len -= maxsize
                 chunks = [ assembled[maxsize:] ]
                 block = assembled[:maxsize]
                 yield to_numpy(block)
 
         if total_len:
-            yield to_numpy("".join(chunks))
+            yield to_numpy(b"".join(chunks))
 
     @contextlib.contextmanager
     def stream_insert_numpy_context(self, path, start = None, end = None,

nilmdb/server/bulkdata.py

@@ -276,7 +276,7 @@ class Table(object):
     @classmethod
     def valid_path(cls, root):
         """Return True if a root path is a valid name"""
-        return "_format" not in root.split(b"/")
+        return b"_format" not in root.split(b"/")
 
     @classmethod
     def exists(cls, root):
@@ -538,7 +538,10 @@ class Table(object):
                 ret.append(f.extract_string(offset, count))
             remaining -= count
             row += count
-        return "".join(ret)
+        if binary:
+            return b"".join(ret)
+        else:
+            return "".join(ret)
 
     def __getitem__(self, row):
         """Extract timestamps from a row, with table[n] notation."""

nilmdb/server/nilmdb.py

@@ -638,7 +638,7 @@ class NilmDB(object):
 
         if count:
             return matched
-        return ("".join(result), restart)
+        return (b"".join(result), restart)
 
     def stream_remove(self, path, start = None, end = None):
         """

nilmdb/server/rocket.c

@@ -279,10 +279,7 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 	union64_t t64;
 	int i;
 
-	/* It would be nice to use 't#' instead of 's' for data,
-	   but we need the null termination for strto*.  If we had
-	   strnto* that took a length, we could use t# and not require
-	   a copy. */
+        /* Input data is Unicode */
 	if (!PyArg_ParseTuple(args, "isiiLLL:append_string", &count,
 			      &data, &offset, &linenum,
 			      &ll1, &ll2, &ll3))
@@ -443,7 +440,7 @@ static PyObject *Rocket_append_binary(Rocket *self, PyObject *args)
 	timestamp_t end;
 	timestamp_t last_timestamp;
 
-	if (!PyArg_ParseTuple(args, "it#iiLLL:append_binary",
+	if (!PyArg_ParseTuple(args, "iy#iiLLL:append_binary",
                               &count, &data, &data_len, &offset,
                               &linenum, &ll1, &ll2, &ll3))
 		return NULL;
@@ -493,7 +490,7 @@ static PyObject *Rocket_append_binary(Rocket *self, PyObject *args)
 }
 
 /****
- * Extract to string
+ * Extract to a Unicode string
  */
 
 static PyObject *Rocket_extract_string(Rocket *self, PyObject *args)
@@ -601,7 +598,7 @@ err:
 }
 
 /****
- * Extract to binary string containing raw little-endian binary data
+ * Extract to binary bytes object containing raw little-endian binary data
  */
 static PyObject *Rocket_extract_binary(Rocket *self, PyObject *args)
 {

nilmdb/utils/atomic.py

@@ -7,7 +7,7 @@ def replace_file(filename, content):
     given contents.  This is intended to be 'pretty good on most
     OSes', but not necessarily bulletproof."""
 
-    newfilename = filename + ".new"
+    newfilename = filename + b".new"
 
     # Write to new file, flush it
     with open(newfilename, "wb") as f:

tests/test_client.py

@@ -333,7 +333,7 @@ class TestClient(object):
         with assert_raises(ClientError) as e:
             list(client.stream_extract("/newton/prep",
                                        count = True, binary = True))
-        data = "".join(client.stream_extract("/newton/prep", binary = True))
+        data = b"".join(client.stream_extract("/newton/prep", binary = True))
         # Quick check using struct
         unpacker = struct.Struct("<qffffffff")
         out = []