Revert "Allow batch_size=None in Table.scan() to avoid filter incompa…

…tibilities"

This reverts commit 8481d31, since it
is not a correct fix. See issue #56.

NEWS.rst

@@ -18,11 +18,6 @@ Release date: *not yet released*
   Python 2.6 a separate ```ordereddict``` pacakge has to be installed from PyPI.
   (`issue #39 <https://github.com/wbolster/happybase/issues/39>`_)
 
-* Allow `None` as a valid value for the `batch_size` argument to
-  :py:meth:`Table.scan`, since HBase does not support specifying a batch size
-  when some scanner filters are used. (`issue #54
-  <https://github.com/wbolster/happybase/issues/54>`_).
-
 
 HappyBase 0.7
 -------------

happybase/table.py

@@ -254,15 +254,6 @@ def scan(self, row_start=None, row_stop=None, row_prefix=None,
         this to a low value (or even 1) if your data is large, since a low
         batch size results in added round-trips to the server.
 
-        .. warning::
-
-           Not all HBase filters can be used in combination with a batch
-           size. Explicitly specify `None` for the `batch_size` argument
-           in those cases to override the default value. Failure to do
-           so can result in hard to debug errors (not HappyBase's
-           fault), such as a non-responsive connection. The HBase logs
-           may contain more useful information in these situations.
-
         **Compatibility notes:**
 
         * The `filter` argument is only available when using HBase 0.92
@@ -289,11 +280,11 @@ def scan(self, row_start=None, row_stop=None, row_prefix=None,
         :return: generator yielding the rows matching the scan
         :rtype: iterable of `(row_key, row_data)` tuples
         """
-        if batch_size is not None and batch_size < 1:
-            raise ValueError("'batch_size' must be >= 1 (or None)")
+        if batch_size < 1:
+            raise ValueError("'batch_size' must be >= 1")
 
         if limit is not None and limit < 1:
-            raise ValueError("'limit' must be >= 1 (or None)")
+            raise ValueError("'limit' must be >= 1")
 
         if sorted_columns and self.connection.compat < '0.96':
             raise NotImplementedError(
@@ -358,16 +349,16 @@ def scan(self, row_start=None, row_stop=None, row_prefix=None,
         n_returned = n_fetched = 0
         try:
             while True:
-                if batch_size is None:
-                    how_many = 1
-                else:
+                if limit is None:
                     how_many = batch_size
+                else:
+                    how_many = min(batch_size, limit - n_returned)
 
-                if limit is not None:
-                    how_many = min(how_many, limit - n_returned)
-
-                items = self.connection.client.scannerGetList(
-                    scan_id, how_many)
+                if how_many == 1:
+                    items = self.connection.client.scannerGet(scan_id)
+                else:
+                    items = self.connection.client.scannerGetList(
+                        scan_id, how_many)
 
                 n_fetched += len(items)
 

tests/test_api.py

@@ -357,7 +357,7 @@ def test_scan():
         list(table.scan(row_prefix='foobar', row_start='xyz'))
 
     with assert_raises(ValueError):
-        list(table.scan(batch_size=0))
+        list(table.scan(batch_size=None))
 
     if connection.compat == '0.90':
         with assert_raises(NotImplementedError):
@@ -446,13 +446,6 @@ def test_scan_sorting():
         row.items())
 
 
-def test_scan_filter_and_batch_size():
-    # See issue #54
-    filter = "SingleColumnValueFilter ('cf1', 'qual1', =, 'binary:val1')"
-    for k, v in table.scan(filter=filter, batch_size=None):
-        print v
-
-
 def test_delete():
     row_key = 'row-test-delete'
     data = {'cf1:col1': 'v1',