feat: support filtering on incrementable values (#178)

Document that 'regex' may be either bytes or text, and add an explicit test for that.

Add 'ExactValueFilter' shortcut, wrapping 'ValueRegexFilter', but also convertin integer values
to the equivalent packed 8-octet bytes.

Allow integer values for 'ValueRangeFilter', converting them to the equivalent packed 8-octet bytes values.

Closes #177.

Author: tseaver

packages/google-cloud-bigtable/google/cloud/bigtable/row_filters.py

@@ -14,11 +14,15 @@
 
 """Filters for Google Cloud Bigtable Row classes."""
 
+import struct
+
 
 from google.cloud._helpers import _microseconds_from_datetime
 from google.cloud._helpers import _to_bytes
 from google.cloud.bigtable_v2.proto import data_pb2 as data_v2_pb2
 
+_PACK_I64 = struct.Struct(">q").pack
+
 
 class RowFilter(object):
     """Basic filter to apply to cells in a row.
@@ -115,7 +119,9 @@ class _RegexFilter(RowFilter):
     .. _RE2 reference: https://github.com/google/re2/wiki/Syntax
 
     :type regex: bytes or str
-    :param regex: A regular expression (RE2) for some row filter.
+    :param regex:
+        A regular expression (RE2) for some row filter.  String values
+        will be encoded as ASCII.
     """
 
     def __init__(self, regex):
@@ -439,9 +445,9 @@ class ValueRegexFilter(_RegexFilter):
         character will not match the new line character ``\\n``, which may be
         present in a binary value.
 
-    :type regex: bytes
+    :type regex: bytes or str
     :param regex: A regular expression (RE2) to match cells with values that
-                  match this regex.
+                  match this regex.  String values will be encoded as ASCII.
     """
 
     def to_pb(self):
@@ -453,6 +459,22 @@ def to_pb(self):
         return data_v2_pb2.RowFilter(value_regex_filter=self.regex)
 
 
+class ExactValueFilter(ValueRegexFilter):
+    """Row filter for an exact value.
+
+
+    :type value: bytes or str or int
+    :param value:
+        a literal string encodable as ASCII, or the
+        equivalent bytes, or an integer (which will be packed into 8-bytes).
+    """
+
+    def __init__(self, value):
+        if isinstance(value, int):
+            value = _PACK_I64(value)
+        super(ExactValueFilter, self).__init__(value)
+
+
 class ValueRangeFilter(RowFilter):
     """A range of values to restrict to in a row filter.
 
@@ -496,6 +518,8 @@ def __init__(
             raise ValueError(
                 "Inclusive start was specified but no " "start value was given."
             )
+        if isinstance(start_value, int):
+            start_value = _PACK_I64(start_value)
         self.start_value = start_value
         self.inclusive_start = inclusive_start
 
@@ -505,6 +529,8 @@ def __init__(
             raise ValueError(
                 "Inclusive end was specified but no " "end value was given."
             )
+        if isinstance(end_value, int):
+            end_value = _PACK_I64(end_value)
         self.end_value = end_value
         self.inclusive_end = inclusive_end
 

packages/google-cloud-bigtable/tests/unit/test_row_filters.py

@@ -498,9 +498,53 @@ def _get_target_class():
     def _make_one(self, *args, **kwargs):
         return self._get_target_class()(*args, **kwargs)
 
-    def test_to_pb(self):
-        regex = b"value-regex"
-        row_filter = self._make_one(regex)
+    def test_to_pb_w_bytes(self):
+        value = regex = b"value-regex"
+        row_filter = self._make_one(value)
+        pb_val = row_filter.to_pb()
+        expected_pb = _RowFilterPB(value_regex_filter=regex)
+        self.assertEqual(pb_val, expected_pb)
+
+    def test_to_pb_w_str(self):
+        value = u"value-regex"
+        regex = value.encode("ascii")
+        row_filter = self._make_one(value)
+        pb_val = row_filter.to_pb()
+        expected_pb = _RowFilterPB(value_regex_filter=regex)
+        self.assertEqual(pb_val, expected_pb)
+
+
+class TestExactValueFilter(unittest.TestCase):
+    @staticmethod
+    def _get_target_class():
+        from google.cloud.bigtable.row_filters import ExactValueFilter
+
+        return ExactValueFilter
+
+    def _make_one(self, *args, **kwargs):
+        return self._get_target_class()(*args, **kwargs)
+
+    def test_to_pb_w_bytes(self):
+        value = regex = b"value-regex"
+        row_filter = self._make_one(value)
+        pb_val = row_filter.to_pb()
+        expected_pb = _RowFilterPB(value_regex_filter=regex)
+        self.assertEqual(pb_val, expected_pb)
+
+    def test_to_pb_w_str(self):
+        value = u"value-regex"
+        regex = value.encode("ascii")
+        row_filter = self._make_one(value)
+        pb_val = row_filter.to_pb()
+        expected_pb = _RowFilterPB(value_regex_filter=regex)
+        self.assertEqual(pb_val, expected_pb)
+
+    def test_to_pb_w_int(self):
+        import struct
+
+        value = 1
+        regex = struct.Struct(">q").pack(value)
+        row_filter = self._make_one(value)
         pb_val = row_filter.to_pb()
         expected_pb = _RowFilterPB(value_regex_filter=regex)
         self.assertEqual(pb_val, expected_pb)
@@ -518,6 +562,7 @@ def _make_one(self, *args, **kwargs):
 
     def test_constructor_defaults(self):
         row_filter = self._make_one()
+
         self.assertIsNone(row_filter.start_value)
         self.assertIsNone(row_filter.end_value)
         self.assertTrue(row_filter.inclusive_start)
@@ -528,22 +573,42 @@ def test_constructor_explicit(self):
         end_value = object()
         inclusive_start = object()
         inclusive_end = object()
+
         row_filter = self._make_one(
             start_value=start_value,
             end_value=end_value,
             inclusive_start=inclusive_start,
             inclusive_end=inclusive_end,
         )
+
         self.assertIs(row_filter.start_value, start_value)
         self.assertIs(row_filter.end_value, end_value)
         self.assertIs(row_filter.inclusive_start, inclusive_start)
         self.assertIs(row_filter.inclusive_end, inclusive_end)
 
+    def test_constructor_w_int_values(self):
+        import struct
+
+        start_value = 1
+        end_value = 10
+
+        row_filter = self._make_one(start_value=start_value, end_value=end_value)
+
+        expected_start_value = struct.Struct(">q").pack(start_value)
+        expected_end_value = struct.Struct(">q").pack(end_value)
+
+        self.assertEqual(row_filter.start_value, expected_start_value)
+        self.assertEqual(row_filter.end_value, expected_end_value)
+        self.assertTrue(row_filter.inclusive_start)
+        self.assertTrue(row_filter.inclusive_end)
+
     def test_constructor_bad_start(self):
-        self.assertRaises(ValueError, self._make_one, inclusive_start=True)
+        with self.assertRaises(ValueError):
+            self._make_one(inclusive_start=True)
 
     def test_constructor_bad_end(self):
-        self.assertRaises(ValueError, self._make_one, inclusive_end=True)
+        with self.assertRaises(ValueError):
+            self._make_one(inclusive_end=True)
 
     def test___eq__(self):
         start_value = object()