store chunked binary sample data in uV

processor/reader.py

@@ -7,6 +7,16 @@
 
 log = logging.getLogger()
 
+# Conversion factors to microvolts (uV)
+UNIT_TO_UV = {
+    "volts": 1e6,
+    "v": 1e6,
+    "millivolts": 1e3,
+    "mv": 1e3,
+    "microvolts": 1,
+    "uv": 1,
+}
+
 
 class NWBElectricalSeriesReader:
     """
@@ -223,7 +233,7 @@ def get_chunk(self, start=None, end=None):
             end: End sample index (default: num_samples)
 
         Returns:
-            list of numpy arrays, one per channel, with scaling applied
+            list of numpy arrays, one per channel, with scaling applied in uV
         """
         # Single HDF5 read for all channels
         all_data = self.electrical_series.data[start:end, :]
@@ -239,5 +249,21 @@ def get_chunk(self, start=None, end=None):
         else:
             scaled_data = all_data * base_scale + offset
 
+        # Convert volts to microvolts (uV)
+        # This processor fixes the TimeSeriesChannel unit to uV.
+        # NWB conversion factor outputs values in the unit specified by electrical_series.unit (fixed to 'volts'; see docstring below)
+        # From: https://github.com/NeurodataWithoutBorders/nwb-schema/blob/d65d42257003543c569ea7ac0cd6d7aee01c88d6/core/nwb.ecephys.yaml#L35-L42
+        #     - name: unit
+        #       dtype: text
+        #       value: volts
+        #       doc: Base unit of measurement for working with the data. This value is fixed to
+        #         'volts'. Actual stored values are not necessarily stored in these units. To
+        #         access the data in these units, multiply 'data' by 'conversion', followed by
+        #         'channel_conversion' (if present), and then add 'offset'.
+        unit = getattr(self.electrical_series, "unit", "volts").lower()
+        if unit not in UNIT_TO_UV:
+            raise ValueError(f"Unknown unit '{unit}' - expected one of: {list(UNIT_TO_UV.keys())}")
+        scaled_data = scaled_data * UNIT_TO_UV[unit]
+
         # Split into list of per-channel arrays
         return [scaled_data[:, i] for i in range(self.num_channels)]

processor/timeseries_channel.py

@@ -1,4 +1,7 @@
 class TimeSeriesChannel:
+    # Unit is always microvolts - data is converted to uV during processing
+    UNIT = "uV"
+
     def __init__(
         self,
         index,
@@ -7,7 +10,6 @@ def __init__(
         start,
         end,
         type="CONTINUOUS",
-        unit="uV",
         group="default",
         last_annotation=0,
         properties=None,
@@ -27,7 +29,6 @@ def __init__(
         self.start = int(start)
         self.end = int(end)
 
-        self.unit = unit.strip()
         self.type = type.upper()
         self.group = group.strip()
         self.last_annotation = last_annotation
@@ -38,7 +39,7 @@ def as_dict(self):
             "name": self.name,
             "start": self.start,
             "end": self.end,
-            "unit": self.unit,
+            "unit": self.UNIT,
             "rate": self.rate,
             "type": self.type,
             "group": self.group,
@@ -53,11 +54,11 @@ def as_dict(self):
 
     @staticmethod
     def from_dict(channel, properties=None):
+        # Note: channel["unit"] is ignored - unit is always uV
         return TimeSeriesChannel(
             name=channel["name"],
             start=int(channel["start"]),
             end=int(channel["end"]),
-            unit=channel["unit"],
             rate=channel["rate"],
             type=channel.get("channelType", channel.get("type")),
             group=channel["group"],

processor/writer.py

@@ -89,8 +89,8 @@ def write_chunk(chunk, start_time, end_time, channel_index, output_dir):
 
         file_name = "channel-{}_{}_{}{}".format(
             "{index:05d}".format(index=channel_index),
-            int(start_time * 1e6),
-            int(end_time * 1e6),
+            round(start_time * 1e6),
+            round(end_time * 1e6),
             TIME_SERIES_BINARY_FILE_EXTENSION,
         )
         file_path = os.path.join(output_dir, file_name)

tests/conftest.py

@@ -94,6 +94,7 @@ def mock_electrical_series(sample_electrical_series_data, sample_timestamps):
     series.conversion = 1.0
     series.offset = 0.0
     series.channel_conversion = None
+    series.unit = "volts"
 
     # Mock electrodes table
     mock_electrodes = []

tests/test_reader.py

@@ -15,6 +15,7 @@ def create_mock_electrical_series(
     offset=0.0,
     channel_conversion=None,
     group_names=None,
+    unit="microvolts",
 ):
     """Helper to create mock ElectricalSeries objects."""
     series = Mock()
@@ -29,6 +30,7 @@ def create_mock_electrical_series(
     series.conversion = conversion
     series.offset = offset
     series.channel_conversion = channel_conversion
+    series.unit = unit
 
     # Create mock electrodes as a Mock object that can be iterated and has table attribute
     mock_electrode_list = []
@@ -307,3 +309,29 @@ def test_all_scaling_factors_combined(self):
         # Result: data * conversion * channel_conversion + offset
         np.testing.assert_array_equal(chunks[0], np.ones(10) * 7.0)  # 1 * 2 * 3 + 1 = 7
         np.testing.assert_array_equal(chunks[1], np.ones(10) * 9.0)  # 1 * 2 * 4 + 1 = 9
+
+    def test_volts_to_microvolts_conversion(self):
+        """Test that data in volts is converted to microvolts."""
+        series = create_mock_electrical_series(10, 2, rate=1000.0, unit="volts")
+        series.data = np.ones((10, 2)) * 1e-6  # 1 microvolt in volts
+        session_start = datetime(2023, 1, 1, 12, 0, 0)
+
+        reader = NWBElectricalSeriesReader(series, session_start)
+        chunks = reader.get_chunk()
+
+        # 1e-6 V * 1e6 = 1 uV
+        for chunk in chunks:
+            np.testing.assert_array_almost_equal(chunk, np.ones(10) * 1.0)
+
+    def test_millivolts_to_microvolts_conversion(self):
+        """Test that data in millivolts is converted to microvolts."""
+        series = create_mock_electrical_series(10, 2, rate=1000.0, unit="millivolts")
+        series.data = np.ones((10, 2)) * 0.001  # 1 microvolt in millivolts
+        session_start = datetime(2023, 1, 1, 12, 0, 0)
+
+        reader = NWBElectricalSeriesReader(series, session_start)
+        chunks = reader.get_chunk()
+
+        # 0.001 mV * 1e3 = 1 uV
+        for chunk in chunks:
+            np.testing.assert_array_almost_equal(chunk, np.ones(10) * 1.0)

tests/test_timeseries_channel.py

@@ -15,7 +15,7 @@ def test_basic_initialization(self):
         assert channel.start == 1000000
         assert channel.end == 2000000
         assert channel.type == "CONTINUOUS"
-        assert channel.unit == "uV"
+        assert TimeSeriesChannel.UNIT == "uV"
         assert channel.group == "default"
         assert channel.last_annotation == 0
         assert channel.properties == []
@@ -30,7 +30,6 @@ def test_initialization_with_all_parameters(self):
             start=500000,
             end=1500000,
             type="UNIT",
-            unit="  mV  ",
             group="  electrode_group  ",
             last_annotation=100,
             properties=[{"key": "value"}],
@@ -43,7 +42,7 @@ def test_initialization_with_all_parameters(self):
         assert channel.start == 500000
         assert channel.end == 1500000
         assert channel.type == "UNIT"  # should be uppercased
-        assert channel.unit == "mV"  # should be stripped
+        assert TimeSeriesChannel.UNIT == "uV"  # unit is always uV
         assert channel.group == "electrode_group"  # should be stripped
         assert channel.last_annotation == 100
         assert channel.properties == [{"key": "value"}]
@@ -124,7 +123,7 @@ def test_from_dict_with_type_key(self, sample_channel_dict):
         assert channel.name == "Channel 1"
         assert channel.start == 1000000
         assert channel.end == 2000000
-        assert channel.unit == "uV"
+        assert TimeSeriesChannel.UNIT == "uV"  # unit is always uV
         assert channel.rate == 30000.0
         assert channel.type == "CONTINUOUS"
         assert channel.group == "default"
@@ -281,7 +280,6 @@ def test_as_dict_from_dict_round_trip(self):
             start=500000,
             end=1500000,
             type="UNIT",
-            unit="mV",
             group="test_group",
             last_annotation=50,
             properties=[{"key": "value"}],
@@ -296,7 +294,7 @@ def test_as_dict_from_dict_round_trip(self):
         assert restored.start == original.start
         assert restored.end == original.end
         assert restored.type == original.type
-        assert restored.unit == original.unit
+        assert serialized["unit"] == "uV"  # unit is always uV in serialized output
         assert restored.group == original.group
         assert restored.last_annotation == original.last_annotation
         assert restored.properties == original.properties

tests/test_timeseries_client.py

@@ -76,7 +76,7 @@ def test_create_channel_sends_correct_body(self, mock_session_manager):
 
         client = TimeSeriesClient("https://api.test.com", mock_session_manager)
         channel = TimeSeriesChannel(
-            index=0, name="Ch1", rate=1000.0, start=0, end=1000, type="UNIT", unit="mV", group="test_group"
+            index=0, name="Ch1", rate=1000.0, start=0, end=1000, type="UNIT", group="test_group"
         )
 
         client.create_channel("pkg-123", channel)
@@ -86,7 +86,7 @@ def test_create_channel_sends_correct_body(self, mock_session_manager):
         assert body["rate"] == 1000.0
         assert body["channelType"] == "UNIT"  # 'type' should be renamed to 'channelType'
         assert "type" not in body  # Original 'type' key should be removed
-        assert body["unit"] == "mV"
+        assert body["unit"] == "uV"  # unit is always uV
         assert body["group"] == "test_group"
 
     @responses.activate

tests/test_writer.py

@@ -34,9 +34,9 @@ def test_write_chunk_creates_file(self, temp_output_dir):
 
         TimeSeriesChunkWriter.write_chunk(chunk, start_time, end_time, 0, temp_output_dir)
 
-        # Check file was created
+        # Check file was created (use round() to match writer behavior)
         expected_filename = (
-            f"channel-00000_{int(start_time * 1e6)}_{int(end_time * 1e6)}{TIME_SERIES_BINARY_FILE_EXTENSION}"
+            f"channel-00000_{round(start_time * 1e6)}_{round(end_time * 1e6)}{TIME_SERIES_BINARY_FILE_EXTENSION}"
         )
         file_path = os.path.join(temp_output_dir, expected_filename)
         assert os.path.exists(file_path)
@@ -116,7 +116,7 @@ def test_write_channel_creates_metadata_file(self, temp_output_dir, session_star
         writer = TimeSeriesChunkWriter(session_start_time, temp_output_dir, 1000)
 
         channel = TimeSeriesChannel(
-            index=5, name="Test Channel", rate=30000.0, start=1000000, end=2000000, unit="mV", group="electrode_group"
+            index=5, name="Test Channel", rate=30000.0, start=1000000, end=2000000, group="electrode_group"
         )
 
         writer.write_channel(channel)
@@ -135,7 +135,6 @@ def test_write_channel_json_content(self, temp_output_dir, session_start_time):
             rate=30000.0,
             start=1000000,
             end=2000000,
-            unit="mV",
             type="CONTINUOUS",
             group="test_group",
             last_annotation=100,
@@ -153,7 +152,7 @@ def test_write_channel_json_content(self, temp_output_dir, session_start_time):
         assert data["rate"] == 30000.0
         assert data["start"] == 1000000
         assert data["end"] == 2000000
-        assert data["unit"] == "mV"
+        assert data["unit"] == "uV"  # unit is always uV
         assert data["type"] == "CONTINUOUS"
         assert data["group"] == "test_group"
         assert data["lastAnnotation"] == 100