Add check for TDSm header when reading data (#196)

his should hopefully catch most cases where the tdms_index file doesn't match the tdms data file.

nptdms/reader.py

@@ -103,6 +103,7 @@ def read_raw_data(self):
             raise RuntimeError(
                 "Cannot read data unless metadata has first been read")
         for segment in self._segments:
+            self._verify_segment_start(segment)
             for chunk in segment.read_raw_data(self._file):
                 yield chunk
 
@@ -136,6 +137,7 @@ def read_raw_data_for_channel(self, channel_path, offset=0, length=None):
 
         segment_index = start_segment
         for segment in self._segments[start_segment:end_segment + 1]:
+            self._verify_segment_start(segment)
             # By default, read all chunks in a segment
             chunk_offset = 0
             num_chunks = segment.num_chunks
@@ -197,6 +199,7 @@ def read_channel_chunk_for_index(self, channel_path, index):
         index_in_segment = index - segment_start_index
         chunk_index = index_in_segment // chunk_size
 
+        self._verify_segment_start(segment)
         chunk_data = next(segment.read_raw_data_for_channel(self._file, channel_path, chunk_index, 1))
         chunk_offset = segment_start_index + chunk_index * chunk_size
         return chunk_data, chunk_offset
@@ -272,6 +275,20 @@ def _read_lead_in(self, file, segment_position, is_index_file=False):
         return (segment_position, toc_mask, endianness, data_position, raw_data_offset,
                 next_segment_offset, next_segment_pos)
 
+    def _verify_segment_start(self, segment):
+        """ When reading data for a segment, check for the TDSm tag at the start of the segment in an attempt
+            to detect any mismatch between tdms and tdms_index files.
+        """
+        position = segment.position
+        self._file.seek(segment.position)
+        expected_tag = b'TDSm'
+        tag = self._file.read(4)
+        if tag != expected_tag:
+            raise ValueError(
+                "Attempted to read data segment at position {0} but did not find segment start header. ".format(
+                    position) +
+                "Check that the tdms_index file matches the tdms data file.")
+
     def _get_data_file_size(self):
         current_pos = self._file.tell()
         self._file.seek(0, os.SEEK_END)

nptdms/test/test_tdms_file.py

@@ -4,6 +4,7 @@
 import logging
 import os
 import sys
+from shutil import copyfile
 import tempfile
 from hypothesis import (assume, given, example, settings, strategies)
 import numpy as np
@@ -461,6 +462,65 @@ def test_read_file_passed_as_pathlib_path():
         compare_arrays(channel_obj.data, expected_channel_data)
 
 
+def test_read_with_mismatching_index_file():
+    """ Test that reading data when the index file doesn't match the data file raises an error
+    """
+
+    test_file = GeneratedFile()
+    test_file.add_segment(
+        ("kTocMetaData", "kTocRawData", "kTocNewObjList"),
+        segment_objects_metadata(
+            channel_metadata("/'group'/'channel1'", 3, 2),
+            channel_metadata("/'group'/'channel2'", 3, 2),
+        ),
+        "01 00 00 00" "02 00 00 00"
+        "03 00 00 00" "04 00 00 00"
+    )
+    test_file.add_segment(
+        ("kTocMetaData", "kTocRawData", "kTocNewObjList"),
+        segment_objects_metadata(
+            channel_metadata("/'group'/'channel1'", 3, 2),
+            channel_metadata("/'group'/'channel2'", 3, 2),
+        ),
+        "01 00 00 00" "02 00 00 00"
+        "03 00 00 00" "04 00 00 00"
+    )
+
+    test_file_with_index = GeneratedFile()
+    test_file_with_index.add_segment(
+        ("kTocMetaData", "kTocRawData", "kTocNewObjList"),
+        segment_objects_metadata(
+            channel_metadata("/'group'/'channel1'", 3, 3),
+            channel_metadata("/'group'/'channel2'", 3, 3),
+        ),
+        "01 00 00 00" "02 00 00 00" "03 00 00 00"
+        "04 00 00 00" "05 00 00 00" "06 00 00 00"
+    )
+    test_file_with_index.add_segment(
+        ("kTocMetaData", "kTocRawData", "kTocNewObjList"),
+        segment_objects_metadata(
+            channel_metadata("/'group'/'channel1'", 3, 3),
+            channel_metadata("/'group'/'channel2'", 3, 3),
+        ),
+        "01 00 00 00" "02 00 00 00" "03 00 00 00"
+        "04 00 00 00" "05 00 00 00" "06 00 00 00"
+    )
+
+    with test_file.get_tempfile(delete=False) as tdms_file:
+        with test_file_with_index.get_tempfile_with_index() as tdms_file_with_index_path:
+            # Move index file from second file to match the name of the first file
+            new_index_file = tdms_file.name + '_index'
+            copyfile(tdms_file_with_index_path + '_index', new_index_file)
+            try:
+                tdms_file.file.close()
+                with pytest.raises(ValueError) as exc_info:
+                    _ = TdmsFile.read(tdms_file.name)
+                assert 'Check that the tdms_index file matches the tdms data file' in str(exc_info.value)
+            finally:
+                os.remove(new_index_file)
+                os.remove(tdms_file.name)
+
+
 @pytest.mark.filterwarnings('ignore:.* is deprecated')
 def test_get_objects():
     """Test reading data"""