Merge pull request #1404 from PeterNSteinmetz/factorHeaderHandling

NeuralynxIO: Factor out header handling.

neo/rawio/neuralynxrawio/ncssections.py

@@ -51,19 +51,13 @@ class NcsSection:
 
     _RECORD_SIZE = 512  # nb sample per signal record
 
-    def __init__(self):
-        self.startRec = -1  # index of starting record
-        self.startTime = -1  # starttime of first record
-        self.endRec = -1  # index of last record (inclusive)
-        self.endTime = -1  # end time of last record, that is, the end time of the last
-        # sampling period contained in the last record of the section
-
     def __init__(self, sb, st, eb, et, ns):
-        self.startRec = sb
-        self.startTime = st
-        self.endRec = eb
-        self.endTime = et
-        self.n_samples = ns
+        self.startRec = sb # index of starting record
+        self.startTime = st # starttime of first record
+        self.endRec = eb # index of last record (inclusive)
+        self.endTime = et # end time of last record, that is, the end time of the last
+                          # sampling period contained in the last record of the section
+        self.n_samples = ns # number of samples in record which are valid
 
     def __eq__(self, other):
         return (

neo/rawio/neuralynxrawio/nlxheader.py

@@ -124,18 +124,46 @@ def _to_bool(txt):
         ),
     }
 
-    def __init__(self, filename):
+    def __init__(self, filename, props_only=False):
         """
         Factory function to build NlxHeader for a given file.
+
+        :param filename: name of Neuralynx file
+        :param props_only: if true, will not try and read time and date or check start
         """
         super(OrderedDict, self).__init__()
         with open(filename, "rb") as f:
             txt_header = f.read(NlxHeader.HEADER_SIZE)
         txt_header = txt_header.strip(b"\x00").decode("latin-1")
 
         # must start with 8 # characters
-        assert txt_header.startswith("########"), "Neuralynx files must start with 8 # characters."
+        if not props_only and not txt_header.startswith("########"):
+           ValueError("Neuralynx files must start with 8 # characters.")
+
+        self.read_properties(filename, txt_header)
+
+        if not props_only:
+            self.readTimeDate(txt_header)
+
+    @staticmethod
+    def build_with_properties_only(filename):
+        """
+        Builds a version of the header without time and date or other validity checking.
+
+        Intended mostly for utilities but may also be useful for some recalcitrant header formats.
 
+        :param filename: name of Neuralynx file.
+        :return: NlxHeader with properties from header text
+        """
+        res = OrderedDict()
+
+
+    def read_properties(self, filename, txt_header):
+        """
+        Read properties from header and place in OrderedDictionary which this object is.
+        :param filename: name of ncs file, used for extracting channel number
+        :param txt_header: header text
+        """
         # find keys
         for k1, k2, type_ in NlxHeader.txt_header_keys:
             pattern = r"-(?P<name>" + k1 + r")\s+(?P<value>[\S ]*)"
@@ -149,17 +177,14 @@ def __init__(self, filename):
                 if type_ is not None:
                     value = type_(value)
                 self[name] = value
-
         # if channel_ids or s not in self then the filename is used
         name = os.path.splitext(os.path.basename(filename))[0]
-
         # convert channel ids
         if "channel_ids" in self:
             chid_entries = re.findall(r"\S+", self["channel_ids"])
             self["channel_ids"] = [int(c) for c in chid_entries]
         else:
             self["channel_ids"] = ["unknown"]
-
         # convert channel names
         if "channel_names" in self:
             name_entries = re.findall(r"\S+", self["channel_names"])
@@ -170,7 +195,6 @@ def __init__(self, filename):
             ), "Number of channel ids does not match channel names."
         else:
             self["channel_names"] = ["unknown"] * len(self["channel_ids"])
-
         # version and application name
         # older Cheetah versions with CheetahRev property
         if "CheetahRev" in self:
@@ -192,11 +216,9 @@ def __init__(self, filename):
         else:
             self["ApplicationName"] = "Neuraview"
             app_version = "2"
-
         if " Development" in app_version:
             app_version = app_version.replace(" Development", ".dev0")
         self["ApplicationVersion"] = Version(app_version)
-
         # convert bit_to_microvolt
         if "bit_to_microVolt" in self:
             btm_entries = re.findall(r"\S+", self["bit_to_microVolt"])
@@ -206,7 +228,6 @@ def __init__(self, filename):
             assert len(self["bit_to_microVolt"]) == len(
                 self["channel_ids"]
             ), "Number of channel ids does not match bit_to_microVolt conversion factors."
-
         if "InputRange" in self:
             ir_entries = re.findall(r"\w+", self["InputRange"])
             if len(ir_entries) == 1:
@@ -217,9 +238,13 @@ def __init__(self, filename):
                 chid_entries
             ), "Number of channel ids does not match input range values."
 
-        # Format of datetime depends on app name, app version
-        # :TODO: this works for current examples but is not likely actually related
-        # to app version in this manner.
+    def readTimeDate(self, txt_header):
+        """
+        Read time and date from text of header appropriate for app name and version
+
+        :TODO: this works for current examples but is not likely actually related
+        to app version in this manner.
+        """
         an = self["ApplicationName"]
         if an == "Cheetah":
             av = self["ApplicationVersion"]
@@ -245,7 +270,6 @@ def __init__(self, filename):
             an = "Unknown"
             av = "NA"
             hpd = NlxHeader.header_pattern_dicts["def"]
-
         # opening time
         sr = re.search(hpd["datetime1_regex"], txt_header)
         if not sr:
@@ -257,7 +281,6 @@ def __init__(self, filename):
             self["recording_opened"] = datetime.datetime.strptime(
                 dt1["date"] + " " + dt1["time"], hpd["datetimeformat"]
             )
-
         # close time, if available
         if "datetime2_regex" in hpd:
             sr = re.search(hpd["datetime2_regex"], txt_header)

neo/test/rawiotest/test_neuralynxrawio.py

@@ -27,7 +27,7 @@ class TestNeuralynxRawIO(
         "neuralynx/Cheetah_v5.5.1/original_data",
         "neuralynx/Cheetah_v5.6.3/original_data",
         "neuralynx/Cheetah_v5.7.4/original_data",
-        "neuralynx/Cheetah_v6.3.2/incomplete_blocks",
+        "neuralynx/Cheetah_v6.3.2/incomplete_blocks"
     ]
 
     def test_scan_ncs_files(self):
@@ -356,5 +356,19 @@ def test_equality(self):
         self.assertNotEqual(ns0, ns1)
 
 
+# I comment this now and will put it back when files will be in gin.g-node
+# class TestNlxHeader(TestNeuralynxRawIO, unittest.TestCase):
+#     def test_no_date_time(self):
+#         filename = self.get_local_path("neuralynx/NoDateHeader/NoDateHeader.nev")
+
+#         with self.assertRaises(IOError):
+#             hdr = NlxHeader(filename)
+
+#         hdr = NlxHeader(filename, props_only=True)
+
+#         self.assertEqual(len(hdr), 11)
+#         self.assertEqual(hdr['ApplicationName'], 'Pegasus')
+#         self.assertEqual(hdr['FileType'], 'Event')
+
 if __name__ == "__main__":
     unittest.main()