fix: parse DEVC csv header names with a real CSV parser

FDS only quotes a device name in the _devc.csv header when it contains a
comma or space. The previous parser split on ',"', which found no split
points at all when none of the names needed quoting (e.g. plain identifiers
like TC_Door_Single), causing the entire header line to be treated as a
single device name and raising StopIteration when looking it up. Found while
verifying compatibility against FDS 6.11.1; not version-specific, it can
happen with any FDS version whenever no device name needs quoting.

Author: m-fehr

fdsreader/simulation.py

@@ -1,3 +1,4 @@
+import csv
 import glob
 import logging
 import os
@@ -1023,7 +1024,10 @@ def _register_device(self, smv_file: TextIO) -> Tuple[str, Device]:
     def _load_DEVC_data(self):
         with open(self.devc_path) as infile:
             units = infile.readline().split(",")
-            names = [name.replace('"', "").replace("\n", "").strip() for name in infile.readline().split(',"')]
+            # Device names are only quoted by FDS when they contain a comma or space, so a plain
+            # split on "," or ',"' would misparse a header where none (or all) of the names need
+            # quoting. Use a real CSV parser instead so it doesn't matter which fields are quoted.
+            names = [name.strip() for name in next(csv.reader([infile.readline()]))]
             values = np.genfromtxt(infile, delimiter=",", dtype=np.float32, autostrip=True)
             for k in range(len(names)):
                 if isinstance(self.devices[names[k]], list):

tests/test_devc_csv_parsing.py

@@ -0,0 +1,49 @@
+"""Regression test for the DEVC csv header parsing: FDS only quotes device names that
+contain a comma or space. When none of the names in a given run need quoting (which
+happens to be common, e.g. plain identifiers like "TC_Door_Single"), the previous
+ad-hoc `split(',"')` parser found no quote-comma boundaries at all and treated the
+entire header line as a single device name, raising StopIteration."""
+
+import numpy as np
+
+from fdsreader.devc import Device, DeviceCollection
+from fdsreader.simulation import Simulation
+from fdsreader.utils import Quantity
+
+
+def _make_sim(tmp_path, header_line: str, device_ids):
+    csv_path = tmp_path / "test_devc.csv"
+    units = ",".join(["s"] + ["C"] * len(device_ids))
+    csv_path.write_text(f"{units}\n{header_line}\n0.0,1.0,2.0\n1.0,1.5,2.5\n")
+
+    devices = {
+        device_id: Device(device_id, Quantity("TEMPERATURE", "TEMP", ""), (0.0, 0.0, 0.0), (0.0, 0.0, 0.0))
+        for device_id in device_ids
+    }
+    devices["Time"] = Device("Time", Quantity("TIME", "TIME", "s"), (0.0, 0.0, 0.0), (0.0, 0.0, 0.0))
+
+    sim = object.__new__(Simulation)
+    sim.devc_path = str(csv_path)
+    sim._devices = devices
+    sim.devices = DeviceCollection(devices.values())
+    return sim
+
+
+def test_devc_csv_with_no_quoted_names(tmp_path):
+    # None of these names need quoting (no commas/spaces), so FDS writes them unquoted.
+    sim = _make_sim(tmp_path, "Time,TC_Door_Single,BP_Door_Single", ["TC_Door_Single", "BP_Door_Single"])
+    sim._load_DEVC_data()
+
+    assert np.array_equal(sim._devices["Time"]._data, [0.0, 1.0])
+    assert np.array_equal(sim._devices["TC_Door_Single"]._data, [1.0, 1.5])
+    assert np.array_equal(sim._devices["BP_Door_Single"]._data, [2.0, 2.5])
+
+
+def test_devc_csv_with_mixed_quoted_and_unquoted_names(tmp_path):
+    # "HGL Temp" needs quoting (space), "BP_Door_Single" doesn't.
+    sim = _make_sim(tmp_path, 'Time,"HGL Temp",BP_Door_Single', ["HGL Temp", "BP_Door_Single"])
+    sim._load_DEVC_data()
+
+    assert np.array_equal(sim._devices["Time"]._data, [0.0, 1.0])
+    assert np.array_equal(sim._devices["HGL Temp"]._data, [1.0, 1.5])
+    assert np.array_equal(sim._devices["BP_Door_Single"]._data, [2.0, 2.5])