Support custom truncated Thies LNM format

actris-cloudnet · Apr 18, 2024 · eb94fb7 · eb94fb7
1 parent 446acbd
commit eb94fb7
Show file tree

Hide file tree

Showing 3 changed files with 42 additions and 6 deletions.
diff --git a/cloudnetpy/instruments/disdrometer/thies.py b/cloudnetpy/instruments/disdrometer/thies.py
@@ -194,9 +194,13 @@ def _append_data(self) -> None:
 
     def _read_line(self, line: str, timestamp: datetime.datetime | None = None):
         raw_values = line.split(";")
-        if len(raw_values) != 521:
+        # Support custom truncated format used in Leipzig LIM.
+        expected_columns = self.site_meta.get("truncate_columns", 521)
+        if len(raw_values) != expected_columns:
             return
         for i, key in TELEGRAM4:
+            if i >= expected_columns - 1:
+                break
             value: Any
             if key == "_date":
                 value = _parse_date(raw_values[i])
@@ -227,11 +231,12 @@ def _read_line(self, line: str, timestamp: datetime.datetime | None = None):
             else:
                 value = int(raw_values[i])
             self.raw_data[key].append(value)
-        self.raw_data["spectrum"].append(
-            np.array(list(map(int, raw_values[79:-2])), dtype="i2").reshape(
-                self.n_diameter, self.n_velocity
+        if expected_columns > 79:
+            self.raw_data["spectrum"].append(
+                np.array(list(map(int, raw_values[79:-2])), dtype="i2").reshape(
+                    self.n_diameter, self.n_velocity
+                )
             )
-        )
         if timestamp is not None:
             self.raw_data["time"].append(timestamp)
         else:
@@ -266,7 +271,9 @@ def _create_diameter_vectors(self) -> None:
 
 def _parse_date(date: str) -> datetime.date:
     day, month, year = map(int, date.split("."))
-    return datetime.date(2000 + year, month, day)
+    if year < 100:
+        year += 2000
+    return datetime.date(year, month, day)
 
 
 def _parse_time(time: str) -> datetime.time:

diff --git a/tests/unit/data/thies-lnm/2024041723-leipzig-lim.txt b/tests/unit/data/thies-lnm/2024041723-leipzig-lim.txt
@@ -0,0 +1,3 @@
+00;0747;2.50;17.04.2024;23:00:50;00;00;NP   ;000.000;00;00;NP   ;000.000;000.000;000.000;0087.76;99999;-9.9;100;0.0;0;0;
+00;0747;2.50;17.04.2024;23:01:50;00;00;NP   ;000.000;00;00;NP   ;000.000;000.000;000.000;0087.76;99999;-9.9;100;0.0;0;0;
+00;0747;2.50;17.04.2024;23:02:50;00;00;NP   ;000.000;00;00;NP   ;000.000;000.000;000.000;0087.76;99999;-9.9;100;0.0;0;0;
diff --git a/tests/unit/test_disdrometer.py b/tests/unit/test_disdrometer.py
@@ -306,6 +306,32 @@ def test_processing(self):
         )
 
 
+class TestThies3(Check):
+    date = "2024-04-17"
+    temp_dir = TemporaryDirectory()
+    temp_path = temp_dir.name + "/test.nc"
+    filename = f"{SCRIPT_PATH}/data/thies-lnm/2024041723-leipzig-lim.txt"
+    site_meta = SITE_META
+    uuid = disdrometer.thies2nc(filename, temp_path, {**site_meta, "truncate_columns": 23}, date=date)
+
+    def test_processing(self):
+        assert self.nc.title == f'LNM disdrometer from {self.site_meta["name"]}'
+        assert self.nc.year == "2024"
+        assert self.nc.month == "04"
+        assert self.nc.day == "17"
+        assert self.nc.location == "Kumpula"
+        assert self.nc.cloudnet_file_type == "disdrometer"
+        assert self.nc.serial_number == "0747"
+        assert np.allclose(
+            self.nc["time"][:],
+            [
+                timedelta(hours=23, minutes=0, seconds=50) / timedelta(hours=1),
+                timedelta(hours=23, minutes=1, seconds=50) / timedelta(hours=1),
+                timedelta(hours=23, minutes=2, seconds=50) / timedelta(hours=1),
+            ],
+        )
+
+
 class TestInvalidCharacters(Check):
     temp_dir = TemporaryDirectory()
     temp_path = temp_dir.name + "/test.nc"