use regex for HCS label parsing

iohub/mm_fov.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import re
 from pathlib import Path
 
 from xarray import DataArray
@@ -109,58 +110,47 @@ def stage_positions(self, value):
     @property
     def hcs_position_labels(self):
         """Parse plate position labels generated by the HCS position generator,
-        e.g. 'A1-Site_0' or '1-Pos000_000', and split into row, column, and
-        FOV names.
+        e.g. 'A1-Site_0', '1-Pos000_000', or 'Pos-9-000_000', and split into
+        row, column, and FOV names.
 
         Returns
         -------
         list[tuple[str, str, str]]
-            FOV name paths, e.g. ('A', '1', '0') or ('0', '1', '000000')
+            FOV name paths, e.g. ('A', '1', '0'), ('0', '1', '000000'), or
+            ('0', '9', '000000').
         """
         if not self.stage_positions:
             raise ValueError("Stage position metadata not available.")
-        try:
-            # Look for "'A1-Site_0', 'H12-Site_1', ... " format
-            labels = [
-                pos["Label"].split("-Site_") for pos in self.stage_positions
-            ]
-            return [(well[0], well[1:], fov) for well, fov in labels]
-        except Exception:
-            try:
-                # Look for "'1-Pos000_000', '2-Pos000_001', ... "
-                # and split into ('1', '000_000'), ...
-                labels = [
-                    pos["Label"].split("-Pos") for pos in self.stage_positions
-                ]
-
-                # remove underscore from FOV name, i.e. '000_000'
-                # collect all wells in row '0' so output is
-                # ('0', '1', '000000')
-                return [
-                    ("0", col, fov.replace("_", "")) for col, fov in labels
-                ]
-            except Exception:
-                try:
-                    # Look for "'Pos-1-000_000', 'Pos-2-000_001', ... "
-                    # and split into ('1', '000_000'), ...
-                    # New format following
-                    # https://github.com/micro-manager/micro-manager/pull/1897
-                    labels = [
-                        pos["Label"].split("Pos-")[1].split("-")
-                        for pos in self.stage_positions
-                    ]
-                    return [
-                        ("0", col, fov.replace("_", "")) for col, fov in labels
-                    ]
-
-                except Exception:
-                    labels = [pos.get("Label") for pos in self.stage_positions]
-                    raise ValueError(
-                        "HCS position labels are in the format of "
-                        "'A1-Site_0', 'H12-Site_1', or '1-Pos000_000', "
-                        f"or 'Pos-1-000_000'. Got labels {labels}"
+
+        # See https://chatgpt.com/share/e/67009155-d960-8008-bd64-aa3117b142b6
+        pattern = re.compile(
+            r"([A-Za-z])?(\d+)-(?:Site_(\d+)|Pos(?:-)?(\d+)[-_](\d+))"
+        )
+        labels = [pos.get("Label") for pos in self.stage_positions]
+        row_col_fov = []
+        for label in labels:
+            if label:
+                match = re.match(pattern, label)
+                if match:
+                    row = match.group(1) if match.group(1) else "0"
+                    col = match.group(2)
+                    fov = (
+                        match.group(3)
+                        if match.group(3)
+                        else f"{match.group(4)}{match.group(5)}"
                     )
 
+                    row_col_fov.append((row, col, fov))
+
+        if not row_col_fov:
+            raise ValueError(
+                "HCS position labels are in the format of "
+                "'A1-Site_0', 'H12-Site_1', or '1-Pos000_000', "
+                f"or 'Pos-1-000_000'. Got labels {labels}"
+            )
+
+        return row_col_fov
+
     @property
     def zyx_scale(self) -> tuple[float, float, float]:
         """ZXY pixel size in micrometers."""