Refactor SchismOut2DUIManager and SchismOutputUIDataManager to remove identity_key_columns and update catalog initialization

Author: dwr-psandhu

AGENTS.md

@@ -67,11 +67,30 @@ All `viz` sub-commands support YAML config with CLI overrides:
 
 ## Pitfalls
 
-- When extending with mixed catalogs (raw + math references), guard against NaN in filename columns — see `dvue/AGENTS.md` for the safe pattern.
+- When extending with mixed catalogs (raw + math references), guard against NaN in the
+  **`source` column** (not `filename`) — see `dvue/AGENTS.md` for the safe pattern.
+  `SchismDataUIManager.get_data_reference()` already does this correctly.
 - HoloViews has issues with dots in dataset dimension names; see workaround in [calibplot.py](schismviz/calibplot.py).
 - `schout_reader.py` is pre-xarray; avoid extending it — use xarray-based paths instead.
 - Spatial operations assume UTM Zone 10 N (EPSG:32610); do not add CRS auto-detection without testing.
 
+## SchismDataUIManager — Catalog Pattern
+
+`SchismDataUIManager` uses **dvue Pattern A** (live `DataCatalog`) with:
+
+```python
+catalog = DataCatalog(primary_key=["source_num", "id", "variable"], crs=crs)
+```
+
+- Multi-source by design — each SCHISM study output directory is a separate `source`.
+- The observation datastore uses `source="datastore"`.
+- `source_num` is auto-assigned in the order studies are passed to the constructor.
+- Refs are added with an **explicit `name`** (`f"{source}::{id}/{variable}"`) to preserve
+  human-readable study-path identity; auto-derivation is skipped when `name` is set.
+- `_build_dvue_catalog()` is called after `super().__init__()` once the reader is ready.
+- Do **not** pass `url_column`, `url_num_column`, or `identity_key_columns` to `super().__init__()` —
+  these parameters no longer exist in dvue.
+
 ## Docs & References
 
 - [README.md](README.md) — overview and quick-start

schismviz/out2dui.py

@@ -123,10 +123,6 @@ class SchismOut2DUIManager(TimeSeriesDataUIManager):
     """
 
     study_name = param.String(default="out2d", doc="Label for this study")
-    identity_key_columns = param.List(
-        default=["variable", "node_name"],
-        doc="Columns that identify a unique variable/node pair; used for Transform and Source Compare naming.",
-    )
     show_source_compare = param.Boolean(
         default=True,
         doc="Show the Source Compare action in the Add to Catalog menu.",
@@ -155,7 +151,7 @@ def __init__(
         # calls get_data_catalog() and get_time_range() during setup.
         self._dfcat = self._build_catalog()
 
-        super().__init__(url_column="filename", **kwargs)
+        super().__init__(**kwargs)
 
         # Set visual styling param defaults *after* super().__init__()
         self.study_name = study_name

schismviz/schismui.py

@@ -129,10 +129,6 @@ def create_title(self, v):
 class SchismOutputUIDataManager(TimeSeriesDataUIManager):
 
     convert_units = param.Boolean(default=True, doc="Convert units to SI")
-    identity_key_columns = param.List(
-        default=["id", "variable"],
-        doc="Columns that identify a unique station/variable pair; used for Transform and Source Compare naming.",
-    )
     show_source_compare = param.Boolean(
         default=True,
         doc="Show the Source Compare action in the Add to Catalog menu.",
@@ -162,7 +158,7 @@ def __init__(self, *studies, datastore=None, time_range=None, **kwargs):
         # data_catalog property returns None during base-class init (which triggers
         # the correct get_data_catalog() path rather than crashing with AttributeError).
         self._dvue_catalog = None
-        super().__init__(url_column="filename", **kwargs)
+        super().__init__(**kwargs)
         self.color_cycle_column = "id"
         self.dashed_line_cycle_column = "source"
         self.marker_cycle_column = "variable"
@@ -231,7 +227,7 @@ def _make_plot_action(self):
         return SchismTimeSeriesPlotAction()
 
     def _build_dvue_catalog(self, crs=None) -> DataCatalog:
-        catalog = DataCatalog(crs=crs)
+        catalog = DataCatalog(primary_key=["source_num", "id", "variable"], crs=crs)
         for _, row in self.catalog.iterrows():
             ref_name = f"{row['source']}::{row['id']}/{row['variable']}"
             attrs = {k: v for k, v in row.items() if k != "geometry"}