fixups for score scaling

Author: nickzoic

countess/core/parameters.py

@@ -165,8 +165,11 @@ class NumericParam(ScalarWithOperatorsParam):
 
     def set_value(self, value):
         try:
+            if type(value) is str:
+                if m := re.match(r'(\d+(\.\d*)?)', value):
+                    value = m.group(1)
             self._value = self.var_type(value)
-        except (TypeError, ValueError):
+        except (TypeError, ValueError) as exc:
             self.reset_value()
 
     # Operator methods which apply only to numerics

countess/gui/config.py

@@ -402,8 +402,6 @@ def value_changed_callback(self, *_):
                 self.var.set(self.set_value(self.var.get()))
         else:
             val = self.set_value(self.var.get())
-            if val is not None:
-                self.var.set(val)
 
     def widget_modified_callback(self, *_):
         # only gets called the *first* time a modification happens, unless

countess/plugins/score_scale.py

@@ -22,7 +22,7 @@
 
 class ScaleClassParam(TabularMultiParam):
     col = ColumnChoiceParam("Column")
-    op = ChoiceParam("Operation", "Equals", ["Equals", "Starts With", "Ends With", "Contains"])
+    op = ChoiceParam("Operation", "Equals", ["Equals", "Starts With", "Ends With", "Contains", "Matches"])
     st = StringParam("Value")
     score = FloatParam("Scaled Score")
 
@@ -37,6 +37,8 @@ def filter(self):
             return f"ends_with({col},{val})"
         elif self.op.value == "Contains":
             return f"contains({col},{val})"
+        elif self.op.value == "Matches":
+            return f"regexp_full_match({col},{val})"
         else:
             raise NotImplementedError()
 
@@ -57,19 +59,23 @@ def execute(
         score_col_id = duckdb_escape_identifier(self.score_col.value)
         scaled_col_id = duckdb_escape_identifier(self.scaled_col.value)
 
-        all_columns = ",".join("T0." + duckdb_escape_identifier(c) for c in source.columns if c != self.scaled_col.value)
+        all_columns = ",".join("T0." + duckdb_escape_identifier(c) for c in source.columns if self.scaled_col != c)
 
         if self.group_col.is_not_none():
             group_col_id = "T0." + duckdb_escape_identifier(self.group_col.value)
         else:
             group_col_id = "1"  # dummy value for one big group.
 
+        c0, c1 = self.classifiers
+        s0 = duckdb_escape_literal(c0.score.value)
+        s1 = duckdb_escape_literal(c1.score.value)
+
         sql = f"""
-            select {all_columns}, ({score_col_id} - T1.y) / (T1.z - T1.y) as {scaled_col_id}
+            select {all_columns}, ({s1} - {s0}) * ({score_col_id} - T1.y) / (T1.z - T1.y) + {s0} as {scaled_col_id}
             from {source.alias} T0 join (
                 select {group_col_id} as x,
-                    median({score_col_id}) filter ({self.classifiers[0].filter()}) as y,
-                    median({score_col_id}) filter ({self.classifiers[1].filter()}) as z
+                    median({score_col_id}) filter ({c0.filter()}) as y,
+                    median({score_col_id}) filter ({c1.filter()}) as z
                 from {source.alias} T0
                 group by x
             ) T1 on ({group_col_id} = T1.x)