Fixing proportion plot logic

Value was exceeding 1 and I think it's calcCategories problem

utils.py

@@ -455,57 +455,80 @@ def build_multifile_category_df(validNames, results, summaries, categories, cate
                                 speakerRenames=None):
     """Build df6 (category breakdown per file) for the multi-file expander.
 
-    speakerRenames: {filename: {raw_sp: display_name}} — used to apply display
-    names to unassigned speaker columns (extraCats).
+    Uses su.sumTimes() per speaker (same as the single-file charts) so that:
+    - Each speaker's time = union of their segments (overlaps within one speaker
+      are merged by get_timeline().duration())
+    - Multiple speakers in the same role are subset-unioned before summing so
+      cross-speaker overlaps within a role are counted only once
+    - Values are proportions (0-1) of the file's total duration
+
+    speakerRenames: {filename: {raw_sp: display_name}} — applied to unassigned
+    speaker column headers.
     """
     speakerRenames = speakerRenames or {}
-    df6_dict    = {"files": validNames}
+    df6_dict      = {"files": validNames}
     allCategories = copy.deepcopy(categories)
 
+    # First pass: discover unassigned speaker columns across all files
     for fn in validNames:
         currAnnotation, _ = results[fn]
         prefix = fn + ": "
-        per_file_selections = [
-            [t[len(prefix):] for t in tokens if t.startswith(prefix)]
+        assigned = {
+            t[len(prefix):]
             for tokens in categorySelect
-        ]
-        try:
-            catSummary, extraCats = su.calcCategories(currAnnotation, per_file_selections)
-        except (UnboundLocalError, Exception):
-            summaries[fn]["categories"] = ([], [])
-            continue
-        # Apply display names to raw speaker IDs in extraCats
+            for t in tokens
+            if t.startswith(prefix)
+        }
         renames = speakerRenames.get(fn, {})
-        extraCats = [renames.get(sp, sp) for sp in extraCats]
-        summaries[fn]["categories"] = (catSummary, extraCats)
-        for extra in extraCats:
-            df6_dict.setdefault(extra, [])
-            if extra not in allCategories:
-                allCategories.append(extra)
+        for sp in currAnnotation.labels():
+            if sp not in assigned:
+                display = renames.get(sp, sp)
+                if display not in allCategories:
+                    allCategories.append(display)
+                df6_dict.setdefault(display, [])
 
     for category in categories:
         df6_dict.setdefault(category, [])
 
+    # Second pass: compute proportions per file
     for fn in validNames:
-        summary, extras = summaries[fn]["categories"]
-        # If this file was skipped in the first loop (calcCategories failed),
-        # summary will be [] — fill every category column with 0 so all
-        # arrays stay the same length.
-        if not summary:
-            for category in allCategories:
-                df6_dict[category].append(0)
-            continue
-        theseCategories = categories + extras
-        annotation_end = max(
-            (s.end for s in results[fn][0].itersegments()),
-            default=results[fn][1]
-        )
-        safe_total = max(annotation_end, results[fn][1], 1)
-        for j, timeSlots in enumerate(summary):
-            val = sum(t.duration for _, t in timeSlots) / safe_total
-            df6_dict[theseCategories[j]].append(min(val, 1.0))
+        currAnnotation, totalSeconds = results[fn]
+        safe_total = max(totalSeconds, 1)
+        prefix = fn + ": "
+        renames = speakerRenames.get(fn, {})
+
+        # For each role: union all assigned speakers into one subset, then sum.
+        # su.sumTimes uses get_timeline(False).duration() which merges overlaps.
+        for i, category in enumerate(categories):
+            assigned_sps = [
+                t[len(prefix):]
+                for t in categorySelect[i]
+                if t.startswith(prefix)
+            ] if i < len(categorySelect) else []
+            # Filter to speakers that actually exist in this annotation
+            valid_sps = [sp for sp in assigned_sps if sp in currAnnotation.labels()]
+            if valid_sps:
+                val = su.sumTimes(currAnnotation.subset(valid_sps)) / safe_total
+            else:
+                val = 0.0
+            df6_dict[category].append(min(val, 1.0))
+
+        # For unassigned speakers: each gets their own column
+        assigned_all = {
+            t[len(prefix):]
+            for tokens in categorySelect
+            for t in tokens
+            if t.startswith(prefix)
+        }
+        unassigned = [sp for sp in currAnnotation.labels() if sp not in assigned_all]
+        for sp in unassigned:
+            display = renames.get(sp, sp)
+            val = su.sumTimes(currAnnotation.subset([sp])) / safe_total
+            df6_dict[display].append(min(val, 1.0))
+
+        # Fill 0 for any extra columns this file doesn't have
         for category in allCategories:
-            if category not in theseCategories:
+            if len(df6_dict[category]) < len(df6_dict["files"]):
                 df6_dict[category].append(0)
 
     return pd.DataFrame(df6_dict), allCategories