revert left sidebar naming scheme and add a hover mouse feature

the naming scheme was causing a lot of trouble with dict handling

app.py

@@ -111,6 +111,10 @@ st.markdown(
     "details > summary { font-size: 1rem; font-weight: 500; }"
     ".stTabs [data-baseweb='tab'] { font-size: 1rem; }"
     ".stFileUploader label { font-size: 1rem; }"
+    /* Multiselect: show full label on hover when text is truncated */
+    "[data-baseweb='tag'] span { cursor: default; }"
+    "[data-baseweb='menu'] li { white-space: nowrap; overflow: hidden; text-overflow: ellipsis; max-width: 260px; }"
+    "[data-baseweb='menu'] li:hover { overflow: visible; white-space: normal; z-index: 9999; }"
     "</style>",
     unsafe_allow_html=True,
 )
@@ -312,47 +316,23 @@ try:
         for sp in st.session_state.results[fn][0].labels()
     ]
     # Map display label -> raw token so ui.py can translate back after selection.
-    # Display label uses a truncated filename (first 5 chars + "...") to keep
-    # the dropdown readable when filenames are long.  The raw token always uses
-    # the full filename so the data model is never affected by this truncation.
-    # If two speakers in the same file share a display name, the raw speaker ID
-    # is appended to disambiguate both entries.
-    def _short_fn(fname):
-        """Return first 5 chars of filename stem + '...' if longer than 5."""
-        stem = fname.rsplit(".", 1)[0]   # strip extension
-        return (stem[:5] + "...") if len(stem) > 5 else stem
-
-    # Build token_display_map in two passes to guarantee globally unique display keys.
-    # Pass 1: collect all (display_label -> raw_token) pairs across every file.
-    # Pass 2: any display label that appears more than once (collision between
-    #         files that share the same short prefix AND speaker name) gets the
-    #         raw speaker ID appended to disambiguate.
-    raw_pairs = []   # [(display_label, raw_token), ...]
+    # If two speakers in the same file share a display name, append the raw ID
+    # to disambiguate both entries.
+    token_display_map = {}
     for fn in st.session_state.file_names:
         if not (fn in st.session_state.results and len(st.session_state.results[fn]) == 2):
             continue
-        short = _short_fn(fn)
-        for sp in st.session_state.results[fn][0].labels():
-            raw   = f"{fn}: {sp}"
-            disp  = f"{short}: {get_display_name(sp, fn)}"
-            raw_pairs.append((disp, raw))
-
-    # Count occurrences of each display label globally
-    disp_counts = {}
-    for disp, _ in raw_pairs:
-        disp_counts[disp] = disp_counts.get(disp, 0) + 1
-
-    token_display_map = {}
-    for disp, raw in raw_pairs:
-        fn, sp = raw.split(": ", 1)
-        short  = _short_fn(fn)
-        if disp_counts[disp] > 1:
-            # Disambiguate with raw speaker ID
-            unique_disp = f"{short}: {get_display_name(sp, fn)} ({sp})"
-        else:
-            unique_disp = disp
-        token_display_map[unique_disp] = raw
-
+        sp_labels = {sp: f"{fn}: {get_display_name(sp, fn)}"
+                     for sp in st.session_state.results[fn][0].labels()}
+        label_counts = {}
+        for disp in sp_labels.values():
+            label_counts[disp] = label_counts.get(disp, 0) + 1
+        for sp, disp in sp_labels.items():
+            raw = f"{fn}: {sp}"
+            if label_counts[disp] > 1:
+                token_display_map[f"{fn}: {get_display_name(sp, fn)} ({sp})"] = raw
+            else:
+                token_display_map[disp] = raw
     display_speaker_tokens = list(token_display_map.keys())
 
     # -----------------------------------------------------------------------