Update app_utils.py

app_utils.py

@@ -103,52 +103,88 @@ def wall_of_checkboxes(labels, max_width=10):
                 checkboxes.append(st.empty())
     return checkboxes
 
-def aggregate_genres(genres, legit_genres, verbose=False):
-    genres_output = dict()
+def find_legit_genre(glabel, legit_genres, verbose=False):
     legit_genres_formatted = [lg.replace('-', '').replace(' ', '') for lg in legit_genres]
-    for glabel in genres.keys():
-        if verbose: print('\n', glabel)
-        glabel_formatted = glabel.replace(' ', '').replace('-', '')
-        best_match = None
-        best_match_score = 0
-        for legit_glabel, legit_glabel_formatted in zip(legit_genres, legit_genres_formatted):
-            if 'jazz' in glabel_formatted:
-                best_match = 'jazz'
-                if verbose: print('\t', 'pop')
-                break
-            if 'ukpop' in glabel_formatted:
-                best_match = 'pop'
-                if verbose: print('\t', 'pop')
-                break
-            if legit_glabel_formatted == glabel_formatted:
-                if verbose: print('\t', legit_glabel_formatted)
+    glabel_formatted = glabel.replace(' ', '').replace('-', '')
+    if verbose: print('\n', glabel)
+    best_match = None
+    best_match_score = 0
+    for legit_glabel, legit_glabel_formatted in zip(legit_genres, legit_genres_formatted):
+        if 'jazz' in glabel_formatted:
+            best_match = 'jazz'
+            if verbose: print('\t', 'pop')
+            break
+        if 'ukpop' in glabel_formatted:
+            best_match = 'pop'
+            if verbose: print('\t', 'pop')
+            break
+        if legit_glabel_formatted == glabel_formatted:
+            if verbose: print('\t', legit_glabel_formatted)
+            best_match = legit_glabel
+            break
+        elif glabel_formatted in legit_glabel_formatted:
+            if verbose: print('\t', legit_glabel_formatted)
+            if len(glabel_formatted) > best_match_score:
                 best_match = legit_glabel
-                break
-            elif glabel_formatted in legit_glabel_formatted:
-                if verbose: print('\t', legit_glabel_formatted)
-                if len(glabel_formatted) > best_match_score:
-                    best_match = legit_glabel
-                    best_match_score = len(glabel_formatted)
-            elif legit_glabel_formatted in glabel_formatted:
-                if verbose: print('\t', legit_glabel_formatted)
-                if len(legit_glabel_formatted) > best_match_score:
-                    best_match = legit_glabel
-                    best_match_score = len(legit_glabel_formatted)
-
-        if best_match is not None:
-            if verbose: print('\t', '-->', best_match)
-            if best_match in genres_output.keys():
-                genres_output[best_match] += genres[glabel]
-            else:
-                genres_output[best_match] = genres[glabel]
-        else:
-            if "unknown" in genres_output.keys():
-                genres_output["unknown"] += genres[glabel]
-            else:
-                genres_output["unknown"] = genres[glabel]
-    for k in genres_output.keys():
-        genres_output[k] = sorted(set(genres_output[k]))
-    return genres_output
+                best_match_score = len(glabel_formatted)
+        elif legit_glabel_formatted in glabel_formatted:
+            if verbose: print('\t', legit_glabel_formatted)
+            if len(legit_glabel_formatted) > best_match_score:
+                best_match = legit_glabel
+                best_match_score = len(legit_glabel_formatted)
+
+    if best_match is None:
+        return "unknown"
+    else:
+        return best_match
+
+
+# def aggregate_genres(genres, legit_genres, verbose=False):
+#     genres_output = dict()
+#     legit_genres_formatted = [lg.replace('-', '').replace(' ', '') for lg in legit_genres]
+#     for glabel in genres.keys():
+#         if verbose: print('\n', glabel)
+#         glabel_formatted = glabel.replace(' ', '').replace('-', '')
+#         best_match = None
+#         best_match_score = 0
+#         for legit_glabel, legit_glabel_formatted in zip(legit_genres, legit_genres_formatted):
+#             if 'jazz' in glabel_formatted:
+#                 best_match = 'jazz'
+#                 if verbose: print('\t', 'pop')
+#                 break
+#             if 'ukpop' in glabel_formatted:
+#                 best_match = 'pop'
+#                 if verbose: print('\t', 'pop')
+#                 break
+#             if legit_glabel_formatted == glabel_formatted:
+#                 if verbose: print('\t', legit_glabel_formatted)
+#                 best_match = legit_glabel
+#                 break
+#             elif glabel_formatted in legit_glabel_formatted:
+#                 if verbose: print('\t', legit_glabel_formatted)
+#                 if len(glabel_formatted) > best_match_score:
+#                     best_match = legit_glabel
+#                     best_match_score = len(glabel_formatted)
+#             elif legit_glabel_formatted in glabel_formatted:
+#                 if verbose: print('\t', legit_glabel_formatted)
+#                 if len(legit_glabel_formatted) > best_match_score:
+#                     best_match = legit_glabel
+#                     best_match_score = len(legit_glabel_formatted)
+#
+#         if best_match is not None:
+#             if verbose: print('\t', '-->', best_match)
+#             if best_match in genres_output.keys():
+#                 genres_output[best_match] += genres[glabel]
+#             else:
+#                 genres_output[best_match] = genres[glabel]
+#         else:
+#             if "unknown" in genres_output.keys():
+#                 genres_output["unknown"] += genres[glabel]
+#             else:
+#                 genres_output["unknown"] = genres[glabel]
+#     for k in genres_output.keys():
+#         genres_output[k] = sorted(set(genres_output[k]))
+#     return genres_output
 
 def get_all_playlists_uris_from_users(sp, user_ids):
     all_uris = []