Merge branch 'master' into mic_sound_visualisation

Author: HadronCollider

app/criteria/utils.py

@@ -1,6 +1,7 @@
 from app.root_logger import get_root_logger
 import math
 import traceback
+import string   # to remove punctuation
 from typing import Optional, Callable
 
 from app.audio import Audio
@@ -60,13 +61,20 @@ def get_proportional_result(value: float,
     else:
         return f(upper_bound / value)
 
-
 def get_fillers(fillers: list, audio: Audio) -> list:
     found_fillers = []
+    # пунктуация + пробелы для str.translate
+    removable = string.punctuation + string.whitespace
+    translation_table = str.maketrans('', '', removable)
+    
     for audio_slide in audio.audio_slides:
         found_slide_fillers = []
+        # добавлена предобработка слов - перевод в нижний регистр, очистка от пунктуации
         audio_slide_words = [
-            recognized_word.word.value for recognized_word in audio_slide.recognized_words]
+            recognized_word.word.value.lower().translate(translation_table) 
+            for recognized_word in audio_slide.recognized_words
+            ]
+        
         for i in range(len(audio_slide_words)):
             for filler in fillers:
                 filler_split = filler.split()

app/filters.py

@@ -27,7 +27,7 @@ def __new__(cls):
             cls.init_done = True
         return cls.instance
 
-    def __collect_filters(self, filters: dict) -> (dict, dict):
+    def __collect_filters(self, filters: dict) -> tuple[dict, dict]:
         simple_filters = {}
         complex_filters = {}
 
@@ -64,8 +64,11 @@ def __create_simple_query(self, applicable_filters: dict) -> dict:
                 logger.info("ELSE ID: " + str(values[0]))
                 mongodb_query[key] = ObjectId(values[0])
             elif key == "username" or key == "full_name":
-                # Частичное совпадение
-                mongodb_query[key] = {"$regex": regex.escape(values[0], literal_spaces=True)}
+                # Частичное совпадение без учета регистра
+                mongodb_query[key] = {
+                    "$regex": regex.escape(values[0], literal_spaces=True),
+                    "$options": 'i'
+                }
             elif key == "presentation_record_duration":
                 # Range по длительности воспроизведения
                 start_range_time = datetime.strptime(values[0], "%M:%S").time()