From 356476217d47c17018a4e75afa1ac8531a186f10 Mon Sep 17 00:00:00 2001
From: Alexander-D-Karpov <alexandr.d.karpov@gmail.com>
Date: Mon, 20 Nov 2023 01:06:17 +0300
Subject: [PATCH] added profiling script, optimised imports

---
 .gitignore                        |  2 ++
 akarpov/files/services/lema.py    |  9 +++++++--
 akarpov/files/services/preview.py |  9 ++++++++-
 akarpov/music/services/youtube.py |  3 +--
 test_start.py                     | 32 +++++++++++++++++++++++++++++++
 5 files changed, 50 insertions(+), 5 deletions(-)
 create mode 100644 test_start.py

diff --git a/.gitignore b/.gitignore
index d71c9e4..2c1b25a 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,4 +1,6 @@
 .idea
+django_setup.txt
+django_setup.prof
 
 ### Python template
 # Byte-compiled / optimized / DLL files
diff --git a/akarpov/files/services/lema.py b/akarpov/files/services/lema.py
index 719921b..388d91f 100644
--- a/akarpov/files/services/lema.py
+++ b/akarpov/files/services/lema.py
@@ -8,20 +8,25 @@
 russian_stopwords = set(stopwords.words("russian"))
 
 # Set up lemmatizers
-english_lemmatizer = WordNetLemmatizer()
-russian_lemmatizer = MorphAnalyzer()
+english_lemmatizer = None
+russian_lemmatizer = None
 
 
 def lemmatize_and_remove_stopwords(text, language="english"):
     # Tokenize the text
+    global english_lemmatizer, russian_lemmatizer
     tokens = word_tokenize(text)
 
     # Lemmatize each token based on the specified language
     lemmatized_tokens = []
     for token in tokens:
         if language == "russian":
+            if not russian_lemmatizer:
+                russian_lemmatizer = MorphAnalyzer()
             lemmatized_token = russian_lemmatizer.parse(token)[0].normal_form
         else:  # Default to English
+            if not english_lemmatizer:
+                english_lemmatizer = WordNetLemmatizer()
             lemmatized_token = english_lemmatizer.lemmatize(token)
         lemmatized_tokens.append(lemmatized_token)
 
diff --git a/akarpov/files/services/preview.py b/akarpov/files/services/preview.py
index 4b761c3..fd83a52 100644
--- a/akarpov/files/services/preview.py
+++ b/akarpov/files/services/preview.py
@@ -16,7 +16,7 @@
     "Consola.ttf",
 ]
 
-manager = PreviewManager(cache_path, create_folder=True)
+manager = None
 
 
 def textfile_to_image(textfile_path) -> Image:
@@ -79,7 +79,10 @@ def _font_points_to_pixels(pt):
 
 
 def create_preview(file_path: str) -> str:
+    global manager
     # TODO: add text image generation/code image
+    if not manager:
+        manager = PreviewManager(cache_path, create_folder=True)
     if manager.has_jpeg_preview(file_path):
         return manager.get_jpeg_preview(file_path, height=500)
     return ""
@@ -91,6 +94,10 @@ def get_file_mimetype(file_path: str) -> str:
 
 
 def get_description(file_path: str) -> str:
+    global manager
+    if not manager:
+        manager = PreviewManager(cache_path, create_folder=True)
+
     if manager.has_text_preview(file_path):
         return manager.get_text_preview(file_path)
     return ""
diff --git a/akarpov/music/services/youtube.py b/akarpov/music/services/youtube.py
index a52d676..b95cbb1 100644
--- a/akarpov/music/services/youtube.py
+++ b/akarpov/music/services/youtube.py
@@ -9,7 +9,6 @@
 from PIL import Image
 from pydub import AudioSegment
 from pytube import Search, YouTube
-from yt_dlp import YoutubeDL
 
 from akarpov.music.models import Song
 from akarpov.music.services.db import load_track
@@ -67,7 +66,7 @@ def parse_description(description: str) -> list:
 def download_from_youtube_link(link: str, user_id: int) -> Song:
     song = None
 
-    with YoutubeDL(ydl_opts) as ydl:
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
         info_dict = ydl.extract_info(link, download=False)
         title = info_dict.get("title", None)
         description = info_dict.get("description", None)
diff --git a/test_start.py b/test_start.py
new file mode 100644
index 0000000..71ef5d7
--- /dev/null
+++ b/test_start.py
@@ -0,0 +1,32 @@
+import cProfile
+import os
+import pstats
+
+import django
+
+
+# Function to run the Django setup process, which you want to profile
+def django_setup():
+    os.environ.setdefault("DJANGO_SETTINGS_MODULE", "config.settings.local")
+    django.setup()
+
+
+# Create a Profile object and run the django_setup function
+profiler = cProfile.Profile()
+profiler.enable()
+django_setup()
+profiler.disable()
+
+# Write the stats to a .prof file
+profiler.dump_stats("django_setup.prof")
+
+# Create a Stats object and print the sorted stats to a text file
+with open("django_setup.txt", "w") as stream:
+    stats = pstats.Stats(profiler, stream=stream)
+    stats.sort_stats("cumtime")  # Sort the statistics by cumulative time
+    stats.print_stats()  # Print the statistics to the stream
+
+# Optionally, print the stats to the console as well
+stats = pstats.Stats(profiler)
+stats.sort_stats("cumtime")
+stats.print_stats()