Improve general performance

markovbot.py

@@ -1,6 +1,5 @@
 
 import asyncio
-import gc
 import os
 import time
 
@@ -17,23 +16,22 @@
 #    filename='logs/discord.log', encoding='utf-8', mode='w')
 
 
-def try_load_model() -> markovify.NewlineText:
+async def try_load_model() -> markovify.NewlineText:
     if not os.path.exists("data/markov_model.json"):
         logger.info(
             "markov_model.json not found. Loading messages.txt and creating model...")
-        text_model: markovify.NewlineText = model_manager.build_markov_model()
+        text_model: markovify.NewlineText = await model_manager.build_markov_model()
         logger.info("Saving model to markov_model.json...")
-        model_manager.save_model(botconfig.STATE_SIZE)
+        await model_manager.save_model(botconfig.STATE_SIZE)
         return text_model
     else:
         logger.info("markov_model.json found. Loading model...")
-        return model_manager.load_model()
+        return await model_manager.load_model()
 
-
-load_time_s = time.time()
-text_model: markovify.NewlineText = try_load_model()
+load_time_s: float = time.time()
+text_model: markovify.NewlineText = asyncio.run(try_load_model())
 text_model.compile(inplace=True)  # Compile the model for faster generation
-load_time_e = time.time()
+load_time_e: float = time.time()
 logger.info(
     f"Model loaded in {load_time_e - load_time_s:.4f} seconds.")
 

model_manager.py

@@ -1,34 +1,45 @@
 import os
 
+from aiofile import async_open
 import markovify
 from loguru import logger
 
 import botconfig
 
 
-def save_model(state_size: int) -> None:
-    with open("data/messages.txt", encoding="utf-8") as f:
-        text: str = f.read()
+async def save_model(state_size: int) -> None:
+    async with async_open("data/messages.txt", "r", encoding="utf-8") as f:
+        text: str = await f.read()
 
         text_model = markovify.NewlineText(
-            text, well_formed=False, state_size=state_size)
-        text_model.compile(inplace=True)
-        model_json: str = text_model.to_json()
-        try:
-            with open("data/markov_model.json", "w", encoding="utf-8") as model_file:
-                model_file.write(model_json)
-        except PermissionError as e:
-            logger.error(
-                f"Permission error while trying to save the model: {repr(e)}")
-            logger.info(
-                f"Permission is {os.access('data/markov_model.json', os.W_OK)}")
-
-
-def load_model() -> markovify.NewlineText:
+            text, well_formed=False, state_size=state_size, retain_original=False)
+
+        del text
+
+    text_model.compile(inplace=True)
+    model_json: str = text_model.to_json()
+    try:
+        async with async_open("data/markov_model.json", "w", encoding="utf-8") as af:
+            await af.write(model_json)
+            del model_json
+    except PermissionError as e:
+        logger.error(
+            f"Permission error while trying to save the model: {repr(e)}")
+        logger.info(
+            f"Permission is {os.access('data/markov_model.json', os.W_OK)}")
+    return
+
+
+async def load_model() -> markovify.NewlineText:
     try:
-        with open("data/markov_model.json", "r", encoding="utf-8") as model_file:
-            model_json: str = model_file.read()
-            return markovify.NewlineText.from_json(model_json)
+        async with async_open("data/markov_model.json", "r", encoding="utf-8") as model_file:
+            logger.info("Reading markov_model file...")
+            model_json: str = await model_file.read()
+            logger.info("Loading model from_json...")
+            model = markovify.NewlineText.from_json(model_json)
+            logger.info("Deleting model_json from memory...")
+            del model_json
+            return model
     except FileNotFoundError:
         logger.error(
             "markov_model.json not found. Please build the model first.")
@@ -41,11 +52,11 @@ def load_model() -> markovify.NewlineText:
         raise
 
 
-def build_markov_model() -> markovify.NewlineText:
+async def build_markov_model() -> markovify.NewlineText:
     logger.info("Loading messages.txt...")
     try:
-        with open("data/messages.txt", "r", encoding="utf-8") as f:
-            text: str = f.read()
+        async with async_open("data/messages.txt", "r", encoding="utf-8") as af:
+            text: str = await af.read()
     except FileNotFoundError:
         logger.error(
             "messages.txt not found. Please run the dataset generation script first.")
@@ -58,4 +69,7 @@ def build_markov_model() -> markovify.NewlineText:
         raise
 
     logger.info("Creating NewlineText. This may take a while")
-    return markovify.NewlineText(text, well_formed=False, state_size=botconfig.STATE_SIZE)
+    model = markovify.NewlineText(
+        text, well_formed=False, state_size=botconfig.STATE_SIZE)
+    del text
+    return model

pyproject.toml

@@ -5,14 +5,15 @@ description = "Markovbot is a Discord Bot that utilizes Markovify and Markov Cha
 readme = "README.md"
 requires-python = ">=3.14"
 dependencies = [
+    "aiofile>=3.9.0",
     "discord-py>=2.7.1",
     "loguru>=0.7.3",
     "markovify",
     "tqdm>=4.67.3",
 ]
 
 [tool.uv.sources]
-markovify = { git = "https://github.com/bscout9956/markovify", branch = "orjson" }
+markovify = { git = "https://github.com/bscout9956/markovify" }
 
 [dependency-groups]
-dev = ["snakeviz>=2.2.2"]
+dev = ["pympler>=1.1", "snakeviz>=2.2.2"]