Merge pull request #51 from A-Baji/dev

3.0.5

Author: A-Baji

.coveragerc

@@ -0,0 +1,8 @@
+[run]
+omit = 
+    */tests/*
+
+[report]
+exclude_lines =
+    if __name__ == "__main__":
+    fine_tune = client.fine_tuning.jobs.create

CHANGELOG.md

@@ -2,6 +2,17 @@
 
 Observes [Semantic Versioning](https://semver.org/spec/v2.0.0.html) standard and [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) convention.
 
+## [3.0.5] - 07-9-2024
+
+### Added
+
+- `--force` arg to model deletion
+- error handling for when user was not found in logs
+
+### Changed
+
+- Include `tests/` within the package to allow reuse in parent package
+
 ## [3.0.4] - 07-1-2024
 
 ### Removed
@@ -109,6 +120,7 @@ Observes [Semantic Versioning](https://semver.org/spec/v2.0.0.html) standard and
 
 - switched to `pathlib` for file path parsing
 
+[3.0.5]: https://github.com/A-Baji/discordAI-modelizer/compare/3.0.4...3.0.5
 [3.0.4]: https://github.com/A-Baji/discordAI-modelizer/compare/3.0.3...3.0.4
 [3.0.3]: https://github.com/A-Baji/discordAI-modelizer/compare/3.0.2...3.0.3
 [3.0.2]: https://github.com/A-Baji/discordAI-modelizer/compare/3.0.1...3.0.2

discordai_modelizer/command_line/command_line.py

@@ -55,7 +55,9 @@ def read_modelizer_args(args, model_subcommand, job_subcommand):
                 use_existing=args.use_existing,
             )
         elif args.subcommand == "delete":
-            display(openai_wrapper.delete_model(args.model_id, args.openai_key))
+            display(
+                openai_wrapper.delete_model(args.model_id, args.openai_key, args.force)
+            )
         else:
             raise argparse.ArgumentError(
                 model_subcommand,

discordai_modelizer/command_line/subparsers.py

@@ -177,6 +177,9 @@ def setup_model_delete(model_subcommand, is_parent=False):
     model_delete_required_named = model_delete.add_argument_group(
         "required named arguments"
     )
+    model_delete_optional_named = model_delete.add_argument_group(
+        "optional named arguments"
+    )
 
     model_delete_required_named.add_argument(
         "-o",
@@ -194,6 +197,14 @@ def setup_model_delete(model_subcommand, is_parent=False):
         help="Target model id",
     )
 
+    model_delete_optional_named.add_argument(
+        "--force",
+        action="store_true",
+        required=False,
+        dest="force",
+        help="Skips the deletion confirmation dialogue",
+    )
+
 
 def setup_job_list(job_subcommand, is_parent=False):
     job_list = job_subcommand.add_parser(

discordai_modelizer/customize.py

@@ -5,7 +5,7 @@
 import pathlib
 
 from openai import OpenAI
-from discordai_modelizer.gen_dataset import parse_logs, get_lines
+from discordai_modelizer.gen_dataset import parse_logs, get_lines, UserNotFoundError
 
 MODEL_MAP = {
     "davinci": "davinci-002",
@@ -87,14 +87,18 @@ def create_model(
         print("INFO: Using existing dataset... Skipping download and parsing.")
     else:
         print("INFO: Parsing chat logs into an openAI compatible dataset...")
-        parse_logs(
-            full_logs_path,
-            channel_id,
-            user_id,
-            thought_time,
-            thought_max,
-            thought_min,
-        )
+        try:
+            parse_logs(
+                full_logs_path,
+                channel_id,
+                user_id,
+                thought_time,
+                thought_max,
+                thought_min,
+            )
+        except UserNotFoundError as e:
+            print(f"ERROR: {e}")
+            return
         get_lines(full_dataset_path, max_entry_count, offset, select_mode, reverse)
         if not clean:
             print(f"INFO: Dataset saved to {full_dataset_path}")

discordai_modelizer/gen_dataset.py

@@ -9,6 +9,10 @@
 import pathlib
 
 
+class UserNotFoundError(Exception):
+    pass
+
+
 def parse_logs(
     file: str,
     channel: str,
@@ -78,6 +82,10 @@ def add_to_dataset(thought: str):
             if msg["author"].get("name") == username
             and (user_id is None or msg["author"].get("discriminator") == user_id)
         ]
+        if not messages:
+            raise UserNotFoundError(
+                f"No messages found in chat logs for user: {username}"
+            )
         thought = build_thought("", messages[0])
         for i, msg in enumerate(messages[1::]):
             if msg["content"]:

discordai_modelizer/openai.py

@@ -98,10 +98,14 @@ def cancel_job(job_id: str, openai_key: str = os.getenv("OPENAI_API_KEY")) -> di
 
 
 def delete_model(
-    model_name: str, openai_key: str = os.getenv("OPENAI_API_KEY")
+    model_name: str, openai_key: str = os.getenv("OPENAI_API_KEY"), force=False
 ) -> dict:
-    confirm = input(
-        "Are you sure you want to delete this model? This action is not reversable. Y/N: "
+    confirm = (
+        "yes"
+        if force
+        else input(
+            "Are you sure you want to delete this model? This action is not reversable. Y/N: "
+        )
     )
     if confirm.lower() not in ["y", "yes"]:
         print("Cancelling model deletion...")

tests/__init__.py renamed to discordai_modelizer/tests/__init__.py

@@ -1,9 +1,9 @@
 import os
 import pathlib
 import appdirs
-import pytest
 
 from discordai_modelizer import customize
+from pytest import fixture
 
 CHANNEL_ID = os.environ["CHANNEL_ID"]
 USER = os.environ["USERNAME"]
@@ -18,7 +18,7 @@ def list_dict_comp(x, y):
         assert d1 == d2
 
 
-@pytest.fixture(scope="session")
+@fixture(scope="session")
 def default_file_output():
     # Ensure a clean start by removing any existing files
     if FULL_LOGS_PATH.exists():
@@ -33,3 +33,14 @@ def default_file_output():
         FULL_LOGS_PATH.unlink()
     if FULL_DATASET_PATH.exists():
         FULL_DATASET_PATH.unlink()
+
+
+@fixture(scope="function")
+def unset_envs():
+    DISCORD_BOT_TOKEN = os.environ["DISCORD_BOT_TOKEN"]
+    OPENAI_API_KEY = os.environ["OPENAI_API_KEY"]
+    del os.environ["DISCORD_BOT_TOKEN"]
+    del os.environ["OPENAI_API_KEY"]
+    yield
+    os.environ["DISCORD_BOT_TOKEN"] = DISCORD_BOT_TOKEN
+    os.environ["OPENAI_API_KEY"] = OPENAI_API_KEY

tests/conftest.py renamed to discordai_modelizer/tests/conftest.py

@@ -4,20 +4,20 @@
     {"id": "dall-e-2", "created": "2023-11-01 00:22:57"},
     {"id": "tts-1-hd-1106", "created": "2023-11-03 23:18:53"},
     {"id": "tts-1-hd", "created": "2023-11-03 21:13:35"},
+    {"id": "gpt-4-turbo-2024-04-09", "created": "2024-04-08 18:41:17"},
+    {"id": "gpt-4-turbo", "created": "2024-04-05 23:57:21"},
     {"id": "gpt-3.5-turbo-1106", "created": "2023-11-02 21:15:48"},
+    {"id": "dall-e-3", "created": "2023-10-31 20:46:29"},
     {"id": "gpt-4-0125-preview", "created": "2024-01-23 19:20:12"},
     {"id": "gpt-4-turbo-preview", "created": "2024-01-23 19:22:57"},
     {"id": "text-embedding-3-small", "created": "2024-01-22 18:43:17"},
     {"id": "text-embedding-3-large", "created": "2024-01-22 19:53:00"},
     {"id": "gpt-3.5-turbo-16k", "created": "2023-05-10 22:35:02"},
     {"id": "gpt-4-1106-preview", "created": "2023-11-02 20:33:26"},
-    {"id": "gpt-4-turbo-2024-04-09", "created": "2024-04-08 18:41:17"},
     {"id": "babbage-002", "created": "2023-08-21 16:16:55"},
     {"id": "gpt-4o-2024-05-13", "created": "2024-05-10 19:08:52"},
-    {"id": "gpt-4-turbo", "created": "2024-04-05 23:57:21"},
     {"id": "gpt-4", "created": "2023-06-27 16:13:31"},
     {"id": "gpt-4-0613", "created": "2023-06-12 16:54:56"},
-    {"id": "dall-e-3", "created": "2023-10-31 20:46:29"},
     {"id": "gpt-3.5-turbo-0125", "created": "2024-01-23 22:19:18"},
     {"id": "tts-1-1106", "created": "2023-11-03 23:14:01"},
     {"id": "gpt-3.5-turbo", "created": "2023-02-28 18:56:42"},
@@ -203,12 +203,30 @@
         "object": "model",
         "owned_by": "system",
     },
+    {
+        "id": "gpt-4-turbo-2024-04-09",
+        "created": "2024-04-08 18:41:17",
+        "object": "model",
+        "owned_by": "system",
+    },
+    {
+        "id": "gpt-4-turbo",
+        "created": "2024-04-05 23:57:21",
+        "object": "model",
+        "owned_by": "system",
+    },
     {
         "id": "gpt-3.5-turbo-1106",
         "created": "2023-11-02 21:15:48",
         "object": "model",
         "owned_by": "system",
     },
+    {
+        "id": "dall-e-3",
+        "created": "2023-10-31 20:46:29",
+        "object": "model",
+        "owned_by": "system",
+    },
     {
         "id": "gpt-4-0125-preview",
         "created": "2024-01-23 19:20:12",
@@ -245,12 +263,6 @@
         "object": "model",
         "owned_by": "system",
     },
-    {
-        "id": "gpt-4-turbo-2024-04-09",
-        "created": "2024-04-08 18:41:17",
-        "object": "model",
-        "owned_by": "system",
-    },
     {
         "id": "babbage-002",
         "created": "2023-08-21 16:16:55",
@@ -263,12 +275,6 @@
         "object": "model",
         "owned_by": "system",
     },
-    {
-        "id": "gpt-4-turbo",
-        "created": "2024-04-05 23:57:21",
-        "object": "model",
-        "owned_by": "system",
-    },
     {
         "id": "gpt-4",
         "created": "2023-06-27 16:13:31",
@@ -281,12 +287,6 @@
         "object": "model",
         "owned_by": "openai",
     },
-    {
-        "id": "dall-e-3",
-        "created": "2023-10-31 20:46:29",
-        "object": "model",
-        "owned_by": "system",
-    },
     {
         "id": "gpt-3.5-turbo-0125",
         "created": "2024-01-23 22:19:18",