Rewrote the argparsing functionality to enable autocompletion via the "kislyuk/argcomplete" package.

This essentially means:
- `argparsing.py` does a minimal amount of work to initialize the arg parser.
   Then it attempts dynamic completion.
- `argvalidation.py` processes the raw arguments parsed in
  `argparsing.py`, validates them, issues warnings (if required), and
  splits them into logical groupings
- Commands in `gcli.py` have been moved to `cmd.py`
- `main.py` provides an initial control path to call these functions in
  succession

1 files changed, 2 insertions, 3 deletions

diff --git a/src/gpt_chat_cli/openai_wrappers.py b/src/gpt_chat_cli/openai_wrappers.py
index 3e1ec06..d478531 100644
--- a/src/gpt_chat_cli/openai_wrappers.py
+++ b/src/gpt_chat_cli/openai_wrappers.py
@@ -5,6 +5,8 @@ from typing import Any, List, Optional, Generator
 from dataclasses import dataclass
 from enum import Enum, auto
 
+from .argvalidation import CompletionArguments
+
 @dataclass
 class Delta:
     content: Optional[str] = None
@@ -21,7 +23,6 @@ class FinishReason(Enum):
         if finish_reason_str is None:
             return FinishReason.NONE
         return FinishReason[finish_reason_str.upper()]
-
 @dataclass
 class Choice:
     delta: Delta
@@ -79,8 +80,6 @@ class OpenAIChatResponse:
 
 OpenAIChatResponseStream = Generator[OpenAIChatResponse, None, None]
 
-from .argparsing import CompletionArguments
-
 def create_chat_completion(hist : ChatHistory, args: CompletionArguments) \
     -> OpenAIChatResponseStream: