Merge pull request #32 from atomiechen/dev

Bump version to 0.8.2

.gitignore

@@ -31,3 +31,7 @@ temp*
 *.flac
 *.aac
 
+# test output
+.coverage*
+.pytest_cache
+htmlcov

CHANGELOG.md

@@ -5,6 +5,45 @@ All notable changes to HandyLLM will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 
 
+## [0.8.2] - 2024-06-30
+
+### Added
+
+- `hprompt`: load methods now support `cls` parameter for prompt type specification
+- `ChatPrompt` and `CompletionsPrompt` support optional request and meta
+- `ChatPrompt` :
+  - supports add dict
+  - add `add_message(...)` method
+- `CompletionsPrompt`:
+  - add `add_text(...)` method
+- `PromptConverter`: `yaml.dump` uses `allow_unicode=True` option
+- move all type definitions to `_types.py`
+- support for package development:
+  - add `requirement.txt` for development
+  - add `scripts/test.sh` for running tests
+  - add test scripts in `tests` folder
+
+### Fixed
+
+- `HandyPrompt.eval(...)` should not make directories for output paths
+- `CompletionsPrompt._run_with_client(...)`: misplaced `run_config` param
+- `PromptConverter`
+  - fix variable replacement for `content_array` message
+  - fix wrong return type of `stream_msgs2raw` and `astream_msgs2raw`
+- `requestor`:
+  - `httpx.Response` should use `reason_phrase` to get error reason
+  - `acall()` fix missing brackets for await
+  - `_call_raw()` and `_acall_raw()` intercept and raise new exception without original one
+  - `_acall_raw()`: read the response first to prevent `httpx.ResponseNotRead` before getting error message
+- `_utils.exception2err_msg(...)` should append error message instead of printing
+- change `io.IOBase` to `IO[str]` for file descriptors (e.g. `RunConfig.output_fd`)
+- fix other type hints
+
+### Changed
+
+- move all old files in `tests` folder to `examples` folder
+
+
 ## [0.8.1] - 2024-06-10
 
 ### Fixed

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "HandyLLM"
-version = "0.8.1"
+version = "0.8.2"
 authors = [
   { name="Atomie CHEN", email="[email protected]" },
 ]
@@ -33,3 +33,8 @@ Changelog = "https://github.com/atomiechen/HandyLLM/blob/master/CHANGELOG.md"
 
 [project.scripts]
 handyllm = "handyllm.__main__:cli"
+
+[tool.pytest.ini_options]
+addopts = [
+  "--ignore=examples"
+]

requirements.txt

@@ -0,0 +1,5 @@
+-e .
+
+# tests
+pytest
+pytest-cov

scripts/test.sh

@@ -0,0 +1,4 @@
+set -e
+set -x
+
+pytest --cov --cov-report=term-missing -o console_output_style=progress --cov-report=html

src/handyllm/_str_enum.py

@@ -4,7 +4,7 @@
 class StrEnumMeta(EnumMeta):
     def __contains__(cls, item):
         if isinstance(item, str):
-            return item in iter(cls)
+            return item in iter(cls) # type: ignore
         return super().__contains__(item)
 
 

src/handyllm/_types.py

@@ -0,0 +1,18 @@
+import sys
+from typing import Any, Awaitable, Callable, Dict, MutableMapping, Optional, Union
+from os import PathLike
+
+
+if sys.version_info >= (3, 9):
+    PathType = Union[str, PathLike[str]]
+else:
+    PathType = Union[str, PathLike]
+
+VarMapType = MutableMapping[str, str]
+
+SyncHandlerChat = Callable[[str, Optional[str], Optional[Dict]], Any]
+SyncHandlerCompletions = Callable[[str], Any]
+AsyncHandlerChat = Callable[[str, Optional[str], Optional[Dict]], Awaitable[Any]]
+AsyncHandlerCompletions = Callable[[str], Awaitable[Any]]
+OnChunkType = Union[SyncHandlerChat, SyncHandlerCompletions, AsyncHandlerChat, AsyncHandlerCompletions]
+

src/handyllm/_utils.py

@@ -74,7 +74,7 @@ def exception2err_msg(exception: Exception):
     err_msg = f"Exception: {type(exception).__module__}.{type(exception).__name__}"
     err_msg += f"\nDetailed info: {repr(exception)}"
     if exception.args:
-        print(f"\nException arguments: {exception.args}")
+        err_msg += f"\nException arguments: {exception.args}"
     return err_msg
 
 def _chat_log_response_final(logger, log_marks, kwargs, messages, start_time, role, content, err_msg=None):
@@ -91,7 +91,7 @@ def _chat_log_response(logger, log_marks, kwargs, messages, start_time, response
     if logger is not None:
         if stream:
             if inspect.isasyncgen(response):
-                async def wrapper(response):
+                async def wrapper(response): # type: ignore
                     content = ''
                     role = ''
                     async for data in response:
@@ -155,7 +155,7 @@ def _completions_log_response(logger, log_marks, kwargs, prompt, start_time, res
     if logger is not None:
         if stream:
             if inspect.isasyncgen(response):
-                async def wrapper(response):
+                async def wrapper(response): # type: ignore
                     text = ''
                     async for data in response:
                         try:

src/handyllm/endpoint_manager.py

@@ -65,17 +65,17 @@ def clear(self):
     def __len__(self) -> int:
         return len(self._endpoints)
 
-    def __getitem__(self, idx: int) -> Endpoint:
+    def __getitem__(self, idx):
         return self._endpoints[idx]
 
-    def __setitem__(self, idx: int, endpoint: Endpoint):
+    def __setitem__(self, idx, endpoint):
         self._endpoints[idx] = endpoint
 
-    def __delitem__(self, idx: int):
+    def __delitem__(self, idx):
         del self._endpoints[idx]
 
-    def insert(self, idx: int, endpoint: Endpoint):
-        self._endpoints.insert(idx, endpoint)
+    def insert(self, index: int, value: Endpoint):
+        self._endpoints.insert(index, value)
 
     def add_endpoint_by_info(self, **kwargs):
         endpoint = Endpoint(**kwargs)