flashlite 0.1.1__tar.gz → 0.2.0__tar.gz

{flashlite-0.1.1 → flashlite-0.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: flashlite
-Version: 0.1.1
+Version: 0.2.0
 Summary: Batteries-included wrapper for litellm with rate limiting, retries, templating, and more
 Author-email: ndalton12 <niall.dalton12@gmail.com>
 License-File: LICENSE.md

{flashlite-0.1.1 → flashlite-0.2.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "flashlite"
-version = "0.1.1"
+version = "0.2.0"
 description = "Batteries-included wrapper for litellm with rate limiting, retries, templating, and more"
 readme = "README.md"
 authors = [
@@ -35,7 +35,10 @@ build-backend = "hatchling.build"
 asyncio_mode = "auto"
 asyncio_default_fixture_loop_scope = "function"
 testpaths = ["tests"]
-addopts = "-v --tb=short"
+addopts = "-v --tb=short -m 'not slow'"
+markers = [
+    "slow: marks tests that may require real waiting (deselected by default, run with -m slow)",
+]
 
 [tool.mypy]
 python_version = "3.13"

flashlite-0.2.0/src/flashlite/_spinner.py

@@ -0,0 +1,91 @@
+"""Terminal spinner for user-visible progress during async waits.
+
+Provides a lightweight, non-blocking spinner that renders to stderr
+when — and only when — the output is an interactive terminal.  Multiple
+concurrent ``Spinner`` instances (e.g. from ``complete_many``) are
+gracefully collapsed so only one animation is visible at a time.
+"""
+
+import asyncio
+import sys
+import time
+
+# Braille-dot frames — smooth and compact.
+_FRAMES = ("⠋", "⠙", "⠹", "⠸", "⠼", "⠴", "⠦", "⠧", "⠇", "⠏")
+_INTERVAL = 0.08  # seconds between frame updates
+
+# ANSI helpers
+_CYAN = "\033[36m"
+_DIM = "\033[2m"
+_RESET = "\033[0m"
+_CLEAR_LINE = "\r\033[K"
+
+# Module-level guard — only one spinner renders at a time.
+_active: bool = False
+
+
+class Spinner:
+    """Async context manager that shows a terminal spinner on stderr.
+
+    The spinner only appears when stderr is a TTY **and** no other
+    ``Spinner`` is already active, making it safe for concurrent use
+    inside ``complete_many``.
+
+    Args:
+        message: Text displayed next to the spinner.
+        delay: Grace period (seconds) before the spinner appears.
+            If the wrapped operation finishes within this window the
+            spinner is never rendered, avoiding flicker for fast calls.
+    """
+
+    def __init__(self, message: str = "Working...", *, delay: float = 0.3) -> None:
+        self.message = message
+        self.delay = delay
+        self._task: asyncio.Task[None] | None = None
+        self._owns_active = False
+        self._start: float = 0.0
+
+    # -- internal -----------------------------------------------------
+
+    async def _render(self) -> None:
+        """Background coroutine that draws frames until cancelled."""
+        await asyncio.sleep(self.delay)
+        idx = 0
+        while True:
+            elapsed = time.monotonic() - self._start
+            frame = _FRAMES[idx % len(_FRAMES)]
+            sys.stderr.write(
+                f"{_CLEAR_LINE}{_CYAN}{frame}{_RESET} {self.message} "
+                f"{_DIM}({elapsed:.1f}s){_RESET}"
+            )
+            sys.stderr.flush()
+            idx += 1
+            await asyncio.sleep(_INTERVAL)
+
+    @staticmethod
+    def _clear() -> None:
+        sys.stderr.write(_CLEAR_LINE)
+        sys.stderr.flush()
+
+    # -- context manager ----------------------------------------------
+
+    async def __aenter__(self) -> "Spinner":
+        global _active  # noqa: PLW0603
+        if sys.stderr.isatty() and not _active:
+            _active = True
+            self._owns_active = True
+            self._start = time.monotonic()
+            self._task = asyncio.create_task(self._render())
+        return self
+
+    async def __aexit__(self, *_: object) -> None:
+        global _active  # noqa: PLW0603
+        if self._task is not None:
+            self._task.cancel()
+            try:
+                await self._task
+            except asyncio.CancelledError:
+                pass
+            self._clear()
+        if self._owns_active:
+            _active = False

{flashlite-0.1.1 → flashlite-0.2.0}/src/flashlite/client.py

@@ -7,6 +7,7 @@ from typing import Any, TypeVar, overload
 
 from pydantic import BaseModel
 
+from ._spinner import Spinner
 from .cache import CacheBackend, MemoryCache
 from .config import FlashliteConfig, load_env_files
 from .conversation import ContextManager, Conversation
@@ -223,7 +224,8 @@ class Flashlite:
         if self._config.log_requests:
             logger.info(f"Completion request: model={request.model}")
 
-        response = await core_complete(request)
+        async with Spinner(f"Waiting for {request.model}...", delay=0.2):
+            response = await core_complete(request)
 
         if self._config.log_requests:
             logger.info(
@@ -395,10 +397,12 @@ class Flashlite:
             else:
                 extra_kwargs["tools"] = tools_to_openai(tools)
 
-        # Build request
+        # Build request (template/variables stored for middleware traceability)
         request = CompletionRequest(
             model=resolved_model,
             messages=final_messages,
+            template=template,
+            variables=variables,
             temperature=temperature,
             max_tokens=max_tokens,
             max_completion_tokens=max_completion_tokens,