kash-shell 0.3.18__py3-none-any.whl → 0.3.21__py3-none-any.whl

kash/model/items_model.py

@@ -675,9 +675,21 @@ class Item:
             raise FileFormatError(f"Config item is not YAML: {self.format}: {self}")
         return from_yaml_string(self.body)
 
+    def get_filename(self) -> str | None:
+        """
+        Get the store or external path filename of the item, including the
+        file extension.
+        """
+        if self.store_path:
+            return Path(self.store_path).name
+        elif self.external_path:
+            return Path(self.external_path).name
+        else:
+            return None
+
     def get_file_ext(self) -> FileExt:
         """
-        Get or infer file extension.
+        Get or infer the base file extension for the item.
         """
         if self.file_ext:
             return self.file_ext
@@ -688,7 +700,8 @@ class Item:
 
     def get_full_suffix(self) -> str:
         """
-        Get the full file extension suffix (e.g. "note.md") for this item.
+        Assemble the full file extension suffix (e.g. "resource.yml") for this item.
+        Without a leading dot.
         """
         if self.type == ItemType.extension:
             # Python files cannot have more than one . in them.
@@ -892,12 +905,14 @@ class Item:
 
     def fmt_loc(self) -> str:
         """
-        Formatted store path, external path, or title. For error messages etc.
+        Formatted store path, external path, URL, or title. Use for logging etc.
         """
         if self.store_path:
             return fmt_store_path(self.store_path)
         elif self.external_path:
             return fmt_loc(self.external_path)
+        elif self.url:
+            return fmt_loc(self.url)
         else:
             return repr(self.pick_title())
 

kash/shell/output/shell_output.py

@@ -28,6 +28,7 @@ from kash.config.text_styles import (
     STYLE_HINT,
 )
 from kash.shell.output.kmarkdown import KMarkdown
+from kash.utils.rich_custom.multitask_status import MultiTaskStatus, StatusSettings
 from kash.utils.rich_custom.rich_indent import Indent
 from kash.utils.rich_custom.rich_markdown_fork import Markdown
 
@@ -80,6 +81,20 @@ def console_pager(use_pager: bool = True):
     PrintHooks.after_pager()
 
 
+def multitask_status(
+    settings: StatusSettings | None = None, *, auto_summary: bool = True
+) -> MultiTaskStatus:
+    """
+    Create a `MultiTaskStatus` context manager for displaying multiple task progress
+    using the global shell console.
+    """
+    return MultiTaskStatus(
+        console=get_console(),
+        settings=settings,
+        auto_summary=auto_summary,
+    )
+
+
 null_style = rich.style.Style.null()
 
 

kash/utils/api_utils/api_retries.py

@@ -0,0 +1,305 @@
+from __future__ import annotations
+
+import random
+from collections.abc import Callable
+from dataclasses import dataclass
+
+
+class RetryException(RuntimeError):
+    """
+    Base exception class for retry-related errors.
+    """
+
+
+class RetryExhaustedException(RetryException):
+    """
+    Retries exhausted (this is not retriable).
+    """
+
+    def __init__(self, original_exception: Exception, max_retries: int, total_time: float):
+        self.original_exception = original_exception
+        self.max_retries = max_retries
+        self.total_time = total_time
+
+        super().__init__(
+            f"Max retries ({max_retries}) exhausted after {total_time:.1f}s. "
+            f"Final error: {type(original_exception).__name__}: {original_exception}"
+        )
+
+
+def default_is_retriable(exception: Exception) -> bool:
+    """
+    Default retriable exception checker for common rate limit patterns.
+
+    Args:
+        exception: The exception to check
+
+    Returns:
+        True if the exception should be retried with backoff
+    """
+    # Check for LiteLLM specific exceptions first, as a soft dependency.
+    try:
+        import litellm.exceptions
+
+        # Check for specific LiteLLM exception types
+        if isinstance(
+            exception,
+            (
+                litellm.exceptions.RateLimitError,
+                litellm.exceptions.APIError,
+            ),
+        ):
+            return True
+    except ImportError:
+        # LiteLLM not available, fall back to string-based detection
+        pass
+
+    # Fallback to string-based detection for general patterns
+    exception_str = str(exception).lower()
+    rate_limit_indicators = [
+        "rate limit",
+        "too many requests",
+        "try again later",
+        "429",
+        "quota exceeded",
+        "throttled",
+        "rate_limit_error",
+        "ratelimiterror",
+    ]
+
+    return any(indicator in exception_str for indicator in rate_limit_indicators)
+
+
+@dataclass(frozen=True)
+class RetrySettings:
+    """
+    Retry behavior when handling concurrent requests.
+    """
+
+    max_task_retries: int
+    """Maximum retries per individual task (0 = no retries)"""
+
+    max_total_retries: int | None = None
+    """Maximum retries across all tasks combined (None = no global limit)"""
+
+    initial_backoff: float = 1.0
+    """Base backoff time in seconds"""
+
+    max_backoff: float = 128.0
+    """Maximum backoff time in seconds"""
+
+    backoff_factor: float = 2.0
+    """Exponential backoff multiplier"""
+
+    is_retriable: Callable[[Exception], bool] = default_is_retriable
+    """Function to determine if an exception should be retried"""
+
+
+DEFAULT_RETRIES = RetrySettings(
+    max_task_retries=10,
+    max_total_retries=100,
+    initial_backoff=1.0,
+    max_backoff=128.0,
+    backoff_factor=2.0,
+    is_retriable=default_is_retriable,
+)
+"""Reasonable default retry settings with both per-task and global limits."""
+
+
+NO_RETRIES = RetrySettings(
+    max_task_retries=0,
+    max_total_retries=0,
+    initial_backoff=0.0,
+    max_backoff=0.0,
+    backoff_factor=1.0,
+    is_retriable=lambda _: False,
+)
+"""Disable retries completely."""
+
+
+def extract_retry_after(exception: Exception) -> float | None:
+    """
+    Try to extract retry-after time from exception headers or message.
+
+    Args:
+        exception: The exception to extract retry-after from
+
+    Returns:
+        Retry-after time in seconds, or None if not found
+    """
+    # Check if exception has response headers
+    response = getattr(exception, "response", None)
+    if response:
+        headers = getattr(response, "headers", None)
+        if headers and "retry-after" in headers:
+            try:
+                return float(headers["retry-after"])
+            except (ValueError, TypeError):
+                pass
+
+    # Check for retry_after attribute
+    retry_after = getattr(exception, "retry_after", None)
+    if retry_after is not None:
+        try:
+            return float(retry_after)
+        except (ValueError, TypeError):
+            pass
+
+    return None
+
+
+def calculate_backoff(
+    attempt: int,
+    exception: Exception,
+    *,
+    initial_backoff: float,
+    max_backoff: float,
+    backoff_factor: float,
+) -> float:
+    """
+    Calculate backoff time using exponential backoff with jitter.
+
+    Args:
+        attempt: Current attempt number (0-based)
+        exception: The exception that triggered the backoff
+        initial_backoff: Base backoff time in seconds
+        max_backoff: Maximum backoff time in seconds
+        backoff_factor: Exponential backoff multiplier
+
+    Returns:
+        Backoff time in seconds
+    """
+    # Try to extract retry-after header if available
+    retry_after = extract_retry_after(exception)
+    if retry_after is not None:
+        return min(retry_after, max_backoff)
+
+    # Exponential backoff: initial_backoff * (backoff_factor ^ attempt)
+    exponential_backoff = initial_backoff * (backoff_factor**attempt)
+
+    # Add significant jitter (±50% randomization) to prevent thundering herd
+    jitter_factor = 1 + (random.random() - 0.5) * 1.0
+    backoff_with_jitter = exponential_backoff * jitter_factor
+    # Add a small random base delay (0 to 50% of initial_backoff) to further spread out retries
+    base_delay = random.random() * (initial_backoff * 0.5)
+    total_backoff = backoff_with_jitter + base_delay
+
+    return min(total_backoff, max_backoff)
+
+
+## Tests
+
+
+def test_default_is_retriable():
+    """Test string-based rate limit detection."""
+    # Positive cases
+    assert default_is_retriable(Exception("Rate limit exceeded"))
+    assert default_is_retriable(Exception("Too many requests"))
+    assert default_is_retriable(Exception("HTTP 429 error"))
+    assert default_is_retriable(Exception("Quota exceeded"))
+    assert default_is_retriable(Exception("throttled"))
+    assert default_is_retriable(Exception("RateLimitError"))
+
+    # Negative cases
+    assert not default_is_retriable(Exception("Authentication failed"))
+    assert not default_is_retriable(Exception("Invalid API key"))
+    assert not default_is_retriable(Exception("Network error"))
+
+
+def test_default_is_retriable_litellm():
+    """Test LiteLLM exception detection if available."""
+    try:
+        import litellm.exceptions
+
+        # Test retriable LiteLLM exceptions
+        rate_error = litellm.exceptions.RateLimitError(
+            message="Rate limit", model="test", llm_provider="test"
+        )
+        api_error = litellm.exceptions.APIError(
+            message="API error", model="test", llm_provider="test", status_code=500
+        )
+        assert default_is_retriable(rate_error)
+        assert default_is_retriable(api_error)
+
+        # Test non-retriable exception
+        auth_error = litellm.exceptions.AuthenticationError(
+            message="Auth failed", model="test", llm_provider="test"
+        )
+        assert not default_is_retriable(auth_error)
+
+    except ImportError:
+        # LiteLLM not available, skip
+        pass
+
+
+def test_extract_retry_after():
+    """Test retry-after header extraction."""
+
+    class MockResponse:
+        def __init__(self, headers):
+            self.headers = headers
+
+    class MockException(Exception):
+        def __init__(self, response=None, retry_after=None):
+            self.response = response
+            if retry_after is not None:
+                self.retry_after = retry_after
+            super().__init__()
+
+    # Test response header
+    response = MockResponse({"retry-after": "30"})
+    assert extract_retry_after(MockException(response=response)) == 30.0
+
+    # Test retry_after attribute
+    assert extract_retry_after(MockException(retry_after=45.0)) == 45.0
+
+    # Test no retry info
+    assert extract_retry_after(MockException()) is None
+
+    # Test invalid values
+    invalid_response = MockResponse({"retry-after": "invalid"})
+    assert extract_retry_after(MockException(response=invalid_response)) is None
+
+
+def test_calculate_backoff():
+    """Test backoff calculation."""
+
+    class MockException(Exception):
+        def __init__(self, retry_after=None):
+            self.retry_after = retry_after
+            super().__init__()
+
+    # Test with retry_after header
+    exception = MockException(retry_after=30.0)
+    assert (
+        calculate_backoff(
+            attempt=1,
+            exception=exception,
+            initial_backoff=1.0,
+            max_backoff=60.0,
+            backoff_factor=2.0,
+        )
+        == 30.0
+    )
+
+    # Test exponential backoff with increased jitter and base delay
+    exception = MockException()
+    backoff = calculate_backoff(
+        attempt=1,
+        exception=exception,
+        initial_backoff=1.0,
+        max_backoff=60.0,
+        backoff_factor=2.0,
+    )
+    # base factor * (±50% jitter) + (0-50% of initial_backoff) = range calculation
+    assert 1.0 <= backoff <= 3.5
+
+    # Test max_backoff cap
+    high_backoff = calculate_backoff(
+        attempt=10,
+        exception=exception,
+        initial_backoff=1.0,
+        max_backoff=5.0,
+        backoff_factor=2.0,
+    )
+    assert high_backoff <= 5.0

kash/utils/api_utils/cache_requests_limited.py

@@ -0,0 +1,84 @@
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+from urllib.parse import urlencode
+
+import requests
+from pyrate_limiter import Duration, Limiter, Rate
+from pyrate_limiter.buckets import InMemoryBucket
+from typing_extensions import override
+
+from kash.config.logger import get_logger
+from kash.web_content.file_cache_utils import cache_file
+from kash.web_content.local_file_cache import Loadable
+
+log = get_logger(__name__)
+
+
+class CachingSession(requests.Session):
+    """
+    A `requests.Session` that adds local file caching and optional rate limiting (if
+    `limit` and `limit_interval_secs` are provided). A bit of a hack but enables
+    hot patching libraries that use `requests` without other code changes.
+    """
+
+    def __init__(
+        self,
+        *,
+        limit: int | None = None,
+        limit_interval_secs: int | None = None,
+        max_wait_secs: int = 60 * 5,
+    ):
+        super().__init__()
+        self._limiter: Limiter | None = None
+        if limit and limit_interval_secs:
+            rate = Rate(limit, Duration.SECOND * limit_interval_secs)
+            bucket = InMemoryBucket([rate])
+            # Explicitly set raise_when_fail=False and max_delay to enable blocking.
+            self._limiter = Limiter(
+                bucket, raise_when_fail=False, max_delay=Duration.SECOND * max_wait_secs
+            )
+            log.info(
+                "CachingSession: rate limiting requests with limit=%d, interval=%d, max_wait=%d",
+                limit,
+                limit_interval_secs,
+                max_wait_secs,
+            )
+
+    @override
+    def get(self, url: str | bytes, **kwargs: Any) -> Any:
+        params = kwargs.get("params")
+        # We need a unique key for the cache, so we use the URL and params.
+        url_str = url.decode() if isinstance(url, bytes) else str(url)
+        query_string = urlencode(params or {})
+        url_key = f"{url_str}?{query_string}" if query_string else url_str
+
+        def save(path: Path):
+            if self._limiter:
+                acquired = self._limiter.try_acquire("caching_session_get")
+                if not acquired:
+                    # Generally shouldn't happen.
+                    raise RuntimeError("Rate limiter failed to acquire after maximum delay")
+
+            response = super(CachingSession, self).get(url, **kwargs)
+            response.raise_for_status()
+            content = response.content
+            with open(path, "wb") as f:
+                f.write(content)
+
+        cache_result = cache_file(Loadable(url_key, save))
+
+        if not cache_result.was_cached:
+            log.debug("Cache miss, fetched: %s", url_key)
+        else:
+            log.debug("Cache hit: %s", url_key)
+
+        # A simple hack to make sure response.json() works (e.g. when using wikipediaapi needs).
+        # TODO: Wrap more carefully to ensure other methods work.
+        response = requests.Response()
+        response.status_code = 200
+        response.encoding = "utf-8"
+        response._content = cache_result.content.path.read_bytes()
+        response.url = url_key
+        return response