inspect-ai 0.3.99__py3-none-any.whl → 0.3.101__py3-none-any.whl

inspect_ai/_util/content.py

@@ -1,9 +1,14 @@
 from typing import Literal, Union
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, JsonValue
 
 
-class ContentText(BaseModel):
+class ContentBase(BaseModel):
+    internal: JsonValue | None = Field(default=None)
+    """Model provider specific payload - typically used to aid transformation back to model types."""
+
+
+class ContentText(ContentBase):
     """Text content."""
 
     type: Literal["text"] = Field(default="text")
@@ -16,7 +21,7 @@ class ContentText(BaseModel):
     """Was this a refusal message?"""
 
 
-class ContentReasoning(BaseModel):
+class ContentReasoning(ContentBase):
     """Reasoning content.
 
     See the specification for [thinking blocks](https://docs.anthropic.com/en/docs/build-with-claude/extended-thinking#understanding-thinking-blocks) for Claude models.
@@ -35,7 +40,7 @@ class ContentReasoning(BaseModel):
     """Indicates that the explicit content of this reasoning block has been redacted."""
 
 
-class ContentImage(BaseModel):
+class ContentImage(ContentBase):
     """Image content."""
 
     type: Literal["image"] = Field(default="image")
@@ -51,7 +56,7 @@ class ContentImage(BaseModel):
     """
 
 
-class ContentAudio(BaseModel):
+class ContentAudio(ContentBase):
     """Audio content."""
 
     type: Literal["audio"] = Field(default="audio")
@@ -64,7 +69,7 @@ class ContentAudio(BaseModel):
     """Format of audio data ('mp3' or 'wav')"""
 
 
-class ContentVideo(BaseModel):
+class ContentVideo(ContentBase):
     """Video content."""
 
     type: Literal["video"] = Field(default="video")

inspect_ai/_util/interrupt.py

@@ -1,6 +1,6 @@
 import anyio
 
-from .working import check_sample_working_limit
+from inspect_ai.util._limit import check_working_limit
 
 
 def check_sample_interrupt() -> None:
@@ -12,4 +12,4 @@ def check_sample_interrupt() -> None:
         raise anyio.get_cancelled_exc_class()
 
     # check for working_limit
-    check_sample_working_limit()
+    check_working_limit()

inspect_ai/_util/text.py

@@ -1,12 +1,19 @@
 import random
 import re
 import string
+import textwrap
 from logging import getLogger
 from typing import List, NamedTuple
 
 logger = getLogger(__name__)
 
 
+def truncate_text(text: str, max_length: int) -> str:
+    if len(text) <= max_length:
+        return text
+    return textwrap.shorten(text, width=max_length, placeholder="...")
+
+
 def strip_punctuation(s: str) -> str:
     return s.strip(string.whitespace + string.punctuation)
 

inspect_ai/_util/working.py

@@ -1,19 +1,14 @@
 import time
 from contextvars import ContextVar
 
-from inspect_ai.util._limit import LimitExceededError
+from inspect_ai.util._limit import check_working_limit, record_waiting_time
 
 
-def init_sample_working_limit(start_time: float, working_limit: float | None) -> None:
-    _sample_working_limit.set(working_limit)
+def init_sample_working_time(start_time: float) -> None:
     _sample_start_time.set(start_time)
     _sample_waiting_time.set(0)
 
 
-def end_sample_working_limit() -> None:
-    _sample_working_limit.set(None)
-
-
 def sample_waiting_time() -> float:
     return _sample_waiting_time.get()
 
@@ -23,37 +18,13 @@ def sample_working_time() -> float:
 
 
 def report_sample_waiting_time(waiting_time: float) -> None:
+    # record and check for scoped limits
+    record_waiting_time(waiting_time)
+    check_working_limit()
+
+    # record sample-level limits
     _sample_waiting_time.set(_sample_waiting_time.get() + waiting_time)
-    check_sample_working_limit()
-
-
-def check_sample_working_limit() -> None:
-    from inspect_ai.log._transcript import SampleLimitEvent, transcript
-
-    # no check if we don't have a limit
-    working_limit = _sample_working_limit.get()
-    if working_limit is None:
-        return
-
-    # are we over the limit?
-    running_time = time.monotonic() - _sample_start_time.get()
-    working_time = running_time - sample_waiting_time()
-    if working_time > working_limit:
-        message = f"Exceeded working time limit ({working_limit:,} seconds)"
-        transcript()._event(
-            SampleLimitEvent(type="working", limit=int(working_limit), message=message)
-        )
-        raise LimitExceededError(
-            type="working",
-            value=int(working_time),
-            limit=int(working_limit),
-            message=message,
-        )
-
-
-_sample_working_limit: ContextVar[float | None] = ContextVar(
-    "sample_working_limit", default=None
-)
+
 
 _sample_start_time: ContextVar[float] = ContextVar("sample_start_time", default=0)
 

inspect_ai/_view/schema.py

@@ -45,9 +45,10 @@ def sync_view_schema() -> None:
                 "false",
             ],
             cwd=WWW_DIR,
+            check=True,
         )
 
-        subprocess.run(["yarn", "prettier:write"], cwd=types_path.parent)
+        subprocess.run(["yarn", "prettier:write"], cwd=types_path.parent, check=True)
 
         shutil.copyfile(types_path, vs_code_types_path)
 

inspect_ai/_view/www/CLAUDE.md

@@ -0,0 +1,15 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Build/Lint/Test Commands
+
+- Run type checker, linter, and tests: `yarn check-all`
+- Run type checker: `yarn tsc`
+- Run linter: `yarn lint`
+- Run tests: `yarn test`
+
+## Other Information
+
+- The code in this project is typescript, learn more about the configuration by inspecting package.json.
+- Respect existing code patterns when modifying files. Run linting before committing changes.