eval-studio-client 1.0.0a1__py3-none-any.whl → 1.0.1__py3-none-any.whl

eval_studio_client/tests.py

@@ -11,6 +11,7 @@ from typing import Union
 from eval_studio_client import api
 from eval_studio_client import documents as d7s
 from eval_studio_client import perturbators as p10s
+from eval_studio_client import utils
 from eval_studio_client.api import models
 
 
@@ -85,15 +86,9 @@ class TestCaseGenerator(enum.Enum):
 
 
 @dataclasses.dataclass
-class TestCaseGenerationHandle:
+class _TestCaseGenerationHandle:
 
     name: Any | None
-    create_time: Optional[datetime.datetime] = None
-    creator: Optional[str] = None
-    update_time: Optional[datetime.datetime] = None
-    updater: Optional[str] = None
-    delete_time: Optional[datetime.datetime] = None
-    deleter: Optional[str] = None
     progress: Optional[float] = None
     progress_message: Optional[str] = None
     error: Optional[models.RpcStatus] = None
@@ -102,11 +97,40 @@ class TestCaseGenerationHandle:
     @staticmethod
     def _from_operation(
         res: models.V1GenerateTestCasesResponse | models.V1GetOperationResponse,
-    ) -> "TestCaseGenerationHandle":
+    ) -> "_TestCaseGenerationHandle":
         """Converts an API operation to prompt generation handle."""
         op: models.V1Operation | None = res.operation
         if not op:
-            return TestCaseGenerationHandle(name=None)
+            return _TestCaseGenerationHandle(name=None)
+
+        # progress
+        if hasattr(op, "metadata") and op.metadata:
+            meta_dict = op.metadata.to_dict() or {}
+        else:
+            meta_dict = {}
+
+        return _TestCaseGenerationHandle(
+            name=op.name,
+            progress=meta_dict.get("progress"),
+            progress_message=meta_dict.get("progressMessage"),
+            error=op.error,
+            done=op.done,
+        )
+
+
+@dataclasses.dataclass
+class _TestCaseLibraryGetHandle(_TestCaseGenerationHandle):
+
+    @staticmethod
+    def _from_operation(
+        res: (
+            models.V1ImportTestCasesFromLibraryResponse | models.V1GetOperationResponse
+        ),
+    ) -> "_TestCaseLibraryGetHandle":
+        """Converts an API operation to prompt library handle."""
+        op: models.V1Operation | None = res.operation
+        if not op:
+            return _TestCaseLibraryGetHandle(name=None)
 
         # progress
         if hasattr(op, "metadata") and op.metadata:
@@ -114,14 +138,8 @@ class TestCaseGenerationHandle:
         else:
             meta_dict = {}
 
-        return TestCaseGenerationHandle(
+        return _TestCaseLibraryGetHandle(
             name=op.name,
-            create_time=op.create_time,
-            creator=op.creator,
-            update_time=op.update_time,
-            updater=op.updater,
-            delete_time=op.delete_time,
-            deleter=op.deleter,
             progress=meta_dict.get("progress"),
             progress_message=meta_dict.get("progressMessage"),
             error=op.error,
@@ -129,6 +147,42 @@ class TestCaseGenerationHandle:
         )
 
 
+@dataclasses.dataclass
+class TestCaseLibraryItem:
+    """Represents a single test case library item - test suite."""
+
+    key: str
+    name: str
+    description: str
+    test_suite_url: str
+    test_count: int
+    test_case_count: int
+    evaluates: List[str]
+    categories: List[str]
+
+    @staticmethod
+    def _from_api_items(
+        api_items: List[models.V1PromptLibraryItem],
+    ) -> List["TestCaseLibraryItem"]:
+        return (
+            [
+                TestCaseLibraryItem(
+                    key=api_item.name or "",
+                    name=api_item.display_name or "",
+                    description=api_item.description or "",
+                    test_suite_url=api_item.test_suite_url or "",
+                    test_count=api_item.test_count or 0,
+                    test_case_count=api_item.test_case_count or 0,
+                    evaluates=list(api_item.evaluates) if api_item.evaluates else [],
+                    categories=list(api_item.categories) if api_item.categories else [],
+                )
+                for api_item in api_items
+            ]
+            if api_items
+            else []
+        )
+
+
 @dataclasses.dataclass
 class TestCase:
     """Represents a single test case, which contains tested prompt, expected answer
@@ -193,6 +247,8 @@ class Test:
     create_time: Optional[datetime.datetime] = None
     update_time: Optional[datetime.datetime] = None
     _client: Optional[api.ApiClient] = None
+    _gen_tc_op_name: Optional[str] = None
+    _lib_tc_op_name: Optional[str] = None
 
     def __post_init__(self):
         if self._client:
@@ -272,12 +328,12 @@ class Test:
         base_llm_model: Optional[str] = None,
         generators: Optional[List[TestCaseGenerator]] = None,
         existing_collection: Optional[str] = None,
-    ) -> "TestCaseGenerationHandle":
+    ) -> None:
         """Generates test cases based on the documents of the Test.
 
         Args:
             count (int): Number of test cases to generate (generator may return fewer
-               prompts).
+                prompts).
             model (str): Model to use for generating the prompts.
             base_llm_model (str): Base LLM model to use for generating the prompts.
             generators (List[TestCaseGenerator]): Methods to use for generation.
@@ -296,28 +352,149 @@ class Test:
 
         res = self._test_api.test_service_generate_test_cases(self.key, req)
 
-        return TestCaseGenerationHandle._from_operation(res)
+        op: models.V1Operation | None = res.operation
+        self._gen_tc_op_name = op.name if op else None
 
     def wait_for_test_case_generation(
-        self,
-        handle: TestCaseGenerationHandle,
-        timeout: Optional[float] = None,
-        verbose: bool = False,
-    ) -> TestCaseGenerationHandle:
+        self, timeout: Optional[float] = None, verbose: bool = False
+    ) -> None:
         """Waits for the test case generation to finish.
 
         Args:
-            handle (TestCaseGenerationHandle): Handle of the test case generation.
             timeout (float): The maximum time to wait in seconds.
             verbose (bool): If True, prints the status of the handle while waiting.
         """
-        if not handle.name:
-            raise ValueError("Test case generation handle is not valid.")
-        elif handle.done:
-            return handle
+        if not self._gen_tc_op_name:
+            raise ValueError(
+                "There is no ongoing test case generation - the operation name is not "
+                "set."
+            )
+
+        if verbose:
+            print(
+                f"Waiting for test case generation to finish ({self._gen_tc_op_name}):"
+            )
+        if self._client:
+            # exponential backoff
+            wait_time = 1.0
+            wait_coef = 1.6
+            wait_max = 8.0
+            wait_total = 0.0
+            timeout = timeout or float(2 * 24 * 60 * 60)  # 2 days
+            progress_bar = utils.ProgressBar()
+            while wait_total < timeout:
+                handle = _TestCaseGenerationHandle._from_operation(
+                    self._operation_api.operation_service_get_operation(
+                        self._gen_tc_op_name
+                    )
+                )
+
+                if verbose:
+                    progress_bar.update(handle.progress or 0, handle.progress_message)
+
+                if handle.done:
+                    if handle.error:
+                        raise RuntimeError(
+                            f"Test case generation failed: {handle.error}"
+                        )
+                    return
+
+                wait_time *= wait_coef
+                time.sleep(min(wait_time, wait_max))
+        else:
+            raise ValueError(
+                "Unable to establish a connection to the Eval Studio host."
+            )
+
+        raise TimeoutError("Waiting timeout has been reached.")
+
+    def list_test_suite_library_items(
+        self,
+        filter_by_categories: Optional[List[str]] = None,
+        filter_by_purposes: Optional[List[str]] = None,
+        filter_by_evaluates: Optional[List[str]] = None,
+        filter_by_origin: Optional[str] = None,
+        filter_by_test_case_count: Optional[int] = None,
+        filter_by_test_count: Optional[int] = None,
+        filter_by_fts: Optional[str] = None,
+    ) -> List[TestCaseLibraryItem]:
+        """Retrieves a list of all available items - suites of tests - in the library.
+
+        Args:
+            filter_by_categories (List[str]): List of categories to filter
+                the library items.
+            filter_by_purposes (List[str]): List of purposes to filter
+                the library items.
+            filter_by_evaluates (List[str]): List of evaluates to filter
+                the library items.
+            filter_by_origin (str): Origin to filter the library items.
+            filter_by_test_case_count (int): Test case count to filter
+                the library items.
+            filter_by_test_count (int): Test count to filter the library items.
+            filter_by_fts (str): FTS to filter the library items - phrase to search for.
+
+        Returns:
+            List[TestCaseLibraryItem]: List of library items.
+        """
+        req = models.TestServiceListTestCaseLibraryItemsRequest(
+            filter_by_categories=filter_by_categories,
+            filter_by_purposes=filter_by_purposes,
+            filter_by_evaluates=filter_by_evaluates,
+            filter_by_origin=filter_by_origin,
+            filter_by_test_case_count=filter_by_test_case_count,
+            filter_by_test_count=filter_by_test_count,
+            filter_by_fts=filter_by_fts,
+        )
+
+        res = self._test_api.test_service_list_test_case_library_items(self.key, req)
+        if res and res.prompt_library_items:
+            return TestCaseLibraryItem._from_api_items(res.prompt_library_items)
+
+        return []
+
+    def add_library_test_cases(
+        self, test_suite_url: str, count: int, test_document_urls: Optional[List[str]]
+    ) -> None:
+        """Sample test cases from the test suite library and add them to the test.
+
+        Args:
+            test_suite_url (str): The URL of the library test suite to get TestCases
+                from (sample).
+            count (int): The number of TestCases to get from the library.
+            test_document_urls (List[str]): The list of target Test corpus
+                document URLs to skip when returning library TestCases corpus.
+        """
+        req = models.TestServiceImportTestCasesFromLibraryRequest(
+            test_suite_url=test_suite_url,
+            count=count,
+            test_document_urls=test_document_urls,
+        )
+
+        res = self._test_api.test_service_import_test_cases_from_library(self.key, req)
+
+        op: models.V1Operation | None = res.operation
+        self._lib_tc_op_name = op.name if op else None
+
+    def wait_for_library_test_case_get(
+        self, timeout: Optional[float] = None, verbose: bool = False
+    ) -> None:
+        """Waits for the library test cases(s) sampling  to finish.
+
+        Args:
+            timeout (float): The maximum time to wait in seconds.
+            verbose (bool): If True, prints the status of the handle while waiting.
+        """
+        if not self._lib_tc_op_name:
+            raise ValueError(
+                "There is no ongoing getting of test case(s) from the library - "
+                "the operation name is not set."
+            )
 
         if verbose:
-            print(f"Waiting for test case generation to finish ({handle.name}):")
+            print(
+                f"Waiting for getting library test case(s) operation to finish "
+                f"({self._lib_tc_op_name}):"
+            )
         if self._client:
             # exponential backoff
             wait_time = 1.0
@@ -325,37 +502,30 @@ class Test:
             wait_max = 8.0
             wait_total = 0.0
             timeout = timeout or float(2 * 24 * 60 * 60)  # 2 days
-            # progress
-            p_max = 1.0
-            p_msg = ""
+            progress_bar = utils.ProgressBar()
             while wait_total < timeout:
-                handle = TestCaseGenerationHandle._from_operation(
-                    self._operation_api.operation_service_get_operation(handle.name)
+                handle = _TestCaseLibraryGetHandle._from_operation(
+                    self._operation_api.operation_service_get_operation(
+                        self._lib_tc_op_name
+                    )
                 )
 
                 if verbose:
-                    print(" " * len(p_msg), end="\r")
-                    if handle.progress or handle.progress_message:
-                        try:
-                            h_progress = float(str(handle.progress))
-                        except ValueError:
-                            h_progress = 0.0
-                        h_msg = handle.progress_message or "Processing"
-                    else:
-                        h_progress = 0.0
-                        h_msg = "Initializing"
-                    p_progress = int(h_progress / p_max * 100)
-                    p_hashes = p_progress // 5
-                    p_msg = f"  {p_progress:>3}% |{'#' * p_hashes:<20}| {h_msg}"
-                    print(p_msg, end="\r")
+                    progress_bar.update(handle.progress or 0, handle.progress_message)
 
                 if handle.done:
-                    return handle
+                    if handle.error:
+                        raise RuntimeError(
+                            f"Getting of library test case(s) failed: {handle.error}"
+                        )
+                    return
 
                 wait_time *= wait_coef
                 time.sleep(min(wait_time, wait_max))
         else:
-            raise ValueError("Cannot establish connection to Eval Studio host.")
+            raise ValueError(
+                "Unable to establish a connection to the Eval Studio host."
+            )
 
         raise TimeoutError("Waiting timeout has been reached.")
 

eval_studio_client/utils.py

@@ -0,0 +1,26 @@
+from typing import Optional
+
+
+class ProgressBar:
+    def __init__(self):
+        self.progress = 0.0
+        self.progress_message = "Initializing"
+        self._progress_max = 1.0
+
+    def update(self, progress: float, message: Optional[str] = None):
+        try:
+            self.progress = float(str(progress))
+        except ValueError:
+            self.progress = 0.0
+
+        if message:
+            self.progress_message = message or ""
+
+        self.print()
+
+    def print(self):
+        print(" " * len(self.progress_message), end="\r")
+        p_progress = int(self.progress / self._progress_max * 100)
+        p_hashes = p_progress // 5
+        p_msg = f"  {p_progress:>3}% |{'#' * p_hashes:<20}| {self.progress_message}"
+        print(p_msg, end="\r")

{eval_studio_client-1.0.0a1.dist-info → eval_studio_client-1.0.1.dist-info}/METADATA

@@ -1,10 +1,9 @@
 Metadata-Version: 2.3
 Name: eval-studio-client
-Version: 1.0.0a1
+Version: 1.0.1
 Project-URL: Source, https://github.com/h2oai/eval-studio/tree/main/client-py/src/
 Project-URL: Issues, https://github.com/h2oai/eval-studio/issues
 Author-email: "H2O.ai" <support@h2o.ai>
-License: MIT
 Classifier: Development Status :: 4 - Beta
 Classifier: Programming Language :: Python
 Classifier: Programming Language :: Python :: 3.9