exa-py 1.13.0__py3-none-any.whl → 1.13.2__py3-none-any.whl

exa_py/api.py

@@ -38,6 +38,7 @@ from exa_py.utils import (
 )
 from .websets import WebsetsClient
 from .websets.core.base import ExaJSONEncoder
+from .research.client import ResearchClient, AsyncResearchClient
 
 is_beta = os.getenv("IS_BETA") == "True"
 
@@ -56,7 +57,7 @@ def snake_to_camel(snake_str: str) -> str:
         return "$schema"
     if snake_str == "not_":
         return "not"
-        
+
     components = snake_str.split("_")
     return components[0] + "".join(x.title() for x in components[1:])
 
@@ -261,6 +262,7 @@ class JSONSchema(TypedDict, total=False):
     """Represents a JSON Schema definition used for structured summary output.
     To learn more visit https://json-schema.org/overview/what-is-jsonschema.
     """
+
     schema_: str  # This will be converted to "$schema" in JSON
     title: str
     description: str
@@ -288,7 +290,7 @@ class SummaryContentsOptions(TypedDict, total=False):
 
     query: str
     schema: JSONSchema
-    
+
 
 class ExtrasOptions(TypedDict, total=False):
     """A class representing additional extraction fields (e.g. links, images)"""
@@ -669,7 +671,7 @@ class AnswerResponse:
         citations (List[AnswerResult]): A list of citations used to generate the answer.
     """
 
-    answer: str
+    answer: Union[str, dict[str, Any]]
     citations: List[AnswerResult]
 
     def __str__(self):
@@ -765,9 +767,9 @@ class AsyncStreamAnswerResponse:
                         content = chunk["choices"][0]["delta"].get("content")
 
                 if (
-                        "citations" in chunk
-                        and chunk["citations"]
-                        and chunk["citations"] != "null"
+                    "citations" in chunk
+                    and chunk["citations"]
+                    and chunk["citations"] != "null"
                 ):
                     citations = [
                         AnswerResult(**to_snake_case(s)) for s in chunk["citations"]
@@ -776,6 +778,7 @@ class AsyncStreamAnswerResponse:
                 stream_chunk = StreamChunk(content=content, citations=citations)
                 if stream_chunk.has_data():
                     yield stream_chunk
+
         return generator()
 
     def close(self) -> None:
@@ -834,36 +837,6 @@ def nest_fields(original_dict: Dict, fields_to_nest: List[str], new_key: str):
 
     return original_dict
 
-@dataclass
-class ResearchTaskResponse:
-    """A class representing the response for a research task.
-
-    Attributes:
-        id (str): The unique identifier for the research request.
-        status (str): Status of the research request.
-        output (Optional[Dict[str, Any]]): The answer structured as JSON, if available.
-        citations (Optional[Dict[str, List[_Result]]]): List of citations used to generate the answer, grouped by root field in the output schema.
-    """
-
-    id: str
-    status: str
-    output: Optional[Dict[str, Any]]
-    citations: Dict[str, List[_Result]]
-
-    def __str__(self):
-        output_repr = (
-            json.dumps(self.output, indent=2, ensure_ascii=False)
-            if self.output is not None
-            else "None"
-        )
-        citations_str = "\n\n".join(str(src) for src in self.citations)
-        return (
-            f"ID: {self.id}\n"
-            f"Status: {self.status}\n"
-            f"Output: {output_repr}\n\n"
-            f"Citations:\n{citations_str}"
-        )
-
 
 class Exa:
     """A client for interacting with Exa API."""
@@ -872,7 +845,7 @@ class Exa:
         self,
         api_key: Optional[str],
         base_url: str = "https://api.exa.ai",
-        user_agent: str = "exa-py 1.12.1",
+        user_agent: str = "exa-py 1.12.4",
     ):
         """Initialize the Exa client with the provided API key and optional base URL and user agent.
 
@@ -889,10 +862,22 @@ class Exa:
                     "API key must be provided as an argument or in EXA_API_KEY environment variable"
                 )
         self.base_url = base_url
-        self.headers = {"x-api-key": api_key, "User-Agent": user_agent, "Content-Type": "application/json"}
+        self.headers = {
+            "x-api-key": api_key,
+            "User-Agent": user_agent,
+            "Content-Type": "application/json",
+        }
         self.websets = WebsetsClient(self)
+        # Research tasks client (new, mirrors Websets design)
+        self.research = ResearchClient(self)
 
-    def request(self, endpoint: str, data: Optional[Union[Dict[str, Any], str]] = None, method: str = "POST", params: Optional[Dict[str, Any]] = None) -> Union[Dict[str, Any], requests.Response]:
+    def request(
+        self,
+        endpoint: str,
+        data: Optional[Union[Dict[str, Any], str]] = None,
+        method: str = "POST",
+        params: Optional[Dict[str, Any]] = None,
+    ) -> Union[Dict[str, Any], requests.Response]:
         """Send a request to the Exa API, optionally streaming if data['stream'] is True.
 
         Args:
@@ -915,13 +900,13 @@ class Exa:
         else:
             # Otherwise, serialize the dictionary to JSON if it exists
             json_data = json.dumps(data, cls=ExaJSONEncoder) if data else None
-            
+
         if data and data.get("stream"):
             res = requests.post(
-                self.base_url + endpoint, 
+                self.base_url + endpoint,
                 data=json_data,
-                headers=self.headers, 
-                stream=True
+                headers=self.headers,
+                stream=True,
             )
             return res
 
@@ -931,20 +916,14 @@ class Exa:
             )
         elif method.upper() == "POST":
             res = requests.post(
-                self.base_url + endpoint, 
-                data=json_data,
-                headers=self.headers
+                self.base_url + endpoint, data=json_data, headers=self.headers
             )
         elif method.upper() == "PATCH":
             res = requests.patch(
-                self.base_url + endpoint, 
-                data=json_data,
-                headers=self.headers
+                self.base_url + endpoint, data=json_data, headers=self.headers
             )
         elif method.upper() == "DELETE":
-            res = requests.delete(
-                self.base_url + endpoint, headers=self.headers
-            )
+            res = requests.delete(self.base_url + endpoint, headers=self.headers)
         else:
             raise ValueError(f"Unsupported HTTP method: {method}")
 
@@ -1875,6 +1854,7 @@ class Exa:
         text: Optional[bool] = False,
         system_prompt: Optional[str] = None,
         model: Optional[Literal["exa", "exa-pro"]] = None,
+        output_schema: Optional[dict[str, Any]] = None,
     ) -> Union[AnswerResponse, StreamAnswerResponse]: ...
 
     def answer(
@@ -1885,6 +1865,7 @@ class Exa:
         text: Optional[bool] = False,
         system_prompt: Optional[str] = None,
         model: Optional[Literal["exa", "exa-pro"]] = None,
+        output_schema: Optional[dict[str, Any]] = None,
     ) -> Union[AnswerResponse, StreamAnswerResponse]:
         """Generate an answer to a query using Exa's search and LLM capabilities.
 
@@ -1893,6 +1874,7 @@ class Exa:
             text (bool, optional): Whether to include full text in the results. Defaults to False.
             system_prompt (str, optional): A system prompt to guide the LLM's behavior when generating the answer.
             model (str, optional): The model to use for answering. Either "exa" or "exa-pro". Defaults to None.
+            output_schema (dict[str, Any], optional): JSON schema describing the desired answer structure.
 
         Returns:
             AnswerResponse: An object containing the answer and citations.
@@ -1922,6 +1904,7 @@ class Exa:
         text: bool = False,
         system_prompt: Optional[str] = None,
         model: Optional[Literal["exa", "exa-pro"]] = None,
+        output_schema: Optional[dict[str, Any]] = None,
     ) -> StreamAnswerResponse:
         """Generate a streaming answer response.
 
@@ -1930,7 +1913,7 @@ class Exa:
             text (bool): Whether to include full text in the results. Defaults to False.
             system_prompt (str, optional): A system prompt to guide the LLM's behavior when generating the answer.
             model (str, optional): The model to use for answering. Either "exa" or "exa-pro". Defaults to None.
-
+            output_schema (dict[str, Any], optional): JSON schema describing the desired answer structure.
         Returns:
             StreamAnswerResponse: An object that can be iterated over to retrieve (partial text, partial citations).
                 Each iteration yields a tuple of (Optional[str], Optional[List[AnswerResult]]).
@@ -1941,40 +1924,12 @@ class Exa:
         raw_response = self.request("/answer", options)
         return StreamAnswerResponse(raw_response)
 
-    def researchTask(
-        self,
-        *,
-        input_instructions: str,
-        output_schema: Dict[str, Any],
-    ) -> ResearchTaskResponse:
-        """Submit a research request to Exa.
-
-        Args:
-            input_instructions (str): The instructions for the research task.
-            output_schema (Dict[str, Any]): JSON schema describing the desired answer structure.
-        """
-        # Build the request payload expected by the Exa API
-        options = {
-            "input": {"instructions": input_instructions},
-            "output": {"schema": output_schema},
-        }
-
-        response = self.request("/research/tasks", options)
-
-        return ResearchTaskResponse(
-            id=response["id"],
-            status=response["status"],
-            output=response.get("output"),
-            citations={
-                key: [_Result(**to_snake_case(citation)) for citation in citations_list]
-                for key, citations_list in response.get("citations", {}).items()
-            },
-        )
-
 
 class AsyncExa(Exa):
     def __init__(self, api_key: str, api_base: str = "https://api.exa.ai"):
         super().__init__(api_key, api_base)
+        # Override the synchronous ResearchClient with its async counterpart.
+        self.research = AsyncResearchClient(self)
         self._client = None
 
     @property
@@ -1982,9 +1937,7 @@ class AsyncExa(Exa):
         # this may only be a
         if self._client is None:
             self._client = httpx.AsyncClient(
-                base_url=self.base_url,
-                headers=self.headers,
-                timeout=60
+                base_url=self.base_url, headers=self.headers, timeout=60
             )
         return self._client
 
@@ -2004,15 +1957,14 @@ class AsyncExa(Exa):
         """
         if data.get("stream"):
             request = httpx.Request(
-                'POST',
-                self.base_url + endpoint,
-                json=data,
-                headers=self.headers
+                "POST", self.base_url + endpoint, json=data, headers=self.headers
             )
             res = await self.client.send(request, stream=True)
             return res
 
-        res = await self.client.post(self.base_url + endpoint, json=data, headers=self.headers)
+        res = await self.client.post(
+            self.base_url + endpoint, json=data, headers=self.headers
+        )
         if res.status_code != 200:
             raise ValueError(
                 f"Request failed with status code {res.status_code}: {res.text}"
@@ -2250,6 +2202,7 @@ class AsyncExa(Exa):
         text: Optional[bool] = False,
         system_prompt: Optional[str] = None,
         model: Optional[Literal["exa", "exa-pro"]] = None,
+        output_schema: Optional[dict[str, Any]] = None,
     ) -> Union[AnswerResponse, StreamAnswerResponse]:
         """Generate an answer to a query using Exa's search and LLM capabilities.
 
@@ -2258,6 +2211,7 @@ class AsyncExa(Exa):
             text (bool, optional): Whether to include full text in the results. Defaults to False.
             system_prompt (str, optional): A system prompt to guide the LLM's behavior when generating the answer.
             model (str, optional): The model to use for answering. Either "exa" or "exa-pro". Defaults to None.
+            output_schema (dict[str, Any], optional): JSON schema describing the desired answer structure.
 
         Returns:
             AnswerResponse: An object containing the answer and citations.
@@ -2287,6 +2241,7 @@ class AsyncExa(Exa):
         text: bool = False,
         system_prompt: Optional[str] = None,
         model: Optional[Literal["exa", "exa-pro"]] = None,
+        output_schema: Optional[dict[str, Any]] = None,
     ) -> AsyncStreamAnswerResponse:
         """Generate a streaming answer response.
 
@@ -2295,7 +2250,7 @@ class AsyncExa(Exa):
             text (bool): Whether to include full text in the results. Defaults to False.
             system_prompt (str, optional): A system prompt to guide the LLM's behavior when generating the answer.
             model (str, optional): The model to use for answering. Either "exa" or "exa-pro". Defaults to None.
-
+            output_schema (dict[str, Any], optional): JSON schema describing the desired answer structure.
         Returns:
             AsyncStreamAnswerResponse: An object that can be iterated over to retrieve (partial text, partial citations).
                 Each iteration yields a tuple of (Optional[str], Optional[List[AnswerResult]]).
@@ -2305,36 +2260,3 @@ class AsyncExa(Exa):
         options["stream"] = True
         raw_response = await self.async_request("/answer", options)
         return AsyncStreamAnswerResponse(raw_response)
-
-    async def researchTask(
-        self,
-        *,
-        input_instructions: str,
-        output_schema: Dict[str, Any],
-    ) -> ResearchTaskResponse:
-        """Asynchronously submit a research request to Exa.
-
-        Args:
-            input_instructions (str): The instructions for the research task.
-            output_schema (Dict[str, Any]): JSON schema describing the desired answer structure.
-
-        Returns:
-            ResearchTaskResponse: The parsed response from the Exa API.
-        """
-        # Build the request payload expected by the Exa API
-        options = {
-            "input": {"instructions": input_instructions},
-            "output": {"schema": output_schema},
-        }
-
-        response = await self.async_request("/research/tasks", options)
-
-        return ResearchTaskResponse(
-            id=response["id"],
-            status=response["status"],
-            output=response.get("output"),
-            citations={
-                key: [_Result(**to_snake_case(citation)) for citation in citations_list]
-                for key, citations_list in response.get("citations", {}).items()
-            },
-        )

exa_py/research/__init__.py

@@ -0,0 +1,9 @@
+from .client import ResearchClient, AsyncResearchClient
+from .models import ResearchTask
+
+__all__ = [
+    "ResearchClient",
+    "AsyncResearchClient",
+    "ResearchTaskId",
+    "ResearchTask",
+]

exa_py/research/client.py

@@ -0,0 +1,232 @@
+"""Lightweight research client wrappers for the Exa REST API.
+
+This module purposefully keeps its import surface minimal to avoid circular
+import problems with :pymod:`exa_py.api`.  Any heavy dependencies (including
+`exa_py.api` itself) are imported lazily **inside** functions.  This means
+that type-checkers still see the full, precise types via the ``TYPE_CHECKING``
+block, but at runtime we only pay the cost if/when a helper is actually used.
+"""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any, Dict
+
+if TYPE_CHECKING:  # pragma: no cover – only for static analysers
+    # Import with full type info when static type-checking.  `_Result` still
+    # lives in ``exa_py.api`` but the response model moved to
+    # ``exa_py.research.models``.
+    from ..api import _Result  # noqa: F401
+    from .models import ResearchTask, ResearchTaskId  # noqa: F401
+
+# ---------------------------------------------------------------------------
+# Public, user-facing clients
+# ---------------------------------------------------------------------------
+
+
+class ResearchClient:
+    """Synchronous helper namespace accessed via :pyattr:`Exa.research`."""
+
+    def __init__(self, parent_client):
+        # A reference to the *already-constructed* ``Exa`` instance so that we
+        # can piggy-back on its HTTP plumbing (headers, base URL, retries, …).
+        self._client = parent_client
+
+    def create_task(
+        self,
+        *,
+        input_instructions: str,
+        output_schema: Dict[str, Any],
+    ) -> "ResearchTaskId":
+        """Submit a research request and return the *task identifier*."""
+        payload = {
+            "input": {"instructions": input_instructions},
+            "output": {"schema": output_schema},
+        }
+
+        raw_response: Dict[str, Any] = self._client.request("/research/tasks", payload)
+
+        # Defensive checks so that we fail loudly if the contract changes.
+        if not isinstance(raw_response, dict) or "id" not in raw_response:
+            raise RuntimeError(
+                f"Unexpected response while creating research task: {raw_response}"
+            )
+
+        # Lazily import to avoid circular deps at runtime.
+        from .models import ResearchTaskId  # noqa: WPS433 – runtime import
+
+        return ResearchTaskId(id=raw_response["id"])
+
+    def get_task(
+        self, id: str
+    ) -> "ResearchTask":  # noqa: D401 – imperative mood is fine
+        """Fetch the current status / result for a research task."""
+        endpoint = f"/research/tasks/{id}"
+
+        # The new endpoint is a simple GET.
+        raw_response: Dict[str, Any] = self._client.request(endpoint, method="GET")
+
+        return _build_research_task(raw_response)
+
+    # ------------------------------------------------------------------
+    # Convenience helpers
+    # ------------------------------------------------------------------
+
+    def poll_task(
+        self,
+        id: str,
+        *,
+        poll_interval: float = 1.0,
+        timeout_seconds: int = 15 * 60,
+    ) -> "ResearchTask":
+        """Blocking helper that polls until task completes or fails.
+
+        Parameters
+        ----------
+        id:
+            The ID of the research task to poll.
+        poll_interval:
+            Seconds to wait between successive polls (default 1s).
+        timeout_seconds:
+            Maximum time to wait before raising :class:`TimeoutError` (default 15 min).
+        """
+
+        import time
+
+        deadline = time.monotonic() + timeout_seconds
+
+        while True:
+            task = self.get_task(id)
+            status = task.status.lower() if isinstance(task.status, str) else ""
+
+            if status in {"completed", "failed", "complete", "finished", "done"}:
+                return task
+
+            if time.monotonic() > deadline:
+                raise TimeoutError(
+                    f"Research task {id} did not finish within {timeout_seconds} seconds"
+                )
+
+            time.sleep(poll_interval)
+
+
+class AsyncResearchClient:
+    """Async counterpart used via :pyattr:`AsyncExa.research`."""
+
+    def __init__(self, parent_client):
+        self._client = parent_client
+
+    async def create_task(
+        self,
+        *,
+        input_instructions: str,
+        output_schema: Dict[str, Any],
+    ) -> "ResearchTaskId":
+        """Submit a research request and return the *task identifier* (async)."""
+
+        payload = {
+            "input": {"instructions": input_instructions},
+            "output": {"schema": output_schema},
+        }
+
+        raw_response: Dict[str, Any] = await self._client.async_request(
+            "/research/tasks", payload
+        )
+
+        # Defensive checks so that we fail loudly if the contract changes.
+        if not isinstance(raw_response, dict) or "id" not in raw_response:
+            raise RuntimeError(
+                f"Unexpected response while creating research task: {raw_response}"
+            )
+
+        # Lazily import to avoid circular deps at runtime.
+        from .models import ResearchTaskId  # noqa: WPS433 – runtime import
+
+        return ResearchTaskId(id=raw_response["id"])
+
+    async def get_task(self, id: str) -> "ResearchTask":  # noqa: D401
+        """Fetch the current status / result for a research task (async)."""
+
+        endpoint = f"/research/tasks/{id}"
+
+        # Perform GET using the underlying HTTP client because `async_request`
+        # only supports POST semantics.
+        resp = await self._client.client.get(
+            self._client.base_url + endpoint, headers=self._client.headers
+        )
+
+        if resp.status_code >= 400:
+            raise RuntimeError(
+                f"Request failed with status code {resp.status_code}: {resp.text}"
+            )
+
+        raw_response: Dict[str, Any] = resp.json()
+
+        return _build_research_task(raw_response)
+
+    # ------------------------------------------------------------------
+    # Convenience helpers
+    # ------------------------------------------------------------------
+
+    async def poll_task(
+        self,
+        id: str,
+        *,
+        poll_interval: float = 1.0,
+        timeout_seconds: int = 15 * 60,
+    ) -> "ResearchTask":
+        """Async helper that polls until task completes or fails.
+
+        Mirrors :py:meth:`ResearchClient.poll_task` but uses ``await`` and
+        :pyfunc:`asyncio.sleep`.  Raises :class:`TimeoutError` on timeout.
+        """
+
+        import asyncio
+        import time
+
+        deadline = time.monotonic() + timeout_seconds
+
+        while True:
+            task = await self.get_task(id)
+            status = task.status.lower() if isinstance(task.status, str) else ""
+
+            if status in {"completed", "failed", "complete", "finished", "done"}:
+                return task
+
+            if time.monotonic() > deadline:
+                raise TimeoutError(
+                    f"Research task {id} did not finish within {timeout_seconds} seconds"
+                )
+
+            await asyncio.sleep(poll_interval)
+
+
+# ---------------------------------------------------------------------------
+# Internal helpers (lazy imports to avoid cycles)
+# ---------------------------------------------------------------------------
+
+
+def _build_research_task(raw: Dict[str, Any]):
+    """Convert raw API response into a :class:`ResearchTask` instance."""
+
+    # Defensive check – fail loudly if the API contract changes.
+    if not isinstance(raw, dict) or "id" not in raw:
+        raise RuntimeError(f"Unexpected response while fetching research task: {raw}")
+
+    # Lazily import heavy deps to avoid cycles and unnecessary startup cost.
+    from .models import ResearchTask  # noqa: WPS433 – runtime import
+    from ..api import _Result, to_snake_case  # noqa: WPS433 – runtime import
+
+    citations_raw = raw.get("citations", {}) or {}
+    citations_parsed = {
+        key: [_Result(**to_snake_case(c)) for c in cites]
+        for key, cites in citations_raw.items()
+    }
+
+    return ResearchTask(
+        id=raw["id"],
+        status=raw["status"],
+        instructions=raw.get("instructions", ""),
+        schema=raw.get("schema", {}),
+        data=raw.get("data"),
+        citations=citations_parsed,
+    )

exa_py/research/models.py

@@ -0,0 +1,98 @@
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass
+from typing import Any, Dict, List, Optional
+
+# Local import placed inside TYPE_CHECKING block to avoid runtime cycles.
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:  # pragma: no cover – for static analysers only
+    from ..api import _Result  # noqa: F401
+
+
+@dataclass
+class ResearchTaskId:
+    """Structured research task ID.
+
+    Attributes
+    ----------
+    id:
+        Unique identifier for the research task.
+    """
+
+    id: str
+
+    # ---------------------------------------------------------------------
+    # Pretty representation helpers
+    # ---------------------------------------------------------------------
+    def __str__(self) -> str:  # pragma: no cover – convenience only
+        return f"ID: {self.id}\n"
+
+
+@dataclass
+class ResearchTask:
+    """Structured research task.
+
+    Attributes
+    ----------
+    id:
+        Unique identifier for the research task.
+    status:
+        Current task status
+    instructions:
+        Instructions for the task
+    schema:
+        Output schema defining the task
+    data:
+        JSON-serialisable answer generated by Exa (may be ``None`` until the task
+        completes).
+    citations:
+        Mapping from *root field* in the output schema to the list of search
+        results that were used to generate that part of the answer.
+    """
+
+    id: str
+    status: str
+    instructions: str
+    schema: Dict[str, Any]
+    data: Optional[Dict[str, Any]]
+    citations: Dict[str, List["_Result"]]
+
+    # ---------------------------------------------------------------------
+    # Pretty representation helpers
+    # ---------------------------------------------------------------------
+    def __str__(self) -> str:  # pragma: no cover – convenience only
+        """Human-readable representation including *all* relevant fields."""
+        schema_repr = json.dumps(self.schema, indent=2, ensure_ascii=False)
+        data_repr = (
+            json.dumps(self.data, indent=2, ensure_ascii=False)
+            if self.data is not None
+            else "None"
+        )
+
+        # Render citations grouped by the root field they belong to.
+        if self.citations:
+            # Each key is a root field, each value is a list of _Result objects.
+            citations_lines = []
+            for field, sources in self.citations.items():
+                rendered_sources = "\n    ".join(str(src) for src in sources)
+                citations_lines.append(f"{field}:\n    {rendered_sources}")
+            citations_str = "\n\n".join(citations_lines)
+        else:
+            citations_str = "None"
+
+        return (
+            f"ID: {self.id}\n"
+            f"Status: {self.status}\n"
+            f"Instructions: {self.instructions}\n"
+            f"Schema:\n{schema_repr}\n"
+            f"Data:\n{data_repr}\n\n"
+            f"Citations:\n{citations_str}"
+        )
+
+
+__all__ = [
+    "ResearchTaskId",
+    "ResearchTask",
+]