llama-cloud 0.1.14__py3-none-any.whl → 0.1.16__py3-none-any.whl

llama_cloud/resources/retrievers/client.py

@@ -346,6 +346,68 @@ class RetrieversClient:
             raise ApiError(status_code=_response.status_code, body=_response.text)
         raise ApiError(status_code=_response.status_code, body=_response_json)
 
+    def direct_retrieve(
+        self,
+        *,
+        project_id: typing.Optional[str] = None,
+        organization_id: typing.Optional[str] = None,
+        mode: typing.Optional[CompositeRetrievalMode] = OMIT,
+        rerank_top_n: typing.Optional[int] = OMIT,
+        query: str,
+        pipelines: typing.Optional[typing.List[RetrieverPipeline]] = OMIT,
+    ) -> CompositeRetrievalResult:
+        """
+        Retrieve data using specified pipelines without creating a persistent retriever.
+
+        Parameters:
+            - project_id: typing.Optional[str].
+
+            - organization_id: typing.Optional[str].
+
+            - mode: typing.Optional[CompositeRetrievalMode]. The mode of composite retrieval.
+
+            - rerank_top_n: typing.Optional[int]. The number of nodes to retrieve after reranking over retrieved nodes from all retrieval tools.
+
+            - query: str. The query to retrieve against.
+
+            - pipelines: typing.Optional[typing.List[RetrieverPipeline]]. The pipelines to use for retrieval.
+        ---
+        from llama_cloud import CompositeRetrievalMode
+        from llama_cloud.client import LlamaCloud
+
+        client = LlamaCloud(
+            token="YOUR_TOKEN",
+        )
+        client.retrievers.direct_retrieve(
+            mode=CompositeRetrievalMode.ROUTING,
+            query="string",
+        )
+        """
+        _request: typing.Dict[str, typing.Any] = {"query": query}
+        if mode is not OMIT:
+            _request["mode"] = mode
+        if rerank_top_n is not OMIT:
+            _request["rerank_top_n"] = rerank_top_n
+        if pipelines is not OMIT:
+            _request["pipelines"] = pipelines
+        _response = self._client_wrapper.httpx_client.request(
+            "POST",
+            urllib.parse.urljoin(f"{self._client_wrapper.get_base_url()}/", "api/v1/retrievers/retrieve"),
+            params=remove_none_from_dict({"project_id": project_id, "organization_id": organization_id}),
+            json=jsonable_encoder(_request),
+            headers=self._client_wrapper.get_headers(),
+            timeout=60,
+        )
+        if 200 <= _response.status_code < 300:
+            return pydantic.parse_obj_as(CompositeRetrievalResult, _response.json())  # type: ignore
+        if _response.status_code == 422:
+            raise UnprocessableEntityError(pydantic.parse_obj_as(HttpValidationError, _response.json()))  # type: ignore
+        try:
+            _response_json = _response.json()
+        except JSONDecodeError:
+            raise ApiError(status_code=_response.status_code, body=_response.text)
+        raise ApiError(status_code=_response.status_code, body=_response_json)
+
 
 class AsyncRetrieversClient:
     def __init__(self, *, client_wrapper: AsyncClientWrapper):
@@ -664,3 +726,65 @@ class AsyncRetrieversClient:
         except JSONDecodeError:
             raise ApiError(status_code=_response.status_code, body=_response.text)
         raise ApiError(status_code=_response.status_code, body=_response_json)
+
+    async def direct_retrieve(
+        self,
+        *,
+        project_id: typing.Optional[str] = None,
+        organization_id: typing.Optional[str] = None,
+        mode: typing.Optional[CompositeRetrievalMode] = OMIT,
+        rerank_top_n: typing.Optional[int] = OMIT,
+        query: str,
+        pipelines: typing.Optional[typing.List[RetrieverPipeline]] = OMIT,
+    ) -> CompositeRetrievalResult:
+        """
+        Retrieve data using specified pipelines without creating a persistent retriever.
+
+        Parameters:
+            - project_id: typing.Optional[str].
+
+            - organization_id: typing.Optional[str].
+
+            - mode: typing.Optional[CompositeRetrievalMode]. The mode of composite retrieval.
+
+            - rerank_top_n: typing.Optional[int]. The number of nodes to retrieve after reranking over retrieved nodes from all retrieval tools.
+
+            - query: str. The query to retrieve against.
+
+            - pipelines: typing.Optional[typing.List[RetrieverPipeline]]. The pipelines to use for retrieval.
+        ---
+        from llama_cloud import CompositeRetrievalMode
+        from llama_cloud.client import AsyncLlamaCloud
+
+        client = AsyncLlamaCloud(
+            token="YOUR_TOKEN",
+        )
+        await client.retrievers.direct_retrieve(
+            mode=CompositeRetrievalMode.ROUTING,
+            query="string",
+        )
+        """
+        _request: typing.Dict[str, typing.Any] = {"query": query}
+        if mode is not OMIT:
+            _request["mode"] = mode
+        if rerank_top_n is not OMIT:
+            _request["rerank_top_n"] = rerank_top_n
+        if pipelines is not OMIT:
+            _request["pipelines"] = pipelines
+        _response = await self._client_wrapper.httpx_client.request(
+            "POST",
+            urllib.parse.urljoin(f"{self._client_wrapper.get_base_url()}/", "api/v1/retrievers/retrieve"),
+            params=remove_none_from_dict({"project_id": project_id, "organization_id": organization_id}),
+            json=jsonable_encoder(_request),
+            headers=self._client_wrapper.get_headers(),
+            timeout=60,
+        )
+        if 200 <= _response.status_code < 300:
+            return pydantic.parse_obj_as(CompositeRetrievalResult, _response.json())  # type: ignore
+        if _response.status_code == 422:
+            raise UnprocessableEntityError(pydantic.parse_obj_as(HttpValidationError, _response.json()))  # type: ignore
+        try:
+            _response_json = _response.json()
+        except JSONDecodeError:
+            raise ApiError(status_code=_response.status_code, body=_response.text)
+        raise ApiError(status_code=_response.status_code, body=_response_json)

llama_cloud/types/__init__.py

@@ -103,15 +103,7 @@ from .embedding_model_config_update_embedding_config import (
     EmbeddingModelConfigUpdateEmbeddingConfig_OpenaiEmbedding,
     EmbeddingModelConfigUpdateEmbeddingConfig_VertexaiEmbedding,
 )
-from .eval_dataset import EvalDataset
-from .eval_dataset_job_params import EvalDatasetJobParams
-from .eval_dataset_job_record import EvalDatasetJobRecord
 from .eval_execution_params import EvalExecutionParams
-from .eval_execution_params_override import EvalExecutionParamsOverride
-from .eval_metric import EvalMetric
-from .eval_question import EvalQuestion
-from .eval_question_create import EvalQuestionCreate
-from .eval_question_result import EvalQuestionResult
 from .extract_agent import ExtractAgent
 from .extract_agent_create import ExtractAgentCreate
 from .extract_agent_create_data_schema import ExtractAgentCreateDataSchema
@@ -123,6 +115,9 @@ from .extract_agent_update_data_schema_zero_value import ExtractAgentUpdateDataS
 from .extract_config import ExtractConfig
 from .extract_job import ExtractJob
 from .extract_job_create import ExtractJobCreate
+from .extract_job_create_batch import ExtractJobCreateBatch
+from .extract_job_create_batch_data_schema_override import ExtractJobCreateBatchDataSchemaOverride
+from .extract_job_create_batch_data_schema_override_zero_value import ExtractJobCreateBatchDataSchemaOverrideZeroValue
 from .extract_job_create_data_schema_override import ExtractJobCreateDataSchemaOverride
 from .extract_job_create_data_schema_override_zero_value import ExtractJobCreateDataSchemaOverrideZeroValue
 from .extract_mode import ExtractMode
@@ -175,9 +170,6 @@ from .llama_parse_supported_file_extensions import LlamaParseSupportedFileExtens
 from .llm import Llm
 from .llm_model_data import LlmModelData
 from .llm_parameters import LlmParameters
-from .local_eval import LocalEval
-from .local_eval_results import LocalEvalResults
-from .local_eval_sets import LocalEvalSets
 from .managed_ingestion_status import ManagedIngestionStatus
 from .managed_ingestion_status_response import ManagedIngestionStatusResponse
 from .markdown_element_node_parser import MarkdownElementNodeParser
@@ -188,7 +180,6 @@ from .metadata_filter import MetadataFilter
 from .metadata_filter_value import MetadataFilterValue
 from .metadata_filters import MetadataFilters
 from .metadata_filters_filters_item import MetadataFiltersFiltersItem
-from .metric_result import MetricResult
 from .node_parser import NodeParser
 from .node_relationship import NodeRelationship
 from .none_chunking_config import NoneChunkingConfig
@@ -203,6 +194,7 @@ from .page_screenshot_metadata import PageScreenshotMetadata
 from .page_screenshot_node_with_score import PageScreenshotNodeWithScore
 from .page_segmentation_config import PageSegmentationConfig
 from .page_splitter_node_parser import PageSplitterNodeParser
+from .paginated_extract_runs_response import PaginatedExtractRunsResponse
 from .paginated_jobs_history_with_metrics import PaginatedJobsHistoryWithMetrics
 from .paginated_list_cloud_documents_response import PaginatedListCloudDocumentsResponse
 from .paginated_list_pipeline_files_response import PaginatedListPipelineFilesResponse
@@ -272,8 +264,6 @@ from .progress_event_status import ProgressEventStatus
 from .project import Project
 from .project_create import ProjectCreate
 from .prompt_conf import PromptConf
-from .prompt_mixin_prompts import PromptMixinPrompts
-from .prompt_spec import PromptSpec
 from .pydantic_program_mode import PydanticProgramMode
 from .recurring_credit_grant import RecurringCreditGrant
 from .related_node_info import RelatedNodeInfo
@@ -431,15 +421,7 @@ __all__ = [
     "EmbeddingModelConfigUpdateEmbeddingConfig_HuggingfaceApiEmbedding",
     "EmbeddingModelConfigUpdateEmbeddingConfig_OpenaiEmbedding",
     "EmbeddingModelConfigUpdateEmbeddingConfig_VertexaiEmbedding",
-    "EvalDataset",
-    "EvalDatasetJobParams",
-    "EvalDatasetJobRecord",
     "EvalExecutionParams",
-    "EvalExecutionParamsOverride",
-    "EvalMetric",
-    "EvalQuestion",
-    "EvalQuestionCreate",
-    "EvalQuestionResult",
     "ExtractAgent",
     "ExtractAgentCreate",
     "ExtractAgentCreateDataSchema",
@@ -451,6 +433,9 @@ __all__ = [
     "ExtractConfig",
     "ExtractJob",
     "ExtractJobCreate",
+    "ExtractJobCreateBatch",
+    "ExtractJobCreateBatchDataSchemaOverride",
+    "ExtractJobCreateBatchDataSchemaOverrideZeroValue",
     "ExtractJobCreateDataSchemaOverride",
     "ExtractJobCreateDataSchemaOverrideZeroValue",
     "ExtractMode",
@@ -501,9 +486,6 @@ __all__ = [
     "Llm",
     "LlmModelData",
     "LlmParameters",
-    "LocalEval",
-    "LocalEvalResults",
-    "LocalEvalSets",
     "ManagedIngestionStatus",
     "ManagedIngestionStatusResponse",
     "MarkdownElementNodeParser",
@@ -514,7 +496,6 @@ __all__ = [
     "MetadataFilterValue",
     "MetadataFilters",
     "MetadataFiltersFiltersItem",
-    "MetricResult",
     "NodeParser",
     "NodeRelationship",
     "NoneChunkingConfig",
@@ -529,6 +510,7 @@ __all__ = [
     "PageScreenshotNodeWithScore",
     "PageSegmentationConfig",
     "PageSplitterNodeParser",
+    "PaginatedExtractRunsResponse",
     "PaginatedJobsHistoryWithMetrics",
     "PaginatedListCloudDocumentsResponse",
     "PaginatedListPipelineFilesResponse",
@@ -592,8 +574,6 @@ __all__ = [
     "Project",
     "ProjectCreate",
     "PromptConf",
-    "PromptMixinPrompts",
-    "PromptSpec",
     "PydanticProgramMode",
     "RecurringCreditGrant",
     "RelatedNodeInfo",

llama_cloud/types/chunk_mode.py

@@ -10,12 +10,14 @@ class ChunkMode(str, enum.Enum):
     PAGE = "PAGE"
     DOCUMENT = "DOCUMENT"
     SECTION = "SECTION"
+    GROUPED_PAGES = "GROUPED_PAGES"
 
     def visit(
         self,
         page: typing.Callable[[], T_Result],
         document: typing.Callable[[], T_Result],
         section: typing.Callable[[], T_Result],
+        grouped_pages: typing.Callable[[], T_Result],
     ) -> T_Result:
         if self is ChunkMode.PAGE:
             return page()
@@ -23,3 +25,5 @@ class ChunkMode(str, enum.Enum):
             return document()
         if self is ChunkMode.SECTION:
             return section()
+        if self is ChunkMode.GROUPED_PAGES:
+            return grouped_pages()

llama_cloud/types/extract_config.py

@@ -23,9 +23,6 @@ class ExtractConfig(pydantic.BaseModel):
 
     extraction_target: typing.Optional[ExtractTarget] = pydantic.Field(description="The extraction target specified.")
     extraction_mode: typing.Optional[ExtractMode] = pydantic.Field(description="The extraction mode specified.")
-    handle_missing: typing.Optional[bool] = pydantic.Field(
-        description="Whether to handle missing fields in the schema."
-    )
     system_prompt: typing.Optional[str]
 
     def json(self, **kwargs: typing.Any) -> str:

llama_cloud/types/{local_eval.py → extract_job_create_batch.py}

@@ -4,6 +4,8 @@ import datetime as dt
 import typing
 
 from ..core.datetime_utils import serialize_datetime
+from .extract_config import ExtractConfig
+from .extract_job_create_batch_data_schema_override import ExtractJobCreateBatchDataSchemaOverride
 
 try:
     import pydantic
@@ -14,24 +16,17 @@ except ImportError:
     import pydantic  # type: ignore
 
 
-class LocalEval(pydantic.BaseModel):
+class ExtractJobCreateBatch(pydantic.BaseModel):
     """
-    Evaluation result, EvaluationResult from llama_index.
-
-    Output of an BaseEvaluator.
+    Schema for creating extraction jobs in batch.
     """
 
-    query: typing.Optional[str]
-    contexts: typing.Optional[typing.List[str]]
-    response: typing.Optional[str]
-    passing: typing.Optional[bool]
-    feedback: typing.Optional[str]
-    score: typing.Optional[float]
-    pairwise_source: typing.Optional[str]
-    invalid_result: typing.Optional[bool] = pydantic.Field(
-        description="Whether the evaluation result is an invalid one."
+    extraction_agent_id: str = pydantic.Field(description="The id of the extraction agent")
+    file_ids: typing.List[str] = pydantic.Field(description="The ids of the files")
+    data_schema_override: typing.Optional[ExtractJobCreateBatchDataSchemaOverride] = pydantic.Field(
+        description="The data schema to override the extraction agent's data schema with"
     )
-    invalid_reason: typing.Optional[str]
+    config_override: typing.Optional[ExtractConfig]
 
     def json(self, **kwargs: typing.Any) -> str:
         kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}

llama_cloud/types/extract_job_create_batch_data_schema_override.py

@@ -0,0 +1,9 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing
+
+from .extract_job_create_batch_data_schema_override_zero_value import ExtractJobCreateBatchDataSchemaOverrideZeroValue
+
+ExtractJobCreateBatchDataSchemaOverride = typing.Union[
+    typing.Dict[str, typing.Optional[ExtractJobCreateBatchDataSchemaOverrideZeroValue]], str
+]

llama_cloud/types/extract_job_create_batch_data_schema_override_zero_value.py

@@ -0,0 +1,7 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing
+
+ExtractJobCreateBatchDataSchemaOverrideZeroValue = typing.Union[
+    typing.Dict[str, typing.Any], typing.List[typing.Any], str, int, float, bool
+]

llama_cloud/types/extract_mode.py

@@ -9,9 +9,17 @@ T_Result = typing.TypeVar("T_Result")
 class ExtractMode(str, enum.Enum):
     FAST = "FAST"
     ACCURATE = "ACCURATE"
+    MULTIMODAL = "MULTIMODAL"
 
-    def visit(self, fast: typing.Callable[[], T_Result], accurate: typing.Callable[[], T_Result]) -> T_Result:
+    def visit(
+        self,
+        fast: typing.Callable[[], T_Result],
+        accurate: typing.Callable[[], T_Result],
+        multimodal: typing.Callable[[], T_Result],
+    ) -> T_Result:
         if self is ExtractMode.FAST:
             return fast()
         if self is ExtractMode.ACCURATE:
             return accurate()
+        if self is ExtractMode.MULTIMODAL:
+            return multimodal()

llama_cloud/types/llama_parse_parameters.py

@@ -26,7 +26,6 @@ class LlamaParseParameters(pydantic.BaseModel):
     disable_ocr: typing.Optional[bool]
     annotate_links: typing.Optional[bool]
     adaptive_long_table: typing.Optional[bool]
-    compact_markdown_table: typing.Optional[bool]
     disable_reconstruction: typing.Optional[bool]
     disable_image_extraction: typing.Optional[bool]
     invalidate_cache: typing.Optional[bool]

llama_cloud/types/{local_eval_results.py → paginated_extract_runs_response.py}

@@ -4,7 +4,7 @@ import datetime as dt
 import typing
 
 from ..core.datetime_utils import serialize_datetime
-from .local_eval import LocalEval
+from .extract_run import ExtractRun
 
 try:
     import pydantic
@@ -15,16 +15,15 @@ except ImportError:
     import pydantic  # type: ignore
 
 
-class LocalEvalResults(pydantic.BaseModel):
+class PaginatedExtractRunsResponse(pydantic.BaseModel):
     """
-    Schema for the result of a local evaluation.
+    Schema for paginated extraction runs response.
     """
 
-    project_id: str = pydantic.Field(description="The ID of the project.")
-    eval_set_id: typing.Optional[str]
-    app_name: str = pydantic.Field(description="The name of the app.")
-    eval_name: str = pydantic.Field(description="The name of the eval.")
-    result: LocalEval = pydantic.Field(description="The eval results.")
+    items: typing.List[ExtractRun] = pydantic.Field(description="The list of extraction runs")
+    total: int = pydantic.Field(description="The total number of extraction runs")
+    skip: int = pydantic.Field(description="The number of extraction runs skipped")
+    limit: int = pydantic.Field(description="The maximum number of extraction runs returned")
 
     def json(self, **kwargs: typing.Any) -> str:
         kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}

llama_cloud/types/prompt_conf.py

@@ -18,6 +18,7 @@ class PromptConf(pydantic.BaseModel):
     system_prompt: typing.Optional[str] = pydantic.Field(description="The system prompt to use for the extraction.")
     extraction_prompt: typing.Optional[str] = pydantic.Field(description="The prompt to use for the extraction.")
     error_handling_prompt: typing.Optional[str] = pydantic.Field(description="The prompt to use for error handling.")
+    reasoning_prompt: typing.Optional[str] = pydantic.Field(description="The prompt to use for reasoning.")
 
     def json(self, **kwargs: typing.Any) -> str:
         kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}

llama_cloud/types/report_block.py

@@ -18,6 +18,7 @@ except ImportError:
 class ReportBlock(pydantic.BaseModel):
     idx: int = pydantic.Field(description="The index of the block")
     template: str = pydantic.Field(description="The content of the block")
+    requires_human_review: typing.Optional[bool] = pydantic.Field(description="Whether the block requires human review")
     sources: typing.Optional[typing.List[TextNodeWithScore]] = pydantic.Field(description="The sources for the block")
 
     def json(self, **kwargs: typing.Any) -> str:

llama_cloud/types/struct_mode.py

@@ -10,6 +10,7 @@ class StructMode(str, enum.Enum):
     STRUCT_PARSE = "STRUCT_PARSE"
     JSON_MODE = "JSON_MODE"
     FUNC_CALL = "FUNC_CALL"
+    STRUCT_RELAXED = "STRUCT_RELAXED"
     UNSTRUCTURED = "UNSTRUCTURED"
 
     def visit(
@@ -17,6 +18,7 @@ class StructMode(str, enum.Enum):
         struct_parse: typing.Callable[[], T_Result],
         json_mode: typing.Callable[[], T_Result],
         func_call: typing.Callable[[], T_Result],
+        struct_relaxed: typing.Callable[[], T_Result],
         unstructured: typing.Callable[[], T_Result],
     ) -> T_Result:
         if self is StructMode.STRUCT_PARSE:
@@ -25,5 +27,7 @@ class StructMode(str, enum.Enum):
             return json_mode()
         if self is StructMode.FUNC_CALL:
             return func_call()
+        if self is StructMode.STRUCT_RELAXED:
+            return struct_relaxed()
         if self is StructMode.UNSTRUCTURED:
             return unstructured()

llama_cloud/types/struct_parse_conf.py

@@ -32,6 +32,12 @@ class StructParseConf(pydantic.BaseModel):
     struct_mode: typing.Optional[StructMode] = pydantic.Field(
         description="The struct mode to use for the structured parsing."
     )
+    handle_missing: typing.Optional[bool] = pydantic.Field(
+        description="Whether to handle missing fields in the schema."
+    )
+    use_reasoning: typing.Optional[bool] = pydantic.Field(
+        description="Whether to use reasoning for the structured parsing."
+    )
     prompt_conf: typing.Optional[PromptConf] = pydantic.Field(
         description="The prompt configuration for the structured parsing."
     )

llama_cloud/types/usage.py

@@ -18,7 +18,7 @@ except ImportError:
 
 class Usage(pydantic.BaseModel):
     """
-    Response model; use UsageSubmission for tracking
+    Response model
     """
 
     active_free_credits_usage: typing.Optional[typing.List[FreeCreditsUsage]]
@@ -27,6 +27,7 @@ class Usage(pydantic.BaseModel):
     total_indexed_pages: typing.Optional[int]
     active_alerts: typing.Optional[typing.List[UsageActiveAlertsItem]]
     current_invoice_total_usd_cents: typing.Optional[int]
+    total_extraction_agents: typing.Optional[int]
 
     def json(self, **kwargs: typing.Any) -> str:
         kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}

{llama_cloud-0.1.14.dist-info → llama_cloud-0.1.16.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: llama-cloud
-Version: 0.1.14
+Version: 0.1.16
 Summary: 
 License: MIT
 Author: Logan Markewich