structifyai 1.172.0__py3-none-any.whl → 1.174.0__py3-none-any.whl

structify/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "structify"
-__version__ = "1.172.0"  # x-release-please-version
+__version__ = "1.174.0"  # x-release-please-version

structify/resources/polars.py

@@ -935,20 +935,24 @@ class PolarsResource(SyncAPIResource):
             # Wait for all PDF processing jobs to complete
             self._client.jobs.wait_for_jobs(job_ids=job_ids, title=f"Parsing {table_name} from PDFs", node_id=node_id)
 
-            # Collect results from all processed PDFs
+            # Collect results from all processed PDFs - each result is tagged with its source row_idx
             structured_results: list[dict[str, Any]] = []
 
             def collect_pdf_results(row_idx: int, dataset_name: str) -> List[Dict[str, Any]]:
                 pdf_path = batch_rows[row_idx][path_column]
                 entities_result = self._client.datasets.view_table(dataset=dataset_name, name=table_name)
-                return [{**entity.properties, path_column: pdf_path} for entity in entities_result]
+                return [
+                    {**entity.properties, path_column: pdf_path, "__row_idx__": row_idx} for entity in entities_result
+                ]
 
             with ThreadPoolExecutor(max_workers=MAX_PARALLEL_REQUESTS) as executor:
                 collect_futures = [
                     executor.submit(collect_pdf_results, row_idx, dataset_name)
                     for row_idx, dataset_name in idx_to_dataset.items()
                 ]
-                for future in tqdm(as_completed(collect_futures), total=len(collect_futures), desc="Collecting PDF extractions"):
+                for future in tqdm(
+                    as_completed(collect_futures), total=len(collect_futures), desc="Collecting PDF extractions"
+                ):
                     results = future.result()
                     structured_results.extend(results)
 
@@ -958,17 +962,12 @@ class PolarsResource(SyncAPIResource):
                     if col_name not in result_row:
                         result_row[col_name] = None
 
-            # Create DataFrame with structured results
             if not structured_results:
-                structured_df = pl.DataFrame(
-                    {col: pl.Series([], dtype=polars_schema[col]) for col in polars_schema.names()}
-                )
-            else:
-                structured_df = pl.DataFrame(structured_results, schema=polars_schema)
+                return pl.DataFrame(schema=polars_schema)
 
-            # Join with original batch to preserve any additional columns
-            joined_df = batch_df.join(structured_df, on=path_column, how="left")
-            return joined_df
+            # Build result dataframe directly from structured_results without joining
+            # Each entity is already tagged with path_column from its source PDF
+            return pl.DataFrame(structured_results, schema=polars_schema)
 
         return document_paths.map_batches(structure_batch, schema=polars_schema, no_optimizations=True)
 

structify/types/chat_create_session_params.py

@@ -44,6 +44,7 @@ class Config(TypedDict, total=False):
             "bedrock.claude-sonnet-4-bedrock",
             "bedrock.claude-sonnet-4-5-bedrock",
             "bedrock.claude-opus-4-5-bedrock",
+            "bedrock.claude-haiku-4-5-bedrock",
             "gemini.gemini-2.5-pro",
             "gemini.gemini-2.5-flash",
             "gemini.gemini-3-pro-preview",

structify/types/code_generate_code_params.py

@@ -48,6 +48,7 @@ class Config(TypedDict, total=False):
             "bedrock.claude-sonnet-4-bedrock",
             "bedrock.claude-sonnet-4-5-bedrock",
             "bedrock.claude-opus-4-5-bedrock",
+            "bedrock.claude-haiku-4-5-bedrock",
             "gemini.gemini-2.5-pro",
             "gemini.gemini-2.5-flash",
             "gemini.gemini-3-pro-preview",

structify/types/connector_catalog_list_response.py

@@ -5,10 +5,18 @@ from typing import List
 from .._models import BaseModel
 from .connector_catalog_with_methods import ConnectorCatalogWithMethods
 
-__all__ = ["ConnectorCatalogListResponse"]
+__all__ = ["ConnectorCatalogListResponse", "CategoryCount"]
+
+
+class CategoryCount(BaseModel):
+    category: str
+
+    count: int
 
 
 class ConnectorCatalogListResponse(BaseModel):
+    category_counts: List[CategoryCount]
+
     items: List[ConnectorCatalogWithMethods]
 
     total_count: int

structify/types/job_event_body.py

@@ -18,12 +18,14 @@ __all__ = [
     "DerivedProperty",
     "Failed",
     "Completed",
+    "CacheHit",
     "AttemptedMatch",
     "DatahubPageFetched",
     "DatahubDatabasesCreated",
     "DatahubSchemasCreated",
     "DatahubTablesProcessed",
     "DatahubEmbeddingBatch",
+    "ViewedPdfPage",
 ]
 
 
@@ -106,6 +108,14 @@ class Completed(BaseModel):
     message: Optional[str] = None
 
 
+class CacheHit(BaseModel):
+    cached_from_job_id: str
+
+    event_type: Literal["cache_hit"]
+
+    message: Optional[str] = None
+
+
 class AttemptedMatch(BaseModel):
     candidates: List[Dict[str, Union[str, bool, float]]]
 
@@ -117,6 +127,8 @@ class AttemptedMatch(BaseModel):
 
     match_idx: Optional[int] = None
 
+    raw_text: Optional[str] = None
+
 
 class DatahubPageFetched(BaseModel):
     datasets_in_page: int
@@ -164,6 +176,12 @@ class DatahubEmbeddingBatch(BaseModel):
     total_batches: int
 
 
+class ViewedPdfPage(BaseModel):
+    event_type: Literal["viewed_pdf_page"]
+
+    page_index: int
+
+
 JobEventBody: TypeAlias = Annotated[
     Union[
         AgentNavigated,
@@ -175,12 +193,14 @@ JobEventBody: TypeAlias = Annotated[
         DerivedProperty,
         Failed,
         Completed,
+        CacheHit,
         AttemptedMatch,
         DatahubPageFetched,
         DatahubDatabasesCreated,
         DatahubSchemasCreated,
         DatahubTablesProcessed,
         DatahubEmbeddingBatch,
+        ViewedPdfPage,
     ],
     PropertyInfo(discriminator="event_type"),
 ]

structify/types/tool_invocation.py

@@ -1,10 +1,11 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
-from typing import List, Union, Optional
+from typing import Dict, List, Union, Optional
 from typing_extensions import Literal, Annotated, TypeAlias
 
 from .._utils import PropertyInfo
 from .._models import BaseModel
+from .knowledge_graph import KnowledgeGraph
 
 __all__ = [
     "ToolInvocation",
@@ -12,6 +13,21 @@ __all__ = [
     "WebSearchInput",
     "WebNavigate",
     "WebNavigateInput",
+    "ViewPage",
+    "ViewPageInput",
+    "Save",
+    "SaveInput",
+    "SaveEntities",
+    "SaveEntitiesInput",
+    "Exit",
+    "ExitInput",
+    "APIExecute",
+    "APIExecuteInput",
+    "Javascript",
+    "JavascriptInput",
+    "NavigateToIFrame",
+    "NavigateToIFrameInput",
+    "InfiniteScroll",
     "InspectStep",
     "InspectStepInput",
     "ReadNodeLogs",
@@ -66,6 +82,8 @@ class WebSearch(BaseModel):
 class WebNavigateInput(BaseModel):
     url: str
 
+    output_format: Optional[Literal["Text", "Visual"]] = None
+
 
 class WebNavigate(BaseModel):
     input: WebNavigateInput
@@ -73,6 +91,95 @@ class WebNavigate(BaseModel):
     name: Literal["WebNavigate"]
 
 
+class ViewPageInput(BaseModel):
+    page_number: int
+
+
+class ViewPage(BaseModel):
+    input: ViewPageInput
+
+    name: Literal["ViewPage"]
+
+
+class SaveInput(BaseModel):
+    knowledge_graph: KnowledgeGraph
+    """
+    Knowledge graph info structured to deserialize and display in the same format
+    that the LLM outputs. Also the first representation of an LLM output in the
+    pipeline from raw tool output to being merged into a DB
+    """
+
+    reason: str
+
+    sources: List[str]
+
+
+class Save(BaseModel):
+    input: SaveInput
+
+    name: Literal["Save"]
+
+
+class SaveEntitiesInput(BaseModel):
+    entities: List[Dict[str, Dict[str, object]]]
+
+    reason: str
+
+    sources: List[str]
+
+
+class SaveEntities(BaseModel):
+    input: SaveEntitiesInput
+
+    name: Literal["SaveEntities"]
+
+
+class ExitInput(BaseModel):
+    reason: str
+
+
+class Exit(BaseModel):
+    input: ExitInput
+
+    name: Literal["Exit"]
+
+
+class APIExecuteInput(BaseModel):
+    code: str
+
+
+class APIExecute(BaseModel):
+    input: APIExecuteInput
+
+    name: Literal["ApiExecute"]
+
+
+class JavascriptInput(BaseModel):
+    code: str
+
+
+class Javascript(BaseModel):
+    input: JavascriptInput
+
+    name: Literal["Javascript"]
+
+
+class NavigateToIFrameInput(BaseModel):
+    index: int
+
+
+class NavigateToIFrame(BaseModel):
+    input: NavigateToIFrameInput
+
+    name: Literal["NavigateToIFrame"]
+
+
+class InfiniteScroll(BaseModel):
+    input: object
+
+    name: Literal["InfiniteScroll"]
+
+
 class InspectStepInput(BaseModel):
     step_name: str
 
@@ -329,6 +436,14 @@ ToolInvocation: TypeAlias = Annotated[
     Union[
         WebSearch,
         WebNavigate,
+        ViewPage,
+        Save,
+        SaveEntities,
+        Exit,
+        APIExecute,
+        Javascript,
+        NavigateToIFrame,
+        InfiniteScroll,
         InspectStep,
         ReadNodeLogs,
         DeleteFile,

{structifyai-1.172.0.dist-info → structifyai-1.174.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: structifyai
-Version: 1.172.0
+Version: 1.174.0
 Summary: The official Python library for the structify API
 Project-URL: Homepage, https://github.com/StructifyAI/structify-python
 Project-URL: Repository, https://github.com/StructifyAI/structify-python

{structifyai-1.172.0.dist-info → structifyai-1.174.0.dist-info}/RECORD

@@ -11,7 +11,7 @@ structify/_resource.py,sha256=tJi4pDQooQZ_zJwEwrLj-U-ye2hC-cbmr1GzIwCT10Y,1118
 structify/_response.py,sha256=RuNhMDiZUdPqEbmFJHDVI4FMPDszk8QjK9LVWm1Fagk,28806
 structify/_streaming.py,sha256=n4C9M7ITmANYn9LaWHNoqJdIIyF7svLco2qst7u3M7U,10233
 structify/_types.py,sha256=jj4p-m3vpUma0AdhPWIaljHZXeb4RKnrAusjVdpDy5Y,7597
-structify/_version.py,sha256=Rx9zou1rDU7TWT-wTjB0rgLPSL1e7FlGBViJXZzQX60,163
+structify/_version.py,sha256=jAkquXwE8e7XiNAPLSl0tNkX5WyP7kZ2giJmBdnOmlM,163
 structify/pagination.py,sha256=ycybhWcpKk4ztsMcCA6C0WZiJejGrSx6bSr8LLskJUY,4346
 structify/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 structify/_utils/__init__.py,sha256=7fch0GT9zpNnErbciSpUNa-SjTxxjY6kxHxKMOM4AGs,2305
@@ -39,7 +39,7 @@ structify/resources/external_dataframe_proxy.py,sha256=DSn0YwWIembR__ZtDxVCJtyY3
 structify/resources/jobs.py,sha256=gO1aSByi1dMvW90UDsMmNhLHFCOY4ENLkZcAx4gbLHY,30108
 structify/resources/match.py,sha256=gDWEWnKwEoLbheQAMFltJCk2ysa_L9AuJMOaauM7c4Y,12248
 structify/resources/nango.py,sha256=Zl0M1XhlVe61jHVd-SdYI9uEbEhIRmskvlk7Xp0Lh8g,9166
-structify/resources/polars.py,sha256=nE-I3kX552BOjcyn4ZJFlAY2-_Y0LM0Pta1653-uYCg,61613
+structify/resources/polars.py,sha256=--kVVUoJ7AvGirnc462r0TRsK7lcXXDQD0XWhqtG3K8,61552
 structify/resources/projects.py,sha256=YDikBDB9D1EXyZ2GyRx4GlpQ83snw51YlNuU1sLHqho,14117
 structify/resources/public_sessions.py,sha256=_JmssE0MMjeGdxT0FWtrkcceSV4skeEkVGYeO2FkJes,9976
 structify/resources/sandbox.py,sha256=Gc7uxZAOmbXA477UHvS244BokI1Tte_34xxIj0dC5PA,16984
@@ -90,7 +90,7 @@ structify/types/chat_admin_issue_found_params.py,sha256=N5YMgm6O-Yr3sXZb2sWDHdyN
 structify/types/chat_copy_node_output_by_code_hash_params.py,sha256=gP4V2Y0_Rb4-kwoEceClJLgjVdWv7bgcI4qqQIvdNv8,356
 structify/types/chat_copy_node_output_by_code_hash_response.py,sha256=Cst1RK9jDrMAKEyzLIk5_-D6YUbO-AIVllYui2sy2aQ,272
 structify/types/chat_copy_params.py,sha256=tfA3jdduKDJeHiYjWWKdHifRMp1sCGyGQcYHFuMFdnE,425
-structify/types/chat_create_session_params.py,sha256=_ugvArrIDJDP1zT8o_mFCEGkZOwv67xl5nsMPIrMvQE,1762
+structify/types/chat_create_session_params.py,sha256=QSkL1Z0Vj0sT8AJ8mcDZQIqAwUkbhiQHuF6811veMmc,1810
 structify/types/chat_delete_files_params.py,sha256=Rv24bWe1CK524xobl9-_APx5GG8KTqIwEtUgYIUlMXc,343
 structify/types/chat_delete_files_response.py,sha256=ZoFJjfZqn_rVuwiFhsuFigN_AQHh_DRkb5KRj7J_49g,225
 structify/types/chat_dependency.py,sha256=J8JLY6kBjFt4dgf-_Vk_HEjxS5R6_6VnTQue2JzZRug,415
@@ -114,13 +114,13 @@ structify/types/chat_update_session_favorite_params.py,sha256=UNi0YujSs1gCSry2jK
 structify/types/chat_update_session_params.py,sha256=5NQZx-j-W2xQGlX0GU-9vwyQt0QunlifWY2MpRtrEv8,382
 structify/types/chat_update_visibility_params.py,sha256=uWtgq4zb_TSO4hIsY0FmdWaljB_20wpjl6IQhRMibTg,361
 structify/types/chat_visibility.py,sha256=GpTgjpcupE1z9Iul8fxOrlMoP_2wjQ_7Z0rJ2y9qO5Q,244
-structify/types/code_generate_code_params.py,sha256=NzDMzMT7IpzrL5bHhC4m3k0B7R5N5WUbiyCmTudM2oE,2066
+structify/types/code_generate_code_params.py,sha256=cX5HYhmrv-9gBZcYMy9jJ8h4vMewt-Zy5jVfAYJsirA,2114
 structify/types/code_interrupt_generation_params.py,sha256=1Y9VOgObIJFyYgAEkUuWZRKKV5-4HcoRA6p5iSEnF3s,410
 structify/types/connector.py,sha256=hqfk8x1ZM39idvAd4wXLm1QNrnT3kRgxEuhk8O28-B0,1069
 structify/types/connector_auth_method.py,sha256=iHBmcNbi74mDjFd_m4-HrGrZoV9_WRSFtrOY0fz9NhQ,562
 structify/types/connector_auth_method_with_fields.py,sha256=EABCugmJ8ahZNhSqvNQAESjHpD3kozh4GPop2OZpSMw,519
 structify/types/connector_catalog_list_params.py,sha256=vqyYfA7M3STeDKKshpD9FthkmoSh4LJecrBFhVGbBJs,470
-structify/types/connector_catalog_list_response.py,sha256=Mr9KRSF1oi2HauT_Hvkb8CYjdMFIG3Zc2lb5U8itOhE,376
+structify/types/connector_catalog_list_response.py,sha256=0e1AQU59zTfN-bvx1GCUElnaEiyW104XHUIYvZ4Gkmg,503
 structify/types/connector_catalog_with_methods.py,sha256=EZdXqN6oZggX-SYXUrn_hDgCRHeMY1zNo7pY1q0vX4E,532
 structify/types/connector_category.py,sha256=2YnDoj4lXtyfRTVhSi-7a5InFuOUn26SqeVeaCSS4lE,245
 structify/types/connector_create_params.py,sha256=Xj2MWkiWlMhgCc-Oo3udVGlXwSjU7jy8f0WpKgRFbj8,961
@@ -259,7 +259,7 @@ structify/types/granularity.py,sha256=At6biWApGE7uE8jr5KnHP9Jr1yPFkuqLwXjExaswtB
 structify/types/image.py,sha256=FpYU3gDZnet0wO17e2uHzcyRUD6E1ssSgv63Ew0DzjU,269
 structify/types/invitation_details_response.py,sha256=TmyeM4mW4Kb6L0d7Ook9cH3g8vzfQYPnvZDIBdgVAO4,272
 structify/types/job_cancel_response.py,sha256=y8M8qPkcXT-pTi4IwQ0JBJQzXeAQIs3u2OsaVeGBTtc,1224
-structify/types/job_event_body.py,sha256=CUZt8TK6d46qf4KYBxH0mFYaiyoDJT4pGAUZMhGrwC4,3564
+structify/types/job_event_body.py,sha256=pO11fTm5sGvQ6cx7NvWTitUaunSqdRo-d5tkyh4KPsk,3906
 structify/types/job_get_scrapers_response.py,sha256=-E9DaG9YCkrJeuffZ3RiI6MW7ZsbPjmlVU3jUQdggSY,694
 structify/types/job_get_source_entities_response.py,sha256=vlGKFkMKjII48EexIEBbls6-PAVYDDkmR1__aKjuiUo,3379
 structify/types/job_list_params.py,sha256=i_MYi4vIFC6Dq5I4E4htYwZejqRE5qAf2lgg_SRBuKg,1012
@@ -367,7 +367,7 @@ structify/types/team_wiki_page.py,sha256=PDcFqG8mc10Ejnh6YeoSdalNorn40axi7K3ppmh
 structify/types/team_with_role.py,sha256=t-UgnD90UmxlMpLHdxN8L7kK7VxMVLj-PM26NvkV_2U,327
 structify/types/teams_link_code_response.py,sha256=p81nvCP4KeGVUgQiiaFwAlLMRaGnG4Iv7rp03WFG0uA,269
 structify/types/token_response.py,sha256=N75RMh9ZF05z5Ubc_Pu2pePe3MN7Yqn2GB1FJ8FQJZ0,408
-structify/types/tool_invocation.py,sha256=5D_zxMZmI3omAvGHTJSmCeuZMgMQXhbtomWh0f2Ll-k,5987
+structify/types/tool_invocation.py,sha256=HCc_nFmfjm5MnoDKDbtnh7TIvLEH7gvsBI4IIU0AxO4,7995
 structify/types/tool_metadata.py,sha256=drtuS9ZQm4cB2KScB1pycX1ifjhGwSUMJm9TFv1SHuI,382
 structify/types/tool_result.py,sha256=C7CSxU0PfW4WMiWlwDckEmsjTbZmWzEoSt2LbcYdAtc,2013
 structify/types/update_member_role_response.py,sha256=q1jTM0lFmzvH8ki7GcSQAh-5cN7VnLF6Jg_IRdA4AqE,222
@@ -489,7 +489,7 @@ structify/types/user/stripe_create_portal_session_params.py,sha256=5AYRC8z_SlKmd
 structify/types/user/stripe_create_session_params.py,sha256=DFcNLNzEWeupkGQ9J5PafsuL_bIU9cLEIhAmFPsRlfo,387
 structify/types/user/stripe_create_subscription_params.py,sha256=d8HfiC94gJbG-cC_WvBz6xYCvxKJO_EP2yyVmVvufrU,424
 structify/types/user/subscription_plan.py,sha256=qKJMM-zPpYolYC1DlypOwPpxlyJBLkQqFK_0VpwktJs,222
-structifyai-1.172.0.dist-info/METADATA,sha256=wA_nnCaPjieihaBzi0fUawgTBLp4syLr_bLKFFZKz9o,16399
-structifyai-1.172.0.dist-info/WHEEL,sha256=C2FUgwZgiLbznR-k0b_5k3Ai_1aASOXDss3lzCUsUug,87
-structifyai-1.172.0.dist-info/licenses/LICENSE,sha256=9CwgrmGz3rZSTT-KqGc1gua-7g8B4ThTgMtUgPALh5c,11339
-structifyai-1.172.0.dist-info/RECORD,,
+structifyai-1.174.0.dist-info/METADATA,sha256=_cBl4jOw0K_TtGioyYQoTH0ezXr8x4NPfipk2wWlzVk,16399
+structifyai-1.174.0.dist-info/WHEEL,sha256=C2FUgwZgiLbznR-k0b_5k3Ai_1aASOXDss3lzCUsUug,87
+structifyai-1.174.0.dist-info/licenses/LICENSE,sha256=9CwgrmGz3rZSTT-KqGc1gua-7g8B4ThTgMtUgPALh5c,11339
+structifyai-1.174.0.dist-info/RECORD,,