Skip to main content

AddPowerPointPresentationPipeline

What it is​

A Pipeline that reads a PowerPoint presentation (from local src... path or object storage), extracts core document properties plus all slides/shapes, builds an RDF graph using the ABI/PPT namespaces, and inserts the triples into a configured triple store (unless the presentation already exists).

Public API​

  • AddPowerPointPresentationPipelineConfiguration (dataclass)

    • Purpose: Configuration container for the pipeline.
    • Fields:
      • powerpoint_configuration: PowerPointIntegrationConfiguration
      • triple_store: ITripleStoreService

  • AddPowerPointPresentationPipelineParameters (PipelineParameters)

    • Purpose: Validated inputs for run().
    • Fields:
      • presentation_name: str — Name of the presentation.
      • storage_path: str — Storage path for the presentation.
      • download_url: Optional[str] — Optional download URL.
      • template_uri: Optional[str] — Optional template URI (must match URI_REGEX).

  • AddPowerPointPresentationPipeline (Pipeline)

    • run(parameters) -> rdflib.Graph
      • Purpose: Build and insert RDF triples for a presentation, its slides, and shapes.
      • Behavior:
        • Loads the presentation:
          • If storage_path starts with "src": uses PowerPointIntegration.create_presentation(storage_path).
          • Else: fetches bytes via StorageUtils.get_powerpoint_presentation(dir, filename) and loads with pptx.Presentation(...).
        • Extracts core properties (author, created, modified, last_modified_by) and combines them with presentation_name to form a signature string.
        • Hashes the signature (create_hash_from_string) and checks for existing entity via SPARQLUtils.get_identifier(...).
          • If found: returns SPARQLUtils.get_subject_graph(existing_uri, depth=2) and does not insert.
        • Otherwise: creates new URIs and adds triples for:
          • Presentation metadata (label, unique_id, storage_path, optional download_url, optional template links).
          • Slides (slide_number) and shapes (id/type/text/alt-text/position/size/rotation), plus relationships.
        • Inserts the graph into the triple store under graph name http://ontology.naas.ai/graph/default when non-empty.
        • Returns the built rdflib.Graph (or an empty Graph() if core property extraction fails).
    • as_tools() -> list[BaseTool]
      • Purpose: Exposes a LangChain StructuredTool named add_powerpoint_presentation that calls run() with AddPowerPointPresentationPipelineParameters.
    • as_api(...) -> None
      • Purpose: Present but not implemented; always returns None and does not register routes.

Configuration/Dependencies​

  • Requires:
    • PowerPointIntegrationConfiguration (passed into pipeline configuration).
    • ITripleStoreService (used for insert()).
  • Uses runtime services from ABIModule.get_instance().engine.services:
    • Triple store (wrapped via SPARQLUtils).
    • Object storage (wrapped via StorageUtils).
  • External libraries:
    • python-pptx (pptx.Presentation)
    • rdflib
    • fastapi (only for typing in as_api)
    • langchain_core tools

Usage​

from naas_abi_marketplace.applications.powerpoint.pipelines.AddPowerPointPresentationPipeline import (
    AddPowerPointPresentationPipeline,
    AddPowerPointPresentationPipelineConfiguration,
    AddPowerPointPresentationPipelineParameters,
)
from naas_abi_marketplace.applications.powerpoint.integrations.PowerPointIntegration import (
    PowerPointIntegrationConfiguration,
)

# Provide concrete instances appropriate for your environment:
ppt_cfg = PowerPointIntegrationConfiguration(...)   # integration-specific
triple_store = ...  # must implement ITripleStoreService

pipeline = AddPowerPointPresentationPipeline(
    AddPowerPointPresentationPipelineConfiguration(
        powerpoint_configuration=ppt_cfg,
        triple_store=triple_store,
    )
)

g = pipeline.run(
    AddPowerPointPresentationPipelineParameters(
        presentation_name="Quarterly Review",
        storage_path="src/presentations/review.pptx",  # or a non-src path to load from object storage
        download_url="https://example.com/review.pptx",
        template_uri="http://ontology.naas.ai/abi/some-template-uri",
    )
)

print(len(g))  # number of triples produced/returned

Caveats​

  • If extracting presentation.core_properties fails for any reason, the pipeline logs an error and returns an empty RDF Graph without inserting anything.
  • as_api() is a no-op; it does not expose a FastAPI endpoint.
  • Deduplication is based on a hash of presentation_name plus available core properties; different files with identical properties may be treated as the same presentation.