Reuse file_path instead of introducing another document_path parameter

Author: stellasia

docs/source/user_guide_kg_builder.rst

@@ -224,7 +224,6 @@ Run Parameters
 
 SimpleKGPipeline also accepts addition runtime parameters:
 
-- ``document_path`` (str): only used when ``from_pdf=False``, this is the path property of the ``Document`` node.
 - ``document_metadata`` (dict): each item will be saved as a property attached to the ``Document`` node.
 
 

examples/build_graph/simple_kg_builder_from_text.py

@@ -81,9 +81,9 @@ async def define_and_run_pipeline(
     )
     return await kg_builder.run_async(
         text=TEXT,
-        # optional, specify document path for the Document node
+        # optional, specify file path for the Document node
         # if not, a random name will be generated
-        # document_path="my_document.txt"
+        # file_path="my_document.txt"
         # optional, add document metadata, each item will
         # be saved as a property of the Document node
         # document_metadata={"author": "Frank Herbert"},

src/neo4j_graphrag/experimental/pipeline/config/template_pipeline/simple_kg_builder.py

@@ -339,8 +339,8 @@ def _get_connections(self) -> list[ConnectionDefinition]:
     def get_run_params(self, user_input: dict[str, Any]) -> dict[str, Any]:
         text = user_input.get("text")
         file_path = user_input.get("file_path")
-        if not ((text is None) ^ (file_path is None)):
-            # exactly one of text or user_input must be set
+        if text is None and file_path is None:
+            # use must provide either text or file_path or both
             raise PipelineDefinitionError(
                 "Use either 'text' (when from_pdf=False) or 'file_path' (when from_pdf=True) argument."
             )
@@ -367,7 +367,7 @@ def get_run_params(self, user_input: dict[str, Any]) -> dict[str, Any]:
                 run_params["schema"]["text"] = text
             run_params["extractor"]["document_info"] = dict(
                 path=user_input.get(
-                    "document_path",
+                    "file_path",
                 )
                 or "document.txt",
                 metadata=user_input.get("document_metadata"),

src/neo4j_graphrag/experimental/pipeline/kg_builder.py

@@ -148,16 +148,14 @@ async def run_async(
         self,
         file_path: Optional[str] = None,
         text: Optional[str] = None,
-        document_path: Optional[str] = None,
         document_metadata: Optional[dict[str, Any]] = None,
     ) -> PipelineResult:
         """
         Asynchronously runs the knowledge graph building process.
 
         Args:
-            file_path (Optional[str]): The path to the PDF file to process. Required if `from_pdf` is True.
+            file_path (Optional[str]): The path to the PDF file to process. Required if `from_pdf` is True. If `from_pdf` is False, can be used to set the Document node path property.
             text (Optional[str]): The text content to process. Required if `from_pdf` is False.
-            document_path (Optional[str]): The path to the document to process. Required if `from_pdf` is True.
             document_metadata (Optional[dict[str, Any]]): The metadata to attach to the document.
 
         Returns:
@@ -167,7 +165,6 @@ async def run_async(
             {
                 "file_path": file_path,
                 "text": text,
-                "document_path": document_path,
                 "document_metadata": document_metadata,
             }
         )

tests/unit/experimental/pipeline/config/template_pipeline/test_simple_kg_builder.py

@@ -322,16 +322,6 @@ def test_simple_kg_pipeline_config_run_params_no_file_no_text() -> None:
     )
 
 
-def test_simple_kg_pipeline_config_run_params_both_file_and_text() -> None:
-    config = SimpleKGPipelineConfig(from_pdf=False)
-    with pytest.raises(PipelineDefinitionError) as excinfo:
-        config.get_run_params({"text": "my text", "file_path": "my file"})
-    assert (
-        "Use either 'text' (when from_pdf=False) or 'file_path' (when from_pdf=True) argument."
-        in str(excinfo)
-    )
-
-
 def test_simple_kg_pipeline_config_process_schema_with_precedence_legacy() -> None:
     entities: list[EntityInputType] = [
         "Person",

tests/unit/experimental/pipeline/test_kg_builder.py

@@ -90,7 +90,7 @@ async def test_knowledge_graph_builder_document_info_with_text(_: Mock) -> None:
     ) as mock_run:
         await kg_builder.run_async(
             text=text_input,
-            document_path="my_document.txt",
+            file_path="my_document.txt",
             document_metadata={"source": "google drive"},
         )