Merge branch 'cau/input-format-abstraction' of github.com:DS4SD/docling into cau/input-format-abstraction

docling/backend/mspowerpoint_backend.py

@@ -91,8 +91,12 @@ class MsPowerpointDocumentBackend(DeclarativeDocumentBackend, PaginatedDocumentB
             mimetype="application/vnd.ms-powerpoint",
             binary_hash=self.document_hash,
         )
+        if len(fname) > 0:
+            docname = Path(fname).stem
+        else:
+            docname = "stream"
         doc = DoclingDocument(
-            description=DescriptionItem(), name="name_without_extension", origin=origin
+            description=DescriptionItem(), name=docname, origin=origin
         )  # must add origin information
         doc = self.walk_linear(self.pptx_obj, doc)
 

docling/backend/msword_backend.py

@@ -8,6 +8,7 @@ from docling_core.types.doc import (
     DescriptionItem,
     DocItemLabel,
     DoclingDocument,
+    DocumentOrigin,
     GroupLabel,
     TableCell,
     TableData,
@@ -84,7 +85,23 @@ class MsWordDocumentBackend(DeclarativeDocumentBackend):
 
     def convert(self) -> DoclingDocument:
         # Parses the DOCX into a structured document model.
-        doc = DoclingDocument(description=DescriptionItem(), name="dummy")
+
+        fname = ""
+        if isinstance(self.path_or_stream, Path):
+            fname = self.path_or_stream.name
+
+        origin = DocumentOrigin(
+            filename=fname,
+            mimetype="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+            binary_hash=self.document_hash,
+        )
+        if len(fname) > 0:
+            docname = Path(fname).stem
+        else:
+            docname = "stream"
+        doc = DoclingDocument(
+            description=DescriptionItem(), name=docname, origin=origin
+        )
         if self.is_valid():
             assert self.docx_obj is not None
             doc = self.walk_linear(self.docx_obj.element.body, self.docx_obj, doc)