docling-core 2.46.0__py3-none-any.whl → 2.47.0__py3-none-any.whl

docling_core/types/doc/document.py

@@ -4045,7 +4045,7 @@ class DoclingDocument(BaseModel):
             root=root,
             with_groups=with_groups,
             traverse_pictures=traverse_pictures,
-            page_no=page_no,
+            page_nrs={page_no} if page_no is not None else None,
             included_content_layers=included_content_layers,
         ):
             yield item, len(stack)
@@ -4055,7 +4055,7 @@ class DoclingDocument(BaseModel):
         root: Optional[NodeItem] = None,
         with_groups: bool = False,
         traverse_pictures: bool = False,
-        page_no: Optional[int] = None,
+        page_nrs: Optional[set[int]] = None,
         included_content_layers: Optional[set[ContentLayer]] = None,
         _stack: Optional[list[int]] = None,
     ) -> typing.Iterable[Tuple[NodeItem, list[int]]]:  # tuple of node and level
@@ -4078,8 +4078,8 @@ class DoclingDocument(BaseModel):
             and (
                 not isinstance(root, DocItem)
                 or (
-                    page_no is None
-                    or any(prov.page_no == page_no for prov in root.prov)
+                    page_nrs is None
+                    or any(prov.page_no in page_nrs for prov in root.prov)
                 )
             )
             and root.content_layer in my_layers
@@ -4113,7 +4113,7 @@ class DoclingDocument(BaseModel):
                     child,
                     with_groups=with_groups,
                     traverse_pictures=traverse_pictures,
-                    page_no=page_no,
+                    page_nrs=page_nrs,
                     _stack=my_stack,
                     included_content_layers=my_layers,
                 )
@@ -5603,7 +5603,9 @@ class DoclingDocument(BaseModel):
         def get_item_list(self, key: str) -> list[NodeItem]:
             return getattr(self, key)
 
-        def index(self, doc: "DoclingDocument") -> None:
+        def index(
+            self, doc: "DoclingDocument", page_nrs: Optional[set[int]] = None
+        ) -> None:
 
             orig_ref_to_new_ref: dict[str, str] = {}
             page_delta = self._max_page - min(doc.pages.keys()) + 1 if doc.pages else 0
@@ -5614,10 +5616,11 @@ class DoclingDocument(BaseModel):
             self._names.append(doc.name)
 
             # collect items in traversal order
-            for item, _ in doc.iterate_items(
+            for item, _ in doc._iterate_items_with_stack(
                 with_groups=True,
                 traverse_pictures=True,
                 included_content_layers={c for c in ContentLayer},
+                page_nrs=page_nrs,
             ):
                 key = item.self_ref.split("/")[1]
                 is_body = key == "body"
@@ -5686,12 +5689,13 @@ class DoclingDocument(BaseModel):
             # update pages
             new_max_page = None
             for page_nr in doc.pages:
-                new_page = copy.deepcopy(doc.pages[page_nr])
-                new_page_nr = page_nr + page_delta
-                new_page.page_no = new_page_nr
-                self.pages[new_page_nr] = new_page
-                if new_max_page is None or new_page_nr > new_max_page:
-                    new_max_page = new_page_nr
+                if page_nrs is None or page_nr in page_nrs:
+                    new_page = copy.deepcopy(doc.pages[page_nr])
+                    new_page_nr = page_nr + page_delta
+                    new_page.page_no = new_page_nr
+                    self.pages[new_page_nr] = new_page
+                    if new_max_page is None or new_page_nr > new_max_page:
+                        new_max_page = new_page_nr
             if new_max_page is not None:
                 self._max_page = new_max_page
 
@@ -5715,6 +5719,14 @@ class DoclingDocument(BaseModel):
         doc_index.index(doc=self)
         self._update_from_index(doc_index)
 
+    def filter(self, page_nrs: Optional[set[int]] = None) -> "DoclingDocument":
+        """Create a new document based on the provided filter parameters."""
+        doc_index = DoclingDocument._DocIndex()
+        doc_index.index(doc=self, page_nrs=page_nrs)
+        res_doc = DoclingDocument(name=self.name)
+        res_doc._update_from_index(doc_index)
+        return res_doc
+
     @classmethod
     def concatenate(cls, docs: Sequence["DoclingDocument"]) -> "DoclingDocument":
         """Concatenate multiple documents into a single document."""

{docling_core-2.46.0.dist-info → docling_core-2.47.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docling-core
-Version: 2.46.0
+Version: 2.47.0
 Summary: A python library to define and validate data types in Docling.
 Author-email: Cesar Berrospi Ramis <ceb@zurich.ibm.com>, Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>
 Maintainer-email: Panos Vagenas <pva@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Christoph Auer <cau@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>, Cesar Berrospi Ramis <ceb@zurich.ibm.com>

{docling_core-2.46.0.dist-info → docling_core-2.47.0.dist-info}/RECORD

@@ -43,7 +43,7 @@ docling_core/types/__init__.py,sha256=MVRSgsk5focwGyAplh_TRR3dEecIXpd98g_u3zZ5HX
 docling_core/types/base.py,sha256=PusJskRVL19y-hq0BgXr5e8--QEqSqLnFNJ8UbOqW88,8318
 docling_core/types/doc/__init__.py,sha256=Vsl3oJV3_BLpS7rIwvahhcWOwmEBvj7ZbQzQCCl-IQk,1678
 docling_core/types/doc/base.py,sha256=i98y4IF250adR-8BSS374K90fwfwG-vBfWh14tLC5Cs,15906
-docling_core/types/doc/document.py,sha256=Ab-JOc6fkzocXP3PcxPRXJPjLOhOTYo_0571vSr6VXo,202093
+docling_core/types/doc/document.py,sha256=jyMcK1oiu8X8juNa9DuI3S1imn4hXwjOS7iTLQ1HykU,202707
 docling_core/types/doc/labels.py,sha256=-W1-LW6z0J9F9ExJqR0Wd1WeqWTaY3Unm-j1UkQGlC4,7330
 docling_core/types/doc/page.py,sha256=35h1xdtCM3-AaN8Dim9jDseZIiw-3GxpB-ofF-H2rQQ,41878
 docling_core/types/doc/tokens.py,sha256=z22l9J81_sg9CYMvOuLmPuLsNT7h_s7wao2UT89DvI8,9278
@@ -76,9 +76,9 @@ docling_core/utils/generate_jsonschema.py,sha256=uNX1O5XnjyB5nA66XqZXTt3YbGuR2ty
 docling_core/utils/legacy.py,sha256=G7ed8fkBpIO8hG3DKEY83cHsrKJHyvDst_1jSdgBXMI,24406
 docling_core/utils/validate.py,sha256=aQ11UbFyl8iD_N7yTTZmm_VVeXz8KcCyn3GLXgkfYRM,2049
 docling_core/utils/validators.py,sha256=azcrndLzhNkTWnbFSu9shJ5D3j_znnLrIFA5R8hzmGU,2798
-docling_core-2.46.0.dist-info/licenses/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
-docling_core-2.46.0.dist-info/METADATA,sha256=txMHh-7y8N3RiJ_M_HbrsvzRyGPJVXv8UcA6_DpAfok,6453
-docling_core-2.46.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-docling_core-2.46.0.dist-info/entry_points.txt,sha256=ER4zROQWkFMHIrY-oqY5E4HeCcCIg8dLkNztYGxdb7c,59
-docling_core-2.46.0.dist-info/top_level.txt,sha256=O-tcXpGiurlud-1ZxMq1b-OmrfAVA4sajcgWU32RtfA,13
-docling_core-2.46.0.dist-info/RECORD,,
+docling_core-2.47.0.dist-info/licenses/LICENSE,sha256=2M9-6EoQ1sxFztTOkXGAtwUDJvnWaAHdB9BYWVwGkIw,1087
+docling_core-2.47.0.dist-info/METADATA,sha256=jW4Zdx0WwStnLDifSsvYyGLw-5C2IYiEeK4IQRGQi-I,6453
+docling_core-2.47.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+docling_core-2.47.0.dist-info/entry_points.txt,sha256=ER4zROQWkFMHIrY-oqY5E4HeCcCIg8dLkNztYGxdb7c,59
+docling_core-2.47.0.dist-info/top_level.txt,sha256=O-tcXpGiurlud-1ZxMq1b-OmrfAVA4sajcgWU32RtfA,13
+docling_core-2.47.0.dist-info/RECORD,,