fix(markdown): fix case of empty page break string (#298)

Signed-off-by: Panos Vagenas <[email protected]>

Author: vagenas

docling_core/transforms/serializer/markdown.py

@@ -535,14 +535,14 @@ def serialize_doc(
     ) -> SerializationResult:
         """Serialize a document out of its parts."""
         text_res = "\n\n".join([p.text for p in parts if p.text])
-        if self.params.page_break_placeholder:
+        if self.requires_page_break():
             page_sep = self.params.page_break_placeholder or ""
             for full_match, _, _ in self._get_page_breaks(text=text_res):
                 text_res = text_res.replace(full_match, page_sep)
 
         return create_ser_result(text=text_res, span_source=parts)
 
     @override
-    def requires_page_break(self):
+    def requires_page_break(self) -> bool:
         """Whether to add page breaks."""
         return self.params.page_break_placeholder is not None

test/data/doc/activities_pb_empty.gt.md

@@ -0,0 +1,35 @@
+## Summer activities
+
+## Swimming in the lake
+
+Duck
+
+Figure 1: This is a cute duckling
+
+## Let's swim!
+
+To get started with swimming, first lay down in a water and try not to drown:
+
+- ∞ You can relax and look around
+- ∞ Paddle about
+- ∞ Enjoy summer warmth
+
+Also, don't forget:
+
+- 1. Wear sunglasses
+- 2. Don't forget to drink water
+- 3. Use sun cream
+
+Hmm, what else…
+
+- -Another activity item
+
+- -Yet another one
+- -Stopping it here
+
+Some text.
+
+
+
+- -Starting the next page with a list item.
+- -Second item.

test/data/doc/activities_pb_non_empty.gt.md

@@ -0,0 +1,35 @@
+## Summer activities
+
+## Swimming in the lake
+
+Duck
+
+Figure 1: This is a cute duckling
+
+## Let's swim!
+
+To get started with swimming, first lay down in a water and try not to drown:
+
+- ∞ You can relax and look around
+- ∞ Paddle about
+- ∞ Enjoy summer warmth
+
+Also, don't forget:
+
+- 1. Wear sunglasses
+- 2. Don't forget to drink water
+- 3. Use sun cream
+
+Hmm, what else…
+
+- -Another activity item
+<!-- page-break -->
+- -Yet another one
+- -Stopping it here
+
+Some text.
+
+<!-- page-break -->
+
+- -Starting the next page with a list item.
+- -Second item.

test/data/doc/activities_pb_none.gt.md

@@ -0,0 +1,32 @@
+## Summer activities
+
+## Swimming in the lake
+
+Duck
+
+Figure 1: This is a cute duckling
+
+## Let's swim!
+
+To get started with swimming, first lay down in a water and try not to drown:
+
+- ∞ You can relax and look around
+- ∞ Paddle about
+- ∞ Enjoy summer warmth
+
+Also, don't forget:
+
+- 1. Wear sunglasses
+- 2. Don't forget to drink water
+- 3. Use sun cream
+
+Hmm, what else…
+
+- -Another activity item
+- -Yet another one
+- -Stopping it here
+
+Some text.
+
+- -Starting the next page with a list item.
+- -Second item.

test/test_serialization.py

@@ -48,6 +48,57 @@ def test_md_cross_page_list_page_break():
     verify(exp_file=src.parent / f"{src.stem}.gt.md", actual=actual)
 
 
+def test_md_cross_page_list_page_break_none():
+    src = Path("./test/data/doc/activities.json")
+    doc = DoclingDocument.load_from_json(src)
+
+    ser = MarkdownDocSerializer(
+        doc=doc,
+        params=MarkdownParams(
+            image_mode=ImageRefMode.PLACEHOLDER,
+            image_placeholder="<!-- image -->",
+            page_break_placeholder=None,
+            labels=_DEFAULT_LABELS - {DocItemLabel.PICTURE},
+        ),
+    )
+    actual = ser.serialize().text
+    verify(exp_file=src.parent / f"{src.stem}_pb_none.gt.md", actual=actual)
+
+
+def test_md_cross_page_list_page_break_empty():
+    src = Path("./test/data/doc/activities.json")
+    doc = DoclingDocument.load_from_json(src)
+
+    ser = MarkdownDocSerializer(
+        doc=doc,
+        params=MarkdownParams(
+            image_mode=ImageRefMode.PLACEHOLDER,
+            image_placeholder="<!-- image -->",
+            page_break_placeholder="",
+            labels=_DEFAULT_LABELS - {DocItemLabel.PICTURE},
+        ),
+    )
+    actual = ser.serialize().text
+    verify(exp_file=src.parent / f"{src.stem}_pb_empty.gt.md", actual=actual)
+
+
+def test_md_cross_page_list_page_break_non_empty():
+    src = Path("./test/data/doc/activities.json")
+    doc = DoclingDocument.load_from_json(src)
+
+    ser = MarkdownDocSerializer(
+        doc=doc,
+        params=MarkdownParams(
+            image_mode=ImageRefMode.PLACEHOLDER,
+            image_placeholder="<!-- image -->",
+            page_break_placeholder="<!-- page-break -->",
+            labels=_DEFAULT_LABELS - {DocItemLabel.PICTURE},
+        ),
+    )
+    actual = ser.serialize().text
+    verify(exp_file=src.parent / f"{src.stem}_pb_non_empty.gt.md", actual=actual)
+
+
 def test_md_cross_page_list_page_break_p2():
     src = Path("./test/data/doc/activities.json")
     doc = DoclingDocument.load_from_json(src)