get_textbox(), get_text_selection(): leave out text hidden by a clip path

src/__init__.py

@@ -12881,7 +12881,9 @@ def get_textbox(
             ) -> str:
         tp = textpage
         if tp is None:
-            tp = page.get_textpage()
+            # Leave out text that a clip path hides, as get_text() and
+            # search_for() do. Otherwise the default flags (0) are kept.
+            tp = page.get_textpage(flags=TEXT_MEDIABOX_CLIP)
         elif getattr(tp, "parent") != page:
             raise ValueError("not a textpage of this page")
         rc = tp.extractTextbox(rect)

src/utils.py

@@ -276,7 +276,9 @@ def get_textbox(
 ) -> str:
     tp = textpage
     if tp is None:
-        tp = page.get_textpage()
+        # Leave out text that a clip path hides, as get_text() and
+        # search_for() do. Otherwise the default flags (0) are kept.
+        tp = page.get_textpage(flags=pymupdf.TEXT_MEDIABOX_CLIP)
     elif getattr(tp, "parent") != page:
         raise ValueError("not a textpage of this page")
     rc = tp.extractTextbox(rect)
@@ -295,7 +297,10 @@ def get_text_selection(
     pymupdf.CheckParent(page)
     tp = textpage
     if tp is None:
-        tp = page.get_textpage(clip=clip, flags=pymupdf.TEXT_DEHYPHENATE)
+        tp = page.get_textpage(
+            clip=clip,
+            flags=pymupdf.TEXT_DEHYPHENATE | pymupdf.TEXT_MEDIABOX_CLIP,
+        )
     elif getattr(tp, "parent") != page:
         raise ValueError("not a textpage of this page")
     rc = tp.extractSelection(p1, p2)

tests/test_textextract.py

@@ -1033,3 +1033,27 @@ def test_4994():
             assert len(search_results) == 2
             #assert text == 'planet-warming gas driving climate change.'
     sys.stdout.flush()
+
+
+def test_textbox_clipped_text():
+    # Text hidden by a clip path is not on the rendered page. get_text() and
+    # search_for() leave it out; get_textbox() and get_text_selection() must
+    # leave it out too, unless the caller passes a textpage of their own.
+    src = pymupdf.open()
+    src_page = src.new_page(width=400, height=200)
+    src_page.insert_text((20, 50), 'VISIBLE left part', fontsize=12)
+    src_page.insert_text((250, 50), 'HIDDEN right part', fontsize=12)
+
+    doc = pymupdf.open()
+    page = doc.new_page(width=400, height=200)
+    left = pymupdf.Rect(0, 0, 200, 200)
+    page.show_pdf_page(left, src, 0, clip=left)  # clips away the right half
+
+    assert 'HIDDEN' not in page.get_text()
+    assert page.search_for('HIDDEN') == []
+    assert page.get_textbox(page.rect) == 'VISIBLE left part'
+    assert page.get_text_selection((0, 40), (400, 60)) == 'VISIBLE left part'
+
+    # An explicit textpage still decides what is extracted.
+    tp = page.get_textpage(flags=0)
+    assert 'HIDDEN' in page.get_textbox(page.rect, textpage=tp)