From 9cdcda60aa6490337637c7241fb8f5a6c5c23310 Mon Sep 17 00:00:00 2001
From: "Jorj X. McKie" <jorj.x.mckie@outlook.de>
Date: Sat, 10 Oct 2026 07:11:32 -0400
Subject: [PATCH] Port 4llm functions to PyMuPDF API

This fix makes PyMuPDF4LLM's API functions available in PyMuPDF.
---
 src/__init__.py              | 41 ++++++++++++++++++++++++++++++++++--
 tests/test_configurations.py | 30 ++++++++++++++++++++++++++
 2 files changed, 69 insertions(+), 2 deletions(-)
 create mode 100644 tests/test_configurations.py

diff --git a/src/__init__.py b/src/__init__.py
index 4fae080dc..66853afc1 100644
--- a/src/__init__.py
+++ b/src/__init__.py
@@ -26742,7 +26742,44 @@ def deprecated_function( *args, **kwargs):
         f'Python {sys.version_info[0]}.{sys.version_info[1]} running on {sys.platform} ({64 if sys.maxsize > 2**32 else 32}-bit).\n'
         )
 
+
 try:
     import pymupdf4llm   # noqa: F401
-except ImportError:
-    pass
+except (ImportError, ModuleNotFoundError):
+    pymupdf4llm = None
+
+def _to_markdown(*args, **kwargs):
+    if pymupdf4llm:
+        return pymupdf4llm.to_markdown(*args, **kwargs)
+    else:
+        raise NotImplementedError('pymupdf4llm functions are not available.')
+
+def _to_text(*args, **kwargs):
+    if pymupdf4llm:
+        return pymupdf4llm.to_text(*args, **kwargs)
+    else:
+        raise NotImplementedError('pymupdf4llm functions are not available.')
+
+def _to_json(*args, **kwargs):
+    if pymupdf4llm:
+        return pymupdf4llm.to_json(*args, **kwargs)
+    else:
+        raise NotImplementedError('pymupdf4llm functions are not available.')
+
+def _to_chunks(*args, **kwargs):
+    if pymupdf4llm:
+        return pymupdf4llm.to_chunks(*args, **kwargs)
+    else:
+        raise NotImplementedError('pymupdf4llm functions are not available.')
+
+def _convert_batch(*args, **kwargs):
+    if pymupdf4llm:
+        return pymupdf4llm.convert_batch(*args, **kwargs)
+    else:
+        raise NotImplementedError('pymupdf4llm functions are not available.')
+
+Document.to_markdown = _to_markdown
+Document.to_text = _to_text
+Document.to_json = _to_json
+Document.to_chunks = _to_chunks
+convert_batch = _convert_batch
diff --git a/tests/test_configurations.py b/tests/test_configurations.py
new file mode 100644
index 000000000..df3ff4d2e
--- /dev/null
+++ b/tests/test_configurations.py
@@ -0,0 +1,30 @@
+import pymupdf
+import pathlib
+import util
+
+
+def test_configuration_layout():
+    if not util._use_layout() :
+        print("not testing: layout feature not available")
+        return
+
+    doc = pymupdf.open()
+    page = doc.new_page()
+    page.insert_text((72, 72), "Hallo Welt!\nSeid umschlungen, Millionen!")
+
+    out1 = doc.to_markdown()
+    out2 = pymupdf.pymupdf4llm.to_markdown(doc)
+    assert out1 == out2
+
+    out1 = doc.to_text()
+    out2 = pymupdf.pymupdf4llm.to_text(doc)
+    assert out1 == out2
+
+    out1 = doc.to_chunks()
+    out2 = pymupdf.pymupdf4llm.to_chunks(doc)
+    assert repr(out1) == repr(out2), f"{out1=} != {out2=}"
+
+    out1 = doc.to_json()
+    out2 = pymupdf.pymupdf4llm.to_json(doc)
+    assert out1 == out2
+