remove references from paper body

core/utils.py

@@ -1,2 +1,13 @@
+import regex as re
+
+
 def escape_newlines(text: str):
     return text.replace("\n", "\\n")
+
+def remove_references(text: str):
+    """
+    Removes reference sections from a scientific paper.
+    """
+    pattern = r"References\n([\s\S]*?)(?:\n\n|\Z)"
+    new_text = re.sub(pattern, '', text)
+    return new_text.strip()

tests/utils/test_regex.py

@@ -0,0 +1,26 @@
+from core.utils import remove_references
+import unittest
+
+class TestRemoveReferences(unittest.TestCase):
+
+    def test_remove_without_following_section(self):
+        text = """
+        Some initial text.
+        References
+        1. Reference one details.
+        2. Reference two details.
+        """
+        result = remove_references(text)
+        self.assertNotIn("References", result)
+        self.assertIn("Some initial text.", result)
+
+    def test_no_references_section(self):
+        text = """
+        Some initial text.
+        No references here.
+        """
+        result = remove_references(text)
+        self.assertEqual(text.strip(), result)
+
+if __name__ == '__main__':
+    unittest.main()