fix: preserve all PubMed structured abstract sections (#123)

Author: njzjz-bot

tests/test_pubmed_structured_abstract.py

@@ -0,0 +1,58 @@
+"""Regression tests for PubMed structured abstracts."""
+
+from __future__ import annotations
+
+from wenxian.feeder.pubmed import Pubmed
+
+
+def _xml(abstract: str) -> bytes:
+    return f"""
+<PubmedArticleSet>
+  <PubmedArticle>
+    <MedlineCitation>
+      <Article>
+        <ArticleTitle>Example.</ArticleTitle>
+        {abstract}
+        <Journal>
+          <Title>Example Journal</Title>
+          <JournalIssue><PubDate><Year>2024</Year></PubDate></JournalIssue>
+        </Journal>
+      </Article>
+    </MedlineCitation>
+    <PubmedData />
+  </PubmedArticle>
+</PubmedArticleSet>
+""".encode()
+
+
+def test_structured_abstract_keeps_all_sections():
+    """Test all sibling AbstractText sections are preserved in document order."""
+    reference = Pubmed()._from_content(
+        _xml(
+            """
+        <Abstract>
+          <AbstractText Label="BACKGROUND">Background text.</AbstractText>
+          <AbstractText Label="METHODS">Methods <i>nested</i> text.</AbstractText>
+          <AbstractText Label="RESULTS">Results text.</AbstractText>
+        </Abstract>
+        """
+        )
+    )
+    assert reference is not None
+    assert reference.annote == "Background text. Methods nested text. Results text."
+
+
+def test_single_abstract_section_is_unchanged():
+    """Test a conventional single-section abstract retains its content."""
+    reference = Pubmed()._from_content(
+        _xml("<Abstract><AbstractText>Only section.</AbstractText></Abstract>")
+    )
+    assert reference is not None
+    assert reference.annote == "Only section."
+
+
+def test_missing_abstract_remains_none():
+    """Test records without an abstract still produce no abstract value."""
+    reference = Pubmed()._from_content(_xml(""))
+    assert reference is not None
+    assert reference.annote is None

wenxian/feeder/pubmed.py

@@ -152,7 +152,16 @@ def _from_content(
             return None
         rets = {}
         for key, path in self.PUBMED_PATH.items():
-            if key != "author":
+            if key == "abstract":
+                abstract_sections = [
+                    self._text(node)
+                    for node in tree.findall(self.PUBMED_PATH["abstract"])
+                ]
+                rets[key] = (
+                    " ".join(section for section in abstract_sections if section)
+                    or None
+                )
+            elif key != "author":
                 rets[key] = self._text(tree.find(path))
 
         if rets["journal"] == "Physical chemistry chemical physics : PCCP":