fix: corrige extração de texto de abstract em document.py

publication/utils/document.py

@@ -169,7 +169,7 @@ def get_abstracts(self):
                     continue
                 yield {"language": lang, "text": text}
         except Exception as e:
-            return []
+            return
 
     def get_keywords(self):
         for lang, keywords in (

publication/utils/test_document.py

@@ -110,3 +110,98 @@ def test_get_contribs_with_no_contribs(self):
 
         self.assertEqual(result["names"], [])
         self.assertEqual(result["collabs"], [])
+
+
+class XMLArticleGetAbstractsTest(TestCase):
+    def test_get_abstracts_with_p_tag(self):
+        xml_string = """<article xml:lang="en">
+            <front>
+                <article-meta>
+                    <abstract>
+                        <title>Abstract</title>
+                        <p>Simple abstract with a p tag.</p>
+                    </abstract>
+                    <trans-abstract xml:lang="pt">
+                        <title>Resumo</title>
+                        <p>Resumo simples com uma tag p.</p>
+                    </trans-abstract>
+                </article-meta>
+            </front>
+        </article>"""
+
+        article_xml = _create_xml_article(xml_string)
+        result = list(article_xml.get_abstracts())
+
+        self.assertEqual(
+            result,
+            [
+                {"language": "en", "text": "Simple abstract with a p tag."},
+                {"language": "pt", "text": "Resumo simples com uma tag p."},
+            ],
+        )
+
+    def test_get_abstracts_with_sections(self):
+        xml_string = """<article xml:lang="en">
+            <front>
+                <article-meta>
+                    <abstract>
+                        <title>Abstract</title>
+                        <sec><title>Objective:</title><p>obj text.</p></sec>
+                        <sec><title>Method:</title><p>method text.</p></sec>
+                    </abstract>
+                </article-meta>
+            </front>
+        </article>"""
+
+        article_xml = _create_xml_article(xml_string)
+        result = list(article_xml.get_abstracts())
+
+        self.assertEqual(len(result), 1)
+        self.assertEqual(result[0]["language"], "en")
+        self.assertEqual(
+            result[0]["text"], "Objective: obj text. Method: method text."
+        )
+
+    def test_get_abstracts_without_p_wrapper_regression_issue_1031(self):
+        """Regression test for scieloorg/scms-upload#1031: legacy migrated
+        XML (e.g. URY collection) may have abstract text sitting directly
+        under <abstract>/<trans-abstract>, not wrapped in <p>. Extraction
+        for this case lives in packtools (scieloorg/packtools#1272,
+        released in 4.16.11); this test only confirms it comes through
+        get_abstracts() transparently."""
+        xml_string = """<article xml:lang="es">
+            <front>
+                <article-meta>
+                    <abstract>
+                        <title>Resumen:</title> Texto do resumo sem tag p.
+                    </abstract>
+                    <trans-abstract xml:lang="en">
+                        <title>Abstract:</title> Abstract text without a p tag.
+                    </trans-abstract>
+                </article-meta>
+            </front>
+        </article>"""
+
+        article_xml = _create_xml_article(xml_string)
+        result = list(article_xml.get_abstracts())
+
+        self.assertEqual(
+            result,
+            [
+                {"language": "es", "text": "Texto do resumo sem tag p."},
+                {"language": "en", "text": "Abstract text without a p tag."},
+            ],
+        )
+
+    def test_get_abstracts_with_no_abstract(self):
+        xml_string = """<article xml:lang="en">
+            <front>
+                <article-meta>
+                </article-meta>
+            </front>
+        </article>"""
+
+        article_xml = _create_xml_article(xml_string)
+        result = list(article_xml.get_abstracts())
+
+        self.assertEqual(result, [])

requirements/base.txt

@@ -90,7 +90,7 @@ iso639-lang==2.6.3  # Mantendo versão maior
 # SciELO Specific Packages
 # ========================================
 # Using specific versions for stability
--e git+https://github.com/scieloorg/packtools.git@4.16.8#egg=packtools
+-e git+https://github.com/scieloorg/packtools.git@4.16.11#egg=packtools
 -e git+https://github.com/scieloorg/scielo_scholarly_data#egg=scielo_scholarly_data
 -e git+https://github.com/scieloorg/opac_schema.git@v2.66#egg=opac_schema
 -e git+https://github.com/scieloorg/scielo_migration.git@1.10.8#egg=scielo_classic_website