Merge pull request #186 from elifesciences/develop

PR for version 0.77.0 release
elifesciences · Jan 29, 2025 · b997fac · b997fac
2 parents e6f890d + 2e52e50
commit b997fac
Show file tree

Hide file tree

Showing 5 changed files with 97 additions and 1 deletion.
diff --git a/elifecleaner/__init__.py b/elifecleaner/__init__.py
@@ -1,7 +1,7 @@
 import logging
 
 
-__version__ = "0.76.0"
+__version__ = "0.77.0"
 
 
 LOGGER = logging.getLogger(__name__)

diff --git a/elifecleaner/equation.py b/elifecleaner/equation.py
@@ -103,6 +103,18 @@ def equation_inline_graphic_hrefs(sub_article_root, identifier):
     return href_list
 
 
+def formula_graphic_hrefs(sub_article_root, identifier):
+    "get disp-formula graphic href values"
+    sub_article_id, body_tag = block.sub_article_tag_parts(sub_article_root)
+    href_list = []
+    if body_tag is not None:
+        for graphic_tag in body_tag.findall(".//disp-formula/graphic"):
+            image_href = utils.xlink_href(graphic_tag)
+            if image_href:
+                href_list.append(image_href)
+    return href_list
+
+
 def inline_equation_inline_graphic_hrefs(sub_article_root, identifier):
     "get inline-graphic xlink:href values to be inline-formula"
     body_tag = block.sub_article_tag_parts(sub_article_root)[1]

diff --git a/elifecleaner/table.py b/elifecleaner/table.py
@@ -36,6 +36,18 @@ def table_inline_graphic_hrefs(sub_article_root, identifier):
     return href_list
 
 
+def table_graphic_hrefs(sub_article_root, identifier):
+    "get table-wrap graphic href values"
+    sub_article_id, body_tag = block.sub_article_tag_parts(sub_article_root)
+    href_list = []
+    if body_tag is not None:
+        for graphic_tag in body_tag.findall(".//table-wrap/graphic"):
+            image_href = utils.xlink_href(graphic_tag)
+            if image_href:
+                href_list.append(image_href)
+    return href_list
+
+
 def transform_table_group(body_tag, table_index, table_group, sub_article_id):
     "transform one set of p tags into table-wrap tags as specified in the table_group dict"
     inline_graphic_p_tag = body_tag[table_group.get("inline_graphic_index")]

diff --git a/tests/test_equation.py b/tests/test_equation.py
@@ -234,6 +234,40 @@ def test_equation_inline(self):
         self.assertEqual(result, expected)
 
 
+class TestFormulaGraphicHrefs(unittest.TestCase):
+    "tests for equation.formula_graphic_hrefs()"
+
+    def test_formula_graphic_hrefs(self):
+        "get a list of xlink:href values from disp-formula graphic tags"
+        xml_string = (
+            b'<sub-article id="sa1" xmlns:xlink="http://www.w3.org/1999/xlink"><body>'
+            b"<p>Following is a display formula:</p>\n"
+            b'<disp-formula id="sa1equ1">\n'
+            b'<graphic mimetype="image" mime-subtype="jpg"'
+            b' xlink:href="elife-sa1-equ1.jpg"/>\n'
+            b"</disp-formula>\n"
+            b"</body></sub-article>"
+        )
+        identifier = "test.zip"
+        tag = ElementTree.fromstring(xml_string)
+        expected = ["elife-sa1-equ1.jpg"]
+        result = equation.formula_graphic_hrefs(tag, identifier)
+        self.assertEqual(result, expected)
+
+    def test_graphic_hrefs_no_match(self):
+        "empty list of xlink:href values when there is no graphic tag"
+        xml_string = (
+            b'<sub-article id="sa1" xmlns:xlink="http://www.w3.org/1999/xlink">'
+            b"<body><p/></body>"
+            b"</sub-article>"
+        )
+        identifier = "test.zip"
+        tag = ElementTree.fromstring(xml_string)
+        expected = []
+        result = equation.formula_graphic_hrefs(tag, identifier)
+        self.assertEqual(result, expected)
+
+
 class TestInlineEquationInlineGraphicHrefs(unittest.TestCase):
     "tests for inline_equation_inline_graphic_hrefs()"
 

diff --git a/tests/test_table.py b/tests/test_table.py
@@ -62,6 +62,44 @@ def test_table_inline_graphic_hrefs(self):
         self.assertEqual(result, expected)
 
 
+class TestTableGraphicHrefs(unittest.TestCase):
+    "tests for table.table_graphic_hrefs()"
+
+    def test_table_graphic_hrefs(self):
+        "get a list of xlink:href values from table-wrap graphic tags"
+        xml_string = (
+            b'<sub-article id="sa1" xmlns:xlink="http://www.w3.org/1999/xlink"><body>'
+            b'<table-wrap id="sa1table1">\n'
+            b"<label>Review table 1.</label>\n"
+            b"<caption>\n"
+            b"<title>Caption title.</title>\n"
+            b"<p>Caption paragraph.</p>\n"
+            b"</caption>\n"
+            b'<graphic mimetype="image" mime-subtype="jpg"'
+            b' xlink:href="elife-95901-sa1-table1.jpg"/>\n'
+            b"</table-wrap>\n"
+            b"</body></sub-article>"
+        )
+        identifier = "test.zip"
+        tag = ElementTree.fromstring(xml_string)
+        expected = ["elife-95901-sa1-table1.jpg"]
+        result = table.table_graphic_hrefs(tag, identifier)
+        self.assertEqual(result, expected)
+
+    def test_graphic_hrefs_no_match(self):
+        "empty list of xlink:href values when there is no graphic tag"
+        xml_string = (
+            b'<sub-article id="sa1" xmlns:xlink="http://www.w3.org/1999/xlink">'
+            b"<body><p/></body>"
+            b"</sub-article>"
+        )
+        identifier = "test.zip"
+        tag = ElementTree.fromstring(xml_string)
+        expected = []
+        result = table.table_graphic_hrefs(tag, identifier)
+        self.assertEqual(result, expected)
+
+
 class TestTransformTable(unittest.TestCase):
     "tests for table.transform_table()"