Make sure to preserve CDATA blocks in merging

Author: tmadlener

edm4hep/schema_evolution/merge_selection_xml.py

@@ -14,6 +14,31 @@ def escape_name_content(match):
     return f'name="{escaped_value}"'
 
 
+def preserve_cdata_sections(content):
+    """Replace CDATA sections with placeholders to preserve them through XML parsing"""
+    cdata_sections = []
+    placeholder_pattern = "CDATA_PLACEHOLDER_{}"
+
+    def replace_cdata(match):
+        cdata_sections.append(match.group(0))
+        return placeholder_pattern.format(len(cdata_sections) - 1)
+
+    # Find and replace all CDATA sections with placeholders
+    processed_content = re.sub(r"<!\[CDATA\[.*?\]\]>", replace_cdata, content, flags=re.DOTALL)
+    return processed_content, cdata_sections
+
+
+def restore_cdata_sections(content, cdata_sections):
+    """Restore CDATA sections from placeholders"""
+    placeholder_pattern = "CDATA_PLACEHOLDER_{}"
+
+    for i, cdata_section in enumerate(cdata_sections):
+        placeholder = placeholder_pattern.format(i)
+        content = content.replace(placeholder, cdata_section)
+
+    return content
+
+
 def escape_xml_content(content):
     """Escape < and > characters in XML content while preserving XML structure"""
     lines = content.split("\n")
@@ -43,9 +68,14 @@ def unescape_xml_output(content):
 podio_gen_file, manual_file, output_file = sys.argv[1:]
 
 with open(podio_gen_file, "r", encoding="utf-8") as f:
-    gen_content = escape_xml_content(f.read())
+    gen_content, gen_cdata = preserve_cdata_sections(f.read())
+
 with open(manual_file, "r", encoding="utf-8") as f:
-    manual_content = escape_xml_content(f.read())
+    manual_content, manual_cdata = preserve_cdata_sections(f.read())
+
+# Escape XML content for parsing
+gen_content = escape_xml_content(gen_content)
+manual_content = escape_xml_content(manual_content)
 
 gen_tree = ET.fromstring(gen_content)
 manual_tree = ET.fromstring(manual_content)
@@ -57,5 +87,10 @@ def unescape_xml_output(content):
 
 ET.indent(gen_tree, space="  ", level=0)
 output_content = unescape_xml_output(ET.tostring(gen_tree, encoding="unicode"))
+
+# Restore CDATA sections in the final output
+all_cdata = gen_cdata + manual_cdata
+output_content = restore_cdata_sections(output_content, all_cdata)
+
 with open(output_file, "w", encoding="utf-8") as f:
     f.write(output_content)