Preserve original graphic string as CDATA

Captions might contain xml markup
EditionOpenAccess · Mar 5, 2018 · e43eded · e43eded
1 parent 035a952
commit e43eded
Showing 1 changed file with 7 additions and 8 deletions.
diff --git a/prepare_tei.py b/prepare_tei.py
@@ -150,6 +150,7 @@ def validate_citations(used_citekeys, bibdata):
 def convert_figures(string):
     """Find figures shorthands"""
 
+    # negative lookbehind assertion. Real + characters must be escaped by \
     graphic_pattern = re.compile(r"(?<!\\)\+(.*?)\+")
 
     # +Fig.1CarteDuCielPotsdam!Glass photographic plate from the Carte
@@ -162,7 +163,7 @@ def convert_figures(string):
     logging.info("Found %s figures" % len(figures))
 
     for figure in figures:
-        string = re.sub(graphic_pattern, r"<graphic orig_string='\g<1>'/>", string)
+        string = re.sub(graphic_pattern, r"<graphic><![CDATA[\g<1>]]></graphic>", string)
 
     return string
 # def convert_figures ends here
@@ -177,14 +178,12 @@ def make_figure_elements(list_of_figures, figure_directory):
     for img in available_images_long:
         available_images.append(os.path.splitext(img)[0])
 
-    print(available_images)
-
     for figure in list_of_figures:
         parent_tag = figure.getparent()
         parent_tag.tag = "figure"
 
-        original_string = figure.get("orig_string")
-        figure.attrib.pop("orig_string")
+        original_string = figure.text
+        figure.clear()
 
         parts = original_string.split("!")
 
@@ -196,9 +195,9 @@ def make_figure_elements(list_of_figures, figure_directory):
             else:
                 bad_images.append(original_string)
 
-            head_element = etree.Element("head")
+            caption = "<head>" + parts[1] + "</head>"
+            head_element = etree.fromstring(caption)
             parent_tag.insert(1, head_element)
-            head_element.text = parts[1]
 
             if len(parts) == 3:
                 logging.info("This figure contains hyperimage directions")
@@ -316,7 +315,7 @@ def plural(num, noun):
 # def plural ends here
 
 def evaluate_report(report):
-    """Print collection of not found items"""
+    """Print report of conversion."""
 
     print("="*60)
     print(' '*4, "Conversion report")