From 6a5e3eb969a0c15ec19281398cb5d89e6415f5dd Mon Sep 17 00:00:00 2001
From: kthoden <kthoden@gwdg.de>
Date: Mon, 11 Feb 2019 16:28:12 +0100
Subject: [PATCH] Math conversion

---
 fix_tei.py | 44 ++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 42 insertions(+), 2 deletions(-)

diff --git a/fix_tei.py b/fix_tei.py
index 963da3f..1797f93 100644
--- a/fix_tei.py
+++ b/fix_tei.py
@@ -176,6 +176,27 @@ def convert_citations(string, citedrangetext):
     return (string, citations)
 # def convert_citations ends here
 
+
+def convert_math(string):
+    """Find math shorthand using regex.
+
+    <formula notation="tex" rend="inline">2^2 = 4</formula>
+    """
+    formulae = []
+
+    math_pattern = re.compile(r"(\$|&#x24;|&#36;)(?P<contents>.+?)(\$|&#x24;|&#36;)")
+    found_math = re.findall(math_pattern, string)
+    logging.info(f"Found {len(found_math)} formulae.")
+
+    string = re.sub(math_pattern, r'<formula notation="tex" rend="inline">\g<contents></formula>', string)
+
+    for formula in found_math:
+        formulae.append(formula[1])
+
+    return string
+# def convert_math ends here
+
+
 def parse_cited_range(list_of_xml_elements):
     """citedRange: split up parameters or remove element if attributes are empty"""
 
@@ -340,6 +361,19 @@ def cleanup_xml(xml_tree):
     for attribute in xml_preserve:
         attribute.attrib.pop("{http://www.w3.org/XML/1998/namespace}space")
 
+    formulae = xmltree.xpath("//t:formula", namespaces=NS_MAP)
+    logging.info(f"Found {len(formulae)} formulae.")
+
+    for formula in formulae:
+        hi_in_formula = formula.xpath("t:hi", namespaces=NS_MAP)
+        for hi in hi_in_formula:
+            logging.debug(f"Found something in the formula: {etree.tostring(hi)}")
+            if hi.attrib["rend"] == "italic":
+                hi.attrib.pop("rend")
+            else:
+                logging.warning(f"Found another rend attribute: {hi.get('rend')}")
+            hi.tag = "tagtobestripped"
+
     empty_rend = xml_tree.xpath("//t:hi[(@rend='')]", namespaces=NS_MAP)
     logging.info(f"Found {len(empty_rend)} empty rend attributes.")
 
@@ -353,6 +387,10 @@ def cleanup_xml(xml_tree):
             xml_parent = attribute.getparent()
             attribute.tag = "tagtobestripped"
 
+    seg_element = xml_tree.findall("//t:seg", namespaces=NS_MAP)
+    for seg in seg_element:
+        seg.tag = "tagtobestripped"
+
     footnotes = xml_tree.xpath("//t:note[@place='foot']", namespaces=NS_MAP)
     for footnote in footnotes:
         footnote.set("place", "bottom")
@@ -642,14 +680,16 @@ def main():
 
     mod_string3 = convert_figures(mod_string2)
 
+    math_string = convert_math(mod_string3)
+
     debug_output = TMP_DIR + os.path.sep + args.teifile.replace(".xml", "-modified.xml")
     with open(debug_output, "w") as debugfile:
-        debugfile.write(mod_string3)
+        debugfile.write(math_string)
     logging.info("Wrote %s." % debug_output)
 
     # check for wellformedness, read again as xml
     try:
-        xml_tree2 = etree.fromstring(mod_string3)
+        xml_tree2 = etree.fromstring(math_string)
     except etree.XMLSyntaxError:
         logging.error("\nXML syntax error when trying to parse modified tree. Dumped it to %s." % debug_output)
         print("-"*60)