From b02aad21480b1edb458ff907090dad097b4e1846 Mon Sep 17 00:00:00 2001
From: lhess2021 <lorenzohess@tutanota.com>
Date: Fri, 20 May 2022 23:25:19 -0400
Subject: [PATCH] add xml header. add c_num, canticle attr to <canto>. start
 addTEIheader() for later

---
 scripts.d/py-scripts.d/parser-txt-to-xml.py | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/scripts.d/py-scripts.d/parser-txt-to-xml.py b/scripts.d/py-scripts.d/parser-txt-to-xml.py
index 63b561c..b5ee18e 100644
--- a/scripts.d/py-scripts.d/parser-txt-to-xml.py
+++ b/scripts.d/py-scripts.d/parser-txt-to-xml.py
@@ -138,8 +138,12 @@ def wrapTags(tag, xmlCantoFile):
     used to add <tercets>, <canto>, and <text> tags around <canto> body
     """
     xmlCanto = getXMLFileName(xmlCantoFile)
+    cantoNum = str(int(xmlCanto.split('-')[0]))
+    canticle = xmlCanto.split('-')[1]
     if tag == 'TEI':
         openTag = '<TEI xmlns="http://www.tei-c.org/ns/1.0">\n'
+    elif tag == 'canto':
+        openTag = '<' + tag + ' canticle="' + canticle + '"' + ' c_num="' + cantoNum + '">\n'
     else:
         openTag = '<' + tag + '>\n'
     closeTag = '\n</' + tag + '>'
@@ -151,7 +155,7 @@ def wrapTags(tag, xmlCantoFile):
         xmlCantoContents.write(closeTag)
     xmlCantoContents.close()
 
-def addTEIHeaderTags(xmlCantoFile):
+def addTEIHeader(xmlCantoFile):
     """
     The following is the minimum required TEI header
     <TEI xmlns="http://www.tei-c.org/ns/1.0">
@@ -177,6 +181,16 @@ def addTEIHeaderTags(xmlCantoFile):
     """
     xmlCanto = getXMLFileName(xmlCantoFile)
 
+def addXMLHeader(xmlCantoFile):
+    xmlCanto = getXMLFileName(xmlCantoFile)
+    xmlHeader = '<?xml version="1.0" encoding="UTF-8"?>\n\n'
+    with open(xmlCanto, 'r') as xmlCantoContents:
+        contents = xmlCantoContents.read()
+    with open(xmlCanto, 'w') as xmlCantoContents:
+        xmlCantoContents.write(xmlHeader)
+        xmlCantoContents.write(contents)
+    xmlCantoContents.close()
+
 def addSpaces(xmlCantoFile):
     """
     Add a given number of spaces before each opening and closing tag.
@@ -238,9 +252,9 @@ def main():
             wrapTags('tercets', xmlCantoName)
             wrapTags('canto', xmlCantoName)
             wrapTags('text', xmlCantoName)
-            #addTEIHeaderTags()
             wrapTags('TEI', xmlCantoName)
             addSpaces(xmlCantoName)
+            addXMLHeader(xmlCantoName)
             os.remove(cantoFile) # don't delete txt files while debugging so you don't have to keep copying them