#5 Support for endnotes

We have support for footnotes but we need to be able to read endnotes
also. Just like footnotes, endnotes are placed in separate file
endnotes.xml. We need to parse it and store in similar to footnotes.

Author: aerkalov

ooxml/doc.py

@@ -154,6 +154,7 @@ def reset(self):
         self.elements = []
         self.relationships = {}
         self.footnotes = {}
+        self.endnotes = {}
         self.numbering = {}
         self.abstruct_numbering = {}
         self.styles = StylesCollection()
@@ -290,6 +291,18 @@ def value(self):
         return self.rid
 
 
+class Endnote(Element):
+    "Represents endnote element."
+
+    def __init__(self, rid):
+        super(Endnote, self).__init__()
+
+        self.rid = rid
+
+    def value(self):
+        return self.rid
+
+
 class TextBox(Element):
     "Represents TextBox element."
 

ooxml/parse.py

@@ -196,6 +196,15 @@ def parse_footnote(document, container, elem):
     container.elements.append(foot)
 
 
+def parse_endnote(document, container, elem):
+    "Parse the endnote element."
+
+    _rid =  elem.attrib[_name('{{{w}}}id')]
+
+    note = doc.Endnote(_rid)
+    container.elements.append(note)
+
+
 def parse_alternate(document, container, elem):
     txtbx = elem.find('.//'+_name('{{{w}}}txbxContent'))
     paragraphs = []
@@ -256,6 +265,11 @@ def parse_text(document, container, element):
     if foot is not None:
         parse_footnote(document, container, foot)
 
+    end = element.find(_name('{{{w}}}endnoteReference'))
+
+    if end is not None:
+        parse_endnote(document, container, end)
+
     sym = element.find(_name('{{{w}}}sym'))
 
     if sym is not None:
@@ -522,6 +536,21 @@ def parse_footnotes(document, xmlcontent):
         document.footnotes[style.attrib[_name('{{{w}}}id')]] = p
 
 
+def parse_endnotes(document, xmlcontent):
+    """Parse endnotes document.
+
+    Endnotes are defined in file 'endnotes.xml'
+    """
+
+    endnotes = etree.fromstring(xmlcontent)
+    document.endnotes = {}
+
+    for note in endnotes.xpath('.//w:endnote', namespaces=NAMESPACES):
+        paragraphs = [parse_paragraph(document, para) for para in note.xpath('.//w:p', namespaces=NAMESPACES)]
+
+        document.endnotes[note.attrib[_name('{{{w}}}id')]] = paragraphs
+
+
 def parse_numbering(document, xmlcontent):
     """Parse numbering document.
 
@@ -590,6 +619,12 @@ def parse_from_file(file_object):
     except KeyError:
         logger.warning('Could not read footnotes.')
 
+    try:    
+        endnotes_content = file_object.read_file('endnotes.xml')
+        parse_endnotes(document, endnotes_content)    
+    except KeyError:
+        logger.warning('Could not read endnotes.')
+
     try:
         numbering_content = file_object.read_file('numbering.xml')
         parse_numbering(document, numbering_content)    

ooxml/serialize.py

@@ -347,8 +347,9 @@ def get_style_fontsize(node):
       Font size as int number or 0 if it is not defined
 
     """
-    if 'sz' in node.rpr:
-        return int(node.rpr['sz']) / 2
+    if hasattr(node, 'rpr'):
+        if 'sz' in node.rpr:
+            return int(node.rpr['sz']) / 2
 
     return 0
 
@@ -697,6 +698,27 @@ def serialize_footnote(ctx, document, el, root):
     return root
 
 
+def serialize_endnote(ctx, document, el, root):
+    "Serializes endnotes."
+
+    footnote_num = el.rid
+
+    if el.rid not in ctx.endnote_list:
+        ctx.endnote_id += 1
+        ctx.endnote_list[el.rid] = ctx.endnote_id
+
+    footnote_num = ctx.endnote_list[el.rid]
+
+    note = etree.SubElement(root, 'sup')
+    link = etree.SubElement(note, 'a')
+    link.set('href', '#')
+    link.text = u'{}'.format(footnote_num)
+
+    fire_hooks(ctx, document, el, note, ctx.get_hook('endnote'))
+
+    return root
+
+
 def serialize_table(ctx, document, table, root):
     """Serializes table element.
     """
@@ -907,6 +929,7 @@ def get_header(self, elem, style, node):
         doc.Break: serialize_break,
         doc.TextBox: serialize_textbox,
         doc.Footnote: serialize_footnote,
+        doc.Endnote: serialize_endnote,
         doc.Symbol: serialize_symbol
     },
 
@@ -997,6 +1020,8 @@ def reset(self):
 
         self.footnote_id = 0
         self.footnote_list = {}
+        self.endnote_id = 0
+        self.endnote_list = {}
 
         self.in_list = []
         self.header = self.options['header']()