From 6b1db9a285526b692ac481eb06b785e5880e4a55 Mon Sep 17 00:00:00 2001
From: Ben Webb <bjwebb67@googlemail.com>
Date: Mon, 9 Jun 2014 12:06:49 +0100
Subject: [PATCH] [#8] Add code to recreate Country codelist from ISO XML

---
 convert.py            | 48 ++++++++++++++++++++++++++++++++++++++++++-
 templates/Country.xml |  7 +++++++
 2 files changed, 54 insertions(+), 1 deletion(-)
 create mode 100644 templates/Country.xml

diff --git a/convert.py b/convert.py
index cfdbb9f3..a75f6498 100644
--- a/convert.py
+++ b/convert.py
@@ -2,12 +2,18 @@
 
 Converts codelist files from external sources into the format used by IATI.
 
-Currently only supports the IANA Media Types code list (FileFormat).
+Note not all external codelists are converted automatically yet.
 
 """
 
 from lxml import etree as ET
 
+"""
+
+IANA Media Types (FileFormat)
+
+"""
+
 template = ET.parse('templates/FileFormat.xml', ET.XMLParser(remove_blank_text=True))
 codelist_items = template.find('codelist-items')
 
@@ -47,3 +53,43 @@ def indent(elem, level=0, shift=2):
 
 template.write('xml/FileFormat.xml', pretty_print=True)
 
+
+
+"""
+
+ISO Country Alpha 2
+
+"""
+
+XML_LANG = '{http://www.w3.org/XML/1998/namespace}lang'
+
+template = ET.parse('templates/Country.xml', ET.XMLParser(remove_blank_text=True))
+codelist_items = template.find('codelist-items')
+
+countries = ET.parse('source/iso_country_codes.xml')
+for country in countries.findall('country'):
+    if country.find('status').text == 'officially-assigned':
+        codelist_item = ET.Element('codelist-item')
+
+        code = ET.Element('code')
+        code.text = country.find('alpha-2-code').text
+        codelist_item.append(code)
+        
+        for short_name in country.findall('short-name'):
+            if XML_LANG in short_name.attrib:
+                name = ET.Element('name')
+                name.attrib[XML_LANG] = short_name.attrib[XML_LANG]
+                name.text = short_name.text
+                codelist_item.append(name)
+
+        for full_name in country.findall('full-name'):
+            if XML_LANG in full_name.attrib:
+                description = ET.Element('description')
+                description.attrib[XML_LANG] = full_name.attrib[XML_LANG]
+                description.text = full_name.text
+                codelist_item.append(description)
+
+        codelist_items.append(codelist_item)
+
+template.write('xml/Country.xml', pretty_print=True)
+
diff --git a/templates/Country.xml b/templates/Country.xml
new file mode 100644
index 00000000..5e6efe58
--- /dev/null
+++ b/templates/Country.xml
@@ -0,0 +1,7 @@
+<codelist name="Country" xml:lang="en" complete="0">
+  <metadata>
+    <name>Country</name>
+    <url>http://www.iso.org/iso/home/standards/country_codes.htm</url>
+  </metadata>
+  <codelist-items/>
+</codelist>