Convert Intel ARK Product Specification Comparison XML into CSV

intel_ark_comparison_xml_converter.py

# coding: utf-8

import datetime
import re
import sys

import xmltodict

reload(sys)
sys.setdefaultencoding('utf-8')

if len(sys.argv) < 2:
    print "Usage: {} intel_ark.xml".format(sys.argv[0])
    sys.exit(1)

in_file_name = sys.argv[1]
out_file_name = re.sub('\.xml$', '.csv', in_file_name)

print("Loading {}...".format(in_file_name))
with open(in_file_name) as in_file:
    start = datetime.datetime.now()
    intel_dict = xmltodict.parse(in_file)

rows = intel_dict['ss:Workbook']['ss:Worksheet']['ss:Table']['ss:Row']

max_len = 0
keys_len = 0

for x in rows:
    cell = x['ss:Cell']
    if isinstance(cell, dict):
        continue
    else:
        max_len = max(max_len, len(cell))
        keys_len += 1

result = {}
keys = ['Name']
current_row = []
result['Name'] = current_row

for x in rows:
    cell = x['ss:Cell']
    if isinstance(cell, dict):
        data = cell['ss:Data']
        text = data.get('text', data.get('#text', None))
        # print("c=d {}", text)
    else:
        for i, item in enumerate(cell):
            if not item:
                continue
            elif isinstance(item, dict):
                if 'ss:Data' in item:
                    text = item['ss:Data']['#text']
                    if i == 0:
                        current_row = []
                        result[text] = current_row
                        keys.append(text)
                        continue
                    current_row.append(text)
                    # print('{:>4} {}'.format(i, text))
                else:
                    current_row.append(u'N/A')
            else:
                print item.__name__
        current_row = []
    # print('')

print "Saving to {}".format(out_file_name)

with open(out_file_name, "w+") as out_file:
    for j in range(0, len(keys) - 1):
        out_file.write(keys[j])
        if j != len(keys):
            out_file.write(',')
    out_file.write('\n')
    for i in range(0, max_len - 1):
        for j in range(0, len(keys) - 1):
            out_file.write(result[keys[j]][i].replace(',', ''))
            if j != len(keys):
                out_file.write(',')
        out_file.write('\n')
    out_file.flush()

print "Finished, converted {} rows".format(max_len - 1)