Freelancer Pani7

index.py

from lxml import etree

import sys
import os

if len(sys.argv) != 2:
    sys.exit("Make sure to put the path of the file as an argument. Or to drag the file into the script.")

# the input is what we get as an argument, the ouput is that same file, in that same location, 
#   with .xml instead of its current extension
infile = sys.argv[1]
outfile = os.path.splitext(infile)[0]+".index.xml"

doc = etree.parse(infile)

# dictionary of indecies
indecies = {}

# go through every shop individually, getting their address and splitting for keywords, saving
#   the number into the indecies dictionary for all keywords
for shop in doc.getroot():
    number = shop.get("number")
    addr = shop.find("address").text.lower()
    parts = addr.split(" ")

    for part in parts:
        if part in indecies:
            indecies[part].append(number)
        else:
            indecies[part] = [number]

# Creating the index xml with lxml 
out = etree.Element("index")
doc = etree.ElementTree(out)


for entry in indecies:
    el = etree.SubElement(out, "entry")
    etree.SubElement(el, "keyword").text = entry
    accounts = etree.SubElement(el, "accounts")
    for x in indecies[entry]:
        etree.SubElement(accounts, "number").text = x

outFile = open(outfile, 'w')
outFile.write(etree.tostring(doc,pretty_print=True))

toxml.py

from lxml import etree
import json
import sys
import os

if len(sys.argv) != 2:
    sys.exit("Make sure to put the path of the file as an argument. Or to drag the file into the script.")

# the input is what we get as an argument, the ouput is that same file, in that same location, 
#   with .xml instead of its current extension
infile = sys.argv[1]
outfile = os.path.splitext(infile)[0]+".xml"

shops = {}
with open(infile,"r") as f:
    lines = f.readlines()

    for json_line in lines:
        line = json.loads(json_line)

        if "index" in line: #skip the ones that are {"index":{...}}
            continue

        shop_number = str(line['shop_number'])
        shops[shop_number] = line #store the shop in our temporary dictionary, by shop_number

#sort the keys, so that we order the shops by number from least to greatest
k = shops.keys()
k.sort()

#create the XML file and insert the shops and their data following the format discussed over chat.
out = etree.Element("shop_numbers")
doc = etree.ElementTree(out)

for shop_number in k:
    shop = shops[shop_number]
    el = etree.SubElement(out,"shop", number=shop_number)
    
    addr_el = etree.SubElement(el, "address")
    addr_el.text = shop['address']

    email_el = etree.SubElement(el, "email")
    email_el.text = shop['email']

    city_el = etree.SubElement(el, "city")
    city_el.text = shop['city']
    
    state_el = etree.SubElement(el, "state")
    state_el.text = shop['state']

out_file = open(outfile, 'w')
out_file.write(etree.tostring(doc,pretty_print=True))