This puzzle was contributed by Shane Emmons and published on March 15, 2011

UPDATE 1: Due to a BigDecimal bug in all existing implementations of Matz's Ruby Interpreter, our solution is incorrect. Those who have passing solutions to this problem have been bitten by the same bug. We will soon upload a solution that should match what all other Ruby implementations produce, which should also match correct solutions from other languages.

UPDATE 2: Updated solution file so that calculation is correct. We implemented our own rounding function rather than using BigDecimal's. If you're using Matz's Ruby, you'll need to do the same. Folks who submitted earlier will still be given credit for solving the problem, of course.

You have been given two files. The first is an XML file containing the conversion rates for exchanging one currency with another. The second is a CSV file containing sales data by transaction for an international business. Your goal is to parse all the transactions and return the grand total of all sales for a given item.

## Problem

What is the grand total of sales for item `DM1182` across all stores in `USD` currency?

## Notes

• After each conversion, the result should be rounded to 2 decimal places using bankers rounding.
• Some conversion rates are missing; you will need to derive them using the information provided.
• Since we are working with financial transactions, we need to avoid floating point arithmetic errors.

## Example

Given a `RATES.xml` of:

``````<?xml version="1.0"?>
<rates>
<rate>
<from>AUD</from>
<conversion>1.0079</conversion>
</rate>
<rate>
<to>USD</to>
<conversion>1.0090</conversion>
</rate>
<rate>
<from>USD</from>
<conversion>0.9911</conversion>
</rate>
</rates>
``````

and a `TRANS.csv` of:

``````store,sku,amount
Yonkers,DM1210,70.00 USD
Yonkers,DM1182,19.68 AUD
Nashua,DM1182,58.58 AUD
Scranton,DM1210,68.76 USD
Camden,DM1182,54.64 USD
``````

You program should return `134.22` as the total of `DM1182` in `USD`. You can use the sample files from this example to test your program before submitting an answer for the real dataset for this problem.