canada_payroll.py

from mechanize import Browser
from lxml import html as HTML
import pprint


ENDPOINT = "https://apps.cra-arc.gc.ca/ebci/rhpd/startLanguage.do?lang=English"


class PayrollCalculator(object):
    def __init__(self):
        self.b = Browser()
        self.b.set_handle_robots(False)
        self.b.open(ENDPOINT)   

    def calculate(self):
        # 1) Pay Period
        print "PAY"
        print self.b.geturl()
        self.b.select_form(name="payrollData")
        self.b.form["year"] = ["4"]  # January 1, 2009
        self.b.form["employerName"] = "Handi"
        self.b.form["payPeriodEndingDate"] = "2009-02-01" # YYYY-MM-DD
        self.b.form["province"] = ["9"]  # British Columbia
        self.b.form["payPeriod"] = ["2"]  # Biweekly
        self.b.submit(name="fwdSalary")

        # 2) Salary / Bonus Etc.
        print "SALARY"
        print self.b.geturl()
        self.b.select_form(name="payrollData")
        self.b.form["employeeName"] = "John"
        self.b.submit(name="fwdGrossSalary")

        # 3) Gross Income
        print "GROSS"
        print self.b.geturl()
        self.b.select_form(name="grossIncomeData")
        self.b.form["incomeTypeAmount"] = "4000"
        self.b.submit()

        # 4) Salary / Bonus Etc. again
        print "SALARY"
        print self.b.geturl()
        self.b.select_form(name="payrollData")
        self.b.submit(nr=3)  # 3 is the magic number for the Calculate button

        # 5) Results page. Scraping time.
        self.doc = HTML.fromstring(self.b.response().read(), self.b.geturl())
        self._needle()    
    
    def _needle(self):
        # The data we want is trapped somewhere around here.
        fields = ["Salary or wages for the pay period",
                  "Total EI insurable earnings for the pay period",
                  "Taxable income",
                  "Cash income for the pay period",
                  "Federal tax deductions",
                  "Provincial tax deductions",
                  "Requested additional tax deduction",
                  "Total tax on income",
                  "CPP deductions",
                  "EI deductions",
                  "Amounts deducted at source",
                  "Total deductions on income"]
                  
        values = [td.text.strip() for td in self.doc.xpath("//table[3]//td[2]")]
        pprint.pprint(zip(fields, values))
        
        
pc = PayrollCalculator()
pc.calculate()