input.py

#-------------------------------------------------------------------------------
# Name:        input
# Purpose:
#
# Author:      Vaibhav Kapoor and Adrian Barbur
#
# Created:     22/11/2012
# Licence:     GPL
#-------------------------------------------------------------------------------
#!/usr/bin/env python
# input.py

import re,sys,dateutil.parser as dateutil

class Input:
  def __init__(self, filename):
    self.csvfile = open(filename, 'r')
    self.contents = self.csvfile.read()
    self.initPatterns()

    valid = re.search(self.pattern, self.contents, re.MULTILINE)
    rows = re.findall(self.rowPattern, self.contents, re.MULTILINE)
    evals = re.findall(self.evalPattern, self.contents, re.MULTILINE)

    self.valid_file = True if valid else False
    print 'Valid file' if self.valid_file else 'Not a valid file'
    self.tr = map(toTr,rows)
    self.dates = map(toDate,rows)

    self.start = self.dates[0] if not evals else dateutil.parse(evals[0][0])
    self.end = self.dates[-1] if not evals else dateutil.parse(evals[0][1])


  def initPatterns(self):
    datePattern = ('(?:\d\d\d\d-\d\d-\d\d)')
    qp = quotePattern('(?:Name: )')
    dp = quotePattern('(?:Description: )')
    ap = quotePattern('(?:Account#: )')
    ep = quotePattern('(?:Email: )')
    adp = quotePattern('(?:Address: )')
    pp = quotePattern('(?:Phone: )')
    evp = quotePattern('(?:Evaluation Period: )')

    namePattern = r'(?:(?:'+qp+')[^\n]*\n)'
    descPattern = r'(?:(?:'+dp+')[^\n]*\n)?'
    accPattern = r'(?:(?:'+ap+')[^\n]*\n)?'
    emailPattern = r'(?:(?:'+ep+')[^\n]*\n)?'
    addrPattern = r'(?:(?:'+adp+')[^\n]*\n)?'
    phonePattern = r'(?:(?:'+pp+')[^\n]*\n)?'
    colPattern = r'(?:Transaction Date,Market Value,Cash Flow,Agent Fees,Benchmark,\n)'

    floatPattern = quotePattern(r'(?:[-+]?(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][-+])?)')
    # TODO: Ask if benchmark needs to have a % sign or can be represented as decimal.
    benchPattern = quotePattern(r'(?:[-+]?(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][-+])?%)')

    self.evalPattern = r'(?:(?:'+evp+')'+'('+datePattern+')'+' to '+'('+datePattern+')'+',*\n)'
    self.rowPattern = r'(?:(^'+datePattern + '),' +'('+floatPattern+'),'+'('+floatPattern+')?,'+'('+floatPattern+')?,'+'('+benchPattern+')?,*\n)'
    self.pattern = namePattern + descPattern + accPattern + emailPattern + addrPattern + phonePattern + self.evalPattern + '?'
    self.pattern = self.pattern + '(?:[,\n]*)' + colPattern + self.rowPattern + '+(?:[,\n]*)'


def stringToFloat(string, default):
  if string:
    return float(string)
  else:
    return default

def quotePattern(pattern):
  return r'''(?:[^,"']*'''+ pattern+''')|(?:"[^"']*'''+ pattern +'''[^"']*")'''

def toTr(row):
  date = dateutil.parse(row[0])
  mv = re.sub('[,"]*','',row[1])
  mv = float(mv)
  cf = stringToFloat(row[2],0)
  af = stringToFloat(row[3],0)
  bench = re.sub('[%]*','',row[4])
  bench = stringToFloat(bench,None)
  bench = bench/100. if bench else bench

  return (date,mv,cf,af,bench)

def toDate(row):
  return dateutil.parse(row[0])

def main():
    i = Input('input.csv')

if __name__ == '__main__':
    main()