#Read input file
def read_file(filename):
	file = str(filename)
	record_reader = csv.reader(open(file, "rU"), delimiter = ',')
	catList = []
	testList = []
	isFirstLine = True
	for row in record_reader:
		if isFirstLine:
			length = int(len(row))
			ID = row
			ScientNameID = ID.index('species')
			LongID = ID.index('lon')
			LatID = ID.index('lat')
			CatNoID = ID.index('catalogNumber')
			InstCodeID = ID.index('institution')
			CountryID = ID.index('country')
			DateID = ID.index('earliestDateCollected')
			isFirstLine = False
				
		else:
			CatNo = row[CatNoID]
			ScientName = row[ScientNameID]
			Long = row[LongID]
			Lat = row[LatID]
			InstCode = row[InstCodeID]
			Country = row[CountryID]
			Date = row[DateID]
			if CatNo not in testList:
				testList.append(CatNo)
				CatNo = Specimen_Record(CatNo, ScientName, InstCode, Lat, Long, Country, Date)
				catList.append(CatNo)
	return catList
				

#Create and write new file, output
def write_file(records):
	if len(sys.argv) == 3:
		outputfile = sys.argv[2]
		recWriter = open(outputfile, 'w') # csv.writer(open(outputfile, 'w'), dialect='excel', delimiter = "\t")
	else:
		output_stream = sys.stdout
	for instance in records:
		if len(instance.ScientName) > 2:
			if len(instance.Long) > 1:
				if len(instance.Date) > 2:
					recWriter.write("%s\n" % '\t'.join([instance.InstCode, instance.ScientName, instance.Long, instance.Lat, instance.Date]))

#Here's where classes and functions are defined

class Specimen_Record:
	def __init__ (self, CatNo, ScientName, InstCode, Lat, Long, Country, Date):
		#Check to make sure species is the same#
		self.CatNo = CatNo
		self.ScientName = ScientName
		self.InstCode = InstCode
		self.Lat = Lat
		self.Long = Long
		self.Country = Country
		self.Date = Date
		
	def __str__ (self):
		summary_lines = [self.ScientName + self.CatNo + self.Long + self.Lat]
		return '\n'.join(summary_lines)
		

#Here's where the action is
import sys, csv
program_name = sys.argv[0]
if len(sys.argv) > 3 or len(sys.argv) < 2:
    sys.exit(program_name + ": Expecting two arguments: [filename], [outputfile]")
#filename should be csv, outputfile should be txt
    
if len(sys.argv) == 3:
    filename = sys.argv[1]
else:
    filename = sys.stdin

records = read_file(filename)
print len(records)

write_file(records)
#Now you have a txt file that can be imported and saved in excel