Code to parse voter list pdf - ocred by tesseract

parse-voter-list.py

import sys

in_file = sys.argv[1]


content = open(in_file).read()
out = open("result.csv","a")

con  = content.split("வாக்காளர்‌ பெயர்‌")

csv_string = ""

line = 1
for item in con:

    lines = item.split("\n")

    for line in lines:
    #print(item)
        #print(str(line))
        if line.strip().startswith(":"):
            print("line = " + line)
            voter_name = line.split(":")[1]
            print(voter_name)
            csv_string = csv_string + voter_name + ","

        if "கணவர்" in line:
            print("line = " + line)
            if ":" in line:
                husband_name = line.split("கணவர்")[1].split(":")[1]
                print(husband_name)
                csv_string = csv_string + husband_name + ","


        if "தந்தை" in line:
            print("line = " + line)
            if ":" in line:
                father_name = line.split("தந்தை")[1].split(":")[1]
                print(father_name)
            else:
                father_name = line.split("பெயர்‌")[1]

            csv_string = csv_string + father_name + ","



        if "வயது" in line:
            print("line = " + line)
            if "இனம்‌" in line:
                age = str(line.split("வயது")[1].split("இனம்‌")[0].split(":")[1])
                print(age)
                csv_string = csv_string + age + ","

            


        if "இனம்‌" in line:
            print("line = " + line)
            sex = line.split("இனம்‌")[1].split(":")[1]
            print(sex)
            csv_string = csv_string + sex + ","

            
    csv_string = csv_string + "\n"
#        print("====")
      
        
print(csv_string)      
out.write(csv_string)
#வயது: 63 இனம்‌: பெண்‌