from sys import argv, exit
import re
if len(argv) < 2:
    print("mising command-line argument")
    exit(1)

pattern1 = re.compile(r'AGAT')
pattern2 = re.compile(r'AATG')
pattern3 = re.compile(r'TATC')
#with open("small.csv", "r") as file:
#with open(sequences/"1.txt", "r") as f:
#with open(argv[1], "r") as f:
    #contents = f.read()
    #matches = pattern2.finditer(contents)
    #for match in matches:
        #print(match)
#with open(argv[1], "r") as f:
    #contents = f.read()
    #matches = pattern3.finditer(contents)
    #for match in matches:
        #print(match)

#print("Here is a list of all the times the string TATC occurs")
with open(argv[1], "r") as f:
    count = 0
    contents = f.read()
    print("contents prints: ")
    print(contents)
    #print(contents[0:4])
    i = 0
    j = i + 4
    while contents[i:j]: # will read contents until end of file
        span = contents[i:j]
        #count = 0
        #sample output for this count

        #span GAAA reapeats 0 consecutive times
        #span GGAG reapeats 0 consecutive times
        #span GGAT reapeats 0 consecutive times

        while contents[i+4:j+4] == span:
            count += 1
            #print("span reapeats " + str(count) + " times" )
            i += 4
            j += 4
            #count = 0 # resetting count to 0 as suggested

            #sample output for this count

            #span GAAA reapeats 0 consecutive times
            #span GGAG reapeats 0 consecutive times
            #span GGAT reapeats 0 consecutive times
        #count = 0
        #count +=1
        i += 4
        j += 4
        print("span " + span + " reapeats " + str(count) + " consecutive times" )
        #count = 0
    #count = 0 sample output remains unchanged
    #print("span " + span + " reapeats " + str(count) + " consecutive times" )