Find the longest string without repeating a letter

I've written some code to scan large files and search for long strings that you can write without repeating a letter. I'm self-taught in python and I'm wondering what I've missed.

import sys
letters ={} #letters is a dictionary
currentResult=""
ideas=[]
with open(sys.argv[1]) as f:
  while True:
    c = f.read(1)
    if not c:
      print "End of file"
      break
    if c in letters:
        letters={}
        temp=currentResult.rsplit(' ',1)[0]
        ideas.append(temp.rstrip('?:!.,;'))
        currentResult=""
    currentResult+=c
    if not c.isspace():
        letters[c]=True
ideas=sorted(set(ideas))
ideas.sort(lambda x,y: cmp(len(x), len(y)))
for x in ideas:
    print x.strip()

I do believe the algorithm in the pervious answer and in the OP does not do what the OP wanted. It test for the longest string starting with the line's first character, and then repeat from the first non repeating string. For example if the string is

abcaefgb

The algorithm given will return

abc

while the right answer should be

bcaefg

I believe the following algorithm will find the right string. Note that if there are several equal length longest strings it will return the first

def scantillrepeat(line):
    found = ''
    for char in line:
        if not char in found:
            found = found + char
        else:
            break
    return found

def findlongest(f):
    for line in f:
        longestfound = ''
        longestfoundlen = 0
        for k in range(len(line)):
            candidate = scantillrepeat(line[k:])
            if len(candidate) > longestfoundlen:
                longestfound = candidate
                longestfoundlen = len(candidate)
    return longestfound

To test

In [39]: f = ['abcaefgb']

In [40]: findlongest(f)
Out[40]: 'bcaefg'

Note that my version does not allow joining lines separated by newline \n. This should be altered if needed.

StackExchange Code Review Q#133943, answer score: 4