Question

In a comparison function I am bascially looking for a pattern (e.g. “AAA”) inside a long binary object (for an example, aaaAAAbbbBBB)

I’m working backwards through the file (I know the match will be closer to the end than beginning), an adding 1 byte to the variable that is being checked for the match:

1. aaaAAAbbbBB[B]
2. aaaAAAbbbB[BB]
3. aaaAAAbbb[BBB]
4. aaaAAAbb[bBBB]
5. ... 
n. aaa[AAAbbbBBB] 

match found, offset = -n

Given that I know my pattern is 3 elements long, I wondered if I can simply window the search variable rather than incrementing it – it gets very slow when the match is +1,000,000 elements deep in the list – windowed view of the same data would be:

1. aaaAAAbbb[BBB]
2. aaaAAAbb[bBB]B
3. aaaAAAb[bbB]BB
4. aaaAAA[bbb]BBB
5. ...
n. aaa[AAA]bbbBBB

match found, offset = -n

My current search looks like:

if marker in f_data[-counter:]:
    offset = (len(f_data)-counter)+len(marker)
    return offset

In MATLAB I would have used the array addressing to move through the array,(e.g. calling window = a[5:8], window = a[4:7] etc) but I don’t think that’s possible in Python (2.7)

I can see a few suggestions for using a sliding window, ( Rolling or sliding window iterator in Python – this looks like a close match) but I can’t see how to implement it or they reference libs that I don’t know how to use.

Is there a built in function for doing this?

Answer 1

Why not just use rfind() or rindex()?

haystack = "aaaAAAbbbBBB"
needle   = "AAA"

pos = haystack.rfind(needle)

if pos >= 0:
    print "found at", pos - len(haystack)
else:
    print "not found"