Question

I need to parse a large log file (flat file), which contains two column of values (column-A , column-B).

Values in both columns are repeating. I need to find for each unique value in column-A , I need to find a set of column-B values.

Is this can be done using unix shell command or need to write any perl or python script? What are the ways this can be done?

Example:

xxxA 2
xxxA 1
xxxB 2
XXXC 3
XXXA 3
xxxD 4

output:

xxxA - 2,1,3
xxxB - 2
xxxC - 3
xxxD - 4

Answer 1

I would use Python dictionaries where the dictionary keys are column A values and the dictionary values are Python’s built-in Set type holding column B values

def parse_the_file():
    lower = str.lower
    split = str.split
    with open('f.txt') as f:
        d = {}
        lines = f.read().split('\n')
        for A,B in [split(l) for l in lines]:
            try:
                d[lower(A)].add(B)
            except KeyError:
                d[lower(A)] = set(B)

        for a in d:
            print "%s - %s" % (a,",".join(list(d[a])))

if __name__ == "__main__":
    parse_the_file()

The advantage of using a dictionary is that you’ll have a single dictionary key per column A value. The advantage of using a set is that you’ll have a unique set of column B values.

Efficiency notes:

• The use of try-catch is more efficient than using an if\else statement to check for initial cases.
• The evaluation and assignment of the str functions outside of the loop is more efficient than simply using them inside the loop.
• Depending on the proportion of new A values vs. reappearance of A values throughout the file, you may consider using a = lower(A) before the try catch statement
• I used a function, as accessing local variables is more efficient in Python than accessing global variables
• Some of these performance tips are from here

Testing the code above on your input example yields:

xxxd - 4
xxxa - 1,3,2
xxxb - 2
xxxc - 3