parser.py

"""
Parser for amazon appliances data
"""
#TODO: Remove trailing '\n' in entries

import pprint
import sys

FILE_NAME='maestroparseador.txt'

def remove_empty(list_to_process, value=''):
    return [ entry for entry in list_to_process if entry != value ]

def process_line(line):
    task_type = 'TaskType'
    disk_image = 'DISKIMAGE'

    # Process TaskType line
    if line[0] == task_type:
        keys = remove_empty(line)[0::2]
        values = remove_empty(line)[1::2]
        return dict(zip(keys, values))
 
    # Process DISKIMAGE line
    elif line[0] == disk_image:
 
        keys = remove_empty(line)[1::2]
        values = remove_empty(line)[2::2]
        return dict(zip(keys, values))

    elif line == ['\n']:
        return dict()
    else:
        raise ValueError('Line does not begin with {0} or {1}'.format(task_type, disk_image))

def find(app_name, app):
    import_url = 'ImportManifestUrl'
    try:
        url = app[import_url]
        if app_name in url:
            return True
    except KeyError:
        return False

def print_usage():
    usage = """
    parser.py APPLIANCE

        APPLIANCE appliance name to be found. Quote if whitespace is included"""

    print(usage)

if __name__ == '__main__':
    
    with open(FILE_NAME) as f:
        appliances = [process_line(line.split(' ')) for line in f ]
    try:
        app_to_find = sys.argv[1]
    except IndexError:
        print_usage()
        exit(1)

    for app in appliances:
        if find(app_to_find, app):
            pprint.pprint(app)