Parses the entwine output to determine which files weren't completed.

entwineoutput.py

"""Parses the entwine output to determine which files weren't completed at
the time of a crash.

There are two lines from the output that this script is interested in. The
lines with 'Adding' and 'Done'.

Output Example:
Adding 0 - E:/data/Volcano/LAZ/E262N1546_LAZ_PL1.laz
Adding 1 - E:/data/Volcano/LAZ/E262N1547_LAZ_PL1.laz
Adding 2 - E:/data/Volcano/LAZ/E262N1548_LAZ_PL1.laz
Adding 3 - E:/data/Volcano/LAZ/E262N1549_LAZ_PL1.laz
Adding 4 - E:/data/Volcano/LAZ/E262N1550_LAZ_PL1.laz
        Done 1
Adding 5 - E:/data/Volcano/LAZ/E262N1551_LAZ_PL1.laz
        Done 2
Adding 6 - E:/data/Volcano/LAZ/E263N1543_LAZ_PL1.laz
00:10 - 0% - 14,399,664 - 5,183(5,183)M/h - 90W - 0R - 287A
        Done 0
"""


def read(filepath):
    """Read the output from 'entwine build' to determine what files were
    started for processing and which are complete."""
    started = []
    done = []

    started_prefix = 'Adding '

    with open(filepath) as reader:
        for line in reader:
            if line.startswith(started_prefix):
                number, _, path = line[len(started_prefix):].partition(' - ')
                started.append((int(number), path.strip()))
            elif 'Done ' in line:
                _, _, result = line.partition('Done ')
                done.append(int(result.strip()))

    return started, done


def incomplete(started, done):
    """Return items from started that were never done."""
    done = set(done)
    for number, path in started:
        if number not in done:
            yield path


if __name__ == '__main__':
    started, done = read('output.txt')
    remaining = incomplete(started, done)
    print('Items still to do:')
    print('\n'.join(remaining))