create shortened command aliases based on use frequency

cmd_compress.py

import os
import re
import shutil
import sys

cmd_re = re.compile('(\w+)\s+.*')
if 'zsh' in sys.argv[1]:
    cmd_re = re.compile(".*;(\w+)\s+.*")

history= open(sys.argv[1], 'rb').read()
history = history.decode('utf-8', 'ignore')
history=[s.strip() for s in history.split('\n')]

if len(sys.argv) == 3:
    cmd_re = re.compile(sys.argv[2])

cmd_freq = {}
for line in history:
    m = cmd_re.match(line)
    if m:
        cmd = os.path.basename(m.group(1)).lower()
        cmd_path = shutil.which(cmd)
        if cmd_path and os.path.exists(cmd_path):
            cmd_freq[cmd] = cmd_freq.get(cmd, 0) + 1

cmd_tuples = sorted([(c, n) for c, n in cmd_freq.items()], key=lambda t: -t[1])

compressed = {}

for cmd_tuple in cmd_tuples:
    cmd = cmd_tuple[0]

    for i in range(len(cmd)):
        compressed_cmd = cmd[:i+1]
        if not shutil.which(compressed_cmd) and \
            compressed.get(compressed_cmd) is None:
            compressed[compressed_cmd] = cmd
            break

compressed_tuples = sorted([(c1, c2, len(c1)) for c1, c2 in compressed.items()])
compressed_tuples = sorted(compressed_tuples, key=lambda t: t[2])

[print("alias {}={}".format(c1, c2)) for c1, c2, _ in compressed_tuples]