dirhist

#!/usr/bin/env python2
import sys
import os
import re
from collections import OrderedDict
from os.path import expanduser

# Size of the history
HISTSIZE = 10000

# History file
HISTFILE = ".directory_history"

# Home directory of user
home = expanduser("~")

def get_dir_and_command(line):
    parts = line.split(";")
    return (parts[1], ";".join(parts[2:]))

def remove_duplicates_front(commands):
    """Remove duplicate entries from the front in commands.

    Remove from the front since entries in the front are older
    than the ones at the end of commands.
    """
    # [ls,vim,su,cat,ls] -> [vim,su,cat,ls]
    commands.reverse()
    commands_unique = list(OrderedDict.fromkeys(commands))
    commands_unique.reverse()

    return commands_unique

def read_backwards(f, num_lines, delimiter, blocksize=1024):
    """Read num_lines seperated by delimiter from end of file without
    loading whole file."""
    f.seek(0, 2)
    end = f.tell()
    block_number = -1
    blocks = []

    while num_lines > 0 and end > 0:
        if (end - blocksize > 0):
            f.seek(block_number*blocksize, 2)
            blocks.append(f.read(blocksize))
        else:
            # Read remaining data
            f.seek(0,0)
            blocks.append(f.read(end))

        num_lines -= blocks[-1].count(delimiter)
        end -= blocksize
        block_number -= 1

    return ''.join(reversed(blocks))

def get_commands_in_directory(directory):
    """Returns all commands executed in directory.

    Duplicates will be removed and more important commands will be at
    the end of the returned list.
    The most recent command in directory is considered the most
    important command.
    The oldest command from all other directories is considered the
    least important command.
    """
    commands_dir = []
    commands_not_dir = []
    try:
        with open(home + "/" + HISTFILE, "r") as f:
            # Get commands from this directory
            for line in reversed(read_backwards(f, HISTSIZE, '\0').strip().strip('\0').split('\0')):
                try:
                    directory_in_history, command = get_dir_and_command(line.strip())
                except (ValueError, IndexError):
                    continue

                if directory_in_history == directory:
                    commands_dir.append(command)

            # Get commands not in this directory
            for line in reversed(read_backwards(f, HISTSIZE, '\0').strip().strip('\0').split('\0')):
                try:
                    directory_in_history, command = get_dir_and_command(line.strip())
                except (ValueError, IndexError):
                    continue

                if len(commands_not_dir) + len(commands_dir) >= HISTSIZE:
                    break

                if directory_in_history != directory:
                    commands_not_dir.append(command)
    except IOError:
        open(home + "/" + HISTFILE, 'a').close()

    # Least important command needs to be at the start
    commands = list(reversed(commands_not_dir)) + list(reversed(commands_dir))

    # Remove duplicates from the front
    commands = remove_duplicates_front(commands)

    return commands

def get_indices_by_substring_and_directory(directory, substring):
    """Return indices of substrings from a certain directory."""
    commands = get_commands_in_directory(directory)

    # Remove backslashes in front of ][()|\*?#<>~^
    substring_no_escape = re.sub(r'\\([\][()|\\*?#<>~^])', r'\1', substring)

    indices = []
    for cmd in commands:
        if substring_no_escape.lower() in cmd.lower():
            index = commands.index(cmd)+1
            indices.append(index)

    return indices

# Return complete history for a directory if -a/-all -d DIRECTORY are arguments
if len(sys.argv) == 4 and (sys.argv[1] == "-a" or sys.argv[1] == "--all") and (sys.argv[2] == "-d"):
    directory = sys.argv[3]

    print "\0\n".join(get_commands_in_directory(directory))

# Return list of indizes which match a given substring
# dirhist -s/--substring SUBSTRING -d DIRECTORY
if len(sys.argv) == 5 and (sys.argv[1] == "-s" or sys.argv[1] == "--substring") and sys.argv[3] == "-d":
    directory = sys.argv[4]
    substring = sys.argv[2]

    indices = [str(i) for i in get_indices_by_substring_and_directory(directory, substring)]
    if indices:
        print "\n".join(indices)
    else:
        print "NONE"