wting_autojump/bin/autojump

#!/usr/bin/env python
from __future__ import division, print_function

import argparse
from operator import itemgetter
import os
import re
import shutil
import sys
from tempfile import NamedTemporaryFile

VERSION = 'release-v20'
MAX_KEYWEIGHT = 1000
MAX_STORED_PATHS = 1000
COMPLETION_SEPARATOR = '__'
ARGS = None

if 'AUTOJUMP_DATA_DIR' in os.environ:
    CONFIG_DIR = os.environ.get('AUTOJUMP_DATA_DIR')
else:
    xdg_data_dir = os.environ.get('XDG_DATA_HOME') or os.path.join(os.environ['HOME'], '.local', 'share')
    CONFIG_DIR = os.path.join(xdg_data_dir, 'autojump')

if CONFIG_DIR == os.path.expanduser('~'):
    DB_FILE = CONFIG_DIR + '/.autojump.txt'
else:
    DB_FILE = CONFIG_DIR + '/autojump.txt'

class Database:
    """ Object for interfacing with autojump database. """

    def __init__(self, filename):
        self.filename = filename
        self.data = {}
        self.load()

    # TODO: increase logarithmically as opposed to linear
    def add(self, key, increment = 1):
        """ Increment existing paths or initialize new ones to 0. """
        self.data[key] = self.data.get(key, 0.) + increment

    def decay(self):
        """ Decay database entries. """
        for k in self.data.keys():
            self.data[k] *= 0.9

    def load(self, error_recovery = False):
        """ Try to open the database file, recovering from backup if needed. """
        try:
            with open(self.filename, 'r') as aj_file:
                for line in aj_file.readlines():
                    weight, path = line[:-1].split("\t", 1)
                    path = decode(path, 'utf-8')
                    self.data[path] = float(weight)
        except (IOError, EOFError):
            if not error_recovery and os.path.exists(self.filename + ".bak"):
                print('Problem with autojump database,\
                        trying to recover from backup...', file=sys.stderr)
                shutil.copy(self.filename + '.bak', self.filename)
                return self.load(True)
            else:
                # TODO: migration code, will be removed in v22+
                # autojump_py last used in v17
                self.filename = get_db_file('autojump_py')
                if os.path.exists(self.filename):
                    try: # fix to get optimised pickle in python < 3
                        import cPickle as pickle
                    except ImportError:
                        import pickle
                    try:
                        with open(self.filename, 'rb') as aj_file:
                            # encoding is only specified for Python 2 compatibility
                            if sys.version_info[0] > 2:
                                self.data = pickle.load(aj_file, encoding="utf-8")
                            else:
                                self.data = pickle.load(aj_file)
                        unicode_dict = {}
                        for k, v in self.data.items():
                            print(k)
                            unicode_dict[decode(k, errors="replace")] = v
                        return unicode_dict
                    except (IOError, EOFError, pickle.UnpicklingError):
                        pass
                return {} # if everything fails, return an empty dictionary

    def maintenance(self):
        """ Trims and decays database entries when exceeding settings. """
        if sum(self.data.values()) > MAX_KEYWEIGHT:
            self.decay()
        if len(self.data) > MAX_STORED_PATHS:
            self.trim()
        self.save()

    def save(self):
        """ Save database atomically and preserve backup. """
        # check file existence and permissions
        if ((not os.path.exists(self.filename)) or
                os.name == 'nt' or
                os.getuid() == os.stat(self.filename)[4]):
            temp = NamedTemporaryFile(dir = CONFIG_DIR, delete = False)
            for path, weight in sorted(self.data.items(),
                    key=itemgetter(1),
                    reverse=True):
                temp.write((unico("%s\t%s\n")%(weight, path)).encode("utf-8"))

            # catching disk errors and skipping save when file handle can't be closed.
            try:
                # http://thunk.org/tytso/blog/2009/03/15/dont-fear-the-fsync/
                temp.flush()
                os.fsync(temp)
                temp.close()
            except IOError as ex:
                print("Error saving autojump database (disk full?)" %
                        ex, file=sys.stderr)
                return

            shutil.move(temp.name, self.filename)
            try: # backup file
                import time
                if (not os.path.exists(self.filename+".bak") or
                        time.time()-os.path.getmtime(self.filename+".bak") > 86400):
                    shutil.copy(self.filename, self.filename+".bak")
            except OSError as ex:
                print("Error while creating backup autojump file. (%s)" %
                        ex, file=sys.stderr)

    def trim(self):
        """ If database has exceeded MAX_STORED_PATHS, removes bottom 10%. """
        dirs = list(self.data.items())
        dirs.sort(key=itemgetter(1))
        remove_cnt = .1 * MAX_STORED_PATHS
        for path, _ in dirs[:remove_cnt]:
            del self.data[path]


def get_db_file(filename = "autojump.txt"):
    """ Retrieve full database path. """
    # TODO: Remove when migration code is removed.
    if CONFIG_DIR == os.path.expanduser("~"):
        return CONFIG_DIR + "/." + filename
    else:
        return CONFIG_DIR + "/" + filename

def options():
    """ Parse command line options. """
    global ARGS

    parser = argparse.ArgumentParser(description='Automatically jump to directory passed as an argument.',
            epilog="Please see autojump(1) man pages for full documentation.")
    parser.add_argument('directory', metavar='DIR', nargs='*', default='',
            help='directory to jump to')
    parser.add_argument('-a', '--add', metavar='DIR',
            help='manually add path to database')
    parser.add_argument('-b', '--bash', action="store_true", default=False,
            help='enclose directory quotes to prevent errors')
    parser.add_argument('--complete', action="store_true", default=False,
            help='used for bash tab completion')
    parser.add_argument('--preserve', action="store_true", default=False,
            help='prevent key weight decay over time')
    parser.add_argument('--stat', action="store_true", default=False,
            help='show database entries and their key weights')
    parser.add_argument('--version', action="version", version="%(prog)s " + VERSION,
            help='show version information and exit')

    ARGS = parser.parse_args()

    # The home dir can be reached quickly by "cd" and may interfere with other directories
    if (ARGS.add):
        if(ARGS.add != os.path.expanduser("~")):
            db = Database(DB_FILE)
            db.add(decode(ARGS.add))
            db.save()
        return True

    if (ARGS.stat):
        db = Database(DB_FILE)
        dirs = list(db.data.items())
        dirs.sort(key=itemgetter(1))
        for path, count in dirs[-100:]:
            output(unico("%.1f:\t%s") % (count, path))
        print("Total key weight: %d. Number of stored dirs: %d" %
                (sum(db.data.values()), len(dirs)))
        return True
    return False

def decode(text, encoding=None, errors="strict"):
    """ Decoding step for Python 2 which does not default to unicode. """
    if sys.version_info[0] > 2:
        return text
    else:
        if encoding is None:
            encoding = sys.getfilesystemencoding()
        return text.decode(encoding, errors)

def output(unicode_text, encoding=None):
    """ Wrapper for the print function, using the filesystem encoding by default
    to minimize encoding mismatch problems in directory names. """
    if sys.version_info[0] > 2:
        print(unicode_text)
    else:
        if encoding is None:
            encoding = sys.getfilesystemencoding()
        print(unicode_text.encode(encoding))

def unico(text):
    """ If Python 2, convert to a unicode object. """
    if sys.version_info[0] > 2:
        return text
    else:
        return unicode(text)

def match_last(pattern):
    """ If the last pattern contains a full path, jump there.
    The regexp is because we need to support stuff like
    "j wo jo__3__/home/joel/workspace/joel" for zsh. """
    last_pattern_path = re.sub("(.*)"+COMPLETION_SEPARATOR, "", pattern[-1])
    if (len(last_pattern_path) > 0 and
            last_pattern_path[0] == "/" and
            os.path.exists(last_pattern_path)):
        if not ARGS.preserve:
            output(last_pattern_path)
            return True
    return False

def match(path, pattern, only_end=False, ignore_case=False):
    """ Check whether a path matches a particular pattern, and return
       the remaining part of the string. """
    if only_end:
        match_path = "/".join(path.split('/')[-1-pattern.count('/'):])
    else:
        match_path = path

    if ignore_case:
        find_idx = match_path.lower().find(pattern.lower())
    else:
        find_idx = match_path.find(pattern)

    # truncate path to avoid matching a pattern multiple times
    if find_idx != -1:
        return (True, path)
    else:
        return (False, path[find_idx+len(pattern):])

def find_matches(db, patterns, max_matches=1, ignore_case=False):
    """ Find max_matches paths that match the pattern, and add them to the result_list. """
    try:
        current_dir = decode(os.path.realpath(os.curdir))
    except OSError:
        current_dir = None

    dirs = list(db.data.items())
    dirs.sort(key=itemgetter(1), reverse=True)
    results = []
    for path, _ in dirs:
        # avoid jumping to current directory
        if current_dir == path :
            continue

        found, tmp = True, path
        for n, p in enumerate(patterns):
            # for single/last pattern, only check end of path
            if n == len(patterns)-1:
                found, tmp = match(tmp, p, True, ignore_case)
            else:
                found, tmp = match(tmp, p, False, ignore_case)
            if not found: break

        if found and os.path.exists(path):
            if path not in results:
                results.append(path)
            if len(results) >= max_matches:
                break
    return results

def shell_utility():
    """ Run this when autojump is called as a shell utility. """
    if options(): return True
    db = Database(DB_FILE)

    # if no directories, add empty string
    if (ARGS.directory == ''):
        patterns = [unico('')]
    else:
        patterns = [decode(a) for a in ARGS.directory]

    # check last pattern for full path
    # FIXME: disabled until zsh tab completion is fixed on the shell side
    # if match_last(patterns): return True

    # check for tab completion
    tab_choice = -1
    tab_match = re.search(COMPLETION_SEPARATOR+"([0-9]+)", patterns[-1])
    if tab_match: # user has selected a tab completion entry
        tab_choice = int(tab_match.group(1))
        patterns[-1] = re.sub(COMPLETION_SEPARATOR+"[0-9]+.*", "", patterns[-1])
    else: # user hasn't selected a tab completion, display choices again
        tab_match = re.match("(.*)"+COMPLETION_SEPARATOR, patterns[-1])
        if tab_match:
            patterns[-1] = tab_match.group(1)

    # on tab completion always show all results
    if ARGS.complete or tab_choice != -1:
        max_matches = 9
    else:
        max_matches = 1

    results = find_matches(db, patterns, max_matches, False)
    # if no results, try ignoring case
    if ARGS.complete or not results:
        results = find_matches(db, patterns, max_matches, True)

    quotes = ""
    if ARGS.complete and ARGS.bash: quotes = "'"

    if tab_choice != -1:
        if len(results) > tab_choice-1:
            output(unico("%s%s%s") % (quotes,results[tab_choice-1],quotes))
    elif len(results) > 1 and ARGS.complete:
        output("\n".join(("%s%s%d%s%s" % (patterns[-1],
            COMPLETION_SEPARATOR, n+1, COMPLETION_SEPARATOR, r)
            for n, r in enumerate(results[:8]))))
    elif results:
        output(unico("%s%s%s")%(quotes,results[0],quotes))
    else:
        return False

    if not ARGS.preserve:
        db.maintenance()

    return True

if __name__ == "__main__":
    if not shell_utility(): sys.exit(1)