wting_autojump/autojump

#!/usr/bin/env python
"""Copyright Joel Schaerer 2008-2010
This file is part of autojump

autojump is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

autojump is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with autojump.  If not, see <http://www.gnu.org/licenses/>.

Autojump is a small tool that maintains a database of your most
used directories, and finds the best match to help you jump to
frequently used places."""

from __future__ import division, print_function

import getopt
from sys import argv, stderr, version_info, exit, getfilesystemencoding
from tempfile import NamedTemporaryFile
from operator import itemgetter
import os

MAX_KEYWEIGHT = 1000
MAX_STORED_PATHS = 600
COMPLETION_SEPARATOR = '__'

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')

def uniqadd(collection, key):
    """Adds a key to a list only if it is not already present"""
    if key not in collection:
        collection.append(key)

def dicadd(dic, key, increment=1):
    """Increment a value in a dic, set it to 0
    if is is not already present"""
    dic[key] = dic.get(key, 0.)+increment

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 version_info[0] > 2:
        print(unicode_text)
    else:
        if encoding is None:
            encoding = getfilesystemencoding()
        print(unicode_text.encode(encoding))

def decode(text,encoding=None,errors="strict"):
    """Decoding step for python2.x which does not default to unicode"""
    if version_info[0] > 2:
        return text
    else:
        if encoding is None:
            encoding = getfilesystemencoding()
        return text.decode(encoding,errors)

def unico(text):
    """if python2, convert to a unicode object"""
    if version_info[0] > 2:
        return text
    else:
        return unicode(text)

def save(path_dict, dic_file):
    """Save the database in an atomic way, and preserve
       a backup file."""
    # If the dic_file exists, check that it belongs to us
    # Otherwise, fail quietly
    if (not os.path.exists(dic_file)) or os.getuid() == os.stat(dic_file)[4]:
        temp = NamedTemporaryFile(dir=CONFIG_DIR, delete=False)
        for path in path_dict:
            # the db is stored in utf-8
            temp.write((unico("%s\t%s\n")%(path_dict[path],path)).encode("utf-8"))
        temp.flush()
        os.fsync(temp)
        temp.close()
        #cf. http://thunk.org/tytso/blog/2009/03/15/dont-fear-the-fsync/
        os.rename(temp.name, dic_file)
        try: #backup file
            import time 
            if (not os.path.exists(dic_file+".bak") or
                    time.time()-os.path.getmtime(dic_file+".bak")>86400):
                import shutil
                shutil.copy(dic_file, dic_file+".bak")
        except OSError as ex:
            print("Error while creating backup autojump file. (%s)" %
                    ex, file=stderr)

def open_dic(dic_file, error_recovery=False):
    """Try hard to open the database file, recovering
       from backup if needed. """
    try:
        path_dict = {}
        with open(dic_file, 'r') as aj_file:
            for l in aj_file.readlines():
                weight,path = l[:-1].split("\t",1)
                # the db is stored in utf-8
                path = decode(path,"utf-8")
                path_dict[path] = float(weight)
            return path_dict
    except (IOError, EOFError):
        if not error_recovery and os.path.exists(dic_file+".bak"):
            print('Problem with autojump database,\
                    trying to recover from backup...', file=stderr)
            import shutil
            shutil.copy(dic_file+".bak", dic_file)
            return open_dic(dic_file, True)
        else:
            # Temporary migration code
            old_dic_file = get_dic_file("autojump_py")
            if os.path.exists(old_dic_file):
                try: # fix to get optimised pickle in python < 3
                    import cPickle as pickle
                except ImportError:
                    import pickle
                try:
                    with open(old_dic_file, 'rb') as aj_file:
                        if version_info[0] > 2:
                            #encoding is only specified for python2.x compatibility
                            path_dict = pickle.load(aj_file, encoding="utf-8")
                        else:
                            path_dict = pickle.load(aj_file)
                    unicode_dict = {} #we now use unicode internally
                    for k,v in path_dict.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 file


def forget(path_dict, dic_file):
    """Gradually forget about directories. Only call
    from the actual jump since it can take time"""
    keyweight = sum(path_dict.values()) 
    if keyweight > MAX_KEYWEIGHT: 
        for k in path_dict.keys():
            path_dict[k] *= 0.9 * MAX_KEYWEIGHT / keyweight
        save(path_dict, dic_file)

def clean_dict(sorted_dirs, path_dict):
    """Limits the sized of the path_dict to MAX_STORED_PATHS. 
    Returns True if keys were deleted"""
    if len(sorted_dirs) > MAX_STORED_PATHS:
        #remove 25 more than needed, to avoid doing it every time
        for path, dummy in sorted_dirs[MAX_STORED_PATHS-25:]:
            del path_dict[path]
        return True
    else: return False

def match(path, pattern, ignore_case=False, only_end=False):
    """Check whether a path matches a particular pattern, and return
       the remaning part of the string"""
    if only_end:
        match_string = "/".join(path.split('/')[-1-pattern.count('/'):])
    else:
        match_string = path
    if ignore_case:
        find_idx = match_string.lower().find(pattern.lower())
    else:
        find_idx = match_string.find(pattern)
    does_match = (find_idx != -1)
    # Eat the path to avoid two patterns matching the same part of the string
    if does_match:
        eaten_path = path[find_idx+len(pattern):]
    else:
        eaten_path = path
    return (does_match, eaten_path)

def find_matches(dirs, patterns, result_list, ignore_case, max_matches):
    """Find max_matches paths that match the pattern, 
    and add them to the result_list"""
    for path, count in dirs:
        # Don't jump to where we alread are
        try:
            if decode(os.path.realpath(os.curdir)) == path :
                continue
        #Sometimes the current path doesn't exist anymore.
        #In that case, jump if possible.
        except OSError:
            pass
        #If a path doesn't exist, don't jump there
        #We still keep it in db in case it's from a removable drive
        if not os.path.exists(path):
            continue
        does_match, eaten_path = True, path
        for n,p in enumerate(patterns):
            #For the last pattern, only match the end of the pattern
            does_match, eaten_path = match(eaten_path, p, ignore_case, only_end=(n == len(patterns)-1))
            if not does_match: break
        if does_match:
            uniqadd(result_list, path)
            if len(result_list) >= max_matches :
                break

def get_dic_file(filename="autojump.txt"):
    if CONFIG_DIR == os.path.expanduser("~"):
        dic_file = CONFIG_DIR+"/." + filename
    else:
        dic_file = CONFIG_DIR+"/" + filename
    return dic_file

def shell_utility():
    """Run this when autojump is called as a shell utility"""
    try:
        optlist, args = getopt.getopt(argv[1:], 'a',
                ['stat', 'import', 'completion', 'bash']) 
    except getopt.GetoptError as ex:
        print("Unknown command line argument: %s" % ex, file=stderr)
        exit(1)

    dic_file = get_dic_file()
    path_dict = open_dic(dic_file)
    if ('-a', '') in optlist:
        # The home dir can be reached quickly by "cd"
        # and may interfere with other directories
        if(args[-1] != os.path.expanduser("~")): 
            dicadd(path_dict, decode(args[-1]))
            save(path_dict, dic_file)
    elif ('--stat', '') in optlist:
        paths = list(path_dict.items())
        paths.sort(key=itemgetter(1))
        for path, count in paths[-100:]:
            output(unico("%.1f:\t%s") % (count, path))
        print("Total key weight: %d. Number of stored paths: %d" %
                (sum(path_dict.values()), len(paths)))
    else:
        import re
        completion = False
        #userchoice is i if the pattern is __pattern__i, otherwise -1
        userchoice = -1 
        results = []
        if ('--completion', '') in optlist:
            completion = True
        else:
            forget(path_dict, dic_file) #gradually forget about old directories
        if not args: patterns = [unico("")]
        else: patterns = [decode(a) for a in args]

        # 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, "", patterns[-1])
        if (len(last_pattern_path)>0 and
              last_pattern_path[0] == "/" and
              os.path.exists(last_pattern_path)):
            if not completion: output(last_pattern_path)
        else:
            #check for ongoing completion, and act accordingly
            endmatch = re.search(COMPLETION_SEPARATOR+"([0-9]+)", patterns[-1])
            if endmatch:  #user has selected a completion
                userchoice = int(endmatch.group(1))
                patterns[-1] = re.sub(COMPLETION_SEPARATOR+"[0-9]+.*",
                        "", patterns[-1])
            else: #user hasn't selected a completion, display the same choices again
                endmatch = re.match("(.*)"+COMPLETION_SEPARATOR, patterns[-1])
                if endmatch: patterns[-1] = endmatch.group(1)

            dirs = list(path_dict.items())
            dirs.sort(key=itemgetter(1), reverse=True)
            if completion or userchoice != -1:
                max_matches = 9
            else:
                max_matches = 1
            find_matches(dirs, patterns, results, False, max_matches)
            # If not found, try ignoring case.
            # On completion always show all results
            if completion or not results: 
                find_matches(dirs, patterns, results,
                        ignore_case=True, max_matches=max_matches) 
            # Keep the database to a reasonable size
            if not completion and clean_dict(dirs, path_dict):
                save(path_dict, dic_file)

            if completion and ('--bash', '') in optlist: quotes = "'"
            else: quotes = ""

            if userchoice != -1:
                if len(results) > userchoice-1 : 
                    output(unico("%s%s%s") % (quotes,results[userchoice-1],quotes))
            elif len(results) > 1 and completion:
                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
            return True


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