Merge remote branch 'joelthelion/master'

Conflicts:
	autojump

autojump

@@ -21,19 +21,16 @@ frequently used places."""
 
 from __future__ import division, print_function
 
-try: # fix to get optimised pickle in python < 3
-    import cPickle as pickle
-except ImportError:
-    import pickle
-
 import getopt
-from sys import argv, stderr, version_info, exit
+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:
@@ -50,6 +47,32 @@ def dicadd(dic, key, increment=1):
     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."""
@@ -57,12 +80,13 @@ def save(path_dict, dic_file):
     # Otherwise, fail quietly
     if (not os.path.exists(dic_file)) or os.name == 'nt' or os.getuid() == os.stat(dic_file)[4]:
         temp = NamedTemporaryFile(dir=CONFIG_DIR, delete=False)
-        pickle.dump(path_dict, temp, 2)
+        for path,weight in sorted(path_dict.items(),key=itemgetter(1),reverse=True):
+            # the db is stored in utf-8
+            temp.write((unico("%s\t%s\n")%(weight,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) 
         import shutil
         shutil.copy(temp.name, dic_file)
         try: #backup file
@@ -75,6 +99,50 @@ def save(path_dict, dic_file):
             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"""
@@ -113,58 +181,30 @@ def match(path, pattern, ignore_case=False, only_end=False):
         eaten_path = path
     return (does_match, eaten_path)
 
-def find_matches(dirs, patterns, result_list, ignore_case, max_matches):
+def find_matches(dirs, patterns, result_list, ignore_case, max_matches, current_dir):
     """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 current_dir == path :
-            if 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:
+        #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 does_match and os.path.exists(path):
             uniqadd(result_list, path)
             if len(result_list) >= max_matches :
                 break
 
-def open_dic(dic_file, error_recovery=False):
+def get_dic_file(filename="autojump.txt"):
-    """Try hard to open the database file, recovering
-       from backup if needed. """
-    try:
-        aj_file = open(dic_file, 'rb')
-        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)
-        aj_file.close()
-        return path_dict
-    except (IOError, EOFError, pickle.UnpicklingError):
-        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: return {} #if everything fails, return an empty file
-
-def get_dic_file():
     if CONFIG_DIR == os.path.expanduser("~"):
-        dic_file = CONFIG_DIR+"/.autojump_py"
+        dic_file = CONFIG_DIR+"/." + filename
     else:
-        dic_file = CONFIG_DIR+"/autojump_py"
+        dic_file = CONFIG_DIR+"/" + filename
     return dic_file
 
 def shell_utility():
@@ -182,13 +222,13 @@ def shell_utility():
         # The home dir can be reached quickly by "cd"
         # and may interfere with other directories
         if(args[-1] != os.path.expanduser("~")): 
-            dicadd(path_dict, args[-1])
+            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:]:
-            print("%.1f:\t%s" % (count, path))
+            output(unico("%.1f:\t%s") % (count, path))
         print("Total key weight: %d. Number of stored paths: %d" %
                 (sum(path_dict.values()), len(paths)))
     else:
@@ -201,8 +241,8 @@ def shell_utility():
             completion = True
         else:
             forget(path_dict, dic_file) #gradually forget about old directories
-        if not args: patterns = [""]
+        if not args: patterns = [unico("")]
-        else: patterns = args
+        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
@@ -211,7 +251,7 @@ def shell_utility():
         if (len(last_pattern_path)>0 and
               last_pattern_path[0] == "/" and
               os.path.exists(last_pattern_path)):
-            if not completion: print(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])
@@ -229,12 +269,21 @@ def shell_utility():
                 max_matches = 9
             else:
                 max_matches = 1
-            find_matches(dirs, patterns, results, False, max_matches)
+
+            # Don't jump to the current directory
+            try:
+                current_dir = decode(os.path.realpath(os.curdir))
+            #Sometimes the current path doesn't exist anymore.
+            #In that case, jump if possible.
+            except OSError:
+                current_dir = None
+            find_matches(dirs, patterns, results, False, max_matches, current_dir)
             # 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) 
+                        ignore_case=True,
+                        max_matches=max_matches, current_dir=current_dir) 
             # Keep the database to a reasonable size
             if not completion and clean_dict(dirs, path_dict):
                 save(path_dict, dic_file)
@@ -244,12 +293,12 @@ def shell_utility():
 
             if userchoice != -1:
                 if len(results) > userchoice-1 : 
-                    print(quotes+results[userchoice-1]+quotes)
+                    output(unico("%s%s%s") % (quotes,results[userchoice-1],quotes))
             elif len(results) > 1 and completion:
-                print("\n".join(("%s%s%d%s%s" % (patterns[-1],
+                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: print(quotes+results[0]+quotes)
+            elif results: output(unico("%s%s%s")%(quotes,results[0],quotes))
             else:
                 return False
             return True