Merge "Add command fuzzy matching"

This commit is contained in:
Jenkins 2015-07-24 19:36:37 +00:00 committed by Gerrit Code Review
commit 5970766546
3 changed files with 157 additions and 5 deletions

View File

@ -9,9 +9,11 @@ import logging
import logging.handlers
import os
import sys
import operator
from .complete import CompleteCommand
from .help import HelpAction, HelpCommand
from .utils import damerau_levenshtein, COST
# Make sure the cliff library has a logging handler
# in case the app developer doesn't set up logging.
@ -299,14 +301,60 @@ class App(object):
self.interpreter.cmdloop()
return 0
def get_fuzzy_matches(self, cmd):
"""return fuzzy matches of unknown command
"""
sep = '_'
if self.command_manager.convert_underscores:
sep = ' '
all_cmds = [k[0] for k in self.command_manager]
dist = []
for candidate in sorted(all_cmds):
prefix = candidate.split(sep)[0]
# Give prefix match a very good score
if candidate.startswith(cmd):
dist.append((candidate, 0))
continue
# Levenshtein distance
dist.append((candidate, damerau_levenshtein(cmd, prefix, COST)+1))
dist = sorted(dist, key=operator.itemgetter(1, 0))
matches = []
i = 0
# Find the best similarity
while (not dist[i][1]):
matches.append(dist[i][0])
i += 1
best_similarity = dist[i][1]
while (dist[i][1] == best_similarity):
matches.append(dist[i][0])
i += 1
return matches
def run_subcommand(self, argv):
try:
subcommand = self.command_manager.find_command(argv)
except ValueError as err:
if self.options.debug:
raise
# If there was no exact match, try to find a fuzzy match
the_cmd = argv[0]
fuzzy_matches = self.get_fuzzy_matches(the_cmd)
if fuzzy_matches:
article = 'a'
if self.NAME[0] in 'aeiou':
article = 'an'
self.stdout.write('%s: \'%s\' is not %s %s command. '
'See \'%s --help\'.\n'
% (self.NAME, the_cmd, article,
self.NAME, self.NAME))
self.stdout.write('Did you mean one of these?\n')
for match in fuzzy_matches:
self.stdout.write(' %s\n' % match)
else:
self.LOG.error(err)
if self.options.debug:
raise
else:
self.LOG.error(err)
return 2
cmd_factory, cmd_name, sub_argv = subcommand
kwargs = {}

View File

@ -3,8 +3,7 @@ from argparse import ArgumentError
try:
from StringIO import StringIO
except ImportError:
# Probably python 3, that test won't be run so ignore the error
pass
from io import StringIO
import sys
import nose
@ -13,6 +12,7 @@ import mock
from cliff.app import App
from cliff.command import Command
from cliff.commandmanager import CommandManager
from cliff.tests import utils
def make_app(**kwargs):
@ -432,3 +432,19 @@ def test_unknown_cmd_debug():
app.run(['--debug', 'hell']) == 2
except ValueError as err:
assert "['hell']" in ('%s' % err)
def test_list_matching_commands():
stdout = StringIO()
app = App('testing', '1',
utils.TestCommandManager(utils.TEST_NAMESPACE),
stdout=stdout)
app.NAME = 'test'
try:
assert app.run(['t']) == 2
except SystemExit:
pass
output = stdout.getvalue()
assert "test: 't' is not a test command. See 'test --help'." in output
assert 'Did you mean one of these?' in output
assert 'three word command\n two words\n' in output

88
cliff/utils.py Normal file
View File

@ -0,0 +1,88 @@
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
# implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# Each edit operation is assigned different cost, such as:
# 'w' means swap operation, the cost is 0;
# 's' means substitution operation, the cost is 2;
# 'a' means insertion operation, the cost is 1;
# 'd' means deletion operation, the cost is 3;
# The smaller cost results in the better similarity.
COST = {'w': 0, 's': 2, 'a': 1, 'd': 3}
def damerau_levenshtein(s1, s2, cost):
"""Calculates the Damerau-Levenshtein distance between two strings.
The Levenshtein distance says the minimum number of single-character edits
(i.e. insertions, deletions, swap or substitution) required to change one
string to the other.
The idea is to reserve a matrix to hold the Levenshtein distances between
all prefixes of the first string and all prefixes of the second, then we
can compute the values in the matrix in a dynamic programming fashion. To
avoid a large space complexity, only the last three rows in the matrix is
needed.(row2 holds the current row, row1 holds the previous row, and row0
the row before that.)
More details:
https://en.wikipedia.org/wiki/Levenshtein_distance
https://github.com/git/git/commit/8af84dadb142f7321ff0ce8690385e99da8ede2f
"""
if s1 == s2:
return 0
len1 = len(s1)
len2 = len(s2)
if len1 == 0:
return len2 * cost['a']
if len2 == 0:
return len1 * cost['d']
row1 = [i * cost['a'] for i in range(len2 + 1)]
row2 = row1[:]
row0 = row1[:]
for i in range(len1):
row2[0] = (i + 1) * cost['d']
for j in range(len2):
# substitution
sub_cost = row1[j] + (s1[i] != s2[j]) * cost['s']
# insertion
ins_cost = row2[j] + cost['a']
# deletion
del_cost = row1[j + 1] + cost['d']
# swap
swp_condition = ((i > 0)
and (j > 0)
and (s1[i - 1] == s2[j])
and (s1[i] == s2[j - 1])
)
# min cost
if swp_condition:
swp_cost = row0[j - 1] + cost['w']
p_cost = min(sub_cost, ins_cost, del_cost, swp_cost)
else:
p_cost = min(sub_cost, ins_cost, del_cost)
row2[j + 1] = p_cost
row0, row1, row2 = row1, row2, row0
return row1[-1]