lingo 1.8.6 → 1.8.7
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/ChangeLog +40 -4
- data/README +22 -51
- data/Rakefile +3 -17
- data/config/lingo.cfg +24 -15
- data/config/lir.cfg +25 -16
- data/dict/de/test_muh.txt +6 -0
- data/dict/en/lingo-dic.txt +2 -3
- data/lang/de.lang +10 -9
- data/lang/en.lang +1 -1
- data/lib/lingo.rb +4 -4
- data/lib/lingo/attendee.rb +27 -7
- data/lib/lingo/attendee/analysis_filter.rb +81 -0
- data/lib/lingo/attendee/debug_filter.rb +42 -0
- data/lib/lingo/attendee/debugger.rb +2 -11
- data/lib/lingo/attendee/decomposer.rb +6 -3
- data/lib/lingo/attendee/formatter.rb +6 -6
- data/lib/lingo/attendee/hal_filter.rb +94 -0
- data/lib/lingo/attendee/lsi_filter.rb +99 -0
- data/lib/lingo/attendee/multi_worder.rb +69 -43
- data/lib/lingo/attendee/sequencer.rb +32 -19
- data/lib/lingo/attendee/synonymer.rb +2 -2
- data/lib/lingo/attendee/text_reader.rb +63 -92
- data/lib/lingo/attendee/text_writer.rb +12 -21
- data/lib/lingo/attendee/tokenizer.rb +32 -21
- data/lib/lingo/attendee/variator.rb +3 -3
- data/lib/lingo/attendee/vector_filter.rb +7 -9
- data/lib/lingo/attendee/word_searcher.rb +3 -3
- data/lib/lingo/buffered_attendee.rb +3 -36
- data/lib/lingo/config.rb +1 -1
- data/lib/lingo/ctl.rb +7 -155
- data/lib/lingo/ctl/analysis.rb +136 -0
- data/lib/lingo/ctl/files.rb +86 -0
- data/lib/lingo/ctl/other.rb +140 -0
- data/lib/lingo/database.rb +64 -60
- data/lib/lingo/database/crypter.rb +7 -5
- data/lib/lingo/error.rb +5 -4
- data/lib/lingo/language.rb +13 -5
- data/lib/lingo/language/grammar.rb +13 -7
- data/lib/lingo/language/token.rb +6 -0
- data/lib/lingo/language/word.rb +23 -36
- data/lib/lingo/language/word_form.rb +5 -1
- data/lib/lingo/srv.rb +2 -2
- data/lib/lingo/text_utils.rb +96 -0
- data/lib/lingo/version.rb +1 -1
- data/lib/lingo/web/views/index.erb +1 -1
- data/test/attendee/ts_decomposer.rb +23 -5
- data/test/attendee/ts_multi_worder.rb +66 -0
- data/test/attendee/ts_sequencer.rb +28 -4
- data/test/attendee/ts_text_reader.rb +20 -0
- data/test/attendee/ts_tokenizer.rb +20 -0
- data/test/attendee/ts_variator.rb +1 -1
- data/test/attendee/ts_word_searcher.rb +39 -3
- data/test/lir3.txt +12 -0
- data/test/ref/artikel.non +1 -12
- data/test/ref/artikel.seq +3 -1
- data/test/ref/artikel.vec +1 -0
- data/test/ref/artikel.vef +35 -34
- data/test/ref/artikel.ven +8 -7
- data/test/ref/artikel.ver +34 -33
- data/test/ref/artikel.vet +2573 -2563
- data/test/ref/lir.non +77 -78
- data/test/ref/lir.seq +9 -7
- data/test/ref/lir.syn +1 -1
- data/test/ref/lir.vec +41 -41
- data/test/ref/lir.vef +210 -210
- data/test/ref/lir.ven +46 -46
- data/test/ref/lir.ver +72 -72
- data/test/ref/lir.vet +329 -329
- data/test/ts_database.rb +166 -62
- data/test/ts_language.rb +23 -23
- metadata +53 -34
- data/lib/lingo/attendee/dehyphenizer.rb +0 -120
- data/lib/lingo/attendee/noneword_filter.rb +0 -115
- data/test/attendee/ts_noneword_filter.rb +0 -15
@@ -0,0 +1,136 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
#--
|
4
|
+
###############################################################################
|
5
|
+
# #
|
6
|
+
# Lingo -- A full-featured automatic indexing system #
|
7
|
+
# #
|
8
|
+
# Copyright (C) 2005-2007 John Vorhauer #
|
9
|
+
# Copyright (C) 2007-2015 John Vorhauer, Jens Wille #
|
10
|
+
# #
|
11
|
+
# Lingo is free software; you can redistribute it and/or modify it under the #
|
12
|
+
# terms of the GNU Affero General Public License as published by the Free #
|
13
|
+
# Software Foundation; either version 3 of the License, or (at your option) #
|
14
|
+
# any later version. #
|
15
|
+
# #
|
16
|
+
# Lingo is distributed in the hope that it will be useful, but WITHOUT ANY #
|
17
|
+
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS #
|
18
|
+
# FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for #
|
19
|
+
# more details. #
|
20
|
+
# #
|
21
|
+
# You should have received a copy of the GNU Affero General Public License #
|
22
|
+
# along with Lingo. If not, see <http://www.gnu.org/licenses/>. #
|
23
|
+
# #
|
24
|
+
###############################################################################
|
25
|
+
#++
|
26
|
+
|
27
|
+
require 'csv'
|
28
|
+
|
29
|
+
class Lingo
|
30
|
+
|
31
|
+
module Ctl
|
32
|
+
|
33
|
+
{ stats: [:s, 'Extract statistics from analysis file(s)', 'path...'],
|
34
|
+
trans: [:t, 'Transpose columns and rows of analysis file(s)', 'path...']
|
35
|
+
}.each { |n, (s, *a)| cmd("analysis#{n}", "a#{s}", *a) }
|
36
|
+
|
37
|
+
private
|
38
|
+
|
39
|
+
def do_analysisstats
|
40
|
+
require 'nuggets/array/histogram'
|
41
|
+
|
42
|
+
paths, write = paths_write
|
43
|
+
stats, patts = Hash.array(1), Hash.array(1)
|
44
|
+
|
45
|
+
csv_foreach(paths) { |path, _, token, word, pattern|
|
46
|
+
token ? stats[:tokens][path] << token : word ? begin
|
47
|
+
stats[:words][path] << word
|
48
|
+
patts[word][path] << pattern if pattern
|
49
|
+
end : nil
|
50
|
+
}
|
51
|
+
|
52
|
+
stats.each { |k, h| write.(k) { |csv|
|
53
|
+
csv << ['file', *c = columns(g = histograms(h))]
|
54
|
+
histograms_to_csv(csv, c, g)
|
55
|
+
h.values.map(&:size)
|
56
|
+
} }
|
57
|
+
|
58
|
+
write.(:patterns) { |csv|
|
59
|
+
csv << ['file', 'word', *c = columns(patts, :values)]
|
60
|
+
patts.sort.each { |k, h| histograms_to_csv(csv, c, histograms(h), k) }
|
61
|
+
c.size - 1
|
62
|
+
}
|
63
|
+
end
|
64
|
+
|
65
|
+
def do_analysistrans
|
66
|
+
paths, write = paths_write
|
67
|
+
|
68
|
+
rows, comm, more, less = Hash.array(1), {}, Hash.array(1), Hash.array(1)
|
69
|
+
|
70
|
+
csv_foreach(paths) { |path, string, token, word, _|
|
71
|
+
rows[token || word][path] << string }
|
72
|
+
|
73
|
+
c = rows.keys.sort.each { |k|
|
74
|
+
a = (h = rows[k]).first.last
|
75
|
+
|
76
|
+
paths.size == 1 ? comm[k] = a : begin
|
77
|
+
comm[k] = a & (o = h.drop(1)).flat_map(&:last)
|
78
|
+
o.each { |path, b| more[path][k] = b - a; less[path][k] = a - b }
|
79
|
+
end
|
80
|
+
}
|
81
|
+
|
82
|
+
rows.clear
|
83
|
+
|
84
|
+
write.(:transpose) { |csv| transpose_csv(csv, c, comm) }
|
85
|
+
|
86
|
+
{ transmore: more, transless: less }.each { |k, v| v.each { |path, h|
|
87
|
+
csv_writer(path, *paths).(k) { |csv| transpose_csv(csv, c, h) } } }
|
88
|
+
end
|
89
|
+
|
90
|
+
def paths_write
|
91
|
+
ARGV.empty? ? missing_arg(:path) : [a = ARGV.each { |x|
|
92
|
+
abort "No such file: #{x}" unless File.exist?(x) }, csv_writer(*a)]
|
93
|
+
end
|
94
|
+
|
95
|
+
def csv_writer(*paths)
|
96
|
+
name = File.join(File.dirname(paths.first), paths.map { |path|
|
97
|
+
File.basename(path.chomp(File.extname(path))) }.uniq.join('-'))
|
98
|
+
|
99
|
+
lambda { |key, &block| overwrite?(file = "#{name}.#{key}.csv") &&
|
100
|
+
puts("#{file}: #{Array(CSV.open(file, 'wb', &block)).join(' / ')}") }
|
101
|
+
end
|
102
|
+
|
103
|
+
def csv_foreach(paths)
|
104
|
+
paths.each { |path| CSV.foreach(path, headers: true) { |row|
|
105
|
+
yield path, *row.values_at(*%w[string token word pattern]) } }
|
106
|
+
end
|
107
|
+
|
108
|
+
def columns(hash, map = :keys)
|
109
|
+
hash.values.map(&map).flatten.uniq.sort
|
110
|
+
end
|
111
|
+
|
112
|
+
def histograms(hash)
|
113
|
+
hash.each_with_object({}) { |(k, v), h|
|
114
|
+
h[k] = v.histogram.tap { |x| x.default = nil } }
|
115
|
+
end
|
116
|
+
|
117
|
+
def histograms_to_csv(csv, columns, histograms, *args)
|
118
|
+
histograms.each { |key, histogram|
|
119
|
+
others = histograms.values_at(*histograms.keys - [key])
|
120
|
+
others = [{}] if others.empty?
|
121
|
+
|
122
|
+
csv << args.dup.unshift(key).concat(columns.map { |header|
|
123
|
+
value = histogram[header] or next
|
124
|
+
value if others.any? { |other| other[header] != value }
|
125
|
+
})
|
126
|
+
}
|
127
|
+
end
|
128
|
+
|
129
|
+
def transpose_csv(csv, columns, rows)
|
130
|
+
csv << columns; values = rows.values_at(*columns); rows.clear
|
131
|
+
values.map(&:size).max.times { |i| csv << values.map { |v| v[i] } }
|
132
|
+
end
|
133
|
+
|
134
|
+
end
|
135
|
+
|
136
|
+
end
|
@@ -0,0 +1,86 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
#--
|
4
|
+
###############################################################################
|
5
|
+
# #
|
6
|
+
# Lingo -- A full-featured automatic indexing system #
|
7
|
+
# #
|
8
|
+
# Copyright (C) 2005-2007 John Vorhauer #
|
9
|
+
# Copyright (C) 2007-2015 John Vorhauer, Jens Wille #
|
10
|
+
# #
|
11
|
+
# Lingo is free software; you can redistribute it and/or modify it under the #
|
12
|
+
# terms of the GNU Affero General Public License as published by the Free #
|
13
|
+
# Software Foundation; either version 3 of the License, or (at your option) #
|
14
|
+
# any later version. #
|
15
|
+
# #
|
16
|
+
# Lingo is distributed in the hope that it will be useful, but WITHOUT ANY #
|
17
|
+
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS #
|
18
|
+
# FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for #
|
19
|
+
# more details. #
|
20
|
+
# #
|
21
|
+
# You should have received a copy of the GNU Affero General Public License #
|
22
|
+
# along with Lingo. If not, see <http://www.gnu.org/licenses/>. #
|
23
|
+
# #
|
24
|
+
###############################################################################
|
25
|
+
#++
|
26
|
+
|
27
|
+
class Lingo
|
28
|
+
|
29
|
+
module Ctl
|
30
|
+
|
31
|
+
{ config: %w[c configuration],
|
32
|
+
lang: %w[l language],
|
33
|
+
dict: %w[d dictionary dictionaries],
|
34
|
+
store: %w[s store],
|
35
|
+
sample: %w[e sample\ text\ file]
|
36
|
+
}.each { |n, (s, q, r)|
|
37
|
+
t = n == :store
|
38
|
+
|
39
|
+
cmd([:list, :l, n], s, "List available #{r || "#{q}s"}", '[name...]') if !t
|
40
|
+
cmd([:find, :f, n], s, "Find #{q} in Lingo search path", 'name')
|
41
|
+
cmd([:copy, :c, n], s, "Copy #{q} to local Lingo directory", 'name') if !t
|
42
|
+
cmd([:clear, :c, n], s, 'Remove store files to force rebuild', 'name') if t
|
43
|
+
}
|
44
|
+
|
45
|
+
private
|
46
|
+
|
47
|
+
def list(what, doit = true)
|
48
|
+
names = Regexp.union(*ARGV.empty? ? '' : ARGV)
|
49
|
+
|
50
|
+
Lingo.list(what, path: path_for_scope).select { |file|
|
51
|
+
File.basename(file) =~ names ? doit ? puts(file) : true : false
|
52
|
+
}
|
53
|
+
end
|
54
|
+
|
55
|
+
def find(what, doit = true, path = path_for_scope)
|
56
|
+
name = ARGV.shift or missing_arg(:name)
|
57
|
+
no_args
|
58
|
+
|
59
|
+
file = Lingo.find(what, name, path: path) { |err| usage(err) }
|
60
|
+
doit ? puts(file) : file
|
61
|
+
end
|
62
|
+
|
63
|
+
def copy(what)
|
64
|
+
usage('Source and target are the same.') if OPTIONS[:scope] == :local
|
65
|
+
|
66
|
+
local_path = path_for_scope(:local)
|
67
|
+
|
68
|
+
source = find(what, false, path_for_scope || Lingo::PATH - local_path)
|
69
|
+
target = File.expand_path(Lingo.basepath(what, source), local_path[0])
|
70
|
+
|
71
|
+
usage('Source and target are the same.') if source == target
|
72
|
+
|
73
|
+
return unless overwrite?(target)
|
74
|
+
|
75
|
+
FileUtils.mkdir_p(File.dirname(target))
|
76
|
+
FileUtils.cp(source, target, verbose: true)
|
77
|
+
end
|
78
|
+
|
79
|
+
def do_clearstore
|
80
|
+
store = Dir["#{find(:store, false)}.*"]
|
81
|
+
FileUtils.rm(store, verbose: true) unless store.empty?
|
82
|
+
end
|
83
|
+
|
84
|
+
end
|
85
|
+
|
86
|
+
end
|
@@ -0,0 +1,140 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
#--
|
4
|
+
###############################################################################
|
5
|
+
# #
|
6
|
+
# Lingo -- A full-featured automatic indexing system #
|
7
|
+
# #
|
8
|
+
# Copyright (C) 2005-2007 John Vorhauer #
|
9
|
+
# Copyright (C) 2007-2015 John Vorhauer, Jens Wille #
|
10
|
+
# #
|
11
|
+
# Lingo is free software; you can redistribute it and/or modify it under the #
|
12
|
+
# terms of the GNU Affero General Public License as published by the Free #
|
13
|
+
# Software Foundation; either version 3 of the License, or (at your option) #
|
14
|
+
# any later version. #
|
15
|
+
# #
|
16
|
+
# Lingo is distributed in the hope that it will be useful, but WITHOUT ANY #
|
17
|
+
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS #
|
18
|
+
# FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for #
|
19
|
+
# more details. #
|
20
|
+
# #
|
21
|
+
# You should have received a copy of the GNU Affero General Public License #
|
22
|
+
# along with Lingo. If not, see <http://www.gnu.org/licenses/>. #
|
23
|
+
# #
|
24
|
+
###############################################################################
|
25
|
+
#++
|
26
|
+
|
27
|
+
require 'zip'
|
28
|
+
Zip.unicode_names = true
|
29
|
+
|
30
|
+
class Lingo
|
31
|
+
|
32
|
+
module Ctl
|
33
|
+
|
34
|
+
{ demo: [:d, 'Initialize demo directory', '[path]', 'current directory'],
|
35
|
+
archive: [:a, 'Create archive of directory', '[path]', 'current directory'],
|
36
|
+
rackup: [:r, 'Print path to rackup file', 'name'],
|
37
|
+
path: [:p, 'Print search path for dictionaries and configurations'],
|
38
|
+
help: [:h, 'Print help for available commands', '[command...]'],
|
39
|
+
version: [:v, 'Print Lingo version number']
|
40
|
+
}.each { |n, (s, *a)| cmd(n.to_s, s.to_s, *a) }
|
41
|
+
|
42
|
+
private
|
43
|
+
|
44
|
+
def do_archive
|
45
|
+
OPTIONS.update(path: ARGV.shift, scope: :local)
|
46
|
+
no_args
|
47
|
+
|
48
|
+
source = File.expand_path(path_for_scope.first)
|
49
|
+
target = "#{source}.zip"
|
50
|
+
|
51
|
+
abort "No such directory: #{source}" unless Dir.exist?(source)
|
52
|
+
|
53
|
+
return unless overwrite?(target, true)
|
54
|
+
|
55
|
+
base, name = File.split(source)
|
56
|
+
|
57
|
+
Dir.chdir(base) {
|
58
|
+
Zip::File.open(target, Zip::File::CREATE) { |zipfile|
|
59
|
+
Dir[File.join(name, '**', '*')].each { |file|
|
60
|
+
zipfile.add(file, file)
|
61
|
+
}
|
62
|
+
}
|
63
|
+
}
|
64
|
+
|
65
|
+
puts "Directory successfully archived at `#{target}'."
|
66
|
+
end
|
67
|
+
|
68
|
+
def do_demo
|
69
|
+
OPTIONS.update(path: ARGV.shift, scope: :system)
|
70
|
+
no_args
|
71
|
+
|
72
|
+
path = path_for_scope(:local).first
|
73
|
+
|
74
|
+
copy_list(:config) { |i| !File.basename(i).start_with?('test') }
|
75
|
+
copy_list(:lang)
|
76
|
+
copy_list(:dict) { |i| File.basename(i).start_with?('user') }
|
77
|
+
copy_list(:sample)
|
78
|
+
|
79
|
+
puts "Demo directory successfully initialized at `#{path}'."
|
80
|
+
end
|
81
|
+
|
82
|
+
def do_rackup(doit = true)
|
83
|
+
name = ARGV.shift or missing_arg(:name)
|
84
|
+
no_args
|
85
|
+
|
86
|
+
require 'lingo/app'
|
87
|
+
|
88
|
+
if file = Lingo::App.rackup(name)
|
89
|
+
doit ? puts(file) : file
|
90
|
+
else
|
91
|
+
usage("Invalid app name `#{name.inspect}'.")
|
92
|
+
end
|
93
|
+
end
|
94
|
+
|
95
|
+
def do_path
|
96
|
+
no_args
|
97
|
+
puts path_for_scope || PATH
|
98
|
+
end
|
99
|
+
|
100
|
+
def do_help(opts = nil)
|
101
|
+
msg = opts ? [opts, 'Commands:'] : []
|
102
|
+
|
103
|
+
args = ARGV unless opts || ARGV.empty?
|
104
|
+
|
105
|
+
aliases = Hash.array
|
106
|
+
ALIASES.each { |k, v| aliases[v] << k }
|
107
|
+
|
108
|
+
COMMANDS.each { |c, (d, *e)|
|
109
|
+
a = aliases[c]
|
110
|
+
next if args && ([c, *a] & args).empty?
|
111
|
+
|
112
|
+
c = "#{c} (#{a.join(', ')})" unless a.empty?
|
113
|
+
|
114
|
+
if opts
|
115
|
+
msg << " %-#{OPTWIDTH}s %s" % [c, d]
|
116
|
+
else
|
117
|
+
msg << "#{c}" << " - #{d}"
|
118
|
+
e.each { |i| msg << " + #{i}" }
|
119
|
+
end
|
120
|
+
}
|
121
|
+
|
122
|
+
msg.empty? ? abort : abort(msg.join("\n"))
|
123
|
+
end
|
124
|
+
|
125
|
+
def do_version(doit = true)
|
126
|
+
no_args
|
127
|
+
|
128
|
+
msg = "Lingo v#{Lingo::VERSION}"
|
129
|
+
doit ? puts(msg) : msg
|
130
|
+
end
|
131
|
+
|
132
|
+
def copy_list(what)
|
133
|
+
files = list(what, false)
|
134
|
+
files.select! { |i| yield i } if block_given?
|
135
|
+
files.each { |file| ARGV.replace([file]); copy(what) }
|
136
|
+
end
|
137
|
+
|
138
|
+
end
|
139
|
+
|
140
|
+
end
|
data/lib/lingo/database.rb
CHANGED
@@ -6,7 +6,7 @@
|
|
6
6
|
# Lingo -- A full-featured automatic indexing system #
|
7
7
|
# #
|
8
8
|
# Copyright (C) 2005-2007 John Vorhauer #
|
9
|
-
# Copyright (C) 2007-
|
9
|
+
# Copyright (C) 2007-2015 John Vorhauer, Jens Wille #
|
10
10
|
# #
|
11
11
|
# Lingo is free software; you can redistribute it and/or modify it under the #
|
12
12
|
# terms of the GNU Affero General Public License as published by the Free #
|
@@ -67,22 +67,33 @@ class Lingo
|
|
67
67
|
def initialize(id, lingo)
|
68
68
|
@id, @lingo, @config, @db = id, lingo, lingo.database_config(id), nil
|
69
69
|
|
70
|
-
@srcfile =
|
71
|
-
|
72
|
-
|
73
|
-
@val = Hash.nest { [] }
|
70
|
+
@val, @crypt, @srcfile = Hash.array, config.key?('crypt'),
|
71
|
+
Lingo.find(:dict, config['name'], relax: true)
|
74
72
|
|
75
73
|
begin
|
76
74
|
@stofile = Lingo.find(:store, @srcfile)
|
77
75
|
FileUtils.mkdir_p(File.dirname(@stofile))
|
78
76
|
rescue SourceFileNotFoundError => err
|
79
77
|
@stofile = skip_ext = err.id
|
80
|
-
|
78
|
+
|
79
|
+
unless err.name
|
80
|
+
if name = BACKEND_BY_EXT[File.extname(@stofile)]
|
81
|
+
backend = get_backend(name, @stofile)
|
82
|
+
else
|
83
|
+
raise BackendNotFoundError.new(@stofile)
|
84
|
+
end
|
85
|
+
end
|
81
86
|
rescue NoWritableStoreError
|
82
87
|
backend = HashStore
|
83
88
|
end
|
84
89
|
|
85
|
-
|
90
|
+
unless backend ||= get_backend(ENV['LINGO_BACKEND'])
|
91
|
+
BACKENDS.find { |name| backend = get_backend(name, nil, true) }
|
92
|
+
end
|
93
|
+
|
94
|
+
extend(@backend = backend || HashStore)
|
95
|
+
|
96
|
+
@stofile << store_ext unless skip_ext || !respond_to?(:store_ext)
|
86
97
|
|
87
98
|
convert unless uptodate?
|
88
99
|
end
|
@@ -131,7 +142,7 @@ class Lingo
|
|
131
142
|
val.uniq!
|
132
143
|
|
133
144
|
val = val.join(FLD_SEP)
|
134
|
-
@
|
145
|
+
@crypt ? _set(*Crypter.encode(key, val)) : _set(key, val)
|
135
146
|
end
|
136
147
|
|
137
148
|
def warn(*msg)
|
@@ -140,26 +151,13 @@ class Lingo
|
|
140
151
|
|
141
152
|
private
|
142
153
|
|
143
|
-
def
|
144
|
-
|
145
|
-
backend = get_backend(mod) and break if mod
|
146
|
-
} unless backend
|
147
|
-
|
148
|
-
extend(@backend = backend || HashStore)
|
149
|
-
|
150
|
-
@stofile << store_ext if !skip_ext && respond_to?(:store_ext)
|
151
|
-
end
|
152
|
-
|
153
|
-
def get_backend(mod)
|
154
|
-
self.class.const_get("#{mod}Store") if Object.const_defined?(mod)
|
155
|
-
rescue TypeError, NameError
|
156
|
-
end
|
157
|
-
|
158
|
-
def backend_from_file(file)
|
159
|
-
ext = File.extname(file)
|
154
|
+
def get_backend(name, file = nil, relax = false)
|
155
|
+
return unless name
|
160
156
|
|
161
|
-
|
162
|
-
|
157
|
+
Object.const_get(name)
|
158
|
+
self.class.const_get("#{name}Store")
|
159
|
+
rescue TypeError, NameError => err
|
160
|
+
raise BackendNotAvailableError.new(name, file, err) unless relax
|
163
161
|
end
|
164
162
|
|
165
163
|
def config_hash
|
@@ -223,9 +221,9 @@ class Lingo
|
|
223
221
|
end
|
224
222
|
|
225
223
|
def _val(key)
|
226
|
-
if val = _get(@
|
224
|
+
if val = _get(@crypt ? Crypter.digest(key) : key)
|
227
225
|
_encode!(val)
|
228
|
-
@
|
226
|
+
@crypt ? Crypter.decode(key, val) : val
|
229
227
|
end
|
230
228
|
end
|
231
229
|
|
@@ -236,32 +234,41 @@ class Lingo
|
|
236
234
|
def convert(verbose = lingo.config.stderr.tty?)
|
237
235
|
src = Source.get(config.fetch('txt-format', 'key_value'), @id, lingo)
|
238
236
|
|
239
|
-
sep, key_map, val_map = prepare_lex
|
237
|
+
sep, hyphenate, key_map, val_map = prepare_lex
|
240
238
|
|
241
239
|
Progress.new(self, src, verbose) { |progress| create {
|
242
240
|
src.each { |key, val|
|
243
241
|
progress << src.pos
|
244
242
|
|
245
|
-
|
246
|
-
key.
|
247
|
-
|
248
|
-
if sep && key.include?(sep)
|
249
|
-
key = key.split(sep).map!(&key_map).join(sep)
|
250
|
-
val = val.map { |v| val_map[v.split(sep)].join(sep) } if val_map
|
243
|
+
set_key(src, key, val, sep) { |keys, cnt|
|
244
|
+
key = keys.map!(&key_map).join(sep)
|
245
|
+
val = val.map { |v| val_map[v.split(sep)].join(sep) } if val_map
|
251
246
|
|
252
|
-
|
253
|
-
|
254
|
-
end
|
255
|
-
end
|
256
|
-
end
|
247
|
+
hyphenate.repeated_permutation(cnt - 1) { |h| set_key(src, keys.
|
248
|
+
zip(h).join, val, sep) unless h.uniq.size == 1 } if hyphenate
|
257
249
|
|
258
|
-
|
250
|
+
[key, val]
|
251
|
+
}
|
259
252
|
}
|
260
253
|
|
261
254
|
uptodate!
|
262
255
|
} }
|
263
256
|
end
|
264
257
|
|
258
|
+
def set_key(src, key, val, sep, len = 3)
|
259
|
+
if key
|
260
|
+
key.chomp!('.')
|
261
|
+
|
262
|
+
if sep && key.include?(sep)
|
263
|
+
keys = key.split(sep); cnt = keys.size
|
264
|
+
key, val = yield keys, cnt if block_given?
|
265
|
+
self[keys[0, len].join(sep)] = ["#{KEY_REF}#{cnt}"] if cnt > len
|
266
|
+
end
|
267
|
+
end
|
268
|
+
|
269
|
+
src.set(self, key, val)
|
270
|
+
end
|
271
|
+
|
265
272
|
def prepare_lex
|
266
273
|
use_lex = config['use-lex'] or return
|
267
274
|
|
@@ -275,6 +282,9 @@ class Lingo
|
|
275
282
|
|
276
283
|
args = nil
|
277
284
|
|
285
|
+
wac = Language::WA_COMPOUND
|
286
|
+
lac = Language::LA_COMPOUND
|
287
|
+
|
278
288
|
if inflect = config['inflect']
|
279
289
|
inflect, wc = inflect == true ? %w[s e] : inflect.split(SEP_RE), 'a'
|
280
290
|
|
@@ -286,19 +296,16 @@ class Lingo
|
|
286
296
|
end
|
287
297
|
end
|
288
298
|
|
289
|
-
[' ', lambda { |form|
|
299
|
+
[sep = ' ', config['hyphenate'] && [sep, '-'], lambda { |form|
|
290
300
|
word = dic.find_word(form)
|
291
301
|
|
292
302
|
if word.unknown?
|
293
|
-
|
303
|
+
comp = gra.find_compound(form)
|
294
304
|
|
295
|
-
|
296
|
-
|
297
|
-
else
|
298
|
-
compo.norm
|
299
|
-
end
|
305
|
+
comp.attr == wac && comp.lex_form(lac) ||
|
306
|
+
(comp.identified? ? comp.lex_form : comp.form)
|
300
307
|
else
|
301
|
-
word.
|
308
|
+
word.identified? ? word.lex_form : word.form
|
302
309
|
end
|
303
310
|
}, inflect && lambda { |forms|
|
304
311
|
inflectables = []
|
@@ -306,24 +313,21 @@ class Lingo
|
|
306
313
|
forms.each { |form|
|
307
314
|
word = dic.find_word(word_form = form[re])
|
308
315
|
|
309
|
-
if word.identified?
|
310
|
-
inflectables << form if form ==
|
316
|
+
if word.identified? && _form = word.lex_form(wc)
|
317
|
+
inflectables << form if form == _form
|
311
318
|
else
|
312
319
|
unless inflectables.empty?
|
313
|
-
|
314
|
-
word = comp.head || comp if comp && !comp.unknown?
|
320
|
+
word = gra.find_compound_head(word_form) || word if word.unknown?
|
315
321
|
|
316
|
-
if word.attr?(*inflect)
|
317
|
-
|
318
|
-
inflectables.each { |lex_form| lex_form << suffix }
|
322
|
+
if word.attr?(*inflect) && suffix =
|
323
|
+
suffixes[word.genders.compact.first]
|
324
|
+
inflectables.each { |lex_form| lex_form << suffix }
|
319
325
|
end
|
320
326
|
end
|
321
327
|
|
322
|
-
break
|
328
|
+
break forms
|
323
329
|
end
|
324
330
|
}
|
325
|
-
|
326
|
-
forms
|
327
331
|
}]
|
328
332
|
end
|
329
333
|
|