file-digests 0.0.15 → 0.0.20
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/bin/file-digests +0 -3
- data/bin/file-digests-auto +7 -0
- data/bin/file-digests-test +1 -2
- data/lib/file-digests.rb +226 -190
- metadata +3 -1
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 053daa0145db845876fb37b63656102744c524034b87cb91e85f0bce4658aec9
|
4
|
+
data.tar.gz: eac730f10cbe5812078e3461b74b19a740611e9375dd7c34684b2a896e881fc4
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: ce4894b25eb9ecbcbee2ae6c32a9e7bc09fce12286a31896c5d6093ed1ca0f0618e2eaf01facd464c3ebb52f1f9e327198d6a4abffae4f011f9c9d589e6c2d3c
|
7
|
+
data.tar.gz: bdd07cc7c958095b11b57174d8d98cd065d9a83289aeac5ce62dc604c9526f46ab512a7b229e1620c3ba0a49139cae341ac51b11af372e8395eb5a9bfbece1c6
|
data/bin/file-digests
CHANGED
data/bin/file-digests-test
CHANGED
data/lib/file-digests.rb
CHANGED
@@ -1,4 +1,3 @@
|
|
1
|
-
|
2
1
|
require 'date'
|
3
2
|
require 'set'
|
4
3
|
require 'digest'
|
@@ -6,252 +5,289 @@ require 'fileutils'
|
|
6
5
|
require 'pathname'
|
7
6
|
require 'sqlite3'
|
8
7
|
|
9
|
-
|
8
|
+
class FileDigests
|
10
9
|
|
11
10
|
def self.perform_check
|
12
|
-
|
13
|
-
|
11
|
+
options = {
|
12
|
+
auto: (ENV["AUTO"] == "true"),
|
13
|
+
quiet: (ENV["QUIET"] == "true"),
|
14
|
+
test_only: (ENV["TEST_ONLY"] == "true")
|
15
|
+
}
|
16
|
+
file_digests = self.new ARGV[0], ARGV[1], options
|
17
|
+
file_digests.perform_check
|
14
18
|
end
|
15
19
|
|
16
|
-
|
17
|
-
|
18
|
-
|
19
|
-
|
20
|
-
|
21
|
-
execute 'PRAGMA journal_mode = "WAL"'
|
22
|
-
execute 'PRAGMA synchronous = "NORMAL"'
|
23
|
-
execute 'PRAGMA locking_mode = "EXCLUSIVE"'
|
24
|
-
execute 'PRAGMA cache_size = "5000"'
|
25
|
-
|
26
|
-
unless execute("SELECT name FROM sqlite_master WHERE type='table' AND name = 'digests'").length == 1
|
27
|
-
execute 'PRAGMA encoding = "UTF-8"'
|
28
|
-
execute "CREATE TABLE digests (
|
29
|
-
id INTEGER PRIMARY KEY,
|
30
|
-
filename TEXT,
|
31
|
-
mtime TEXT,
|
32
|
-
digest TEXT,
|
33
|
-
digest_check_time TEXT)"
|
34
|
-
execute "CREATE UNIQUE INDEX digests_filename ON digests(filename)"
|
35
|
-
end
|
20
|
+
def self.show_duplicates
|
21
|
+
file_digests = self.new ARGV[0], ARGV[1]
|
22
|
+
file_digests.show_duplicates
|
23
|
+
end
|
36
24
|
|
37
|
-
|
38
|
-
|
25
|
+
def initialize files_path, digest_database_path, options = {}
|
26
|
+
@options = options
|
39
27
|
|
40
|
-
|
41
|
-
|
42
|
-
prepare_method :touch_digest_check_time, "UPDATE digests SET digest_check_time = datetime('now') WHERE id = ?"
|
43
|
-
prepare_method :update_mtime_and_digest, "UPDATE digests SET mtime = ?, digest = ?, digest_check_time = datetime('now') WHERE id = ?"
|
44
|
-
prepare_method :update_mtime, "UPDATE digests SET mtime = ?, digest_check_time = datetime('now') WHERE id = ?"
|
45
|
-
prepare_method :delete_by_filename, "DELETE FROM digests WHERE filename = ?"
|
46
|
-
end
|
28
|
+
@files_path = cleanup_path(files_path || ".")
|
29
|
+
@prefix_to_remove = @files_path.to_s + '/'
|
47
30
|
|
48
|
-
|
49
|
-
result = find_by_filename file_path
|
31
|
+
raise "Files path must be a readable directory" unless (File.directory?(@files_path) && File.readable?(@files_path))
|
50
32
|
|
51
|
-
|
52
|
-
|
33
|
+
@digest_database_path = if digest_database_path
|
34
|
+
cleanup_path(digest_database_path)
|
35
|
+
else
|
36
|
+
@files_path + '.file-digests.sqlite'
|
37
|
+
end
|
53
38
|
|
54
|
-
|
39
|
+
if File.directory?(@digest_database_path)
|
40
|
+
@digest_database_path = @digest_database_path + '.file-digests.sqlite'
|
41
|
+
end
|
55
42
|
|
56
|
-
|
57
|
-
|
58
|
-
# puts "GOOD: #{file_path}" unless QUIET
|
59
|
-
unless TEST_ONLY
|
60
|
-
if found['mtime'] == mtime
|
61
|
-
touch_digest_check_time found['id']
|
62
|
-
else
|
63
|
-
update_mtime mtime, found['id']
|
64
|
-
end
|
65
|
-
end
|
66
|
-
else
|
67
|
-
if found['mtime'] == mtime # Digest is different and mtime is the same
|
68
|
-
counters[:likely_damaged] += 1
|
69
|
-
STDERR.puts "LIKELY DAMAGED: #{file_path}"
|
70
|
-
else
|
71
|
-
counters[:updated] += 1
|
72
|
-
puts "UPDATED: #{file_path}" unless QUIET
|
73
|
-
unless TEST_ONLY
|
74
|
-
update_mtime_and_digest mtime, digest, found['id']
|
75
|
-
end
|
76
|
-
end
|
77
|
-
end
|
78
|
-
else
|
79
|
-
counters[:new] += 1
|
80
|
-
puts "NEW: #{file_path}" unless QUIET
|
81
|
-
unless TEST_ONLY
|
82
|
-
@new_files[file_path] = digest
|
83
|
-
insert file_path, mtime, digest
|
84
|
-
end
|
85
|
-
end
|
43
|
+
if @files_path == @digest_database_path.dirname
|
44
|
+
@skip_file_digests_sqlite = true
|
86
45
|
end
|
87
46
|
|
88
|
-
|
89
|
-
@missing_files.delete_if do |filename, digest|
|
90
|
-
if @new_files.value?(digest)
|
91
|
-
counters[:renamed] += 1
|
92
|
-
unless TEST_ONLY
|
93
|
-
delete_by_filename filename
|
94
|
-
end
|
95
|
-
true
|
96
|
-
end
|
97
|
-
end
|
47
|
+
ensure_dir_exists @digest_database_path.dirname
|
98
48
|
|
99
|
-
|
100
|
-
|
101
|
-
|
102
|
-
puts filename
|
103
|
-
end
|
104
|
-
unless TEST_ONLY
|
105
|
-
puts "Remove missing files from the database (y/n)?"
|
106
|
-
if STDIN.gets.strip.downcase == "y"
|
107
|
-
@db.transaction do
|
108
|
-
@missing_files.each do |filename, digest|
|
109
|
-
delete_by_filename filename
|
110
|
-
end
|
111
|
-
end
|
112
|
-
end
|
113
|
-
end
|
114
|
-
end
|
49
|
+
# Please do not use this flag, support for sha512 is here for backward compatibility, and one day it will be removed.
|
50
|
+
if File.exist?(@digest_database_path.dirname + '.file-digests.sha512')
|
51
|
+
@use_sha512 = true
|
115
52
|
end
|
116
53
|
|
117
|
-
|
54
|
+
initialize_database @digest_database_path
|
55
|
+
end
|
118
56
|
|
119
|
-
|
120
|
-
|
57
|
+
def initialize_database path
|
58
|
+
@db = SQLite3::Database.new path.to_s
|
59
|
+
@db.results_as_hash = true
|
60
|
+
|
61
|
+
execute 'PRAGMA journal_mode = "WAL"'
|
62
|
+
execute 'PRAGMA synchronous = "NORMAL"'
|
63
|
+
execute 'PRAGMA locking_mode = "EXCLUSIVE"'
|
64
|
+
execute 'PRAGMA cache_size = "5000"'
|
65
|
+
|
66
|
+
unless execute("SELECT name FROM sqlite_master WHERE type='table' AND name = 'digests'").length == 1
|
67
|
+
execute 'PRAGMA encoding = "UTF-8"'
|
68
|
+
execute "CREATE TABLE digests (
|
69
|
+
id INTEGER PRIMARY KEY,
|
70
|
+
filename TEXT,
|
71
|
+
mtime TEXT,
|
72
|
+
digest TEXT,
|
73
|
+
digest_check_time TEXT)"
|
74
|
+
execute "CREATE UNIQUE INDEX digests_filename ON digests(filename)"
|
121
75
|
end
|
122
76
|
|
123
|
-
|
124
|
-
|
125
|
-
|
126
|
-
|
127
|
-
|
128
|
-
|
129
|
-
|
77
|
+
prepare_method :insert, "INSERT INTO digests (filename, mtime, digest, digest_check_time) VALUES (?, ?, ?, datetime('now'))"
|
78
|
+
prepare_method :find_by_filename, "SELECT id, mtime, digest FROM digests WHERE filename = ?"
|
79
|
+
prepare_method :touch_digest_check_time, "UPDATE digests SET digest_check_time = datetime('now') WHERE id = ?"
|
80
|
+
prepare_method :update_mtime_and_digest, "UPDATE digests SET mtime = ?, digest = ?, digest_check_time = datetime('now') WHERE id = ?"
|
81
|
+
prepare_method :update_mtime, "UPDATE digests SET mtime = ?, digest_check_time = datetime('now') WHERE id = ?"
|
82
|
+
prepare_method :delete_by_filename, "DELETE FROM digests WHERE filename = ?"
|
83
|
+
prepare_method :query_duplicates, "SELECT digest, filename FROM digests WHERE digest IN (SELECT digest FROM digests GROUP BY digest HAVING count(*) > 1) ORDER BY digest, filename;"
|
130
84
|
end
|
131
85
|
|
132
|
-
|
133
|
-
|
134
|
-
|
135
|
-
|
86
|
+
def perform_check
|
87
|
+
@counters = {good: 0, updated: 0, new: 0, missing: 0, renamed: 0, likely_damaged: 0, exceptions: 0}
|
88
|
+
@missing_files = Hash[@db.prepare("SELECT filename, digest FROM digests").execute!]
|
89
|
+
@new_files = {}
|
136
90
|
|
137
|
-
|
138
|
-
|
139
|
-
|
140
|
-
cleanup_path(digest_database_path)
|
141
|
-
else
|
142
|
-
@files_path + '.file-digests.sqlite'
|
91
|
+
measure_time do
|
92
|
+
walk_files do |filename|
|
93
|
+
process_file filename
|
143
94
|
end
|
95
|
+
end
|
144
96
|
|
145
|
-
|
146
|
-
@digest_database_path = @digest_database_path + '.file-digests.sqlite'
|
147
|
-
end
|
97
|
+
track_renames
|
148
98
|
|
149
|
-
|
150
|
-
|
99
|
+
if any_missing_files?
|
100
|
+
print_missing_files
|
101
|
+
if !@options[:test_only] && (@options[:auto] || confirm("Remove missing files from the database"))
|
102
|
+
remove_missing_files
|
151
103
|
end
|
104
|
+
end
|
152
105
|
|
153
|
-
|
106
|
+
if @counters[:likely_damaged] > 0 || @counters[:exceptions] > 0
|
107
|
+
STDERR.puts "ERRORS WERE OCCURRED"
|
108
|
+
end
|
154
109
|
|
155
|
-
|
156
|
-
|
157
|
-
end
|
110
|
+
puts @counters.inspect
|
111
|
+
end
|
158
112
|
|
159
|
-
|
160
|
-
|
161
|
-
|
113
|
+
def show_duplicates
|
114
|
+
current_digest = nil
|
115
|
+
result = query_duplicates
|
162
116
|
|
163
|
-
|
164
|
-
|
165
|
-
|
166
|
-
|
167
|
-
|
117
|
+
while found = result.next_hash do
|
118
|
+
if current_digest != found['digest']
|
119
|
+
puts "" if current_digest
|
120
|
+
current_digest = found['digest']
|
121
|
+
puts "#{found['digest']}:"
|
168
122
|
end
|
123
|
+
puts " #{found['filename']}"
|
124
|
+
end
|
125
|
+
end
|
169
126
|
|
170
|
-
|
127
|
+
private
|
171
128
|
|
172
|
-
|
173
|
-
|
174
|
-
end
|
129
|
+
def process_file filename
|
130
|
+
return if File.symlink? filename
|
175
131
|
|
176
|
-
|
132
|
+
stat = File.stat filename
|
133
|
+
|
134
|
+
return if stat.blockdev?
|
135
|
+
return if stat.chardev?
|
136
|
+
return if stat.directory?
|
137
|
+
return if stat.pipe?
|
138
|
+
unless stat.readable?
|
139
|
+
raise "File is not readable"
|
140
|
+
end
|
141
|
+
return if stat.socket?
|
142
|
+
|
143
|
+
if @skip_file_digests_sqlite
|
144
|
+
basename = File.basename(filename)
|
145
|
+
return if basename == '.file-digests.sha512'
|
146
|
+
return if basename == '.file-digests.sqlite'
|
147
|
+
return if basename == '.file-digests.sqlite-wal'
|
148
|
+
return if basename == '.file-digests.sqlite-shm'
|
177
149
|
end
|
178
150
|
|
179
|
-
|
151
|
+
insert_or_update(
|
152
|
+
filename.delete_prefix(@prefix_to_remove).encode('utf-8', universal_newline: true).unicode_normalize(:nfkc),
|
153
|
+
stat.mtime.utc.strftime('%Y-%m-%d %H:%M:%S'),
|
154
|
+
get_file_digest(filename)
|
155
|
+
)
|
156
|
+
rescue => exception
|
157
|
+
@counters[:exceptions] += 1
|
158
|
+
STDERR.puts "EXCEPTION: #{filename.encode('utf-8', universal_newline: true)}: #{exception.message}"
|
159
|
+
end
|
180
160
|
|
181
|
-
|
182
|
-
|
161
|
+
def patch_path_string path
|
162
|
+
Gem.win_platform? ? path.gsub(/\\/, '/') : path
|
163
|
+
end
|
183
164
|
|
184
|
-
|
165
|
+
def cleanup_path path
|
166
|
+
Pathname.new(patch_path_string(path)).cleanpath
|
167
|
+
end
|
185
168
|
|
186
|
-
|
187
|
-
|
188
|
-
|
189
|
-
|
190
|
-
unless stat.readable?
|
191
|
-
raise "File is not readable"
|
192
|
-
end
|
193
|
-
return if stat.socket?
|
194
|
-
|
195
|
-
if @skip_file_digests_sqlite
|
196
|
-
basename = File.basename(filename)
|
197
|
-
return if basename == '.file-digests.sha512'
|
198
|
-
return if basename == '.file-digests.sqlite'
|
199
|
-
return if basename == '.file-digests.sqlite-wal'
|
200
|
-
return if basename == '.file-digests.sqlite-shm'
|
169
|
+
def ensure_dir_exists path
|
170
|
+
if File.exist?(path)
|
171
|
+
unless File.directory?(path)
|
172
|
+
raise "#{path} is not a directory"
|
201
173
|
end
|
174
|
+
else
|
175
|
+
FileUtils.mkdir_p path
|
176
|
+
end
|
177
|
+
end
|
202
178
|
|
203
|
-
|
204
|
-
|
205
|
-
|
206
|
-
get_file_digest(filename),
|
207
|
-
@counters
|
208
|
-
)
|
209
|
-
rescue => exception
|
210
|
-
@counters[:exceptions] += 1
|
211
|
-
STDERR.puts "EXCEPTION: #{filename.encode('utf-8', universal_newline: true)}: #{exception.message}"
|
179
|
+
def walk_files
|
180
|
+
Dir.glob(@files_path + '**' + '*', File::FNM_DOTMATCH) do |filename|
|
181
|
+
yield filename
|
212
182
|
end
|
183
|
+
end
|
213
184
|
|
214
|
-
|
215
|
-
|
185
|
+
def get_file_digest filename
|
186
|
+
File.open(filename, 'rb') do |io|
|
187
|
+
digest = (@use_sha512 ? Digest::SHA512 : Digest::SHA256).new
|
188
|
+
buffer = ""
|
189
|
+
while io.read(40960, buffer)
|
190
|
+
digest.update(buffer)
|
191
|
+
end
|
192
|
+
return digest.hexdigest
|
216
193
|
end
|
194
|
+
end
|
217
195
|
|
218
|
-
|
219
|
-
|
196
|
+
def confirm text
|
197
|
+
if STDIN.tty? && STDOUT.tty?
|
198
|
+
puts "#{text} (y/n)?"
|
199
|
+
STDIN.gets.strip.downcase == "y"
|
220
200
|
end
|
201
|
+
end
|
202
|
+
|
203
|
+
def measure_time
|
204
|
+
start = Process.clock_gettime(Process::CLOCK_MONOTONIC)
|
205
|
+
yield
|
206
|
+
elapsed = (Process.clock_gettime(Process::CLOCK_MONOTONIC) - start).to_i
|
207
|
+
puts "Elapsed time: #{elapsed / 3600}h #{(elapsed % 3600) / 60}m #{elapsed % 60}s" unless @options[:quiet]
|
208
|
+
end
|
221
209
|
|
222
|
-
|
223
|
-
|
224
|
-
|
225
|
-
|
210
|
+
def insert_or_update file_path, mtime, digest
|
211
|
+
result = find_by_filename file_path
|
212
|
+
|
213
|
+
if found = result.next_hash
|
214
|
+
raise "Multiple records found" if result.next
|
215
|
+
|
216
|
+
@missing_files.delete(file_path)
|
217
|
+
|
218
|
+
if found['digest'] == digest
|
219
|
+
@counters[:good] += 1
|
220
|
+
# puts "GOOD: #{file_path}" unless @options[:quiet]
|
221
|
+
unless @options[:test_only]
|
222
|
+
if found['mtime'] == mtime
|
223
|
+
touch_digest_check_time found['id']
|
224
|
+
else
|
225
|
+
update_mtime mtime, found['id']
|
226
|
+
end
|
226
227
|
end
|
227
228
|
else
|
228
|
-
|
229
|
+
if found['mtime'] == mtime # Digest is different and mtime is the same
|
230
|
+
@counters[:likely_damaged] += 1
|
231
|
+
STDERR.puts "LIKELY DAMAGED: #{file_path}"
|
232
|
+
else
|
233
|
+
@counters[:updated] += 1
|
234
|
+
puts "UPDATED: #{file_path}" unless @options[:quiet]
|
235
|
+
unless @options[:test_only]
|
236
|
+
update_mtime_and_digest mtime, digest, found['id']
|
237
|
+
end
|
238
|
+
end
|
229
239
|
end
|
230
|
-
|
231
|
-
|
232
|
-
|
233
|
-
|
234
|
-
|
240
|
+
else
|
241
|
+
@counters[:new] += 1
|
242
|
+
puts "NEW: #{file_path}" unless @options[:quiet]
|
243
|
+
unless @options[:test_only]
|
244
|
+
@new_files[file_path] = digest
|
245
|
+
insert file_path, mtime, digest
|
235
246
|
end
|
236
247
|
end
|
248
|
+
end
|
237
249
|
|
238
|
-
|
239
|
-
|
240
|
-
|
241
|
-
|
242
|
-
|
243
|
-
|
250
|
+
def track_renames
|
251
|
+
@missing_files.delete_if do |filename, digest|
|
252
|
+
if @new_files.value?(digest)
|
253
|
+
@counters[:renamed] += 1
|
254
|
+
unless @options[:test_only]
|
255
|
+
delete_by_filename filename
|
244
256
|
end
|
245
|
-
|
257
|
+
true
|
246
258
|
end
|
247
259
|
end
|
260
|
+
@counters[:missing] = @missing_files.length
|
261
|
+
end
|
262
|
+
|
263
|
+
def any_missing_files?
|
264
|
+
@missing_files.length > 0
|
265
|
+
end
|
266
|
+
|
267
|
+
def print_missing_files
|
268
|
+
puts "\nMISSING FILES:"
|
269
|
+
@missing_files.sort.to_h.each do |filename, digest|
|
270
|
+
puts filename
|
271
|
+
end
|
272
|
+
end
|
248
273
|
|
249
|
-
|
250
|
-
|
251
|
-
|
252
|
-
|
253
|
-
|
274
|
+
def remove_missing_files
|
275
|
+
@db.transaction do
|
276
|
+
@missing_files.each do |filename, digest|
|
277
|
+
delete_by_filename filename
|
278
|
+
end
|
254
279
|
end
|
280
|
+
end
|
281
|
+
|
282
|
+
def execute *args, &block
|
283
|
+
@db.execute *args, &block
|
284
|
+
end
|
255
285
|
|
286
|
+
def prepare_method name, query
|
287
|
+
variable = "@#{name}"
|
288
|
+
instance_variable_set(variable, @db.prepare(query))
|
289
|
+
define_singleton_method name do |*args, &block|
|
290
|
+
instance_variable_get(variable).execute(*args, &block)
|
291
|
+
end
|
256
292
|
end
|
257
293
|
end
|
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: file-digests
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.0.
|
4
|
+
version: 0.0.20
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Stanislav Senotrusov
|
@@ -28,11 +28,13 @@ description: Calculate file digests and check for the possible file corruption
|
|
28
28
|
email: stan@senotrusov.com
|
29
29
|
executables:
|
30
30
|
- file-digests
|
31
|
+
- file-digests-auto
|
31
32
|
- file-digests-test
|
32
33
|
extensions: []
|
33
34
|
extra_rdoc_files: []
|
34
35
|
files:
|
35
36
|
- bin/file-digests
|
37
|
+
- bin/file-digests-auto
|
36
38
|
- bin/file-digests-test
|
37
39
|
- lib/file-digests.rb
|
38
40
|
homepage: https://github.com/senotrusov/file-digests
|