cnvrg 1.9.9.9.7
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/bin/cnvrg +9 -0
- data/cnvrg.gemspec +47 -0
- data/lib/cnvrg.rb +7 -0
- data/lib/cnvrg/Images.rb +351 -0
- data/lib/cnvrg/api.rb +247 -0
- data/lib/cnvrg/api_v2.rb +14 -0
- data/lib/cnvrg/auth.rb +79 -0
- data/lib/cnvrg/cli.rb +5715 -0
- data/lib/cnvrg/cli/flow.rb +166 -0
- data/lib/cnvrg/cli/library_cli.rb +33 -0
- data/lib/cnvrg/cli/subcommand.rb +28 -0
- data/lib/cnvrg/cli/task.rb +116 -0
- data/lib/cnvrg/colors.rb +8 -0
- data/lib/cnvrg/connect_job_ssh.rb +31 -0
- data/lib/cnvrg/data.rb +335 -0
- data/lib/cnvrg/datafiles.rb +1325 -0
- data/lib/cnvrg/dataset.rb +892 -0
- data/lib/cnvrg/downloader/client.rb +101 -0
- data/lib/cnvrg/downloader/clients/azure_client.rb +45 -0
- data/lib/cnvrg/downloader/clients/gcp_client.rb +50 -0
- data/lib/cnvrg/downloader/clients/s3_client.rb +78 -0
- data/lib/cnvrg/experiment.rb +209 -0
- data/lib/cnvrg/files.rb +1047 -0
- data/lib/cnvrg/flow.rb +137 -0
- data/lib/cnvrg/helpers.rb +422 -0
- data/lib/cnvrg/helpers/agent.rb +188 -0
- data/lib/cnvrg/helpers/executer.rb +213 -0
- data/lib/cnvrg/hyper.rb +21 -0
- data/lib/cnvrg/image.rb +113 -0
- data/lib/cnvrg/image_cli.rb +25 -0
- data/lib/cnvrg/job_cli.rb +73 -0
- data/lib/cnvrg/job_ssh.rb +48 -0
- data/lib/cnvrg/logger.rb +111 -0
- data/lib/cnvrg/org_helpers.rb +5 -0
- data/lib/cnvrg/project.rb +822 -0
- data/lib/cnvrg/result.rb +29 -0
- data/lib/cnvrg/runner.rb +49 -0
- data/lib/cnvrg/ssh.rb +94 -0
- data/lib/cnvrg/storage.rb +128 -0
- data/lib/cnvrg/task.rb +165 -0
- data/lib/cnvrg/version.rb +3 -0
- metadata +460 -0
@@ -0,0 +1,1325 @@
|
|
1
|
+
require 'mimemagic'
|
2
|
+
require 'aws-sdk-s3'
|
3
|
+
require 'URLcrypt'
|
4
|
+
require 'parallel'
|
5
|
+
require 'fileutils'
|
6
|
+
|
7
|
+
|
8
|
+
module Cnvrg
|
9
|
+
class Datafiles
|
10
|
+
ParallelThreads ||= Cnvrg::Helpers.parallel_threads
|
11
|
+
|
12
|
+
LARGE_FILE=1024*1024*5
|
13
|
+
MULTIPART_SPLIT=10000000
|
14
|
+
|
15
|
+
attr_reader :base_resource
|
16
|
+
|
17
|
+
def initialize(owner, dataset_slug, dataset: nil)
|
18
|
+
@dataset_slug = dataset_slug
|
19
|
+
@owner = owner
|
20
|
+
@dataset = dataset
|
21
|
+
@base_resource = "users/#{owner}/datasets/#{dataset_slug}/"
|
22
|
+
@downloader = @dataset.get_storage_client
|
23
|
+
@token_issue_time = Time.current
|
24
|
+
end
|
25
|
+
|
26
|
+
def refresh_storage_token
|
27
|
+
current_time = Time.current
|
28
|
+
if current_time - @token_issue_time > 1.hours
|
29
|
+
@downloader = @dataset.get_storage_client
|
30
|
+
@token_issue_time = Time.current
|
31
|
+
end
|
32
|
+
end
|
33
|
+
|
34
|
+
def check_file_sha1(filename, org_sha1, tag: 'conflict')
|
35
|
+
file_loc = "#{Dir.pwd}/#{filename}"
|
36
|
+
sha1 = OpenSSL::Digest::SHA1.file(file_loc).hexdigest
|
37
|
+
return 0 if sha1 == org_sha1
|
38
|
+
FileUtils.cp(file_loc, "#{file_loc}.#{tag}")
|
39
|
+
return 1
|
40
|
+
end
|
41
|
+
|
42
|
+
def verify_files_exists(files)
|
43
|
+
paths = []
|
44
|
+
files.each do |file|
|
45
|
+
if File.exists? file
|
46
|
+
if File.directory? file
|
47
|
+
paths += Dir.glob("#{file}/**/*")
|
48
|
+
else
|
49
|
+
paths << file
|
50
|
+
end
|
51
|
+
next
|
52
|
+
end
|
53
|
+
raise SignalException.new(1, "Cant find file #{file}") unless File.exists? "#{Dir.pwd}/#{file}"
|
54
|
+
end
|
55
|
+
paths.map{|p| p.gsub(/^\.\//, '')}
|
56
|
+
end
|
57
|
+
|
58
|
+
def get_files_and_folders(paths)
|
59
|
+
files_and_folders = {}
|
60
|
+
paths.each do |file|
|
61
|
+
if File.exists? file
|
62
|
+
if File.directory? file
|
63
|
+
Dir.glob("#{file}/**/*").select do |f|
|
64
|
+
files_and_folders["#{f}/"] = "folder" if File.directory? f
|
65
|
+
files_and_folders[f] = "file" if File.file? f
|
66
|
+
end
|
67
|
+
files_and_folders["#{file}/"] = "folder"
|
68
|
+
else
|
69
|
+
files_and_folders[file] = "file"
|
70
|
+
end
|
71
|
+
next
|
72
|
+
end
|
73
|
+
raise SignalException.new(1, "Cant find file #{file}") unless File.exists? "#{Dir.pwd}/#{file}"
|
74
|
+
end
|
75
|
+
return files_and_folders
|
76
|
+
end
|
77
|
+
|
78
|
+
def check_files_sha1(files, resolver, tag)
|
79
|
+
conflicts = 0
|
80
|
+
files.each do |file|
|
81
|
+
next if file.ends_with? '/'
|
82
|
+
sha1 = resolver.fetch(file, {}).fetch("sha1", nil)
|
83
|
+
conflicts += self.check_file_sha1(file, sha1, tag: tag)
|
84
|
+
end
|
85
|
+
conflicts
|
86
|
+
end
|
87
|
+
|
88
|
+
def mark_conflicts(results)
|
89
|
+
begin
|
90
|
+
updated = results["updated_on_server"]
|
91
|
+
deleted = results["deleted"]
|
92
|
+
resolver = results['resolved_files']
|
93
|
+
overall = 0
|
94
|
+
overall += self.check_files_sha1(updated, resolver, "conflict")
|
95
|
+
overall += self.check_files_sha1(deleted, resolver, "deleted")
|
96
|
+
overall
|
97
|
+
rescue => e
|
98
|
+
0
|
99
|
+
end
|
100
|
+
end
|
101
|
+
|
102
|
+
def cp_ds(relative: false)
|
103
|
+
prefix = @dataset.get_dataset["bucket_prefix"]
|
104
|
+
batch_size = 10000
|
105
|
+
pbar = ProgressBar.create(:title => "Download Progress",
|
106
|
+
:progress_mark => '=',
|
107
|
+
:format => "%b%i| %c Files downloaded",
|
108
|
+
:starting_at => 0,
|
109
|
+
:total => nil,
|
110
|
+
:autofinish => true)
|
111
|
+
parallel_options = {
|
112
|
+
in_threads: ParallelThreads,
|
113
|
+
in_processes: Cnvrg::CLI::ParallelProcesses,
|
114
|
+
isolation: true,
|
115
|
+
finish: ->(*args) { pbar.progress += 1 }
|
116
|
+
}
|
117
|
+
finished = false
|
118
|
+
while not finished
|
119
|
+
current_batch, marker = @downloader.fetch_files(prefix: prefix, marker: marker, limit: batch_size)
|
120
|
+
if marker.blank?
|
121
|
+
finished = true
|
122
|
+
end
|
123
|
+
Parallel.map(current_batch, parallel_options) do |file|
|
124
|
+
next if file.end_with? "/"
|
125
|
+
cutted_key = relative ? @downloader.cut_prefix(prefix, file) : file
|
126
|
+
dest_path = File.join(@dataset.local_path, cutted_key)
|
127
|
+
@downloader.download(file, dest_path, decrypt: false)
|
128
|
+
file
|
129
|
+
end
|
130
|
+
end
|
131
|
+
end
|
132
|
+
|
133
|
+
# This is for backwards compatibility only and should be removed in future versions:
|
134
|
+
def put_commit(commit_sha1)
|
135
|
+
response = Cnvrg::API.request(
|
136
|
+
"#{@base_resource}/commit/latest",
|
137
|
+
'PUT',
|
138
|
+
{
|
139
|
+
commit_sha1: commit_sha1,
|
140
|
+
ignore: true # tells the new server to ignore this api call since its coming from the new CLI
|
141
|
+
}
|
142
|
+
)
|
143
|
+
if response.present?
|
144
|
+
msg = response['result']
|
145
|
+
else
|
146
|
+
msg = "Can't save changes in the dataset"
|
147
|
+
end
|
148
|
+
|
149
|
+
Cnvrg::Result.new(Cnvrg::CLI.is_response_success(response, false), msg)
|
150
|
+
end
|
151
|
+
|
152
|
+
def create_progressbar(title, total)
|
153
|
+
return ProgressBar.create(
|
154
|
+
:title => title,
|
155
|
+
:progress_mark => '=',
|
156
|
+
:format => "%b>>%i| %p%% %t",
|
157
|
+
:starting_at => 0,
|
158
|
+
:total => total,
|
159
|
+
:autofinish => true
|
160
|
+
)
|
161
|
+
end
|
162
|
+
|
163
|
+
|
164
|
+
def upload_multiple_files(commit_sha1, tree, threads: ParallelThreads, force: false, new_branch: false, prefix: '', partial_commit: nil, total: nil)
|
165
|
+
begin
|
166
|
+
Cnvrg::Logger.log_info("Sending Upload Files request")
|
167
|
+
refresh_storage_token
|
168
|
+
error = nil
|
169
|
+
upload_resp = nil
|
170
|
+
10.times do
|
171
|
+
upload_resp = Cnvrg::API.request(@base_resource + "upload_files", 'POST_JSON', {commit_sha1: commit_sha1, tree: tree, force: force, override: force, is_branch: new_branch, partial_commit: partial_commit})
|
172
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
173
|
+
error = nil
|
174
|
+
break
|
175
|
+
end
|
176
|
+
error = upload_resp
|
177
|
+
Cnvrg::Logger.log_method(bind: binding)
|
178
|
+
Cnvrg::Logger.log_info("Got an error message from server, #{upload_resp.try(:fetch, "message")}, trying again")
|
179
|
+
end
|
180
|
+
raise Exception.new("Can't upload data files: #{error["message"]}") if error.present?
|
181
|
+
|
182
|
+
Cnvrg::Logger.log_info("Uploading files")
|
183
|
+
results = upload_resp['result'].with_indifferent_access
|
184
|
+
|
185
|
+
if results['files'].blank?
|
186
|
+
return 0, []
|
187
|
+
end
|
188
|
+
|
189
|
+
if @temp_upload_progressbar.blank?
|
190
|
+
@temp_upload_progressbar = ProgressBar.create(:title => "Upload Progress",
|
191
|
+
:progress_mark => '=',
|
192
|
+
:format => "%b>>%i| %p%% %t",
|
193
|
+
:starting_at => 0,
|
194
|
+
:total => total,
|
195
|
+
:autofinish => true)
|
196
|
+
end
|
197
|
+
|
198
|
+
files = results['files']
|
199
|
+
|
200
|
+
progress_semaphore = Mutex.new
|
201
|
+
upload_error_files = []
|
202
|
+
@temp_upload_progressbar.progress += tree.keys.length - files.length if @temp_upload_progressbar.present?
|
203
|
+
Parallel.map((files.keys), in_threads: threads) do |k|
|
204
|
+
o = tree[k].merge(files[k])
|
205
|
+
success = upload_single_file(nil, o)
|
206
|
+
if not success
|
207
|
+
upload_error_files << {absolute_path: o[:absolute_path]}
|
208
|
+
files.except!(k)
|
209
|
+
tree.except!(k)
|
210
|
+
Cnvrg::Logger.log_error_message("Error while upload single file #{o["path"]}")
|
211
|
+
end
|
212
|
+
progress_semaphore.synchronize { @temp_upload_progressbar.progress += 1 if @temp_upload_progressbar.present? }
|
213
|
+
end
|
214
|
+
|
215
|
+
blob_ids = files.values.map {|f| f['bv_id']}
|
216
|
+
if blob_ids.present?
|
217
|
+
dirs = tree.keys.select {|k| tree[k].nil?} || []
|
218
|
+
Cnvrg::Logger.info("Sending Upload files save")
|
219
|
+
upload_resp = Cnvrg::API.request(@base_resource + "upload_files_save", "POST", {commit: commit_sha1, blob_ids: blob_ids, dirs: dirs})
|
220
|
+
unless Cnvrg::CLI.is_response_success(upload_resp, false)
|
221
|
+
Cnvrg::Logger.log_method(bind: binding)
|
222
|
+
raise Exception.new("Got an error message from server, #{upload_resp.try(:fetch, "message")}")
|
223
|
+
end
|
224
|
+
end
|
225
|
+
Cnvrg::Logger.log_info("Upload Success")
|
226
|
+
return files.try(:keys).try(:length), upload_error_files
|
227
|
+
rescue => e
|
228
|
+
Cnvrg::Logger.log_method(bind: binding)
|
229
|
+
Cnvrg::Logger.log_error(e)
|
230
|
+
raise e
|
231
|
+
end
|
232
|
+
end
|
233
|
+
|
234
|
+
def delete_multiple_files(commit_sha1, regex_list)
|
235
|
+
begin
|
236
|
+
Cnvrg::Logger.log_info("Sending Delete Files request")
|
237
|
+
resp = Cnvrg::API.request(
|
238
|
+
@base_resource + "delete_files",
|
239
|
+
'POST_JSON',
|
240
|
+
{
|
241
|
+
commit_sha1: commit_sha1,
|
242
|
+
regex_list: regex_list,
|
243
|
+
}
|
244
|
+
)
|
245
|
+
unless Cnvrg::CLI.is_response_success(resp, false)
|
246
|
+
Cnvrg::Logger.log_method(bind: binding)
|
247
|
+
raise Exception.new("Got an error message from server, #{resp.try(:fetch, "message")}")
|
248
|
+
end
|
249
|
+
Cnvrg::Logger.log_info("Delete Files request Successful")
|
250
|
+
return resp["files"], resp["folders"], resp["job_id"]
|
251
|
+
rescue => e
|
252
|
+
Cnvrg::Logger.log_method(bind: binding)
|
253
|
+
Cnvrg::Logger.log_error(e)
|
254
|
+
raise e
|
255
|
+
end
|
256
|
+
end
|
257
|
+
|
258
|
+
def delete_file_chunk(commit_sha1, regex_list, chunk_size, offset)
|
259
|
+
begin
|
260
|
+
resp = Cnvrg::API.request(
|
261
|
+
@base_resource + "delete_files_by_chunk",
|
262
|
+
'POST_JSON',
|
263
|
+
{
|
264
|
+
commit_sha1: commit_sha1,
|
265
|
+
regex_list: regex_list,
|
266
|
+
chunk_size: chunk_size,
|
267
|
+
offset: offset
|
268
|
+
}
|
269
|
+
)
|
270
|
+
unless Cnvrg::CLI.is_response_success(resp, false)
|
271
|
+
Cnvrg::Logger.log_method(bind: binding)
|
272
|
+
raise Exception.new("Got an error message from server, #{resp.try(:fetch, "message")}")
|
273
|
+
end
|
274
|
+
return resp["total_changes"]
|
275
|
+
rescue => e
|
276
|
+
Cnvrg::Logger.log_method(bind: binding)
|
277
|
+
Cnvrg::Logger.log_error(e)
|
278
|
+
raise e
|
279
|
+
end
|
280
|
+
end
|
281
|
+
|
282
|
+
def get_delete_progress(commit_sha1, job_id)
|
283
|
+
begin
|
284
|
+
resp = Cnvrg::API.request(
|
285
|
+
@base_resource + "get_delete_progress",
|
286
|
+
'POST_JSON',
|
287
|
+
{
|
288
|
+
commit_sha1: commit_sha1,
|
289
|
+
job_id: job_id
|
290
|
+
}
|
291
|
+
)
|
292
|
+
unless Cnvrg::CLI.is_response_success(resp, false)
|
293
|
+
Cnvrg::Logger.log_method(bind: binding)
|
294
|
+
raise Exception.new("Got an error message from server, #{resp.try(:fetch, "message")}")
|
295
|
+
end
|
296
|
+
return resp["total_deleted"]
|
297
|
+
rescue => e
|
298
|
+
Cnvrg::Logger.log_method(bind: binding)
|
299
|
+
Cnvrg::Logger.log_error(e)
|
300
|
+
raise e
|
301
|
+
end
|
302
|
+
end
|
303
|
+
|
304
|
+
def request_upload_files(commit_sha1, tree, override, new_branch, partial_commit)
|
305
|
+
retry_count = 0
|
306
|
+
loop do
|
307
|
+
upload_resp = Cnvrg::API.request(@base_resource + "upload_files", 'POST_JSON', {
|
308
|
+
commit_sha1: commit_sha1,
|
309
|
+
tree: tree,
|
310
|
+
override: override,
|
311
|
+
force: override,
|
312
|
+
is_branch: new_branch,
|
313
|
+
partial_commit: partial_commit
|
314
|
+
})
|
315
|
+
if not (Cnvrg::CLI.is_response_success(upload_resp, false))
|
316
|
+
#Cnvrg::Logger.log_method(bind: binding)
|
317
|
+
retry_count += 1
|
318
|
+
|
319
|
+
puts "Failed request upload files: #{Time.current}"
|
320
|
+
puts upload_resp
|
321
|
+
|
322
|
+
if retry_count > 5
|
323
|
+
raise Exception.new("Got an error message from server, #{upload_resp.try(:fetch, "message")}")
|
324
|
+
end
|
325
|
+
sleep 5
|
326
|
+
next
|
327
|
+
end
|
328
|
+
return upload_resp['result'].with_indifferent_access
|
329
|
+
end
|
330
|
+
|
331
|
+
end
|
332
|
+
|
333
|
+
def upload_multiple_files_optimized(files, commit_sha1, threads: 15, chunk_size: 1000, override: false, new_branch: false, prefix: '', partial_commit: nil)
|
334
|
+
cli = CLI.new
|
335
|
+
cli.log_message("Using #{threads} threads with chunk size of #{chunk_size}.", Thor::Shell::Color::GREEN)
|
336
|
+
|
337
|
+
progressbar = create_progressbar("Upload Progress", files.size)
|
338
|
+
cli = CLI.new
|
339
|
+
|
340
|
+
# Vars to handle the parallelism
|
341
|
+
progress_mutex = Mutex.new
|
342
|
+
file_queue = Queue.new
|
343
|
+
progress_queue = Queue.new
|
344
|
+
worker_threads = []
|
345
|
+
|
346
|
+
# Vars to keep track of uploaded files and directories
|
347
|
+
dirs = []
|
348
|
+
uploaded_files = []
|
349
|
+
|
350
|
+
begin
|
351
|
+
# Init working threads that handle the upload of the files:
|
352
|
+
threads.times do |i|
|
353
|
+
worker_threads[i] = Thread.new do
|
354
|
+
# wait for file_queue.close to break the loop
|
355
|
+
while file = file_queue.deq
|
356
|
+
success = upload_single_file(cli, file)
|
357
|
+
file[:success] = success
|
358
|
+
if not success
|
359
|
+
cli.log_message("Error while uploading file: #{file[:absolute_path]}", Thor::Shell::Color::RED)
|
360
|
+
Cnvrg::Logger.log_error_message("Error while upload single file #{file["path"]}")
|
361
|
+
end
|
362
|
+
progress_queue << file
|
363
|
+
end
|
364
|
+
end
|
365
|
+
end
|
366
|
+
|
367
|
+
# init the thread that handles the file upload progress and saving them in the server
|
368
|
+
progress_thread = Thread.new do
|
369
|
+
loop do
|
370
|
+
file = progress_queue.deq(non_block: true) rescue nil # to prevent deadlocks
|
371
|
+
unless file.nil?
|
372
|
+
progress_mutex.synchronize {
|
373
|
+
progressbar.progress += 1
|
374
|
+
uploaded_files.append(file) if file[:success]
|
375
|
+
}
|
376
|
+
|
377
|
+
if uploaded_files.size == chunk_size or progressbar.finished?
|
378
|
+
refresh_storage_token
|
379
|
+
# puts "progress: #{progress_queue.length}"
|
380
|
+
# puts "files: #{file_queue.length}"
|
381
|
+
Cnvrg::Logger.info("Sending Upload files save")
|
382
|
+
blob_ids = uploaded_files.map {|f| f['bv_id']}
|
383
|
+
upload_resp = Cnvrg::API.request(@base_resource + "upload_files_save", "POST", {commit: commit_sha1, blob_ids: blob_ids, dirs: dirs})
|
384
|
+
unless Cnvrg::CLI.is_response_success(upload_resp, false)
|
385
|
+
Cnvrg::Logger.log_method(bind: binding)
|
386
|
+
raise Exception.new("Got an error message from server, #{upload_resp.try(:fetch, "message")}")
|
387
|
+
end
|
388
|
+
# cli.log_message("Saved file chunk to server", Thor::Shell::Color::GREEN)
|
389
|
+
uploaded_files = []
|
390
|
+
dirs = []
|
391
|
+
end
|
392
|
+
else
|
393
|
+
sleep(0.1)
|
394
|
+
end
|
395
|
+
|
396
|
+
if progressbar.finished?
|
397
|
+
# puts "finished"
|
398
|
+
file_queue.close()
|
399
|
+
progress_queue.close()
|
400
|
+
Thread.exit
|
401
|
+
end
|
402
|
+
end
|
403
|
+
end
|
404
|
+
|
405
|
+
file_chunks = files.each_slice(chunk_size).to_a
|
406
|
+
# Fetch the required files from the server:
|
407
|
+
Parallel.map((file_chunks), in_threads: 10) do |files_chunk|
|
408
|
+
|
409
|
+
tree = @dataset.generate_chunked_idx(files_chunk, prefix: prefix, threads: threads)
|
410
|
+
results = request_upload_files(commit_sha1, tree, override, new_branch, partial_commit)
|
411
|
+
|
412
|
+
# puts "Got #{results['files'].size} files to upload from #{files_chunk.size} files"
|
413
|
+
|
414
|
+
if results['files'].blank?
|
415
|
+
progress_mutex.synchronize { progressbar.progress += tree.keys.length }
|
416
|
+
next
|
417
|
+
end
|
418
|
+
|
419
|
+
# Handle directories:
|
420
|
+
new_dirs = tree.keys.select {|k| tree[k].nil?}
|
421
|
+
dirs += new_dirs
|
422
|
+
|
423
|
+
files_to_upload = results['files']
|
424
|
+
progress_mutex.synchronize {
|
425
|
+
progressbar.progress += tree.keys.length - files_to_upload.length
|
426
|
+
}
|
427
|
+
|
428
|
+
files_to_upload.keys.each do |key|
|
429
|
+
while file_queue.size > 5000
|
430
|
+
sleep(0.1)
|
431
|
+
end
|
432
|
+
file_queue.push tree[key].merge(files_to_upload[key])
|
433
|
+
end
|
434
|
+
end
|
435
|
+
|
436
|
+
progress_thread.join()
|
437
|
+
worker_threads.each(&:join)
|
438
|
+
|
439
|
+
rescue => e
|
440
|
+
puts e
|
441
|
+
Cnvrg::Logger.log_method(bind: binding)
|
442
|
+
Cnvrg::Logger.log_error(e)
|
443
|
+
raise e
|
444
|
+
end
|
445
|
+
end
|
446
|
+
|
447
|
+
def upload_single_file(cli, file)
|
448
|
+
success = false
|
449
|
+
begin
|
450
|
+
file = file.as_json
|
451
|
+
Cnvrg::Logger.log_info("Uploading #{file["absolute_path"]}")
|
452
|
+
@downloader.safe_upload(file["path"], file["absolute_path"])
|
453
|
+
success = true
|
454
|
+
Cnvrg::Logger.log_info("#{file["absolute_path"]} uploaded.")
|
455
|
+
rescue => e
|
456
|
+
Cnvrg::Logger.log_error_message("Error while upload single file #{file["path"]}")
|
457
|
+
Cnvrg::Logger.log_error(e)
|
458
|
+
end
|
459
|
+
success
|
460
|
+
end
|
461
|
+
|
462
|
+
def upload_file(absolute_path, relative_path, commit_sha1)
|
463
|
+
file_name = File.basename relative_path
|
464
|
+
file_size = File.size(absolute_path).to_f
|
465
|
+
mime_type = MimeMagic.by_path(absolute_path)
|
466
|
+
content_type = !(mime_type.nil? or mime_type.text?) ? mime_type.type : "text/plain"
|
467
|
+
sha1 = OpenSSL::Digest::SHA1.file(absolute_path).hexdigest
|
468
|
+
if (absolute_path.include? "_tags.yml" or absolute_path.include? "_tags.yaml")
|
469
|
+
is_valid = false
|
470
|
+
begin
|
471
|
+
content = open(absolute_path).read()
|
472
|
+
hash = YAML.load(open(absolute_path).read())
|
473
|
+
# if level 1 keys count is 1
|
474
|
+
if hash.keys.count == 1
|
475
|
+
if hash["tags"].present?
|
476
|
+
is_valid = true
|
477
|
+
elsif hash[hash.keys.first].class != Hash
|
478
|
+
is_valid = true
|
479
|
+
end
|
480
|
+
# if level 1 keys count is greater than 1
|
481
|
+
elsif hash.keys.count > 1
|
482
|
+
if hash["tags"].present? and hash["tags"].class == Hash
|
483
|
+
is_valid = false
|
484
|
+
else
|
485
|
+
is_valid = true
|
486
|
+
end
|
487
|
+
end
|
488
|
+
rescue
|
489
|
+
is_valid = false
|
490
|
+
end
|
491
|
+
|
492
|
+
if is_valid
|
493
|
+
upload_resp = Cnvrg::API.request(@base_resource + "upload_file", 'POST_FILE', {absolute_path: absolute_path, relative_path: relative_path,
|
494
|
+
commit_sha1: commit_sha1, file_name: file_name,
|
495
|
+
file_size: file_size, file_content_type: content_type, sha1: sha1, content: content})
|
496
|
+
else
|
497
|
+
puts("#{absolute_path} is invalid")
|
498
|
+
puts("Please check yaml structure.")
|
499
|
+
end
|
500
|
+
else
|
501
|
+
upload_resp = Cnvrg::API.request(@base_resource + "upload_file", 'POST_FILE', {absolute_path: absolute_path, relative_path: relative_path,
|
502
|
+
commit_sha1: commit_sha1, file_name: file_name,
|
503
|
+
file_size: file_size, file_content_type: content_type, sha1: sha1})
|
504
|
+
puts upload_resp
|
505
|
+
|
506
|
+
end
|
507
|
+
|
508
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
509
|
+
s3_res = upload_large_files_s3(upload_resp, absolute_path)
|
510
|
+
|
511
|
+
return s3_res
|
512
|
+
end
|
513
|
+
end
|
514
|
+
def upload_tar_file(absolute_path, relative_path, commit_sha1)
|
515
|
+
begin
|
516
|
+
file_name = File.basename relative_path
|
517
|
+
file_size = File.size(absolute_path).to_f
|
518
|
+
mime_type = MimeMagic.by_path(absolute_path)
|
519
|
+
content_type = !(mime_type.nil? or mime_type.text?) ? mime_type.type : "text/plain"
|
520
|
+
begin
|
521
|
+
chunked_bytes = [100, (file_size*0.01)].min
|
522
|
+
total_yanked = ""
|
523
|
+
open(absolute_path, "rb") do |f|
|
524
|
+
total_yanked = f.read(chunked_bytes)
|
525
|
+
end
|
526
|
+
if !total_yanked.empty?
|
527
|
+
sha1 = OpenSSL::Digest::SHA1.hexdigest(total_yanked)
|
528
|
+
else
|
529
|
+
sha1 = OpenSSL::Digest::SHA1.file(absolute_path).hexdigest
|
530
|
+
end
|
531
|
+
rescue
|
532
|
+
sha1 = OpenSSL::Digest::SHA1.file(absolute_path).hexdigest
|
533
|
+
end
|
534
|
+
|
535
|
+
upload_resp = Cnvrg::API.request(@base_resource + "upload_tar_file", 'POST_FILE', {absolute_path: absolute_path, relative_path: relative_path,
|
536
|
+
commit_sha1: commit_sha1, file_name: file_name,
|
537
|
+
file_size: file_size, file_content_type: content_type, sha1: sha1,
|
538
|
+
new_version:true})
|
539
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
540
|
+
path = upload_resp["result"]["path"]
|
541
|
+
s3_res = upload_large_files_s3(upload_resp, absolute_path)
|
542
|
+
if s3_res
|
543
|
+
# Cnvrg::API.request(@base_resource + "update_s3", 'POST', {path: path, commit_id: upload_resp["result"]["commit_id"],
|
544
|
+
# blob_id: upload_resp["result"]["id"]})
|
545
|
+
return true
|
546
|
+
end
|
547
|
+
else
|
548
|
+
return false
|
549
|
+
end
|
550
|
+
rescue => e
|
551
|
+
puts e.message
|
552
|
+
return false
|
553
|
+
end
|
554
|
+
end
|
555
|
+
|
556
|
+
|
557
|
+
def upload_log_file(absolute_path, relative_path, log_date)
|
558
|
+
file_name = File.basename relative_path
|
559
|
+
file_size = File.size(absolute_path).to_f
|
560
|
+
content_type = "text/x-log"
|
561
|
+
upload_resp = Cnvrg::API.request("/users/#{@owner}/" + "upload_cli_log", 'POST_FILE', {absolute_path: absolute_path, relative_path: relative_path,
|
562
|
+
file_name: file_name, log_date: log_date,
|
563
|
+
file_size: file_size, file_content_type: content_type})
|
564
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
565
|
+
path = upload_resp["result"]["path"]
|
566
|
+
s3_res = upload_small_files_s3(path, absolute_path, "text/plain")
|
567
|
+
end
|
568
|
+
if s3_res
|
569
|
+
return true
|
570
|
+
end
|
571
|
+
return false
|
572
|
+
|
573
|
+
end
|
574
|
+
def upload_data_log_file(absolute_path, relative_path,data_commit_sha)
|
575
|
+
file_name = File.basename relative_path
|
576
|
+
file_size = File.size(absolute_path).to_f
|
577
|
+
content_type = "text/x-log"
|
578
|
+
upload_resp = Cnvrg::API.request("/users/#{@owner}/" + "upload_data_log", 'POST_FILE', {absolute_path: absolute_path, relative_path: relative_path,
|
579
|
+
file_name: file_name, log_date: Time.now,
|
580
|
+
file_size: file_size, file_content_type: content_type,
|
581
|
+
data_commit_sha1:data_commit_sha})
|
582
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
583
|
+
path = upload_resp["result"]["path"]
|
584
|
+
s3_res = upload_small_files_s3(path, absolute_path, "text/plain")
|
585
|
+
end
|
586
|
+
if s3_res
|
587
|
+
return true
|
588
|
+
end
|
589
|
+
return false
|
590
|
+
|
591
|
+
end
|
592
|
+
|
593
|
+
|
594
|
+
def upload_exec_file(absolute_path, image_name, commit_id)
|
595
|
+
file_name = File.basename absolute_path
|
596
|
+
file_size = File.size(absolute_path).to_f
|
597
|
+
content_type = "application/zip"
|
598
|
+
begin
|
599
|
+
upload_resp = Cnvrg::API.request("users/#{@owner}/images/" + "upload_config", 'POST_FILE', {relative_path: absolute_path,
|
600
|
+
file_name: file_name,
|
601
|
+
image_name: image_name,
|
602
|
+
file_size: file_size,
|
603
|
+
file_content_type: content_type,
|
604
|
+
project_slug: @project_slug,
|
605
|
+
commit_id: commit_id})
|
606
|
+
# puts upload_resp
|
607
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
608
|
+
if upload_resp["result"]["image"] == -1
|
609
|
+
return -1
|
610
|
+
end
|
611
|
+
path = upload_resp["result"]["path"]
|
612
|
+
s3_res = upload_small_files_s3(path, absolute_path, content_type)
|
613
|
+
|
614
|
+
end
|
615
|
+
if s3_res
|
616
|
+
return upload_resp["result"]["id"]
|
617
|
+
end
|
618
|
+
return false
|
619
|
+
rescue SignalException
|
620
|
+
|
621
|
+
say "\nAborting"
|
622
|
+
exit(1)
|
623
|
+
end
|
624
|
+
|
625
|
+
end
|
626
|
+
|
627
|
+
|
628
|
+
def upload_image(absolute_path, image_name, owner, is_public, is_base, dpkg, libraries, bash, message, commit_id)
|
629
|
+
file_name = File.basename absolute_path
|
630
|
+
file_size = File.size(absolute_path).to_f
|
631
|
+
if is_base
|
632
|
+
|
633
|
+
content_type = "application/zip"
|
634
|
+
else
|
635
|
+
content_type = "application/gzip"
|
636
|
+
end
|
637
|
+
begin
|
638
|
+
upload_resp = Cnvrg::API.request("users/#{owner}/images/" + "upload_cnvrg", 'POST_FILE', {relative_path: absolute_path,
|
639
|
+
file_name: file_name,
|
640
|
+
image_name: image_name,
|
641
|
+
file_size: file_size,
|
642
|
+
file_content_type: content_type,
|
643
|
+
is_public: is_public,
|
644
|
+
project_slug: @project_slug,
|
645
|
+
commit_id: commit_id,
|
646
|
+
dpkg: dpkg,
|
647
|
+
py2: libraries,
|
648
|
+
py3: libraries,
|
649
|
+
bash_history: bash,
|
650
|
+
commit_message: message,
|
651
|
+
is_base: is_base})
|
652
|
+
# puts upload_resp
|
653
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
654
|
+
path = upload_resp["result"]["path"]
|
655
|
+
s3_res = upload_small_files_s3(path, absolute_path, content_type)
|
656
|
+
if s3_res
|
657
|
+
commit_resp = Cnvrg::API.request("users/#{owner}/images/#{upload_resp["result"]["id"]}/" + "commit", 'GET')
|
658
|
+
if Cnvrg::CLI.is_response_success(commit_resp, false)
|
659
|
+
return commit_resp["result"]["image"]
|
660
|
+
else
|
661
|
+
return false
|
662
|
+
end
|
663
|
+
|
664
|
+
end
|
665
|
+
end
|
666
|
+
return false
|
667
|
+
rescue => e
|
668
|
+
end
|
669
|
+
|
670
|
+
end
|
671
|
+
|
672
|
+
def upload_cnvrg_image(absolute_path, image_name,secret)
|
673
|
+
file_name = File.basename relative_path
|
674
|
+
file_size = File.size(absolute_path).to_f
|
675
|
+
mime_type = MimeMagic.by_path(absolute_path)
|
676
|
+
content_type = !(mime_type.nil? or mime_type.text?) ? mime_type.type : "text/plain"
|
677
|
+
upload_resp = Cnvrg::API.request("images/#{image_name}/upload_file", 'POST_FILE', {absolute_path: absolute_path, relative_path: absolute_path,
|
678
|
+
file_name: file_name,
|
679
|
+
file_size: file_size, file_content_type: content_type,
|
680
|
+
secret: secret})
|
681
|
+
if Cnvrg::CLI.is_response_success(upload_resp, false)
|
682
|
+
path = upload_resp["result"]["path"]
|
683
|
+
s3_res = upload_large_files_s3(upload_resp, absolute_path)
|
684
|
+
if s3_res
|
685
|
+
Cnvrg::API.request(@base_resource + "update_s3", 'POST', {path: path, commit_id: upload_resp["result"]["commit_id"],
|
686
|
+
blob_id: upload_resp["result"]["id"]})
|
687
|
+
return true
|
688
|
+
end
|
689
|
+
end
|
690
|
+
return false
|
691
|
+
|
692
|
+
end
|
693
|
+
|
694
|
+
def download_image(file_path_to_store, image_slug, owner)
|
695
|
+
|
696
|
+
|
697
|
+
download_resp = Cnvrg::API.request("users/#{owner}/images/#{image_slug}/" + "download", 'GET')
|
698
|
+
path = download_resp["result"]["path"]
|
699
|
+
|
700
|
+
if Cnvrg::CLI.is_response_success(download_resp, false)
|
701
|
+
begin
|
702
|
+
open(file_path_to_store, 'w+') do |file|
|
703
|
+
file << open(path).read
|
704
|
+
end
|
705
|
+
|
706
|
+
return true
|
707
|
+
rescue => e
|
708
|
+
return false
|
709
|
+
end
|
710
|
+
|
711
|
+
return true
|
712
|
+
else
|
713
|
+
return false
|
714
|
+
end
|
715
|
+
|
716
|
+
|
717
|
+
end
|
718
|
+
|
719
|
+
def upload_large_files_s3(upload_resp, file_path)
|
720
|
+
begin
|
721
|
+
sts_path = upload_resp["result"]["path_sts"]
|
722
|
+
|
723
|
+
retries = 0
|
724
|
+
success= false
|
725
|
+
while !success and retries < 20
|
726
|
+
begin
|
727
|
+
if !Helpers.is_verify_ssl
|
728
|
+
body = open(sts_path, {ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE}).read
|
729
|
+
else
|
730
|
+
body = open(sts_path).read
|
731
|
+
end
|
732
|
+
success = true
|
733
|
+
rescue => e
|
734
|
+
retries +=1
|
735
|
+
sleep(5)
|
736
|
+
|
737
|
+
end
|
738
|
+
end
|
739
|
+
if !success
|
740
|
+
return false
|
741
|
+
end
|
742
|
+
split = body.split("\n")
|
743
|
+
key = split[0]
|
744
|
+
iv = split[1]
|
745
|
+
is_s3 = upload_resp["result"]["is_s3"]
|
746
|
+
access = Cnvrg::Helpers.decrypt(key, iv, upload_resp["result"]["sts_a"])
|
747
|
+
|
748
|
+
secret = Cnvrg::Helpers.decrypt(key,iv, upload_resp["result"]["sts_s"])
|
749
|
+
|
750
|
+
session = Cnvrg::Helpers.decrypt(key,iv, upload_resp["result"]["sts_st"])
|
751
|
+
region = Cnvrg::Helpers.decrypt(key,iv, upload_resp["result"]["region"])
|
752
|
+
|
753
|
+
bucket = Cnvrg::Helpers.decrypt(key,iv, upload_resp["result"]["bucket"])
|
754
|
+
server_side_encryption =upload_resp["result"]["server_side_encryption"]
|
755
|
+
use_accelerate_endpoint = false
|
756
|
+
if is_s3 or is_s3.nil?
|
757
|
+
|
758
|
+
use_accelerate_endpoint =true
|
759
|
+
client = Aws::S3::Client.new(
|
760
|
+
:access_key_id =>access,
|
761
|
+
:secret_access_key => secret,
|
762
|
+
:session_token => session,
|
763
|
+
:region => region,
|
764
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
765
|
+
else
|
766
|
+
endpoint = Cnvrg::Helpers.decrypt(key,iv, upload_resp["result"]["endpoint"])
|
767
|
+
use_accelerate_endpoint = false
|
768
|
+
client = Aws::S3::Client.new(
|
769
|
+
:access_key_id =>access,
|
770
|
+
:secret_access_key => secret,
|
771
|
+
:region => region,
|
772
|
+
:endpoint=> endpoint,:force_path_style=> true,:ssl_verify_peer=>false,
|
773
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
774
|
+
end
|
775
|
+
|
776
|
+
|
777
|
+
if !server_side_encryption
|
778
|
+
options = {:use_accelerate_endpoint => use_accelerate_endpoint}
|
779
|
+
else
|
780
|
+
options = {:use_accelerate_endpoint => use_accelerate_endpoint, :server_side_encryption => server_side_encryption}
|
781
|
+
end
|
782
|
+
|
783
|
+
|
784
|
+
|
785
|
+
|
786
|
+
s3 = Aws::S3::Resource.new(client: client)
|
787
|
+
|
788
|
+
resp = s3.bucket(bucket).
|
789
|
+
object(upload_resp["result"]["path"]+"/"+File.basename(file_path)).
|
790
|
+
upload_file(file_path,options)
|
791
|
+
|
792
|
+
|
793
|
+
return resp
|
794
|
+
|
795
|
+
rescue => e
|
796
|
+
puts e.message
|
797
|
+
return false
|
798
|
+
|
799
|
+
end
|
800
|
+
return true
|
801
|
+
|
802
|
+
end
|
803
|
+
|
804
|
+
|
805
|
+
def upload_small_files_s3(url_path, file_path, content_type)
|
806
|
+
url = URI.parse(url_path)
|
807
|
+
file = File.open(file_path, "rb")
|
808
|
+
body = file.read
|
809
|
+
begin
|
810
|
+
Net::HTTP.start(url.host) do |http|
|
811
|
+
http.send_request("PUT", url.request_uri, body, {
|
812
|
+
"content-type" => content_type,
|
813
|
+
})
|
814
|
+
end
|
815
|
+
return true
|
816
|
+
rescue Interrupt
|
817
|
+
return false
|
818
|
+
rescue
|
819
|
+
return false
|
820
|
+
end
|
821
|
+
end
|
822
|
+
|
823
|
+
def upload_url(file_path)
|
824
|
+
response = Cnvrg::API.request(@base_resource + "upload_url", 'POST', {file_s3_path: file_path})
|
825
|
+
if Cnvrg::CLI.is_response_success(response, false)
|
826
|
+
return response
|
827
|
+
else
|
828
|
+
return nil
|
829
|
+
end
|
830
|
+
|
831
|
+
end
|
832
|
+
|
833
|
+
def delete_file(absolute_path, relative_path, commit_sha1)
|
834
|
+
response = Cnvrg::API.request(@base_resource + "delete_file", 'DELETE', {absolute_path: absolute_path, relative_path: relative_path, commit_sha1: commit_sha1})
|
835
|
+
return Cnvrg::CLI.is_response_success(response, false)
|
836
|
+
end
|
837
|
+
|
838
|
+
def delete_dir(absolute_path, relative_path, commit_sha1)
|
839
|
+
response = Cnvrg::API.request(@base_resource + "delete_dir", 'DELETE', {absolute_path: absolute_path, relative_path: relative_path, commit_sha1: commit_sha1})
|
840
|
+
return Cnvrg::CLI.is_response_success(response, false)
|
841
|
+
end
|
842
|
+
|
843
|
+
def create_dir(absolute_path, relative_path, commit_sha1)
|
844
|
+
response = Cnvrg::API.request(@base_resource + "create_dir", 'POST', {absolute_path: absolute_path, relative_path: relative_path, commit_sha1: commit_sha1})
|
845
|
+
return Cnvrg::CLI.is_response_success(response, false)
|
846
|
+
end
|
847
|
+
def download_list_files_in_query(response, dataset_home)
|
848
|
+
sts_path = response["path_sts"]
|
849
|
+
if !Helpers.is_verify_ssl
|
850
|
+
body = open(sts_path, {ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE}).read
|
851
|
+
else
|
852
|
+
body = open(sts_path).read
|
853
|
+
end
|
854
|
+
split = body.split("\n")
|
855
|
+
key = split[0]
|
856
|
+
iv = split[1]
|
857
|
+
|
858
|
+
access = Cnvrg::Helpers.decrypt(key, iv, response["sts_a"])
|
859
|
+
|
860
|
+
secret = Cnvrg::Helpers.decrypt(key,iv, response["sts_s"])
|
861
|
+
|
862
|
+
session = Cnvrg::Helpers.decrypt(key,iv, response["sts_st"])
|
863
|
+
region = Cnvrg::Helpers.decrypt(key,iv, response["region"])
|
864
|
+
|
865
|
+
bucket = Cnvrg::Helpers.decrypt(key,iv, response["bucket"])
|
866
|
+
is_s3 = response["is_s3"]
|
867
|
+
if is_s3 or is_s3.nil?
|
868
|
+
client = Aws::S3::Client.new(
|
869
|
+
:access_key_id =>access,
|
870
|
+
:secret_access_key => secret,
|
871
|
+
:session_token => session,
|
872
|
+
:region => region,
|
873
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
874
|
+
else
|
875
|
+
endpoint = Cnvrg::Helpers.decrypt(key,iv, response["endpoint_url"])
|
876
|
+
client = Aws::S3::Client.new(
|
877
|
+
:access_key_id =>access,
|
878
|
+
:secret_access_key => secret,
|
879
|
+
:region => region,
|
880
|
+
:endpoint=> endpoint,:force_path_style=> true,:ssl_verify_peer=>false,
|
881
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
882
|
+
end
|
883
|
+
list_files = response["files"]
|
884
|
+
parallel_options = {
|
885
|
+
:progress => {
|
886
|
+
:title => "Download Progress",
|
887
|
+
:progress_mark => '=',
|
888
|
+
:format => "%b>>%i| %p%% %t",
|
889
|
+
:starting_at => 0,
|
890
|
+
:total => list_files.size,
|
891
|
+
:autofinish => true
|
892
|
+
},
|
893
|
+
in_threads: ParallelThreads
|
894
|
+
}
|
895
|
+
download_count = 0
|
896
|
+
Parallel.map((list_files), parallel_options) do |f|
|
897
|
+
file_key = Cnvrg::Helpers.decrypt(key,iv, f["path"])
|
898
|
+
begin
|
899
|
+
begin
|
900
|
+
dir = File.dirname f["fullpath"]
|
901
|
+
FileUtils.mkdir_p(dataset_home+"/"+ dir) unless File.exist? (dataset_home+"/"+ dir)
|
902
|
+
end
|
903
|
+
|
904
|
+
File.open(dataset_home+"/"+f["fullpath"], 'w+') do |file|
|
905
|
+
resp = client.get_object({bucket:bucket,
|
906
|
+
key:file_key}, target: file)
|
907
|
+
end
|
908
|
+
download_count += 1
|
909
|
+
rescue
|
910
|
+
end
|
911
|
+
|
912
|
+
end
|
913
|
+
if download_count == list_files.size
|
914
|
+
return true
|
915
|
+
else
|
916
|
+
return false
|
917
|
+
end
|
918
|
+
|
919
|
+
|
920
|
+
|
921
|
+
|
922
|
+
end
|
923
|
+
|
924
|
+
|
925
|
+
def download_file_s3(absolute_path, relative_path, project_home, conflict=false, commit_sha1=nil, as_link=false)
|
926
|
+
begin
|
927
|
+
res = Cnvrg::API.request(@base_resource + "download_file", 'POST', {absolute_path: absolute_path, relative_path: relative_path, commit_sha1: commit_sha1 ,new_version:true, as_link:as_link})
|
928
|
+
Cnvrg::CLI.is_response_success(res, false)
|
929
|
+
if res["result"]
|
930
|
+
file_url = res["result"]["file_url"]
|
931
|
+
|
932
|
+
if as_link
|
933
|
+
return res["result"]
|
934
|
+
end
|
935
|
+
# begin
|
936
|
+
# if !Helpers.is_verify_ssl
|
937
|
+
# tempfile = Down.download(file_url,open_timeout: 60,ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE )
|
938
|
+
#
|
939
|
+
# else
|
940
|
+
# tempfile = Down.download(file_url,open_timeout: 60)
|
941
|
+
#
|
942
|
+
# end
|
943
|
+
#
|
944
|
+
# FileUtils.move(tempfile.path, project_home+"/"+ absolute_path)
|
945
|
+
# return true
|
946
|
+
# rescue
|
947
|
+
#
|
948
|
+
# end
|
949
|
+
download_resp = res
|
950
|
+
filename = download_resp["result"]["filename"]
|
951
|
+
|
952
|
+
absolute_path += ".conflict" if conflict
|
953
|
+
sts_path = download_resp["result"]["path_sts"]
|
954
|
+
retries = 0
|
955
|
+
success= false
|
956
|
+
while !success and retries < 20
|
957
|
+
begin
|
958
|
+
if !Helpers.is_verify_ssl
|
959
|
+
body = open(sts_path, {ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE}).read
|
960
|
+
else
|
961
|
+
body = open(sts_path).read
|
962
|
+
end
|
963
|
+
success = true
|
964
|
+
rescue => e
|
965
|
+
retries +=1
|
966
|
+
sleep(5)
|
967
|
+
|
968
|
+
end
|
969
|
+
end
|
970
|
+
if !success
|
971
|
+
return false
|
972
|
+
end
|
973
|
+
split = body.split("\n")
|
974
|
+
key = split[0]
|
975
|
+
iv = split[1]
|
976
|
+
|
977
|
+
access = Cnvrg::Helpers.decrypt(key, iv, download_resp["result"]["sts_a"])
|
978
|
+
|
979
|
+
secret = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["sts_s"])
|
980
|
+
|
981
|
+
session = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["sts_st"])
|
982
|
+
region = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["region"])
|
983
|
+
|
984
|
+
bucket = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["bucket"])
|
985
|
+
file_key = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["key"])
|
986
|
+
|
987
|
+
is_s3 = download_resp["result"]["is_s3"]
|
988
|
+
if is_s3 or is_s3.nil?
|
989
|
+
client = Aws::S3::Client.new(
|
990
|
+
:access_key_id =>access,
|
991
|
+
:secret_access_key => secret,
|
992
|
+
:session_token => session,
|
993
|
+
:region => region,
|
994
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
995
|
+
else
|
996
|
+
endpoint = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["endpoint_url"])
|
997
|
+
client = Aws::S3::Client.new(
|
998
|
+
:access_key_id =>access,
|
999
|
+
:secret_access_key => secret,
|
1000
|
+
:region => region,
|
1001
|
+
:endpoint=> endpoint,:force_path_style=> true,:ssl_verify_peer=>false,
|
1002
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
1003
|
+
end
|
1004
|
+
|
1005
|
+
File.open(project_home+"/"+absolute_path, 'w+') do |file|
|
1006
|
+
resp = client.get_object({bucket:bucket,
|
1007
|
+
key:file_key}, target: file)
|
1008
|
+
end
|
1009
|
+
return true
|
1010
|
+
end
|
1011
|
+
|
1012
|
+
rescue => e
|
1013
|
+
return false
|
1014
|
+
|
1015
|
+
end
|
1016
|
+
end
|
1017
|
+
|
1018
|
+
def download_data_file(commit_sha1, dataset_home)
|
1019
|
+
begin
|
1020
|
+
res = Cnvrg::API.request(@base_resource + "download_data_file", 'POST', {commit_sha1: commit_sha1,new_version:true})
|
1021
|
+
Cnvrg::CLI.is_response_success(res, false)
|
1022
|
+
if res["result"]
|
1023
|
+
download_resp = res
|
1024
|
+
filename = download_resp["result"]["filename"]
|
1025
|
+
|
1026
|
+
sts_path = download_resp["result"]["path_sts"]
|
1027
|
+
if !Helpers.is_verify_ssl
|
1028
|
+
body = open(sts_path, {ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE}).read
|
1029
|
+
else
|
1030
|
+
body = open(sts_path).read
|
1031
|
+
end
|
1032
|
+
split = body.split("\n")
|
1033
|
+
key = split[0]
|
1034
|
+
iv = split[1]
|
1035
|
+
|
1036
|
+
|
1037
|
+
access = Cnvrg::Helpers.decrypt(key, iv, download_resp["result"]["sts_a"])
|
1038
|
+
|
1039
|
+
secret = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["sts_s"])
|
1040
|
+
|
1041
|
+
session = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["sts_st"])
|
1042
|
+
region = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["region"])
|
1043
|
+
|
1044
|
+
bucket = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["bucket"])
|
1045
|
+
file_key = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["key"])
|
1046
|
+
|
1047
|
+
is_s3 = download_resp["result"]["is_s3"]
|
1048
|
+
if is_s3 or is_s3.nil?
|
1049
|
+
client = Aws::S3::Client.new(
|
1050
|
+
:access_key_id =>access,
|
1051
|
+
:secret_access_key => secret,
|
1052
|
+
:session_token => session,
|
1053
|
+
:region => region,
|
1054
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
1055
|
+
else
|
1056
|
+
endpoint = Cnvrg::Helpers.decrypt(key,iv, download_resp["result"]["endpoint_url"])
|
1057
|
+
client = Aws::S3::Client.new(
|
1058
|
+
:access_key_id =>access,
|
1059
|
+
:secret_access_key => secret,
|
1060
|
+
:region => region,
|
1061
|
+
:endpoint=> endpoint,:force_path_style=> true,:ssl_verify_peer=>false,
|
1062
|
+
:http_open_timeout => 60, :retry_limit => 20)
|
1063
|
+
end
|
1064
|
+
|
1065
|
+
|
1066
|
+
File.open(dataset_home+"/"+filename, 'w+') do |file|
|
1067
|
+
resp = client.get_object({bucket: bucket,
|
1068
|
+
key: file_key }, target: file)
|
1069
|
+
end
|
1070
|
+
return filename
|
1071
|
+
end
|
1072
|
+
|
1073
|
+
rescue => e
|
1074
|
+
return false
|
1075
|
+
|
1076
|
+
end
|
1077
|
+
end
|
1078
|
+
|
1079
|
+
def download_file(absolute_path, relative_path, project_home, conflict=false)
|
1080
|
+
res = Cnvrg::API.request(@base_resource + "download_file", 'POST', {absolute_path: absolute_path, relative_path: relative_path})
|
1081
|
+
Cnvrg::CLI.is_response_success(res, false)
|
1082
|
+
if res["result"]
|
1083
|
+
res = res["result"]
|
1084
|
+
return false if res["link"].empty? or res["filename"].empty?
|
1085
|
+
filename = res["filename"]
|
1086
|
+
file_location = absolute_path.gsub(/#{filename}\/?$/, "")
|
1087
|
+
|
1088
|
+
FileUtils.mkdir_p project_home + "/" + file_location
|
1089
|
+
filename += ".conflict" if conflict
|
1090
|
+
|
1091
|
+
File.open("#{project_home}/#{file_location}/#{filename}", "w+") do |file|
|
1092
|
+
file.write open(res["link"]).read
|
1093
|
+
end
|
1094
|
+
else
|
1095
|
+
return false
|
1096
|
+
end
|
1097
|
+
return true
|
1098
|
+
end
|
1099
|
+
|
1100
|
+
def delete_commit_files_local(deleted)
|
1101
|
+
begin
|
1102
|
+
FileUtils.rm_rf(deleted) unless (deleted.nil? or deleted.empty?)
|
1103
|
+
return Cnvrg::Result.new(true, '')
|
1104
|
+
rescue => e
|
1105
|
+
return Cnvrg::Result.new(false, '')
|
1106
|
+
end
|
1107
|
+
end
|
1108
|
+
|
1109
|
+
def download_dir(dataset_home, absolute_path)
|
1110
|
+
FileUtils.mkdir_p("#{dataset_home}/#{absolute_path}")
|
1111
|
+
end
|
1112
|
+
|
1113
|
+
def revoke_download_dir(absolute_path)
|
1114
|
+
puts FileUtils.rmtree("#{absolute_path}")
|
1115
|
+
end
|
1116
|
+
|
1117
|
+
def revoke_download_file(absolute_path, filename, conflict=false)
|
1118
|
+
begin
|
1119
|
+
file_location = absolute_path.gsub(/#{filename}\/?$/, "")
|
1120
|
+
|
1121
|
+
filename += ".conflict" if conflict
|
1122
|
+
FileUtils.remove("#{file_location}/#{filename}")
|
1123
|
+
return true
|
1124
|
+
rescue
|
1125
|
+
return false
|
1126
|
+
end
|
1127
|
+
end
|
1128
|
+
|
1129
|
+
def revoke_download(tar_files, extracted_files)
|
1130
|
+
begin
|
1131
|
+
FileUtils.rm_rf(tar_files) unless (tar_files.nil? or tar_files.empty?)
|
1132
|
+
FileUtils.rm_rf(extracted_files) unless (extracted_files.nil? or extracted_files.empty?)
|
1133
|
+
rescue => e
|
1134
|
+
return false
|
1135
|
+
end
|
1136
|
+
return true
|
1137
|
+
end
|
1138
|
+
|
1139
|
+
def delete_commit(commit_sha1)
|
1140
|
+
response = Cnvrg::API.request("#{base_resource}/commit/#{commit_sha1}", 'DELETE')
|
1141
|
+
Cnvrg::CLI.is_response_success(response, true)
|
1142
|
+
return response
|
1143
|
+
end
|
1144
|
+
|
1145
|
+
def get_commit(commit_sha1)
|
1146
|
+
response = Cnvrg::API.request("#{base_resource}/commit/#{commit_sha1}", 'GET')
|
1147
|
+
Cnvrg::CLI.is_response_success(response, true)
|
1148
|
+
return response
|
1149
|
+
end
|
1150
|
+
|
1151
|
+
def start_commit(new_branch, force=false, chunks: 0, dataset: @dataset, message:nil)
|
1152
|
+
#if we are pushing with force or to branch we dont need to send current/next commit cause we want to
|
1153
|
+
# create a new commit.
|
1154
|
+
idx = {}
|
1155
|
+
commit = idx[:commit]
|
1156
|
+
next_commit = idx[:next_commit]
|
1157
|
+
response = Cnvrg::API.request(
|
1158
|
+
"#{base_resource}/commit/start",
|
1159
|
+
'POST',
|
1160
|
+
{
|
1161
|
+
dataset_slug: @dataset_slug,
|
1162
|
+
new_branch: new_branch,
|
1163
|
+
force:force,
|
1164
|
+
username: @owner,
|
1165
|
+
current_commit: commit,
|
1166
|
+
next_commit: next_commit,
|
1167
|
+
total_chunks: chunks,
|
1168
|
+
message: message
|
1169
|
+
}
|
1170
|
+
)
|
1171
|
+
Cnvrg::CLI.is_response_success(response, true)
|
1172
|
+
response
|
1173
|
+
rescue => e
|
1174
|
+
false
|
1175
|
+
end
|
1176
|
+
|
1177
|
+
def end_commit(commit_sha1, force, success: true, uploaded_files: 0, commit_type: nil)
|
1178
|
+
begin
|
1179
|
+
response = Cnvrg::API.request(
|
1180
|
+
"#{base_resource}/commit/end",
|
1181
|
+
'POST',
|
1182
|
+
{
|
1183
|
+
commit_sha1: commit_sha1,
|
1184
|
+
force:force,
|
1185
|
+
success: success,
|
1186
|
+
uploaded_files: uploaded_files,
|
1187
|
+
commit_type: commit_type
|
1188
|
+
}
|
1189
|
+
)
|
1190
|
+
Cnvrg::CLI.is_response_success(response, true)
|
1191
|
+
return response
|
1192
|
+
rescue => e
|
1193
|
+
return false
|
1194
|
+
end
|
1195
|
+
end
|
1196
|
+
|
1197
|
+
def end_commit_tar(commit_sha1, cur_idx)
|
1198
|
+
response = Cnvrg::API.request("#{base_resource}/commit/end_tar", 'POST', {commit_sha1: commit_sha1, idx: cur_idx})
|
1199
|
+
return response
|
1200
|
+
end
|
1201
|
+
|
1202
|
+
def rollback_commit(commit_sha1)
|
1203
|
+
response = Cnvrg::API.request("#{base_resource}/commit/rollback", 'POST', {commit_sha1: commit_sha1})
|
1204
|
+
Cnvrg::CLI.is_response_success(response, false)
|
1205
|
+
end
|
1206
|
+
|
1207
|
+
def clone_in_chunks(commit: 'latest', chunk_size: 1000)
|
1208
|
+
begin
|
1209
|
+
|
1210
|
+
end
|
1211
|
+
end
|
1212
|
+
|
1213
|
+
def get_trees(commit: "latest")
|
1214
|
+
response = Cnvrg::API.request("#{@base_resource}/clone_trees", 'POST',{commit: commit})
|
1215
|
+
return nil unless Cnvrg::CLI.is_response_success(response, false)
|
1216
|
+
response['result']['files']
|
1217
|
+
end
|
1218
|
+
|
1219
|
+
def get_clone_chunk(latest_id: nil, chunk_size: 1000, commit: 'latest')
|
1220
|
+
response = Cnvrg::API.request("#{@base_resource}/clone_chunk", 'POST',{commit: commit, chunk_size: chunk_size, latest_id: latest_id})
|
1221
|
+
unless Cnvrg::CLI.is_response_success(response, false)
|
1222
|
+
Cnvrg::Logger.log_info("#{{commit: commit, chunk_size: chunk_size, latest_id: latest_id}}")
|
1223
|
+
return nil
|
1224
|
+
end
|
1225
|
+
response['result']['files']
|
1226
|
+
end
|
1227
|
+
|
1228
|
+
def download_files_in_chunks(files, chunk_size: 1000, conflict: false, commit: 'latest', progress: nil)
|
1229
|
+
begin
|
1230
|
+
files.each_slice(chunk_size).each do |files|
|
1231
|
+
download_files_chunk(files, conflict: conflict, progress: progress)
|
1232
|
+
end
|
1233
|
+
return Cnvrg::Result.new(true, "Download Completed")
|
1234
|
+
rescue Exception => e
|
1235
|
+
return Cnvrg::Result.new(false, "Can`t download files")
|
1236
|
+
end
|
1237
|
+
end
|
1238
|
+
|
1239
|
+
def download_files_chunk(files, conflict: false, progress: nil)
|
1240
|
+
(1..5).each do |i|
|
1241
|
+
response = Cnvrg::API.request("users/#{@owner}/datasets/#{@dataset_slug}/download_multi", 'POST', {files: files})
|
1242
|
+
next unless Cnvrg::CLI.is_response_success(response, false) #trying to api request 5 times.
|
1243
|
+
files_to_download = response['files']
|
1244
|
+
data_home = "#{Dir.pwd}/#{response['name']}"
|
1245
|
+
res = download_multiple_files_s3(files_to_download, data_home, conflict: conflict, read_only: false, progressbar: progress)
|
1246
|
+
next unless res.is_success? #try again..
|
1247
|
+
return files_to_download['keys'].length
|
1248
|
+
end
|
1249
|
+
end
|
1250
|
+
|
1251
|
+
def download_multiple_chunks(commit, chunk_size=1000, progress: nil)
|
1252
|
+
begin
|
1253
|
+
last_chunk_size = chunk_size
|
1254
|
+
q = { commit: commit, chunk_size: chunk_size}
|
1255
|
+
overall = 0
|
1256
|
+
while last_chunk_size > 0
|
1257
|
+
response = Cnvrg::API.request("users/#{@owner}/datasets/#{@dataset_slug}/clone", 'POST', q)
|
1258
|
+
if Cnvrg::CLI.is_response_success(response, false)
|
1259
|
+
files = response['files']
|
1260
|
+
data_home = "#{Dir.pwd}/#{response['name']}"
|
1261
|
+
last_chunk_size = files['keys'].length
|
1262
|
+
break if last_chunk_size == 0
|
1263
|
+
res = download_multiple_files_s3(files, data_home, read_only: false, progressbar: progress)
|
1264
|
+
overall += last_chunk_size
|
1265
|
+
q[:latest] = files['latest']
|
1266
|
+
else
|
1267
|
+
last_chunk_size = 0
|
1268
|
+
end
|
1269
|
+
end
|
1270
|
+
Cnvrg::Result.new(true, "Cloned #{overall} files!")
|
1271
|
+
rescue => e
|
1272
|
+
Cnvrg::Result.new(false, "Cant download chunk", e.message, e.backtrace)
|
1273
|
+
|
1274
|
+
end
|
1275
|
+
end
|
1276
|
+
|
1277
|
+
def last_valid_commit()
|
1278
|
+
begin
|
1279
|
+
response = Cnvrg::API.request("#{base_resource}/last_valid_commit", 'GET')
|
1280
|
+
Cnvrg::CLI.is_response_success(response, true)
|
1281
|
+
return response
|
1282
|
+
rescue => e
|
1283
|
+
return false
|
1284
|
+
end
|
1285
|
+
end
|
1286
|
+
|
1287
|
+
def download_multiple_files_s3(files, project_home, conflict: false, progressbar: nil, read_only:false, flatten: false)
|
1288
|
+
begin
|
1289
|
+
refresh_storage_token
|
1290
|
+
parallel_options = {
|
1291
|
+
in_threads: ParallelThreads,
|
1292
|
+
isolation: true
|
1293
|
+
}
|
1294
|
+
Parallel.map(files["keys"], parallel_options) do |f|
|
1295
|
+
begin
|
1296
|
+
file_path = f['name']
|
1297
|
+
file_path = File.basename(f['name']) if flatten
|
1298
|
+
local_path = @dataset.local_path + '/' + file_path
|
1299
|
+
Cnvrg::Logger.log_info("Downloading #{local_path}")
|
1300
|
+
progressbar.progress += 1 if progressbar.present?
|
1301
|
+
if local_path.end_with? "/"
|
1302
|
+
@downloader.mkdir(local_path, recursive: true)
|
1303
|
+
next
|
1304
|
+
end
|
1305
|
+
# blob
|
1306
|
+
local_path = "#{local_path}.conflict" if conflict
|
1307
|
+
storage_path = f["path"]
|
1308
|
+
# if File.exists? local_path
|
1309
|
+
# Cnvrg::Logger.log_info("Trying to download #{local_path} but its already exists, skipping..")
|
1310
|
+
# next
|
1311
|
+
# end
|
1312
|
+
resp = @downloader.download(storage_path, local_path)
|
1313
|
+
Cnvrg::Logger.log_info("Download #{local_path} success resp: #{resp}")
|
1314
|
+
rescue => e
|
1315
|
+
Cnvrg::Logger.log_error(e)
|
1316
|
+
end
|
1317
|
+
end
|
1318
|
+
return Cnvrg::Result.new(true,"Downloaded successfully")
|
1319
|
+
rescue => e
|
1320
|
+
Cnvrg::Logger.log_error(e)
|
1321
|
+
return Cnvrg::Result.new(false,"Could not download some files", e.message, e.backtrace)
|
1322
|
+
end
|
1323
|
+
end
|
1324
|
+
end
|
1325
|
+
end
|