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 +1,892 @@
|
|
1
|
+
require 'fileutils'
|
2
|
+
module Cnvrg
|
3
|
+
class Dataset
|
4
|
+
attr_reader :slug, :owner, :title, :local_path, :working_dir
|
5
|
+
|
6
|
+
RemoteURL ||= "https://cnvrg.io"
|
7
|
+
IDXParallelThreads ||= Cnvrg::Helpers.parallel_threads
|
8
|
+
IDXParallelProcesses ||= Parallel.processor_count
|
9
|
+
|
10
|
+
def initialize(project_home = '', dataset_url: '', dataset_info: '')
|
11
|
+
begin
|
12
|
+
@info = {}
|
13
|
+
if project_home.present?
|
14
|
+
@local_path = project_home
|
15
|
+
@working_dir = project_home
|
16
|
+
config = YAML.load_file(project_home + "/.cnvrg/config.yml")
|
17
|
+
@title = config[:dataset_name]
|
18
|
+
@slug = config[:dataset_slug]
|
19
|
+
@owner = config[:owner]
|
20
|
+
elsif dataset_info.present?
|
21
|
+
@title = dataset_info[:slug]
|
22
|
+
@slug = dataset_info[:slug]
|
23
|
+
@owner = dataset_info[:owner]
|
24
|
+
@local_path = Dir.pwd
|
25
|
+
else
|
26
|
+
owner, slug = Cnvrg::Helpers.extract_owner_slug_from_url(dataset_url, 'datasets')
|
27
|
+
@title = slug
|
28
|
+
@slug = slug
|
29
|
+
@owner = owner
|
30
|
+
@local_path = Dir.pwd
|
31
|
+
end
|
32
|
+
rescue => e
|
33
|
+
|
34
|
+
end
|
35
|
+
end
|
36
|
+
|
37
|
+
def soft_linked?
|
38
|
+
@dataset_call["dataset_type"] == "soft_link_dataset"
|
39
|
+
end
|
40
|
+
|
41
|
+
def init_home(remote: false)
|
42
|
+
dataset_home = File.join(Dir.pwd, @slug)
|
43
|
+
if Dir.exists? dataset_home
|
44
|
+
if !remote
|
45
|
+
Cnvrg::CLI.log_message("Error: Conflict with dir #{@slug}", Thor::Shell::Color::RED)
|
46
|
+
if Thor::Shell::Basic.new.no? "Sync to repository anyway? (current data might lost)", Thor::Shell::Color::YELLOW
|
47
|
+
Cnvrg::CLI.log_message("Remove dir in order to clone #{@slug}", Thor::Shell::Color::RED)
|
48
|
+
exit(1)
|
49
|
+
end
|
50
|
+
end
|
51
|
+
FileUtils.rm_rf(dataset_home)
|
52
|
+
end
|
53
|
+
|
54
|
+
# if Dataset.clone(owner, dataset_name, slug, remote)
|
55
|
+
Dataset.clone(@owner, @slug, @slug, remote)
|
56
|
+
@local_path = dataset_home
|
57
|
+
Cnvrg::CLI.log_message('')
|
58
|
+
true
|
59
|
+
end
|
60
|
+
|
61
|
+
def get_dataset(commit: nil, query: nil)
|
62
|
+
if @dataset_call
|
63
|
+
return @dataset_call
|
64
|
+
end
|
65
|
+
response = Cnvrg::API.request("users/#{owner}/datasets/#{slug}/clone", 'POST',{ commit: commit, query:query})
|
66
|
+
Cnvrg::CLI.is_response_success(response,true)
|
67
|
+
@dataset_call = response["result"]
|
68
|
+
@dataset_call
|
69
|
+
end
|
70
|
+
|
71
|
+
def softlinked?
|
72
|
+
get_dataset["dataset_type"] == "soft_link_dataset"
|
73
|
+
end
|
74
|
+
|
75
|
+
|
76
|
+
def get_storage_client
|
77
|
+
response = Cnvrg::API.request("users/#{@owner}/datasets/#{@slug}/client", 'GET')
|
78
|
+
if Cnvrg::CLI.is_response_success(response, false)
|
79
|
+
client_params = response['client']
|
80
|
+
else
|
81
|
+
client_params = get_storage_client_fallback
|
82
|
+
end
|
83
|
+
Cnvrg::Downloader::Client.factory(client_params)
|
84
|
+
end
|
85
|
+
|
86
|
+
def get_storage_client_fallback
|
87
|
+
response = Cnvrg::API.request("users/#{@owner}/datasets/#{@slug}/download_multi", "POST", {files: []})
|
88
|
+
raise StandardError.new("Can't find dataset credentials") unless Cnvrg::CLI.is_response_success(response, false)
|
89
|
+
files = response['files']
|
90
|
+
storage = files['is_s3'] ? 's3' : 'minio'
|
91
|
+
files['storage'] = storage
|
92
|
+
files
|
93
|
+
end
|
94
|
+
|
95
|
+
def get_stats(commit: nil, query: nil)
|
96
|
+
response = Cnvrg::API.request("users/#{@owner}/datasets/#{@slug}/clone", 'POST', {commit: commit, query: query})
|
97
|
+
Cnvrg::CLI.is_response_success(response, true)
|
98
|
+
response['result']
|
99
|
+
end
|
100
|
+
|
101
|
+
def get_clone_chunk(latest_id: nil, chunk_size: 1000, offset: 0, commit: 'latest')
|
102
|
+
response = Cnvrg::API.request("users/#{@owner}/datasets/#{@slug}/clone_chunk", 'POST', {commit: commit, chunk_size: chunk_size, latest_id: latest_id, offset: offset})
|
103
|
+
return nil unless Cnvrg::CLI.is_response_success(response, false)
|
104
|
+
response['result']['files']['keys']
|
105
|
+
end
|
106
|
+
|
107
|
+
def backup_idx
|
108
|
+
Cnvrg::Logger.log_info("Backup idx")
|
109
|
+
if File.exists? "#{self.local_path}/.cnvrg/idx.yml"
|
110
|
+
FileUtils.cp "#{self.local_path}/.cnvrg/idx.yml", "#{self.local_path}/.cnvrg/idx.yml.backup"
|
111
|
+
else
|
112
|
+
idx = {commit: nil, tree: {}}
|
113
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml.backup", 'w') {|f| f.write idx.to_yaml}
|
114
|
+
end
|
115
|
+
end
|
116
|
+
|
117
|
+
def restore_idx
|
118
|
+
Cnvrg::Logger.log_info("Restore idx because an error.")
|
119
|
+
Cnvrg::Logger.log_method(bind: binding)
|
120
|
+
idx = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml.backup")
|
121
|
+
self.set_idx(idx)
|
122
|
+
end
|
123
|
+
|
124
|
+
def change_url(owner: '', slug: '', title: '')
|
125
|
+
config = {dataset_home: title, dataset_slug: slug, owner: owner}
|
126
|
+
File.open(".cnvrg/config.yml", "w+") {|f| f.write config.to_yaml}
|
127
|
+
end
|
128
|
+
|
129
|
+
def self.delete(dataset_slug, owner)
|
130
|
+
response = Cnvrg::API.request("users/#{owner}/datasets/#{dataset_slug}/delete", 'DELETE')
|
131
|
+
return response
|
132
|
+
end
|
133
|
+
|
134
|
+
def last_local_commit
|
135
|
+
if !File.exist? "#{self.local_path}/.cnvrg/idx.yml"
|
136
|
+
return nil
|
137
|
+
end
|
138
|
+
idx = YAML.load_file(@local_path + "/.cnvrg/idx.yml")
|
139
|
+
return idx[:commit]
|
140
|
+
end
|
141
|
+
|
142
|
+
def snapshot
|
143
|
+
commit = last_local_commit
|
144
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/volumes/create", 'POST', {data_commit: commit})
|
145
|
+
CLI.is_response_success(response)
|
146
|
+
return response
|
147
|
+
end
|
148
|
+
|
149
|
+
def list(owner)
|
150
|
+
response = Cnvrg::API.request("users/#{owner}/datasets/list", 'GET')
|
151
|
+
CLI.is_response_success(response)
|
152
|
+
return response
|
153
|
+
end
|
154
|
+
|
155
|
+
def search_queries
|
156
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/queries/list", 'GET')
|
157
|
+
CLI.is_response_success(response)
|
158
|
+
row = [["name", "id", "created_at", "username"]]
|
159
|
+
response["results"]["queries"].each do |query|
|
160
|
+
row << [query["name"], query["slug"], query["created_at"].in_time_zone.to_s, query["username"]]
|
161
|
+
end
|
162
|
+
return row
|
163
|
+
end
|
164
|
+
|
165
|
+
def get_query_file(query_slug)
|
166
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/search/#{query_slug}", 'GET')
|
167
|
+
CLI.is_response_success(response)
|
168
|
+
row = [["Name", "Full path", "URL"]]
|
169
|
+
response["results"]["query_files"].each do |file|
|
170
|
+
row << [file["name"], file["fullpath"], file["s3_url"]]
|
171
|
+
end
|
172
|
+
return row
|
173
|
+
end
|
174
|
+
|
175
|
+
def download_tags_yaml
|
176
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/tags_yml", 'GET')
|
177
|
+
CLI.is_response_success(response)
|
178
|
+
begin
|
179
|
+
path = self.working_dir
|
180
|
+
File.open("#{path}/#{response["results"]["filename"]}", "w+") {|f| f.write response["results"]["file_content"]}
|
181
|
+
return true
|
182
|
+
rescue
|
183
|
+
return false
|
184
|
+
end
|
185
|
+
end
|
186
|
+
|
187
|
+
def list_commits(commit_sha1: nil)
|
188
|
+
response = Cnvrg::API.request(
|
189
|
+
"users/#{self.owner}/datasets/#{self.slug}/list_commits?commit=#{commit_sha1}",
|
190
|
+
'GET'
|
191
|
+
)
|
192
|
+
CLI.is_response_success(response)
|
193
|
+
return response
|
194
|
+
end
|
195
|
+
|
196
|
+
def upload_tags_via_yml(tag_file = nil)
|
197
|
+
records_yml = YAML.load_file(tag_file)
|
198
|
+
tag_file.close
|
199
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/data_tags_create", 'POST', {records_yml: records_yml})
|
200
|
+
if response["status"] == 200
|
201
|
+
return true
|
202
|
+
else
|
203
|
+
return false
|
204
|
+
end
|
205
|
+
end
|
206
|
+
|
207
|
+
def url
|
208
|
+
url = Cnvrg::Helpers.remote_url
|
209
|
+
"#{url}/#{self.owner}/projects/#{self.slug}"
|
210
|
+
end
|
211
|
+
|
212
|
+
def self.verify_cnvrgignore_exist(dataset_name, remote)
|
213
|
+
path = ".cnvrgignore"
|
214
|
+
if !File.exist? path
|
215
|
+
path = "#{dataset_name}/.cnvrgignore"
|
216
|
+
end
|
217
|
+
ignore_exits = File.exist? path
|
218
|
+
if !ignore_exits
|
219
|
+
begin
|
220
|
+
list_files = [
|
221
|
+
path
|
222
|
+
]
|
223
|
+
FileUtils.touch list_files
|
224
|
+
cnvrgignore = Helpers.cnvrgignore_content
|
225
|
+
File.open(path, "w+") {|f| f.write cnvrgignore}
|
226
|
+
rescue => e
|
227
|
+
return false
|
228
|
+
end
|
229
|
+
|
230
|
+
end
|
231
|
+
end
|
232
|
+
|
233
|
+
def update_ignore_list(new_ignore)
|
234
|
+
|
235
|
+
if new_ignore.nil? or new_ignore.empty?
|
236
|
+
return true
|
237
|
+
end
|
238
|
+
begin
|
239
|
+
File.open(self.local_path + "/.cnvrgignore", "a+") do |f|
|
240
|
+
f.puts("\n")
|
241
|
+
|
242
|
+
new_ignore.each do |i|
|
243
|
+
f.puts("#{i}\n")
|
244
|
+
end
|
245
|
+
end
|
246
|
+
return true
|
247
|
+
rescue
|
248
|
+
return false
|
249
|
+
end
|
250
|
+
end
|
251
|
+
|
252
|
+
def get_ignore_list
|
253
|
+
### handle case when after clone .cnvrgignore doesnt exists
|
254
|
+
if not File.exists?(self.local_path + "/.cnvrgignore")
|
255
|
+
self.generate_cnvrg_ignore
|
256
|
+
end
|
257
|
+
|
258
|
+
ignore_list = []
|
259
|
+
if not File.exists? "#{self.local_path}/.cnvrgignore"
|
260
|
+
return ignore_list
|
261
|
+
end
|
262
|
+
File.open(self.local_path + "/.cnvrgignore", "r").each_line do |line|
|
263
|
+
line = line.strip
|
264
|
+
if line.start_with? "#" or ignore_list.include? line or line.empty?
|
265
|
+
next
|
266
|
+
end
|
267
|
+
if line.end_with? "/" or File.directory?(line)
|
268
|
+
ignore_list << line
|
269
|
+
all_sub = Dir.glob("#{line}/**/*", File::FNM_DOTMATCH).flatten
|
270
|
+
|
271
|
+
ignore_list << all_sub.flatten
|
272
|
+
elsif line.include? "*"
|
273
|
+
regex_list = Dir.glob("**/*#{line}", File::FNM_DOTMATCH).flatten
|
274
|
+
ignore_list << regex_list
|
275
|
+
else
|
276
|
+
ignore_list << line
|
277
|
+
end
|
278
|
+
end
|
279
|
+
return ignore_list.flatten
|
280
|
+
end
|
281
|
+
|
282
|
+
|
283
|
+
def self.init(owner, dataset_name, is_public = false, bucket: nil)
|
284
|
+
list_dirs = [".cnvrg"
|
285
|
+
]
|
286
|
+
list_files = [
|
287
|
+
".cnvrg/config.yml"
|
288
|
+
]
|
289
|
+
create_ignore = false
|
290
|
+
if !File.exist? ".cnvrgignore"
|
291
|
+
list_files << ".cnvrgignore"
|
292
|
+
create_ignore = true
|
293
|
+
end
|
294
|
+
|
295
|
+
cnvrgignore = Helpers.cnvrgignore_content
|
296
|
+
begin
|
297
|
+
response = Cnvrg::API.request("cli/create_dataset", 'POST', {title: dataset_name, owner: owner, is_public: is_public, bucket: bucket})
|
298
|
+
Cnvrg::CLI.is_response_success(response)
|
299
|
+
response = JSON.parse response["result"]
|
300
|
+
dataset_slug = response["slug"]
|
301
|
+
|
302
|
+
config = {dataset_name: dataset_name,
|
303
|
+
dataset_slug: dataset_slug,
|
304
|
+
owner: owner}
|
305
|
+
|
306
|
+
FileUtils.mkdir_p list_dirs
|
307
|
+
FileUtils.touch list_files
|
308
|
+
File.open(".cnvrg/config.yml", "w+") {|f| f.write config.to_yaml}
|
309
|
+
File.open(".cnvrgignore", "w+") {|f| f.write cnvrgignore} unless !create_ignore
|
310
|
+
rescue => e
|
311
|
+
return false
|
312
|
+
end
|
313
|
+
return true
|
314
|
+
end
|
315
|
+
|
316
|
+
def self.link_dataset(owner: nil, slug: nil)
|
317
|
+
begin
|
318
|
+
return false if owner.blank? or slug.blank?
|
319
|
+
|
320
|
+
response = Cnvrg::API.request("users/#{owner}/datasets/#{slug}", 'GET')
|
321
|
+
success = Cnvrg::CLI.is_response_success(response, false)
|
322
|
+
return unless success
|
323
|
+
result = response["result"]
|
324
|
+
|
325
|
+
sha1 = result["init_commit_sha1"]
|
326
|
+
|
327
|
+
# We need to write init IDX that contain init commit sha1 so the user will be able to doing actions on the dataset
|
328
|
+
# so it only relevant for new server
|
329
|
+
raise Exception.new("This feature is not available for your cnvrg version. Please contact support for more information") if sha1.blank? ## means this is old version of server
|
330
|
+
|
331
|
+
config = {dataset_name: result["title"],
|
332
|
+
dataset_slug: result["slug"],
|
333
|
+
owner: owner}
|
334
|
+
|
335
|
+
list_dirs = [".cnvrg"]
|
336
|
+
list_files = [".cnvrg/config.yml"]
|
337
|
+
|
338
|
+
|
339
|
+
FileUtils.mkdir_p list_dirs
|
340
|
+
FileUtils.touch list_files
|
341
|
+
File.open(".cnvrg/config.yml", "w+") {|f| f.write config.to_yaml}
|
342
|
+
|
343
|
+
dataset = Dataset.new(Dir.pwd)
|
344
|
+
dataset.write_idx({}, sha1)
|
345
|
+
true
|
346
|
+
rescue => e
|
347
|
+
raise Exception.new(e)
|
348
|
+
end
|
349
|
+
end
|
350
|
+
|
351
|
+
def self.blank_clone(owner, dataset_name, dataset_slug)
|
352
|
+
list_dirs = ["#{dataset_slug}/.cnvrg"
|
353
|
+
]
|
354
|
+
list_files = [
|
355
|
+
"#{dataset_slug}/.cnvrg/config.yml"
|
356
|
+
]
|
357
|
+
create_ignore = false
|
358
|
+
if !File.exist? ".cnvrgignore"
|
359
|
+
list_files << "#{dataset_slug}/.cnvrgignore"
|
360
|
+
create_ignore = true
|
361
|
+
end
|
362
|
+
|
363
|
+
|
364
|
+
cnvrgignore = Helpers.cnvrgignore_content
|
365
|
+
begin
|
366
|
+
|
367
|
+
config = {dataset_name: dataset_name,
|
368
|
+
dataset_slug: dataset_slug,
|
369
|
+
owner: owner}
|
370
|
+
|
371
|
+
FileUtils.mkdir_p list_dirs
|
372
|
+
FileUtils.touch list_files
|
373
|
+
File.open("#{dataset_slug}/.cnvrg/config.yml", "w+") {|f| f.write config.to_yaml}
|
374
|
+
File.open("#{dataset_slug}/.cnvrgignore", "w+") {|f| f.write cnvrgignore} unless !create_ignore
|
375
|
+
rescue => e
|
376
|
+
return false
|
377
|
+
end
|
378
|
+
return true
|
379
|
+
end
|
380
|
+
|
381
|
+
def generate_cnvrg_ignore
|
382
|
+
cnvrgignore = Helpers.cnvrgignore_content
|
383
|
+
File.open(self.local_path + "/.cnvrgignore", "w+") {|f| f.write cnvrgignore}
|
384
|
+
end
|
385
|
+
|
386
|
+
def self.verify_dataset(dataset_slug)
|
387
|
+
config = YAML.load_file("/data/#{dataset_title}/.cnvrg/config.yml") rescue {}
|
388
|
+
config[:success] == true
|
389
|
+
end
|
390
|
+
|
391
|
+
def self.verify_datasets(dataset_titles, timeout = nil)
|
392
|
+
start_time = Time.now.to_i
|
393
|
+
Cnvrg::Logger.log_info("Verifying datasets #{dataset_titles}")
|
394
|
+
Cnvrg::Logger.log_info("Timeout is #{timeout}")
|
395
|
+
while true
|
396
|
+
begin
|
397
|
+
current_time = Time.now.to_i
|
398
|
+
return false if (timeout.present? and timeout < current_time - start_time)
|
399
|
+
all_are_ready = dataset_titles.all? do |dataset_title|
|
400
|
+
config = YAML.load_file("#{dataset_title}/.cnvrg/config.yml")
|
401
|
+
config[:success] == true
|
402
|
+
end
|
403
|
+
return true if all_are_ready
|
404
|
+
Cnvrg::Logger.log_info("Sleeping..")
|
405
|
+
sleep 10
|
406
|
+
rescue => e
|
407
|
+
Cnvrg::Logger.log_info("Got error")
|
408
|
+
Cnvrg::Logger.log_error(e)
|
409
|
+
sleep 10
|
410
|
+
end
|
411
|
+
end
|
412
|
+
end
|
413
|
+
|
414
|
+
def self.scan_datasets()
|
415
|
+
Cnvrg::Logger.log_info("Looking up datasets")
|
416
|
+
datasets = Dir.entries(Dir.pwd).map do |entry|
|
417
|
+
if File.directory? File.join(Dir.pwd,entry) and !(entry =='.' || entry == '..')
|
418
|
+
begin
|
419
|
+
config = YAML.load_file("#{Dir.pwd}/#{entry}/.cnvrg/config.yml") rescue nil
|
420
|
+
local_commit = YAML.load_file("#{Dir.pwd}/#{entry}/.cnvrg/idx.yml")[:commit] rescue nil
|
421
|
+
if config.present? and config[:success] == true and config[:dataset_name].present? and config[:dataset_slug].present? and local_commit.present?
|
422
|
+
{
|
423
|
+
"dataset_slug": config[:dataset_slug],
|
424
|
+
"dataset_name": config[:dataset_name],
|
425
|
+
"local_commit": local_commit,
|
426
|
+
}
|
427
|
+
else
|
428
|
+
nil
|
429
|
+
end
|
430
|
+
rescue
|
431
|
+
nil
|
432
|
+
end
|
433
|
+
end
|
434
|
+
end.compact.uniq
|
435
|
+
datasets
|
436
|
+
end
|
437
|
+
|
438
|
+
def clone(commit)
|
439
|
+
return
|
440
|
+
end
|
441
|
+
|
442
|
+
def self.clone(owner, dataset_name, dataset_slug, remote = false)
|
443
|
+
begin
|
444
|
+
list_dirs = []
|
445
|
+
if !remote
|
446
|
+
list_dirs << dataset_name
|
447
|
+
end
|
448
|
+
list_dirs << "#{dataset_name}/.cnvrg"
|
449
|
+
list_files = [
|
450
|
+
"#{dataset_name}/.cnvrg/config.yml",
|
451
|
+
]
|
452
|
+
|
453
|
+
config = {dataset_name: dataset_name,
|
454
|
+
dataset_slug: dataset_slug,
|
455
|
+
owner: owner}
|
456
|
+
|
457
|
+
|
458
|
+
FileUtils.mkdir_p list_dirs
|
459
|
+
FileUtils.touch list_files
|
460
|
+
File.open("#{dataset_name}/.cnvrg/config.yml", "w+") {|f| f.write config.to_yaml}
|
461
|
+
rescue => e
|
462
|
+
puts "Exception in clone request:#{e.message}"
|
463
|
+
return false
|
464
|
+
end
|
465
|
+
return true
|
466
|
+
end
|
467
|
+
|
468
|
+
def list_files(commit_sha1: "latest", limit: 1000, offset: 0, expires: 3600)
|
469
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/list", 'GET', {commit_sha1: commit_sha1, limit: limit, offset: offset, expires: expires})
|
470
|
+
return nil if response.blank?
|
471
|
+
response.to_json
|
472
|
+
end
|
473
|
+
|
474
|
+
def self.clone_tree(commit: 'latest', dataset_home: nil)
|
475
|
+
@dataset = Cnvrg::Dataset.new(dataset_home)
|
476
|
+
@files = Cnvrg::Datafiles.new(@dataset.owner, @dataset.slug, dataset: @dataset)
|
477
|
+
trees = @files.get_trees(commit: commit)
|
478
|
+
return false if trees.nil?
|
479
|
+
pb = ProgressBar.create(:title => "Download Progress",
|
480
|
+
:progress_mark => '=',
|
481
|
+
:format => "%b>>%i| %p%% %t",
|
482
|
+
:starting_at => 0,
|
483
|
+
:total => trees.size,
|
484
|
+
:autofinish => true)
|
485
|
+
trees.each do |tree|
|
486
|
+
pb.progress += 1
|
487
|
+
@files.download_dir(dataset_home, tree)
|
488
|
+
end
|
489
|
+
pb.finish
|
490
|
+
@dataset.write_success
|
491
|
+
true
|
492
|
+
end
|
493
|
+
|
494
|
+
def write_success(in_folder = false)
|
495
|
+
file_path = ".cnvrg/config.yml"
|
496
|
+
file_path = File.join(@local_path || @working_dir, file_path)
|
497
|
+
if File.exist?(file_path)
|
498
|
+
File.open(file_path, "a") {|f| f.puts(":success: true")}
|
499
|
+
end
|
500
|
+
end
|
501
|
+
|
502
|
+
def self.init_container(owner, dataset_slug, dataset_name)
|
503
|
+
|
504
|
+
cnvrgignore = Helpers.cnvrgignore_content
|
505
|
+
begin
|
506
|
+
list_dirs = [".cnvrg"
|
507
|
+
]
|
508
|
+
list_files = [
|
509
|
+
".cnvrgignore",
|
510
|
+
".cnvrg/config.yml"
|
511
|
+
]
|
512
|
+
FileUtils.mkdir_p list_dirs
|
513
|
+
FileUtils.touch list_files
|
514
|
+
|
515
|
+
config = {dataset_name: dataset_name,
|
516
|
+
dataset_slug: dataset_slug,
|
517
|
+
owner: owner}
|
518
|
+
File.open(".cnvrg/config.yml", "w+") {|f| f.write config.to_yaml}
|
519
|
+
|
520
|
+
File.open(".cnvrgignore", "w+") {|f| f.write cnvrgignore} unless File.exist? ".cnvrgignore"
|
521
|
+
rescue => e
|
522
|
+
return false
|
523
|
+
end
|
524
|
+
return true
|
525
|
+
end
|
526
|
+
|
527
|
+
|
528
|
+
def get_idx
|
529
|
+
if File.exists? "#{self.local_path}/.cnvrg/idx.yml"
|
530
|
+
return YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
531
|
+
end
|
532
|
+
{commit: nil, tree: {}}
|
533
|
+
end
|
534
|
+
|
535
|
+
def set_idx(idx)
|
536
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w+') {|f| f.write idx.to_yaml}
|
537
|
+
end
|
538
|
+
|
539
|
+
def url
|
540
|
+
url = Cnvrg::Helpers.remote_url
|
541
|
+
"#{url}/#{self.owner}/datasets/#{self.slug}"
|
542
|
+
end
|
543
|
+
|
544
|
+
def generate_chunked_idx(list_files = [], threads: 15, prefix: '')
|
545
|
+
tree = {}
|
546
|
+
Parallel.map(list_files, in_threads: threads) do |file|
|
547
|
+
#check if prefix exists do prefix/path otherwise path
|
548
|
+
label = file.gsub(self.local_path + "/", "")
|
549
|
+
label = "#{prefix}/#{label}" if prefix.present?
|
550
|
+
if not Cnvrg::Files.valid_file_name?(label)
|
551
|
+
raise StandardError.new("#{label} is not a valid file name.")
|
552
|
+
end
|
553
|
+
if File.directory? file
|
554
|
+
tree[label + "/"] = nil
|
555
|
+
else
|
556
|
+
sha1 = OpenSSL::Digest::SHA1.file(file).hexdigest
|
557
|
+
file_name = File.basename file
|
558
|
+
file_size = File.size(file).to_f
|
559
|
+
mime_type = MimeMagic.by_path(file)
|
560
|
+
content_type = !(mime_type.nil? or mime_type.text?) ? mime_type.type : "text/plain"
|
561
|
+
relative_path = file.gsub(/^#{@local_path + "/"}/, "")
|
562
|
+
relative_path = "#{prefix}/#{relative_path}" if prefix.present?
|
563
|
+
tree[label] = {sha1: sha1, file_name: file_name, file_size: file_size, content_type: content_type, absolute_path: file, relative_path: relative_path}
|
564
|
+
end
|
565
|
+
end
|
566
|
+
if prefix.present? #add the prefix as dirs to the files
|
567
|
+
#lets say the prefix is a/b/c so we want that a/, a/b/, a/b/c/ will be in our files_list
|
568
|
+
dirs = prefix.split('/')
|
569
|
+
curr_path = []
|
570
|
+
dirs.each do |dir|
|
571
|
+
curr_path << dir
|
572
|
+
list_files << curr_path.join('/')
|
573
|
+
end
|
574
|
+
end
|
575
|
+
return tree
|
576
|
+
end
|
577
|
+
|
578
|
+
def revert_to_last_commit(commit: nil)
|
579
|
+
if commit.blank?
|
580
|
+
resp = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/last_valid_commit", 'GET')
|
581
|
+
if CLI.is_response_success(resp, false)
|
582
|
+
commit = resp['result']['commit_sha1']
|
583
|
+
end
|
584
|
+
end
|
585
|
+
self.update_idx_with_commit(commit) if commit.present?
|
586
|
+
self.revert_next_commit
|
587
|
+
end
|
588
|
+
|
589
|
+
def list_all_files(with_ignore = false)
|
590
|
+
list = Dir.glob("#{self.local_path}/**/*", File::FNM_DOTMATCH).reject {|x| (x =~ /\/\.{1,2}$/) or (x =~ /^#{self.local_path}\/\.cnvrg\/*/) or (x =~ /^#{self.local_path}\/\.cnvrgignore.conflict*/) and not (x =~ /^#{self.local_path}\/\.cnvrgignore/)}
|
591
|
+
|
592
|
+
#we want that big files will
|
593
|
+
list = list.sort_by {|fn| File.size(fn)}
|
594
|
+
return list if with_ignore
|
595
|
+
list_ignore = self.get_ignore_list.map {|ignore_file| "#{self.local_path}/#{ignore_file}"}
|
596
|
+
(list - list_ignore)
|
597
|
+
end
|
598
|
+
|
599
|
+
def write_idx(tree = nil, commit = nil)
|
600
|
+
if tree.nil?
|
601
|
+
tree = self.generate_idx[:tree]
|
602
|
+
tree = tree.map {|k, v| (v.present?) ? [k, {sha1: v[:sha1], commit_time: Time.now}] : [k, v]}.to_h
|
603
|
+
end
|
604
|
+
idx = {tree: tree, commit: commit}
|
605
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx.to_yaml}
|
606
|
+
end
|
607
|
+
|
608
|
+
def write_tree(tree)
|
609
|
+
idx = self.get_idx
|
610
|
+
idx[:tree] = tree
|
611
|
+
self.set_idx(idx)
|
612
|
+
end
|
613
|
+
|
614
|
+
def generate_idx(show_progress = false)
|
615
|
+
if File.exists? "#{self.local_path}/.cnvrg/idx.yml"
|
616
|
+
old_idx = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
617
|
+
else
|
618
|
+
old_idx = nil
|
619
|
+
end
|
620
|
+
tree_idx = Hash.new(0)
|
621
|
+
list = Dir.glob("#{self.local_path}/**/*", File::FNM_DOTMATCH).reject {|x| (x =~ /\/\.{1,2}$/) or (x =~ /^#{self.local_path}\/\.cnvrg\/*/) or (x =~ /^#{self.local_path}\/\.cnvrgignore.conflict*/) and not (x =~ /^#{self.local_path}\/\.cnvrgignore/)}
|
622
|
+
list_ignore = self.get_ignore_list()
|
623
|
+
if show_progress
|
624
|
+
parallel_options = {
|
625
|
+
:progress => {
|
626
|
+
:title => "Checking Dataset",
|
627
|
+
:progress_mark => '=',
|
628
|
+
:format => "%b>>%i| %p%% %t",
|
629
|
+
:starting_at => 0,
|
630
|
+
:total => (list).size,
|
631
|
+
:autofinish => true
|
632
|
+
},
|
633
|
+
in_threads: IDXParallelThreads,
|
634
|
+
isolation: true
|
635
|
+
}
|
636
|
+
else
|
637
|
+
parallel_options = {
|
638
|
+
in_threads: IDXParallelThreads,
|
639
|
+
isolation: true
|
640
|
+
}
|
641
|
+
end
|
642
|
+
|
643
|
+
Parallel.map(list, parallel_options) do |e|
|
644
|
+
label = e.gsub(self.local_path + "/", "")
|
645
|
+
if File.directory? e
|
646
|
+
if list_ignore.include? label
|
647
|
+
next
|
648
|
+
end
|
649
|
+
tree_idx[label + "/"] = nil
|
650
|
+
else
|
651
|
+
if list_ignore.include? label
|
652
|
+
next
|
653
|
+
end
|
654
|
+
sha1 = OpenSSL::Digest::SHA1.file(e).hexdigest
|
655
|
+
if old_idx.nil? or old_idx.to_h["tree"].nil?
|
656
|
+
tree_idx[label] = {sha1: sha1, commit_time: nil}
|
657
|
+
elsif old_idx["tree"][label].nil? or old_idx["tree"][label]["sha1"] != sha1
|
658
|
+
tree_idx[label] = {sha1: sha1, commit_time: nil}
|
659
|
+
else
|
660
|
+
tree_idx[label] = old_idx["tree"][label]
|
661
|
+
end
|
662
|
+
end
|
663
|
+
end
|
664
|
+
if !old_idx.nil? and !old_idx[:next_commit].nil? and !old_idx[:next_commit].empty?
|
665
|
+
idx = {commit: old_idx.to_h[:commit], tree: tree_idx, next_commit: old_idx[:next_commit]}
|
666
|
+
else
|
667
|
+
idx = {commit: old_idx.to_h[:commit], tree: tree_idx}
|
668
|
+
end
|
669
|
+
idx_yaml = idx.to_yaml
|
670
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx_yaml}
|
671
|
+
return idx
|
672
|
+
end
|
673
|
+
|
674
|
+
def create_volume
|
675
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/volumes/create", 'POST')
|
676
|
+
CLI.is_response_success(response)
|
677
|
+
return response
|
678
|
+
end
|
679
|
+
|
680
|
+
def download_updated_data(current_commit)
|
681
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/download_updated_data", 'POST', {current_commit: current_commit})
|
682
|
+
CLI.is_response_success(response, false)
|
683
|
+
return response
|
684
|
+
end
|
685
|
+
|
686
|
+
def compare_idx(new_branch, commit = last_local_commit, local_idx = nil, force = false, next_commit = nil)
|
687
|
+
if local_idx.nil?
|
688
|
+
local_idx = self.generate_idx
|
689
|
+
end
|
690
|
+
ignore_list = self.get_ignore_list()
|
691
|
+
if force
|
692
|
+
added = []
|
693
|
+
if local_idx[:tree]
|
694
|
+
added << local_idx[:tree].keys
|
695
|
+
added.flatten!
|
696
|
+
end
|
697
|
+
|
698
|
+
response = {"result" => {"commit" => next_commit, "tree" => {"added" => added,
|
699
|
+
"updated_on_server" => [],
|
700
|
+
"updated_on_local" => [],
|
701
|
+
"deleted" => [],
|
702
|
+
"conflicts" => []}}}
|
703
|
+
return response
|
704
|
+
|
705
|
+
end
|
706
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/status", 'POST', {idx: local_idx, new_branch: new_branch, current_commit: commit, ignore: ignore_list, next_commit: next_commit})
|
707
|
+
CLI.is_response_success(response, false)
|
708
|
+
return response
|
709
|
+
end
|
710
|
+
|
711
|
+
def compare_idx_download(all_files: false, desired_commit: nil)
|
712
|
+
current_commit = self.last_local_commit
|
713
|
+
next_commit = self.get_next_commit
|
714
|
+
ignore_list = self.send_ignore_list()
|
715
|
+
return Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/download_status", 'POST', {current_commit: current_commit, next_commit: next_commit, ignore: ignore_list, all_files: all_files, desired_commit: desired_commit.presence})
|
716
|
+
end
|
717
|
+
|
718
|
+
def set_partial_commit(commit_sha1)
|
719
|
+
idx = self.get_idx
|
720
|
+
idx[:partial_commit] = commit_sha1
|
721
|
+
self.set_idx(idx)
|
722
|
+
end
|
723
|
+
|
724
|
+
def get_partial_commit
|
725
|
+
idx = self.get_idx
|
726
|
+
idx.try(:fetch, :partial_commit)
|
727
|
+
end
|
728
|
+
|
729
|
+
def current_status(new_branch)
|
730
|
+
commit = last_local_commit
|
731
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/status_current", 'POST', {current_commit: commit, new_branch: new_branch})
|
732
|
+
CLI.is_response_success(response, true)
|
733
|
+
return response
|
734
|
+
end
|
735
|
+
|
736
|
+
def send_ignore_list()
|
737
|
+
begin
|
738
|
+
ignore_list = []
|
739
|
+
File.open(self.local_path + "/.cnvrgignore", "r").each_line do |line|
|
740
|
+
line = line.strip
|
741
|
+
if line.start_with? "#" or ignore_list.include? line or line.empty?
|
742
|
+
next
|
743
|
+
end
|
744
|
+
if line.end_with? "/"
|
745
|
+
ignore_list << line.gsub("/", "")
|
746
|
+
ignore_list << line + "."
|
747
|
+
elsif line.include? "*"
|
748
|
+
line = line.gsub("*", ".*")
|
749
|
+
ignore_list << line
|
750
|
+
else
|
751
|
+
ignore_list << line
|
752
|
+
end
|
753
|
+
end
|
754
|
+
return ignore_list.flatten
|
755
|
+
rescue
|
756
|
+
return []
|
757
|
+
end
|
758
|
+
end
|
759
|
+
|
760
|
+
|
761
|
+
def compare_commits(commit)
|
762
|
+
response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/compare_commits", 'POST', {compare_commit: commit, current_commit: last_local_commit})
|
763
|
+
CLI.is_response_success(response, false)
|
764
|
+
return response
|
765
|
+
end
|
766
|
+
|
767
|
+
def set_next_commit(commit_sha1)
|
768
|
+
if !File.exist? "#{self.local_path}/.cnvrg/idx.yml"
|
769
|
+
idx_hash = Hash.new()
|
770
|
+
idx_hash[:commit] = ""
|
771
|
+
idx_hash[:tree] = ""
|
772
|
+
else
|
773
|
+
idx_hash = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
774
|
+
end
|
775
|
+
idx_hash[:next_commit] = commit_sha1
|
776
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx_hash.to_yaml}
|
777
|
+
return true
|
778
|
+
|
779
|
+
end
|
780
|
+
|
781
|
+
def get_next_commit()
|
782
|
+
if !File.exist? "#{self.local_path}/.cnvrg/idx.yml"
|
783
|
+
return nil
|
784
|
+
end
|
785
|
+
idx_hash = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
786
|
+
return idx_hash[:next_commit]
|
787
|
+
end
|
788
|
+
|
789
|
+
def remove_next_commit()
|
790
|
+
if !File.exist? "#{self.local_path}/.cnvrg/idx.yml"
|
791
|
+
return nil
|
792
|
+
end
|
793
|
+
idx_hash = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
794
|
+
idx = Hash.new()
|
795
|
+
idx[:commit] = idx_hash[:next_commit]
|
796
|
+
idx[:tree] = idx_hash[:tree]
|
797
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx.to_yaml}
|
798
|
+
end
|
799
|
+
|
800
|
+
def revert_next_commit()
|
801
|
+
if !File.exist? "#{self.local_path}/.cnvrg/idx.yml"
|
802
|
+
return nil
|
803
|
+
end
|
804
|
+
idx_hash = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
805
|
+
idx_hash = idx_hash.except(:next_commit)
|
806
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx_hash.to_yaml}
|
807
|
+
end
|
808
|
+
|
809
|
+
def compare_commit(commit)
|
810
|
+
if commit.nil? or commit.empty?
|
811
|
+
commit = last_local_commit
|
812
|
+
end
|
813
|
+
response = Cnvrg::API.request("users/#{self.owner}/projects/#{self.slug}/commit/compare", 'POST', {current_commit: commit})
|
814
|
+
CLI.is_response_success(response, false)
|
815
|
+
update_is_new_branch(response["result"]["new_branch"])
|
816
|
+
return response["result"]["new_branch"]
|
817
|
+
end
|
818
|
+
|
819
|
+
def update_idx_with_files_commits!(files, commit_time)
|
820
|
+
# files.flatten!
|
821
|
+
idx_hash = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
822
|
+
# idx_hash[:commit] = commit
|
823
|
+
|
824
|
+
files.each do |path|
|
825
|
+
idx_hash[:tree].to_h[path].to_h[:commit_time] = commit_time
|
826
|
+
end
|
827
|
+
idx_hash[:next_commit] = idx_hash[:next_commit]
|
828
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx_hash.to_yaml}
|
829
|
+
|
830
|
+
return true
|
831
|
+
end
|
832
|
+
|
833
|
+
def update_idx(idx)
|
834
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx.to_yaml}
|
835
|
+
return true
|
836
|
+
end
|
837
|
+
|
838
|
+
def update_idx_with_commit!(commit)
|
839
|
+
idx_hash = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
|
840
|
+
idx_hash[:commit] = commit
|
841
|
+
|
842
|
+
File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') {|f| f.write idx_hash.to_yaml}
|
843
|
+
return true
|
844
|
+
end
|
845
|
+
|
846
|
+
def revert(working_dir)
|
847
|
+
FileUtils.rm_rf working_dir
|
848
|
+
# response = Cnvrg::API.request("users/#{self.owner}/projects/#{self.slug}/revert", 'GET')
|
849
|
+
# CLI.is_response_success(response)
|
850
|
+
end
|
851
|
+
|
852
|
+
def self.validate_config
|
853
|
+
## check that the .cnvrg folder exists:
|
854
|
+
dot_cnvrg_exists = Dir[".cnvrg"].present?
|
855
|
+
return {validation: Data::ConfigValidation::FAILED, message: ".cnvrg folder does not exists"} if not dot_cnvrg_exists
|
856
|
+
|
857
|
+
## check that the config.yml exists:
|
858
|
+
config_file_exists = Dir[".cnvrg/*"].include? ".cnvrg/config.yml"
|
859
|
+
return {validation: Data::ConfigValidation::FAILED, message: "config.yml exists"} if not config_file_exists
|
860
|
+
|
861
|
+
## check that the config.yml file not empty:
|
862
|
+
config = YAML.load_file("#{Dir.getwd}/.cnvrg/config.yml")
|
863
|
+
return {validation: Data::ConfigValidation::FAILED, message: "config.yml is empty"} if not config
|
864
|
+
|
865
|
+
## check that config.yml is valid:
|
866
|
+
title = config[:dataset_name]
|
867
|
+
slug = config[:dataset_slug]
|
868
|
+
owner = config[:owner]
|
869
|
+
return {validation: Data::ConfigValidation::FAILED, message: "config.yml is not valid or some keys are missing"} if title.blank? or slug.blank? or owner.blank?
|
870
|
+
|
871
|
+
## everything OK:
|
872
|
+
return {validation: Data::ConfigValidation::SUCCESS, message: "Directory is already linked to #{slug}"}
|
873
|
+
end
|
874
|
+
|
875
|
+
def self.stop_if_dataset_present(dataset_home, dataset_name, commit: nil)
|
876
|
+
|
877
|
+
cli = Cnvrg::CLI.new()
|
878
|
+
config = YAML.load_file(dataset_home + "/.cnvrg/config.yml")
|
879
|
+
if commit.present?
|
880
|
+
local_commit = YAML.load_file(dataset_home + "/.cnvrg/idx.yml")[:commit] rescue nil
|
881
|
+
return if commit != local_commit or local_commit.blank?
|
882
|
+
end
|
883
|
+
if config[:dataset_name] == dataset_name
|
884
|
+
cli.log_message("Dataset already present, clone aborted")
|
885
|
+
exit(0)
|
886
|
+
end
|
887
|
+
rescue => e
|
888
|
+
nil
|
889
|
+
end
|
890
|
+
|
891
|
+
end
|
892
|
+
end
|