fedora-migrate 0.1.0 → 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.gitignore +1 -0
- data/lib/fedora-migrate.rb +3 -5
- data/lib/fedora_migrate/content_mover.rb +32 -10
- data/lib/fedora_migrate/datastream_mover.rb +4 -7
- data/lib/fedora_migrate/datastream_verification.rb +1 -3
- data/lib/fedora_migrate/dates_mover.rb +19 -6
- data/lib/fedora_migrate/migration_report.rb +40 -0
- data/lib/fedora_migrate/mover.rb +11 -14
- data/lib/fedora_migrate/object_mover.rb +25 -9
- data/lib/fedora_migrate/permissions_mover.rb +2 -2
- data/lib/fedora_migrate/rdf_datastream_mover.rb +1 -1
- data/lib/fedora_migrate/rels_ext_datastream_mover.rb +5 -2
- data/lib/fedora_migrate/repository_migrator.rb +35 -22
- data/lib/fedora_migrate/target_constructor.rb +1 -2
- data/lib/fedora_migrate/version.rb +1 -1
- data/lib/tasks/fedora-migrate.rake +8 -2
- data/spec/fixtures/failed-report.json +339 -0
- data/spec/fixtures/sample-report.json +166 -0
- data/spec/integration/missing_relationships_spec.rb +15 -6
- data/spec/integration/repository_migration_spec.rb +25 -6
- data/spec/unit/content_mover_spec.rb +17 -10
- data/spec/unit/datastream_verification_spec.rb +2 -8
- data/spec/unit/migration_report_spec.rb +58 -0
- data/spec/unit/object_mover_spec.rb +2 -2
- data/spec/unit/repository_migrator_spec.rb +39 -10
- data/spec/unit/rubydora_connection_spec.rb +4 -0
- data/tasks/dev.rake +9 -3
- metadata +10 -3
@@ -3,53 +3,63 @@ module FedoraMigrate
|
|
3
3
|
|
4
4
|
include MigrationOptions
|
5
5
|
|
6
|
-
attr_accessor :source_objects, :namespace, :
|
6
|
+
attr_accessor :source_objects, :namespace, :report
|
7
|
+
|
8
|
+
SingleObjectReport = Struct.new(:status, :object, :relationships)
|
7
9
|
|
8
10
|
def initialize namespace = nil, options = {}
|
9
11
|
@namespace = namespace || repository_namespace
|
10
12
|
@options = options
|
11
|
-
@
|
13
|
+
@report = MigrationReport.new(@options.fetch(:report, nil))
|
12
14
|
@source_objects = get_source_objects
|
13
15
|
conversion_options
|
14
16
|
end
|
15
17
|
|
16
|
-
# TODO: need a reporting mechanism for results (issue #4)
|
17
18
|
def migrate_objects
|
18
19
|
source_objects.each { |source| migrate_object(source) }
|
19
|
-
@failed == 0
|
20
20
|
end
|
21
21
|
|
22
|
-
# TODO: need a reporting mechanism for results (issue #4)
|
23
22
|
def migrate_relationships
|
24
|
-
return "Relationship migration halted because #{
|
23
|
+
return "Relationship migration halted because #{failures.to_s} objects didn't migrate successfully." if failures > 0 && not_forced?
|
25
24
|
source_objects.each { |source| migrate_relationship(source) }
|
26
|
-
@failed == 0
|
27
25
|
end
|
28
26
|
|
29
27
|
def get_source_objects
|
30
|
-
|
28
|
+
if report.empty?
|
29
|
+
FedoraMigrate.source.connection.search(nil).collect { |o| qualifying_object(o) }.compact
|
30
|
+
else
|
31
|
+
report.failed_objects.map { |o| FedoraMigrate.source.connection.find(o) }
|
32
|
+
end
|
33
|
+
end
|
34
|
+
|
35
|
+
def failures
|
36
|
+
report.failed_objects.count
|
31
37
|
end
|
32
38
|
|
33
39
|
private
|
34
40
|
|
35
41
|
def migrate_object source
|
36
|
-
|
37
|
-
|
38
|
-
|
39
|
-
|
40
|
-
|
42
|
+
object_report = SingleObjectReport.new
|
43
|
+
begin
|
44
|
+
object_report.object = FedoraMigrate::ObjectMover.new(source, nil, options).migrate
|
45
|
+
object_report.status = true
|
46
|
+
rescue StandardError => e
|
47
|
+
object_report.object = e.inspect
|
48
|
+
object_report.status = false
|
49
|
+
end
|
50
|
+
report.results[source.pid] = object_report
|
41
51
|
end
|
42
52
|
|
43
53
|
def migrate_relationship source
|
44
|
-
|
45
|
-
|
46
|
-
|
47
|
-
|
48
|
-
|
49
|
-
|
50
|
-
|
51
|
-
|
52
|
-
[
|
54
|
+
relationship_report = find_or_create_single_object_report(source)
|
55
|
+
begin
|
56
|
+
relationship_report.relationships = FedoraMigrate::RelsExtDatastreamMover.new(source).migrate
|
57
|
+
relationship_report.status = true
|
58
|
+
rescue StandardError => e
|
59
|
+
relationship_report.relationships = e.inspect
|
60
|
+
relationship_report.status = false
|
61
|
+
end
|
62
|
+
report.results[source.pid] = relationship_report
|
53
63
|
end
|
54
64
|
|
55
65
|
def repository_namespace
|
@@ -61,5 +71,8 @@ module FedoraMigrate
|
|
61
71
|
return object if name.match(namespace)
|
62
72
|
end
|
63
73
|
|
74
|
+
def find_or_create_single_object_report source
|
75
|
+
report.results[source.pid] || SingleObjectReport.new
|
76
|
+
end
|
64
77
|
end
|
65
78
|
end
|
@@ -30,9 +30,8 @@ module FedoraMigrate
|
|
30
30
|
|
31
31
|
def vet model
|
32
32
|
@target = FedoraMigrate::Mover.id_component(model).constantize
|
33
|
-
Logger.info "using #{model} for target"
|
34
33
|
rescue NameError
|
35
|
-
Logger.
|
34
|
+
Logger.debug "rejecting #{model} for target"
|
36
35
|
end
|
37
36
|
|
38
37
|
end
|
@@ -8,8 +8,8 @@ namespace :fedora do
|
|
8
8
|
namespace :migrate do
|
9
9
|
desc "Migrates all objects in a Sufia-based application"
|
10
10
|
task sufia: :environment do
|
11
|
-
|
12
|
-
|
11
|
+
migrator = FedoraMigrate.migrate_repository(namespace: "sufia", options: {convert: "descMetadata"})
|
12
|
+
migrator.report.save
|
13
13
|
end
|
14
14
|
|
15
15
|
desc "Migrates only relationships in a Sufia-based application"
|
@@ -39,6 +39,12 @@ namespace :fedora do
|
|
39
39
|
raise "Please provide a pid, example changeme:1234" if args[:pid].nil?
|
40
40
|
FedoraMigrate::RelsExtDatastreamMover.new(FedoraMigrate.source.connection.find(args[:pid])).migrate
|
41
41
|
end
|
42
|
+
|
43
|
+
desc "Report the results of a migration"
|
44
|
+
task :report, [:file] => :environment do |t, args|
|
45
|
+
raise "Please provide a path to a report.json file" if args[:file].nil?
|
46
|
+
FedoraMigrate::MigrationReport.new(args[:file]).report_failures
|
47
|
+
end
|
42
48
|
|
43
49
|
end
|
44
50
|
|
@@ -0,0 +1,339 @@
|
|
1
|
+
{
|
2
|
+
"sufia:5m60qr94g": {
|
3
|
+
"status": false,
|
4
|
+
"object": "Sample object migration failure",
|
5
|
+
"relationships": [
|
6
|
+
|
7
|
+
]
|
8
|
+
},
|
9
|
+
"sufia:5m60qr95r": {
|
10
|
+
"status": true,
|
11
|
+
"object": {
|
12
|
+
"id": "5m60qr95r",
|
13
|
+
"class": "GenericFile",
|
14
|
+
"content_datastreams": [
|
15
|
+
{
|
16
|
+
"ds": "content",
|
17
|
+
"versions": [
|
18
|
+
{
|
19
|
+
"name": "file2.txt",
|
20
|
+
"mime_type": "text/plain",
|
21
|
+
"original_date": "2015-01-19T21:32:47Z"
|
22
|
+
}
|
23
|
+
]
|
24
|
+
},
|
25
|
+
{
|
26
|
+
"ds": "thumbnail",
|
27
|
+
"versions": [
|
28
|
+
{
|
29
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
30
|
+
}
|
31
|
+
]
|
32
|
+
},
|
33
|
+
{
|
34
|
+
"ds": "characterization",
|
35
|
+
"versions": [
|
36
|
+
{
|
37
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
38
|
+
}
|
39
|
+
]
|
40
|
+
}
|
41
|
+
],
|
42
|
+
"rdf_datastreams": [
|
43
|
+
{
|
44
|
+
"ds": "descMetadata",
|
45
|
+
"status": [
|
46
|
+
|
47
|
+
]
|
48
|
+
}
|
49
|
+
],
|
50
|
+
"permissions": [
|
51
|
+
"read_groups = []",
|
52
|
+
"edit_groups = []",
|
53
|
+
"discover_groups = []",
|
54
|
+
"read_users = []",
|
55
|
+
"edit_users = [\"awead@psu.edu\"]",
|
56
|
+
"discover_users = []"
|
57
|
+
],
|
58
|
+
"dates": {
|
59
|
+
"uploaded": null,
|
60
|
+
"modified": null
|
61
|
+
}
|
62
|
+
},
|
63
|
+
"relationships": [
|
64
|
+
"http://localhost:8983/fedora/rest/test/5m60qr95r--info:fedora/fedora-system:def/relations-external#isPartOf--http://localhost:8983/fedora/rest/test/5m60qr94g"
|
65
|
+
]
|
66
|
+
},
|
67
|
+
"sufia:5m60qr961": {
|
68
|
+
"status": true,
|
69
|
+
"object": {
|
70
|
+
"id": "5m60qr961",
|
71
|
+
"class": "GenericFile",
|
72
|
+
"content_datastreams": [
|
73
|
+
{
|
74
|
+
"ds": "content",
|
75
|
+
"versions": [
|
76
|
+
{
|
77
|
+
"name": "file1.txt",
|
78
|
+
"mime_type": "text/plain",
|
79
|
+
"original_date": "2015-01-19T21:32:49Z"
|
80
|
+
}
|
81
|
+
]
|
82
|
+
},
|
83
|
+
{
|
84
|
+
"ds": "thumbnail",
|
85
|
+
"versions": [
|
86
|
+
{
|
87
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
88
|
+
}
|
89
|
+
]
|
90
|
+
},
|
91
|
+
{
|
92
|
+
"ds": "characterization",
|
93
|
+
"versions": [
|
94
|
+
{
|
95
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
96
|
+
}
|
97
|
+
]
|
98
|
+
}
|
99
|
+
],
|
100
|
+
"rdf_datastreams": [
|
101
|
+
{
|
102
|
+
"ds": "descMetadata",
|
103
|
+
"status": [
|
104
|
+
|
105
|
+
]
|
106
|
+
}
|
107
|
+
],
|
108
|
+
"permissions": [
|
109
|
+
"read_groups = []",
|
110
|
+
"edit_groups = []",
|
111
|
+
"discover_groups = []",
|
112
|
+
"read_users = []",
|
113
|
+
"edit_users = [\"awead@psu.edu\"]",
|
114
|
+
"discover_users = []"
|
115
|
+
],
|
116
|
+
"dates": {
|
117
|
+
"uploaded": null,
|
118
|
+
"modified": null
|
119
|
+
}
|
120
|
+
},
|
121
|
+
"relationships": [
|
122
|
+
"http://localhost:8983/fedora/rest/test/5m60qr961--info:fedora/fedora-system:def/relations-external#isPartOf--http://localhost:8983/fedora/rest/test/5m60qr94g"
|
123
|
+
]
|
124
|
+
},
|
125
|
+
"sufia:5m60qr979": {
|
126
|
+
"status": true,
|
127
|
+
"object": {
|
128
|
+
"id": "5m60qr979",
|
129
|
+
"class": "Collection",
|
130
|
+
"content_datastreams": [
|
131
|
+
|
132
|
+
],
|
133
|
+
"rdf_datastreams": [
|
134
|
+
{
|
135
|
+
"ds": "descMetadata",
|
136
|
+
"status": [
|
137
|
+
|
138
|
+
]
|
139
|
+
}
|
140
|
+
],
|
141
|
+
"permissions": [
|
142
|
+
"read_groups = [\"public\"]",
|
143
|
+
"edit_groups = []",
|
144
|
+
"discover_groups = []",
|
145
|
+
"read_users = []",
|
146
|
+
"edit_users = [\"awead@psu.edu\"]",
|
147
|
+
"discover_users = []"
|
148
|
+
],
|
149
|
+
"dates": {
|
150
|
+
"uploaded": "2015-01-19T21:34:23.805Z",
|
151
|
+
"modified": "2015-03-03T18:48:51.16Z"
|
152
|
+
}
|
153
|
+
},
|
154
|
+
"relationships": [
|
155
|
+
"http://localhost:8983/fedora/rest/test/5m60qr979--info:fedora/fedora-system:def/relations-external#hasCollectionMember--http://localhost:8983/fedora/rest/test/5m60qr95r",
|
156
|
+
"http://localhost:8983/fedora/rest/test/5m60qr979--info:fedora/fedora-system:def/relations-external#hasCollectionMember--http://localhost:8983/fedora/rest/test/5m60qr961"
|
157
|
+
]
|
158
|
+
},
|
159
|
+
"sufia:rb68xc089": {
|
160
|
+
"status": false,
|
161
|
+
"object": "Sample object migration failure",
|
162
|
+
"relationships": [
|
163
|
+
|
164
|
+
]
|
165
|
+
},
|
166
|
+
"sufia:rb68xc09k": {
|
167
|
+
"status": true,
|
168
|
+
"object": {
|
169
|
+
"id": "rb68xc09k",
|
170
|
+
"class": "Batch",
|
171
|
+
"content_datastreams": [
|
172
|
+
|
173
|
+
],
|
174
|
+
"rdf_datastreams": [
|
175
|
+
|
176
|
+
],
|
177
|
+
"permissions": null,
|
178
|
+
"dates": {
|
179
|
+
"uploaded": null,
|
180
|
+
"modified": null
|
181
|
+
}
|
182
|
+
},
|
183
|
+
"relationships": [
|
184
|
+
|
185
|
+
]
|
186
|
+
},
|
187
|
+
"sufia:rb68xc10b": {
|
188
|
+
"status": true,
|
189
|
+
"object": {
|
190
|
+
"id": "rb68xc10b",
|
191
|
+
"class": "GenericFile",
|
192
|
+
"content_datastreams": [
|
193
|
+
{
|
194
|
+
"ds": "content",
|
195
|
+
"versions": [
|
196
|
+
|
197
|
+
]
|
198
|
+
},
|
199
|
+
{
|
200
|
+
"ds": "thumbnail",
|
201
|
+
"versions": [
|
202
|
+
{
|
203
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
204
|
+
}
|
205
|
+
]
|
206
|
+
},
|
207
|
+
{
|
208
|
+
"ds": "characterization",
|
209
|
+
"versions": [
|
210
|
+
{
|
211
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
212
|
+
}
|
213
|
+
]
|
214
|
+
}
|
215
|
+
],
|
216
|
+
"rdf_datastreams": [
|
217
|
+
|
218
|
+
],
|
219
|
+
"permissions": [
|
220
|
+
"read_groups = []",
|
221
|
+
"edit_groups = []",
|
222
|
+
"discover_groups = []",
|
223
|
+
"read_users = []",
|
224
|
+
"edit_users = [\"jilluser@example.com\"]",
|
225
|
+
"discover_users = []"
|
226
|
+
],
|
227
|
+
"dates": {
|
228
|
+
"uploaded": null,
|
229
|
+
"modified": null
|
230
|
+
}
|
231
|
+
},
|
232
|
+
"relationships": [
|
233
|
+
"http://localhost:8983/fedora/rest/test/rb68xc10b--info:fedora/fedora-system:def/relations-external#isPartOf--http://localhost:8983/fedora/rest/test/rb68xc09k"
|
234
|
+
]
|
235
|
+
},
|
236
|
+
"sufia:rb68xc11m": {
|
237
|
+
"status": true,
|
238
|
+
"object": {
|
239
|
+
"id": "rb68xc11m",
|
240
|
+
"class": "GenericFile",
|
241
|
+
"content_datastreams": [
|
242
|
+
{
|
243
|
+
"ds": "content",
|
244
|
+
"versions": [
|
245
|
+
|
246
|
+
]
|
247
|
+
},
|
248
|
+
{
|
249
|
+
"ds": "thumbnail",
|
250
|
+
"versions": [
|
251
|
+
{
|
252
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
253
|
+
}
|
254
|
+
]
|
255
|
+
},
|
256
|
+
{
|
257
|
+
"ds": "characterization",
|
258
|
+
"versions": [
|
259
|
+
{
|
260
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
261
|
+
}
|
262
|
+
]
|
263
|
+
}
|
264
|
+
],
|
265
|
+
"rdf_datastreams": [
|
266
|
+
|
267
|
+
],
|
268
|
+
"permissions": [
|
269
|
+
"read_groups = []",
|
270
|
+
"edit_groups = []",
|
271
|
+
"discover_groups = []",
|
272
|
+
"read_users = []",
|
273
|
+
"edit_users = [\"otherUser\"]",
|
274
|
+
"discover_users = []"
|
275
|
+
],
|
276
|
+
"dates": {
|
277
|
+
"uploaded": null,
|
278
|
+
"modified": null
|
279
|
+
}
|
280
|
+
},
|
281
|
+
"relationships": [
|
282
|
+
"http://localhost:8983/fedora/rest/test/rb68xc11m--info:fedora/fedora-system:def/relations-external#isPartOf--http://localhost:8983/fedora/rest/test/rb68xc09k"
|
283
|
+
]
|
284
|
+
},
|
285
|
+
"sufia:xp68km39w": {
|
286
|
+
"status": true,
|
287
|
+
"object": {
|
288
|
+
"id": "xp68km39w",
|
289
|
+
"class": "GenericFile",
|
290
|
+
"content_datastreams": [
|
291
|
+
{
|
292
|
+
"ds": "content",
|
293
|
+
"versions": [
|
294
|
+
|
295
|
+
]
|
296
|
+
},
|
297
|
+
{
|
298
|
+
"ds": "thumbnail",
|
299
|
+
"versions": [
|
300
|
+
{
|
301
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
302
|
+
}
|
303
|
+
]
|
304
|
+
},
|
305
|
+
{
|
306
|
+
"ds": "characterization",
|
307
|
+
"versions": [
|
308
|
+
{
|
309
|
+
"error": "Nil source -- it's probably defined in the target but not present in the source"
|
310
|
+
}
|
311
|
+
]
|
312
|
+
}
|
313
|
+
],
|
314
|
+
"rdf_datastreams": [
|
315
|
+
{
|
316
|
+
"ds": "descMetadata",
|
317
|
+
"status": [
|
318
|
+
|
319
|
+
]
|
320
|
+
}
|
321
|
+
],
|
322
|
+
"permissions": [
|
323
|
+
"read_groups = []",
|
324
|
+
"edit_groups = []",
|
325
|
+
"discover_groups = []",
|
326
|
+
"read_users = []",
|
327
|
+
"edit_users = [\"awead@psu.edu\"]",
|
328
|
+
"discover_users = []"
|
329
|
+
],
|
330
|
+
"dates": {
|
331
|
+
"uploaded": null,
|
332
|
+
"modified": null
|
333
|
+
}
|
334
|
+
},
|
335
|
+
"relationships": [
|
336
|
+
|
337
|
+
]
|
338
|
+
}
|
339
|
+
}
|