mets_converter 0.2.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (28) hide show
  1. checksums.yaml +7 -0
  2. data/.gitignore +13 -0
  3. data/.rspec +2 -0
  4. data/.travis.yml +5 -0
  5. data/CODE_OF_CONDUCT.md +74 -0
  6. data/Gemfile +4 -0
  7. data/LICENSE +15 -0
  8. data/README.md +49 -0
  9. data/Rakefile +6 -0
  10. data/bin/console +14 -0
  11. data/bin/mets_to_yaml +84 -0
  12. data/bin/setup +8 -0
  13. data/lib/indent.rb +5 -0
  14. data/lib/mets_converter/logging.rb +26 -0
  15. data/lib/mets_converter/mets_parser.rb +37 -0
  16. data/lib/mets_converter/version.rb +3 -0
  17. data/lib/mets_converter/yaml_builder.rb +147 -0
  18. data/lib/mets_converter.rb +13 -0
  19. data/mets_converter.gemspec +27 -0
  20. data/spec/fixtures/35556002332765.mets.xml +2510 -0
  21. data/spec/fixtures/test.yml +257 -0
  22. data/spec/fixtures/test_with_options.yml +256 -0
  23. data/spec/mets_converter/mets_parser_spec.rb +33 -0
  24. data/spec/mets_converter/yaml_builder_spec.rb +28 -0
  25. data/spec/mets_converter_spec.rb +7 -0
  26. data/spec/spec_helper.rb +11 -0
  27. data/xsd/mets.xsd +1768 -0
  28. metadata +146 -0
data/spec/fixtures/test.yml ADDED
@@ -0,0 +1,257 @@
1
+ capture_date: 2015-09-03T14:28:35-06:00
2
+ scanner_make: Kirtas
3
+ scanner_model: APT 1200
4
+ scanner_user: "Northwestern University Library: Repository & Digital Curation"
5
+ contone_resolution_dpi: 300
6
+ image_compression_date: 2015-09-03T14:28:35-06:00
7
+ image_compression_agent: northwestern
8
+ image_compression_tool: ["LIMB v3.1.0.0"]
9
+ scanning_order: left-to-right
10
+ reading_order: left-to-right
11
+ pagedata:
12
+ 00000001.jp2: { label: "FRONT_COVER" }
13
+ 00000007.jp2: { label: "TITLE" }
14
+ 00000013.jp2: { orderlabel: "v", label: "PREFACE" }
15
+ 00000014.jp2: { orderlabel: "vi" }
16
+ 00000016.jp2: { orderlabel: "viii" }
17
+ 00000017.jp2: { orderlabel: "ix" }
18
+ 00000018.jp2: { orderlabel: "x" }
19
+ 00000019.jp2: { orderlabel: "xi" }
20
+ 00000020.jp2: { orderlabel: "xii" }
21
+ 00000021.jp2: { orderlabel: "xiii" }
22
+ 00000022.jp2: { orderlabel: "xiv" }
23
+ 00000023.jp2: { orderlabel: "xv" }
24
+ 00000024.jp2: { orderlabel: "xvi" }
25
+ 00000025.jp2: { orderlabel: "xvii" }
26
+ 00000027.jp2: { label: "TABLE_OF_CONTENTS" }
27
+ 00000029.jp2: { orderlabel: "1", label: "FIRST_CONTENT_CHAPTER_START" }
28
+ 00000030.jp2: { orderlabel: "2" }
29
+ 00000031.jp2: { orderlabel: "3" }
30
+ 00000032.jp2: { orderlabel: "4" }
31
+ 00000033.jp2: { orderlabel: "5" }
32
+ 00000034.jp2: { orderlabel: "6" }
33
+ 00000035.jp2: { orderlabel: "7" }
34
+ 00000036.jp2: { orderlabel: "8" }
35
+ 00000037.jp2: { orderlabel: "9" }
36
+ 00000038.jp2: { orderlabel: "10" }
37
+ 00000039.jp2: { orderlabel: "11" }
38
+ 00000040.jp2: { orderlabel: "12" }
39
+ 00000041.jp2: { orderlabel: "13" }
40
+ 00000042.jp2: { orderlabel: "14" }
41
+ 00000043.jp2: { orderlabel: "15" }
42
+ 00000044.jp2: { orderlabel: "16", label: "CHAPTER_START" }
43
+ 00000045.jp2: { orderlabel: "17" }
44
+ 00000046.jp2: { orderlabel: "18" }
45
+ 00000047.jp2: { orderlabel: "19" }
46
+ 00000048.jp2: { orderlabel: "20" }
47
+ 00000049.jp2: { orderlabel: "21" }
48
+ 00000050.jp2: { orderlabel: "22" }
49
+ 00000051.jp2: { orderlabel: "23" }
50
+ 00000052.jp2: { orderlabel: "24" }
51
+ 00000053.jp2: { orderlabel: "25" }
52
+ 00000054.jp2: { orderlabel: "26" }
53
+ 00000055.jp2: { orderlabel: "27" }
54
+ 00000056.jp2: { orderlabel: "28" }
55
+ 00000057.jp2: { orderlabel: "29" }
56
+ 00000058.jp2: { orderlabel: "30" }
57
+ 00000059.jp2: { orderlabel: "31" }
58
+ 00000060.jp2: { orderlabel: "32" }
59
+ 00000061.jp2: { orderlabel: "33" }
60
+ 00000062.jp2: { orderlabel: "34", label: "CHAPTER_START" }
61
+ 00000063.jp2: { orderlabel: "35" }
62
+ 00000064.jp2: { orderlabel: "36" }
63
+ 00000065.jp2: { orderlabel: "37" }
64
+ 00000066.jp2: { orderlabel: "38" }
65
+ 00000067.jp2: { orderlabel: "39" }
66
+ 00000068.jp2: { orderlabel: "40" }
67
+ 00000069.jp2: { orderlabel: "41" }
68
+ 00000070.jp2: { orderlabel: "42" }
69
+ 00000071.jp2: { orderlabel: "43" }
70
+ 00000072.jp2: { orderlabel: "44" }
71
+ 00000073.jp2: { orderlabel: "45" }
72
+ 00000074.jp2: { orderlabel: "46" }
73
+ 00000075.jp2: { orderlabel: "47" }
74
+ 00000076.jp2: { orderlabel: "48" }
75
+ 00000077.jp2: { orderlabel: "49" }
76
+ 00000078.jp2: { orderlabel: "50" }
77
+ 00000079.jp2: { orderlabel: "51" }
78
+ 00000080.jp2: { orderlabel: "52" }
79
+ 00000081.jp2: { orderlabel: "53" }
80
+ 00000082.jp2: { orderlabel: "54", label: "CHAPTER_START" }
81
+ 00000083.jp2: { orderlabel: "55" }
82
+ 00000084.jp2: { orderlabel: "56" }
83
+ 00000085.jp2: { orderlabel: "57" }
84
+ 00000086.jp2: { orderlabel: "58" }
85
+ 00000087.jp2: { orderlabel: "59" }
86
+ 00000088.jp2: { orderlabel: "60" }
87
+ 00000089.jp2: { orderlabel: "61" }
88
+ 00000090.jp2: { orderlabel: "62" }
89
+ 00000091.jp2: { orderlabel: "63" }
90
+ 00000092.jp2: { orderlabel: "64" }
91
+ 00000093.jp2: { orderlabel: "65" }
92
+ 00000094.jp2: { orderlabel: "66" }
93
+ 00000095.jp2: { orderlabel: "67" }
94
+ 00000096.jp2: { orderlabel: "68", label: "CHAPTER_START" }
95
+ 00000097.jp2: { orderlabel: "69" }
96
+ 00000098.jp2: { orderlabel: "70" }
97
+ 00000099.jp2: { orderlabel: "71" }
98
+ 00000100.jp2: { orderlabel: "72" }
99
+ 00000101.jp2: { orderlabel: "73" }
100
+ 00000102.jp2: { orderlabel: "74" }
101
+ 00000103.jp2: { orderlabel: "75" }
102
+ 00000104.jp2: { orderlabel: "76" }
103
+ 00000105.jp2: { orderlabel: "77" }
104
+ 00000106.jp2: { orderlabel: "78" }
105
+ 00000107.jp2: { orderlabel: "79" }
106
+ 00000108.jp2: { orderlabel: "80" }
107
+ 00000109.jp2: { orderlabel: "81" }
108
+ 00000110.jp2: { orderlabel: "82" }
109
+ 00000111.jp2: { orderlabel: "83" }
110
+ 00000112.jp2: { orderlabel: "84" }
111
+ 00000113.jp2: { orderlabel: "85" }
112
+ 00000114.jp2: { orderlabel: "86" }
113
+ 00000115.jp2: { orderlabel: "87" }
114
+ 00000116.jp2: { orderlabel: "88" }
115
+ 00000117.jp2: { orderlabel: "89" }
116
+ 00000118.jp2: { orderlabel: "90" }
117
+ 00000119.jp2: { orderlabel: "91" }
118
+ 00000120.jp2: { orderlabel: "92" }
119
+ 00000121.jp2: { orderlabel: "93", label: "CHAPTER_START" }
120
+ 00000122.jp2: { orderlabel: "94" }
121
+ 00000123.jp2: { orderlabel: "95" }
122
+ 00000124.jp2: { orderlabel: "96" }
123
+ 00000125.jp2: { orderlabel: "97" }
124
+ 00000126.jp2: { orderlabel: "98" }
125
+ 00000127.jp2: { orderlabel: "99" }
126
+ 00000128.jp2: { orderlabel: "100" }
127
+ 00000129.jp2: { orderlabel: "101" }
128
+ 00000130.jp2: { orderlabel: "102" }
129
+ 00000131.jp2: { orderlabel: "103" }
130
+ 00000132.jp2: { orderlabel: "104" }
131
+ 00000133.jp2: { orderlabel: "105" }
132
+ 00000134.jp2: { orderlabel: "106" }
133
+ 00000135.jp2: { orderlabel: "107" }
134
+ 00000136.jp2: { orderlabel: "108" }
135
+ 00000137.jp2: { orderlabel: "109" }
136
+ 00000138.jp2: { orderlabel: "110" }
137
+ 00000139.jp2: { orderlabel: "111" }
138
+ 00000140.jp2: { orderlabel: "112", label: "CHAPTER_START" }
139
+ 00000141.jp2: { orderlabel: "113" }
140
+ 00000142.jp2: { orderlabel: "114" }
141
+ 00000143.jp2: { orderlabel: "115" }
142
+ 00000144.jp2: { orderlabel: "116" }
143
+ 00000145.jp2: { orderlabel: "117" }
144
+ 00000146.jp2: { orderlabel: "118" }
145
+ 00000147.jp2: { orderlabel: "119" }
146
+ 00000148.jp2: { orderlabel: "120" }
147
+ 00000149.jp2: { orderlabel: "121" }
148
+ 00000150.jp2: { orderlabel: "122" }
149
+ 00000151.jp2: { orderlabel: "123" }
150
+ 00000152.jp2: { orderlabel: "124" }
151
+ 00000153.jp2: { orderlabel: "125" }
152
+ 00000154.jp2: { orderlabel: "126" }
153
+ 00000155.jp2: { orderlabel: "127" }
154
+ 00000156.jp2: { orderlabel: "128" }
155
+ 00000157.jp2: { orderlabel: "129" }
156
+ 00000158.jp2: { orderlabel: "130", label: "CHAPTER_START" }
157
+ 00000159.jp2: { orderlabel: "131" }
158
+ 00000160.jp2: { orderlabel: "132" }
159
+ 00000161.jp2: { orderlabel: "133" }
160
+ 00000162.jp2: { orderlabel: "134" }
161
+ 00000163.jp2: { orderlabel: "135" }
162
+ 00000164.jp2: { orderlabel: "136" }
163
+ 00000165.jp2: { orderlabel: "137" }
164
+ 00000166.jp2: { orderlabel: "138" }
165
+ 00000167.jp2: { orderlabel: "139" }
166
+ 00000168.jp2: { orderlabel: "140" }
167
+ 00000169.jp2: { orderlabel: "141" }
168
+ 00000170.jp2: { orderlabel: "142" }
169
+ 00000171.jp2: { orderlabel: "143" }
170
+ 00000172.jp2: { orderlabel: "144" }
171
+ 00000173.jp2: { orderlabel: "145" }
172
+ 00000174.jp2: { orderlabel: "146" }
173
+ 00000175.jp2: { orderlabel: "147" }
174
+ 00000176.jp2: { orderlabel: "148" }
175
+ 00000177.jp2: { orderlabel: "149", label: "CHAPTER_START" }
176
+ 00000178.jp2: { orderlabel: "150" }
177
+ 00000179.jp2: { orderlabel: "151" }
178
+ 00000180.jp2: { orderlabel: "152" }
179
+ 00000181.jp2: { orderlabel: "153" }
180
+ 00000182.jp2: { orderlabel: "154" }
181
+ 00000183.jp2: { orderlabel: "155" }
182
+ 00000184.jp2: { orderlabel: "156" }
183
+ 00000185.jp2: { orderlabel: "157" }
184
+ 00000186.jp2: { orderlabel: "158" }
185
+ 00000187.jp2: { orderlabel: "159" }
186
+ 00000188.jp2: { orderlabel: "160" }
187
+ 00000189.jp2: { orderlabel: "161" }
188
+ 00000190.jp2: { orderlabel: "162" }
189
+ 00000191.jp2: { orderlabel: "163" }
190
+ 00000192.jp2: { orderlabel: "164" }
191
+ 00000193.jp2: { orderlabel: "165" }
192
+ 00000194.jp2: { orderlabel: "166" }
193
+ 00000195.jp2: { orderlabel: "167" }
194
+ 00000196.jp2: { orderlabel: "168" }
195
+ 00000197.jp2: { orderlabel: "169" }
196
+ 00000198.jp2: { orderlabel: "170" }
197
+ 00000199.jp2: { orderlabel: "171" }
198
+ 00000200.jp2: { orderlabel: "172" }
199
+ 00000201.jp2: { orderlabel: "173" }
200
+ 00000202.jp2: { orderlabel: "174" }
201
+ 00000203.jp2: { orderlabel: "175", label: "CHAPTER_START" }
202
+ 00000204.jp2: { orderlabel: "176" }
203
+ 00000205.jp2: { orderlabel: "177" }
204
+ 00000206.jp2: { orderlabel: "178" }
205
+ 00000207.jp2: { orderlabel: "179" }
206
+ 00000208.jp2: { orderlabel: "180" }
207
+ 00000209.jp2: { orderlabel: "181" }
208
+ 00000210.jp2: { orderlabel: "182" }
209
+ 00000211.jp2: { orderlabel: "183" }
210
+ 00000212.jp2: { orderlabel: "184" }
211
+ 00000213.jp2: { orderlabel: "185" }
212
+ 00000214.jp2: { orderlabel: "186" }
213
+ 00000215.jp2: { orderlabel: "187" }
214
+ 00000216.jp2: { orderlabel: "188" }
215
+ 00000217.jp2: { orderlabel: "189" }
216
+ 00000218.jp2: { orderlabel: "190" }
217
+ 00000219.jp2: { orderlabel: "191" }
218
+ 00000220.jp2: { orderlabel: "192" }
219
+ 00000221.jp2: { orderlabel: "193" }
220
+ 00000222.jp2: { orderlabel: "194" }
221
+ 00000223.jp2: { orderlabel: "195", label: "CHAPTER_START" }
222
+ 00000224.jp2: { orderlabel: "196" }
223
+ 00000225.jp2: { orderlabel: "197" }
224
+ 00000226.jp2: { orderlabel: "198" }
225
+ 00000227.jp2: { orderlabel: "199" }
226
+ 00000228.jp2: { orderlabel: "200" }
227
+ 00000229.jp2: { orderlabel: "201" }
228
+ 00000230.jp2: { orderlabel: "202" }
229
+ 00000231.jp2: { orderlabel: "203" }
230
+ 00000232.jp2: { orderlabel: "204" }
231
+ 00000233.jp2: { orderlabel: "205" }
232
+ 00000234.jp2: { orderlabel: "206" }
233
+ 00000235.jp2: { orderlabel: "207" }
234
+ 00000236.jp2: { orderlabel: "208" }
235
+ 00000237.jp2: { orderlabel: "209" }
236
+ 00000238.jp2: { orderlabel: "210" }
237
+ 00000239.jp2: { orderlabel: "211", label: "REFERENCES" }
238
+ 00000240.jp2: { orderlabel: "212" }
239
+ 00000241.jp2: { orderlabel: "213" }
240
+ 00000242.jp2: { orderlabel: "214" }
241
+ 00000243.jp2: { orderlabel: "215" }
242
+ 00000244.jp2: { orderlabel: "216" }
243
+ 00000245.jp2: { orderlabel: "217" }
244
+ 00000246.jp2: { orderlabel: "218" }
245
+ 00000247.jp2: { orderlabel: "219" }
246
+ 00000248.jp2: { orderlabel: "220" }
247
+ 00000249.jp2: { orderlabel: "221" }
248
+ 00000250.jp2: { orderlabel: "222" }
249
+ 00000251.jp2: { orderlabel: "223" }
250
+ 00000252.jp2: { orderlabel: "224" }
251
+ 00000253.jp2: { label: "INDEX" }
252
+ 00000255.jp2: { orderlabel: "227" }
253
+ 00000256.jp2: { orderlabel: "228" }
254
+ 00000257.jp2: { orderlabel: "229" }
255
+ 00000258.jp2: { orderlabel: "230" }
256
+ 00000259.jp2: { orderlabel: "231" }
257
+ 00000268.jp2: { label: "BACK_COVER" }
data/spec/fixtures/test_with_options.yml ADDED
@@ -0,0 +1,256 @@
1
+ capture_date: 2015-09-03T14:28:35-06:00
2
+ scanner_make: SupraScan
3
+ scanner_model: Quartz A1
4
+ scanner_user: "Northwestern University Library: Repository & Digital Curation"
5
+ contone_resolution_dpi: 2000
6
+ image_compression_date: 2015-09-03T14:28:35-06:00
7
+ image_compression_agent: northwestern
8
+ image_compression_tool: ["LIMB v3.1.0.0"]
9
+ scanning_order: right-to-left
10
+ reading_order: right-to-left
11
+ 00000001.jp2: { label: "FRONT_COVER" }
12
+ 00000007.jp2: { label: "TITLE" }
13
+ 00000013.jp2: { orderlabel: "v", label: "PREFACE" }
14
+ 00000014.jp2: { orderlabel: "vi" }
15
+ 00000016.jp2: { orderlabel: "viii" }
16
+ 00000017.jp2: { orderlabel: "ix" }
17
+ 00000018.jp2: { orderlabel: "x" }
18
+ 00000019.jp2: { orderlabel: "xi" }
19
+ 00000020.jp2: { orderlabel: "xii" }
20
+ 00000021.jp2: { orderlabel: "xiii" }
21
+ 00000022.jp2: { orderlabel: "xiv" }
22
+ 00000023.jp2: { orderlabel: "xv" }
23
+ 00000024.jp2: { orderlabel: "xvi" }
24
+ 00000025.jp2: { orderlabel: "xvii" }
25
+ 00000027.jp2: { label: "TABLE_OF_CONTENTS" }
26
+ 00000029.jp2: { orderlabel: "1", label: "FIRST_CONTENT_CHAPTER_START" }
27
+ 00000030.jp2: { orderlabel: "2" }
28
+ 00000031.jp2: { orderlabel: "3" }
29
+ 00000032.jp2: { orderlabel: "4" }
30
+ 00000033.jp2: { orderlabel: "5" }
31
+ 00000034.jp2: { orderlabel: "6" }
32
+ 00000035.jp2: { orderlabel: "7" }
33
+ 00000036.jp2: { orderlabel: "8" }
34
+ 00000037.jp2: { orderlabel: "9" }
35
+ 00000038.jp2: { orderlabel: "10" }
36
+ 00000039.jp2: { orderlabel: "11" }
37
+ 00000040.jp2: { orderlabel: "12" }
38
+ 00000041.jp2: { orderlabel: "13" }
39
+ 00000042.jp2: { orderlabel: "14" }
40
+ 00000043.jp2: { orderlabel: "15" }
41
+ 00000044.jp2: { orderlabel: "16", label: "CHAPTER_START" }
42
+ 00000045.jp2: { orderlabel: "17" }
43
+ 00000046.jp2: { orderlabel: "18" }
44
+ 00000047.jp2: { orderlabel: "19" }
45
+ 00000048.jp2: { orderlabel: "20" }
46
+ 00000049.jp2: { orderlabel: "21" }
47
+ 00000050.jp2: { orderlabel: "22" }
48
+ 00000051.jp2: { orderlabel: "23" }
49
+ 00000052.jp2: { orderlabel: "24" }
50
+ 00000053.jp2: { orderlabel: "25" }
51
+ 00000054.jp2: { orderlabel: "26" }
52
+ 00000055.jp2: { orderlabel: "27" }
53
+ 00000056.jp2: { orderlabel: "28" }
54
+ 00000057.jp2: { orderlabel: "29" }
55
+ 00000058.jp2: { orderlabel: "30" }
56
+ 00000059.jp2: { orderlabel: "31" }
57
+ 00000060.jp2: { orderlabel: "32" }
58
+ 00000061.jp2: { orderlabel: "33" }
59
+ 00000062.jp2: { orderlabel: "34", label: "CHAPTER_START" }
60
+ 00000063.jp2: { orderlabel: "35" }
61
+ 00000064.jp2: { orderlabel: "36" }
62
+ 00000065.jp2: { orderlabel: "37" }
63
+ 00000066.jp2: { orderlabel: "38" }
64
+ 00000067.jp2: { orderlabel: "39" }
65
+ 00000068.jp2: { orderlabel: "40" }
66
+ 00000069.jp2: { orderlabel: "41" }
67
+ 00000070.jp2: { orderlabel: "42" }
68
+ 00000071.jp2: { orderlabel: "43" }
69
+ 00000072.jp2: { orderlabel: "44" }
70
+ 00000073.jp2: { orderlabel: "45" }
71
+ 00000074.jp2: { orderlabel: "46" }
72
+ 00000075.jp2: { orderlabel: "47" }
73
+ 00000076.jp2: { orderlabel: "48" }
74
+ 00000077.jp2: { orderlabel: "49" }
75
+ 00000078.jp2: { orderlabel: "50" }
76
+ 00000079.jp2: { orderlabel: "51" }
77
+ 00000080.jp2: { orderlabel: "52" }
78
+ 00000081.jp2: { orderlabel: "53" }
79
+ 00000082.jp2: { orderlabel: "54", label: "CHAPTER_START" }
80
+ 00000083.jp2: { orderlabel: "55" }
81
+ 00000084.jp2: { orderlabel: "56" }
82
+ 00000085.jp2: { orderlabel: "57" }
83
+ 00000086.jp2: { orderlabel: "58" }
84
+ 00000087.jp2: { orderlabel: "59" }
85
+ 00000088.jp2: { orderlabel: "60" }
86
+ 00000089.jp2: { orderlabel: "61" }
87
+ 00000090.jp2: { orderlabel: "62" }
88
+ 00000091.jp2: { orderlabel: "63" }
89
+ 00000092.jp2: { orderlabel: "64" }
90
+ 00000093.jp2: { orderlabel: "65" }
91
+ 00000094.jp2: { orderlabel: "66" }
92
+ 00000095.jp2: { orderlabel: "67" }
93
+ 00000096.jp2: { orderlabel: "68", label: "CHAPTER_START" }
94
+ 00000097.jp2: { orderlabel: "69" }
95
+ 00000098.jp2: { orderlabel: "70" }
96
+ 00000099.jp2: { orderlabel: "71" }
97
+ 00000100.jp2: { orderlabel: "72" }
98
+ 00000101.jp2: { orderlabel: "73" }
99
+ 00000102.jp2: { orderlabel: "74" }
100
+ 00000103.jp2: { orderlabel: "75" }
101
+ 00000104.jp2: { orderlabel: "76" }
102
+ 00000105.jp2: { orderlabel: "77" }
103
+ 00000106.jp2: { orderlabel: "78" }
104
+ 00000107.jp2: { orderlabel: "79" }
105
+ 00000108.jp2: { orderlabel: "80" }
106
+ 00000109.jp2: { orderlabel: "81" }
107
+ 00000110.jp2: { orderlabel: "82" }
108
+ 00000111.jp2: { orderlabel: "83" }
109
+ 00000112.jp2: { orderlabel: "84" }
110
+ 00000113.jp2: { orderlabel: "85" }
111
+ 00000114.jp2: { orderlabel: "86" }
112
+ 00000115.jp2: { orderlabel: "87" }
113
+ 00000116.jp2: { orderlabel: "88" }
114
+ 00000117.jp2: { orderlabel: "89" }
115
+ 00000118.jp2: { orderlabel: "90" }
116
+ 00000119.jp2: { orderlabel: "91" }
117
+ 00000120.jp2: { orderlabel: "92" }
118
+ 00000121.jp2: { orderlabel: "93", label: "CHAPTER_START" }
119
+ 00000122.jp2: { orderlabel: "94" }
120
+ 00000123.jp2: { orderlabel: "95" }
121
+ 00000124.jp2: { orderlabel: "96" }
122
+ 00000125.jp2: { orderlabel: "97" }
123
+ 00000126.jp2: { orderlabel: "98" }
124
+ 00000127.jp2: { orderlabel: "99" }
125
+ 00000128.jp2: { orderlabel: "100" }
126
+ 00000129.jp2: { orderlabel: "101" }
127
+ 00000130.jp2: { orderlabel: "102" }
128
+ 00000131.jp2: { orderlabel: "103" }
129
+ 00000132.jp2: { orderlabel: "104" }
130
+ 00000133.jp2: { orderlabel: "105" }
131
+ 00000134.jp2: { orderlabel: "106" }
132
+ 00000135.jp2: { orderlabel: "107" }
133
+ 00000136.jp2: { orderlabel: "108" }
134
+ 00000137.jp2: { orderlabel: "109" }
135
+ 00000138.jp2: { orderlabel: "110" }
136
+ 00000139.jp2: { orderlabel: "111" }
137
+ 00000140.jp2: { orderlabel: "112", label: "CHAPTER_START" }
138
+ 00000141.jp2: { orderlabel: "113" }
139
+ 00000142.jp2: { orderlabel: "114" }
140
+ 00000143.jp2: { orderlabel: "115" }
141
+ 00000144.jp2: { orderlabel: "116" }
142
+ 00000145.jp2: { orderlabel: "117" }
143
+ 00000146.jp2: { orderlabel: "118" }
144
+ 00000147.jp2: { orderlabel: "119" }
145
+ 00000148.jp2: { orderlabel: "120" }
146
+ 00000149.jp2: { orderlabel: "121" }
147
+ 00000150.jp2: { orderlabel: "122" }
148
+ 00000151.jp2: { orderlabel: "123" }
149
+ 00000152.jp2: { orderlabel: "124" }
150
+ 00000153.jp2: { orderlabel: "125" }
151
+ 00000154.jp2: { orderlabel: "126" }
152
+ 00000155.jp2: { orderlabel: "127" }
153
+ 00000156.jp2: { orderlabel: "128" }
154
+ 00000157.jp2: { orderlabel: "129" }
155
+ 00000158.jp2: { orderlabel: "130", label: "CHAPTER_START" }
156
+ 00000159.jp2: { orderlabel: "131" }
157
+ 00000160.jp2: { orderlabel: "132" }
158
+ 00000161.jp2: { orderlabel: "133" }
159
+ 00000162.jp2: { orderlabel: "134" }
160
+ 00000163.jp2: { orderlabel: "135" }
161
+ 00000164.jp2: { orderlabel: "136" }
162
+ 00000165.jp2: { orderlabel: "137" }
163
+ 00000166.jp2: { orderlabel: "138" }
164
+ 00000167.jp2: { orderlabel: "139" }
165
+ 00000168.jp2: { orderlabel: "140" }
166
+ 00000169.jp2: { orderlabel: "141" }
167
+ 00000170.jp2: { orderlabel: "142" }
168
+ 00000171.jp2: { orderlabel: "143" }
169
+ 00000172.jp2: { orderlabel: "144" }
170
+ 00000173.jp2: { orderlabel: "145" }
171
+ 00000174.jp2: { orderlabel: "146" }
172
+ 00000175.jp2: { orderlabel: "147" }
173
+ 00000176.jp2: { orderlabel: "148" }
174
+ 00000177.jp2: { orderlabel: "149", label: "CHAPTER_START" }
175
+ 00000178.jp2: { orderlabel: "150" }
176
+ 00000179.jp2: { orderlabel: "151" }
177
+ 00000180.jp2: { orderlabel: "152" }
178
+ 00000181.jp2: { orderlabel: "153" }
179
+ 00000182.jp2: { orderlabel: "154" }
180
+ 00000183.jp2: { orderlabel: "155" }
181
+ 00000184.jp2: { orderlabel: "156" }
182
+ 00000185.jp2: { orderlabel: "157" }
183
+ 00000186.jp2: { orderlabel: "158" }
184
+ 00000187.jp2: { orderlabel: "159" }
185
+ 00000188.jp2: { orderlabel: "160" }
186
+ 00000189.jp2: { orderlabel: "161" }
187
+ 00000190.jp2: { orderlabel: "162" }
188
+ 00000191.jp2: { orderlabel: "163" }
189
+ 00000192.jp2: { orderlabel: "164" }
190
+ 00000193.jp2: { orderlabel: "165" }
191
+ 00000194.jp2: { orderlabel: "166" }
192
+ 00000195.jp2: { orderlabel: "167" }
193
+ 00000196.jp2: { orderlabel: "168" }
194
+ 00000197.jp2: { orderlabel: "169" }
195
+ 00000198.jp2: { orderlabel: "170" }
196
+ 00000199.jp2: { orderlabel: "171" }
197
+ 00000200.jp2: { orderlabel: "172" }
198
+ 00000201.jp2: { orderlabel: "173" }
199
+ 00000202.jp2: { orderlabel: "174" }
200
+ 00000203.jp2: { orderlabel: "175", label: "CHAPTER_START" }
201
+ 00000204.jp2: { orderlabel: "176" }
202
+ 00000205.jp2: { orderlabel: "177" }
203
+ 00000206.jp2: { orderlabel: "178" }
204
+ 00000207.jp2: { orderlabel: "179" }
205
+ 00000208.jp2: { orderlabel: "180" }
206
+ 00000209.jp2: { orderlabel: "181" }
207
+ 00000210.jp2: { orderlabel: "182" }
208
+ 00000211.jp2: { orderlabel: "183" }
209
+ 00000212.jp2: { orderlabel: "184" }
210
+ 00000213.jp2: { orderlabel: "185" }
211
+ 00000214.jp2: { orderlabel: "186" }
212
+ 00000215.jp2: { orderlabel: "187" }
213
+ 00000216.jp2: { orderlabel: "188" }
214
+ 00000217.jp2: { orderlabel: "189" }
215
+ 00000218.jp2: { orderlabel: "190" }
216
+ 00000219.jp2: { orderlabel: "191" }
217
+ 00000220.jp2: { orderlabel: "192" }
218
+ 00000221.jp2: { orderlabel: "193" }
219
+ 00000222.jp2: { orderlabel: "194" }
220
+ 00000223.jp2: { orderlabel: "195", label: "CHAPTER_START" }
221
+ 00000224.jp2: { orderlabel: "196" }
222
+ 00000225.jp2: { orderlabel: "197" }
223
+ 00000226.jp2: { orderlabel: "198" }
224
+ 00000227.jp2: { orderlabel: "199" }
225
+ 00000228.jp2: { orderlabel: "200" }
226
+ 00000229.jp2: { orderlabel: "201" }
227
+ 00000230.jp2: { orderlabel: "202" }
228
+ 00000231.jp2: { orderlabel: "203" }
229
+ 00000232.jp2: { orderlabel: "204" }
230
+ 00000233.jp2: { orderlabel: "205" }
231
+ 00000234.jp2: { orderlabel: "206" }
232
+ 00000235.jp2: { orderlabel: "207" }
233
+ 00000236.jp2: { orderlabel: "208" }
234
+ 00000237.jp2: { orderlabel: "209" }
235
+ 00000238.jp2: { orderlabel: "210" }
236
+ 00000239.jp2: { orderlabel: "211", label: "REFERENCES" }
237
+ 00000240.jp2: { orderlabel: "212" }
238
+ 00000241.jp2: { orderlabel: "213" }
239
+ 00000242.jp2: { orderlabel: "214" }
240
+ 00000243.jp2: { orderlabel: "215" }
241
+ 00000244.jp2: { orderlabel: "216" }
242
+ 00000245.jp2: { orderlabel: "217" }
243
+ 00000246.jp2: { orderlabel: "218" }
244
+ 00000247.jp2: { orderlabel: "219" }
245
+ 00000248.jp2: { orderlabel: "220" }
246
+ 00000249.jp2: { orderlabel: "221" }
247
+ 00000250.jp2: { orderlabel: "222" }
248
+ 00000251.jp2: { orderlabel: "223" }
249
+ 00000252.jp2: { orderlabel: "224" }
250
+ 00000253.jp2: { label: "INDEX" }
251
+ 00000255.jp2: { orderlabel: "227" }
252
+ 00000256.jp2: { orderlabel: "228" }
253
+ 00000257.jp2: { orderlabel: "229" }
254
+ 00000258.jp2: { orderlabel: "230" }
255
+ 00000259.jp2: { orderlabel: "231" }
256
+ 00000268.jp2: { label: "BACK_COVER" }
data/spec/mets_converter/mets_parser_spec.rb ADDED
@@ -0,0 +1,33 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe MetsConverter::MetsParser do
4
+ subject(:mets) { described_class.new('spec/fixtures/35556002332765.mets.xml') }
5
+
6
+ describe '#initialize' do
7
+ it 'creates a Nokogiri XML document' do
8
+ expect(mets.document.class).to eq(Nokogiri::XML::Document)
9
+ end
10
+ end
11
+
12
+ describe '#capture_date' do
13
+ it 'extracts the capture date from the header' do
14
+ expect(mets.capture_date).to eq '2015-09-03T14:28:35-06:00'
15
+ end
16
+ end
17
+
18
+ describe '#input_file_location' do
19
+ it 'returns the input file path' do
20
+ expect(mets.input_file_location).to eq 'spec/fixtures'
21
+ end
22
+ end
23
+
24
+ describe '#pages' do
25
+ it 'creates a Nokogiri NodeSet' do
26
+ expect(mets.pages.class).to eq(Nokogiri::XML::NodeSet)
27
+ end
28
+
29
+ it 'extracts the correct number of page elements from the mets file' do
30
+ expect(mets.pages.length).to eq(268)
31
+ end
32
+ end
33
+ end
data/spec/mets_converter/yaml_builder_spec.rb ADDED
@@ -0,0 +1,28 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe MetsConverter::YamlBuilder do
4
+ subject(:mets) { MetsConverter::MetsParser.new('spec/fixtures/35556002332765.mets.xml') }
5
+ subject(:opts) { {suprascan: true, resolution: "2000", scanning_order_rtl: true, reading_order_rtl: true} }
6
+ subject(:builder) { described_class.new(mets) }
7
+ subject(:builder_with_options) { described_class.new(mets, opts) }
8
+
9
+ it { is_expected.to respond_to(:document) }
10
+ it { is_expected.to respond_to(:pages) }
11
+ it { is_expected.to respond_to(:options) }
12
+
13
+ context 'builds a yaml with default values' do
14
+ describe '#build' do
15
+ it 'outputs yaml' do
16
+ expect(builder.build).to eq(File.read('spec/fixtures/test.yml'))
17
+ end
18
+ end
19
+ end
20
+
21
+ context 'defaults can be overriden with options' do
22
+ describe '#build' do
23
+ it 'outputs yaml' do
24
+ expect(builder_with_options.build).to eq(File.read('spec/fixtures/test_with_options.yml'))
25
+ end
26
+ end
27
+ end
28
+ end
data/spec/mets_converter_spec.rb ADDED
@@ -0,0 +1,7 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe MetsConverter do
4
+ it 'has a version number' do
5
+ expect(MetsConverter::VERSION).not_to be nil
6
+ end
7
+ end
data/spec/spec_helper.rb ADDED
@@ -0,0 +1,11 @@
1
+ require 'bundler/setup'
2
+ require 'mets_converter'
3
+
4
+ RSpec.configure do |config|
5
+ # Enable flags like --only-failures and --next-failure
6
+ config.example_status_persistence_file_path = '.rspec_status'
7
+
8
+ config.expect_with :rspec do |c|
9
+ c.syntax = :expect
10
+ end
11
+ end