libis-format 0.9.41 → 0.9.44
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +365 -0
- data/bin/droid +1 -1
- data/bin/fido +1 -1
- data/bin/pdf_copy +1 -1
- data/lib/libis/format/config.rb +1 -0
- data/lib/libis/format/converter/audio_converter.rb +1 -1
- data/lib/libis/format/converter/base.rb +2 -1
- data/lib/libis/format/converter/office_converter.rb +2 -2
- data/lib/libis/format/converter/pdf_converter.rb +6 -6
- data/lib/libis/format/converter/video_converter.rb +96 -2
- data/lib/libis/format/identifier.rb +12 -12
- data/lib/libis/format/tool/droid.rb +108 -0
- data/lib/libis/format/tool/extension_identification.rb +58 -0
- data/lib/libis/format/tool/ffmpeg.rb +43 -0
- data/lib/libis/format/tool/fido.rb +91 -0
- data/lib/libis/format/tool/file_tool.rb +78 -0
- data/lib/libis/format/tool/identification_tool.rb +175 -0
- data/lib/libis/format/tool/office_to_pdf.rb +54 -0
- data/lib/libis/format/tool/pdf_copy.rb +42 -0
- data/lib/libis/format/tool/pdf_merge.rb +43 -0
- data/lib/libis/format/tool/pdf_optimizer.rb +38 -0
- data/lib/libis/format/tool/pdf_split.rb +41 -0
- data/lib/libis/format/tool/pdf_to_pdfa.rb +78 -0
- data/lib/libis/format/tool/pdfa_validator.rb +63 -0
- data/lib/libis/format/tool.rb +23 -0
- data/lib/libis/format/version.rb +1 -1
- data/lib/libis/format.rb +1 -15
- data/libis-format.gemspec +1 -2
- data/spec/converter_audio_spec.rb +66 -0
- data/spec/converter_image_spec.rb +166 -0
- data/spec/converter_office_spec.rb +84 -0
- data/spec/converter_pdf_spec.rb +30 -0
- data/spec/converter_repository_spec.rb +91 -0
- data/spec/converter_video_spec.rb +97 -0
- data/spec/data/video/copyright.png +0 -0
- data/spec/identifier_spec.rb +3 -15
- metadata +32 -33
- data/lib/libis/format/droid.rb +0 -106
- data/lib/libis/format/extension_identification.rb +0 -55
- data/lib/libis/format/ffmpeg.rb +0 -41
- data/lib/libis/format/fido.rb +0 -89
- data/lib/libis/format/file_tool.rb +0 -76
- data/lib/libis/format/identification_tool.rb +0 -174
- data/lib/libis/format/office_to_pdf.rb +0 -52
- data/lib/libis/format/pdf_copy.rb +0 -40
- data/lib/libis/format/pdf_merge.rb +0 -41
- data/lib/libis/format/pdf_optimizer.rb +0 -36
- data/lib/libis/format/pdf_split.rb +0 -39
- data/lib/libis/format/pdf_to_pdfa.rb +0 -74
- data/lib/libis/format/pdfa_validator.rb +0 -61
- data/spec/converter_spec.rb +0 -433
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 22fe2305d1461bf58825e88933f215cd64fffdba
|
4
|
+
data.tar.gz: 454e2fc45afeae5b338723df845923c7f4364d10
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: ecf6f26241e61b34c6162c31ee1f55896a8d9bb67b01e2484db1714d1c5dfc75ce5666fad215dc8bc99d6e12d6f50fac3e81f92c9f47a5d8f26bdb43c0766d7b
|
7
|
+
data.tar.gz: 8743beb8af9bc17f9953e9488d0fc78ccb8e709a5fcbc551470926fea972e39560b6ab092faa52789109ff796b7309f45bba9844e395c1e2cc0f3c10d94b27b8
|
data/README.md
CHANGED
@@ -30,6 +30,371 @@ Or install it yourself as:
|
|
30
30
|
|
31
31
|
The type database is the core of the format services. It stores information about all the known formats.
|
32
32
|
|
33
|
+
### ::Libis::Format::Identifier
|
34
|
+
|
35
|
+
This class enables support for format identification. It will use 3 external tools to identify a file format:
|
36
|
+
droid, fido and the unix file tool. Based on several criteria, the results of these tools will be given a
|
37
|
+
score and the result with the highest confidence is returned. The output of all tools can also be returned.
|
38
|
+
The format identification result contains a MIME-type and a PRONOM PUID if they are known. Also the calculated
|
39
|
+
score, the tool that produced the outcome and recognition method are listed. Some tools also return more
|
40
|
+
information like the format name and format version. If these are present on the most confident output, their
|
41
|
+
values will also be part of the result info.
|
42
|
+
|
43
|
+
Unidentified files will be listed with MIME-type 'application/octet-stream' and PUID 'fmt/unknown'.
|
44
|
+
|
45
|
+
The main method #get that performs the format identification takes two parameters:
|
46
|
+
- file: can be a string representing a file or directory path, but also an array of file paths.
|
47
|
+
- options: optional hash with parameters that influence the behaviour of the format identification process
|
48
|
+
|
49
|
+
The options has accespts the following keys (Symbol types):
|
50
|
+
- :droid
|
51
|
+
Boolean if set to false will not use the Droid tool. Default: true
|
52
|
+
- :fido
|
53
|
+
Boolean if set to false will not use the Fido tool. Default: true
|
54
|
+
- :file
|
55
|
+
Boolean if set to false will not use the Unix file tool. Default: true
|
56
|
+
- :tool
|
57
|
+
Symbol either :droid, :fido or :file. A shortcut option for setting only the given value to true.
|
58
|
+
- :xml_validation
|
59
|
+
Boolean if set to false will not perform XML validation (see below)
|
60
|
+
- :recursive
|
61
|
+
Boolean if set to true and a directory is passed as first argument, Identifier will parse the
|
62
|
+
directory tree recursively and will process all files found. Default: false
|
63
|
+
- :base_dir
|
64
|
+
String. Normally file will be referenced by their absolute path in the results. If a value is
|
65
|
+
supplied for this option, the given path will be stripped from the start of the file references in the
|
66
|
+
result. Default: nil.
|
67
|
+
- :keep_output
|
68
|
+
Boolean if set will store the output of each identification tool in the result structure. Default: false
|
69
|
+
|
70
|
+
The result of the identification is a Hash with the following keys:
|
71
|
+
- :messages
|
72
|
+
a list (Array) of error and warning messages that the Identifier or one of the tools used has generated.
|
73
|
+
Each entry is another Array with the first argument the severity of the message and the second one the
|
74
|
+
message itself.
|
75
|
+
- :output
|
76
|
+
optional output of each tool as an Array with an entry per tool and grouped into a Hash by file reference.
|
77
|
+
- :formats
|
78
|
+
a Hash with file references as key and the outcome of the identification as value.
|
79
|
+
|
80
|
+
The identification outcome is a Hash of key-value pairs for each property returned. Keys are lower-case
|
81
|
+
symbols. If the tools gave results that deffer in a significant part, their outcome will be added to a list as
|
82
|
+
:alternatives (see example).
|
83
|
+
|
84
|
+
#### XML Validation
|
85
|
+
|
86
|
+
None of the tools used is able to identify an XML file by its content. In our context this is especially
|
87
|
+
annoying for EAD XML files as we want to identify them in order to process them different from the other XML
|
88
|
+
files. The Identifier solves this by performing an extra check on files identified as XML files. It will
|
89
|
+
validate each XML file against a list of XML schemas and return the matching MIME-type in the result. If the
|
90
|
+
Type Database contains an entry for this format, the result will be extended with the information from the
|
91
|
+
Type Database.
|
92
|
+
|
93
|
+
With the :xml_validation option this behaviour can be turned off, for instance in cases where no EAD files are
|
94
|
+
to be expected or no further identification of XML files is needed. Note that if no XML files are present, the
|
95
|
+
Identifier will not spend any time on XML validation anyway.
|
96
|
+
|
97
|
+
The list of validation schemas can be extended with the class method #add_xml_validation which takes two
|
98
|
+
parameters: a MIME-type and the path to an XML schema (XSD). If you want to also assign a fictuous PUID to the
|
99
|
+
XML type you should add an entry to your Type Database with the same MIME-type.
|
100
|
+
|
101
|
+
### Example
|
102
|
+
|
103
|
+
```
|
104
|
+
```
|
105
|
+
|
106
|
+
```
|
107
|
+
{
|
108
|
+
:messages => [
|
109
|
+
[0] [
|
110
|
+
[0] :debug,
|
111
|
+
[1] "XML file validated against XML Schema: [...]/data/ead.xsd"
|
112
|
+
],
|
113
|
+
[1] [
|
114
|
+
[0] :debug,
|
115
|
+
[1] "XML file validated against XML Schema: [...]/data/ead.xsd"
|
116
|
+
],
|
117
|
+
[2] [
|
118
|
+
[0] :debug,
|
119
|
+
[1] "XML file validated against XML Schema: [...]/data/ead.xsd"
|
120
|
+
]
|
121
|
+
],
|
122
|
+
:output => {
|
123
|
+
"[...]/data/Cevennes2.bmp" => [
|
124
|
+
[0] {
|
125
|
+
:matchtype => "signature",
|
126
|
+
:ext_mismatch => "false",
|
127
|
+
:puid => "fmt/116",
|
128
|
+
:mimetype => "image/bmp",
|
129
|
+
:format_name => "Windows Bitmap",
|
130
|
+
:format_version => "3.0",
|
131
|
+
:source => :droid,
|
132
|
+
:TYPE => :BMP,
|
133
|
+
:GROUP => :IMAGE,
|
134
|
+
:score => 7
|
135
|
+
},
|
136
|
+
[1] {
|
137
|
+
:puid => "fmt/116",
|
138
|
+
:format_name => "Windows Bitmap",
|
139
|
+
:format_version => "Windows Bitmap 3.0",
|
140
|
+
:mimetype => "image/bmp",
|
141
|
+
:matchtype => "signature",
|
142
|
+
:source => :fido,
|
143
|
+
:TYPE => :BMP,
|
144
|
+
:GROUP => :IMAGE,
|
145
|
+
:score => 7
|
146
|
+
},
|
147
|
+
[2] {
|
148
|
+
:mimetype => "image/bmp",
|
149
|
+
:matchtype => "magic",
|
150
|
+
:source => :file,
|
151
|
+
:TYPE => :BMP,
|
152
|
+
:GROUP => :IMAGE,
|
153
|
+
:score => 2
|
154
|
+
}
|
155
|
+
],
|
156
|
+
"[...]/data/Cevennes2.jp2" => [
|
157
|
+
[0] {
|
158
|
+
:matchtype => "signature",
|
159
|
+
:ext_mismatch => "false",
|
160
|
+
:puid => "x-fmt/392",
|
161
|
+
:mimetype => "image/jp2",
|
162
|
+
:format_name => "JP2 (JPEG 2000 part 1)",
|
163
|
+
:format_version => "",
|
164
|
+
:source => :droid,
|
165
|
+
:TYPE => :JP2,
|
166
|
+
:GROUP => :IMAGE,
|
167
|
+
:score => 7
|
168
|
+
},
|
169
|
+
[1] {
|
170
|
+
:puid => "x-fmt/392",
|
171
|
+
:format_name => "JP2 (JPEG 2000 part 1)",
|
172
|
+
:format_version => "JPEG2000",
|
173
|
+
:mimetype => "image/jp2",
|
174
|
+
:matchtype => "signature",
|
175
|
+
:source => :fido,
|
176
|
+
:TYPE => :JP2,
|
177
|
+
:GROUP => :IMAGE,
|
178
|
+
:score => 7
|
179
|
+
},
|
180
|
+
[2] {
|
181
|
+
:mimetype => "image/jp2",
|
182
|
+
:matchtype => "magic",
|
183
|
+
:source => :file,
|
184
|
+
:TYPE => :JP2,
|
185
|
+
:GROUP => :IMAGE,
|
186
|
+
:score => 2
|
187
|
+
}
|
188
|
+
],
|
189
|
+
[...]
|
190
|
+
},
|
191
|
+
:formats => {
|
192
|
+
"[...]/data/Cevennes2.bmp" => {
|
193
|
+
:matchtype => "signature",
|
194
|
+
:ext_mismatch => "false",
|
195
|
+
:puid => "fmt/116",
|
196
|
+
:mimetype => "image/bmp",
|
197
|
+
:format_name => "Windows Bitmap",
|
198
|
+
:format_version => "3.0",
|
199
|
+
:source => :droid,
|
200
|
+
:TYPE => :BMP,
|
201
|
+
:GROUP => :IMAGE,
|
202
|
+
:score => 7
|
203
|
+
},
|
204
|
+
"[...]/data/Cevennes2.jp2" => {
|
205
|
+
:matchtype => "signature",
|
206
|
+
:ext_mismatch => "false",
|
207
|
+
:puid => "x-fmt/392",
|
208
|
+
:mimetype => "image/jp2",
|
209
|
+
:format_name => "JP2 (JPEG 2000 part 1)",
|
210
|
+
:format_version => "JPEG2000",
|
211
|
+
:source => :droid,
|
212
|
+
:TYPE => :JP2,
|
213
|
+
:GROUP => :IMAGE,
|
214
|
+
:score => 7
|
215
|
+
},
|
216
|
+
"[...]/data/NikonRaw-CameraRaw.TIF" => {
|
217
|
+
:matchtype => "signature",
|
218
|
+
:ext_mismatch => "false",
|
219
|
+
:puid => "fmt/353",
|
220
|
+
:mimetype => "image/tiff",
|
221
|
+
:format_name => "Tagged Image File Format",
|
222
|
+
:format_version => "TIFF generic (little-endian)",
|
223
|
+
:source => :droid,
|
224
|
+
:TYPE => :TIFF,
|
225
|
+
:GROUP => :IMAGE,
|
226
|
+
:score => 7
|
227
|
+
},
|
228
|
+
"[...]/data/NikonRaw-CaptureOne.tif" => {
|
229
|
+
:matchtype => "signature",
|
230
|
+
:ext_mismatch => "false",
|
231
|
+
:puid => "x-fmt/387",
|
232
|
+
:mimetype => "image/tiff",
|
233
|
+
:format_name => "Exchangeable Image File Format (Uncompressed)",
|
234
|
+
:format_version => "2.2",
|
235
|
+
:source => :droid,
|
236
|
+
:TYPE => :TIFF,
|
237
|
+
:GROUP => :IMAGE,
|
238
|
+
:score => 7,
|
239
|
+
:alternatives => [
|
240
|
+
[0] {
|
241
|
+
:puid => "fmt/353",
|
242
|
+
:format_name => "Tagged Image File Format",
|
243
|
+
:format_version => "TIFF generic (little-endian)",
|
244
|
+
:mimetype => "image/tiff",
|
245
|
+
:matchtype => "signature",
|
246
|
+
:source => :fido,
|
247
|
+
:TYPE => :TIFF,
|
248
|
+
:GROUP => :IMAGE,
|
249
|
+
:score => 7
|
250
|
+
},
|
251
|
+
[1] {
|
252
|
+
:matchtype => "signature",
|
253
|
+
:ext_mismatch => "false",
|
254
|
+
:puid => "x-fmt/387",
|
255
|
+
:mimetype => "image/tiff",
|
256
|
+
:format_name => "Exchangeable Image File Format (Uncompressed)",
|
257
|
+
:format_version => "2.2",
|
258
|
+
:source => :droid,
|
259
|
+
:TYPE => :TIFF,
|
260
|
+
:GROUP => :IMAGE,
|
261
|
+
:score => 7
|
262
|
+
}
|
263
|
+
]
|
264
|
+
},
|
265
|
+
"[...]/data/test-ead.xml" => {
|
266
|
+
:matchtype => "signature",
|
267
|
+
:ext_mismatch => "false",
|
268
|
+
:puid => "fmt/101",
|
269
|
+
:mimetype => "archive/ead",
|
270
|
+
:format_name => "Encoded Archival Description (EAD)",
|
271
|
+
:format_version => "",
|
272
|
+
:source => :xsd_validation,
|
273
|
+
:TYPE => :EAD,
|
274
|
+
:GROUP => :ARCHIVE,
|
275
|
+
:score => 7,
|
276
|
+
:tool => :droid,
|
277
|
+
:match_type => "xsd_validation"
|
278
|
+
},
|
279
|
+
"[...]/data/test.doc" => {
|
280
|
+
:matchtype => "container",
|
281
|
+
:ext_mismatch => "false",
|
282
|
+
:puid => "fmt/40",
|
283
|
+
:mimetype => "application/msword",
|
284
|
+
:format_name => "Microsoft Word Document",
|
285
|
+
:format_version => "97-2003",
|
286
|
+
:source => :droid,
|
287
|
+
:TYPE => :MSDOC,
|
288
|
+
:GROUP => :TEXT,
|
289
|
+
:score => 9,
|
290
|
+
:alternatives => [
|
291
|
+
[0] {
|
292
|
+
:matchtype => "container",
|
293
|
+
:ext_mismatch => "false",
|
294
|
+
:puid => "fmt/40",
|
295
|
+
:mimetype => "application/msword",
|
296
|
+
:format_name => "Microsoft Word Document",
|
297
|
+
:format_version => "97-2003",
|
298
|
+
:source => :droid,
|
299
|
+
:TYPE => :MSDOC,
|
300
|
+
:GROUP => :TEXT,
|
301
|
+
:score => 9
|
302
|
+
},
|
303
|
+
[1] {
|
304
|
+
:puid => "fmt/111",
|
305
|
+
:format_name => "OLE2 Compound Document Format",
|
306
|
+
:format_version => "OLE2 Compound Document Format",
|
307
|
+
:mimetype => nil,
|
308
|
+
:matchtype => "signature",
|
309
|
+
:source => :fido,
|
310
|
+
:score => 3
|
311
|
+
}
|
312
|
+
]
|
313
|
+
},
|
314
|
+
"[...]/data/test.docx" => {
|
315
|
+
:matchtype => "container",
|
316
|
+
:ext_mismatch => "false",
|
317
|
+
:puid => "fmt/412",
|
318
|
+
:mimetype => "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
|
319
|
+
:format_name => "Microsoft Word for Windows",
|
320
|
+
:format_version => "2007 onwards",
|
321
|
+
:source => :droid,
|
322
|
+
:TYPE => :MSDOCX,
|
323
|
+
:GROUP => :TEXT,
|
324
|
+
:score => 9
|
325
|
+
},
|
326
|
+
"[...]/data/test.xlsx" => {
|
327
|
+
:matchtype => "container",
|
328
|
+
:ext_mismatch => "false",
|
329
|
+
:puid => "fmt/214",
|
330
|
+
:mimetype => "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
|
331
|
+
:format_name => "Microsoft Excel for Windows",
|
332
|
+
:format_version => "2007 onwards",
|
333
|
+
:source => :droid,
|
334
|
+
:TYPE => :MSXLSX,
|
335
|
+
:GROUP => :TABULAR,
|
336
|
+
:score => 9,
|
337
|
+
:alternatives => [
|
338
|
+
[0] {
|
339
|
+
:matchtype => "container",
|
340
|
+
:ext_mismatch => "false",
|
341
|
+
:puid => "fmt/214",
|
342
|
+
:mimetype => "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
|
343
|
+
:format_name => "Microsoft Excel for Windows",
|
344
|
+
:format_version => "2007 onwards",
|
345
|
+
:source => :droid,
|
346
|
+
:TYPE => :MSXLSX,
|
347
|
+
:GROUP => :TABULAR,
|
348
|
+
:score => 9
|
349
|
+
},
|
350
|
+
[1] {
|
351
|
+
:mimetype => "application/octet-stream",
|
352
|
+
:matchtype => "magic",
|
353
|
+
:source => :file,
|
354
|
+
:score => -2
|
355
|
+
}
|
356
|
+
]
|
357
|
+
},
|
358
|
+
"[...]/data/test_pdfa.pdf" => {
|
359
|
+
:matchtype => "signature",
|
360
|
+
:ext_mismatch => "false",
|
361
|
+
:puid => "fmt/354",
|
362
|
+
:mimetype => "application/pdf",
|
363
|
+
:format_name => "Acrobat PDF/A - Portable Document Format",
|
364
|
+
:format_version => "1b",
|
365
|
+
:source => :droid,
|
366
|
+
:TYPE => :PDFA,
|
367
|
+
:GROUP => :TEXT,
|
368
|
+
:score => 7,
|
369
|
+
:alternatives => [
|
370
|
+
[0] {
|
371
|
+
:puid => "fmt/19",
|
372
|
+
:format_name => "Acrobat PDF 1.5 - Portable Document Format",
|
373
|
+
:format_version => "PDF 1.5",
|
374
|
+
:mimetype => "application/pdf",
|
375
|
+
:matchtype => "signature",
|
376
|
+
:source => :fido,
|
377
|
+
:TYPE => :PDF,
|
378
|
+
:GROUP => :TEXT,
|
379
|
+
:score => 7
|
380
|
+
},
|
381
|
+
[1] {
|
382
|
+
:matchtype => "signature",
|
383
|
+
:ext_mismatch => "false",
|
384
|
+
:puid => "fmt/354",
|
385
|
+
:mimetype => "application/pdf",
|
386
|
+
:format_name => "Acrobat PDF/A - Portable Document Format",
|
387
|
+
:format_version => "1b",
|
388
|
+
:source => :droid,
|
389
|
+
:TYPE => :PDFA,
|
390
|
+
:GROUP => :TEXT,
|
391
|
+
:score => 7
|
392
|
+
}
|
393
|
+
]
|
394
|
+
}
|
395
|
+
}
|
396
|
+
}
|
397
|
+
```
|
33
398
|
## Contributing
|
34
399
|
|
35
400
|
1. Fork it ( https://github.com/Kris-LIBIS/LIBIS_Format/fork )
|
data/bin/droid
CHANGED
data/bin/fido
CHANGED
data/bin/pdf_copy
CHANGED
data/lib/libis/format/config.rb
CHANGED
@@ -22,6 +22,7 @@ module Libis
|
|
22
22
|
Config[:xml_validations] = [['archive/ead', File.join(data_dir, 'ead.xsd')]]
|
23
23
|
Config[:type_database] = File.join(data_dir, 'types.yml')
|
24
24
|
Config[:raw_audio_convert_cmd] = 'sox %s -e signed -b 16 -t wav %s rate %d channels %d'
|
25
|
+
Config[:watermark_font] = '/usr/share/fonts/truetype/liberation/LiberationSans-Regular.ttf'
|
25
26
|
|
26
27
|
end
|
27
28
|
end
|
@@ -168,7 +168,7 @@ module Libis
|
|
168
168
|
opts[:output] << '-ar' << @options[:sampling_freq] if @options[:sampling_freq]
|
169
169
|
opts[:output] << '-ac' << @options[:channels] if @options[:channels]
|
170
170
|
opts[:output] << '-f' << @options[:format] if @options[:format]
|
171
|
-
result = Libis::Format::FFMpeg.run(source, target, opts)
|
171
|
+
result = Libis::Format::Tool::FFMpeg.run(source, target, opts)
|
172
172
|
info "FFMpeg output: #{result}"
|
173
173
|
result
|
174
174
|
target
|
@@ -3,6 +3,7 @@
|
|
3
3
|
### require 'tools/string'
|
4
4
|
require 'tmpdir'
|
5
5
|
require 'libis/tools/logger'
|
6
|
+
require 'libis/tools/temp_file'
|
6
7
|
require 'libis/format/type_database'
|
7
8
|
|
8
9
|
require_relative 'repository'
|
@@ -43,7 +44,7 @@ module Libis
|
|
43
44
|
end
|
44
45
|
|
45
46
|
def Base.using_temp(target)
|
46
|
-
tempfile =
|
47
|
+
tempfile = Tools::TempFile.name("convert-#{File.basename(target, '.*').gsub(/\s/, '_')}", File.extname(target))
|
47
48
|
result = yield tempfile
|
48
49
|
return nil unless result
|
49
50
|
FileUtils.move result, target
|
@@ -2,7 +2,7 @@
|
|
2
2
|
|
3
3
|
require_relative 'base'
|
4
4
|
|
5
|
-
require 'libis/format/office_to_pdf'
|
5
|
+
require 'libis/format/tool/office_to_pdf'
|
6
6
|
require 'libis/format/type_database'
|
7
7
|
|
8
8
|
module Libis
|
@@ -38,7 +38,7 @@ module Libis
|
|
38
38
|
def convert(source, target, format, opts = {})
|
39
39
|
super
|
40
40
|
|
41
|
-
return nil unless OfficeToPdf.run(source, target)
|
41
|
+
return nil unless Format::Tool::OfficeToPdf.run(source, target)
|
42
42
|
|
43
43
|
target
|
44
44
|
|
@@ -3,9 +3,9 @@
|
|
3
3
|
require_relative 'base'
|
4
4
|
|
5
5
|
require 'libis/tools/extend/hash'
|
6
|
-
require 'libis/format/pdf_copy'
|
7
|
-
require 'libis/format/pdf_to_pdfa'
|
8
|
-
require 'libis/format/pdf_optimizer'
|
6
|
+
require 'libis/format/tool/pdf_copy'
|
7
|
+
require 'libis/format/tool/pdf_to_pdfa'
|
8
|
+
require 'libis/format/tool/pdf_optimizer'
|
9
9
|
|
10
10
|
module Libis
|
11
11
|
module Format
|
@@ -126,7 +126,7 @@ module Libis
|
|
126
126
|
def optimize_pdf(source, target, quality)
|
127
127
|
|
128
128
|
using_temp(target) do |tmpname|
|
129
|
-
result = Libis::Format::PdfOptimizer.run(source, tmpname, quality)
|
129
|
+
result = Libis::Format::Tool::PdfOptimizer.run(source, tmpname, quality)
|
130
130
|
unless result[:status] == 0
|
131
131
|
error("Pdf optimization encountered errors:\n%s", (result[:err] + result[:out]).join('\n'))
|
132
132
|
next nil
|
@@ -138,7 +138,7 @@ module Libis
|
|
138
138
|
def convert_pdf(source, target)
|
139
139
|
|
140
140
|
using_temp(target) do |tmpname|
|
141
|
-
result = Libis::Format::PdfCopy.run(
|
141
|
+
result = Libis::Format::Tool::PdfCopy.run(
|
142
142
|
source, tmpname,
|
143
143
|
@options.map { |k, v|
|
144
144
|
if v.nil?
|
@@ -159,7 +159,7 @@ module Libis
|
|
159
159
|
def pdf_to_pdfa(source, target)
|
160
160
|
|
161
161
|
using_temp(target) do |tmpname|
|
162
|
-
result = Libis::Format::PdfToPdfa.run source, tmpname
|
162
|
+
result = Libis::Format::Tool::PdfToPdfa.run source, tmpname
|
163
163
|
unless result[:status] == 0
|
164
164
|
error("Pdf/A conversion encountered errors:\n%s", result[:err].join('\n'))
|
165
165
|
next nil
|
@@ -1,5 +1,5 @@
|
|
1
1
|
require_relative 'base'
|
2
|
-
require 'libis/format/ffmpeg'
|
2
|
+
require 'libis/format/tool/ffmpeg'
|
3
3
|
|
4
4
|
require 'fileutils'
|
5
5
|
|
@@ -82,6 +82,49 @@ module Libis
|
|
82
82
|
@options[:scale] = width_x_height
|
83
83
|
end
|
84
84
|
|
85
|
+
# @param [String] file Image file path to use as watermark
|
86
|
+
def watermark_image(file)
|
87
|
+
@options[:watermark_image] = file
|
88
|
+
end
|
89
|
+
|
90
|
+
# @param [String] value text for watermark. No watermark if nil (default)
|
91
|
+
def watermark_text(value)
|
92
|
+
@options[:watermark_text] = value
|
93
|
+
end
|
94
|
+
|
95
|
+
# @param [Integer] value Font size for watermark text. Default: 10
|
96
|
+
# Note that the font is selected by the Config[:watermark_font] setting
|
97
|
+
def watermark_text_size(value)
|
98
|
+
@options[:watermark_text_size] = value.to_i
|
99
|
+
end
|
100
|
+
|
101
|
+
# @param [String] value Text color for the watermark text. Default: white
|
102
|
+
def watermark_text_color(value)
|
103
|
+
@options[:watermark_text_color] = value
|
104
|
+
end
|
105
|
+
|
106
|
+
# @param [String] value Text color for the watermark text shadow. Default: black
|
107
|
+
def watermark_text_shadow_color(value)
|
108
|
+
@options[:watermark_text_shadow_color] = value
|
109
|
+
end
|
110
|
+
|
111
|
+
# @param [Integer] value Offset of the watermark text shadow. Used for both x and y offset; default: 1
|
112
|
+
# If the offset is set to 0, no shadow will be printed
|
113
|
+
def watermark_text_shadow_offset(value)
|
114
|
+
@options[:watermark_text_offset] = value.to_i
|
115
|
+
end
|
116
|
+
|
117
|
+
# @param [String] value one of 'bottom_left' (default), 'top_left', 'bottom_right', 'top_right', 'center'
|
118
|
+
def watermark_position(value)
|
119
|
+
@options[:watermark_position] = value
|
120
|
+
end
|
121
|
+
|
122
|
+
# @param [Number] value watermark opacity (0-1) with 0 = invisible and 1 = 100% opaque. Default: 0.5
|
123
|
+
def watermark_opacity(value)
|
124
|
+
@options[:watermark_opacity] = value.to_f
|
125
|
+
end
|
126
|
+
|
127
|
+
# @param [Boolean] value If set to true automatically selects optimal format for web viewing. Default: false
|
85
128
|
def web_stream(value)
|
86
129
|
if value
|
87
130
|
@options[:video_codec] = 'h264'
|
@@ -89,14 +132,17 @@ module Libis
|
|
89
132
|
end
|
90
133
|
end
|
91
134
|
|
135
|
+
# @param [String] name name of a preset. See FFMpeg documentation for more info
|
92
136
|
def preset(name)
|
93
137
|
@options[:preset] = name
|
94
138
|
end
|
95
139
|
|
140
|
+
# @param [String] name name of an audio preset. See FFMpeg documentation for more info
|
96
141
|
def audio_preset(name)
|
97
142
|
@options[:audio_preset] = name
|
98
143
|
end
|
99
144
|
|
145
|
+
# @param [String] name name of a video preset. See FFMpeg documentation for more info
|
100
146
|
def video_preset(name)
|
101
147
|
@options[:video_preset] = name
|
102
148
|
end
|
@@ -163,6 +209,36 @@ module Libis
|
|
163
209
|
opts = {global: [], input: [], filter: [], output: []}
|
164
210
|
opts[:global] << '-hide_banner'
|
165
211
|
opts[:global] << '-loglevel' << (@options[:quiet] ? 'fatal' : 'warning')
|
212
|
+
|
213
|
+
# Watermark info
|
214
|
+
@options[:watermark_opacity] ||= 0.5
|
215
|
+
if @options[:watermark_image]
|
216
|
+
opts[:filter] << '-i' << @options[:watermark_image] << '-filter_complex'
|
217
|
+
opts[:filter] << "[1:v]format=argb,colorchannelmixer=aa=%f[wm];[0:v][wm]overlay=%s" %
|
218
|
+
[@options[:watermark_opacity], watermark_position_text]
|
219
|
+
elsif @options[:watermark_text]
|
220
|
+
@options[:watermark_text_size] ||= 10
|
221
|
+
@options[:watermark_text_color] ||= 'white'
|
222
|
+
@options[:watermark_text_shadow_color] ||= 'black'
|
223
|
+
@options[:watermark_text_shadow_offset] ||= 1
|
224
|
+
filter_text = "drawtext=text='%s':%s:fontfile=%s:fontsize=%d:fontcolor=%s@%f" %
|
225
|
+
[
|
226
|
+
@options[:watermark_text],
|
227
|
+
watermark_position_text(true),
|
228
|
+
Config[:watermark_font],
|
229
|
+
@options[:watermark_text_size],
|
230
|
+
@options[:watermark_text_color],
|
231
|
+
@options[:watermark_opacity]
|
232
|
+
]
|
233
|
+
filter_text += ':shadowcolor=%s@%f:shadowx=%d:shadowy=%d' %
|
234
|
+
[
|
235
|
+
@options[:watermark_text_shadow_color],
|
236
|
+
@options[:watermark_opacity],
|
237
|
+
@options[:watermark_text_shadow_offset],
|
238
|
+
@options[:watermark_text_shadow_offset]
|
239
|
+
] if @options[:watermark_text_shadow_offset] > 0
|
240
|
+
opts[:filter] << '-vf' << filter_text
|
241
|
+
end
|
166
242
|
opts[:output] << '-ac' << @options[:audio_channels] if @options[:audio_channels]
|
167
243
|
opts[:output] << '-c:a' << @options[:audio_codec] if @options[:audio_codec]
|
168
244
|
opts[:output] << '-c:v' << @options[:video_codec] if @options[:video_codec]
|
@@ -188,11 +264,29 @@ module Libis
|
|
188
264
|
opts[:output] << '-pre' << @options[:preset] if @options[:preset]
|
189
265
|
opts[:output] << '-apre' << @options[:audio_preset] if @options[:audio_preset]
|
190
266
|
opts[:output] << '-vpre' << @options[:video_preset] if @options[:video_preset]
|
191
|
-
|
267
|
+
info "FFMpeg options: #{opts}"
|
268
|
+
result = Libis::Format::Tool::FFMpeg.run(source, target, opts)
|
192
269
|
info "FFMpeg output: #{result}"
|
193
270
|
target
|
194
271
|
end
|
195
272
|
|
273
|
+
def watermark_position_text(for_text = false, margin = 10)
|
274
|
+
w = for_text ? 'tw' : 'w'
|
275
|
+
h = for_text ? 'th' : 'h'
|
276
|
+
case @options[:watermark_position]
|
277
|
+
when 'bottom_left'
|
278
|
+
"x=#{margin}:y=H-#{h}-#{margin}"
|
279
|
+
when 'top_left'
|
280
|
+
"x=#{margin}:y=#{margin}"
|
281
|
+
when 'bottom_right'
|
282
|
+
"x=W-#{w}-#{margin}:y=H-#{h}-#{margin}"
|
283
|
+
when 'top_right'
|
284
|
+
"x=W-#{w}-#{margin}:y=#{margin}"
|
285
|
+
else
|
286
|
+
"x=#{margin}:y=H-#{h}-#{margin}"
|
287
|
+
end
|
288
|
+
end
|
289
|
+
|
196
290
|
end
|
197
291
|
|
198
292
|
end
|