libis-format 1.0.5 → 2.0.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.gitignore +5 -1
- data/.travis.yml +32 -24
- data/README.md +2 -2
- data/base/Dockerfile +24 -2
- data/base/Dockerfile.alpine +20 -0
- data/base/Dockerfile.rvm +56 -0
- data/base/rework_path +20 -0
- data/docker_cfg.yml +1 -0
- data/lib/libis/format.rb +12 -3
- data/lib/libis/format/cli/convert.rb +4 -4
- data/lib/libis/format/config.rb +16 -12
- data/lib/libis/format/converter/audio_converter.rb +2 -36
- data/lib/libis/format/converter/base.rb +22 -8
- data/lib/libis/format/converter/chain.rb +3 -3
- data/lib/libis/format/converter/image_assembler.rb +82 -0
- data/lib/libis/format/converter/image_converter.rb +20 -138
- data/lib/libis/format/converter/image_splitter.rb +84 -0
- data/lib/libis/format/converter/image_watermarker.rb +261 -0
- data/lib/libis/format/converter/jp2_converter.rb +1 -1
- data/lib/libis/format/converter/office_converter.rb +2 -2
- data/lib/libis/format/converter/pdf_assembler.rb +66 -0
- data/lib/libis/format/converter/pdf_converter.rb +6 -132
- data/lib/libis/format/converter/pdf_metadata.rb +82 -0
- data/lib/libis/format/converter/pdf_optimizer.rb +67 -0
- data/lib/libis/format/converter/pdf_protecter.rb +147 -0
- data/lib/libis/format/converter/pdf_selecter.rb +83 -0
- data/lib/libis/format/converter/pdf_splitter.rb +70 -0
- data/lib/libis/format/converter/pdf_watermarker_header.rb +71 -0
- data/lib/libis/format/converter/pdf_watermarker_image.rb +76 -0
- data/lib/libis/format/converter/pdf_watermarker_text.rb +93 -0
- data/lib/libis/format/converter/spreadsheet_converter.rb +2 -2
- data/lib/libis/format/converter/video_converter.rb +1 -1
- data/lib/libis/format/identifier.rb +3 -3
- data/lib/libis/format/info.rb +27 -0
- data/lib/libis/format/library.rb +147 -0
- data/lib/libis/format/tool.rb +4 -1
- data/lib/libis/format/tool/extension_identification.rb +4 -4
- data/lib/libis/format/tool/identification_tool.rb +6 -6
- data/lib/libis/format/tool/pdf_merge.rb +3 -3
- data/lib/libis/format/tool/{pdf_copy.rb → pdf_metadata.rb} +5 -5
- data/lib/libis/format/tool/pdf_protect.rb +47 -0
- data/lib/libis/format/tool/pdf_select.rb +47 -0
- data/lib/libis/format/tool/pdf_split.rb +4 -4
- data/lib/libis/format/tool/pdf_watermark.rb +47 -0
- data/lib/libis/format/tool/spreadsheet_to_ods.rb +1 -0
- data/lib/libis/format/version.rb +1 -1
- data/lib/libis/format/yaml_loader.rb +71 -0
- data/libis-format.gemspec +3 -2
- data/tools/PdfTool.jar +0 -0
- data/tools/bcpkix-jdk15on-167.jar +0 -0
- data/tools/bcprov-jdk15on-167.jar +0 -0
- metadata +32 -13
- data/lib/libis/format/type_database.rb +0 -134
- data/lib/libis/format/type_database_impl.rb +0 -120
- data/tools/bcpkix-jdk15on-1.49.jar +0 -0
- data/tools/bcprov-jdk15on-1.49.jar +0 -0
@@ -0,0 +1,27 @@
|
|
1
|
+
module Libis
|
2
|
+
module Format
|
3
|
+
class Info
|
4
|
+
attr_reader :name, :category, :description, :puids, :mimetypes, :extensions
|
5
|
+
|
6
|
+
def initialize(name:, category:, description: '', puids: [], mimetypes: [], extensions: [])
|
7
|
+
@name = name
|
8
|
+
@category = category
|
9
|
+
@description = description
|
10
|
+
@puids = puids
|
11
|
+
@mimetypes = mimetypes
|
12
|
+
@extensions = extensions
|
13
|
+
end
|
14
|
+
|
15
|
+
def to_hash
|
16
|
+
{
|
17
|
+
name: name,
|
18
|
+
description: description.dup,
|
19
|
+
category: category,
|
20
|
+
puids: puids.dup,
|
21
|
+
mimetypes: mimetypes.dup,
|
22
|
+
extensions: extensions.dup
|
23
|
+
}
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|
@@ -0,0 +1,147 @@
|
|
1
|
+
# froze_string_litteral: true
|
2
|
+
# coding: utf-8
|
3
|
+
require 'singleton'
|
4
|
+
|
5
|
+
module Libis
|
6
|
+
module Format
|
7
|
+
|
8
|
+
class Library
|
9
|
+
include Singleton
|
10
|
+
|
11
|
+
class << self
|
12
|
+
def implementation=(impl)
|
13
|
+
instance.implementation = impl
|
14
|
+
end
|
15
|
+
|
16
|
+
def get_info(format)
|
17
|
+
instance.get_info(format)
|
18
|
+
end
|
19
|
+
|
20
|
+
def get_info_by(key, value)
|
21
|
+
instance.get_info_by(key, value)
|
22
|
+
end
|
23
|
+
|
24
|
+
def get_infos_by(key, value)
|
25
|
+
instance.get_infos_by(key, value)
|
26
|
+
end
|
27
|
+
|
28
|
+
def get_field(format, field)
|
29
|
+
instance.get_field(format, field)
|
30
|
+
end
|
31
|
+
|
32
|
+
def get_field_by(key, value, field)
|
33
|
+
instance.get_field_by(key, value, field)
|
34
|
+
end
|
35
|
+
|
36
|
+
def get_fields_by(key, value, field)
|
37
|
+
instance.get_fields_by(key, value, field)
|
38
|
+
end
|
39
|
+
|
40
|
+
def known?(key, value)
|
41
|
+
instance.known?(key, value)
|
42
|
+
end
|
43
|
+
|
44
|
+
def enrich(info, map_keys = {})
|
45
|
+
instance.enrich(info, map_keys)
|
46
|
+
end
|
47
|
+
|
48
|
+
def normalize(info, map_keys = {})
|
49
|
+
instance.normalize(info, map_keys)
|
50
|
+
end
|
51
|
+
end
|
52
|
+
|
53
|
+
def implementation=(impl)
|
54
|
+
@implementation = impl
|
55
|
+
end
|
56
|
+
|
57
|
+
def get_field(format, field)
|
58
|
+
get_field_by(:name, format, field)
|
59
|
+
end
|
60
|
+
|
61
|
+
def get_field_by(key, value, field)
|
62
|
+
info = get_info_by(key, value)
|
63
|
+
return nil unless info
|
64
|
+
case field
|
65
|
+
when :mimetype
|
66
|
+
info[:mimetypes]&.first
|
67
|
+
when :puid
|
68
|
+
info[:puids]&.first
|
69
|
+
when :extension
|
70
|
+
info[:extensions]&.first
|
71
|
+
else
|
72
|
+
info[field]
|
73
|
+
end
|
74
|
+
end
|
75
|
+
|
76
|
+
def get_fields_by(key, value, field)
|
77
|
+
get_infos_by(key, value)&.map { |info| info[field] }.compact
|
78
|
+
end
|
79
|
+
|
80
|
+
def get_info(format)
|
81
|
+
get_info_by(:name, format)
|
82
|
+
end
|
83
|
+
|
84
|
+
def get_info_by(key, value)
|
85
|
+
get_infos_by(key, value)&.first
|
86
|
+
end
|
87
|
+
|
88
|
+
def get_infos_by(key, value)
|
89
|
+
result = @implementation.query(key, value)
|
90
|
+
result.map(&:to_hash)
|
91
|
+
end
|
92
|
+
|
93
|
+
def known?(key, value)
|
94
|
+
!get_info_by(key, value).nil?
|
95
|
+
end
|
96
|
+
|
97
|
+
def enrich(info, map_keys = {})
|
98
|
+
info = normalize(info, map_keys)
|
99
|
+
mapper = Hash.new { |hash, key| hash[key] = key }.merge(map_keys)
|
100
|
+
unless (format = info[mapper[:name]]).nil?
|
101
|
+
lib_info = get_info(format)
|
102
|
+
mapper.keys.each do |key|
|
103
|
+
case key
|
104
|
+
when :mimetype
|
105
|
+
info[mapper[key]] = lib_info[:mimetypes].first if lib_info[:mimetypes].first
|
106
|
+
when :puid
|
107
|
+
info[mapper[key]] = lib_info[:puids].first if lib_info[:puids].first
|
108
|
+
when :extension
|
109
|
+
info[mapper[key]] = lib_info[:extensions].first if lib_info[:extensions].first
|
110
|
+
else
|
111
|
+
info[mapper[key]] = lib_info[key] if lib_info[key]
|
112
|
+
end
|
113
|
+
end
|
114
|
+
end
|
115
|
+
info
|
116
|
+
end
|
117
|
+
|
118
|
+
# Derive name from the available info
|
119
|
+
def normalize(info, map_keys = {})
|
120
|
+
return {} unless info.is_a? Hash
|
121
|
+
mapper = Hash.new { |hash, key| hash[key] = key }.merge(map_keys)
|
122
|
+
# fill format from looking up by puid
|
123
|
+
unless (puid = info[mapper[:puid]]).blank?
|
124
|
+
info[mapper[:name]] ||= get_field_by(:puid, puid, :name)
|
125
|
+
end
|
126
|
+
# fill format from looking up by mimetype
|
127
|
+
unless (mime = info[mapper[:mimetype]]).blank?
|
128
|
+
info[mapper[:name]] ||= get_field_by(:mimetype, mime, :name)
|
129
|
+
end
|
130
|
+
# finally complete the information from looking up by format name
|
131
|
+
unless (format = info[mapper[:name]]).nil?
|
132
|
+
info[mapper[:mimetype]] = get_field(format, :mimetype)
|
133
|
+
info[mapper[:category]] = get_field(format, :category)
|
134
|
+
end
|
135
|
+
info
|
136
|
+
end
|
137
|
+
|
138
|
+
private
|
139
|
+
|
140
|
+
def initialize
|
141
|
+
@implementation = eval(Libis::Format::Config[:format_library_implementation])
|
142
|
+
end
|
143
|
+
|
144
|
+
end
|
145
|
+
|
146
|
+
end
|
147
|
+
end
|
data/lib/libis/format/tool.rb
CHANGED
@@ -12,10 +12,13 @@ module Libis
|
|
12
12
|
autoload :OfficeToPdf, 'libis/format/tool/office_to_pdf'
|
13
13
|
autoload :FFMpeg, 'libis/format/tool/ffmpeg'
|
14
14
|
autoload :FopPdf, 'libis/format/tool/fop_pdf'
|
15
|
-
autoload :PdfCopy, 'libis/format/tool/pdf_copy'
|
16
15
|
autoload :PdfMerge, 'libis/format/tool/pdf_merge'
|
16
|
+
autoload :PdfMetadata, 'libis/format/tool/pdf_metadata'
|
17
17
|
autoload :PdfOptimizer, 'libis/format/tool/pdf_optimizer'
|
18
|
+
autoload :PdfProtect, 'libis/format/tool/pdf_protect'
|
19
|
+
autoload :PdfSelect, 'libis/format/tool/pdf_select'
|
18
20
|
autoload :PdfSplit, 'libis/format/tool/pdf_split'
|
21
|
+
autoload :PdfWatermark, 'libis/format/tool/pdf_watermark'
|
19
22
|
autoload :PdfToPdfa, 'libis/format/tool/pdf_to_pdfa'
|
20
23
|
autoload :PdfaValidator, 'libis/format/tool/pdfa_validator'
|
21
24
|
|
@@ -37,14 +37,14 @@ module Libis
|
|
37
37
|
def runner(*args)
|
38
38
|
|
39
39
|
args.map do |file|
|
40
|
-
info = ::Libis::Format::
|
40
|
+
info = ::Libis::Format::Library.get_info_by(:extension, File.extname(file))
|
41
41
|
if info
|
42
42
|
{
|
43
43
|
filepath: file,
|
44
|
-
mimetype: (info[:
|
45
|
-
puid: (info[:
|
44
|
+
mimetype: (info[:mimetypes].first rescue nil),
|
45
|
+
puid: (info[:puids].first rescue nil),
|
46
46
|
matchtype: 'extension',
|
47
|
-
tool: :
|
47
|
+
tool: :format_library
|
48
48
|
}
|
49
49
|
end
|
50
50
|
end.cleanup
|
@@ -7,7 +7,7 @@ require 'libis/tools/logger'
|
|
7
7
|
require 'libis/tools/command'
|
8
8
|
|
9
9
|
require 'libis/format/config'
|
10
|
-
require 'libis/format/
|
10
|
+
require 'libis/format/library'
|
11
11
|
|
12
12
|
module Libis
|
13
13
|
module Format
|
@@ -99,7 +99,7 @@ module Libis
|
|
99
99
|
end
|
100
100
|
|
101
101
|
# Normalize the mimetype
|
102
|
-
Libis::Format::
|
102
|
+
Libis::Format::Library.normalize(result)
|
103
103
|
|
104
104
|
# Default score is 5
|
105
105
|
result[:score] = 5
|
@@ -117,14 +117,14 @@ module Libis
|
|
117
117
|
# Signature match increases score with 2
|
118
118
|
when 'signature'
|
119
119
|
result[:score] += 2
|
120
|
-
# typeinfo = ::Libis::Format::
|
120
|
+
# typeinfo = ::Libis::Format::Library.get_info_by(:puid, result[:puid])
|
121
121
|
# ext = File.extname(result[:filename])
|
122
122
|
# result[:score] += 1 if typeinfo and typeinfo[:EXTENSIONS].include?(ext)
|
123
123
|
|
124
124
|
# Container match increases score with 4
|
125
125
|
when 'container'
|
126
126
|
result[:score] += 4
|
127
|
-
# typeinfo = ::Libis::Format::
|
127
|
+
# typeinfo = ::Libis::Format::Library.get_info_by(:puid, result[:puid])
|
128
128
|
# ext = File.extname(result[:filename])
|
129
129
|
# result[:score] += 1 if typeinfo and typeinfo[:EXTENSIONS].include?(ext)
|
130
130
|
|
@@ -152,11 +152,11 @@ module Libis
|
|
152
152
|
end
|
153
153
|
|
154
154
|
def get_mimetype(puid)
|
155
|
-
::Libis::Format::
|
155
|
+
::Libis::Format::Library.get_field_by(:puid, puid, :mimetype) rescue nil
|
156
156
|
end
|
157
157
|
|
158
158
|
def get_puid(mimetype)
|
159
|
-
::Libis::Format::
|
159
|
+
::Libis::Format::Library.get_field_by(:mimetype, mimetype, :puid) rescue nil
|
160
160
|
end
|
161
161
|
|
162
162
|
attr_accessor :bad_mimetypes, :bad_puids
|
@@ -27,9 +27,9 @@ module Libis
|
|
27
27
|
timeout = Libis::Format::Config[:timeouts][:pdf_merge]
|
28
28
|
result = Libis::Tools::Command.run(
|
29
29
|
Libis::Format::Config[:java_cmd],
|
30
|
-
'-
|
31
|
-
'
|
32
|
-
'
|
30
|
+
'-jar', Libis::Format::Config[:pdf_tool],
|
31
|
+
'merge',
|
32
|
+
'-o', target,
|
33
33
|
*options,
|
34
34
|
*source,
|
35
35
|
timeout: timeout,
|
@@ -10,7 +10,7 @@ module Libis
|
|
10
10
|
module Format
|
11
11
|
module Tool
|
12
12
|
|
13
|
-
class
|
13
|
+
class PdfMetadata
|
14
14
|
include ::Libis::Tools::Logger
|
15
15
|
|
16
16
|
def self.run(source, target, options = [])
|
@@ -26,10 +26,10 @@ module Libis
|
|
26
26
|
timeout = Libis::Format::Config[:timeouts][:pdf_copy]
|
27
27
|
result = Libis::Tools::Command.run(
|
28
28
|
Libis::Format::Config[:java_cmd],
|
29
|
-
'-
|
30
|
-
'
|
31
|
-
'
|
32
|
-
'
|
29
|
+
'-jar', Libis::Format::Config[:pdf_tool],
|
30
|
+
'metadata',
|
31
|
+
'-i', source,
|
32
|
+
'-o', target,
|
33
33
|
*options,
|
34
34
|
timeout: timeout,
|
35
35
|
kill_after: timeout * 2
|
@@ -0,0 +1,47 @@
|
|
1
|
+
require 'os'
|
2
|
+
|
3
|
+
require 'libis/tools/extend/string'
|
4
|
+
require 'libis/tools/logger'
|
5
|
+
require 'libis/tools/command'
|
6
|
+
|
7
|
+
require 'libis/format/config'
|
8
|
+
|
9
|
+
module Libis
|
10
|
+
module Format
|
11
|
+
module Tool
|
12
|
+
|
13
|
+
class PdfProtect
|
14
|
+
include ::Libis::Tools::Logger
|
15
|
+
|
16
|
+
def self.run(source, target, options = [])
|
17
|
+
self.new.run source, target, options
|
18
|
+
end
|
19
|
+
|
20
|
+
def run(source, target, options = [])
|
21
|
+
|
22
|
+
if OS.java?
|
23
|
+
# TODO: import library and execute in current VM. For now do exactly as in MRI.
|
24
|
+
end
|
25
|
+
|
26
|
+
timeout = Libis::Format::Config[:timeouts][:pdf_copy]
|
27
|
+
result = Libis::Tools::Command.run(
|
28
|
+
Libis::Format::Config[:java_cmd],
|
29
|
+
'-jar', Libis::Format::Config[:pdf_tool],
|
30
|
+
'protect',
|
31
|
+
'-i', source,
|
32
|
+
'-o', target,
|
33
|
+
*options,
|
34
|
+
timeout: timeout,
|
35
|
+
kill_after: timeout * 2
|
36
|
+
)
|
37
|
+
|
38
|
+
raise RuntimeError, "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
|
39
|
+
raise RuntimeError, "#{self.class} errors: #{result[:err].join("\n")}" unless result[:status] == 0 && result[:err].empty?
|
40
|
+
|
41
|
+
result
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
end
|
46
|
+
end
|
47
|
+
end
|
@@ -0,0 +1,47 @@
|
|
1
|
+
require 'os'
|
2
|
+
|
3
|
+
require 'libis/tools/extend/string'
|
4
|
+
require 'libis/tools/logger'
|
5
|
+
require 'libis/tools/command'
|
6
|
+
|
7
|
+
require 'libis/format/config'
|
8
|
+
|
9
|
+
module Libis
|
10
|
+
module Format
|
11
|
+
module Tool
|
12
|
+
|
13
|
+
class PdfSelect
|
14
|
+
include ::Libis::Tools::Logger
|
15
|
+
|
16
|
+
def self.run(source, target, options = [])
|
17
|
+
self.new.run source, target, options
|
18
|
+
end
|
19
|
+
|
20
|
+
def run(source, target, options = [])
|
21
|
+
|
22
|
+
if OS.java?
|
23
|
+
# TODO: import library and execute in current VM. For now do exactly as in MRI.
|
24
|
+
end
|
25
|
+
|
26
|
+
timeout = Libis::Format::Config[:timeouts][:pdf_select]
|
27
|
+
result = Libis::Tools::Command.run(
|
28
|
+
Libis::Format::Config[:java_cmd],
|
29
|
+
'-jar', Libis::Format::Config[:pdf_tool],
|
30
|
+
'select',
|
31
|
+
'-i', source,
|
32
|
+
'-o', target,
|
33
|
+
*options,
|
34
|
+
timeout: timeout,
|
35
|
+
kill_after: timeout * 2
|
36
|
+
)
|
37
|
+
|
38
|
+
raise RuntimeError, "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
|
39
|
+
raise RuntimeError, "#{self.class} errors: #{result[:err].join("\n")}" unless result[:status] == 0 && result[:err].empty?
|
40
|
+
|
41
|
+
result
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
end
|
46
|
+
end
|
47
|
+
end
|
@@ -26,10 +26,10 @@ module Libis
|
|
26
26
|
timeout = Libis::Format::Config[:timeouts][:pdf_split]
|
27
27
|
result = Libis::Tools::Command.run(
|
28
28
|
Libis::Format::Config[:java_cmd],
|
29
|
-
'-
|
30
|
-
'
|
31
|
-
'
|
32
|
-
'
|
29
|
+
'-jar', Libis::Format::Config[:pdf_tool],
|
30
|
+
'split',
|
31
|
+
'-i', source,
|
32
|
+
'-o', target,
|
33
33
|
*options,
|
34
34
|
timeout: timeout,
|
35
35
|
kill_after: timeout * 2
|
@@ -0,0 +1,47 @@
|
|
1
|
+
require 'os'
|
2
|
+
|
3
|
+
require 'libis/tools/extend/string'
|
4
|
+
require 'libis/tools/logger'
|
5
|
+
require 'libis/tools/command'
|
6
|
+
|
7
|
+
require 'libis/format/config'
|
8
|
+
|
9
|
+
module Libis
|
10
|
+
module Format
|
11
|
+
module Tool
|
12
|
+
|
13
|
+
class PdfWatermark
|
14
|
+
include ::Libis::Tools::Logger
|
15
|
+
|
16
|
+
def self.run(source, target, wm_type, options = [])
|
17
|
+
self.new.run source, target, wm_type, options
|
18
|
+
end
|
19
|
+
|
20
|
+
def run(source, target, wm_type, options = [])
|
21
|
+
|
22
|
+
if OS.java?
|
23
|
+
# TODO: import library and execute in current VM. For now do exactly as in MRI.
|
24
|
+
end
|
25
|
+
|
26
|
+
timeout = Libis::Format::Config[:timeouts][:pdf_watermark]
|
27
|
+
result = Libis::Tools::Command.run(
|
28
|
+
Libis::Format::Config[:java_cmd],
|
29
|
+
'-jar', Libis::Format::Config[:pdf_tool],
|
30
|
+
'watermark', wm_type,
|
31
|
+
'-i', source,
|
32
|
+
'-o', target,
|
33
|
+
*options,
|
34
|
+
timeout: timeout,
|
35
|
+
kill_after: timeout * 2
|
36
|
+
)
|
37
|
+
|
38
|
+
raise RuntimeError, "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
|
39
|
+
raise RuntimeError, "#{self.class} errors: #{result[:err].join("\n")}" unless result[:status] == 0 && result[:err].empty?
|
40
|
+
|
41
|
+
result
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
end
|
46
|
+
end
|
47
|
+
end
|