xar unarchiver (.xar, .pkg, .xip)
Fully implement TOC parsing
+128
-56
+2
-2
README.md
+2
-2
README.md
+126
-54
czar.cr
+126
-54
czar.cr
···
9
9
exit 1
10
10
end
11
11
12
-
CHECKSUM_ALGO = {
13
-
0 => :NONE,
14
-
1 => :SHA1,
15
-
2 => :MD5,
16
-
}
12
+
def xml_select(xml : XML::Node, node : String)
13
+
perror "error in xml_select" if xml.nil?
14
+
xml.children.select { |e| e.name == node }
15
+
end
16
+
17
+
def xml_value(xml : XML::Node, name : String)
18
+
perror "error in xml_value" if xml.nil?
19
+
xml.children.select { |e| e.name == name }.map { |e| e.content }
20
+
end
21
+
22
+
enum XARChecksumAlgo
23
+
NONE
24
+
SHA1
25
+
MD5
26
+
end
27
+
28
+
enum XARFileType
29
+
FILE
30
+
DIRECTORY
31
+
end
17
32
18
-
class XARHeader < BinaryParser
19
-
endian :big
20
-
string :magic, {count: 4}
21
-
uint16 :header_size
22
-
uint16 :version
23
-
uint64 :length_compressed
24
-
uint64 :length_uncompressed
25
-
uint32 :checksum_algo
33
+
enum XARFileEncoding
34
+
NONE
35
+
GZIP
36
+
BZIP2
26
37
end
27
38
28
39
class SliceIO < IO
···
46
57
end
47
58
end
48
59
49
-
def xml_select(xml : XML::Node, node : String)
50
-
perror "error in xml_select" if xml.nil?
51
-
xml.children.select { |e| e.name == node }
60
+
class XARHeader < BinaryParser
61
+
endian :big
62
+
string :magic, {count: 4}
63
+
uint16 :header_size
64
+
uint16 :version
65
+
uint64 :length_compressed
66
+
uint64 :length_uncompressed
67
+
uint32 :checksum_algo
52
68
end
53
69
54
-
def xml_value(xml : XML::Node, name : String)
55
-
perror "error in xml_value" if xml.nil?
56
-
xml.children.select { |e| e.name == name }.map { |e| e.content }
70
+
class XARFileData
71
+
property offset : UInt64 = 0
72
+
property size : UInt64 = 0
73
+
property length : UInt64 = 0
74
+
property checksum_extracted : String = ""
75
+
property checksum_extracted_style : XARChecksumAlgo = XARChecksumAlgo::NONE
76
+
property checksum_archived : String = ""
77
+
property checksum_archived_style : XARChecksumAlgo = XARChecksumAlgo::NONE
78
+
property encoding : XARFileEncoding = XARFileEncoding::NONE
79
+
end
80
+
81
+
class XARFileEAttrs < XARFileData
82
+
property name : String = ""
83
+
end
84
+
85
+
class XARChecksum
86
+
property style : XARChecksumAlgo = XARChecksumAlgo::NONE
87
+
property size : UInt64 = 0
88
+
property offset : UInt64 = 0
89
+
end
90
+
91
+
class XARFile
92
+
property path : String = ""
93
+
property name : String = ""
94
+
property type : XARFileType = XARFileType::FILE
95
+
property mode : Array(UInt8) = [0_u8, 0_u8, 0_u8, 0_u8]
96
+
property uid : UInt64 = 0
97
+
property gid : UInt64 = 0
98
+
property user : String = ""
99
+
property group : String = ""
100
+
property size : UInt64 = 0
101
+
property data : XARFileData = XARFileData.new
102
+
property ea : XARFileEAttrs = XARFileEAttrs.new
57
103
end
58
104
59
-
def xar_decode(entity : XML::Node, path : String = "./")
60
-
entity_name = xml_value(entity, "name")
61
-
if entity_name.size < 1
62
-
puts "entity without name: #{entity.to_s}"
63
-
return
105
+
class XAR
106
+
property checksum : XARChecksum = XARChecksum.new
107
+
property files : Array(XARFile) = [] of XARFile
108
+
end
109
+
110
+
def xar_decode_data(entity : XML::Node, data : XARFileData = XARFileData.new)
111
+
data.offset = (xml_value(entity, "offset").first rescue 0).to_u64
112
+
data.size = (xml_value(entity, "size").first rescue 0).to_u64
113
+
data.length = (xml_value(entity, "length").first rescue 0).to_u64
114
+
data.checksum_extracted = xml_value(entity, "extracted-checksum").first rescue ""
115
+
data.checksum_extracted_style = XARChecksumAlgo.parse(xml_select(entity, "extracted-checksum").first["style"]) rescue XARChecksumAlgo::NONE
116
+
data.checksum_archived = xml_value(entity, "archived-checksum").first rescue ""
117
+
data.checksum_archived_style = XARChecksumAlgo.parse(xml_select(entity, "archived-checksum").first["style"]) rescue XARChecksumAlgo::NONE
118
+
data.encoding = XARFileEncoding.parse(xml_select(entity, "encoding").first["style"].split("/x-").last) rescue XARFileEncoding::NONE
119
+
data
120
+
end
121
+
122
+
def xar_decode_ea(entity : XML::Node, ea : XARFileEAttrs = XARFileEAttrs.new)
123
+
xar_decode_data entity, ea
124
+
ea.name = xml_value(entity, "name").first rescue ""
125
+
ea
126
+
end
127
+
128
+
def xar_decode_file(entity : XML::Node, path : String = "./")
129
+
file = XARFile.new
130
+
file.path = path
131
+
file.name = xml_value(entity, "name").first rescue ""
132
+
file.type = XARFileType.parse(xml_value(entity, "type").first) rescue XARFileType::FILE
133
+
file.mode = (xml_value(entity, "mode").first rescue "0000").split("").map { |p| p.to_u8 }
134
+
file.uid = (xml_value(entity, "uid").first rescue 0).to_u64
135
+
file.gid = (xml_value(entity, "gid").first rescue 0).to_u64
136
+
file.user = xml_value(entity, "user").first rescue ""
137
+
file.group = xml_value(entity, "group").first rescue ""
138
+
file.size = (xml_value(entity, "size").first rescue 0).to_u64
139
+
140
+
data = xml_select(entity, "data")
141
+
unless data.size < 1
142
+
xar_decode_data data.first, file.data
64
143
end
65
-
entity_name = entity_name.first
66
144
67
-
entity_type = xml_value(entity, "type")
68
-
if entity_type.size < 1
69
-
puts "entity without type: #{entity.to_s}"
70
-
return
145
+
ea = xml_select(entity, "ea")
146
+
unless ea.size < 1
147
+
xar_decode_ea ea.first, file.ea
71
148
end
72
-
entity_type = entity_type.first
73
149
74
-
puts "#{path}#{entity_name} (#{entity_type})"
150
+
files = [file]
75
151
children = xml_select(entity, "file")
76
152
if children.size > 0
77
-
if entity_type != "directory"
78
-
puts "warn: found a #{entity_type} with #{children.size} children"
153
+
if file.type != XARFileType::DIRECTORY
154
+
puts "warn: found a #{file.type} with #{children.size} children"
79
155
end
80
-
81
156
children.each do |child|
82
-
xar_decode child, "#{path}#{entity_name}/"
157
+
files += xar_decode_file child, "#{path}#{file.name}/"
83
158
end
84
159
end
160
+
files
85
161
end
86
162
87
163
File.open "test.xar", "r" do |file|
···
95
171
puts "format version #{header.version}"
96
172
puts "TOC length (compressed) #{header.length_compressed}"
97
173
puts "TOC length (uncompressed) #{header.length_uncompressed}"
98
-
puts "checksum algo #{CHECKSUM_ALGO[header.checksum_algo]}"
174
+
puts "checksum algo #{XARChecksumAlgo.new(header.checksum_algo.to_i32).to_s}"
99
175
100
176
toc_data = Bytes.new header.length_uncompressed
101
177
file.seek header.header_size
···
103
179
zfile.read toc_data
104
180
end
105
181
106
-
xar = XML.parse SliceIO.new toc_data
182
+
xar_xml = XML.parse SliceIO.new toc_data
183
+
toc = xml_select(xml_select(xar_xml, "xar").first, "toc").first
107
184
108
-
toc = xml_select(xar, "xar").first
109
-
toc = xml_select(toc, "toc").first
185
+
puts "reading TOC"
186
+
xar = XAR.new
110
187
111
-
data_offset = header.header_size + header.length_compressed
112
-
data_checksum = xml_select(toc, "checksum").first
113
-
data_checksum_offset = xml_value(data_checksum, "offset").first.to_i
114
-
data_checksum_size = xml_value(data_checksum, "size").first.to_i
115
-
data_checksum_type = data_checksum["style"]
116
-
117
-
puts "TOC is checksummed as #{data_checksum_type}, #{data_checksum_size} bytes stored at offset #{data_checksum_offset}"
118
-
119
-
file.seek data_offset
120
-
121
-
file.seek data_checksum_offset, IO::Seek::Current
188
+
elem = xml_select(toc, "checksum").first
189
+
xar.checksum.style = XARChecksumAlgo.parse elem["style"]
190
+
xar.checksum.size = xml_value(elem, "size").first.to_u64
191
+
xar.checksum.offset = xml_value(elem, "offset").first.to_u64
122
192
123
-
data_files = xml_select(toc, "file")
124
-
puts "found #{data_files.size} files at the root"
193
+
puts "TOC is checksummed as #{xar.checksum.style}, #{xar.checksum.size} bytes at offset #{xar.checksum.offset}"
125
194
126
-
data_files.each do |entity|
127
-
xar_decode entity
195
+
xml_select(toc, "file").each do |entity|
196
+
xar.files += xar_decode_file entity
128
197
end
198
+
199
+
puts "#{xar.files.select { |e| e.type == XARFileType::FILE }.size} files"
200
+
puts "#{xar.files.select { |e| e.type == XARFileType::DIRECTORY }.size} directories"
129
201
end