ruby-zstds 1.0.3 → 1.1.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +133 -170
- data/ext/extconf.rb +13 -1
- data/ext/zstds_ext/buffer.c +4 -3
- data/ext/zstds_ext/buffer.h +5 -5
- data/ext/zstds_ext/dictionary.c +104 -42
- data/ext/zstds_ext/dictionary.h +1 -1
- data/ext/zstds_ext/error.c +12 -12
- data/ext/zstds_ext/error.h +2 -1
- data/ext/zstds_ext/gvl.h +24 -0
- data/ext/zstds_ext/io.c +229 -137
- data/ext/zstds_ext/option.c +44 -43
- data/ext/zstds_ext/option.h +62 -51
- data/ext/zstds_ext/stream/compressor.c +103 -50
- data/ext/zstds_ext/stream/compressor.h +4 -1
- data/ext/zstds_ext/stream/decompressor.c +46 -22
- data/ext/zstds_ext/stream/decompressor.h +4 -1
- data/ext/zstds_ext/string.c +93 -62
- data/lib/zstds/dictionary.rb +2 -0
- data/lib/zstds/file.rb +6 -2
- data/lib/zstds/option.rb +18 -7
- data/lib/zstds/stream/abstract.rb +6 -9
- data/lib/zstds/stream/raw/abstract.rb +6 -2
- data/lib/zstds/stream/raw/compressor.rb +3 -7
- data/lib/zstds/stream/raw/decompressor.rb +1 -5
- data/lib/zstds/stream/reader.rb +72 -52
- data/lib/zstds/stream/reader_helpers.rb +1 -1
- data/lib/zstds/stream/writer.rb +9 -4
- data/lib/zstds/stream/writer_helpers.rb +3 -2
- data/lib/zstds/validation.rb +4 -2
- data/lib/zstds/version.rb +1 -1
- metadata +62 -19
data/lib/zstds/file.rb
CHANGED
@@ -19,9 +19,11 @@ module ZSTDS
|
|
19
19
|
|
20
20
|
options[:pledged_size] = ::File.size source
|
21
21
|
|
22
|
-
open_files
|
22
|
+
open_files source, destination do |source_io, destination_io|
|
23
23
|
ZSTDS._native_compress_io source_io, destination_io, options
|
24
24
|
end
|
25
|
+
|
26
|
+
nil
|
25
27
|
end
|
26
28
|
|
27
29
|
def self.decompress(source, destination, options = {})
|
@@ -30,9 +32,11 @@ module ZSTDS
|
|
30
32
|
|
31
33
|
options = Option.get_decompressor_options options, BUFFER_LENGTH_NAMES
|
32
34
|
|
33
|
-
open_files
|
35
|
+
open_files source, destination do |source_io, destination_io|
|
34
36
|
ZSTDS._native_decompress_io source_io, destination_io, options
|
35
37
|
end
|
38
|
+
|
39
|
+
nil
|
36
40
|
end
|
37
41
|
|
38
42
|
private_class_method def self.open_files(source, destination, &_block)
|
data/lib/zstds/option.rb
CHANGED
@@ -12,6 +12,7 @@ module ZSTDS
|
|
12
12
|
DEFAULT_BUFFER_LENGTH = 0
|
13
13
|
|
14
14
|
COMPRESSOR_DEFAULTS = {
|
15
|
+
:gvl => false,
|
15
16
|
:compression_level => nil,
|
16
17
|
:window_log => nil,
|
17
18
|
:hash_log => nil,
|
@@ -36,6 +37,7 @@ module ZSTDS
|
|
36
37
|
.freeze
|
37
38
|
|
38
39
|
DECOMPRESSOR_DEFAULTS = {
|
40
|
+
:gvl => false,
|
39
41
|
:window_log_max => nil,
|
40
42
|
:dictionary => nil
|
41
43
|
}
|
@@ -44,11 +46,16 @@ module ZSTDS
|
|
44
46
|
def self.get_compressor_options(options, buffer_length_names)
|
45
47
|
Validation.validate_hash options
|
46
48
|
|
47
|
-
buffer_length_defaults = buffer_length_names.each_with_object({})
|
48
|
-
|
49
|
+
buffer_length_defaults = buffer_length_names.each_with_object({}) do |name, defaults|
|
50
|
+
defaults[name] = DEFAULT_BUFFER_LENGTH
|
51
|
+
end
|
52
|
+
|
53
|
+
options = COMPRESSOR_DEFAULTS.merge(buffer_length_defaults).merge options
|
49
54
|
|
50
55
|
buffer_length_names.each { |name| Validation.validate_not_negative_integer options[name] }
|
51
56
|
|
57
|
+
Validation.validate_bool options[:gvl]
|
58
|
+
|
52
59
|
compression_level = options[:compression_level]
|
53
60
|
unless compression_level.nil?
|
54
61
|
Validation.validate_integer compression_level
|
@@ -166,9 +173,8 @@ module ZSTDS
|
|
166
173
|
end
|
167
174
|
|
168
175
|
dictionary = options[:dictionary]
|
169
|
-
|
170
|
-
|
171
|
-
end
|
176
|
+
raise ValidateError, "invalid dictionary" unless
|
177
|
+
dictionary.nil? || dictionary.is_a?(Dictionary)
|
172
178
|
|
173
179
|
options
|
174
180
|
end
|
@@ -176,11 +182,16 @@ module ZSTDS
|
|
176
182
|
def self.get_decompressor_options(options, buffer_length_names)
|
177
183
|
Validation.validate_hash options
|
178
184
|
|
179
|
-
buffer_length_defaults = buffer_length_names.each_with_object({})
|
180
|
-
|
185
|
+
buffer_length_defaults = buffer_length_names.each_with_object({}) do |name, defaults|
|
186
|
+
defaults[name] = DEFAULT_BUFFER_LENGTH
|
187
|
+
end
|
188
|
+
|
189
|
+
options = DECOMPRESSOR_DEFAULTS.merge(buffer_length_defaults).merge options
|
181
190
|
|
182
191
|
buffer_length_names.each { |name| Validation.validate_not_negative_integer options[name] }
|
183
192
|
|
193
|
+
Validation.validate_bool options[:gvl]
|
194
|
+
|
184
195
|
window_log_max = options[:window_log_max]
|
185
196
|
unless window_log_max.nil?
|
186
197
|
Validation.validate_not_negative_integer window_log_max
|
@@ -12,17 +12,13 @@ module ZSTDS
|
|
12
12
|
# Native stream is not seekable by design.
|
13
13
|
# Related methods like "seek" and "pos=" can't be implemented.
|
14
14
|
|
15
|
-
# It is not possible to maintain correspondance between bytes
|
15
|
+
# It is not possible to maintain correspondance between bytes
|
16
|
+
# consumed from source and bytes written to destination by design.
|
16
17
|
# We will consume all source bytes and maintain buffer with remaining destination data.
|
17
18
|
|
18
19
|
include Delegates
|
19
20
|
|
20
|
-
attr_reader :io
|
21
|
-
attr_reader :stat
|
22
|
-
attr_reader :external_encoding
|
23
|
-
attr_reader :internal_encoding
|
24
|
-
attr_reader :transcode_options
|
25
|
-
attr_reader :pos
|
21
|
+
attr_reader :io, :stat, :external_encoding, :internal_encoding, :transcode_options, :pos
|
26
22
|
|
27
23
|
alias tell pos
|
28
24
|
|
@@ -94,8 +90,9 @@ module ZSTDS
|
|
94
90
|
end
|
95
91
|
|
96
92
|
internal_encoding = args[1]
|
97
|
-
|
98
|
-
|
93
|
+
unless internal_encoding.nil? || internal_encoding.is_a?(::Encoding)
|
94
|
+
Validation.validate_string internal_encoding
|
95
|
+
end
|
99
96
|
|
100
97
|
transcode_options = args[2]
|
101
98
|
Validation.validate_hash transcode_options unless transcode_options.nil?
|
@@ -19,11 +19,13 @@ module ZSTDS
|
|
19
19
|
|
20
20
|
def flush(&writer)
|
21
21
|
write_result(&writer)
|
22
|
+
|
23
|
+
nil
|
22
24
|
end
|
23
25
|
|
24
|
-
protected def
|
26
|
+
protected def more_destination(&writer)
|
25
27
|
result_bytesize = write_result(&writer)
|
26
|
-
raise NotEnoughDestinationError, "not enough destination" if result_bytesize
|
28
|
+
raise NotEnoughDestinationError, "not enough destination" if result_bytesize.zero?
|
27
29
|
end
|
28
30
|
|
29
31
|
protected def write_result(&_writer)
|
@@ -44,6 +46,8 @@ module ZSTDS
|
|
44
46
|
|
45
47
|
@native_stream.close
|
46
48
|
@is_closed = true
|
49
|
+
|
50
|
+
nil
|
47
51
|
end
|
48
52
|
|
49
53
|
def closed?
|
@@ -39,7 +39,7 @@ module ZSTDS
|
|
39
39
|
|
40
40
|
if need_more_destination
|
41
41
|
source = source.byteslice bytes_written, source.bytesize - bytes_written
|
42
|
-
|
42
|
+
more_destination(&writer)
|
43
43
|
next
|
44
44
|
end
|
45
45
|
|
@@ -65,7 +65,7 @@ module ZSTDS
|
|
65
65
|
need_more_destination = @native_stream.flush
|
66
66
|
|
67
67
|
if need_more_destination
|
68
|
-
|
68
|
+
more_destination(&writer)
|
69
69
|
next
|
70
70
|
end
|
71
71
|
|
@@ -73,8 +73,6 @@ module ZSTDS
|
|
73
73
|
end
|
74
74
|
|
75
75
|
super
|
76
|
-
|
77
|
-
nil
|
78
76
|
end
|
79
77
|
|
80
78
|
def close(&writer)
|
@@ -86,7 +84,7 @@ module ZSTDS
|
|
86
84
|
need_more_destination = @native_stream.finish
|
87
85
|
|
88
86
|
if need_more_destination
|
89
|
-
|
87
|
+
more_destination(&writer)
|
90
88
|
next
|
91
89
|
end
|
92
90
|
|
@@ -94,8 +92,6 @@ module ZSTDS
|
|
94
92
|
end
|
95
93
|
|
96
94
|
super
|
97
|
-
|
98
|
-
nil
|
99
95
|
end
|
100
96
|
end
|
101
97
|
end
|
@@ -34,7 +34,7 @@ module ZSTDS
|
|
34
34
|
|
35
35
|
if need_more_destination
|
36
36
|
source = source.byteslice bytes_read, source.bytesize - bytes_read
|
37
|
-
|
37
|
+
more_destination(&writer)
|
38
38
|
next
|
39
39
|
end
|
40
40
|
|
@@ -51,8 +51,6 @@ module ZSTDS
|
|
51
51
|
Validation.validate_proc writer
|
52
52
|
|
53
53
|
super
|
54
|
-
|
55
|
-
nil
|
56
54
|
end
|
57
55
|
|
58
56
|
def close(&writer)
|
@@ -61,8 +59,6 @@ module ZSTDS
|
|
61
59
|
Validation.validate_proc writer
|
62
60
|
|
63
61
|
super
|
64
|
-
|
65
|
-
nil
|
66
62
|
end
|
67
63
|
end
|
68
64
|
end
|
data/lib/zstds/stream/reader.rb
CHANGED
@@ -20,79 +20,61 @@ module ZSTDS
|
|
20
20
|
|
21
21
|
initialize_source_buffer_length
|
22
22
|
reset_io_remainder
|
23
|
+
reset_need_to_flush
|
23
24
|
|
24
25
|
@lineno = 0
|
25
26
|
end
|
26
27
|
|
28
|
+
protected def create_raw_stream
|
29
|
+
Raw::Decompressor.new @options
|
30
|
+
end
|
31
|
+
|
27
32
|
protected def initialize_source_buffer_length
|
28
33
|
source_buffer_length = @options[:source_buffer_length]
|
29
34
|
Validation.validate_not_negative_integer source_buffer_length unless source_buffer_length.nil?
|
30
35
|
|
31
|
-
source_buffer_length
|
32
|
-
|
36
|
+
if source_buffer_length.nil? || source_buffer_length.zero?
|
37
|
+
source_buffer_length = Buffer::DEFAULT_SOURCE_BUFFER_LENGTH_FOR_DECOMPRESSOR
|
38
|
+
end
|
33
39
|
|
34
40
|
@source_buffer_length = source_buffer_length
|
35
41
|
end
|
36
42
|
|
37
|
-
protected def create_raw_stream
|
38
|
-
Raw::Decompressor.new @options
|
39
|
-
end
|
40
|
-
|
41
43
|
protected def reset_io_remainder
|
42
44
|
@io_remainder = ::String.new :encoding => ::Encoding::BINARY
|
43
45
|
end
|
44
46
|
|
47
|
+
protected def reset_need_to_flush
|
48
|
+
@need_to_flush = false
|
49
|
+
end
|
50
|
+
|
45
51
|
# -- synchronous --
|
46
52
|
|
47
53
|
def read(bytes_to_read = nil, out_buffer = nil)
|
48
54
|
Validation.validate_not_negative_integer bytes_to_read unless bytes_to_read.nil?
|
49
55
|
Validation.validate_string out_buffer unless out_buffer.nil?
|
50
56
|
|
51
|
-
return ::String.new :encoding => ::Encoding::BINARY if bytes_to_read == 0
|
52
|
-
|
53
57
|
unless bytes_to_read.nil?
|
58
|
+
return ::String.new :encoding => ::Encoding::BINARY if bytes_to_read.zero?
|
54
59
|
return nil if eof?
|
55
60
|
|
56
|
-
|
61
|
+
append_io_data @io.read(@source_buffer_length) while @buffer.bytesize < bytes_to_read && !@io.eof?
|
62
|
+
flush_io_data if @buffer.bytesize < bytes_to_read
|
57
63
|
|
58
64
|
return read_bytes_from_buffer bytes_to_read, out_buffer
|
59
65
|
end
|
60
66
|
|
61
|
-
|
62
|
-
|
63
|
-
result = @buffer
|
64
|
-
reset_buffer
|
65
|
-
@pos += result.bytesize
|
66
|
-
|
67
|
-
result.force_encoding @external_encoding unless @external_encoding.nil?
|
68
|
-
result = transcode_to_internal result
|
69
|
-
result = out_buffer.replace result unless out_buffer.nil?
|
70
|
-
|
71
|
-
result
|
72
|
-
end
|
73
|
-
|
74
|
-
protected def read_more_to_buffer
|
75
|
-
io_data = @io.read @source_buffer_length
|
76
|
-
append_io_data_to_buffer io_data
|
77
|
-
end
|
78
|
-
|
79
|
-
def readpartial(bytes_to_read = nil, out_buffer = nil)
|
80
|
-
raise ::EOFError if eof?
|
81
|
-
|
82
|
-
readpartial_to_buffer until @buffer.bytesize >= bytes_to_read || @io.eof?
|
83
|
-
|
84
|
-
read_bytes_from_buffer bytes_to_read, out_buffer
|
85
|
-
end
|
67
|
+
append_io_data @io.read(@source_buffer_length) until @io.eof?
|
68
|
+
flush_io_data
|
86
69
|
|
87
|
-
|
88
|
-
io_data = @io.readpartial @source_buffer_length
|
89
|
-
append_io_data_to_buffer io_data
|
70
|
+
read_buffer out_buffer
|
90
71
|
end
|
91
72
|
|
92
73
|
def rewind
|
93
74
|
raw_wrapper :close
|
94
75
|
|
95
76
|
reset_io_remainder
|
77
|
+
reset_need_to_flush
|
96
78
|
|
97
79
|
super
|
98
80
|
end
|
@@ -103,25 +85,61 @@ module ZSTDS
|
|
103
85
|
super
|
104
86
|
end
|
105
87
|
|
88
|
+
def eof?
|
89
|
+
empty? && @io.eof?
|
90
|
+
end
|
91
|
+
|
106
92
|
# -- asynchronous --
|
107
93
|
|
94
|
+
def readpartial(bytes_to_read, out_buffer = nil)
|
95
|
+
read_more_nonblock(bytes_to_read, out_buffer) { @io.readpartial @source_buffer_length }
|
96
|
+
end
|
97
|
+
|
108
98
|
def read_nonblock(bytes_to_read, out_buffer = nil, *options)
|
109
|
-
|
99
|
+
read_more_nonblock(bytes_to_read, out_buffer) { @io.read_nonblock(@source_buffer_length, *options) }
|
100
|
+
end
|
101
|
+
|
102
|
+
protected def read_more_nonblock(bytes_to_read, out_buffer, &_block)
|
103
|
+
Validation.validate_not_negative_integer bytes_to_read
|
104
|
+
Validation.validate_string out_buffer unless out_buffer.nil?
|
105
|
+
|
106
|
+
return ::String.new :encoding => ::Encoding::BINARY if bytes_to_read.zero?
|
107
|
+
|
108
|
+
io_provided_eof_error = false
|
110
109
|
|
111
|
-
|
110
|
+
if @buffer.bytesize < bytes_to_read
|
111
|
+
begin
|
112
|
+
append_io_data yield
|
113
|
+
rescue ::EOFError
|
114
|
+
io_provided_eof_error = true
|
115
|
+
end
|
116
|
+
end
|
117
|
+
|
118
|
+
flush_io_data if @buffer.bytesize < bytes_to_read
|
119
|
+
raise ::EOFError if empty? && io_provided_eof_error
|
112
120
|
|
113
121
|
read_bytes_from_buffer bytes_to_read, out_buffer
|
114
122
|
end
|
115
123
|
|
116
|
-
|
117
|
-
|
118
|
-
|
124
|
+
# -- common --
|
125
|
+
|
126
|
+
protected def append_io_data(io_data)
|
127
|
+
io_portion = @io_remainder + io_data
|
128
|
+
bytes_read = raw_wrapper :read, io_portion
|
129
|
+
@io_remainder = io_portion.byteslice bytes_read, io_portion.bytesize - bytes_read
|
130
|
+
|
131
|
+
# Even empty io data may require flush.
|
132
|
+
@need_to_flush = true
|
119
133
|
end
|
120
134
|
|
121
|
-
|
135
|
+
protected def flush_io_data
|
136
|
+
raw_wrapper :flush
|
122
137
|
|
123
|
-
|
124
|
-
|
138
|
+
@need_to_flush = false
|
139
|
+
end
|
140
|
+
|
141
|
+
protected def empty?
|
142
|
+
!@need_to_flush && @buffer.bytesize.zero?
|
125
143
|
end
|
126
144
|
|
127
145
|
protected def read_bytes_from_buffer(bytes_to_read, out_buffer)
|
@@ -136,14 +154,16 @@ module ZSTDS
|
|
136
154
|
result
|
137
155
|
end
|
138
156
|
|
139
|
-
protected def
|
140
|
-
|
141
|
-
|
142
|
-
@
|
157
|
+
protected def read_buffer(out_buffer)
|
158
|
+
result = @buffer
|
159
|
+
reset_buffer
|
160
|
+
@pos += result.bytesize
|
143
161
|
|
144
|
-
|
145
|
-
|
146
|
-
|
162
|
+
result.force_encoding @external_encoding unless @external_encoding.nil?
|
163
|
+
result = transcode_to_internal result
|
164
|
+
|
165
|
+
result = out_buffer.replace result unless out_buffer.nil?
|
166
|
+
result
|
147
167
|
end
|
148
168
|
|
149
169
|
protected def transcode_to_internal(data)
|
data/lib/zstds/stream/writer.rb
CHANGED
@@ -64,7 +64,7 @@ module ZSTDS
|
|
64
64
|
end
|
65
65
|
|
66
66
|
protected def write_remaining_buffer
|
67
|
-
return nil if @buffer.bytesize
|
67
|
+
return nil if @buffer.bytesize.zero?
|
68
68
|
|
69
69
|
@io.write @buffer
|
70
70
|
|
@@ -77,6 +77,11 @@ module ZSTDS
|
|
77
77
|
|
78
78
|
# -- asynchronous --
|
79
79
|
|
80
|
+
# IO write nonblock can raise wait writable error.
|
81
|
+
# After resolving this error user may provide same content again.
|
82
|
+
# It is not possible to revert accepted content after error.
|
83
|
+
# So we have to accept content after processing IO write nonblock.
|
84
|
+
# It means that first write nonblock won't call IO write nonblock.
|
80
85
|
def write_nonblock(object, *options)
|
81
86
|
return 0 unless write_remaining_buffer_nonblock(*options)
|
82
87
|
|
@@ -120,14 +125,14 @@ module ZSTDS
|
|
120
125
|
end
|
121
126
|
|
122
127
|
protected def write_remaining_buffer_nonblock(*options)
|
123
|
-
return true if @buffer.bytesize
|
128
|
+
return true if @buffer.bytesize.zero?
|
124
129
|
|
125
130
|
bytes_written = @io.write_nonblock @buffer, *options
|
126
|
-
return false if bytes_written
|
131
|
+
return false if bytes_written.zero?
|
127
132
|
|
128
133
|
@buffer = @buffer.byteslice bytes_written, @buffer.bytesize - bytes_written
|
129
134
|
|
130
|
-
@buffer.bytesize
|
135
|
+
@buffer.bytesize.zero?
|
131
136
|
end
|
132
137
|
|
133
138
|
protected def raw_nonblock_wrapper(method_name, *args)
|