8
9
10
11
12
13
14
15
16
17
18
19
20
21
|
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
|
+
+
+
+
+
+
+
+
+
|
)
type context struct {
table [1 << 16]byte
input []byte
hash uint16
}
// The following hash code is the heart of the algorithm:
// It builds a sliding hash sum of the previous 3-and-a-bit
// characters which will be used to index the guess table.
// A better hash function would result in additional compression,
// at the expense of time.
func (ctx *context) update(val uint16) {
ctx.hash = (ctx.hash << 4) ^ val
}
type compressor func([]byte) error
func (w compressor) Write(data []byte) (int, error) {
return len(data), w(data)
}
|
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
|
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
|
-
+
|
// Guess was right - don't output
buf[0] |= 1 << uint(i)
} else {
// Guess was wrong, output char
ctx.table[ctx.hash] = current
buf = append(buf, current)
}
ctx.hash = (ctx.hash << 4) ^ uint16(current)
ctx.update(uint16(current))
}
if _, err := writer.Write(buf); err != nil {
return err
}
// Reset the flags and buffer for the next iteration
|
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
|
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
|
-
-
-
+
+
+
|
// and decompresses data according to the predictor algorithm
func Decompressor(reader io.Reader) io.Reader {
var ctx context
ctx.input = make([]byte, 0, 8)
return decompressor(func(output []byte) (int, error) {
var (
err error
flags byte
rc, available, predicted, total int
err error
flags, predicted byte
rc, total, copied int
)
// Sanity check for space to read into
if len(output) == 0 {
return 0, nil
}
|
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
|
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
|
-
+
-
-
+
-
+
-
+
-
+
-
+
-
+
-
-
+
+
-
+
|
return total, err
}
// Extend the buffer, copy the prediction header
// and calculate the number of subsequent bytes to read
ctx.input = ctx.input[:8]
flags = ctx.input[0]
predicted = int(bits.Hamming(flags))
predicted = bits.Hamming(flags)
available = 8 - predicted
if rc < int(available) && err == nil {
if rc < int(8-predicted) && err == nil {
// Retry the read if we have fewer bytes than what the prediction header indicates
var r int
r, err = reader.Read(ctx.input[predicted+rc:])
r, err = reader.Read(ctx.input[int(predicted)+rc:])
rc += r
goto retryData
} // Continue on any error, try to decompress and return it along the result
// rc now contains the amount of actual bytes in this cycle (usually 8)
rc += predicted
rc += int(predicted)
// Walk the buffer, filling in the predicted blanks,
// relocating read bytes and and updating the guess table
for i, a := 0, predicted; i < rc; i++ {
if (flags & (1 << uint(i))) > 0 {
// Guess succeeded, fill in from the table
ctx.input[i] = ctx.table[ctx.hash]
} else {
// Relocate a read byte
ctx.input[i], a = ctx.input[a], a+1
// Guess failed, update the table
ctx.table[ctx.hash] = ctx.input[i]
}
// Update the hash
ctx.hash = (ctx.hash << 4) ^ uint16(ctx.input[i])
ctx.update(uint16(ctx.input[i]))
}
// Copy the decompressed data to the output
ctx.input = ctx.input[:rc]
available = copy(output, ctx.input)
copied = copy(output, ctx.input)
total += available
total += copied
// Check for remaining bytes that dont fit in the output buffer
if available < rc {
ctx.input = ctx.input[:copy(ctx.input, ctx.input[available:])]
if copied < rc {
ctx.input = ctx.input[:copy(ctx.input, ctx.input[copied:])]
} else {
// Clear the buffer
ctx.input = ctx.input[:0]
output = output[available:]
output = output[copied:]
if len(output) > 0 && err == nil {
goto readHeader
}
}
return total, err
})
}
|