id_pack 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +12 -0
- data/.rspec +2 -0
- data/.travis.yml +5 -0
- data/CODE_OF_CONDUCT.md +74 -0
- data/Gemfile +4 -0
- data/Gemfile.lock +35 -0
- data/LICENSE +21 -0
- data/README.md +39 -0
- data/Rakefile +6 -0
- data/app/assets/javascripts/lib/id-packer.js +296 -0
- data/app/assets/javascripts/lib/lz-string.js +511 -0
- data/app/assets/javascripts/lib/uuid-packer.js +372 -0
- data/bin/console +14 -0
- data/bin/setup +8 -0
- data/id_pack.gemspec +37 -0
- data/lib/id_pack/engine.rb +3 -0
- data/lib/id_pack/id_packer.rb +395 -0
- data/lib/id_pack/lz_string.rb +579 -0
- data/lib/id_pack/uuid_packer.rb +352 -0
- data/lib/id_pack/version.rb +3 -0
- data/lib/id_pack.rb +6 -0
- data/vendor/assets/javascripts/require.js +2145 -0
- metadata +108 -0
@@ -0,0 +1,352 @@
|
|
1
|
+
# Based on work by Oleksandr Korniienko
|
2
|
+
module IdPack
|
3
|
+
class UuidPacker
|
4
|
+
|
5
|
+
# calculate bits in number
|
6
|
+
def bin_pow(num)
|
7
|
+
pow = 0
|
8
|
+
pow += 1 until num >> pow == 0
|
9
|
+
pow
|
10
|
+
end
|
11
|
+
|
12
|
+
# transform string of valid characters to useful array (del = true if we need
|
13
|
+
# delimiter)
|
14
|
+
def alphanum_to_array(alphanum_string, del)
|
15
|
+
|
16
|
+
alphanum_array = []
|
17
|
+
el = alphanum_string.length
|
18
|
+
|
19
|
+
# max number of bits coding by one character (some characters will be one
|
20
|
+
# bit less)
|
21
|
+
pow = bin_pow (el - 1)
|
22
|
+
|
23
|
+
# how many characters will be one bit less
|
24
|
+
lowhi = 2**pow - el
|
25
|
+
|
26
|
+
# if delimited we can't use last characters
|
27
|
+
if del
|
28
|
+
el -= 1
|
29
|
+
pow = bin_pow(el - 1)
|
30
|
+
lowhi = 2**pow - el
|
31
|
+
lowhi = -1 if lowhi.zero?
|
32
|
+
# first element include main data about alphabet and delimiter character
|
33
|
+
alphanum_array.push [lowhi, alphanum_string[el], pow]
|
34
|
+
else
|
35
|
+
# first element include main data about alphabet
|
36
|
+
lowhi = -1 if lowhi.zero?
|
37
|
+
alphanum_array.push [lowhi, '', pow]
|
38
|
+
end
|
39
|
+
|
40
|
+
lowhi = 0 if lowhi == -1
|
41
|
+
char_item = 0
|
42
|
+
|
43
|
+
# loop by characters and get code for each one
|
44
|
+
until char_item == el
|
45
|
+
if char_item < lowhi
|
46
|
+
alphanum_array.push [char_item, alphanum_string[char_item], pow - 1]
|
47
|
+
else
|
48
|
+
alphanum_array.push [lowhi + char_item, alphanum_string[char_item], pow]
|
49
|
+
end
|
50
|
+
char_item += 1
|
51
|
+
end
|
52
|
+
alphanum_array
|
53
|
+
end
|
54
|
+
|
55
|
+
# compress UUIDs array
|
56
|
+
def alphanum_compress(arr, alphanum_string, order)
|
57
|
+
# length of UUID in bits
|
58
|
+
uuid_bit_length = 128
|
59
|
+
|
60
|
+
# compress without delta
|
61
|
+
nresult = ''
|
62
|
+
alphanum_array = alphanum_to_array alphanum_string, false
|
63
|
+
pow = alphanum_array[0][2]
|
64
|
+
lowhi = alphanum_array[0][0]
|
65
|
+
|
66
|
+
# first bit equal 0 => compress without delta
|
67
|
+
achr = 0
|
68
|
+
rest = 1
|
69
|
+
|
70
|
+
# loop by UUIDs
|
71
|
+
arr.each do |item|
|
72
|
+
|
73
|
+
# remove '-' characters from UUID
|
74
|
+
curr = item.delete('-').to_i(16)
|
75
|
+
|
76
|
+
# get base binary code (BBC)
|
77
|
+
achr += (curr << rest)
|
78
|
+
|
79
|
+
# look for number of bits in BBC
|
80
|
+
rest += uuid_bit_length
|
81
|
+
|
82
|
+
# create symbols to compressed string
|
83
|
+
until rest < pow
|
84
|
+
|
85
|
+
power_c = pow - 1
|
86
|
+
code = (
|
87
|
+
(achr & (2**power_c - 1)) + 2**power_c
|
88
|
+
).to_s(2).reverse.to_i(2) >> 1
|
89
|
+
|
90
|
+
power_c += 1 if code >= lowhi
|
91
|
+
|
92
|
+
# decrease number of bits in BBC
|
93
|
+
rest -= power_c
|
94
|
+
|
95
|
+
# get reverse bits from the end of BBC to create new symbol
|
96
|
+
code = (
|
97
|
+
(achr & (2**power_c - 1)) + 2**power_c
|
98
|
+
).to_s(2).reverse.to_i(2) >> 1
|
99
|
+
|
100
|
+
# add new symbol
|
101
|
+
nresult += alphanum_array.assoc(code)[1]
|
102
|
+
|
103
|
+
# remove used bits from BBC
|
104
|
+
achr >>= power_c
|
105
|
+
end
|
106
|
+
end
|
107
|
+
|
108
|
+
# check if we have tail of BBC
|
109
|
+
if rest > 0
|
110
|
+
code = ((achr & (2**rest - 1)) + 2**rest).to_s(2).reverse.to_i(2) >> 1
|
111
|
+
code <<= pow - rest - 1
|
112
|
+
code <<= 1 if code >= lowhi
|
113
|
+
|
114
|
+
# add tail symbol
|
115
|
+
nresult += alphanum_array.assoc(code)[1]
|
116
|
+
end
|
117
|
+
|
118
|
+
# compress with delta
|
119
|
+
arr = arr.sort
|
120
|
+
|
121
|
+
# first character is delimiter => compress with delta : delimiter (last
|
122
|
+
# character in alphabet) always has code of all ones
|
123
|
+
dresult = alphanum_array[-1][1]
|
124
|
+
alphanum_array = alphanum_to_array alphanum_string, true
|
125
|
+
pow = alphanum_array[0][2]
|
126
|
+
|
127
|
+
if pow > 1
|
128
|
+
lowhi = alphanum_array[0][0]
|
129
|
+
prev = 0
|
130
|
+
|
131
|
+
# loop by UUIDs
|
132
|
+
arr.each do |item|
|
133
|
+
|
134
|
+
# remove '-' characters from UUID
|
135
|
+
curr = item.delete('-').to_i(16)
|
136
|
+
|
137
|
+
# calculate delta
|
138
|
+
curr -= prev
|
139
|
+
prev = item.delete('-').to_i(16)
|
140
|
+
binlog = bin_pow curr
|
141
|
+
binlog = uuid_bit_length if binlog >= uuid_bit_length - pow
|
142
|
+
|
143
|
+
# get BBC for only current UUID
|
144
|
+
achr = curr
|
145
|
+
|
146
|
+
# look for number of bits in BBC (also for only current UUID)
|
147
|
+
rest = binlog
|
148
|
+
|
149
|
+
# create symbols to compressed string
|
150
|
+
until rest < pow
|
151
|
+
power_c = pow - 1
|
152
|
+
code = (
|
153
|
+
(achr & (2**power_c - 1)) +
|
154
|
+
2**power_c
|
155
|
+
).to_s(2).reverse.to_i(2) >> 1
|
156
|
+
|
157
|
+
power_c += 1 if code >= lowhi
|
158
|
+
|
159
|
+
# decrease number of bits in BBC
|
160
|
+
rest -= power_c
|
161
|
+
|
162
|
+
# get reverse bits from the end of BBC to create new symbol
|
163
|
+
code = (
|
164
|
+
(achr & (2**power_c - 1)) +
|
165
|
+
2**power_c
|
166
|
+
).to_s(2).reverse.to_i(2) >> 1
|
167
|
+
|
168
|
+
# add new symbol
|
169
|
+
dresult += alphanum_array.assoc(code)[1]
|
170
|
+
|
171
|
+
# remove used bits from BBC
|
172
|
+
achr >>= power_c
|
173
|
+
end
|
174
|
+
|
175
|
+
# check if we have tail of BBC for current UUID
|
176
|
+
if rest > 0
|
177
|
+
code = (
|
178
|
+
(achr & (2**rest - 1)) + 2**rest
|
179
|
+
).to_s(2).reverse.to_i(2) >> 1
|
180
|
+
|
181
|
+
code <<= pow - rest - 1
|
182
|
+
code <<= 1 if code >= lowhi
|
183
|
+
|
184
|
+
# add tail symbol for current UUID
|
185
|
+
dresult += alphanum_array.assoc(code)[1]
|
186
|
+
end
|
187
|
+
|
188
|
+
# add delimiter if we use less symbols than for whole UUID
|
189
|
+
dresult += alphanum_array[0][1] if binlog < uuid_bit_length
|
190
|
+
end
|
191
|
+
else
|
192
|
+
order = true
|
193
|
+
end
|
194
|
+
|
195
|
+
result = nresult
|
196
|
+
|
197
|
+
# get better result or non delta if we need to keep order
|
198
|
+
result = dresult if dresult.length < nresult.length && !order
|
199
|
+
result
|
200
|
+
end
|
201
|
+
|
202
|
+
# decompress UUIDs array
|
203
|
+
def alphanum_decompress(str, alphanum_string)
|
204
|
+
# length of UUID in bits
|
205
|
+
uuid_bit_length = 128
|
206
|
+
|
207
|
+
result = []
|
208
|
+
alphanum_array = alphanum_to_array alphanum_string, false
|
209
|
+
|
210
|
+
# check if delta used when compress
|
211
|
+
if (
|
212
|
+
alphanum_array.rassoc(str[0])[0] &
|
213
|
+
(2**(alphanum_array.rassoc(str[0])[2] - 1))
|
214
|
+
) != 0
|
215
|
+
|
216
|
+
# delta used
|
217
|
+
alphanum_array = alphanum_to_array alphanum_string, true
|
218
|
+
pow = alphanum_array[0][2]
|
219
|
+
lowhi = alphanum_array[0][0]
|
220
|
+
prev = 0
|
221
|
+
item = 1
|
222
|
+
achr = 0
|
223
|
+
rest = 0
|
224
|
+
|
225
|
+
# loop by symbols of compressed string starting from second (the first is
|
226
|
+
# header) to next after last (for BBC length processing after last)
|
227
|
+
while item <= str.length
|
228
|
+
|
229
|
+
# we catch delimiter or we get BBC with length equal whole UUID
|
230
|
+
if str[item] == alphanum_array[0][1] || rest >= uuid_bit_length
|
231
|
+
|
232
|
+
# if BBC length than we need to look to current symbol one more time
|
233
|
+
# if it is delimiter
|
234
|
+
item -= 1 if rest >= uuid_bit_length
|
235
|
+
|
236
|
+
# calculate UUID from delta
|
237
|
+
achr += prev
|
238
|
+
prev = achr
|
239
|
+
|
240
|
+
# transform UUID to hexadecimal
|
241
|
+
curr = prev.to_s(16)
|
242
|
+
|
243
|
+
# add first characters if UUID start with 0
|
244
|
+
curr = '0' * (uuid_bit_length / 4 - curr.length) + curr
|
245
|
+
|
246
|
+
# add '-' characters from UUID
|
247
|
+
curr = [
|
248
|
+
curr[0..7],
|
249
|
+
curr[8..11],
|
250
|
+
curr[12..15],
|
251
|
+
curr[16..19],
|
252
|
+
curr[20..31],
|
253
|
+
].join('-')
|
254
|
+
|
255
|
+
# add new UUID to array
|
256
|
+
result.push curr
|
257
|
+
achr = 0
|
258
|
+
rest = 0
|
259
|
+
|
260
|
+
# if we become last symbol we need no to symbol processing
|
261
|
+
elsif item < str.length
|
262
|
+
|
263
|
+
# reverse symbol code to BBC bits
|
264
|
+
code = (
|
265
|
+
alphanum_array.rassoc(str[item])[0] + 2**alphanum_array.rassoc(str[item])[2]
|
266
|
+
).to_s(2).reverse.to_i(2) >> 1
|
267
|
+
|
268
|
+
# add bits to BBC
|
269
|
+
achr += code << rest
|
270
|
+
|
271
|
+
# look for number of bits in BBC
|
272
|
+
rest += pow
|
273
|
+
rest -= 1 if code < lowhi
|
274
|
+
|
275
|
+
end
|
276
|
+
|
277
|
+
item += 1
|
278
|
+
|
279
|
+
end
|
280
|
+
|
281
|
+
else
|
282
|
+
# delta not used
|
283
|
+
achr = 0
|
284
|
+
rest = 0
|
285
|
+
pow = alphanum_array[0][2]
|
286
|
+
lowhi = alphanum_array[0][0]
|
287
|
+
|
288
|
+
# for first bit processing
|
289
|
+
frst = true
|
290
|
+
item = 0
|
291
|
+
|
292
|
+
# loop by symbols of compressed string
|
293
|
+
while item < str.length
|
294
|
+
|
295
|
+
# reverse symbol code to BBC bits
|
296
|
+
code =
|
297
|
+
(
|
298
|
+
alphanum_array.rassoc(str[item])[0] +
|
299
|
+
2**alphanum_array.rassoc(str[item])[2]
|
300
|
+
)
|
301
|
+
.to_s(2).reverse.to_i(2) >> 1
|
302
|
+
|
303
|
+
# add bits to BBC
|
304
|
+
achr += code << rest
|
305
|
+
|
306
|
+
# look for number of bits in BBC
|
307
|
+
rest += pow
|
308
|
+
rest -= 1 if alphanum_array.rassoc(str[item])[0] < lowhi
|
309
|
+
|
310
|
+
# first bit processing
|
311
|
+
if frst
|
312
|
+
frst = false
|
313
|
+
achr >>= 1
|
314
|
+
rest -= 1
|
315
|
+
end
|
316
|
+
|
317
|
+
# we get BBC with length equal whole UUID
|
318
|
+
if rest >= uuid_bit_length
|
319
|
+
|
320
|
+
# calculate number of bits in BBC
|
321
|
+
rest -= uuid_bit_length
|
322
|
+
|
323
|
+
# transform UUID to hexadecimal
|
324
|
+
curr = (achr & (2**uuid_bit_length - 1)).to_s(16)
|
325
|
+
|
326
|
+
# add first characters if UUID start with 0
|
327
|
+
curr = '0' * (uuid_bit_length / 4 - curr.length) + curr
|
328
|
+
|
329
|
+
# add '-' characters from UUID
|
330
|
+
curr = [
|
331
|
+
curr[0..7],
|
332
|
+
curr[8..11],
|
333
|
+
curr[12..15],
|
334
|
+
curr[16..19],
|
335
|
+
curr[20..31],
|
336
|
+
].join('-')
|
337
|
+
|
338
|
+
# add new UUID to array
|
339
|
+
result.push curr
|
340
|
+
|
341
|
+
# remove used bits from BBC
|
342
|
+
achr >>= uuid_bit_length
|
343
|
+
|
344
|
+
end
|
345
|
+
|
346
|
+
item += 1
|
347
|
+
end
|
348
|
+
end
|
349
|
+
result
|
350
|
+
end
|
351
|
+
end
|
352
|
+
end
|