trace_visualization 0.0.3 → 0.0.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/Gemfile +1 -1
- data/bin/trace_visualization +21 -26
- data/lib/trace_visualization.rb +2 -2
- data/lib/trace_visualization/algorithm.rb +42 -0
- data/lib/trace_visualization/data/irepetition.rb +5 -1
- data/lib/trace_visualization/data/repetition.rb +14 -6
- data/lib/trace_visualization/data/sorted_array.rb +44 -0
- data/lib/trace_visualization/data/{lexeme.rb → token.rb} +7 -8
- data/lib/trace_visualization/lexeme_overlap_filter.rb +18 -18
- data/lib/trace_visualization/mapping.rb +72 -185
- data/lib/trace_visualization/repetitions/concatenation.rb +136 -0
- data/lib/trace_visualization/repetitions/context.rb +70 -0
- data/lib/trace_visualization/repetitions/filter.rb +153 -0
- data/lib/trace_visualization/repetitions/incrementation.rb +89 -0
- data/lib/trace_visualization/repetitions_psy.rb +12 -5
- data/lib/trace_visualization/utils.rb +5 -42
- data/lib/trace_visualization/version.rb +1 -1
- data/lib/trace_visualization/visualization/console_color_print.rb +20 -4
- data/spec/algorithm_spec.rb +69 -0
- data/spec/bwt_spec.rb +2 -5
- data/spec/data/sorted_array_spec.rb +27 -0
- data/spec/lexeme_overlap_filter_spec.rb +22 -22
- data/spec/longest_common_prefix_spec.rb +3 -8
- data/spec/mapping_spec.rb +72 -69
- data/spec/repetitions/concatenation_spec.rb +65 -0
- data/spec/repetitions/filter_spec.rb +180 -0
- data/spec/repetitions/incrementation_spec.rb +29 -0
- data/spec/repetitions_psy_spec.rb +7 -16
- data/spec/suffix_array_spec.rb +11 -31
- data/spec/utils_spec.rb +21 -9
- data/spec/visualization/console_color_print_spec.rb +26 -0
- data/trace_visualization.gemspec +2 -2
- metadata +27 -13
- data/lib/trace_visualization/repetitions_concatenation.rb +0 -134
- data/lib/trace_visualization/repetitions_context.rb +0 -18
- data/spec/repetitions_concatenation_spec.rb +0 -64
@@ -0,0 +1,136 @@
|
|
1
|
+
require 'trace_visualization'
|
2
|
+
|
3
|
+
module TraceVisualization
|
4
|
+
module Repetitions
|
5
|
+
module Concatenation
|
6
|
+
|
7
|
+
def self.process(context, options = {})
|
8
|
+
Utils.set_default_options(options, { :positions_min_size => 3, :k => 3 })
|
9
|
+
k = options[:k]
|
10
|
+
|
11
|
+
result = []
|
12
|
+
|
13
|
+
useful_cnt = 0
|
14
|
+
|
15
|
+
pairs_cnt = {}
|
16
|
+
context.repetitions_by_line.each do |item|
|
17
|
+
for i in 0 ... item.size
|
18
|
+
for j in i + 1 ... item.size
|
19
|
+
left, right = item[i][0], item[j][0]
|
20
|
+
delta = k - left.k - right.k
|
21
|
+
|
22
|
+
next if not concat_condition(left, right, delta, options[:positions_min_size])
|
23
|
+
|
24
|
+
key = (left.id << 32) + right.id
|
25
|
+
val = (pairs_cnt[key] || 0) + 1
|
26
|
+
pairs_cnt[key] = val
|
27
|
+
next if val != options[:positions_min_size]
|
28
|
+
|
29
|
+
lps, rps = process_common_positions(left, right, delta, context)
|
30
|
+
|
31
|
+
if lps.size >= options[:positions_min_size]
|
32
|
+
result << create_repetition(left, right, delta, lps, rps)
|
33
|
+
end
|
34
|
+
|
35
|
+
useful_cnt += 1
|
36
|
+
end
|
37
|
+
end
|
38
|
+
end
|
39
|
+
|
40
|
+
options[:counter] << [k, useful_cnt] if options[:counter]
|
41
|
+
|
42
|
+
process_new_repetitions(result, context)
|
43
|
+
|
44
|
+
context.repetitions.concat(result)
|
45
|
+
end
|
46
|
+
|
47
|
+
def self.process_new_repetitions(new_repetitions, context)
|
48
|
+
context.init_repetitions_by_line(new_repetitions)
|
49
|
+
end
|
50
|
+
|
51
|
+
def self.process_full_search(rs, k, context, options = {})
|
52
|
+
opts = {
|
53
|
+
:positions_min_size => 3
|
54
|
+
}.merge options
|
55
|
+
|
56
|
+
result = []
|
57
|
+
|
58
|
+
useful_cnt = 0
|
59
|
+
|
60
|
+
for left in rs
|
61
|
+
for right in rs
|
62
|
+
delta = k - left.k - right.k
|
63
|
+
next if not concat_condition(left, right, delta, options[:positions_min_size])
|
64
|
+
|
65
|
+
# @@processed_path.add(key(left, right, delta))
|
66
|
+
|
67
|
+
lps, rps = process_common_positions(left, right, delta, context)
|
68
|
+
|
69
|
+
if lps.size >= options[:positions_min_size]
|
70
|
+
result << create_repetition(left, right, delta, lps, rps)
|
71
|
+
end
|
72
|
+
|
73
|
+
useful_cnt += 1
|
74
|
+
end
|
75
|
+
end
|
76
|
+
|
77
|
+
puts "Total: #{rs.size ** 2} #{useful_cnt} #{result.size}"
|
78
|
+
|
79
|
+
rs.concat(result)
|
80
|
+
end
|
81
|
+
|
82
|
+
# Condition for potential concatenation
|
83
|
+
def self.concat_condition(left, right, delta, positions_min_size)
|
84
|
+
delta >= 0 && left.id != right.id &&
|
85
|
+
left.positions_size >= positions_min_size &&
|
86
|
+
right.positions_size >= positions_min_size
|
87
|
+
end
|
88
|
+
|
89
|
+
# *Attention* Position arrays are modified in place which can lead to side
|
90
|
+
# effects. Don't send left == right!
|
91
|
+
def self.process_common_positions(left, right, delta, context)
|
92
|
+
lr_pos = left.left_positions
|
93
|
+
lr_pos.collect! { |pos| pos + left.length + delta }
|
94
|
+
|
95
|
+
rr_pos = right.left_positions
|
96
|
+
|
97
|
+
cpr = lr_pos & rr_pos
|
98
|
+
cpl = cpr.collect { |pos| pos - left.length - delta }
|
99
|
+
|
100
|
+
idx = 0
|
101
|
+
while idx < cpr.size
|
102
|
+
xxx = context.mapping[cpl[idx] + left.length ... cpr[idx]]
|
103
|
+
xxx = xxx.join if xxx.instance_of? Array
|
104
|
+
if xxx.scan(TraceVisualization::FORBIDDEN_CHARS).size != 0
|
105
|
+
cpr.delete_at(idx)
|
106
|
+
cpl.delete_at(idx)
|
107
|
+
else
|
108
|
+
idx += 1
|
109
|
+
end
|
110
|
+
end
|
111
|
+
|
112
|
+
lr_pos.collect! { |lpos| lpos - left.length - delta }
|
113
|
+
|
114
|
+
[cpl, cpr]
|
115
|
+
end
|
116
|
+
|
117
|
+
# Create repetition
|
118
|
+
# @param left [TraceVisualization::Repetition] left parent
|
119
|
+
# @param right [TraceVisualization::Repetition] right parent
|
120
|
+
# @param delta
|
121
|
+
# @param lps
|
122
|
+
# @param rps
|
123
|
+
def self.create_repetition(left, right, delta, lps, rps)
|
124
|
+
r = left.class.new(left.length + right.length + delta, lps, rps)
|
125
|
+
|
126
|
+
r.k = left.k + right.k + delta
|
127
|
+
r.pcount = left.pcount + right.pcount
|
128
|
+
r.left = left
|
129
|
+
r.right = right
|
130
|
+
r.strict_ids = left.strict_ids + right.strict_ids
|
131
|
+
|
132
|
+
r
|
133
|
+
end
|
134
|
+
end
|
135
|
+
end
|
136
|
+
end
|
@@ -0,0 +1,70 @@
|
|
1
|
+
require 'trace_visualization/utils'
|
2
|
+
|
3
|
+
module TraceVisualization
|
4
|
+
module Repetitions
|
5
|
+
class Context
|
6
|
+
attr_accessor :mapping, :repetitions, :repetitions_by_line, :hashes
|
7
|
+
|
8
|
+
def initialize mapping
|
9
|
+
@mapping = mapping
|
10
|
+
@repetitions_by_line = Array.new(@mapping.lines.size) { [] }
|
11
|
+
@hashes = []
|
12
|
+
end
|
13
|
+
|
14
|
+
def repetitions=(repetitions)
|
15
|
+
@repetitions = repetitions
|
16
|
+
init_repetitions_by_line(repetitions)
|
17
|
+
end
|
18
|
+
|
19
|
+
def lines
|
20
|
+
@mapping.lines
|
21
|
+
end
|
22
|
+
|
23
|
+
def init_repetitions_by_line(new_repetitions)
|
24
|
+
for r in new_repetitions
|
25
|
+
r_pos = r.left_positions
|
26
|
+
r.lines.clear
|
27
|
+
i, j = 0, 0
|
28
|
+
|
29
|
+
while (i < @mapping.lines.size && j < r_pos.size)
|
30
|
+
a, b = @mapping.lines[i], (i + 1 < @mapping.lines.size ? @mapping.lines[i + 1] : 2**32)
|
31
|
+
|
32
|
+
if a <= r_pos[j] && r_pos[j] < b
|
33
|
+
@repetitions_by_line[i] << [r, r_pos[j]]
|
34
|
+
r.lines << i
|
35
|
+
|
36
|
+
j += 1
|
37
|
+
else
|
38
|
+
i += 1
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
42
|
+
|
43
|
+
@repetitions_by_line.each { |item| item.sort! { |a, b| a[1] <=> b[1] } }
|
44
|
+
end
|
45
|
+
|
46
|
+
def delete_repetition(repetition)
|
47
|
+
repetition.lines.each do |line|
|
48
|
+
@repetitions_by_line[line].delete_if { |item| item[0].id == repetition.id }
|
49
|
+
end
|
50
|
+
end
|
51
|
+
|
52
|
+
def merge_repetitions(x, y)
|
53
|
+
x.left_positions.concat(y.left_positions).uniq!.sort!
|
54
|
+
|
55
|
+
if x.right_positions
|
56
|
+
x.right_positions.concat(y.right_positions).uniq!.sort!
|
57
|
+
end
|
58
|
+
|
59
|
+
y.lines.each do |line|
|
60
|
+
for i in 0 ... @repetitions_by_line[line].size
|
61
|
+
if @repetitions_by_line[line][i][0].id == y.id
|
62
|
+
@repetitions_by_line[line][i][0] = x
|
63
|
+
x.lines << line unless x.lines.index line
|
64
|
+
end
|
65
|
+
end
|
66
|
+
end
|
67
|
+
end
|
68
|
+
end
|
69
|
+
end
|
70
|
+
end
|
@@ -0,0 +1,153 @@
|
|
1
|
+
require 'trace_visualization/utils'
|
2
|
+
require 'trace_visualization/repetitions/concatenation'
|
3
|
+
|
4
|
+
module TraceVisualization
|
5
|
+
module Repetitions
|
6
|
+
module Filter
|
7
|
+
|
8
|
+
# Filter for strict repetitions
|
9
|
+
def self.strict_repetitions_filter(str, context, options = {})
|
10
|
+
Utils.set_default_options(options, { :positions_min_size => 3 })
|
11
|
+
|
12
|
+
repetitions_filter(str, context, options)
|
13
|
+
split(str, context)
|
14
|
+
merge(context)
|
15
|
+
end
|
16
|
+
|
17
|
+
# Filter repetitions
|
18
|
+
def self.repetitions_filter(str, context, options = {})
|
19
|
+
Utils.set_default_options(options, { :positions_min_size => 3 })
|
20
|
+
|
21
|
+
context.repetitions.delete_if do |repetition|
|
22
|
+
flag = repetition.positions_size < options[:positions_min_size]
|
23
|
+
|
24
|
+
context.delete_repetition(repetition) if flag
|
25
|
+
|
26
|
+
flag
|
27
|
+
end
|
28
|
+
|
29
|
+
fix_boundaries(str, context.repetitions)
|
30
|
+
delete_duplicates(context)
|
31
|
+
end
|
32
|
+
|
33
|
+
# Filter repetitions: Boundaries
|
34
|
+
# * Change repetitions with '\n' at the beginning or end
|
35
|
+
def self.fix_boundaries(str, rs)
|
36
|
+
rs.each do |r|
|
37
|
+
if str[r.get_left_pos(0)].to_str == "\n" && r.length > 1
|
38
|
+
for i in 0 ... r.positions_size
|
39
|
+
r.set_left_pos(i, r.get_left_pos(i) + 1)
|
40
|
+
end
|
41
|
+
r.length -= 1
|
42
|
+
end
|
43
|
+
|
44
|
+
if str[r.get_left_pos(0) + r.length - 1].to_str == "\n"
|
45
|
+
r.length -= 1
|
46
|
+
end
|
47
|
+
end
|
48
|
+
end
|
49
|
+
|
50
|
+
# Filter repetitions: Split
|
51
|
+
# * Split repetitions with '\n' at the center
|
52
|
+
# * Delete duplicate after split (one repeat a subset of the other)
|
53
|
+
#
|
54
|
+
# *NB* It has meaning only for the strict repetitions, approximate
|
55
|
+
# repetitions doesn't consist '\n'
|
56
|
+
def self.split(str, context)
|
57
|
+
splitted = []
|
58
|
+
context.repetitions.delete_if do |r|
|
59
|
+
pos = nil
|
60
|
+
|
61
|
+
for i in r.get_left_pos(0) ... r.get_left_pos(0) + r.length
|
62
|
+
if str[i].to_str == "\n"
|
63
|
+
pos = i
|
64
|
+
break
|
65
|
+
end
|
66
|
+
end
|
67
|
+
|
68
|
+
if pos != nil
|
69
|
+
# right part of split
|
70
|
+
if r.length - pos - 1 >= 2
|
71
|
+
length = r.length - pos - 1
|
72
|
+
positions = (0 ... r.positions_size).collect { |i| r.get_left_pos(i) + pos + 1}
|
73
|
+
|
74
|
+
nr = r.class.new(length, positions)
|
75
|
+
|
76
|
+
splitted << nr
|
77
|
+
end
|
78
|
+
|
79
|
+
# left part of split
|
80
|
+
if pos >= 2
|
81
|
+
r.length = pos
|
82
|
+
splitted << r
|
83
|
+
end
|
84
|
+
end
|
85
|
+
|
86
|
+
context.delete_repetition(r) if pos != nil
|
87
|
+
|
88
|
+
(pos != nil)
|
89
|
+
end
|
90
|
+
|
91
|
+
splitted.delete_if do |r|
|
92
|
+
flag = false
|
93
|
+
|
94
|
+
context.repetitions.each do |x|
|
95
|
+
flag = ((x.left_positions & r.left_positions).size == r.left_positions.size && x.length >= r.length)
|
96
|
+
break if flag
|
97
|
+
end
|
98
|
+
|
99
|
+
flag
|
100
|
+
end
|
101
|
+
|
102
|
+
Concatenation.process_new_repetitions(splitted, context)
|
103
|
+
|
104
|
+
context.repetitions.concat(splitted)
|
105
|
+
end
|
106
|
+
|
107
|
+
# Filter repetitions: delete duplicate
|
108
|
+
def self.delete_duplicates(context)
|
109
|
+
i = 0
|
110
|
+
while i < context.repetitions.size
|
111
|
+
j = i + 1
|
112
|
+
while j < context.repetitions.size
|
113
|
+
if context.repetitions[i] == context.repetitions[j]
|
114
|
+
context.delete_repetition(context.repetitions[j])
|
115
|
+
context.repetitions.delete_at(j)
|
116
|
+
else
|
117
|
+
j += 1
|
118
|
+
end
|
119
|
+
end
|
120
|
+
i += 1
|
121
|
+
end
|
122
|
+
end
|
123
|
+
|
124
|
+
# Merge repetitions with common positions
|
125
|
+
def self.merge(context)
|
126
|
+
i = 0
|
127
|
+
while i < context.repetitions.size
|
128
|
+
j = 0
|
129
|
+
while j < context.repetitions.size
|
130
|
+
|
131
|
+
if i != j
|
132
|
+
x, y = context.repetitions[i], context.repetitions[j]
|
133
|
+
if x.length == y.length && x.k == y.k
|
134
|
+
common_positions = x.left_positions & y.left_positions
|
135
|
+
if common_positions.size == y.left_positions.size
|
136
|
+
context.delete_repetition(y)
|
137
|
+
context.repetitions.delete_at(j)
|
138
|
+
elsif common_positions.size > 0
|
139
|
+
context.merge_repetitions(x, y)
|
140
|
+
context.repetitions.delete_at(j)
|
141
|
+
end
|
142
|
+
end
|
143
|
+
end
|
144
|
+
|
145
|
+
j += 1
|
146
|
+
end
|
147
|
+
i += 1
|
148
|
+
end
|
149
|
+
end
|
150
|
+
|
151
|
+
end # Filter
|
152
|
+
end # Repetitions
|
153
|
+
end # TraceVisualization
|
@@ -0,0 +1,89 @@
|
|
1
|
+
require 'trace_visualization'
|
2
|
+
|
3
|
+
module TraceVisualization
|
4
|
+
module Repetitions
|
5
|
+
module Incrementation
|
6
|
+
|
7
|
+
# Process
|
8
|
+
def self.process(context, options = {})
|
9
|
+
Utils.set_default_options(options, {
|
10
|
+
:k => 1, :common_positions_size => 2
|
11
|
+
})
|
12
|
+
|
13
|
+
k = options[:k]
|
14
|
+
repetitions = context.repetitions
|
15
|
+
|
16
|
+
result = []
|
17
|
+
|
18
|
+
for core in repetitions
|
19
|
+
next if core.k != 0
|
20
|
+
|
21
|
+
# Positions for Right Incrementation
|
22
|
+
positions_ri = []
|
23
|
+
|
24
|
+
# Positions for Left Incrementation
|
25
|
+
positions_li = []
|
26
|
+
|
27
|
+
for r in repetitions
|
28
|
+
next if r == core || core.length + k != r.length
|
29
|
+
|
30
|
+
clps, rlps = core.left_positions, r.left_positions
|
31
|
+
|
32
|
+
common_positions_ri = clps & rlps
|
33
|
+
common_positions_li = clps.collect { |p| p - k } & rlps
|
34
|
+
|
35
|
+
positions_ri |= common_positions_ri if common_positions_ri.size >= options[:common_positions_size]
|
36
|
+
positions_li |= common_positions_li if common_positions_li.size >= options[:common_positions_size]
|
37
|
+
end
|
38
|
+
|
39
|
+
if positions_ri.size > 1
|
40
|
+
left_positions = positions_ri.sort
|
41
|
+
right_positions = left_positions.collect { |pos| pos + core.length + k }
|
42
|
+
|
43
|
+
hash = TraceVisualization::Utils.rhash(left_positions, right_positions)
|
44
|
+
|
45
|
+
unless context.hashes.include? hash
|
46
|
+
context.hashes << hash
|
47
|
+
result << create_repetition(core, k, left_positions, right_positions, "right")
|
48
|
+
end
|
49
|
+
end
|
50
|
+
|
51
|
+
if positions_li.size > 1
|
52
|
+
left_positions = positions_li.sort
|
53
|
+
right_positions = left_positions.collect { |pos| pos + k}
|
54
|
+
|
55
|
+
hash = TraceVisualization::Utils.rhash(left_positions, right_positions)
|
56
|
+
|
57
|
+
unless context.hashes.include? hash
|
58
|
+
context.hashes << hash
|
59
|
+
result << create_repetition(core, k, left_positions, right_positions, "left")
|
60
|
+
end
|
61
|
+
end
|
62
|
+
end
|
63
|
+
|
64
|
+
context.repetitions.concat(result)
|
65
|
+
end
|
66
|
+
|
67
|
+
# Create repetition
|
68
|
+
def self.create_repetition(core, k, left_positions, right_positions, type)
|
69
|
+
repetition = core.class.new(core.length + k, left_positions, right_positions)
|
70
|
+
repetition.k = k
|
71
|
+
|
72
|
+
fake = fake_repetition(core.class, left_positions, right_positions, type)
|
73
|
+
|
74
|
+
if type == "left"
|
75
|
+
repetition.left, repetition.right = fake, core
|
76
|
+
else
|
77
|
+
repetition.left, repetition.right = core, fake
|
78
|
+
end
|
79
|
+
|
80
|
+
repetition
|
81
|
+
end
|
82
|
+
|
83
|
+
# Create fake repetition
|
84
|
+
def self.fake_repetition(cls, left_positions, right_positions, type)
|
85
|
+
cls.new(0, type == "left" ? left_positions : right_positions)
|
86
|
+
end
|
87
|
+
end
|
88
|
+
end
|
89
|
+
end
|