rbbt-util 2.1.0 → 3.0.2
Sign up to get free protection for your applications and to get access to all the features.
- data/bin/rbbt_query.rb +63 -0
- data/lib/rbbt-util.rb +5 -5
- data/lib/rbbt.rb +2 -11
- data/lib/rbbt/util/cmd.rb +1 -1
- data/lib/rbbt/util/fix_width_table.rb +9 -3
- data/lib/rbbt/util/log.rb +23 -7
- data/lib/rbbt/util/misc.rb +121 -15
- data/lib/rbbt/util/open.rb +14 -4
- data/lib/rbbt/util/persistence.rb +52 -21
- data/lib/rbbt/util/rake.rb +108 -21
- data/lib/rbbt/util/resource.rb +338 -0
- data/lib/rbbt/util/simpleDSL.rb +1 -1
- data/lib/rbbt/util/simpleopt.rb +1 -1
- data/lib/rbbt/util/task.rb +340 -0
- data/lib/rbbt/util/tc_hash.rb +19 -2
- data/lib/rbbt/util/tsv.rb +15 -10
- data/lib/rbbt/util/tsv/accessor.rb +16 -7
- data/lib/rbbt/util/tsv/attach.rb +220 -17
- data/lib/rbbt/util/tsv/index.rb +6 -1
- data/lib/rbbt/util/tsv/manipulate.rb +4 -5
- data/lib/rbbt/util/tsv/parse.rb +45 -21
- data/lib/rbbt/util/tsv/resource.rb +74 -0
- data/lib/rbbt/util/workflow.rb +99 -75
- data/test/rbbt/util/test_filecache.rb +2 -2
- data/test/rbbt/util/test_misc.rb +7 -2
- data/test/rbbt/util/test_persistence.rb +40 -5
- data/test/rbbt/util/test_resource.rb +92 -0
- data/test/rbbt/util/test_task.rb +118 -0
- data/test/rbbt/util/test_tsv.rb +5 -1
- data/test/rbbt/util/test_workflow.rb +77 -62
- data/test/rbbt/util/tsv/test_attach.rb +95 -7
- data/test/rbbt/util/tsv/test_index.rb +0 -1
- data/test/rbbt/util/tsv/test_manipulate.rb +20 -0
- data/test/rbbt/util/tsv/test_resource.rb +9 -0
- data/test/test_helper.rb +10 -0
- data/test/test_rbbt.rb +2 -37
- metadata +16 -18
- data/lib/rbbt/util/data_module.rb +0 -93
- data/lib/rbbt/util/path.rb +0 -155
- data/lib/rbbt/util/pkg_config.rb +0 -78
- data/lib/rbbt/util/pkg_data.rb +0 -119
- data/lib/rbbt/util/pkg_software.rb +0 -145
- data/test/rbbt/util/test_data_module.rb +0 -50
- data/test/rbbt/util/test_path.rb +0 -10
- data/test/rbbt/util/test_pkg_data.rb +0 -129
- data/test/test_pkg.rb +0 -28
@@ -1,3 +1,4 @@
|
|
1
|
+
require 'rbbt/util/resource'
|
1
2
|
require 'rbbt/util/misc'
|
2
3
|
|
3
4
|
class TSV
|
@@ -68,12 +69,12 @@ class TSV
|
|
68
69
|
when (TSV === identifiers.first or identifiers.empty?)
|
69
70
|
identifiers
|
70
71
|
when
|
71
|
-
identifiers.collect{|f| Path.path(f,
|
72
|
+
identifiers.collect{|f| Resource::Path.path(f, nil, namespace)}
|
72
73
|
end
|
73
74
|
when (identifiers and not Array === identifiers)
|
74
|
-
[Path.path(identifiers,
|
75
|
+
[Resource::Path.path(identifiers, nil, namespace)]
|
75
76
|
when filename
|
76
|
-
Path.path(filename,
|
77
|
+
Resource::Path.path(filename, nil, namespace).identifier_files
|
77
78
|
else
|
78
79
|
[]
|
79
80
|
end
|
@@ -176,7 +177,10 @@ class TSV
|
|
176
177
|
if String === value && value =~ /__Ref:(.*)/
|
177
178
|
return self[$1]
|
178
179
|
else
|
179
|
-
|
180
|
+
|
181
|
+
if Array === value and fields
|
182
|
+
value = NamedArray.name value, fields
|
183
|
+
end
|
180
184
|
value
|
181
185
|
end
|
182
186
|
end
|
@@ -245,13 +249,18 @@ class TSV
|
|
245
249
|
end
|
246
250
|
|
247
251
|
def include?(key)
|
248
|
-
data.include? key
|
252
|
+
@data.include? key
|
249
253
|
end
|
250
254
|
|
251
|
-
def to_s(keys = nil)
|
255
|
+
def to_s(keys = nil, no_options = false)
|
256
|
+
if FalseClass === keys or TrueClass === keys
|
257
|
+
no_options = keys
|
258
|
+
keys = nil
|
259
|
+
end
|
260
|
+
|
252
261
|
str = ""
|
253
262
|
|
254
|
-
str << "#: " << Misc.hash2string(EXTRA_ACCESSORS.collect{|key| [key, self.send(key)]}) << "\n"
|
263
|
+
str << "#: " << Misc.hash2string(EXTRA_ACCESSORS.collect{|key| [key, self.send(key)]}) << "\n" unless no_options
|
255
264
|
if fields
|
256
265
|
str << "#" << key_field << "\t" << fields * "\t" << "\n"
|
257
266
|
end
|
data/lib/rbbt/util/tsv/attach.rb
CHANGED
@@ -1,5 +1,182 @@
|
|
1
1
|
class TSV
|
2
|
+
def self.paste_merge(file1, file2, output, sep = "\t")
|
3
|
+
case
|
4
|
+
when (String === file1 and not file1.index("\n") and file1.length < 250 and File.exists?(file1))
|
5
|
+
file1 = CMD.cmd("sort -k1,1 -t'#{sep}' #{ file1 } | grep -v '^#{sep}' ", :pipe => true)
|
6
|
+
when (String === file1 or StringIO === file1)
|
7
|
+
file1 = CMD.cmd("sort -k1,1 -t'#{sep}' | grep -v '^#{sep}'", :in => file1, :pipe => true)
|
8
|
+
end
|
9
|
+
|
10
|
+
case
|
11
|
+
when (String === file2 and not file2.index("\n") and file2.length < 250 and File.exists?(file2))
|
12
|
+
file2 = CMD.cmd("sort -k1,1 -t'#{sep}' #{ file2 } | grep -v '^#{sep}' ", :pipe => true)
|
13
|
+
when (String === file2 or StringIO === file2)
|
14
|
+
file2 = CMD.cmd("sort -k1,1 -t'#{sep}' | grep -v '^#{sep}'", :in => file2, :pipe => true)
|
15
|
+
end
|
16
|
+
|
17
|
+
output = File.open(output, 'w') if String === output
|
18
|
+
|
19
|
+
cols1 = nil
|
20
|
+
cols2 = nil
|
21
|
+
|
22
|
+
done1 = false
|
23
|
+
done2 = false
|
24
|
+
|
25
|
+
key1 = key2 = nil
|
26
|
+
while key1.nil?
|
27
|
+
while (line1 = file1.gets) =~ /#/; end
|
28
|
+
key1, *parts1 = line1.sub("\n",'').split(sep, -1)
|
29
|
+
cols1 = parts1.length
|
30
|
+
end
|
31
|
+
|
32
|
+
while key2.nil?
|
33
|
+
while (line2 = file2.gets) =~ /#/; end
|
34
|
+
key2, *parts2 = line2.sub("\n",'').split(sep, -1)
|
35
|
+
cols2 = parts2.length
|
36
|
+
end
|
37
|
+
|
38
|
+
key = key1 < key2 ? key1 : key2
|
39
|
+
parts = [""] * (cols1 + cols2)
|
40
|
+
while not (done1 and done2)
|
41
|
+
while (not done1 and key1 == key)
|
42
|
+
parts1.each_with_index do |part, i|
|
43
|
+
parts[i] = (parts[i].nil? or parts[i].empty?) ? part : parts[i] << "|" << part
|
44
|
+
end
|
45
|
+
key1 = nil
|
46
|
+
while key1.nil? and not done1
|
47
|
+
if file1.eof?; done1 = true; else key1, *parts1 = file1.gets.sub("\n",'').split(sep, -1) end
|
48
|
+
end
|
49
|
+
end
|
50
|
+
while (not done2 and key2 == key)
|
51
|
+
parts2.each_with_index do |part, i|
|
52
|
+
i += cols1
|
53
|
+
parts[i] = (parts[i].nil? or parts[i].empty?) ? part : parts[i] << "|" << part
|
54
|
+
end
|
55
|
+
key2 = nil
|
56
|
+
while key2.nil? and not done2
|
57
|
+
if file2.eof?; done2 = true; else key2, *parts2 = file2.gets.sub("\n",'').split(sep, -1) end
|
58
|
+
end
|
59
|
+
end
|
60
|
+
|
61
|
+
output.puts [key, parts].flatten * sep
|
62
|
+
parts = [""] * (cols1 + cols2)
|
63
|
+
|
64
|
+
case
|
65
|
+
when done1
|
66
|
+
key = key2
|
67
|
+
when done2
|
68
|
+
key = key1
|
69
|
+
else
|
70
|
+
key = key1 < key2 ? key1 : key2
|
71
|
+
end
|
72
|
+
end
|
73
|
+
|
74
|
+
output.close
|
75
|
+
end
|
76
|
+
|
77
|
+
def self.paste(file1, file2, output, sep = "\t")
|
78
|
+
case
|
79
|
+
when (String === file1 and not file1.index("\n") and file1.length < 250 and File.exists?(file1))
|
80
|
+
file1 = CMD.cmd("sort -k1,1 -t'#{sep}' #{ file1 } ", :pipe => true)
|
81
|
+
when String === file1
|
82
|
+
file1 = CMD.cmd("sort -k1,1 -t'#{sep}'", :in => file1, :pipe => true)
|
83
|
+
end
|
84
|
+
|
85
|
+
case
|
86
|
+
when (String === file2 and not file2.index("\n") and file2.length < 250 and File.exists?(file2))
|
87
|
+
file2 = CMD.cmd("sort -k1,1 -t'#{sep}' #{ file2 } ", :pipe => true)
|
88
|
+
when String === file2
|
89
|
+
file2 = CMD.cmd("sort -k1,1 -t'#{sep}'", :in => file2, :pipe => true)
|
90
|
+
end
|
91
|
+
|
92
|
+
output = File.open(output, 'w') if String === output
|
93
|
+
|
94
|
+
cols1 = nil
|
95
|
+
cols2 = nil
|
96
|
+
|
97
|
+
done1 = false
|
98
|
+
done2 = false
|
99
|
+
|
100
|
+
while (line1 = file1.gets) =~ /#/; end
|
101
|
+
line1.strip!
|
102
|
+
parts1 = line1.split(sep)
|
103
|
+
key1 = parts1.shift
|
104
|
+
cols1 = parts1.length
|
2
105
|
|
106
|
+
while (line2 = file2.gets) =~ /#/; end
|
107
|
+
line2.strip!
|
108
|
+
parts2 = line2.split(sep)
|
109
|
+
key2 = parts2.shift
|
110
|
+
cols2 = parts2.length
|
111
|
+
while not (done1 or done2)
|
112
|
+
case
|
113
|
+
when key1 < key2
|
114
|
+
output.puts [key1, parts1, [""] * cols2] * sep
|
115
|
+
if file1.eof?
|
116
|
+
done1 = true
|
117
|
+
else
|
118
|
+
line1 = file1.gets
|
119
|
+
line1.strip!
|
120
|
+
parts1 = line1.split(sep)
|
121
|
+
key1 = parts1.shift
|
122
|
+
end
|
123
|
+
when key2 < key1
|
124
|
+
output.puts [key2, [""] * cols1, parts2] * sep
|
125
|
+
if file2.eof?
|
126
|
+
done2 = true
|
127
|
+
else
|
128
|
+
line2 = file2.gets
|
129
|
+
line2.strip!
|
130
|
+
parts2 = line2.split(sep)
|
131
|
+
key2 = parts2.shift
|
132
|
+
end
|
133
|
+
when key1 == key2
|
134
|
+
output.puts [key1, parts1, parts2] * sep
|
135
|
+
if file1.eof?
|
136
|
+
done1 = true
|
137
|
+
else
|
138
|
+
line1 = file1.gets
|
139
|
+
line1.strip!
|
140
|
+
parts1 = line1.split(sep)
|
141
|
+
key1 = parts1.shift
|
142
|
+
end
|
143
|
+
if file2.eof?
|
144
|
+
done2 = true
|
145
|
+
else
|
146
|
+
line2 = file2.gets
|
147
|
+
line2.strip!
|
148
|
+
parts2 = line2.split(sep)
|
149
|
+
key2 = parts2.shift
|
150
|
+
end
|
151
|
+
end
|
152
|
+
end
|
153
|
+
|
154
|
+
while not done1
|
155
|
+
output.puts [key1, parts1, [""] * cols2] * sep
|
156
|
+
if file1.eof?
|
157
|
+
done1 = true
|
158
|
+
else
|
159
|
+
line1 = file1.gets
|
160
|
+
line1.strip!
|
161
|
+
parts1 = line1.split(sep)
|
162
|
+
key1 = parts1.shift
|
163
|
+
end
|
164
|
+
end
|
165
|
+
|
166
|
+
while not done2
|
167
|
+
output.puts [key2, [""] * cols1, parts2] * sep
|
168
|
+
if file2.eof?
|
169
|
+
done2 = true
|
170
|
+
else
|
171
|
+
line2 = file2.gets
|
172
|
+
line2.strip!
|
173
|
+
parts2 = line2.split(sep)
|
174
|
+
key2 = parts2.shift
|
175
|
+
end
|
176
|
+
end
|
177
|
+
|
178
|
+
output.close
|
179
|
+
end
|
3
180
|
#{{{ Attach Methods
|
4
181
|
|
5
182
|
def attach_same_key(other, fields = nil)
|
@@ -26,7 +203,7 @@ class TSV
|
|
26
203
|
def attach_source_key(other, source, fields = nil)
|
27
204
|
fields = other.fields - [key_field].concat(self.fields) if fields.nil?
|
28
205
|
|
29
|
-
other = other.tsv unless TSV === other
|
206
|
+
other = other.tsv(:persistence => :no_create) unless TSV === other
|
30
207
|
field_positions = fields.collect{|field| other.identify_field field}
|
31
208
|
field_names = field_positions.collect{|pos| pos == :key ? other.key_field : other.fields[pos] }
|
32
209
|
|
@@ -87,7 +264,11 @@ class TSV
|
|
87
264
|
next unless other.include? source_key
|
88
265
|
new_values = field_positions.collect do |pos|
|
89
266
|
if pos == :key
|
90
|
-
|
267
|
+
if other.type == :double
|
268
|
+
[source_key]
|
269
|
+
else
|
270
|
+
source_key
|
271
|
+
end
|
91
272
|
else
|
92
273
|
other[source_key][pos]
|
93
274
|
end
|
@@ -119,7 +300,10 @@ class TSV
|
|
119
300
|
#{{{ Attach Helper
|
120
301
|
|
121
302
|
# May make an extra index!
|
122
|
-
def self.find_path(files,
|
303
|
+
def self.find_path(files, options = {})
|
304
|
+
options = Misc.add_defaults options, :in_namespace => false
|
305
|
+
in_namespace = options[:in_namespace]
|
306
|
+
|
123
307
|
if in_namespace
|
124
308
|
ids = [files.first.all_namespace_fields(in_namespace)]
|
125
309
|
ids += files[1..-1].collect{|f| f.all_fields}
|
@@ -128,9 +312,6 @@ class TSV
|
|
128
312
|
end
|
129
313
|
id_list = []
|
130
314
|
|
131
|
-
ids.flatten.each do |field|
|
132
|
-
end
|
133
|
-
|
134
315
|
ids.each_with_index do |list, i|
|
135
316
|
break if i == ids.length - 1
|
136
317
|
match = list.select{|field|
|
@@ -148,8 +329,12 @@ class TSV
|
|
148
329
|
end
|
149
330
|
end
|
150
331
|
|
151
|
-
def self.build_traverse_index(files,
|
152
|
-
|
332
|
+
def self.build_traverse_index(files, options = {})
|
333
|
+
options = Misc.add_defaults options, :in_namespace => false, :persist_input => false
|
334
|
+
in_namespace = options[:in_namespace]
|
335
|
+
persist_input = options[:persist_input]
|
336
|
+
|
337
|
+
path = find_path(files, options)
|
153
338
|
|
154
339
|
return nil if path.nil?
|
155
340
|
|
@@ -157,11 +342,10 @@ class TSV
|
|
157
342
|
|
158
343
|
Log.medium "Found Traversal: #{traversal_ids * " => "}"
|
159
344
|
|
160
|
-
current_key = files.first.all_fields.first
|
161
|
-
target = files.last.all_fields.first
|
162
|
-
target = nil
|
163
345
|
current_id, current_file = path.shift
|
164
|
-
|
346
|
+
current_key = current_file.all_fields.first
|
347
|
+
|
348
|
+
index = current_file.index :target => current_id, :fields => current_key, :persistence => persist_input
|
165
349
|
|
166
350
|
while not path.empty?
|
167
351
|
current_id, current_file = path.shift
|
@@ -175,7 +359,10 @@ class TSV
|
|
175
359
|
index
|
176
360
|
end
|
177
361
|
|
178
|
-
def self.find_traversal(tsv1, tsv2,
|
362
|
+
def self.find_traversal(tsv1, tsv2, options = {})
|
363
|
+
options = Misc.add_defaults options, :in_namespace => false
|
364
|
+
in_namespace = options[:in_namespace]
|
365
|
+
|
179
366
|
identifiers1 = tsv1.identifier_files || []
|
180
367
|
identifiers2 = tsv2.identifier_files || []
|
181
368
|
|
@@ -188,7 +375,7 @@ class TSV
|
|
188
375
|
files1.push identifiers1.shift
|
189
376
|
identifiers2.each_with_index do |e,i|
|
190
377
|
files2 = identifiers2[(0..i)]
|
191
|
-
index = build_traverse_index(files1 + files2.reverse,
|
378
|
+
index = build_traverse_index(files1 + files2.reverse, options)
|
192
379
|
return index if not index.nil?
|
193
380
|
end
|
194
381
|
end
|
@@ -197,8 +384,8 @@ class TSV
|
|
197
384
|
end
|
198
385
|
|
199
386
|
def attach(other, fields = nil, options = {})
|
200
|
-
options
|
201
|
-
in_namespace =
|
387
|
+
options = Misc.add_defaults options, :in_namespace => false
|
388
|
+
in_namespace = options[:in_namespace]
|
202
389
|
|
203
390
|
fields = other.fields - [key_field].concat(self.fields) if fields == :all
|
204
391
|
fields = other.fields_in_namespace - [key_field].concat(self.fields) if fields.nil?
|
@@ -211,7 +398,7 @@ class TSV
|
|
211
398
|
when (in_namespace and self.fields_in_namespace.include?(other.key_field))
|
212
399
|
attach_source_key other, other.key_field, fields
|
213
400
|
else
|
214
|
-
index = TSV.find_traversal(self, other,
|
401
|
+
index = TSV.find_traversal(self, other, options)
|
215
402
|
raise "Cannot traverse identifiers" if index.nil?
|
216
403
|
attach_index other, index, fields
|
217
404
|
end
|
@@ -225,4 +412,20 @@ class TSV
|
|
225
412
|
reorder :key, detached_fields
|
226
413
|
end
|
227
414
|
|
415
|
+
def paste(other, options = {})
|
416
|
+
tmpfile = TmpFile.tmp_file
|
417
|
+
TSV.paste(self.to_s, other.to_s, tmpfile)
|
418
|
+
|
419
|
+
new = TSV.new(tmpfile, options)
|
420
|
+
|
421
|
+
new.key_field = self.key_field unless self.key_field.nil?
|
422
|
+
if self.fields and other.fields
|
423
|
+
new.fields = self.fields + other.fields
|
424
|
+
end
|
425
|
+
|
426
|
+
FileUtils.rm tmpfile if File.exists? tmpfile
|
427
|
+
|
428
|
+
new
|
429
|
+
end
|
430
|
+
|
228
431
|
end
|
data/lib/rbbt/util/tsv/index.rb
CHANGED
@@ -232,6 +232,7 @@ class TSV
|
|
232
232
|
end
|
233
233
|
|
234
234
|
def self.field_matches(tsv, values)
|
235
|
+
values = [values] if not Array === values
|
235
236
|
if values.flatten.sort[0..9].compact.collect{|n| n.to_i} == (1..10).to_a
|
236
237
|
return {}
|
237
238
|
end
|
@@ -247,7 +248,7 @@ class TSV
|
|
247
248
|
if tsv.type == :double
|
248
249
|
tsv.through do |key,entry_values|
|
249
250
|
fields.zip(entry_values).each do |field,entry_field_values|
|
250
|
-
field_values[field].concat entry_field_values
|
251
|
+
field_values[field].concat entry_field_values unless entry_field_values.nil?
|
251
252
|
end
|
252
253
|
end
|
253
254
|
else
|
@@ -271,6 +272,10 @@ class TSV
|
|
271
272
|
TSV.field_matches(self, values)
|
272
273
|
end
|
273
274
|
|
275
|
+
def guess_field(values)
|
276
|
+
field_matches(values).sort_by{|field, matches| matches.uniq.length}.last
|
277
|
+
end
|
278
|
+
|
274
279
|
def sorted_index(pos_start = nil, pos_end = nil)
|
275
280
|
raise "Please specify indexing fields" if (pos_start.nil? and fields.length > 2)
|
276
281
|
|
@@ -1,4 +1,3 @@
|
|
1
|
-
|
2
1
|
class TSV
|
3
2
|
|
4
3
|
def through(new_key_field = :key, new_fields = nil, &block)
|
@@ -177,7 +176,7 @@ class TSV
|
|
177
176
|
new.filename = filename
|
178
177
|
new.case_insensitive = case_insensitive
|
179
178
|
|
180
|
-
|
179
|
+
case
|
181
180
|
when (method.nil? and block_given?)
|
182
181
|
through do |key, values|
|
183
182
|
new[key] = values if yield key, values
|
@@ -199,7 +198,7 @@ class TSV
|
|
199
198
|
method = method.values.first
|
200
199
|
case
|
201
200
|
when (Array === method and (key == :key or key_field == key))
|
202
|
-
method.each{|item|
|
201
|
+
method.each{|item| new[item] = self[item] if self.include? item}
|
203
202
|
when Array === method
|
204
203
|
through :key, key do |key, values|
|
205
204
|
new[key] = self[key] if (values.flatten & method).any?
|
@@ -250,9 +249,9 @@ class TSV
|
|
250
249
|
def add_field(name = nil)
|
251
250
|
each do |key, values|
|
252
251
|
new_values = yield(key, values)
|
253
|
-
new_values = [new_values] if type == :double and not Array
|
252
|
+
new_values = [new_values] if type == :double and not Array === new_values
|
254
253
|
|
255
|
-
self[key] = values + [
|
254
|
+
self[key] = values + [new_values]
|
256
255
|
end
|
257
256
|
|
258
257
|
self.fields = self.fields + [name] if fields != nil and name != nil
|
data/lib/rbbt/util/tsv/parse.rb
CHANGED
@@ -7,6 +7,7 @@ class TSV
|
|
7
7
|
## split with delimiter, do not remove empty
|
8
8
|
fields = io.split(delimiter, -1)
|
9
9
|
|
10
|
+
|
10
11
|
fields
|
11
12
|
end
|
12
13
|
|
@@ -116,6 +117,8 @@ class TSV
|
|
116
117
|
fix, exclude, select, grep =
|
117
118
|
Misc.process_options options, :fix, :exclude, :select, :grep
|
118
119
|
|
120
|
+
exclude ||= Misc.process_options options, :reject if options.include? :reject
|
121
|
+
|
119
122
|
#{{{ Process rest
|
120
123
|
data = {}
|
121
124
|
single = type.to_sym != :double
|
@@ -152,9 +155,10 @@ class TSV
|
|
152
155
|
if single
|
153
156
|
ids = parse_fields(parts[key_pos], sep2)
|
154
157
|
ids.collect!{|id| id.downcase} if case_insensitive
|
158
|
+
ids = ids.reject{|_id| _id.empty?}.uniq
|
155
159
|
|
156
160
|
id = ids.shift
|
157
|
-
ids.each do |id2| data[id2] = "__Ref:#{id}"
|
161
|
+
ids.each do |id2| data[id2] = "__Ref:#{id}" unless data.include? id2 end
|
158
162
|
|
159
163
|
next if data.include?(id) and type != :flat
|
160
164
|
|
@@ -171,8 +175,8 @@ class TSV
|
|
171
175
|
|
172
176
|
extra.collect! do |elem|
|
173
177
|
case
|
174
|
-
when String === cast
|
175
|
-
elem.send(cast)
|
178
|
+
when (String === cast or Symbol === cast)
|
179
|
+
elem.send(cast.to_s)
|
176
180
|
when Proc === cast
|
177
181
|
cast.call elem
|
178
182
|
end
|
@@ -195,9 +199,26 @@ class TSV
|
|
195
199
|
else
|
196
200
|
ids = parse_fields(parts[key_pos], sep2)
|
197
201
|
ids.collect!{|id| id.downcase} if case_insensitive
|
202
|
+
ids = ids.reject{|_id| _id.empty?}.uniq
|
203
|
+
|
204
|
+
next if ids.empty?
|
198
205
|
|
199
206
|
id = ids.shift
|
200
|
-
|
207
|
+
while data.include? id and data[id] =~ /__Ref:(.*)/
|
208
|
+
data[id] = data[$1].collect{|e| e.dup}
|
209
|
+
end
|
210
|
+
|
211
|
+
all_ids = [id]
|
212
|
+
ids.each do |id2|
|
213
|
+
if data.include? id2
|
214
|
+
while data[id2] =~ /__Ref:(.*)/
|
215
|
+
data[id2] = data[$1].collect{|e| e.dup}
|
216
|
+
end
|
217
|
+
all_ids << id2
|
218
|
+
else
|
219
|
+
data[id2] = "__Ref:#{id}"
|
220
|
+
end
|
221
|
+
end
|
201
222
|
|
202
223
|
if other_pos.nil? or (fields == nil and type == :flat)
|
203
224
|
other_pos = (0..(parts.length - 1)).to_a
|
@@ -207,31 +228,34 @@ class TSV
|
|
207
228
|
extra = parts.values_at(*other_pos).collect{|f| parse_fields(f, sep2)}
|
208
229
|
extra.collect! do |list|
|
209
230
|
case
|
210
|
-
when String === cast
|
211
|
-
list.collect{|elem| elem.send(cast)}
|
231
|
+
when (String === cast or Symbol === cast)
|
232
|
+
list.collect{|elem| elem.send(cast.to_s)}
|
212
233
|
when Proc === cast
|
213
234
|
list.collect{|elem| cast.call elem}
|
214
235
|
end
|
215
236
|
end if cast
|
216
237
|
|
217
238
|
max_cols = extra.size if extra.size > (max_cols || 0)
|
218
|
-
|
219
|
-
|
220
|
-
|
221
|
-
|
222
|
-
data[id] = extra
|
239
|
+
|
240
|
+
all_ids.each do |id|
|
241
|
+
if not merge
|
242
|
+
data[id] = extra unless data.include? id
|
223
243
|
else
|
224
|
-
|
225
|
-
|
226
|
-
|
227
|
-
|
228
|
-
|
229
|
-
|
244
|
+
if not data.include? id
|
245
|
+
data[id] = extra
|
246
|
+
else
|
247
|
+
entry = data[id]
|
248
|
+
while entry =~ /__Ref:(.*)/ do entry = data[$1] end
|
249
|
+
extra.each_with_index do |f, i|
|
250
|
+
if f.empty?
|
251
|
+
next unless keep_empty
|
252
|
+
f= [""]
|
253
|
+
end
|
254
|
+
entry[i] ||= []
|
255
|
+
entry[i] = entry[i].concat f
|
230
256
|
end
|
231
|
-
|
232
|
-
entry[i] = entry[i].concat f
|
257
|
+
data[id] = entry
|
233
258
|
end
|
234
|
-
data[id] = entry
|
235
259
|
end
|
236
260
|
end
|
237
261
|
end
|
@@ -254,7 +278,7 @@ class TSV
|
|
254
278
|
|
255
279
|
fields = nil if Fixnum === fields or (Array === fields and fields.select{|f| Fixnum === f}.any?)
|
256
280
|
fields ||= other_fields
|
257
|
-
[data, {:key_field => key_field, :fields => fields, :type => type, :case_insensitive => case_insensitive, :namespace => namespace, :datadir => options[:datadir], :identifiers => options[:identifiers], :cast =>
|
281
|
+
[data, {:key_field => key_field, :fields => fields, :type => type, :case_insensitive => case_insensitive, :namespace => namespace, :datadir => options[:datadir], :identifiers => options[:identifiers], :cast => (cast.nil? ? false : cast)}]
|
258
282
|
end
|
259
283
|
|
260
284
|
end
|