shalmaneser-rosy 1.2.0.rc4 → 1.2.rc5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +47 -18
- data/bin/rosy +14 -7
- data/lib/rosy/FailedParses.rb +22 -20
- data/lib/rosy/FeatureInfo.rb +35 -31
- data/lib/rosy/GfInduce.rb +132 -130
- data/lib/rosy/GfInduceFeature.rb +86 -68
- data/lib/rosy/InputData.rb +59 -55
- data/lib/rosy/RosyConfusability.rb +47 -40
- data/lib/rosy/RosyEval.rb +55 -55
- data/lib/rosy/RosyFeatureExtractors.rb +295 -290
- data/lib/rosy/RosyFeaturize.rb +54 -67
- data/lib/rosy/RosyInspect.rb +52 -50
- data/lib/rosy/RosyIterator.rb +73 -67
- data/lib/rosy/RosyPhase2FeatureExtractors.rb +48 -48
- data/lib/rosy/RosyPruning.rb +39 -31
- data/lib/rosy/RosyServices.rb +116 -115
- data/lib/rosy/RosySplit.rb +55 -53
- data/lib/rosy/RosyTask.rb +7 -3
- data/lib/rosy/RosyTest.rb +174 -191
- data/lib/rosy/RosyTrain.rb +46 -50
- data/lib/rosy/RosyTrainingTestTable.rb +101 -99
- data/lib/rosy/TargetsMostFrequentFrame.rb +13 -9
- data/lib/rosy/{AbstractFeatureAndExternal.rb → abstract_feature_extractor.rb} +22 -97
- data/lib/rosy/abstract_single_feature_extractor.rb +52 -0
- data/lib/rosy/external_feature_extractor.rb +35 -0
- data/lib/rosy/opt_parser.rb +231 -201
- data/lib/rosy/rosy.rb +63 -64
- data/lib/rosy/rosy_conventions.rb +66 -0
- data/lib/rosy/rosy_error.rb +15 -0
- data/lib/rosy/var_var_restriction.rb +16 -0
- data/lib/shalmaneser/rosy.rb +1 -0
- metadata +26 -19
- data/lib/rosy/ExternalConfigData.rb +0 -58
- data/lib/rosy/View.rb +0 -418
- data/lib/rosy/rosy_config_data.rb +0 -121
- data/test/frprep/test_opt_parser.rb +0 -94
- data/test/functional/functional_test_helper.rb +0 -58
- data/test/functional/test_fred.rb +0 -47
- data/test/functional/test_frprep.rb +0 -99
- data/test/functional/test_rosy.rb +0 -40
data/lib/rosy/rosy.rb
CHANGED
@@ -8,71 +8,70 @@ require 'rosy/RosyTrain'
|
|
8
8
|
require 'rosy/RosyInspect'
|
9
9
|
require 'rosy/RosyEval'
|
10
10
|
require 'rosy/RosyServices'
|
11
|
+
require 'logging'
|
12
|
+
require 'rosy/rosy_error'
|
11
13
|
|
12
|
-
module
|
13
|
-
|
14
|
+
module Shalmaneser
|
15
|
+
module Rosy
|
16
|
+
class Rosy
|
14
17
|
|
15
|
-
|
16
|
-
|
17
|
-
|
18
|
-
end
|
19
|
-
|
20
|
-
def assign
|
21
|
-
|
22
|
-
# make rosy directory pattern:
|
23
|
-
# main rosy directory name (data_dir) plus subdirectory
|
24
|
-
# named after the experiment ID
|
25
|
-
rosy_dir_pattern = File.new_dir(@exp.get("data_dir")) + "<exp_ID>/"
|
26
|
-
@exp.set_entry("rosy_dir", rosy_dir_pattern)
|
27
|
-
|
28
|
-
##
|
29
|
-
# open database
|
30
|
-
|
31
|
-
rosy_dir = File.new_dir(@exp.instantiate("rosy_dir",
|
32
|
-
"exp_ID" => @exp.get("experiment_ID")))
|
33
|
-
database = get_db_interface(@exp, rosy_dir, "features")
|
34
|
-
|
35
|
-
table_obj = RosyTrainingTestTable.new(@exp, database)
|
36
|
-
|
37
|
-
##
|
38
|
-
# start the actual processing,
|
39
|
-
# according to given arguments
|
40
|
-
|
41
|
-
# initialize task object
|
42
|
-
#begin
|
43
|
-
case @task
|
44
|
-
when "featurize"
|
45
|
-
task_obj = RosyFeaturize.new(@exp, @opts, table_obj)
|
46
|
-
when "split"
|
47
|
-
task_obj = RosySplit.new(@exp, @opts, table_obj)
|
48
|
-
when "train"
|
49
|
-
task_obj = RosyTrain.new(@exp, @opts, table_obj)
|
50
|
-
when "test"
|
51
|
-
task_obj = RosyTest.new(@exp, @opts, table_obj)
|
52
|
-
when "eval"
|
53
|
-
task_obj = RosyEvalTask.new(@exp, @opts, table_obj)
|
54
|
-
when "inspect"
|
55
|
-
task_obj = RosyInspect.new(@exp, @opts, table_obj)
|
56
|
-
when "services"
|
57
|
-
task_obj = RosyServices.new(@exp, @opts, table_obj)
|
58
|
-
else
|
59
|
-
raise "Shouldn't be here"
|
18
|
+
def initialize(options)
|
19
|
+
@exp, @opts = options
|
20
|
+
@task = @opts['--task']
|
60
21
|
end
|
61
|
-
|
62
|
-
|
63
|
-
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
70
|
-
|
22
|
+
|
23
|
+
def assign
|
24
|
+
|
25
|
+
# make rosy directory pattern:
|
26
|
+
# main rosy directory name (data_dir) plus subdirectory
|
27
|
+
# named after the experiment ID
|
28
|
+
rosy_dir_pattern = File.new_dir(@exp.get("data_dir")) + "<exp_ID>/"
|
29
|
+
@exp.set_entry("rosy_dir", rosy_dir_pattern)
|
30
|
+
|
31
|
+
##
|
32
|
+
# open database
|
33
|
+
|
34
|
+
rosy_dir = File.new_dir(@exp.instantiate("rosy_dir",
|
35
|
+
"exp_ID" => @exp.get("experiment_ID")))
|
36
|
+
database = DBInterface.get_db_interface(@exp, rosy_dir, "features")
|
37
|
+
|
38
|
+
table_obj = RosyTrainingTestTable.new(@exp, database)
|
39
|
+
|
40
|
+
##
|
41
|
+
# start the actual processing,
|
42
|
+
# according to given arguments
|
43
|
+
|
44
|
+
# initialize task object
|
45
|
+
task = case @task
|
46
|
+
when "featurize"
|
47
|
+
RosyFeaturize.new(@exp, @opts, table_obj)
|
48
|
+
when "split"
|
49
|
+
RosySplit.new(@exp, @opts, table_obj)
|
50
|
+
when "train"
|
51
|
+
RosyTrain.new(@exp, @opts, table_obj)
|
52
|
+
when "test"
|
53
|
+
RosyTest.new(@exp, @opts, table_obj)
|
54
|
+
when "eval"
|
55
|
+
RosyEvalTask.new(@exp, @opts, table_obj)
|
56
|
+
when "inspect"
|
57
|
+
RosyInspect.new(@exp, @opts, table_obj)
|
58
|
+
when "services"
|
59
|
+
RosyServices.new(@exp, @opts, table_obj)
|
60
|
+
else
|
61
|
+
raise "Shouldn't be here"
|
62
|
+
end
|
63
|
+
|
64
|
+
# execute task
|
65
|
+
begin
|
66
|
+
task.perform
|
67
|
+
rescue => e
|
68
|
+
raise RosyError.new 'Rosy is doing bad!', e
|
69
|
+
ensure
|
70
|
+
database.close
|
71
|
+
end
|
72
|
+
|
73
|
+
LOGGER.info "#{PROGRAM_NAME} finished performing the task: #{task}!"
|
71
74
|
end
|
72
|
-
|
73
|
-
|
74
|
-
|
75
|
-
end
|
76
|
-
|
77
|
-
end # class Rosy
|
78
|
-
end # module Rosy
|
75
|
+
end # class Rosy
|
76
|
+
end # module Rosy
|
77
|
+
end
|
@@ -0,0 +1,66 @@
|
|
1
|
+
require 'monkey_patching/file'
|
2
|
+
|
3
|
+
module Shalmaneser
|
4
|
+
module Rosy
|
5
|
+
|
6
|
+
module_function
|
7
|
+
|
8
|
+
# transforming feature output to a format that classifiers can handle
|
9
|
+
# @note Used only under Rosy.
|
10
|
+
def prepare_output_for_classifiers(string)
|
11
|
+
# change punctuation to _PUNCT_
|
12
|
+
# and change empty space to _
|
13
|
+
# because otherwise some classifiers may spit
|
14
|
+
string.gsub(/[.":';`]/,"_PUNCT_").gsub(/\s/,"_")
|
15
|
+
end
|
16
|
+
|
17
|
+
###
|
18
|
+
# classifier directory:
|
19
|
+
# either user-given classifier_dir or our own default classifier directory,
|
20
|
+
# then argrec/arglab/onestep, plus the splitID, if there is one
|
21
|
+
# @note Need the extended File class.
|
22
|
+
# @note Used only under Rosy.
|
23
|
+
def classifier_directory_name(exp, # RosyConfigData object
|
24
|
+
step, # argrec, arglab, onestep
|
25
|
+
splitID) # string or nil
|
26
|
+
|
27
|
+
if exp.get("classifier_dir")
|
28
|
+
base_dir = File.new_dir(exp.get("classifier_dir"))
|
29
|
+
else
|
30
|
+
base_dir = File.new_dir(exp.instantiate("rosy_dir",
|
31
|
+
"exp_ID" => exp.get("experiment_ID")))
|
32
|
+
end
|
33
|
+
classif_base_dir = File.new_dir(base_dir, "classif_dir")
|
34
|
+
|
35
|
+
if splitID
|
36
|
+
return File.new_dir(classif_base_dir, step + "." + splitID.to_s)
|
37
|
+
else
|
38
|
+
return File.new_dir(classif_base_dir, step)
|
39
|
+
end
|
40
|
+
end
|
41
|
+
|
42
|
+
#################################################################
|
43
|
+
#################################################################
|
44
|
+
|
45
|
+
###
|
46
|
+
# @note Used only under Rosy.
|
47
|
+
# instance ID: sentence ID plus frame ID
|
48
|
+
def construct_instance_id(sentence_id, frame_id)
|
49
|
+
sentence_id.to_s + "---" + frame_id.to_s
|
50
|
+
end
|
51
|
+
|
52
|
+
# @note Not used anywhere.
|
53
|
+
def deconstruct_instance_id(instance_id)
|
54
|
+
instance_id.split("---")
|
55
|
+
end
|
56
|
+
|
57
|
+
#################################################################
|
58
|
+
#################################################################
|
59
|
+
|
60
|
+
# default test ID given when the user didn't specify one
|
61
|
+
# @note Used only under Rosy.
|
62
|
+
def default_test_ID
|
63
|
+
"apply"
|
64
|
+
end
|
65
|
+
end
|
66
|
+
end
|
@@ -0,0 +1,15 @@
|
|
1
|
+
module Shalmaneser
|
2
|
+
module Rosy
|
3
|
+
class RosyError < StandardError
|
4
|
+
# @param [String] msg A custom message for this exception.
|
5
|
+
# @param [Exception] nested_exception An external exception
|
6
|
+
# which is reused to provide more information.
|
7
|
+
def initialize(msg = nil, nested_exception = nil)
|
8
|
+
if nested_exception
|
9
|
+
msg = "#{nested_exception.class}: #{nested_exception.message}\n#{msg}" << nested_exception.backtrace.join("\n")
|
10
|
+
end
|
11
|
+
super(msg)
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
require 'value_restriction'
|
2
|
+
###
|
3
|
+
# value restrictions saying that variable1 = variable2:
|
4
|
+
# here, value is a variable name, and the table names
|
5
|
+
# must be already included
|
6
|
+
module Shalmaneser
|
7
|
+
module Rosy
|
8
|
+
class VarVarRestriction < ValueRestriction
|
9
|
+
def initialize(column, value, var_hash = {})
|
10
|
+
super(column, value, var_hash)
|
11
|
+
@val_is_variable = true
|
12
|
+
@table_name_included = true
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1 @@
|
|
1
|
+
# A dummy file to require for now.
|
metadata
CHANGED
@@ -1,15 +1,29 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: shalmaneser-rosy
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 1.2.
|
4
|
+
version: 1.2.rc5
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Andrei Beliankou
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date:
|
11
|
+
date: 2016-01-13 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
|
+
- !ruby/object:Gem::Dependency
|
14
|
+
name: shalmaneser-lib
|
15
|
+
requirement: !ruby/object:Gem::Requirement
|
16
|
+
requirements:
|
17
|
+
- - '='
|
18
|
+
- !ruby/object:Gem::Version
|
19
|
+
version: 1.2.rc5
|
20
|
+
type: :runtime
|
21
|
+
prerelease: false
|
22
|
+
version_requirements: !ruby/object:Gem::Requirement
|
23
|
+
requirements:
|
24
|
+
- - '='
|
25
|
+
- !ruby/object:Gem::Version
|
26
|
+
version: 1.2.rc5
|
13
27
|
- !ruby/object:Gem::Dependency
|
14
28
|
name: mysql
|
15
29
|
requirement: !ruby/object:Gem::Requirement
|
@@ -24,7 +38,7 @@ dependencies:
|
|
24
38
|
- - "~>"
|
25
39
|
- !ruby/object:Gem::Version
|
26
40
|
version: '2.9'
|
27
|
-
description: ROSY -
|
41
|
+
description: ROSY - ROle assignment SYstem.
|
28
42
|
email: arbox@yandex.ru
|
29
43
|
executables:
|
30
44
|
- rosy
|
@@ -39,8 +53,6 @@ files:
|
|
39
53
|
- LICENSE.md
|
40
54
|
- README.md
|
41
55
|
- bin/rosy
|
42
|
-
- lib/rosy/AbstractFeatureAndExternal.rb
|
43
|
-
- lib/rosy/ExternalConfigData.rb
|
44
56
|
- lib/rosy/FailedParses.rb
|
45
57
|
- lib/rosy/FeatureInfo.rb
|
46
58
|
- lib/rosy/GfInduce.rb
|
@@ -61,15 +73,15 @@ files:
|
|
61
73
|
- lib/rosy/RosyTrain.rb
|
62
74
|
- lib/rosy/RosyTrainingTestTable.rb
|
63
75
|
- lib/rosy/TargetsMostFrequentFrame.rb
|
64
|
-
- lib/rosy/
|
76
|
+
- lib/rosy/abstract_feature_extractor.rb
|
77
|
+
- lib/rosy/abstract_single_feature_extractor.rb
|
78
|
+
- lib/rosy/external_feature_extractor.rb
|
65
79
|
- lib/rosy/opt_parser.rb
|
66
80
|
- lib/rosy/rosy.rb
|
67
|
-
- lib/rosy/
|
68
|
-
-
|
69
|
-
-
|
70
|
-
-
|
71
|
-
- test/functional/test_frprep.rb
|
72
|
-
- test/functional/test_rosy.rb
|
81
|
+
- lib/rosy/rosy_conventions.rb
|
82
|
+
- lib/rosy/rosy_error.rb
|
83
|
+
- lib/rosy/var_var_restriction.rb
|
84
|
+
- lib/shalmaneser/rosy.rb
|
73
85
|
homepage: https://github.com/arbox/shalmaneser
|
74
86
|
licenses:
|
75
87
|
- GPL-2.0
|
@@ -92,14 +104,9 @@ required_rubygems_version: !ruby/object:Gem::Requirement
|
|
92
104
|
version: 1.3.1
|
93
105
|
requirements: []
|
94
106
|
rubyforge_project:
|
95
|
-
rubygems_version: 2.
|
107
|
+
rubygems_version: 2.5.1
|
96
108
|
signing_key:
|
97
109
|
specification_version: 4
|
98
110
|
summary: ROSY
|
99
|
-
test_files:
|
100
|
-
- test/frprep/test_opt_parser.rb
|
101
|
-
- test/functional/functional_test_helper.rb
|
102
|
-
- test/functional/test_fred.rb
|
103
|
-
- test/functional/test_frprep.rb
|
104
|
-
- test/functional/test_rosy.rb
|
111
|
+
test_files: []
|
105
112
|
has_rdoc:
|
@@ -1,58 +0,0 @@
|
|
1
|
-
# ExternalConfigData
|
2
|
-
# Katrin Erk January 2006
|
3
|
-
#
|
4
|
-
# All scripts that compute additional external knowledge sources
|
5
|
-
# for Fred and Rosy:
|
6
|
-
# access to configuration and experiment description file
|
7
|
-
|
8
|
-
require 'common/config_data'
|
9
|
-
|
10
|
-
##############################
|
11
|
-
# Class ExternalConfigData
|
12
|
-
#
|
13
|
-
# inherits from ConfigData,
|
14
|
-
# sets variable names appropriate to tasks of external knowledge sources
|
15
|
-
|
16
|
-
class ExternalConfigData < ConfigData
|
17
|
-
def initialize(filename)
|
18
|
-
|
19
|
-
# initialize config data object
|
20
|
-
super(filename, # config file
|
21
|
-
{ "directory" => "string", # features
|
22
|
-
|
23
|
-
"experiment_id" => "string",
|
24
|
-
|
25
|
-
"gfmap_restrict_to_downpath" => "bool",
|
26
|
-
"gfmap_restrict_pathlen" => "integer",
|
27
|
-
"gfmap_remove_gf" => "list"
|
28
|
-
},
|
29
|
-
[] # variables
|
30
|
-
)
|
31
|
-
|
32
|
-
# set access functions for list features
|
33
|
-
set_list_feature_access("gfmap_remove_gf",
|
34
|
-
method("access_as_stringlist"))
|
35
|
-
end
|
36
|
-
|
37
|
-
###
|
38
|
-
protected
|
39
|
-
|
40
|
-
#####
|
41
|
-
# access_as_stringlist
|
42
|
-
#
|
43
|
-
# assumed format:
|
44
|
-
#
|
45
|
-
# lhs = rhs1 rhs2 ... rhsN
|
46
|
-
#
|
47
|
-
# given in val_list as string tuples [rhs1,...,rhsN]
|
48
|
-
#
|
49
|
-
# join the rhs strings by spaces, return as string
|
50
|
-
# "rhs1 rhs2 ... rhsN"
|
51
|
-
#
|
52
|
-
def access_as_stringlist(val_list) # array:array:string
|
53
|
-
return val_list.map { |rhs| rhs.join(" ") }
|
54
|
-
end
|
55
|
-
end
|
56
|
-
|
57
|
-
|
58
|
-
|
data/lib/rosy/View.rb
DELETED
@@ -1,418 +0,0 @@
|
|
1
|
-
# class DBView
|
2
|
-
# KE, SP 27.1.05
|
3
|
-
#
|
4
|
-
# builds on class DBTable, which offers access to a database table
|
5
|
-
# extract views of the table (select columns, select rows)
|
6
|
-
# and offers access methods for these views.
|
7
|
-
# Rows of the table can be returned either as hashes or as arrays.
|
8
|
-
#
|
9
|
-
# There is a special column of the table (the name of which we get in the new() method),
|
10
|
-
# the gold column.
|
11
|
-
# It can be returned directly, or modified by some "dynamic feature object",
|
12
|
-
# and its value (modified or unmodified) will always be last in the array representation of a row.
|
13
|
-
|
14
|
-
require 'db/sql_query'
|
15
|
-
require "common/ruby_class_extensions"
|
16
|
-
require "common/RosyConventions"
|
17
|
-
|
18
|
-
class DBView
|
19
|
-
|
20
|
-
################
|
21
|
-
# new
|
22
|
-
#
|
23
|
-
# prepare a view.
|
24
|
-
# given a list of DB tables to access, each with its
|
25
|
-
# set of features to be returned in the view,
|
26
|
-
# a set of value restrictions,
|
27
|
-
# the name of the gold feature,
|
28
|
-
# and a list of objects that manipulate the gold feature into alternate
|
29
|
-
# gold features.
|
30
|
-
#
|
31
|
-
# value_restrictions restricts the view to those rows for which the value restrictions hold,
|
32
|
-
# e.g. only those rows where frame = Bla, or only those rows where partofspeech = Blupp
|
33
|
-
#
|
34
|
-
# The view remembers the indices of the _first_ table in the list of tables
|
35
|
-
# it is given.
|
36
|
-
#
|
37
|
-
# A standard dynamic ID can be given: DynGold objects all have an id() method,
|
38
|
-
# which returns a string, by which the use of the object can be requested
|
39
|
-
# of the view. If no dynamic ID is given below in methods each_array,
|
40
|
-
# each_hash, each_sentence, the system falls back to the standard dynamic ID.
|
41
|
-
# if none is given here, the standard DynGold object is the one that doesn't
|
42
|
-
# change the gold column. If one is given here, it will be used by default
|
43
|
-
# when no ID is given in each_hash, each_array, each_sentence
|
44
|
-
#
|
45
|
-
# The last parameter is a hash with the following optional entries:
|
46
|
-
# "gold":
|
47
|
-
# string: name of the gold feature
|
48
|
-
# If you want the gold feature to be mapped using a DynGold object,
|
49
|
-
# you need to specify this parameter -- and you need to include
|
50
|
-
# the gold feature in some feature_list.
|
51
|
-
# Warning: if a feature of this name appears in several of the
|
52
|
-
# feature lists, only the first one is mapped
|
53
|
-
# "dynamic_feature_list":
|
54
|
-
# array:DynGold objects, list of objects that map the gold feature
|
55
|
-
# to a different feature value (e.g. to "FE", "NONE")
|
56
|
-
# DynGold objects have one method make: string -> string
|
57
|
-
# that maps one gold feature,
|
58
|
-
# and one method id: -> string that gives an ID unique to this DynGold class
|
59
|
-
# and by which this DynGold class can be chosen.
|
60
|
-
# "standard_dyngold_id":
|
61
|
-
# string: standard DynGold object ID (see above)
|
62
|
-
# "sentence_id_feature":
|
63
|
-
# string: feature name for the sentence ID column, needed for each_sentence()
|
64
|
-
#
|
65
|
-
# further parameters that are passed on to SQLQuery.select: see there
|
66
|
-
|
67
|
-
def initialize(table_col_pairs, # array:SelectTableAndColumns objects
|
68
|
-
value_restrictions, # array:ValueRestriction objects
|
69
|
-
db_obj, # MySql object (from mysql.rb) that already has access to the correct database
|
70
|
-
parameters = {}) # hash with further parameters: see above
|
71
|
-
|
72
|
-
@db_obj = db_obj
|
73
|
-
@table_col_pairs = table_col_pairs
|
74
|
-
@parameters = parameters
|
75
|
-
|
76
|
-
# view empty?
|
77
|
-
if @table_col_pairs.empty? or
|
78
|
-
@table_col_pairs.big_and { |tc| tc.columns.class.to_s == "Array" and tc.columns.empty? }
|
79
|
-
@view_empty = true
|
80
|
-
return
|
81
|
-
else
|
82
|
-
@view_empty = false
|
83
|
-
end
|
84
|
-
|
85
|
-
# okay, we can make the view, it contains at least one table and
|
86
|
-
# at least one column:
|
87
|
-
# do one view for all columns requested, and one for the indices of each table
|
88
|
-
#
|
89
|
-
# @main_table is a DBResult object
|
90
|
-
@main_table = execute_command(SQLQuery.select(@table_col_pairs,
|
91
|
-
value_restrictions, parameters))
|
92
|
-
|
93
|
-
# index_tables: Hash: table name => DBResult object
|
94
|
-
@index_tables = Hash.new
|
95
|
-
table_col_pairs.each_with_index { |tc, index|
|
96
|
-
# read index column of this table, add all the other tables
|
97
|
-
# with empty column lists
|
98
|
-
index_table_col_pairs = @table_col_pairs.map_with_index { |other_tc, other_index|
|
99
|
-
if other_index == index
|
100
|
-
# the current table
|
101
|
-
SelectTableAndColumns.new(tc.table_obj,
|
102
|
-
[tc.table_obj.index_name])
|
103
|
-
else
|
104
|
-
# other table: keep just the table, not the columns
|
105
|
-
SelectTableAndColumns.new(other_tc.table_obj, nil)
|
106
|
-
end
|
107
|
-
}
|
108
|
-
@index_tables[tc.table_obj.table_name] = execute_command(SQLQuery.select(index_table_col_pairs,
|
109
|
-
value_restrictions, parameters))
|
110
|
-
}
|
111
|
-
|
112
|
-
# map gold to something else?
|
113
|
-
# yes, if parameters[gold] has been set
|
114
|
-
if @parameters["gold"]
|
115
|
-
@map_gold = true
|
116
|
-
# remember which column in the DB table is the gold column
|
117
|
-
@gold_index = column_names().index(@parameters["gold"])
|
118
|
-
else
|
119
|
-
@map_gold = false
|
120
|
-
end
|
121
|
-
end
|
122
|
-
|
123
|
-
################
|
124
|
-
# close
|
125
|
-
#
|
126
|
-
# to be called when the view is no longer needed:
|
127
|
-
# frees the DBResult objects underlying this view
|
128
|
-
def close()
|
129
|
-
unless @view_empty
|
130
|
-
@main_table.free()
|
131
|
-
@index_tables.each_value { |t| t.free() }
|
132
|
-
end
|
133
|
-
end
|
134
|
-
|
135
|
-
################
|
136
|
-
# write_to_file
|
137
|
-
#
|
138
|
-
# writes instances to a file
|
139
|
-
# each instance given as a comma-separated list of features
|
140
|
-
# The features are the ones given in my_feature_list
|
141
|
-
# (parameter to the new() method) above, in that order,
|
142
|
-
# plus (dynamic) gold, which is last.
|
143
|
-
#
|
144
|
-
# guarantees that comma is used only to separate features -- but no other
|
145
|
-
# changes in the feature values
|
146
|
-
def write_to_file(file, # stream to write to
|
147
|
-
dyn_gold_id=nil) #string: ID of a DynGold object from the dynamic_feature_list.
|
148
|
-
# if nil, main gold is used
|
149
|
-
|
150
|
-
each_instance_s(dyn_gold_id) { |instance_string|
|
151
|
-
file.puts instance_string
|
152
|
-
}
|
153
|
-
end
|
154
|
-
|
155
|
-
|
156
|
-
################
|
157
|
-
# each_instance_s
|
158
|
-
#
|
159
|
-
# yields each instance as a string:
|
160
|
-
# a comma-separated list of features
|
161
|
-
# The features are the ones given in my_feature_list
|
162
|
-
# (parameter to the new() method) above, in that order,
|
163
|
-
# plus (dynamic) gold, which is last.
|
164
|
-
#
|
165
|
-
# guarantees that comma is used only to separate features -- but no other
|
166
|
-
# changes in the feature values
|
167
|
-
def each_instance_s(dyn_gold_id=nil) #string: ID of a DynGold object from the dynamic_feature_list.
|
168
|
-
# if nil, main gold is used
|
169
|
-
each_array(dyn_gold_id) {|array|
|
170
|
-
yield array.map { |entry| entry.to_s.gsub(/,/, "COMMA") }.join(",")
|
171
|
-
}
|
172
|
-
end
|
173
|
-
|
174
|
-
################
|
175
|
-
# each_hash
|
176
|
-
#
|
177
|
-
# iterates over hashes representing rows
|
178
|
-
# in each row, there is a gold key/value pair
|
179
|
-
# specified by the optional argument dyn_gold_id.
|
180
|
-
# which is the string ID of a DynGold object
|
181
|
-
# from the dynamic_feature_list.
|
182
|
-
# If arg is not present, main gold is used
|
183
|
-
#
|
184
|
-
# The key for the gold is the dyn_gold_id
|
185
|
-
# If that is nil, the key is 'gold'
|
186
|
-
#
|
187
|
-
# yields: hashes column_name -> column_value
|
188
|
-
def each_hash(dyn_gold_id=nil) #string: ID of a DynGold object from the dynamic_feature_list, or nil
|
189
|
-
if @view_empty
|
190
|
-
return
|
191
|
-
end
|
192
|
-
if @map_gold
|
193
|
-
dyn_gold_obj = fetch_dyn_gold_obj(dyn_gold_id)
|
194
|
-
end
|
195
|
-
@main_table.reset()
|
196
|
-
|
197
|
-
@main_table.each_hash { |row_hash|
|
198
|
-
if @map_gold
|
199
|
-
row_hash[@parameters["gold"]] = dyn_gold_obj.make(row_hash[@parameters["gold"]])
|
200
|
-
end
|
201
|
-
|
202
|
-
yield row_hash
|
203
|
-
}
|
204
|
-
end
|
205
|
-
|
206
|
-
################
|
207
|
-
# each_array
|
208
|
-
#
|
209
|
-
# iterates over arrays representing rows
|
210
|
-
# the last item of each row is the gold column
|
211
|
-
# selected by the optional argument dyn_gold_id.
|
212
|
-
# which is the string ID of a DynGold object
|
213
|
-
# from the dynamic_feature_list.
|
214
|
-
# If arg is not present, main gold is used
|
215
|
-
#
|
216
|
-
# yields: arrays of column values,
|
217
|
-
# values are in the order of my_feature_list given
|
218
|
-
# to the new() method, (dynamic) gold is last
|
219
|
-
def each_array(dyn_gold_id=nil) #string: ID of a DynGold object from the dynamic_feature_list, or nil
|
220
|
-
|
221
|
-
if @view_empty
|
222
|
-
return
|
223
|
-
end
|
224
|
-
if @map_gold
|
225
|
-
dyn_gold_obj = fetch_dyn_gold_obj(dyn_gold_id)
|
226
|
-
end
|
227
|
-
@main_table.reset()
|
228
|
-
|
229
|
-
@main_table.each {|row|
|
230
|
-
if @gold_index
|
231
|
-
gold = row.delete_at(@gold_index)
|
232
|
-
if @map_gold
|
233
|
-
row.push dyn_gold_obj.make(gold)
|
234
|
-
else
|
235
|
-
row.push gold
|
236
|
-
end
|
237
|
-
end
|
238
|
-
|
239
|
-
yield row
|
240
|
-
}
|
241
|
-
end
|
242
|
-
|
243
|
-
################
|
244
|
-
# update_column
|
245
|
-
#
|
246
|
-
# update a column for all rows of this view
|
247
|
-
#
|
248
|
-
# Given a column name to be updated, and a list of value tuples,
|
249
|
-
# update each row of the view, or rather the appropriate column of each row of the view,
|
250
|
-
# with values for that row.
|
251
|
-
#
|
252
|
-
# the list has the same length as the view, as there must be a value tuple
|
253
|
-
# for each row of the view.
|
254
|
-
#
|
255
|
-
# returns: nothing
|
256
|
-
def update_column(name, # string: column name
|
257
|
-
values) # array of Objects
|
258
|
-
|
259
|
-
if @view_empty
|
260
|
-
raise "Cannot update empty view"
|
261
|
-
end
|
262
|
-
|
263
|
-
# find the first table in @table_col_pairs that has
|
264
|
-
# a column with this name
|
265
|
-
# and update that column
|
266
|
-
@table_col_pairs.each { |tc|
|
267
|
-
if (tc.columns.class.to_s == "Array" and tc.columns.include? name) or
|
268
|
-
(tc.columns == "*" and tc.table_obj.list_column_names().include? name)
|
269
|
-
|
270
|
-
table_name = tc.table_obj.table_name
|
271
|
-
|
272
|
-
# sanity check: number of update entries must match
|
273
|
-
# number of entries in this view
|
274
|
-
unless values.length() == @index_tables[table_name].num_rows()
|
275
|
-
$stderr.puts "Error: length of value array (#{values.length}) is not equal to length of view (#{@index_tables[table_name].num_rows})!"
|
276
|
-
exit 1
|
277
|
-
end
|
278
|
-
|
279
|
-
@index_tables[tc.table_obj.table_name].reset()
|
280
|
-
|
281
|
-
values.each { |value|
|
282
|
-
index = @index_tables[table_name].fetch_row().first
|
283
|
-
tc.table_obj.update_row(index, [[name, value]])
|
284
|
-
}
|
285
|
-
|
286
|
-
return
|
287
|
-
end
|
288
|
-
}
|
289
|
-
|
290
|
-
# no match found
|
291
|
-
$stderr.puts "View.rb Error: cannot update a column that is not in this view: #{name}"
|
292
|
-
exit 1
|
293
|
-
end
|
294
|
-
|
295
|
-
|
296
|
-
################
|
297
|
-
# each_sentence
|
298
|
-
#
|
299
|
-
# like each_hash, but it groups the row hashes sentence-wise
|
300
|
-
# sentence boundaries in the view are detected by the change in a
|
301
|
-
# special column describing sentence IDs
|
302
|
-
#
|
303
|
-
# also needs a dyngold object id
|
304
|
-
#
|
305
|
-
# returns: an array of hashes column_name -> column_value
|
306
|
-
def each_sentence(dyn_gold_id = nil) # string: ID of a DynGold object from the dynamic_feature_list, or nil
|
307
|
-
|
308
|
-
# sanity check 1: need to know what the sentence ID is
|
309
|
-
unless @parameters["sentence_id_feature"]
|
310
|
-
raise "I need the name of the sentence ID feature for each_sentence()"
|
311
|
-
end
|
312
|
-
# sanity check 2: the view needs to include the sentence ID
|
313
|
-
unless column_names().include? @parameters["sentence_id_feature"]
|
314
|
-
raise "View.each_sentence: Cannot do this without sentence ID in the view"
|
315
|
-
end
|
316
|
-
|
317
|
-
last_sent_id = nil
|
318
|
-
sentence = Array.new
|
319
|
-
each_hash(dyn_gold_id) {|row_hash|
|
320
|
-
if last_sent_id != row_hash[@parameters["sentence_id_feature"]] and
|
321
|
-
(!(last_sent_id.nil?))
|
322
|
-
yield sentence
|
323
|
-
sentence = Array.new
|
324
|
-
end
|
325
|
-
last_sent_id = row_hash[@parameters["sentence_id_feature"]]
|
326
|
-
sentence << row_hash
|
327
|
-
}
|
328
|
-
unless sentence.empty?
|
329
|
-
yield sentence
|
330
|
-
end
|
331
|
-
end
|
332
|
-
|
333
|
-
######################
|
334
|
-
# length
|
335
|
-
#
|
336
|
-
# returns the length of the view: the number of its rows
|
337
|
-
def length()
|
338
|
-
return @index_tables[@table_col_pairs.first.table_obj.table_name].num_rows
|
339
|
-
end
|
340
|
-
|
341
|
-
###
|
342
|
-
private
|
343
|
-
|
344
|
-
################
|
345
|
-
# column_names
|
346
|
-
#
|
347
|
-
# returns: array:string
|
348
|
-
# the list of column names for this view
|
349
|
-
# in the right order
|
350
|
-
def column_names()
|
351
|
-
if @view_empty
|
352
|
-
return []
|
353
|
-
else
|
354
|
-
return @main_table.list_column_names()
|
355
|
-
end
|
356
|
-
end
|
357
|
-
|
358
|
-
######
|
359
|
-
# fetch_dyn_gold_obj
|
360
|
-
#
|
361
|
-
# given an ID of a gold object, look for the DynGold object
|
362
|
-
# with this ID in the dynamic_feature_list and return it
|
363
|
-
# If the ID is nil, use the standard dynamic gold ID that
|
364
|
-
# has been set in the new() method.
|
365
|
-
# If that is nil too, take the non-modified gold as a
|
366
|
-
# default: return a dummy object with a make() method
|
367
|
-
# that just returns its parameter.
|
368
|
-
#
|
369
|
-
# returns: object offering a make() method
|
370
|
-
|
371
|
-
def fetch_dyn_gold_obj(dyn_gold_id) # string or nil
|
372
|
-
# find a DynGold object that will transform the gold column
|
373
|
-
if dyn_gold_id.nil?
|
374
|
-
dyn_gold_id = @parameters["standard_dyngold_id"]
|
375
|
-
end
|
376
|
-
|
377
|
-
dyn_gold_obj = "we need an object that can do 'make'"
|
378
|
-
if dyn_gold_id
|
379
|
-
unless @parameters["dynamic_feature_list"]
|
380
|
-
raise "No dynamic features given"
|
381
|
-
end
|
382
|
-
|
383
|
-
dyn_gold_obj = @parameters["dynamic_feature_list"].detect { |obj|
|
384
|
-
obj.id() == dyn_gold_id
|
385
|
-
}
|
386
|
-
if dyn_gold_obj.nil?
|
387
|
-
$stderr.puts "View.rb: Unknown DynGold ID " + dyn_gold_id
|
388
|
-
$stderr.puts "Using unchanged gold"
|
389
|
-
dyn_gold_id = nil
|
390
|
-
end
|
391
|
-
end
|
392
|
-
|
393
|
-
unless dyn_gold_id
|
394
|
-
# no dynamic gold ID: use unchanged gold by default
|
395
|
-
class << dyn_gold_obj
|
396
|
-
def make(x)
|
397
|
-
x
|
398
|
-
end
|
399
|
-
def id()
|
400
|
-
return "gold"
|
401
|
-
end
|
402
|
-
end
|
403
|
-
end
|
404
|
-
return dyn_gold_obj
|
405
|
-
end
|
406
|
-
|
407
|
-
def execute_command(command)
|
408
|
-
begin
|
409
|
-
return @db_obj.query(command)
|
410
|
-
rescue MysqlError => e
|
411
|
-
$stderr.puts "Error executing SQL query. Command was:\n" + command
|
412
|
-
$stderr.puts "Error code: #{e.errno}"
|
413
|
-
$stderr.puts "Error message: #{e.error}"
|
414
|
-
raise e
|
415
|
-
end
|
416
|
-
end
|
417
|
-
|
418
|
-
end
|