stats_package_syntax_file_generator 1.0.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (24) hide show
  1. checksums.yaml +7 -0
  2. data/README +128 -0
  3. data/lib/stats_package_syntax_file_generator.rb +19 -0
  4. data/lib/stats_package_syntax_file_generator/controller.rb +323 -0
  5. data/lib/stats_package_syntax_file_generator/maker.rb +126 -0
  6. data/lib/stats_package_syntax_file_generator/maker_sas.rb +306 -0
  7. data/lib/stats_package_syntax_file_generator/maker_spss.rb +194 -0
  8. data/lib/stats_package_syntax_file_generator/maker_stata.rb +300 -0
  9. data/lib/stats_package_syntax_file_generator/maker_sts.rb +181 -0
  10. data/lib/stats_package_syntax_file_generator/value.rb +29 -0
  11. data/lib/stats_package_syntax_file_generator/variable.rb +56 -0
  12. data/tests/input_all_vars.yaml +2012 -0
  13. data/tests/input_controller.yaml +13 -0
  14. data/tests/setup.rb +103 -0
  15. data/tests/tc_controller.rb +378 -0
  16. data/tests/tc_maker.rb +172 -0
  17. data/tests/tc_maker_sas.rb +251 -0
  18. data/tests/tc_maker_spss.rb +121 -0
  19. data/tests/tc_maker_stata.rb +224 -0
  20. data/tests/tc_maker_sts.rb +190 -0
  21. data/tests/tc_value.rb +23 -0
  22. data/tests/tc_variable.rb +53 -0
  23. data/tests/ts_all.rb +20 -0
  24. metadata +67 -0
data/tests/tc_maker_stata.rb ADDED
@@ -0,0 +1,224 @@
1
+ # This file is part of the Minnesota Population Center's stats_package_syntax_file_generator project.
2
+ # For copyright and licensing information, see the NOTICE and LICENSE files
3
+ # in this project's top-level directory, and also on-line at:
4
+ # https://github.com/mnpopcenter/stats_package_syntax_file_generator
5
+
6
+ require File.expand_path(File.join(File.dirname(__FILE__), 'setup.rb'))
7
+
8
+ module StatsPackageSyntaxFileGeneratorTest
9
+ class MakerSTATA < Test::Unit::TestCase
10
+
11
+ include StatsPackageSyntaxFileGeneratorTestSetup
12
+
13
+ def test_create_maker_stata
14
+ msg = 'Try to create an object.'
15
+ mk = new_maker('stata')
16
+ assert_instance_of StatsPackageSyntaxFileGenerator::MakerSTATA, mk, msg
17
+ end
18
+
19
+ def test_syn_infix_start
20
+ msg = 'Compare against hardcoded result.'
21
+ mk = new_maker('stata')
22
+ expected = [
23
+ 'clear',
24
+ 'quietly infix ///',
25
+ ]
26
+ actual = mk.syn_infix_start
27
+ assert_equal expected, actual, msg
28
+ end
29
+
30
+ def test_syn_infix_var_locs
31
+ msg = 'Compare against hardcoded result.'
32
+ mk = new_maker('stata')
33
+ expected = [
34
+ ' str rectype 1-1 ///',
35
+ ' long dwnum 2-7 ///',
36
+ ' byte hhnum 8-8 ///',
37
+ ' int fbig_nd 45-48 ///',
38
+ ' float canton 67-69 ///',
39
+ ' byte lit 101-101 ///',
40
+ ]
41
+ var_list = names_to_vars( mk.sfc, %w(RECTYPE DWNUM HHNUM FBIG_ND CANTON LIT) )
42
+ actual = mk.syn_infix_var_locs(var_list)
43
+ assert_equal expected, actual, msg
44
+ end
45
+
46
+ def test_syn_dfh_infix_gen
47
+ mk = new_maker('stata')
48
+
49
+ msg = 'Compare against hardcoded result: rectangularize=false.'
50
+ expected = [
51
+ 'gen _line_num = _n',
52
+ ]
53
+ mk.sfc.rectangularize = false
54
+ actual = mk.syn_dfh_infix_gen
55
+ assert_equal expected, actual, msg
56
+
57
+ msg = 'Compare against hardcoded result: rectangularize=true.'
58
+ expected = [
59
+ 'gen _line_numH = _n',
60
+ 'gen _line_numP = _n',
61
+ 'replace _line_numH = _line_numH[_n - 1] if _n > 1 & rectype != `"H"\'',
62
+ 'replace _line_numP = _line_numP[_n - 1] if _n > 1 & rectype != `"P"\'',
63
+ ]
64
+ mk.sfc.rectangularize = true
65
+ actual = mk.syn_dfh_infix_gen
66
+ assert_equal expected, actual, msg
67
+ end
68
+
69
+ def test_syn_dfh_combine_append
70
+ mk = new_maker('stata')
71
+
72
+ msg = 'Compare against hardcoded result: rectangularize=false.'
73
+ expected = [
74
+ 'use __temp_ipums_hier_H.dta',
75
+ 'append using __temp_ipums_hier_P.dta',
76
+ ]
77
+ mk.sfc.rectangularize = false
78
+ actual = mk.syn_dfh_combine_append
79
+ assert_equal expected, actual, msg
80
+
81
+ msg = 'Compare against hardcoded result: rectangularize=true.'
82
+ expected = [
83
+ 'use __temp_ipums_hier_P.dta',
84
+ 'merge m:1 _line_numH using __temp_ipums_hier_H.dta, keep(master match)',
85
+ 'drop _merge',
86
+ ]
87
+ mk.sfc.rectangularize = true
88
+ actual = mk.syn_dfh_combine_append
89
+ assert_equal expected, actual, msg
90
+ end
91
+
92
+ def test_syn_dfh_combine_save
93
+ mk = new_maker('stata')
94
+
95
+ msg = 'Compare against hardcoded result: rectangularize=false.'
96
+ expected = [
97
+ 'sort _line_num',
98
+ 'drop _line_num',
99
+ # 'save xx9999a.dta',
100
+ ]
101
+ mk.sfc.rectangularize = false
102
+ actual = mk.syn_dfh_combine_save
103
+ assert_equal expected, actual, msg
104
+
105
+ msg = 'Compare against hardcoded result: rectangularize=true.'
106
+ expected = [
107
+ 'sort _line_numH _line_numP',
108
+ 'drop _line_numH _line_numP',
109
+ # 'save xx9999a.dta',
110
+ ]
111
+ mk.sfc.rectangularize = true
112
+ actual = mk.syn_dfh_combine_save
113
+ assert_equal expected, actual, msg
114
+ end
115
+
116
+ def test_syn_dfh_combine_erase
117
+ mk = new_maker('stata')
118
+
119
+ msg = 'Compare against hardcoded result.'
120
+ expected = [
121
+ 'erase __temp_ipums_hier_H.dta',
122
+ 'erase __temp_ipums_hier_P.dta',
123
+ ]
124
+ mk.sfc.rectangularize = false
125
+ actual = mk.syn_dfh_combine_erase
126
+ assert_equal expected, actual, msg
127
+ end
128
+
129
+ def test_syn_convert_implied_decim
130
+ msg = 'Compare against hardcoded result.'
131
+ mk = new_maker('stata')
132
+ expected = [
133
+ 'replace canton = canton / 10',
134
+ 'replace resprev2 = resprev2 / 1000',
135
+ "replace bigdec = bigdec / 100000"
136
+ ]
137
+ actual = mk.syn_convert_implied_decim
138
+ assert_equal expected, actual, msg
139
+ end
140
+
141
+ def test_syn_display_format
142
+ msg = 'Compare against hardcoded result.'
143
+ mk = new_maker('stata')
144
+ expected = [
145
+ "format canton %3.1f",
146
+ "format resprev2 %4.3f",
147
+ "format bigdec %10.5f",
148
+ "format bigint %19.0f"
149
+ ]
150
+ actual = mk.syn_display_format
151
+ assert_equal expected, actual, msg
152
+ end
153
+
154
+ def test_syn_var_labs
155
+ msg = 'Compare against hardcoded result.'
156
+ mk = new_maker('stata')
157
+ expected = [
158
+ 'label var rectype `"Record type"\'',
159
+ 'label var hdfirstd `"Head not first ["dwelling-wide"] {note}"\'',
160
+ 'label var canton `"Canton "geo area""\'',
161
+ 'label var age `"Age"\'',
162
+ ]
163
+ var_list = names_to_vars( mk.sfc, %w(RECTYPE HDFIRSTD CANTON AGE) )
164
+ actual = mk.syn_var_labs(var_list)
165
+ assert_equal expected, actual, msg
166
+ end
167
+
168
+ def test_syn_val_labs_for_var
169
+ msg = 'Compare against hardcoded result.'
170
+ mk = new_maker('stata')
171
+ expected = [
172
+ [
173
+ 'label define urban_lbl 1 `"Urban"\'',
174
+ 'label define urban_lbl 2 `"Rural"\', add',
175
+ ],
176
+ [
177
+ 'label define sex_lbl 1 `"Male"\'',
178
+ 'label define sex_lbl 2 `"Female"\', add',
179
+ ],
180
+ ]
181
+ var_list = names_to_vars( mk.sfc, %w(URBAN SEX) )
182
+ var_list.each_index { |i|
183
+ actual = mk.syn_val_labs_for_var(var_list[i])
184
+ assert_equal expected[i], actual, msg
185
+ }
186
+ end
187
+
188
+ def test_temp_file_names
189
+ msg = 'Compare against hardcoded result.'
190
+ mk = new_maker('stata')
191
+ expected = [
192
+ '__temp_ipums_hier_H.dta',
193
+ '__temp_ipums_hier_P.dta',
194
+ ]
195
+ actual = mk.temp_file_names
196
+ assert_equal expected, actual, msg
197
+
198
+ msg = 'Should return [] if there are no record types.'
199
+ mk.sfc.record_types = []
200
+ actual = mk.temp_file_names
201
+ assert_equal [], actual, msg
202
+ end
203
+
204
+ def test_sort_vars
205
+ msg = 'Compare against hardcoded result.'
206
+ mk = new_maker('stata')
207
+ mk.sfc.rectangularize = true
208
+ expected = ['_line_numH', '_line_numP']
209
+ assert_equal expected, mk.sort_vars, msg
210
+
211
+ mk.sfc.rectangularize = false
212
+ expected = ['_line_num']
213
+ assert_equal expected, mk.sort_vars, msg
214
+ end
215
+
216
+ def test_rt_ne_statement
217
+ msg = 'Compare against hardcoded result.'
218
+ mk = new_maker('stata')
219
+ expected = 'rectype != `"H"\''
220
+ assert_equal expected, mk.rt_ne_statement('H'), msg
221
+ end
222
+
223
+ end
224
+ end
data/tests/tc_maker_sts.rb ADDED
@@ -0,0 +1,190 @@
1
+ # This file is part of the Minnesota Population Center's stats_package_syntax_file_generator project.
2
+ # For copyright and licensing information, see the NOTICE and LICENSE files
3
+ # in this project's top-level directory, and also on-line at:
4
+ # https://github.com/mnpopcenter/stats_package_syntax_file_generator
5
+
6
+ require File.expand_path(File.join(File.dirname(__FILE__), 'setup.rb'))
7
+
8
+ module StatsPackageSyntaxFileGeneratorTest
9
+ class MakerSTS < Test::Unit::TestCase
10
+
11
+ include StatsPackageSyntaxFileGeneratorTestSetup
12
+
13
+ def test_create_maker_sts
14
+ msg = 'Try to create an object.'
15
+ mk = new_maker('sts')
16
+ assert_instance_of StatsPackageSyntaxFileGenerator::MakerSTS, mk, msg
17
+ end
18
+
19
+ def test_syn_quoting
20
+ msg = 'Quoting empty string.'
21
+ mk = new_maker('sts')
22
+ expected = "\"\""
23
+ actual = mk.q('')
24
+ assert_equal expected, actual, msg
25
+
26
+ msg = 'Quoting string with quotes.'
27
+ expected = "\"before''this is a ''test''after\""
28
+ actual = mk.q('before"this is a "test"after')
29
+ assert_equal expected, actual, msg
30
+ end
31
+
32
+ def test_syn_var_fmt
33
+ msg = 'Compare against hardcoded result.'
34
+ mk = new_maker('sts')
35
+ expected = [
36
+ ' (A)',
37
+ ' (F3.1)',
38
+ '',
39
+ ' (F4.3)',
40
+ ]
41
+ var_list = names_to_vars( mk.sfc, %w(RECTYPE CANTON URBAN RESPREV2) )
42
+ actual = var_list.map { |v| mk.var_fmt(v) }
43
+ assert_equal expected, actual, msg
44
+ end
45
+
46
+ def test_syn_var_loc_with_fmt
47
+ msg = 'Compare against hardcoded result.'
48
+ mk = new_maker('sts')
49
+ expected = [
50
+ '1-1 (A)',
51
+ '67 (F3.1)',
52
+ '80-80',
53
+ '80 (F4.3)',
54
+ ]
55
+ var_list = names_to_vars( mk.sfc, %w(RECTYPE CANTON URBAN RESPREV2) )
56
+ actual = var_list.map { |v| mk.var_loc_with_fmt(v) }
57
+ assert_equal expected, actual, msg
58
+ end
59
+
60
+ def test_supported_val
61
+ msg = 'Incompat value label -- empty label'
62
+ mk = new_maker('sts')
63
+ expected = false
64
+ actual = mk.supported_val?(StatsPackageSyntaxFileGenerator::Value.new(:value => 'FOO', :label => ''))
65
+ assert_equal expected, actual, msg
66
+
67
+ msg = 'Incompat value label -- pure whitespace label'
68
+ mk = new_maker('sts')
69
+ expected = false
70
+ actual = mk.supported_val?(StatsPackageSyntaxFileGenerator::Value.new(:value => 'FOO', :label => ' '))
71
+ assert_equal expected, actual, msg
72
+
73
+ msg = 'Incompat value value -- pure whitespace value'
74
+ mk = new_maker('sts')
75
+ expected = false
76
+ actual = mk.supported_val?(StatsPackageSyntaxFileGenerator::Value.new(:value => ' ', :label => 'FOO'))
77
+ assert_equal expected, actual, msg
78
+
79
+ msg = 'Incompat value value -- blank value'
80
+ mk = new_maker('sts')
81
+ expected = false
82
+ actual = mk.supported_val?(StatsPackageSyntaxFileGenerator::Value.new(:value => '', :label => 'FOO'))
83
+ assert_equal expected, actual, msg
84
+
85
+ msg = 'Incompat value value -- nonalphanumeric value'
86
+ mk = new_maker('sts')
87
+ expected = false
88
+ actual = mk.supported_val?(StatsPackageSyntaxFileGenerator::Value.new(:value => '@#(BAR)', :label => 'FOO'))
89
+ assert_equal expected, actual, msg
90
+
91
+ msg = 'Incompat value value -- value with space'
92
+ mk = new_maker('sts')
93
+ expected = false
94
+ actual = mk.supported_val?(StatsPackageSyntaxFileGenerator::Value.new(:value => 'a thing', :label => 'FOO'))
95
+ assert_equal expected, actual, msg
96
+
97
+ msg = 'Compat value'
98
+ mk = new_maker('sts')
99
+ expected = true
100
+ actual = mk.supported_val?(StatsPackageSyntaxFileGenerator::Value.new(:value => 'BAR', :label => 'FOO'))
101
+ assert_equal expected, actual, msg
102
+ end
103
+
104
+ def test_syn_val_labs_for_var
105
+ msg = 'Compare against hardcoded result.'
106
+ mk = new_maker('sts')
107
+ expected = [
108
+ [
109
+ " \\URBAN",
110
+ " 1 \"Urban\"",
111
+ " 2 \"Rural\""
112
+ ],
113
+ [
114
+ " \\RECTYPE",
115
+ " 'H' \"Household\"",
116
+ " 'P' \"Person\""
117
+ ],
118
+ [
119
+ ],
120
+ [" \\RENT",
121
+ " '......................................40........50........60........70........80........90.......100.......110.......120.......130.......140.......150.......160.......170.......180.......190.......200.......210.......220.......230.......240.......250.......260.......270.......280.......290.......300' \"[no label]............................40........50........60........70........80........90.......100.......110.......120.......130.......140.......150.......160.......170.......180.......190.......200.......210.......220.......230.......240.......250.......260.......270.......280.......290.......300\"",
122
+ " '0050' \"[no label]\"",
123
+ " '0070' \"[no label]\"",
124
+ " '0075' \"[no label]\"",
125
+ " '0080' \"[no label]\"",
126
+ " '0085' \"[no label]\"",
127
+ " '0100' \"[no label]\"",
128
+ " '0120' \"[no label]\"",
129
+ " '0125' \"[no label]\"",
130
+ " '0130' \"[no label]\"",
131
+ " '0140' \"[no label]\"",
132
+ " '0150' \"[no label]\"",
133
+ " '0160' \"[no label]\"",
134
+ " '0175' \"[no label]\"",
135
+ " '0200' \"[no label]\"",
136
+ " '0210' \"[no label]\"",
137
+ " '0225' \"[no label]\"",
138
+ " '0240' \"[no label]\"",
139
+ " '0250' \"[no label]\"",
140
+ " '0300' \"[no label]\"",
141
+ " '0400' \"[no label]\"",
142
+ " '0425' \"[no label]\"",
143
+ " '0454' \"[no label]\"",
144
+ " '0500' \"[no label]\"",
145
+ " '0550' \"[no label]\"",
146
+ " '0700' \"[no label]\"",
147
+ " '1800' \"[no label]\"",
148
+ " '2000' \"[no label]\"",
149
+ " '8501' \"[no label]\""]
150
+ ]
151
+ var_list = names_to_vars( mk.sfc, %w(URBAN RECTYPE DWNUM RENT) )
152
+ var_list.each_index { |i|
153
+ actual = mk.syn_val_labs_for_var(var_list[i])
154
+ assert_equal expected[i], actual, msg
155
+ }
156
+ end
157
+
158
+ def test_syn_var_labs
159
+ msg = 'Compare against hardcoded result.'
160
+ mk = new_maker('sts')
161
+ expected = ["VARIABLE LABELS",
162
+ " HHNUM \"''Household number''\"",
163
+ " HDFIRSTD \"Head not first [''dwelling-wide''] {note}\"",
164
+ " CANTON \"Canton ''geo area''\"",
165
+ " SEX \".........x.........x.........x.........x.........x.........x.........x.........x.........x.........x.........x.........x.........x.........x.........x\"",
166
+ "",
167
+ ""]
168
+ var_list = names_to_vars( mk.sfc, %w(HHNUM HDFIRSTD CANTON SEX) )
169
+ var_list[-1].label = '.........x' * 15
170
+ actual = mk.syn_var_labs(var_list)
171
+ assert_equal expected, actual, msg
172
+ end
173
+
174
+ def test_syn_var_locations
175
+ msg = 'Compare against hardcoded result.'
176
+ mk = new_maker('sts')
177
+ expected = [
178
+ "VARIABLES",
179
+ " RECTYPE 1-1 (A) \\RECTYPE",
180
+ " CANTON 67 (F3.1) ",
181
+ " AGE 69-71 ",
182
+ " LIT 101-101 \\LIT"
183
+ ]
184
+ var_list = names_to_vars( mk.sfc, %w(RECTYPE CANTON AGE LIT) )
185
+ actual = mk.syn_var_locations(var_list)
186
+ assert_equal expected, actual, msg
187
+ end
188
+
189
+ end
190
+ end
data/tests/tc_value.rb ADDED
@@ -0,0 +1,23 @@
1
+ # This file is part of the Minnesota Population Center's stats_package_syntax_file_generator project.
2
+ # For copyright and licensing information, see the NOTICE and LICENSE files
3
+ # in this project's top-level directory, and also on-line at:
4
+ # https://github.com/mnpopcenter/stats_package_syntax_file_generator
5
+
6
+ require File.expand_path(File.join(File.dirname(__FILE__), 'setup.rb'))
7
+
8
+ module StatsPackageSyntaxFileGeneratorTest
9
+ class Value < Test::Unit::TestCase
10
+
11
+ include StatsPackageSyntaxFileGeneratorTestSetup
12
+
13
+ def test_create_value
14
+ msg = 'Try to create an object.'
15
+ v = StatsPackageSyntaxFileGenerator::Value.new(:value => 1234)
16
+ assert_instance_of StatsPackageSyntaxFileGenerator::Value, v, msg
17
+
18
+ msg = 'Try to create an object with required parameters missing.'
19
+ assert_raise(ArgumentError, msg) { StatsPackageSyntaxFileGenerator::Value.new }
20
+ end
21
+
22
+ end
23
+ end
data/tests/tc_variable.rb ADDED
@@ -0,0 +1,53 @@
1
+ # This file is part of the Minnesota Population Center's stats_package_syntax_file_generator project.
2
+ # For copyright and licensing information, see the NOTICE and LICENSE files
3
+ # in this project's top-level directory, and also on-line at:
4
+ # https://github.com/mnpopcenter/stats_package_syntax_file_generator
5
+
6
+ require File.expand_path(File.join(File.dirname(__FILE__), 'setup.rb'))
7
+
8
+ module StatsPackageSyntaxFileGeneratorTest
9
+ class Variable < Test::Unit::TestCase
10
+
11
+ include StatsPackageSyntaxFileGeneratorTestSetup
12
+
13
+ def test_create_variable
14
+ msg = 'Try to create an object.'
15
+ v = new_variable()
16
+ assert_instance_of StatsPackageSyntaxFileGenerator::Variable, v, msg
17
+
18
+ msg = 'Try to create an object with required parameters missing.'
19
+ assert_raise(ArgumentError, msg) { StatsPackageSyntaxFileGenerator::Variable.new }
20
+ end
21
+
22
+ def test_column_locations_as_s
23
+ msg = 'Compare against hardcoded result.'
24
+ v = new_variable()
25
+ e = params_variable_lookup(:column_locations_as_s)
26
+ assert_equal e, v.column_locations_as_s, msg
27
+ end
28
+
29
+ def test_end_column
30
+ msg = 'Compare against hardcoded result.'
31
+ v = new_variable()
32
+ e = params_variable_lookup(:end_column)
33
+ assert_equal e, v.end_column, msg
34
+ end
35
+
36
+ def test_add_and_clear_values
37
+ v = new_variable()
38
+
39
+ msg = 'New variable should have no values.'
40
+ assert_equal 0, v.values.size, msg
41
+
42
+ msg = 'Adding a value N times should yield a variable with N values.'
43
+ n = params_values().size
44
+ add_new_values_to_var(v)
45
+ assert_equal n, v.values.size, msg
46
+
47
+ msg = 'After clearing its values, a variable should have no values.'
48
+ v.clear_values
49
+ assert_equal 0, v.values.size, msg
50
+ end
51
+
52
+ end
53
+ end