factbook 2.0.0 → 2.0.1
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.
- checksums.yaml +4 -4
- data/CHANGELOG.md +1 -1
- data/Manifest.txt +0 -61
- data/README.md +8 -506
- data/Rakefile +4 -9
- data/lib/factbook.rb +4 -64
- metadata +6 -124
- data/data/attributes.yml +0 -337
- data/data/categories.csv +0 -164
- data/data/codes.csv +0 -262
- data/data/codesxref.csv +0 -280
- data/data/comparisons.csv +0 -75
- data/lib/factbook/almanac.rb +0 -72
- data/lib/factbook/attributes.rb +0 -74
- data/lib/factbook/builder.rb +0 -212
- data/lib/factbook/builder_item.rb +0 -126
- data/lib/factbook/builder_json.rb +0 -79
- data/lib/factbook/codes.rb +0 -119
- data/lib/factbook/comparisons.rb +0 -50
- data/lib/factbook/counter.rb +0 -48
- data/lib/factbook/db/importer.rb +0 -92
- data/lib/factbook/db/models.rb +0 -11
- data/lib/factbook/db/schema.rb +0 -36
- data/lib/factbook/normalize.rb +0 -43
- data/lib/factbook/page.rb +0 -148
- data/lib/factbook/page_info.rb +0 -12
- data/lib/factbook/reader_json.rb +0 -51
- data/lib/factbook/sanitizer.rb +0 -178
- data/lib/factbook/sect.rb +0 -29
- data/lib/factbook/subsect.rb +0 -18
- data/lib/factbook/table.rb +0 -52
- data/lib/factbook/utils.rb +0 -85
- data/lib/factbook/utils_info.rb +0 -129
- data/lib/factbook/version.rb +0 -21
- data/script/almanac.rb +0 -48
- data/script/attributes.rb +0 -34
- data/script/build.rb +0 -28
- data/script/counter.rb +0 -145
- data/script/json.rb +0 -19
- data/script/testbr.rb +0 -33
- data/script/testcodes.rb +0 -11
- data/test/data/au.html +0 -579
- data/test/data/au.yml +0 -8
- data/test/data/be.html +0 -596
- data/test/data/be.yml +0 -8
- data/test/data/json/au.json +0 -892
- data/test/data/src/ag.html +0 -716
- data/test/data/src/au-2015-09-24.html +0 -2006
- data/test/data/src/au.html +0 -658
- data/test/data/src/be-2015-09-24.html +0 -2011
- data/test/data/src/be.html +0 -648
- data/test/helper.rb +0 -11
- data/test/test_attribs.rb +0 -87
- data/test/test_attribs_def.rb +0 -20
- data/test/test_builder.rb +0 -35
- data/test/test_codes.rb +0 -76
- data/test/test_comparisons.rb +0 -19
- data/test/test_convert.rb +0 -30
- data/test/test_counter.rb +0 -31
- data/test/test_fields.rb +0 -52
- data/test/test_importer.rb +0 -56
- data/test/test_item_builder.rb +0 -99
- data/test/test_json.rb +0 -45
- data/test/test_json_builder.rb +0 -25
- data/test/test_normalize.rb +0 -23
- data/test/test_page.rb +0 -38
- data/test/test_sanitizer.rb +0 -39
- data/test/test_sanitizer_regex.rb +0 -89
data/test/test_json.rb
DELETED
@@ -1,45 +0,0 @@
|
|
1
|
-
# encoding: utf-8
|
2
|
-
|
3
|
-
###
|
4
|
-
# to run use
|
5
|
-
# ruby -I ./lib -I ./test test/test_json.rb
|
6
|
-
|
7
|
-
|
8
|
-
require 'helper'
|
9
|
-
|
10
|
-
|
11
|
-
class TestJson < MiniTest::Test
|
12
|
-
|
13
|
-
|
14
|
-
def test_json
|
15
|
-
Dir.mkdir( 'tmp' ) unless Dir.exists?( 'tmp' )
|
16
|
-
|
17
|
-
codes = [ 'au',
|
18
|
-
'ag',
|
19
|
-
'be',
|
20
|
-
#'br',
|
21
|
-
#'mx',
|
22
|
-
#'ls',
|
23
|
-
#'vt',
|
24
|
-
#'ee',
|
25
|
-
#'xx'
|
26
|
-
]
|
27
|
-
|
28
|
-
codes.each do |code|
|
29
|
-
|
30
|
-
html = File.read( "#{Factbook.root}/test/data/src/#{code}.html" )
|
31
|
-
page = Factbook::Page.new( code, html: html )
|
32
|
-
|
33
|
-
h = page.data
|
34
|
-
pp h
|
35
|
-
|
36
|
-
### save to json
|
37
|
-
puts "saving a copy to #{code}.json for debugging"
|
38
|
-
File.open( "tmp/#{code}.json", 'w' ) do |f|
|
39
|
-
f.write JSON.pretty_generate( h )
|
40
|
-
## f.write page.to_json
|
41
|
-
end
|
42
|
-
end
|
43
|
-
end
|
44
|
-
|
45
|
-
end # class TestJson
|
data/test/test_json_builder.rb
DELETED
@@ -1,25 +0,0 @@
|
|
1
|
-
# encoding: utf-8
|
2
|
-
|
3
|
-
###
|
4
|
-
# to run use
|
5
|
-
# ruby -I ./lib -I ./test test/test_json_builder.rb
|
6
|
-
|
7
|
-
|
8
|
-
require 'helper'
|
9
|
-
|
10
|
-
|
11
|
-
class TestJsonBuilder < MiniTest::Test
|
12
|
-
|
13
|
-
def test_read
|
14
|
-
code = 'au'
|
15
|
-
b = Factbook::JsonBuilder.from_file( "#{Factbook.root}/test/data/json/#{code}.json" )
|
16
|
-
|
17
|
-
assert_equal 10, b.sects.size
|
18
|
-
assert_equal 1, b.sects[0].subsects.size ## e.g. Introduction/Background
|
19
|
-
assert_equal 'Central Europe, north of Italy and Slovenia', b.json['Geography']['Location']['text']
|
20
|
-
|
21
|
-
assert true ## assume everthing ok
|
22
|
-
end
|
23
|
-
|
24
|
-
end # class TestJsonBuilder
|
25
|
-
|
data/test/test_normalize.rb
DELETED
@@ -1,23 +0,0 @@
|
|
1
|
-
# encoding: utf-8
|
2
|
-
|
3
|
-
###
|
4
|
-
# to run use
|
5
|
-
# ruby -I ./lib -I ./test test/test_normalize.rb
|
6
|
-
|
7
|
-
|
8
|
-
require 'helper'
|
9
|
-
|
10
|
-
|
11
|
-
class TestNormalizer < MiniTest::Test
|
12
|
-
|
13
|
-
include Factbook::NormalizeHelper
|
14
|
-
|
15
|
-
def test_normalize
|
16
|
-
assert_equal 'border countries', normalize_category( 'border countries:' )
|
17
|
-
assert_equal 'border countries', normalize_category( 'border countries: ' )
|
18
|
-
assert_equal 'border countries', normalize_category( 'border countries (8):' )
|
19
|
-
assert_equal 'border countries', normalize_category( 'border countries (10): ' )
|
20
|
-
end
|
21
|
-
|
22
|
-
end # class TestNormalizer
|
23
|
-
|
data/test/test_page.rb
DELETED
@@ -1,38 +0,0 @@
|
|
1
|
-
# encoding: utf-8
|
2
|
-
|
3
|
-
###
|
4
|
-
# to run use
|
5
|
-
# ruby -I ./lib -I ./test test/test_page.rb
|
6
|
-
|
7
|
-
|
8
|
-
require 'helper'
|
9
|
-
|
10
|
-
|
11
|
-
class TestPage < MiniTest::Test
|
12
|
-
|
13
|
-
|
14
|
-
def test_sects
|
15
|
-
pages = [
|
16
|
-
[ 'au', 10 ],
|
17
|
-
[ 'be', 10 ],
|
18
|
-
# [ 'br', 10 ],
|
19
|
-
# [ 'ee', 10 ],
|
20
|
-
# [ 'mx', 10 ],
|
21
|
-
# [ 'xx', 10 ],
|
22
|
-
# [ 'ls', 9 ],
|
23
|
-
# [ 'vt', 8 ],
|
24
|
-
]
|
25
|
-
|
26
|
-
pages.each do |rec|
|
27
|
-
code = rec[0]
|
28
|
-
sects_size = rec[1]
|
29
|
-
|
30
|
-
html = File.read( "#{Factbook.root}/test/data/src/#{code}.html" )
|
31
|
-
page = Factbook::Page.new( code, html: html )
|
32
|
-
|
33
|
-
assert_equal sects_size, page.sects.size
|
34
|
-
end
|
35
|
-
end
|
36
|
-
|
37
|
-
|
38
|
-
end # class TestPage
|
data/test/test_sanitizer.rb
DELETED
@@ -1,39 +0,0 @@
|
|
1
|
-
# encoding: utf-8
|
2
|
-
|
3
|
-
###
|
4
|
-
# to run use
|
5
|
-
# ruby -I ./lib -I ./test test/test_sanitizer.rb
|
6
|
-
|
7
|
-
|
8
|
-
require 'helper'
|
9
|
-
|
10
|
-
|
11
|
-
class TestSanitizer < MiniTest::Test
|
12
|
-
|
13
|
-
def test_sanitize
|
14
|
-
|
15
|
-
## austria (au)
|
16
|
-
## algeria (ag)
|
17
|
-
## belgium (be)
|
18
|
-
## ['au'].each do |cnty|
|
19
|
-
['au','ag','be'].each do |cnty|
|
20
|
-
|
21
|
-
## use/fix: ASCII-8BIT (e.g.keep as is) -???
|
22
|
-
html_ascii = File.read( "#{Factbook.root}/test/data/src/#{cnty}.html" ) ## fix/todo: use ASCII8BIT/binary reader ??
|
23
|
-
|
24
|
-
html, info, errors = Factbook::Sanitizer.new.sanitize( html_ascii )
|
25
|
-
|
26
|
-
File.open( "./tmp/#{cnty}.profile.html", 'w' ) do |f|
|
27
|
-
f.write "** info:\n"
|
28
|
-
f.write info.inspect + "\n\n"
|
29
|
-
f.write "** errors:\n"
|
30
|
-
f.write errors.inspect + "\n\n"
|
31
|
-
f.write "** html:\n"
|
32
|
-
f.write html
|
33
|
-
end
|
34
|
-
end
|
35
|
-
|
36
|
-
assert true ## assume everthing ok
|
37
|
-
end
|
38
|
-
|
39
|
-
end # class TestSanitizer
|
@@ -1,89 +0,0 @@
|
|
1
|
-
# encoding: utf-8
|
2
|
-
|
3
|
-
###
|
4
|
-
# to run use
|
5
|
-
# ruby -I ./lib -I ./test test/test_sanitizer_regex.rb
|
6
|
-
|
7
|
-
|
8
|
-
require 'helper'
|
9
|
-
|
10
|
-
|
11
|
-
class TestSanitizerRegex < MiniTest::Test
|
12
|
-
|
13
|
-
def test_area_map
|
14
|
-
|
15
|
-
html =<<HTML
|
16
|
-
<div class='disTable areaComp'>
|
17
|
-
<span class='category tCell' style='margin-bottom:0px; vertical-align:bottom;'>Area comparison map:</span>
|
18
|
-
<span class="tCell"><a data-toggle="modal" href="#areaCompModal"><img src="../graphics/areacomparison_icon.jpg" border="0" style="cursor:pointer; border: 0px solid #CCC;"></a></span></div>
|
19
|
-
|
20
|
-
<div class="modal fade" id="areaCompModal" role="dialog">
|
21
|
-
<div class="wfb-modal-dialog">
|
22
|
-
<div class="modal-content" >
|
23
|
-
<div class="wfb-modal-header" style="border-radius: 4px; font-family: Verdana,Arial,sans-serif; font-size: 14px !important; font-weight: bold; padding: 0.4em 16px 0.4em 1em; background: #cccccc url("..images/ui-bg_highlight-soft_75_cccccc_1x100.png") repeat-x scroll 50% 50%;" >
|
24
|
-
<span style="font-size: 14px !important; margin: 0.1em 16px 0.1em 0;" class="modal-title wfb-title">The World Factbook</span><span style="float: right; margin-top: -4px;">
|
25
|
-
<button type="button" class="close" title="close" data-dismiss="modal">×</button></span>
|
26
|
-
</div>
|
27
|
-
<div class="wfb-modal-body">
|
28
|
-
...
|
29
|
-
<div id='field'
|
30
|
-
HTML
|
31
|
-
|
32
|
-
m = Factbook::Sanitizer::AREA_COMP_CATEGORY_REGEX.match( html )
|
33
|
-
pp m
|
34
|
-
|
35
|
-
assert m.nil? == false
|
36
|
-
end
|
37
|
-
|
38
|
-
|
39
|
-
def test_pop_pyramid
|
40
|
-
|
41
|
-
html =<<HTML
|
42
|
-
<div class='disTable popPyramid'>
|
43
|
-
<span class='category tCell' style='margin-bottom:0px; vertical-align:bottom;'>population pyramid:</span>
|
44
|
-
<span class="tCell"><a data-toggle="modal" href="#popPyramidModal"><img title="" src="../graphics/poppyramid_icon.jpg" style="cursor:pointer; border: 0px solid #CCC;"></span></a></div>
|
45
|
-
|
46
|
-
<div class="modal fade" id="popPyramidModal" role="dialog">
|
47
|
-
<div class="wfb-modal-dialog">
|
48
|
-
<div class="modal-content" >
|
49
|
-
<div class="wfb-modal-header" style="border-radius: 4px; font-family: Verdana,Arial,sans-serif; font-size: 14px !important; font-weight: bold; padding: 0.4em 16px 0.4em 1em; background: #cccccc url("..images/ui-bg_highlight-soft_75_cccccc_1x100.png") repeat-x scroll 50% 50%;" >
|
50
|
-
<span style="font-size: 14px !important; margin: 0.1em 16px 0.1em 0;" class="modal-title wfb-title">The World Factbook</span><span style="float: right; margin-top: -4px;">
|
51
|
-
<button type="button" class="close" title="close" data-dismiss="modal">×</button></span>
|
52
|
-
</div>
|
53
|
-
<div class="wfb-modal-body">
|
54
|
-
...
|
55
|
-
<div id='field'
|
56
|
-
HTML
|
57
|
-
|
58
|
-
m = Factbook::Sanitizer::POP_PYRAMID_CATEGORY_REGEX.match( html )
|
59
|
-
pp m
|
60
|
-
|
61
|
-
assert m.nil? == false
|
62
|
-
end # method test_pop_pyramid
|
63
|
-
|
64
|
-
|
65
|
-
def test_rel_affiliation
|
66
|
-
|
67
|
-
html =<<HTML
|
68
|
-
<div class='disTable relAffiliation'><span class='category tCell' style='margin-bottom:0px; vertical-align:bottom;'>religious affiliation:</span>
|
69
|
-
<span class="tCell"><a data-toggle="modal" href="#relAffiliationModal"><img title="" src="../graphics/middle-east-religion-icon.jpg" style="cursor:pointer; border: 0px solid #CCC;"></span></a></div>
|
70
|
-
|
71
|
-
<div class="modal fade" id="relAffiliationModal" role="dialog">
|
72
|
-
<div class="wfb-modal-dialog">
|
73
|
-
<div class="modal-content" >
|
74
|
-
<div class="wfb-modal-header" style="border-radius: 4px; font-family: Verdana,Arial,sans-serif; font-size: 14px !important; font-weight: bold; padding: 0.4em 16px 0.4em 1em; background: #cccccc url("..images/ui-bg_highlight-soft_75_cccccc_1x100.png") repeat-x scroll 50% 50%;" >
|
75
|
-
<span style="font-size: 14px !important; margin: 0.1em 16px 0.1em 0;" class="modal-title wfb-title">The World Factbook</span><span style="float: right; margin-top: -4px;">
|
76
|
-
<button type="button" class="close" title="close" data-dismiss="modal">×</button></span>
|
77
|
-
</div>
|
78
|
-
<div class="wfb-modal-body">
|
79
|
-
...
|
80
|
-
<div id='field'
|
81
|
-
HTML
|
82
|
-
|
83
|
-
m = Factbook::Sanitizer::REL_AFFILIATION_CATEGORY_REGEX.match( html )
|
84
|
-
pp m
|
85
|
-
|
86
|
-
assert m.nil? == false
|
87
|
-
end # method test_rel_affiliation
|
88
|
-
|
89
|
-
end # class TestSanitizerRegex
|