remote_table 0.2.32 → 1.0.0
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +5 -0
- data/Gemfile +4 -0
- data/Gemfile.lock +65 -0
- data/LICENSE +1 -1
- data/README.rdoc +21 -7
- data/Rakefile +12 -61
- data/lib/remote_table/cleaner.rb +19 -0
- data/lib/remote_table/executor.rb +29 -0
- data/lib/remote_table/format/delimited.rb +62 -0
- data/lib/remote_table/format/excel.rb +10 -0
- data/lib/remote_table/format/excelx.rb +10 -0
- data/lib/remote_table/format/fixed_width.rb +47 -0
- data/lib/remote_table/format/html.rb +43 -0
- data/lib/remote_table/format/mixins/rooable.rb +47 -0
- data/lib/remote_table/format/mixins/textual.rb +34 -0
- data/lib/remote_table/format/open_office.rb +10 -0
- data/lib/remote_table/format.rb +35 -0
- data/lib/remote_table/hasher.rb +25 -0
- data/lib/remote_table/local_file.rb +92 -0
- data/lib/remote_table/properties.rb +209 -0
- data/lib/remote_table/transformer.rb +17 -0
- data/lib/remote_table/version.rb +3 -0
- data/lib/remote_table.rb +91 -99
- data/remote_table.gemspec +32 -77
- data/test/{test_helper.rb → helper.rb} +9 -2
- data/test/test_big.rb +61 -0
- data/test/test_errata.rb +46 -0
- data/test/test_old_syntax.rb +229 -0
- data/test/test_old_transform.rb +49 -0
- data/test/test_remote_table.rb +13 -0
- metadata +176 -53
- data/VERSION +0 -1
- data/lib/remote_table/file/csv.rb +0 -49
- data/lib/remote_table/file/fixed_width.rb +0 -19
- data/lib/remote_table/file/html.rb +0 -37
- data/lib/remote_table/file/ods.rb +0 -11
- data/lib/remote_table/file/roo_spreadsheet.rb +0 -44
- data/lib/remote_table/file/xls.rb +0 -11
- data/lib/remote_table/file/xlsx.rb +0 -11
- data/lib/remote_table/file.rb +0 -100
- data/lib/remote_table/package.rb +0 -89
- data/lib/remote_table/request.rb +0 -44
- data/lib/remote_table/transform.rb +0 -58
- data/test/remote_table_test.rb +0 -386
data/lib/remote_table/package.rb
DELETED
@@ -1,89 +0,0 @@
|
|
1
|
-
class RemoteTable
|
2
|
-
class Package
|
3
|
-
attr_accessor :url, :compression, :packing, :filename, :glob
|
4
|
-
|
5
|
-
def initialize(bus)
|
6
|
-
@url = bus[:url] or raise "need url"
|
7
|
-
@compression = bus[:compression] || compression_from_basename
|
8
|
-
@packing = bus[:packing] || packing_from_basename_and_compression
|
9
|
-
@filename = bus[:filename] || filename_from_basename_and_compression_and_packing
|
10
|
-
@glob = bus[:glob]
|
11
|
-
add_hints!(bus)
|
12
|
-
end
|
13
|
-
|
14
|
-
def add_hints!(hash)
|
15
|
-
hash[:filename] = filename unless hash.has_key?(:filename)
|
16
|
-
end
|
17
|
-
|
18
|
-
def stage(path)
|
19
|
-
decompress(path)
|
20
|
-
unpack(path)
|
21
|
-
identify(path)
|
22
|
-
file_path(path)
|
23
|
-
end
|
24
|
-
|
25
|
-
private
|
26
|
-
|
27
|
-
def decompress(path)
|
28
|
-
return unless compression
|
29
|
-
cmd, args = case compression
|
30
|
-
when :zip, :exe
|
31
|
-
["unzip", "-d #{Escape.shell_single_word ::File.dirname(path)}"]
|
32
|
-
when :bz2
|
33
|
-
'bunzip2'
|
34
|
-
when :gz
|
35
|
-
'gunzip'
|
36
|
-
end
|
37
|
-
move_and_process path, compression, cmd, args
|
38
|
-
end
|
39
|
-
|
40
|
-
def unpack(path)
|
41
|
-
return unless packing
|
42
|
-
cmd, args = case packing
|
43
|
-
when :tar
|
44
|
-
['tar -xf', "-C #{::File.dirname(path)}"]
|
45
|
-
end
|
46
|
-
move_and_process path, packing, cmd, args
|
47
|
-
end
|
48
|
-
|
49
|
-
def move_and_process(path, extname, cmd, args)
|
50
|
-
new_path = "#{path}.#{extname}"
|
51
|
-
FileUtils.mv path, new_path
|
52
|
-
RemoteTable.backtick_with_reporting "#{cmd} #{Escape.shell_single_word new_path} #{args}"
|
53
|
-
end
|
54
|
-
|
55
|
-
# ex. A: 2007-01.csv.gz (compression not capable of storing multiple files)
|
56
|
-
# ex. B: 2007-01.tar.gz (packing)
|
57
|
-
# ex. C: 2007-01.zip (compression capable of storing multiple files)
|
58
|
-
def identify(path)
|
59
|
-
if glob.present?
|
60
|
-
FileUtils.mv Dir[::File.dirname(path) + glob].first, file_path(path)
|
61
|
-
elsif !packing and [ nil, :bz2, :gz ].include?(compression)
|
62
|
-
FileUtils.mv path, file_path(path)
|
63
|
-
end
|
64
|
-
end
|
65
|
-
|
66
|
-
def file_path(path)
|
67
|
-
::File.join(::File.dirname(path), filename)
|
68
|
-
end
|
69
|
-
|
70
|
-
def basename_parts
|
71
|
-
::File.basename(URI.parse(url).path).split('.').map(&:to_sym)
|
72
|
-
end
|
73
|
-
|
74
|
-
def compression_from_basename
|
75
|
-
[ :zip, :exe, :bz2, :gz ].detect { |i| i == basename_parts.last }
|
76
|
-
end
|
77
|
-
|
78
|
-
def packing_from_basename_and_compression
|
79
|
-
[ :tar ].detect { |i| i == ((basename_parts.last == compression) ? basename_parts[-2] : basename_parts.last) }
|
80
|
-
end
|
81
|
-
|
82
|
-
def filename_from_basename_and_compression_and_packing
|
83
|
-
ary = basename_parts
|
84
|
-
ary.pop if ary.last == compression
|
85
|
-
ary.pop if ary.last == packing
|
86
|
-
ary.join('.')
|
87
|
-
end
|
88
|
-
end
|
89
|
-
end
|
data/lib/remote_table/request.rb
DELETED
@@ -1,44 +0,0 @@
|
|
1
|
-
class RemoteTable
|
2
|
-
class Request
|
3
|
-
attr_accessor :parsed_url, :post_data, :username, :password
|
4
|
-
attr_accessor :form_data
|
5
|
-
|
6
|
-
# TODO: support HTTP basic auth
|
7
|
-
def initialize(bus)
|
8
|
-
raise(ArgumentError, "RemoteTable needs :url option") unless bus[:url].present?
|
9
|
-
@parsed_url = URI.parse bus[:url]
|
10
|
-
if @parsed_url.host == 'spreadsheets.google.com'
|
11
|
-
if bus[:format].blank? or bus[:format].to_s == 'csv'
|
12
|
-
@parsed_url.query = 'output=csv&' + @parsed_url.query.sub(/\&?output=.*?(\&|\z)/, '\1')
|
13
|
-
end
|
14
|
-
end
|
15
|
-
@form_data = bus[:form_data]
|
16
|
-
end
|
17
|
-
|
18
|
-
def download
|
19
|
-
path = ::File.join staging_dir_path, 'REMOTE_TABLE_PACKAGE'
|
20
|
-
if parsed_url.scheme == 'file'
|
21
|
-
::FileUtils.cp parsed_url.path, path
|
22
|
-
else
|
23
|
-
RemoteTable.backtick_with_reporting %{
|
24
|
-
curl
|
25
|
-
--header "Expect: "
|
26
|
-
--location
|
27
|
-
#{"--data #{Escape.shell_single_word form_data}" if form_data.present?}
|
28
|
-
#{Escape.shell_single_word parsed_url.to_s}
|
29
|
-
--output #{Escape.shell_single_word path}
|
30
|
-
2>&1
|
31
|
-
}
|
32
|
-
end
|
33
|
-
path
|
34
|
-
end
|
35
|
-
|
36
|
-
def staging_dir_path
|
37
|
-
return @_staging_dir_path if @_staging_dir_path
|
38
|
-
@_staging_dir_path = ::File.join Dir.tmpdir, 'remote_table_gem', rand.to_s
|
39
|
-
FileUtils.mkdir_p @_staging_dir_path
|
40
|
-
RemoteTable.remove_at_exit @_staging_dir_path
|
41
|
-
@_staging_dir_path
|
42
|
-
end
|
43
|
-
end
|
44
|
-
end
|
@@ -1,58 +0,0 @@
|
|
1
|
-
class RemoteTable
|
2
|
-
class Transform
|
3
|
-
attr_accessor :select, :reject, :transform_class, :transform_options, :transform, :raw_table
|
4
|
-
attr_accessor :errata
|
5
|
-
|
6
|
-
def initialize(bus)
|
7
|
-
if transform_params = bus.delete(:transform)
|
8
|
-
@transform_class = transform_params.delete(:class)
|
9
|
-
@transform_options = transform_params
|
10
|
-
@transform = @transform_class.new(@transform_options)
|
11
|
-
@transform.add_hints!(bus)
|
12
|
-
end
|
13
|
-
@select = bus[:select]
|
14
|
-
@reject = bus[:reject]
|
15
|
-
@errata = bus[:errata]
|
16
|
-
end
|
17
|
-
|
18
|
-
# the null transformation
|
19
|
-
def apply(raw_table)
|
20
|
-
self.raw_table = raw_table
|
21
|
-
self
|
22
|
-
end
|
23
|
-
|
24
|
-
# - convert it to a plain hash for whatever ruby version you're on
|
25
|
-
# - dump it
|
26
|
-
# - digest it
|
27
|
-
def self.row_hash(row)
|
28
|
-
plain_hsh = if RUBY_VERSION >= '1.9'
|
29
|
-
row.keys.sort.inject(::Hash.new) do |memo, key|
|
30
|
-
value = row[key]
|
31
|
-
key = key.to_s.toutf8
|
32
|
-
value = value.to_s.toutf8 if value.respond_to? :to_s
|
33
|
-
memo[key] = value
|
34
|
-
memo
|
35
|
-
end
|
36
|
-
else
|
37
|
-
::Hash.new.replace(row)
|
38
|
-
end
|
39
|
-
::Digest::MD5.hexdigest ::Marshal.dump(plain_hsh)
|
40
|
-
end
|
41
|
-
|
42
|
-
def each_row(&block)
|
43
|
-
raw_table.each_row do |row|
|
44
|
-
row['row_hash'] = self.class.row_hash(row)
|
45
|
-
virtual_rows = transform ? transform.apply(row) : row # allow transform.apply(row) to return multiple rows
|
46
|
-
Array.wrap(virtual_rows).each do |virtual_row|
|
47
|
-
if errata
|
48
|
-
next if errata.rejects? virtual_row
|
49
|
-
errata.correct! virtual_row
|
50
|
-
end
|
51
|
-
next if select and !select.call(virtual_row)
|
52
|
-
next if reject and reject.call(virtual_row)
|
53
|
-
yield virtual_row
|
54
|
-
end
|
55
|
-
end
|
56
|
-
end
|
57
|
-
end
|
58
|
-
end
|
data/test/remote_table_test.rb
DELETED
@@ -1,386 +0,0 @@
|
|
1
|
-
require 'test_helper'
|
2
|
-
|
3
|
-
class FuelOilParser
|
4
|
-
def initialize(options = {})
|
5
|
-
# nothing
|
6
|
-
end
|
7
|
-
def add_hints!(bus)
|
8
|
-
bus[:sheet] = 'Data 1'
|
9
|
-
bus[:skip] = 2
|
10
|
-
bus[:select] = lambda { |row| row['year'] > 1989 }
|
11
|
-
end
|
12
|
-
def apply(row)
|
13
|
-
virtual_rows = []
|
14
|
-
row.keys.grep(/(.+) Residual Fuel Oil/) do |location_column_name|
|
15
|
-
first_part = $1
|
16
|
-
next if (cost = row[location_column_name]).blank? or (date = row['Date']).blank?
|
17
|
-
if first_part.start_with?('U.S.')
|
18
|
-
locatable = "united_states (Country)"
|
19
|
-
elsif first_part.include?('PADD')
|
20
|
-
/\(PADD (.*)\)/.match(first_part)
|
21
|
-
padd_part = $1
|
22
|
-
next if padd_part == '1' # skip PADD 1 because we always prefer subdistricts
|
23
|
-
locatable = "#{padd_part} (PetroleumAdministrationForDefenseDistrict)"
|
24
|
-
else
|
25
|
-
locatable = "#{first_part} (State)"
|
26
|
-
end
|
27
|
-
date = Time.parse(date)
|
28
|
-
virtual_rows << {
|
29
|
-
'locatable' => locatable,
|
30
|
-
'cost' => cost,
|
31
|
-
'year' => date.year,
|
32
|
-
'month' => date.month
|
33
|
-
}
|
34
|
-
end
|
35
|
-
virtual_rows
|
36
|
-
end
|
37
|
-
end
|
38
|
-
|
39
|
-
class AircraftGuru
|
40
|
-
def is_a_dc_plane?(row)
|
41
|
-
row['Designator'] =~ /^DC\d/i
|
42
|
-
end
|
43
|
-
|
44
|
-
# def is_a_crj_900?(row)
|
45
|
-
# row['Designator'].downcase == 'crj9'
|
46
|
-
# end
|
47
|
-
|
48
|
-
def is_a_g159?(row)
|
49
|
-
row['Designator'] =~ /^G159$/
|
50
|
-
end
|
51
|
-
|
52
|
-
def is_a_galx?(row)
|
53
|
-
row['Designator'] =~ /^GALX$/
|
54
|
-
end
|
55
|
-
|
56
|
-
def method_missing(method_id, *args, &block)
|
57
|
-
if method_id.to_s =~ /\Ais_n?o?t?_?attributed_to_([^\?]+)/
|
58
|
-
manufacturer_name = $1
|
59
|
-
manufacturer_regexp = Regexp.new(manufacturer_name.gsub('_', ' ?'), Regexp::IGNORECASE)
|
60
|
-
matches = manufacturer_regexp.match(args.first['Manufacturer']) # row['Manufacturer'] =~ /mcdonnell douglas/i
|
61
|
-
method_id.to_s.include?('not_attributed') ? matches.nil? : !matches.nil?
|
62
|
-
else
|
63
|
-
super
|
64
|
-
end
|
65
|
-
end
|
66
|
-
end
|
67
|
-
|
68
|
-
class RemoteTableTest < Test::Unit::TestCase
|
69
|
-
def setup
|
70
|
-
@test2_rows_with_blanks = [
|
71
|
-
{ 'header4' => '', 'header5' => '', 'header6' => '' },
|
72
|
-
{ 'header4' => '1 at 4', 'header5' => '1 at 5', 'header6' => '1 at 6' },
|
73
|
-
{ 'header4' => '', 'header5' => '', 'header6' => '' },
|
74
|
-
{ 'header4' => '2 at 4', 'header5' => '2 at 5', 'header6' => '2 at 6' },
|
75
|
-
]
|
76
|
-
@test2_rows = [
|
77
|
-
{ 'header4' => '1 at 4', 'header5' => '1 at 5', 'header6' => '1 at 6' },
|
78
|
-
{ 'header4' => '2 at 4', 'header5' => '2 at 5', 'header6' => '2 at 6' },
|
79
|
-
]
|
80
|
-
end
|
81
|
-
|
82
|
-
if ENV['ALL'] == 'true' or ENV['SLOW'] == 'true'
|
83
|
-
should "open an XLS inside a zip file" do
|
84
|
-
t = RemoteTable.new(:url => 'http://www.fueleconomy.gov/FEG/epadata/02data.zip', :filename => 'guide_jan28.xls')
|
85
|
-
assert_equal 'ACURA', t.rows.first['Manufacturer']
|
86
|
-
assert_equal 'NSX', t.rows.first['carline name']
|
87
|
-
assert_equal 'VOLVO', t.rows.last['Manufacturer']
|
88
|
-
assert_equal 'V70 XC AWD', t.rows.last['carline name']
|
89
|
-
end
|
90
|
-
|
91
|
-
should "not have indifferent string/symbol hash access" do
|
92
|
-
t = RemoteTable.new(:url => 'http://www.fueleconomy.gov/FEG/epadata/02data.zip', :filename => 'guide_jan28.xls')
|
93
|
-
assert_equal 'ACURA', t.rows.first['Manufacturer']
|
94
|
-
assert_equal nil, t.rows.first[:Manufacturer]
|
95
|
-
end
|
96
|
-
|
97
|
-
should "open a CSV inside a zip file" do
|
98
|
-
t = RemoteTable.new(:url => 'http://www.fueleconomy.gov/FEG/epadata/98guide6.zip', :filename => '98guide6.csv')
|
99
|
-
assert_equal 'ACURA', t.rows.first['Manufacturer']
|
100
|
-
assert_equal 'NSX', t.rows.first['carline name']
|
101
|
-
assert_equal 'TOYOTA', t.rows.last['Manufacturer']
|
102
|
-
assert_equal 'RAV4 SOFT TOP 4WD', t.rows.last['carline name']
|
103
|
-
end
|
104
|
-
|
105
|
-
should "open a fixed-width file with an inline schema inside a zip file" do
|
106
|
-
t = RemoteTable.new(:url => 'http://www.fueleconomy.gov/FEG/epadata/00data.zip',
|
107
|
-
:filename => 'Gd6-dsc.txt',
|
108
|
-
:format => :fixed_width,
|
109
|
-
:crop => 21..26, # inclusive
|
110
|
-
:cut => '2-',
|
111
|
-
:select => lambda { |row| /\A[A-Z]/.match row['code'] },
|
112
|
-
:schema => [[ 'code', 2, { :type => :string } ],
|
113
|
-
[ 'spacer', 2 ],
|
114
|
-
[ 'name', 52, { :type => :string } ]])
|
115
|
-
assert_equal 'regular grade gasoline (octane number of 87)', t.rows.first['name']
|
116
|
-
assert_equal 'R', t.rows.first['code']
|
117
|
-
assert_equal 'electricity', t.rows.last['name']
|
118
|
-
assert_equal 'El', t.rows.last['code']
|
119
|
-
end
|
120
|
-
|
121
|
-
should "send form data, follow redirects and use a filename glob" do
|
122
|
-
url = 'http://www.transtats.bts.gov/DownLoad_Table.asp?Table_ID=293&Has_Group=3&Is_Zipped=0'
|
123
|
-
form_data = 'UserTableName=T_100_Segment__All_Carriers&DBShortName=Air_Carriers&RawDataTable=T_T100_SEGMENT_ALL_CARRIER&sqlstr=+SELECT+DEPARTURES_SCHEDULED%2CDEPARTURES_PERFORMED%2CPAYLOAD%2CSEATS%2CPASSENGERS%2CFREIGHT%2CMAIL%2CDISTANCE%2CRAMP_TO_RAMP%2CAIR_TIME%2CUNIQUE_CARRIER%2CAIRLINE_ID%2CUNIQUE_CARRIER_NAME%2CUNIQUE_CARRIER_ENTITY%2CREGION%2CCARRIER%2CCARRIER_NAME%2CCARRIER_GROUP%2CCARRIER_GROUP_NEW%2CORIGIN%2CORIGIN_CITY_NAME%2CORIGIN_CITY_NUM%2CORIGIN_STATE_ABR%2CORIGIN_STATE_FIPS%2CORIGIN_STATE_NM%2CORIGIN_COUNTRY%2CORIGIN_COUNTRY_NAME%2CORIGIN_WAC%2CDEST%2CDEST_CITY_NAME%2CDEST_CITY_NUM%2CDEST_STATE_ABR%2CDEST_STATE_FIPS%2CDEST_STATE_NM%2CDEST_COUNTRY%2CDEST_COUNTRY_NAME%2CDEST_WAC%2CAIRCRAFT_GROUP%2CAIRCRAFT_TYPE%2CAIRCRAFT_CONFIG%2CYEAR%2CQUARTER%2CMONTH%2CDISTANCE_GROUP%2CCLASS%2CDATA_SOURCE+FROM++T_T100_SEGMENT_ALL_CARRIER+WHERE+Month+%3D1+AND+YEAR%3D2008&varlist=DEPARTURES_SCHEDULED%2CDEPARTURES_PERFORMED%2CPAYLOAD%2CSEATS%2CPASSENGERS%2CFREIGHT%2CMAIL%2CDISTANCE%2CRAMP_TO_RAMP%2CAIR_TIME%2CUNIQUE_CARRIER%2CAIRLINE_ID%2CUNIQUE_CARRIER_NAME%2CUNIQUE_CARRIER_ENTITY%2CREGION%2CCARRIER%2CCARRIER_NAME%2CCARRIER_GROUP%2CCARRIER_GROUP_NEW%2CORIGIN%2CORIGIN_CITY_NAME%2CORIGIN_CITY_NUM%2CORIGIN_STATE_ABR%2CORIGIN_STATE_FIPS%2CORIGIN_STATE_NM%2CORIGIN_COUNTRY%2CORIGIN_COUNTRY_NAME%2CORIGIN_WAC%2CDEST%2CDEST_CITY_NAME%2CDEST_CITY_NUM%2CDEST_STATE_ABR%2CDEST_STATE_FIPS%2CDEST_STATE_NM%2CDEST_COUNTRY%2CDEST_COUNTRY_NAME%2CDEST_WAC%2CAIRCRAFT_GROUP%2CAIRCRAFT_TYPE%2CAIRCRAFT_CONFIG%2CYEAR%2CQUARTER%2CMONTH%2CDISTANCE_GROUP%2CCLASS%2CDATA_SOURCE&grouplist=&suml=&sumRegion=&filter1=title%3D&filter2=title%3D&geo=All%A0&time=January&timename=Month&GEOGRAPHY=All&XYEAR=2008&FREQUENCY=1&AllVars=All&VarName=DEPARTURES_SCHEDULED&VarDesc=DepScheduled&VarType=Num&VarName=DEPARTURES_PERFORMED&VarDesc=DepPerformed&VarType=Num&VarName=PAYLOAD&VarDesc=Payload&VarType=Num&VarName=SEATS&VarDesc=Seats&VarType=Num&VarName=PASSENGERS&VarDesc=Passengers&VarType=Num&VarName=FREIGHT&VarDesc=Freight&VarType=Num&VarName=MAIL&VarDesc=Mail&VarType=Num&VarName=DISTANCE&VarDesc=Distance&VarType=Num&VarName=RAMP_TO_RAMP&VarDesc=RampToRamp&VarType=Num&VarName=AIR_TIME&VarDesc=AirTime&VarType=Num&VarName=UNIQUE_CARRIER&VarDesc=UniqueCarrier&VarType=Char&VarName=AIRLINE_ID&VarDesc=AirlineID&VarType=Num&VarName=UNIQUE_CARRIER_NAME&VarDesc=UniqueCarrierName&VarType=Char&VarName=UNIQUE_CARRIER_ENTITY&VarDesc=UniqCarrierEntity&VarType=Char&VarName=REGION&VarDesc=CarrierRegion&VarType=Char&VarName=CARRIER&VarDesc=Carrier&VarType=Char&VarName=CARRIER_NAME&VarDesc=CarrierName&VarType=Char&VarName=CARRIER_GROUP&VarDesc=CarrierGroup&VarType=Num&VarName=CARRIER_GROUP_NEW&VarDesc=CarrierGroupNew&VarType=Num&VarName=ORIGIN&VarDesc=Origin&VarType=Char&VarName=ORIGIN_CITY_NAME&VarDesc=OriginCityName&VarType=Char&VarName=ORIGIN_CITY_NUM&VarDesc=OriginCityNum&VarType=Num&VarName=ORIGIN_STATE_ABR&VarDesc=OriginState&VarType=Char&VarName=ORIGIN_STATE_FIPS&VarDesc=OriginStateFips&VarType=Char&VarName=ORIGIN_STATE_NM&VarDesc=OriginStateName&VarType=Char&VarName=ORIGIN_COUNTRY&VarDesc=OriginCountry&VarType=Char&VarName=ORIGIN_COUNTRY_NAME&VarDesc=OriginCountryName&VarType=Char&VarName=ORIGIN_WAC&VarDesc=OriginWac&VarType=Num&VarName=DEST&VarDesc=Dest&VarType=Char&VarName=DEST_CITY_NAME&VarDesc=DestCityName&VarType=Char&VarName=DEST_CITY_NUM&VarDesc=DestCityNum&VarType=Num&VarName=DEST_STATE_ABR&VarDesc=DestState&VarType=Char&VarName=DEST_STATE_FIPS&VarDesc=DestStateFips&VarType=Char&VarName=DEST_STATE_NM&VarDesc=DestStateName&VarType=Char&VarName=DEST_COUNTRY&VarDesc=DestCountry&VarType=Char&VarName=DEST_COUNTRY_NAME&VarDesc=DestCountryName&VarType=Char&VarName=DEST_WAC&VarDesc=DestWac&VarType=Num&VarName=AIRCRAFT_GROUP&VarDesc=AircraftGroup&VarType=Num&VarName=AIRCRAFT_TYPE&VarDesc=AircraftType&VarType=Char&VarName=AIRCRAFT_CONFIG&VarDesc=AircraftConfig&VarType=Num&VarName=YEAR&VarDesc=Year&VarType=Num&VarName=QUARTER&VarDesc=Quarter&VarType=Num&VarName=MONTH&VarDesc=Month&VarType=Num&VarName=DISTANCE_GROUP&VarDesc=DistanceGroup&VarType=Num&VarName=CLASS&VarDesc=Class&VarType=Char&VarName=DATA_SOURCE&VarDesc=DataSource&VarType=Char'
|
124
|
-
t = RemoteTable.new :url => url, :form_data => form_data, :compression => :zip, :glob => '/*.csv'
|
125
|
-
assert_equal 'United States of America', t.rows.first['DEST_COUNTRY_NAME']
|
126
|
-
end
|
127
|
-
end
|
128
|
-
|
129
|
-
if ENV['ALL'] == 'true' or ENV['NEW'] == 'true'
|
130
|
-
end
|
131
|
-
|
132
|
-
if ENV['ALL'] == 'true' or ENV['FAST'] == 'true'
|
133
|
-
should "open an XLSX like an array (numbered columns)" do
|
134
|
-
t = RemoteTable.new(:url => 'www.customerreferenceprogram.org/uploads/CRP_RFP_template.xlsx', :headers => false)
|
135
|
-
assert_equal "Secure encryption of all data", t.rows[5][0]
|
136
|
-
end
|
137
|
-
|
138
|
-
should "open an XLSX with custom headers" do
|
139
|
-
t = RemoteTable.new(:url => 'www.customerreferenceprogram.org/uploads/CRP_RFP_template.xlsx', :headers => %w{foo bar baz})
|
140
|
-
assert_equal "Secure encryption of all data", t.rows[5]['foo']
|
141
|
-
end
|
142
|
-
|
143
|
-
should "open an XLSX" do
|
144
|
-
t = RemoteTable.new(:url => 'www.customerreferenceprogram.org/uploads/CRP_RFP_template.xlsx')
|
145
|
-
assert_equal "Secure encryption of all data", t.rows[5]["Requirements"]
|
146
|
-
end
|
147
|
-
|
148
|
-
should "work on filenames with spaces, using globbing" do
|
149
|
-
t = RemoteTable.new :url => 'http://www.fueleconomy.gov/FEG/epadata/08data.zip', :glob => '/*.csv'
|
150
|
-
assert_equal 'ASTON MARTIN', t.rows.first['MFR']
|
151
|
-
end
|
152
|
-
|
153
|
-
should "work on filenames with spaces" do
|
154
|
-
t = RemoteTable.new :url => 'http://www.fueleconomy.gov/FEG/epadata/08data.zip', :filename => '2008_FE_guide_ALL_rel_dates_-no sales-for DOE-5-1-08.csv'
|
155
|
-
assert_equal 'ASTON MARTIN', t.rows.first['MFR']
|
156
|
-
end
|
157
|
-
|
158
|
-
should "ignore UTF-8 byte order marks" do
|
159
|
-
t = RemoteTable.new :url => 'http://www.freebase.com/type/exporttypeinstances/base/horses/horse_breed?page=0&filter_mode=type&filter_view=table&show%01p%3D%2Ftype%2Fobject%2Fname%01index=0&show%01p%3D%2Fcommon%2Ftopic%2Fimage%01index=1&show%01p%3D%2Fcommon%2Ftopic%2Farticle%01index=2&sort%01p%3D%2Ftype%2Fobject%2Ftype%01p%3Dlink%01p%3D%2Ftype%2Flink%2Ftimestamp%01index=false&=&exporttype=csv-8'
|
160
|
-
assert_equal 'Tawleed', t.rows.first['name']
|
161
|
-
end
|
162
|
-
|
163
|
-
should "be able to apply errata files" do
|
164
|
-
t = RemoteTable.new :url => "http://www.faa.gov/air_traffic/publications/atpubs/CNT/5-2-G.htm",
|
165
|
-
:encoding => 'windows-1252',
|
166
|
-
:row_xpath => '//table/tr[2]/td/table/tr',
|
167
|
-
:column_xpath => 'td',
|
168
|
-
:errata => Errata.new(:table => RemoteTable.new(:url => 'http://spreadsheets.google.com/pub?key=tObVAGyqOkCBtGid0tJUZrw'),
|
169
|
-
:responder => AircraftGuru.new)
|
170
|
-
g1 = t.rows.detect { |row| row['Model'] =~ /Gulfstream I/ }
|
171
|
-
assert g1
|
172
|
-
assert_equal 'GULFSTREAM AEROSPACE', g1['Manufacturer']
|
173
|
-
assert_equal 'Gulfstream I', g1['Model']
|
174
|
-
end
|
175
|
-
|
176
|
-
# this will die with an error about libcurl if your curl doesn't support ssl
|
177
|
-
should "connect using HTTPS if available" do
|
178
|
-
t = RemoteTable.new(:url => 'https://spreadsheets.google.com/pub?key=t5HM1KbaRngmTUbntg8JwPA')
|
179
|
-
assert_equal 'Gulf Coast', t.rows.first['PAD district name']
|
180
|
-
assert_equal 'AL', t.rows.first['State']
|
181
|
-
assert_equal 'Rocky Mountain', t.rows.last['PAD district name']
|
182
|
-
assert_equal 'WY', t.rows.last['State']
|
183
|
-
end
|
184
|
-
|
185
|
-
should "read an HTML table made with frontpage" do
|
186
|
-
t = RemoteTable.new :url => "http://www.faa.gov/air_traffic/publications/atpubs/CNT/5-2-E.htm",
|
187
|
-
:encoding => 'US-ASCII',
|
188
|
-
:row_xpath => '//table/tr[2]/td/table/tr',
|
189
|
-
:column_xpath => 'td'
|
190
|
-
assert_equal 'E110', t.rows.first['Designator']
|
191
|
-
assert_equal 'EMBRAER', t.rows.first['Manufacturer']
|
192
|
-
assert_equal 'EZKC', t.rows.last['Designator']
|
193
|
-
assert_equal 'EZ King Cobra', t.rows.last['Model']
|
194
|
-
end
|
195
|
-
|
196
|
-
should "hash rows without paying attention to order" do
|
197
|
-
x = ActiveSupport::OrderedHash.new
|
198
|
-
x[:a] = 1
|
199
|
-
x[:b] = 2
|
200
|
-
|
201
|
-
y = ActiveSupport::OrderedHash.new
|
202
|
-
y[:b] = 2
|
203
|
-
y[:a] = 1
|
204
|
-
|
205
|
-
assert_not_equal Marshal.dump(x), Marshal.dump(y)
|
206
|
-
assert_equal RemoteTable::Transform.row_hash(x), RemoteTable::Transform.row_hash(y)
|
207
|
-
end
|
208
|
-
|
209
|
-
should "open a Google Docs url (as a CSV)" do
|
210
|
-
t = RemoteTable.new(:url => 'http://spreadsheets.google.com/pub?key=t5HM1KbaRngmTUbntg8JwPA')
|
211
|
-
assert_equal 'Gulf Coast', t.rows.first['PAD district name']
|
212
|
-
assert_equal 'AL', t.rows.first['State']
|
213
|
-
assert_equal 'Rocky Mountain', t.rows.last['PAD district name']
|
214
|
-
assert_equal 'WY', t.rows.last['State']
|
215
|
-
end
|
216
|
-
|
217
|
-
should "open a Google Docs url (as a CSV, with sheet options)" do
|
218
|
-
t = RemoteTable.new(:url => 'http://spreadsheets.google.com/pub?key=t5HM1KbaRngmTUbntg8JwPA&single=true&gid=0')
|
219
|
-
assert_equal 'Gulf Coast', t.rows.first['PAD district name']
|
220
|
-
assert_equal 'AL', t.rows.first['State']
|
221
|
-
assert_equal 'Rocky Mountain', t.rows.last['PAD district name']
|
222
|
-
assert_equal 'WY', t.rows.last['State']
|
223
|
-
end
|
224
|
-
|
225
|
-
should "open a Google Docs url as a CSV without headers" do
|
226
|
-
t = RemoteTable.new(:url => 'http://spreadsheets.google.com/pub?key=t5HM1KbaRngmTUbntg8JwPA', :skip => 1, :headers => false)
|
227
|
-
assert_equal 'AL', t.rows.first[0]
|
228
|
-
assert_equal 'Gulf Coast', t.rows.first[4]
|
229
|
-
assert_equal 'WY', t.rows.last[0]
|
230
|
-
assert_equal 'Rocky Mountain', t.rows.last[4]
|
231
|
-
end
|
232
|
-
|
233
|
-
should "take the last of values if the header is duplicated" do
|
234
|
-
t = RemoteTable.new(:url => 'http://spreadsheets.google.com/pub?key=tujrgUOwDSLWb-P4KCt1qBg')
|
235
|
-
assert_equal '2', t.rows.first['dup_header']
|
236
|
-
end
|
237
|
-
|
238
|
-
should "respect field order in CSVs without headers" do
|
239
|
-
t = RemoteTable.new(:url => 'http://spreadsheets.google.com/pub?key=t5HM1KbaRngmTUbntg8JwPA', :skip => 1, :headers => false)
|
240
|
-
last_k = -1
|
241
|
-
saw_string = false
|
242
|
-
t.rows.each do |row|
|
243
|
-
row.each do |k, v|
|
244
|
-
if k.is_a?(Fixnum) and last_k.is_a?(Fixnum)
|
245
|
-
assert !saw_string
|
246
|
-
assert k > last_k
|
247
|
-
end
|
248
|
-
last_k = k
|
249
|
-
saw_string = k.is_a?(String)
|
250
|
-
end
|
251
|
-
end
|
252
|
-
end
|
253
|
-
|
254
|
-
%w{ csv ods xls }.each do |format|
|
255
|
-
eval %{
|
256
|
-
should "read #{format}" do
|
257
|
-
t = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/test2.#{format}')
|
258
|
-
# no blank headers
|
259
|
-
assert t.rows.all? { |row| row.keys.all?(&:present?) }
|
260
|
-
# correct values
|
261
|
-
t.rows.each_with_index do |row, index|
|
262
|
-
assert_equal row.except('row_hash'), @test2_rows[index]
|
263
|
-
end
|
264
|
-
end
|
265
|
-
|
266
|
-
should "read #{format}, keeping blank rows" do
|
267
|
-
t = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/test2.#{format}', :keep_blank_rows => true)
|
268
|
-
# no blank headers
|
269
|
-
assert t.rows.all? { |row| row.keys.all?(&:present?) }
|
270
|
-
# correct values
|
271
|
-
t.rows.each_with_index do |row, index|
|
272
|
-
assert_equal row.except('row_hash'), @test2_rows_with_blanks[index]
|
273
|
-
end
|
274
|
-
end
|
275
|
-
}
|
276
|
-
end
|
277
|
-
|
278
|
-
should "read fixed width correctly" do
|
279
|
-
t = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/test2.fixed_width.txt',
|
280
|
-
:format => :fixed_width,
|
281
|
-
:skip => 1,
|
282
|
-
:schema => [[ 'header4', 10, { :type => :string } ],
|
283
|
-
[ 'spacer', 1 ],
|
284
|
-
[ 'header5', 10, { :type => :string } ],
|
285
|
-
[ 'spacer', 12 ],
|
286
|
-
[ 'header6', 10, { :type => :string } ]])
|
287
|
-
|
288
|
-
# no blank headers
|
289
|
-
assert t.rows.all? { |row| row.keys.all?(&:present?) }
|
290
|
-
# correct values
|
291
|
-
t.rows.each_with_index do |row, index|
|
292
|
-
assert_equal row.except('row_hash'), @test2_rows[index]
|
293
|
-
end
|
294
|
-
end
|
295
|
-
|
296
|
-
should "read fixed width correctly, keeping blank rows" do
|
297
|
-
t = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/test2.fixed_width.txt',
|
298
|
-
:format => :fixed_width,
|
299
|
-
:keep_blank_rows => true,
|
300
|
-
:skip => 1,
|
301
|
-
:schema => [[ 'header4', 10, { :type => :string } ],
|
302
|
-
[ 'spacer', 1 ],
|
303
|
-
[ 'header5', 10, { :type => :string } ],
|
304
|
-
[ 'spacer', 12 ],
|
305
|
-
[ 'header6', 10, { :type => :string } ]])
|
306
|
-
|
307
|
-
# no blank headers
|
308
|
-
assert t.rows.all? { |row| row.keys.all?(&:present?) }
|
309
|
-
# correct values
|
310
|
-
t.rows.each_with_index do |row, index|
|
311
|
-
assert_equal row.except('row_hash'), @test2_rows_with_blanks[index]
|
312
|
-
end
|
313
|
-
end
|
314
|
-
|
315
|
-
should "have the same row hash across formats" do
|
316
|
-
csv = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.csv')
|
317
|
-
ods = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.ods')
|
318
|
-
xls = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.xls')
|
319
|
-
fixed_width = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.fixed_width.txt',
|
320
|
-
:format => :fixed_width,
|
321
|
-
:skip => 1,
|
322
|
-
:schema => [[ 'header1', 10, { :type => :string } ],
|
323
|
-
[ 'spacer', 1 ],
|
324
|
-
[ 'header2', 10, { :type => :string } ],
|
325
|
-
[ 'spacer', 12 ],
|
326
|
-
[ 'header3', 10, { :type => :string } ]])
|
327
|
-
|
328
|
-
csv2 = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.alternate_order.csv')
|
329
|
-
ods2 = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.alternate_order.ods')
|
330
|
-
xls2 = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.alternate_order.xls')
|
331
|
-
fixed_width2 = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/remote_table_row_hash_test.alternate_order.fixed_width.txt',
|
332
|
-
:format => :fixed_width,
|
333
|
-
:skip => 1,
|
334
|
-
:schema => [[ 'spacer', 11 ],
|
335
|
-
[ 'header2', 10, { :type => :string } ],
|
336
|
-
[ 'spacer', 1 ],
|
337
|
-
[ 'header3', 10, { :type => :string } ],
|
338
|
-
[ 'spacer', 1 ],
|
339
|
-
[ 'header1', 10, { :type => :string } ]])
|
340
|
-
|
341
|
-
|
342
|
-
reference = csv.rows[0]['row_hash']
|
343
|
-
|
344
|
-
# same row hashes
|
345
|
-
assert_equal reference, ods.rows[0]['row_hash']
|
346
|
-
assert_equal reference, xls.rows[0]['row_hash']
|
347
|
-
assert_equal reference, fixed_width.rows[0]['row_hash']
|
348
|
-
# same row hashes with different order
|
349
|
-
assert_equal reference, csv2.rows[0]['row_hash']
|
350
|
-
assert_equal reference, ods2.rows[0]['row_hash']
|
351
|
-
assert_equal reference, xls2.rows[0]['row_hash']
|
352
|
-
assert_equal reference, fixed_width2.rows[0]['row_hash']
|
353
|
-
end
|
354
|
-
|
355
|
-
should "open an ODS" do
|
356
|
-
t = RemoteTable.new(:url => 'http://www.worldmapper.org/data/opendoc/2_worldmapper_data.ods', :sheet => 'Data', :keep_blank_rows => true)
|
357
|
-
|
358
|
-
assert_equal 'Central Africa', t.rows[5]['name']
|
359
|
-
assert_equal 99, t.rows[5]['MAP DATA population (millions) 2002'].to_i
|
360
|
-
end
|
361
|
-
|
362
|
-
should "open an XLS with a parser" do
|
363
|
-
ma_1990_01 = {"month"=>1, "cost"=>"54.0", "locatable"=>"Massachusetts (State)", "year"=>1990}
|
364
|
-
ga_1990_01 = {"month"=>1, "cost"=>"50.7", "locatable"=>"Georgia (State)", "year"=>1990}
|
365
|
-
|
366
|
-
t = RemoteTable.new(:url => 'http://tonto.eia.doe.gov/dnav/pet/xls/PET_PRI_RESID_A_EPPR_PTA_CPGAL_M.xls',
|
367
|
-
:transform => { :class => FuelOilParser })
|
368
|
-
|
369
|
-
assert t.rows.include?(ma_1990_01)
|
370
|
-
assert t.rows.include?(ga_1990_01)
|
371
|
-
end
|
372
|
-
|
373
|
-
# should "provide a row_hash on demand" do
|
374
|
-
# t = RemoteTable.new(:url => 'http://www.fueleconomy.gov/FEG/epadata/00data.zip',
|
375
|
-
# :filename => 'Gd6-dsc.txt',
|
376
|
-
# :format => :fixed_width,
|
377
|
-
# :crop => 21..26, # inclusive
|
378
|
-
# :cut => '2-',
|
379
|
-
# :select => lambda { |row| /\A[A-Z]/.match row['code'] },
|
380
|
-
# :schema => [[ 'code', 2, { :type => :string } ],
|
381
|
-
# [ 'spacer', 2 ],
|
382
|
-
# [ 'name', 52, { :type => :string } ]])
|
383
|
-
# assert_equal 'a8a5d7f17b56772723c657eb62b0f238', t.rows.first['row_hash']
|
384
|
-
# end
|
385
|
-
end
|
386
|
-
end
|