wjordan213-csvlint 0.2.8
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.coveralls.yml +1 -0
- data/.gitattributes +2 -0
- data/.gitignore +28 -0
- data/.ruby-version +1 -0
- data/.travis.yml +32 -0
- data/CHANGELOG.md +361 -0
- data/Gemfile +7 -0
- data/LICENSE.md +22 -0
- data/README.md +328 -0
- data/Rakefile +17 -0
- data/bin/create_schema +32 -0
- data/bin/csvlint +10 -0
- data/features/check_format.feature +46 -0
- data/features/cli.feature +210 -0
- data/features/csv_options.feature +35 -0
- data/features/csvupload.feature +145 -0
- data/features/csvw_schema_validation.feature +127 -0
- data/features/fixtures/cr-line-endings.csv +0 -0
- data/features/fixtures/crlf-line-endings.csv +0 -0
- data/features/fixtures/inconsistent-line-endings-unquoted.csv +0 -0
- data/features/fixtures/inconsistent-line-endings.csv +0 -0
- data/features/fixtures/invalid-byte-sequence.csv +0 -0
- data/features/fixtures/invalid_many_rows.csv +0 -0
- data/features/fixtures/lf-line-endings.csv +0 -0
- data/features/fixtures/spreadsheet.xls +0 -0
- data/features/fixtures/spreadsheet.xlsx +0 -0
- data/features/fixtures/title-row.csv +0 -0
- data/features/fixtures/valid.csv +0 -0
- data/features/fixtures/valid_many_rows.csv +0 -0
- data/features/fixtures/windows-line-endings.csv +0 -0
- data/features/information.feature +22 -0
- data/features/parse_csv.feature +90 -0
- data/features/schema_validation.feature +105 -0
- data/features/sources.feature +17 -0
- data/features/step_definitions/cli_steps.rb +11 -0
- data/features/step_definitions/csv_options_steps.rb +24 -0
- data/features/step_definitions/information_steps.rb +13 -0
- data/features/step_definitions/parse_csv_steps.rb +42 -0
- data/features/step_definitions/schema_validation_steps.rb +33 -0
- data/features/step_definitions/sources_steps.rb +7 -0
- data/features/step_definitions/validation_errors_steps.rb +90 -0
- data/features/step_definitions/validation_info_steps.rb +22 -0
- data/features/step_definitions/validation_warnings_steps.rb +60 -0
- data/features/support/aruba.rb +56 -0
- data/features/support/env.rb +26 -0
- data/features/support/load_tests.rb +114 -0
- data/features/support/webmock.rb +1 -0
- data/features/validation_errors.feature +147 -0
- data/features/validation_info.feature +16 -0
- data/features/validation_warnings.feature +86 -0
- data/lib/csvlint.rb +27 -0
- data/lib/csvlint/cli.rb +165 -0
- data/lib/csvlint/csvw/column.rb +359 -0
- data/lib/csvlint/csvw/date_format.rb +182 -0
- data/lib/csvlint/csvw/metadata_error.rb +13 -0
- data/lib/csvlint/csvw/number_format.rb +211 -0
- data/lib/csvlint/csvw/property_checker.rb +761 -0
- data/lib/csvlint/csvw/table.rb +204 -0
- data/lib/csvlint/csvw/table_group.rb +165 -0
- data/lib/csvlint/error_collector.rb +27 -0
- data/lib/csvlint/error_message.rb +15 -0
- data/lib/csvlint/field.rb +196 -0
- data/lib/csvlint/schema.rb +92 -0
- data/lib/csvlint/validate.rb +599 -0
- data/lib/csvlint/version.rb +3 -0
- data/spec/csvw/column_spec.rb +112 -0
- data/spec/csvw/date_format_spec.rb +49 -0
- data/spec/csvw/number_format_spec.rb +417 -0
- data/spec/csvw/table_group_spec.rb +143 -0
- data/spec/csvw/table_spec.rb +90 -0
- data/spec/field_spec.rb +252 -0
- data/spec/schema_spec.rb +211 -0
- data/spec/spec_helper.rb +17 -0
- data/spec/validator_spec.rb +619 -0
- data/wjordan213_csvlint.gemspec +46 -0
- metadata +490 -0
@@ -0,0 +1,86 @@
|
|
1
|
+
Feature: Validation warnings
|
2
|
+
|
3
|
+
Scenario: UTF-8 Encoding
|
4
|
+
Given I have a CSV with the following content:
|
5
|
+
"""
|
6
|
+
"col1","col2","col3"
|
7
|
+
"abc","2","3"
|
8
|
+
"""
|
9
|
+
And it is encoded as "utf-8"
|
10
|
+
And it is stored at the url "http://example.com/example1.csv"
|
11
|
+
When I ask if there are warnings
|
12
|
+
Then there should be 0 warnings
|
13
|
+
|
14
|
+
Scenario: ISO-8859-1 Encoding
|
15
|
+
Given I have a CSV with the following content:
|
16
|
+
"""
|
17
|
+
"col1","col2","col3"
|
18
|
+
"1","2","3"
|
19
|
+
"""
|
20
|
+
And it is encoded as "iso-8859-1"
|
21
|
+
And it is stored at the url "http://example.com/example1.csv"
|
22
|
+
When I ask if there are warnings
|
23
|
+
Then there should be 1 warnings
|
24
|
+
|
25
|
+
Scenario: Correct content type
|
26
|
+
Given I have a CSV with the following content:
|
27
|
+
"""
|
28
|
+
"col1","col2","col3"
|
29
|
+
"abc","2","3"
|
30
|
+
"""
|
31
|
+
And the content type is set to "text/csv"
|
32
|
+
And it is stored at the url "http://example.com/example1.csv"
|
33
|
+
And I ask if there are warnings
|
34
|
+
Then there should be 0 warnings
|
35
|
+
|
36
|
+
Scenario: No extension
|
37
|
+
Given I have a CSV with the following content:
|
38
|
+
"""
|
39
|
+
"col1","col2","col3"
|
40
|
+
"abc","2","3"
|
41
|
+
"""
|
42
|
+
And the content type is set to "text/csv"
|
43
|
+
And it is stored at the url "http://example.com/example1"
|
44
|
+
And I ask if there are warnings
|
45
|
+
Then there should be 0 warnings
|
46
|
+
|
47
|
+
Scenario: Allow query params after extension
|
48
|
+
Given I have a CSV with the following content:
|
49
|
+
"""
|
50
|
+
"col1","col2","col3"
|
51
|
+
"abc","2","3"
|
52
|
+
"""
|
53
|
+
And the content type is set to "text/csv"
|
54
|
+
And it is stored at the url "http://example.com/example1.csv?query=param"
|
55
|
+
And I ask if there are warnings
|
56
|
+
Then there should be 0 warnings
|
57
|
+
|
58
|
+
Scenario: User doesn't supply encoding
|
59
|
+
Given I have a CSV with the following content:
|
60
|
+
"""
|
61
|
+
"col1","col2","col3"
|
62
|
+
"abc","2","3"
|
63
|
+
"""
|
64
|
+
And it is stored at the url "http://example.com/example1.csv" with no character set
|
65
|
+
When I ask if there are warnings
|
66
|
+
Then there should be 1 warnings
|
67
|
+
And that warning should have the type "no_encoding"
|
68
|
+
|
69
|
+
Scenario: Title rows
|
70
|
+
Given I have a CSV file called "title-row.csv"
|
71
|
+
And it is stored at the url "http://example.com/example1.csv"
|
72
|
+
And I ask if there are warnings
|
73
|
+
Then there should be 1 warnings
|
74
|
+
And that warning should have the type "title_row"
|
75
|
+
|
76
|
+
Scenario: catch excel warnings
|
77
|
+
Given I parse a file called "spreadsheet.xls"
|
78
|
+
And I ask if there are warnings
|
79
|
+
Then there should be 1 warnings
|
80
|
+
And that warning should have the type "excel"
|
81
|
+
|
82
|
+
Scenario: catch excel warnings
|
83
|
+
Given I parse a file called "spreadsheet.xlsx"
|
84
|
+
And I ask if there are warnings
|
85
|
+
Then there should be 1 warnings
|
86
|
+
And that warning should have the type "excel"
|
data/lib/csvlint.rb
ADDED
@@ -0,0 +1,27 @@
|
|
1
|
+
require 'csv'
|
2
|
+
require 'date'
|
3
|
+
require 'open-uri'
|
4
|
+
require 'set'
|
5
|
+
require 'tempfile'
|
6
|
+
require 'typhoeus'
|
7
|
+
|
8
|
+
require 'active_support/core_ext/date/conversions'
|
9
|
+
require 'active_support/core_ext/time/conversions'
|
10
|
+
require 'mime/types'
|
11
|
+
require 'open_uri_redirections'
|
12
|
+
require 'uri_template'
|
13
|
+
|
14
|
+
require 'csvlint/error_message'
|
15
|
+
require 'csvlint/error_collector'
|
16
|
+
require 'csvlint/validate'
|
17
|
+
require 'csvlint/field'
|
18
|
+
|
19
|
+
require 'csvlint/csvw/metadata_error'
|
20
|
+
require 'csvlint/csvw/number_format'
|
21
|
+
require 'csvlint/csvw/date_format'
|
22
|
+
require 'csvlint/csvw/property_checker'
|
23
|
+
require 'csvlint/csvw/column'
|
24
|
+
require 'csvlint/csvw/table'
|
25
|
+
require 'csvlint/csvw/table_group'
|
26
|
+
|
27
|
+
require 'csvlint/schema'
|
data/lib/csvlint/cli.rb
ADDED
@@ -0,0 +1,165 @@
|
|
1
|
+
require 'csvlint'
|
2
|
+
require 'colorize'
|
3
|
+
require 'json'
|
4
|
+
require 'pp'
|
5
|
+
require 'thor'
|
6
|
+
|
7
|
+
module Csvlint
|
8
|
+
class Cli < Thor
|
9
|
+
|
10
|
+
desc "myfile.csv OR csvlint http://example.com/myfile.csv", "Supports validating CSV files to check their syntax and contents"
|
11
|
+
option :dump_errors, desc: "Pretty print error and warning objects.", type: :boolean, aliases: :d
|
12
|
+
option :schema, banner: "FILENAME OR URL", desc: "Schema file", aliases: :s
|
13
|
+
def validate(source = nil)
|
14
|
+
source = read_source(source)
|
15
|
+
@schema = get_schema(options[:schema]) if options[:schema]
|
16
|
+
fetch_schema_tables(@schema, options) if source.nil?
|
17
|
+
|
18
|
+
valid = validate_csv(source, @schema, options[:dump])
|
19
|
+
exit 1 unless valid
|
20
|
+
end
|
21
|
+
|
22
|
+
def help
|
23
|
+
self.class.command_help(shell, :validate)
|
24
|
+
end
|
25
|
+
|
26
|
+
default_task :validate
|
27
|
+
|
28
|
+
private
|
29
|
+
|
30
|
+
def read_source(source)
|
31
|
+
if source.nil?
|
32
|
+
# If no source is present, try reading from stdin
|
33
|
+
if !$stdin.tty?
|
34
|
+
source = StringIO.new(STDIN.read) rescue nil
|
35
|
+
return_error "No CSV data to validate" if !options[:schema] && source.nil?
|
36
|
+
end
|
37
|
+
else
|
38
|
+
# If the source isn't a URL, it's a file
|
39
|
+
unless source =~ /^http(s)?/
|
40
|
+
begin
|
41
|
+
source = File.new( source )
|
42
|
+
rescue Errno::ENOENT
|
43
|
+
return_error "#{source} not found"
|
44
|
+
end
|
45
|
+
end
|
46
|
+
end
|
47
|
+
|
48
|
+
source
|
49
|
+
end
|
50
|
+
|
51
|
+
def get_schema(schema)
|
52
|
+
begin
|
53
|
+
schema = Csvlint::Schema.load_from_json(schema, false)
|
54
|
+
rescue Csvlint::Csvw::MetadataError => e
|
55
|
+
return_error "invalid metadata: #{e.message}#{" at " + e.path if e.path}"
|
56
|
+
rescue OpenURI::HTTPError, Errno::ENOENT
|
57
|
+
return_error "#{options[:schema]} not found"
|
58
|
+
end
|
59
|
+
|
60
|
+
if schema.class == Csvlint::Schema && schema.description == "malformed"
|
61
|
+
return_error "invalid metadata: malformed JSON"
|
62
|
+
end
|
63
|
+
|
64
|
+
schema
|
65
|
+
end
|
66
|
+
|
67
|
+
def fetch_schema_tables(schema, options)
|
68
|
+
valid = true
|
69
|
+
|
70
|
+
unless schema.instance_of? Csvlint::Csvw::TableGroup
|
71
|
+
return_error "No CSV data to validate."
|
72
|
+
end
|
73
|
+
schema.tables.keys.each do |source|
|
74
|
+
begin
|
75
|
+
source = source.sub("file:","")
|
76
|
+
source = File.new( source )
|
77
|
+
rescue Errno::ENOENT
|
78
|
+
return_error "#{source} not found"
|
79
|
+
end unless source =~ /^http(s)?/
|
80
|
+
valid &= validate_csv(source, schema, options[:dump])
|
81
|
+
end
|
82
|
+
|
83
|
+
exit 1 unless valid
|
84
|
+
end
|
85
|
+
|
86
|
+
def print_error(index, error, dump, color)
|
87
|
+
location = ""
|
88
|
+
location += error.row.to_s if error.row
|
89
|
+
location += "#{error.row ? "," : ""}#{error.column.to_s}" if error.column
|
90
|
+
if error.row || error.column
|
91
|
+
location = "#{error.row ? "Row" : "Column"}: #{location}"
|
92
|
+
end
|
93
|
+
output_string = "#{index+1}. "
|
94
|
+
if error.column && @schema && @schema.class == Csvlint::Schema
|
95
|
+
output_string += "#{@schema.fields[error.column - 1].name}: "
|
96
|
+
end
|
97
|
+
output_string += "#{error.type}"
|
98
|
+
output_string += ". #{location}" unless location.empty?
|
99
|
+
output_string += ". #{error.content}" if error.content
|
100
|
+
|
101
|
+
if $stdout.tty?
|
102
|
+
puts output_string.colorize(color)
|
103
|
+
else
|
104
|
+
puts output_string
|
105
|
+
end
|
106
|
+
|
107
|
+
if dump
|
108
|
+
pp error
|
109
|
+
end
|
110
|
+
end
|
111
|
+
|
112
|
+
def print_errors(errors, dump)
|
113
|
+
if errors.size > 0
|
114
|
+
errors.each_with_index { |error, i| print_error(i, error, dump, :red) }
|
115
|
+
end
|
116
|
+
end
|
117
|
+
|
118
|
+
def return_error(message)
|
119
|
+
if $stdout.tty?
|
120
|
+
puts message.colorize(:red)
|
121
|
+
else
|
122
|
+
puts message
|
123
|
+
end
|
124
|
+
exit 1
|
125
|
+
end
|
126
|
+
|
127
|
+
def validate_csv(source, schema, dump)
|
128
|
+
@error_count = 0
|
129
|
+
|
130
|
+
validator = Csvlint::Validator.new( source, {}, schema, { lambda: report_lines } )
|
131
|
+
|
132
|
+
if source.class == String
|
133
|
+
csv = source
|
134
|
+
elsif source.class == File
|
135
|
+
csv = source.path
|
136
|
+
else
|
137
|
+
csv = "CSV"
|
138
|
+
end
|
139
|
+
|
140
|
+
if $stdout.tty?
|
141
|
+
puts "\r\n#{csv} is #{validator.valid? ? "VALID".green : "INVALID".red}"
|
142
|
+
else
|
143
|
+
puts "\r\n#{csv} is #{validator.valid? ? "VALID" : "INVALID"}"
|
144
|
+
end
|
145
|
+
|
146
|
+
print_errors(validator.errors, dump)
|
147
|
+
print_errors(validator.warnings, dump)
|
148
|
+
|
149
|
+
return validator.valid?
|
150
|
+
end
|
151
|
+
|
152
|
+
def report_lines
|
153
|
+
lambda do |row|
|
154
|
+
new_errors = row.errors.count
|
155
|
+
if new_errors > @error_count
|
156
|
+
print "!".red
|
157
|
+
else
|
158
|
+
print ".".green
|
159
|
+
end
|
160
|
+
@error_count = new_errors
|
161
|
+
end
|
162
|
+
end
|
163
|
+
|
164
|
+
end
|
165
|
+
end
|
@@ -0,0 +1,359 @@
|
|
1
|
+
module Csvlint
|
2
|
+
module Csvw
|
3
|
+
class Column
|
4
|
+
include Csvlint::ErrorCollector
|
5
|
+
|
6
|
+
attr_reader :id, :about_url, :datatype, :default, :lang, :name, :null, :number, :ordered, :property_url, :required, :separator, :source_number, :suppress_output, :text_direction, :titles, :value_url, :virtual, :annotations
|
7
|
+
|
8
|
+
def initialize(number, name, id: nil, about_url: nil, datatype: { "@id" => "http://www.w3.org/2001/XMLSchema#string" }, default: "", lang: "und", null: [""], ordered: false, property_url: nil, required: false, separator: nil, source_number: nil, suppress_output: false, text_direction: :inherit, titles: {}, value_url: nil, virtual: false, annotations: [], warnings: [])
|
9
|
+
@number = number
|
10
|
+
@name = name
|
11
|
+
@id = id
|
12
|
+
@about_url = about_url
|
13
|
+
@datatype = datatype
|
14
|
+
@default = default
|
15
|
+
@lang = lang
|
16
|
+
@null = null
|
17
|
+
@ordered = ordered
|
18
|
+
@property_url = property_url
|
19
|
+
@required = required
|
20
|
+
@separator = separator
|
21
|
+
@source_number = source_number || number
|
22
|
+
@suppress_output = suppress_output
|
23
|
+
@text_direction = text_direction
|
24
|
+
@titles = titles
|
25
|
+
@value_url = value_url
|
26
|
+
@virtual = virtual
|
27
|
+
@annotations = annotations
|
28
|
+
reset
|
29
|
+
@warnings += warnings
|
30
|
+
end
|
31
|
+
|
32
|
+
def self.from_json(number, column_desc, base_url=nil, lang="und", inherited_properties={})
|
33
|
+
annotations = {}
|
34
|
+
warnings = []
|
35
|
+
column_properties = {}
|
36
|
+
inherited_properties = inherited_properties.clone
|
37
|
+
|
38
|
+
column_desc.each do |property,value|
|
39
|
+
if property == "@type"
|
40
|
+
raise Csvlint::Csvw::MetadataError.new("columns[#{number}].@type"), "@type of column is not 'Column'" if value != 'Column'
|
41
|
+
else
|
42
|
+
v, warning, type = Csvw::PropertyChecker.check_property(property, value, base_url, lang)
|
43
|
+
warnings += Array(warning).map{ |w| Csvlint::ErrorMessage.new(w, :metadata, nil, nil, "#{property}: #{value}", nil) } unless warning.nil? || warning.empty?
|
44
|
+
if type == :annotation
|
45
|
+
annotations[property] = v
|
46
|
+
elsif type == :common || type == :column
|
47
|
+
column_properties[property] = v
|
48
|
+
elsif type == :inherited
|
49
|
+
inherited_properties[property] = v
|
50
|
+
else
|
51
|
+
warnings << Csvlint::ErrorMessage.new(:invalid_property, :metadata, nil, nil, "column: #{property}", nil)
|
52
|
+
end
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
return self.new(number, column_properties["name"],
|
57
|
+
id: column_properties["@id"],
|
58
|
+
datatype: inherited_properties["datatype"] || { "@id" => "http://www.w3.org/2001/XMLSchema#string" },
|
59
|
+
lang: inherited_properties["lang"] || "und",
|
60
|
+
null: inherited_properties["null"] || [""],
|
61
|
+
property_url: column_desc["propertyUrl"],
|
62
|
+
required: inherited_properties["required"] || false,
|
63
|
+
separator: inherited_properties["separator"],
|
64
|
+
titles: column_properties["titles"],
|
65
|
+
virtual: column_properties["virtual"] || false,
|
66
|
+
annotations: annotations,
|
67
|
+
warnings: warnings
|
68
|
+
)
|
69
|
+
end
|
70
|
+
|
71
|
+
def validate_header(header)
|
72
|
+
reset
|
73
|
+
valid_headers = @titles ? @titles.map{ |l,v| v if Column.languages_match(l, lang) }.flatten : []
|
74
|
+
build_errors(:invalid_header, :schema, 1, @number, header, @titles) unless valid_headers.include? header
|
75
|
+
return valid?
|
76
|
+
end
|
77
|
+
|
78
|
+
def validate(string_value, row=nil)
|
79
|
+
reset
|
80
|
+
values = parse(string_value || "", row)
|
81
|
+
# STDERR.puts "#{name} - #{string_value.inspect} - #{values.inspect}"
|
82
|
+
values.each do |value|
|
83
|
+
validate_required(value, row)
|
84
|
+
validate_format(value, row)
|
85
|
+
validate_length(value, row)
|
86
|
+
validate_value(value, row)
|
87
|
+
end unless values.nil?
|
88
|
+
validate_required(values, row) if values.nil?
|
89
|
+
return valid?
|
90
|
+
end
|
91
|
+
|
92
|
+
def parse(string_value, row=nil)
|
93
|
+
return nil if null.include? string_value
|
94
|
+
string_values = @separator.nil? ? [string_value] : string_value.split(@separator)
|
95
|
+
values = []
|
96
|
+
string_values.each do |s|
|
97
|
+
value, warning = DATATYPE_PARSER[@datatype["base"] || @datatype["@id"]].call(s, @datatype["format"])
|
98
|
+
if warning.nil?
|
99
|
+
values << value
|
100
|
+
else
|
101
|
+
build_errors(warning, :schema, row, @number, s, @datatype)
|
102
|
+
values << s
|
103
|
+
end
|
104
|
+
end
|
105
|
+
return values
|
106
|
+
end
|
107
|
+
|
108
|
+
private
|
109
|
+
class << self
|
110
|
+
|
111
|
+
def create_date_parser(type, warning)
|
112
|
+
return lambda { |value, format|
|
113
|
+
format = Csvlint::Csvw::DateFormat.new(nil, type) if format.nil?
|
114
|
+
v = format.parse(value)
|
115
|
+
return nil, warning if v.nil?
|
116
|
+
return v, nil
|
117
|
+
}
|
118
|
+
end
|
119
|
+
|
120
|
+
def create_regexp_based_parser(regexp, warning)
|
121
|
+
return lambda { |value, format|
|
122
|
+
return nil, warning unless value =~ regexp
|
123
|
+
return value, nil
|
124
|
+
}
|
125
|
+
end
|
126
|
+
|
127
|
+
def languages_match(l1, l2)
|
128
|
+
return true if l1 == l2 || l1 == "und" || l2 == "und"
|
129
|
+
return true if l1 =~ Regexp.new("^#{l2}-") || l2 =~ Regexp.new("^#{l1}-")
|
130
|
+
return false
|
131
|
+
end
|
132
|
+
end
|
133
|
+
|
134
|
+
def validate_required(value, row)
|
135
|
+
build_errors(:required, :schema, row, number, value, { "required" => @required }) if @required && value.nil?
|
136
|
+
end
|
137
|
+
|
138
|
+
def validate_length(value, row)
|
139
|
+
if datatype["length"] || datatype["minLength"] || datatype["maxLength"]
|
140
|
+
length = value.length
|
141
|
+
length = value.gsub(/==?$/,"").length * 3 / 4 if datatype["@id"] == "http://www.w3.org/2001/XMLSchema#base64Binary" || datatype["base"] == "http://www.w3.org/2001/XMLSchema#base64Binary"
|
142
|
+
length = value.length / 2 if datatype["@id"] == "http://www.w3.org/2001/XMLSchema#hexBinary" || datatype["base"] == "http://www.w3.org/2001/XMLSchema#hexBinary"
|
143
|
+
|
144
|
+
build_errors(:min_length, :schema, row, number, value, { "minLength" => datatype["minLength"] }) if datatype["minLength"] && length < datatype["minLength"]
|
145
|
+
build_errors(:max_length, :schema, row, number, value, { "maxLength" => datatype["maxLength"] }) if datatype["maxLength"] && length > datatype["maxLength"]
|
146
|
+
build_errors(:length, :schema, row, number, value, { "length" => datatype["length"] }) if datatype["length"] && length != datatype["length"]
|
147
|
+
end
|
148
|
+
end
|
149
|
+
|
150
|
+
def validate_format(value, row)
|
151
|
+
if datatype["format"]
|
152
|
+
build_errors(:format, :schema, row, number, value, { "format" => datatype["format"] }) unless DATATYPE_FORMAT_VALIDATION[datatype["base"]].call(value, datatype["format"])
|
153
|
+
end
|
154
|
+
end
|
155
|
+
|
156
|
+
def validate_value(value, row)
|
157
|
+
build_errors(:min_inclusive, :schema, row, number, value, { "minInclusive" => datatype["minInclusive"] }) if datatype["minInclusive"] && value < datatype["minInclusive"]
|
158
|
+
build_errors(:max_inclusive, :schema, row, number, value, { "maxInclusive" => datatype["maxInclusive"] }) if datatype["maxInclusive"] && value > datatype["maxInclusive"]
|
159
|
+
build_errors(:min_exclusive, :schema, row, number, value, { "minExclusive" => datatype["minExclusive"] }) if datatype["minExclusive"] && value <= datatype["minExclusive"]
|
160
|
+
build_errors(:max_exclusive, :schema, row, number, value, { "maxExclusive" => datatype["maxExclusive"] }) if datatype["maxExclusive"] && value >= datatype["maxExclusive"]
|
161
|
+
end
|
162
|
+
|
163
|
+
REGEXP_VALIDATION = lambda { |value, format| value =~ format }
|
164
|
+
|
165
|
+
NO_ADDITIONAL_VALIDATION = lambda { |value, format| true }
|
166
|
+
|
167
|
+
DATATYPE_FORMAT_VALIDATION = {
|
168
|
+
"http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral" => REGEXP_VALIDATION,
|
169
|
+
"http://www.w3.org/1999/02/22-rdf-syntax-ns#HTML" => REGEXP_VALIDATION,
|
170
|
+
"http://www.w3.org/ns/csvw#JSON" => REGEXP_VALIDATION,
|
171
|
+
"http://www.w3.org/2001/XMLSchema#anyAtomicType" => REGEXP_VALIDATION,
|
172
|
+
"http://www.w3.org/2001/XMLSchema#anyURI" => REGEXP_VALIDATION,
|
173
|
+
"http://www.w3.org/2001/XMLSchema#base64Binary" => REGEXP_VALIDATION,
|
174
|
+
"http://www.w3.org/2001/XMLSchema#boolean" => NO_ADDITIONAL_VALIDATION,
|
175
|
+
"http://www.w3.org/2001/XMLSchema#date" => NO_ADDITIONAL_VALIDATION,
|
176
|
+
"http://www.w3.org/2001/XMLSchema#dateTime" => NO_ADDITIONAL_VALIDATION,
|
177
|
+
"http://www.w3.org/2001/XMLSchema#dateTimeStamp" => NO_ADDITIONAL_VALIDATION,
|
178
|
+
"http://www.w3.org/2001/XMLSchema#decimal" => NO_ADDITIONAL_VALIDATION,
|
179
|
+
"http://www.w3.org/2001/XMLSchema#integer" => NO_ADDITIONAL_VALIDATION,
|
180
|
+
"http://www.w3.org/2001/XMLSchema#long" => NO_ADDITIONAL_VALIDATION,
|
181
|
+
"http://www.w3.org/2001/XMLSchema#int" => NO_ADDITIONAL_VALIDATION,
|
182
|
+
"http://www.w3.org/2001/XMLSchema#short" => NO_ADDITIONAL_VALIDATION,
|
183
|
+
"http://www.w3.org/2001/XMLSchema#byte" => NO_ADDITIONAL_VALIDATION,
|
184
|
+
"http://www.w3.org/2001/XMLSchema#nonNegativeInteger" => NO_ADDITIONAL_VALIDATION,
|
185
|
+
"http://www.w3.org/2001/XMLSchema#positiveInteger" => NO_ADDITIONAL_VALIDATION,
|
186
|
+
"http://www.w3.org/2001/XMLSchema#unsignedLong" => NO_ADDITIONAL_VALIDATION,
|
187
|
+
"http://www.w3.org/2001/XMLSchema#unsignedInt" => NO_ADDITIONAL_VALIDATION,
|
188
|
+
"http://www.w3.org/2001/XMLSchema#unsignedShort" => NO_ADDITIONAL_VALIDATION,
|
189
|
+
"http://www.w3.org/2001/XMLSchema#unsignedByte" => NO_ADDITIONAL_VALIDATION,
|
190
|
+
"http://www.w3.org/2001/XMLSchema#nonPositiveInteger" => NO_ADDITIONAL_VALIDATION,
|
191
|
+
"http://www.w3.org/2001/XMLSchema#negativeInteger" => NO_ADDITIONAL_VALIDATION,
|
192
|
+
"http://www.w3.org/2001/XMLSchema#double" => NO_ADDITIONAL_VALIDATION,
|
193
|
+
"http://www.w3.org/2001/XMLSchema#duration" => REGEXP_VALIDATION,
|
194
|
+
"http://www.w3.org/2001/XMLSchema#dayTimeDuration" => REGEXP_VALIDATION,
|
195
|
+
"http://www.w3.org/2001/XMLSchema#yearMonthDuration" => REGEXP_VALIDATION,
|
196
|
+
"http://www.w3.org/2001/XMLSchema#float" => NO_ADDITIONAL_VALIDATION,
|
197
|
+
"http://www.w3.org/2001/XMLSchema#gDay" => NO_ADDITIONAL_VALIDATION,
|
198
|
+
"http://www.w3.org/2001/XMLSchema#gMonth" => NO_ADDITIONAL_VALIDATION,
|
199
|
+
"http://www.w3.org/2001/XMLSchema#gMonthDay" => NO_ADDITIONAL_VALIDATION,
|
200
|
+
"http://www.w3.org/2001/XMLSchema#gYear" => NO_ADDITIONAL_VALIDATION,
|
201
|
+
"http://www.w3.org/2001/XMLSchema#gYearMonth" => NO_ADDITIONAL_VALIDATION,
|
202
|
+
"http://www.w3.org/2001/XMLSchema#hexBinary" => REGEXP_VALIDATION,
|
203
|
+
"http://www.w3.org/2001/XMLSchema#QName" => REGEXP_VALIDATION,
|
204
|
+
"http://www.w3.org/2001/XMLSchema#string" => REGEXP_VALIDATION,
|
205
|
+
"http://www.w3.org/2001/XMLSchema#normalizedString" => REGEXP_VALIDATION,
|
206
|
+
"http://www.w3.org/2001/XMLSchema#token" => REGEXP_VALIDATION,
|
207
|
+
"http://www.w3.org/2001/XMLSchema#language" => REGEXP_VALIDATION,
|
208
|
+
"http://www.w3.org/2001/XMLSchema#Name" => REGEXP_VALIDATION,
|
209
|
+
"http://www.w3.org/2001/XMLSchema#NMTOKEN" => REGEXP_VALIDATION,
|
210
|
+
"http://www.w3.org/2001/XMLSchema#time" => NO_ADDITIONAL_VALIDATION
|
211
|
+
}
|
212
|
+
|
213
|
+
ALL_VALUES_VALID = lambda { |value, format| return value, nil }
|
214
|
+
|
215
|
+
NUMERIC_PARSER = lambda { |value, format|
|
216
|
+
format = Csvlint::Csvw::NumberFormat.new() if format.nil?
|
217
|
+
v = format.parse(value)
|
218
|
+
return nil, :invalid_number if v.nil?
|
219
|
+
return v, nil
|
220
|
+
}
|
221
|
+
|
222
|
+
DATATYPE_PARSER = {
|
223
|
+
"http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral" => ALL_VALUES_VALID,
|
224
|
+
"http://www.w3.org/1999/02/22-rdf-syntax-ns#HTML" => ALL_VALUES_VALID,
|
225
|
+
"http://www.w3.org/ns/csvw#JSON" => ALL_VALUES_VALID,
|
226
|
+
"http://www.w3.org/2001/XMLSchema#anyAtomicType" => ALL_VALUES_VALID,
|
227
|
+
"http://www.w3.org/2001/XMLSchema#anyURI" => ALL_VALUES_VALID,
|
228
|
+
"http://www.w3.org/2001/XMLSchema#base64Binary" => ALL_VALUES_VALID,
|
229
|
+
"http://www.w3.org/2001/XMLSchema#boolean" => lambda { |value, format|
|
230
|
+
if format.nil?
|
231
|
+
return true, nil if ["true", "1"].include? value
|
232
|
+
return false, nil if ["false", "0"].include? value
|
233
|
+
else
|
234
|
+
return true, nil if value == format[0]
|
235
|
+
return false, nil if value == format[1]
|
236
|
+
end
|
237
|
+
return value, :invalid_boolean
|
238
|
+
},
|
239
|
+
"http://www.w3.org/2001/XMLSchema#date" =>
|
240
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#date", :invalid_date),
|
241
|
+
"http://www.w3.org/2001/XMLSchema#dateTime" =>
|
242
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#dateTime", :invalid_date_time),
|
243
|
+
"http://www.w3.org/2001/XMLSchema#dateTimeStamp" =>
|
244
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#dateTimeStamp", :invalid_date_time_stamp),
|
245
|
+
"http://www.w3.org/2001/XMLSchema#decimal" => lambda { |value, format|
|
246
|
+
return nil, :invalid_decimal if value =~ /(E|^(NaN|INF|-INF)$)/
|
247
|
+
return NUMERIC_PARSER.call(value, format)
|
248
|
+
},
|
249
|
+
"http://www.w3.org/2001/XMLSchema#integer" => lambda { |value, format|
|
250
|
+
v, w = NUMERIC_PARSER.call(value, format)
|
251
|
+
return v, :invalid_integer unless w.nil?
|
252
|
+
return nil, :invalid_integer unless v.kind_of? Integer
|
253
|
+
return v, w
|
254
|
+
},
|
255
|
+
"http://www.w3.org/2001/XMLSchema#long" => lambda { |value, format|
|
256
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
257
|
+
return v, :invalid_long unless w.nil?
|
258
|
+
return nil, :invalid_long unless v <= 9223372036854775807 && v >= -9223372036854775808
|
259
|
+
return v, w
|
260
|
+
},
|
261
|
+
"http://www.w3.org/2001/XMLSchema#int" => lambda { |value, format|
|
262
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
263
|
+
return v, :invalid_int unless w.nil?
|
264
|
+
return nil, :invalid_int unless v <= 2147483647 && v >= -2147483648
|
265
|
+
return v, w
|
266
|
+
},
|
267
|
+
"http://www.w3.org/2001/XMLSchema#short" => lambda { |value, format|
|
268
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
269
|
+
return v, :invalid_short unless w.nil?
|
270
|
+
return nil, :invalid_short unless v <= 32767 && v >= -32768
|
271
|
+
return v, w
|
272
|
+
},
|
273
|
+
"http://www.w3.org/2001/XMLSchema#byte" => lambda { |value, format|
|
274
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
275
|
+
return v, :invalid_byte unless w.nil?
|
276
|
+
return nil, :invalid_byte unless v <= 127 && v >= -128
|
277
|
+
return v, w
|
278
|
+
},
|
279
|
+
"http://www.w3.org/2001/XMLSchema#nonNegativeInteger" => lambda { |value, format|
|
280
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
281
|
+
return v, :invalid_nonNegativeInteger unless w.nil?
|
282
|
+
return nil, :invalid_nonNegativeInteger unless v >= 0
|
283
|
+
return v, w
|
284
|
+
},
|
285
|
+
"http://www.w3.org/2001/XMLSchema#positiveInteger" => lambda { |value, format|
|
286
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
287
|
+
return v, :invalid_positiveInteger unless w.nil?
|
288
|
+
return nil, :invalid_positiveInteger unless v > 0
|
289
|
+
return v, w
|
290
|
+
},
|
291
|
+
"http://www.w3.org/2001/XMLSchema#unsignedLong" => lambda { |value, format|
|
292
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#nonNegativeInteger"].call(value, format)
|
293
|
+
return v, :invalid_unsignedLong unless w.nil?
|
294
|
+
return nil, :invalid_unsignedLong unless v <= 18446744073709551615
|
295
|
+
return v, w
|
296
|
+
},
|
297
|
+
"http://www.w3.org/2001/XMLSchema#unsignedInt" => lambda { |value, format|
|
298
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#nonNegativeInteger"].call(value, format)
|
299
|
+
return v, :invalid_unsignedInt unless w.nil?
|
300
|
+
return nil, :invalid_unsignedInt unless v <= 4294967295
|
301
|
+
return v, w
|
302
|
+
},
|
303
|
+
"http://www.w3.org/2001/XMLSchema#unsignedShort" => lambda { |value, format|
|
304
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#nonNegativeInteger"].call(value, format)
|
305
|
+
return v, :invalid_unsignedShort unless w.nil?
|
306
|
+
return nil, :invalid_unsignedShort unless v <= 65535
|
307
|
+
return v, w
|
308
|
+
},
|
309
|
+
"http://www.w3.org/2001/XMLSchema#unsignedByte" => lambda { |value, format|
|
310
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#nonNegativeInteger"].call(value, format)
|
311
|
+
return v, :invalid_unsignedByte unless w.nil?
|
312
|
+
return nil, :invalid_unsignedByte unless v <= 255
|
313
|
+
return v, w
|
314
|
+
},
|
315
|
+
"http://www.w3.org/2001/XMLSchema#nonPositiveInteger" => lambda { |value, format|
|
316
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
317
|
+
return v, :invalid_nonPositiveInteger unless w.nil?
|
318
|
+
return nil, :invalid_nonPositiveInteger unless v <= 0
|
319
|
+
return v, w
|
320
|
+
},
|
321
|
+
"http://www.w3.org/2001/XMLSchema#negativeInteger" => lambda { |value, format|
|
322
|
+
v, w = DATATYPE_PARSER["http://www.w3.org/2001/XMLSchema#integer"].call(value, format)
|
323
|
+
return v, :invalid_negativeInteger unless w.nil?
|
324
|
+
return nil, :invalid_negativeInteger unless v < 0
|
325
|
+
return v, w
|
326
|
+
},
|
327
|
+
"http://www.w3.org/2001/XMLSchema#double" => NUMERIC_PARSER,
|
328
|
+
# regular expressions here taken from XML Schema datatypes spec
|
329
|
+
"http://www.w3.org/2001/XMLSchema#duration" =>
|
330
|
+
create_regexp_based_parser(/-?P((([0-9]+Y([0-9]+M)?([0-9]+D)?|([0-9]+M)([0-9]+D)?|([0-9]+D))(T(([0-9]+H)([0-9]+M)?([0-9]+(\.[0-9]+)?S)?|([0-9]+M)([0-9]+(\.[0-9]+)?S)?|([0-9]+(\.[0-9]+)?S)))?)|(T(([0-9]+H)([0-9]+M)?([0-9]+(\.[0-9]+)?S)?|([0-9]+M)([0-9]+(\.[0-9]+)?S)?|([0-9]+(\.[0-9]+)?S))))/, :invalid_duration),
|
331
|
+
"http://www.w3.org/2001/XMLSchema#dayTimeDuration" =>
|
332
|
+
create_regexp_based_parser(/-?P(([0-9]+D(T(([0-9]+H)([0-9]+M)?([0-9]+(\.[0-9]+)?S)?|([0-9]+M)([0-9]+(\.[0-9]+)?S)?|([0-9]+(\.[0-9]+)?S)))?)|(T(([0-9]+H)([0-9]+M)?([0-9]+(\.[0-9]+)?S)?|([0-9]+M)([0-9]+(\.[0-9]+)?S)?|([0-9]+(\.[0-9]+)?S))))/, :invalid_dayTimeDuration),
|
333
|
+
"http://www.w3.org/2001/XMLSchema#yearMonthDuration" =>
|
334
|
+
create_regexp_based_parser(/-?P([0-9]+Y([0-9]+M)?|([0-9]+M))/, :invalid_duration),
|
335
|
+
"http://www.w3.org/2001/XMLSchema#float" => NUMERIC_PARSER,
|
336
|
+
"http://www.w3.org/2001/XMLSchema#gDay" =>
|
337
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#gDay", :invalid_gDay),
|
338
|
+
"http://www.w3.org/2001/XMLSchema#gMonth" =>
|
339
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#gMonth", :invalid_gMonth),
|
340
|
+
"http://www.w3.org/2001/XMLSchema#gMonthDay" =>
|
341
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#gMonthDay", :invalid_gMonthDay),
|
342
|
+
"http://www.w3.org/2001/XMLSchema#gYear" =>
|
343
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#gYear", :invalid_gYear),
|
344
|
+
"http://www.w3.org/2001/XMLSchema#gYearMonth" =>
|
345
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#gYearMonth", :invalid_gYearMonth),
|
346
|
+
"http://www.w3.org/2001/XMLSchema#hexBinary" => ALL_VALUES_VALID,
|
347
|
+
"http://www.w3.org/2001/XMLSchema#QName" => ALL_VALUES_VALID,
|
348
|
+
"http://www.w3.org/2001/XMLSchema#string" => ALL_VALUES_VALID,
|
349
|
+
"http://www.w3.org/2001/XMLSchema#normalizedString" => ALL_VALUES_VALID,
|
350
|
+
"http://www.w3.org/2001/XMLSchema#token" => ALL_VALUES_VALID,
|
351
|
+
"http://www.w3.org/2001/XMLSchema#language" => ALL_VALUES_VALID,
|
352
|
+
"http://www.w3.org/2001/XMLSchema#Name" => ALL_VALUES_VALID,
|
353
|
+
"http://www.w3.org/2001/XMLSchema#NMTOKEN" => ALL_VALUES_VALID,
|
354
|
+
"http://www.w3.org/2001/XMLSchema#time" =>
|
355
|
+
create_date_parser("http://www.w3.org/2001/XMLSchema#time", :invalid_time)
|
356
|
+
}
|
357
|
+
end
|
358
|
+
end
|
359
|
+
end
|