Added CSV Schema language

rouge-ruby · pyrmont · Sep 17, 2019 · Nov 21, 2018 · Sep 2, 2019 · Sep 2, 2019
commit 9ea57bafbbcc75bdab0960efd5eaa292eb669ea1
diff --git a/lib/rouge/demos/csvs b/lib/rouge/demos/csvs
@@ -0,0 +1,8 @@
+version 1.1
+@totalColumns 5
+@separator ','
+Transaction_Date: xDate
+Transaction_ID: notEmpty
+Originator_Name: notEmpty
+Originator_Address: any("yes","no")
+Originator_Country: notEmpty
diff --git a/lib/rouge/lexers/csvs.rb b/lib/rouge/lexers/csvs.rb
@@ -0,0 +1,69 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    class CSVS < RegexLexer
+      tag 'csvs'
+      title "csvs"
+      desc 'The CSV Schema Language (http://digital-preservation.github.io/csv-schema/)'
+      mimetypes 'text/x-csvssrc'
+      filenames '*.csvs'
+
+      def self.detect?(text)
+        return true if text.shebang? 'csvs'
+      end
+
+      def self.constants
+        @constants ||= Set.new %w(nil false true)
+      end
+
+      def self.builtins
+        @builtins ||= Set.new %w(
+          args call clone do doFile doString else elseif for if list
+          method return super then
+        )
+      end
+
+      state :root do
+        rule /\s+/m, Text
+        rule %r(//.*?\n), Comment::Single
+        rule %r(#.*?\n), Comment::Single
+        rule %r(/(\\\n)?[*].*?[*](\\\n)?/)m, Comment::Multiline
+        rule %r(/[+]), Comment::Multiline, :nested_comment
+
+        rule /"(\\\\|\\"|[^"])*"/, Str
+
+        rule %r(:?:=), Keyword
+        rule /[()]/, Punctuation
+
+        rule %r([-=;,*+><!/|^.%&\[\]{}]), Operator
+
+        rule /[A-Z]\w*/, Name::Class
+
+        rule /[a-z_]\w*/ do |m|
+          name = m[0]
+
+          if self.class.constants.include? name
+            token Keyword::Constant
+          elsif self.class.builtins.include? name
+            token Name::Builtin
+          else
+            token Name
+          end
+        end
+
+        rule %r((\d+[.]?\d*|\d*[.]\d+)(e[+-]?[0-9]+)?)i, Num::Float
+        rule /\d+/, Num::Integer
+
+        rule /@@?|\'|\:/, Keyword
+      end
+
+      state :nested_comment do
+        rule %r([^/+]+)m, Comment::Multiline
+        rule %r(/[+]), Comment::Multiline, :nested_comment
+        rule %r([+]/), Comment::Multiline, :pop!
+      end
+    end
+  end
+end
diff --git a/spec/lexers/csvs_spec.rb b/spec/lexers/csvs_spec.rb
@@ -0,0 +1,22 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+describe Rouge::Lexers::CSVS do
+  let(:subject) { Rouge::Lexers::CSVS.new }
+
+  describe 'guessing' do
+    include Support::Guessing
+
+    it 'guesses by filename' do
+      assert_guess :filename => 'foo.csvs'
+    end
+
+    it 'guesses by mimetype' do
+      assert_guess :mimetype => 'text/x-csvssrc'
+    end
+
+    it 'guesses by source' do
+      assert_guess :source => '#!/usr/local/bin/csvs'
+    end
+  end
+end
diff --git a/spec/visual/samples/csvs b/spec/visual/samples/csvs
@@ -0,0 +1,22 @@
+version 1.0
+@totalColumns 19
+@separator ','
+Customer_Name: notEmpty
+Customer_Date_of_Birth: xDate
+Customer_Place_of_Birth: notEmpty
+Customer_Address: notEmpty
+Customer_Account_number: notEmpty
+Customer_Legal_Entity_Type: notEmpty
+Customer_Industry: notEmpty
+Adverse_Information_Search: xDate
+Customer_Product_Usage: notEmpty
+Customer_Net_Worth: notEmpty
+Customer_Cash_Balance: notEmpty
+Account_Open_Date: notEmpty
+CDD_Completion_Date: notEmpty
+System_Customer_Risk_Rating: notEmpty
+PEP_Status: any("yes","no")
+Associated_PEP: any("yes","no")
+EDD_Triggered: any("yes","no")
+Last_CDD_Renewal_Date: xDate
+Relationship_Manager_Employee_ID: notEmpty