csv_plus_plus 0.1.2 → 0.2.0

data/lib/csv_plus_plus/lexer/racc_lexer.rb

@@ -0,0 +1,144 @@
+# typed: strict
+# frozen_string_literal: true
+
+module CSVPlusPlus
+  module Lexer
+    # TODO: ugh clean this up
+    RaccToken =
+      ::T.type_alias do
+        ::T.any(
+          [::String, ::Symbol],
+          [::Symbol, ::String],
+          [::String, ::String],
+          [::Symbol, ::Symbol],
+          [::FalseClass, ::FalseClass]
+        )
+      end
+    public_constant :RaccToken
+
+    # Common methods to be mixed into the Racc parsers
+    #
+    # @attr_reader tokens [Array]
+    module RaccLexer
+      extend ::T::Sig
+      extend ::T::Helpers
+      extend ::T::Generic
+      include ::Kernel
+
+      abstract!
+
+      ReturnType = type_member
+      public_constant :ReturnType
+
+      sig { returns(::T::Array[::CSVPlusPlus::Lexer::RaccToken]) }
+      attr_reader :tokens
+
+      sig { params(tokens: ::T::Array[::CSVPlusPlus::Lexer::RaccToken]).void }
+      # Initialize a lexer instance with an empty +@tokens+
+      def initialize(tokens: [])
+        @tokens = ::T.let(tokens, ::T::Array[::CSVPlusPlus::Lexer::RaccToken])
+      end
+
+      sig { returns(::T.nilable(::CSVPlusPlus::Lexer::RaccToken)) }
+      # Used by racc to iterate each token
+      #
+      # @return [Array<(Regexp, Symbol) | (false, false)>]
+      def next_token
+        @tokens.shift
+      end
+
+      sig { params(input: ::String).returns(::CSVPlusPlus::Lexer::RaccLexer::ReturnType) }
+      # Orchestate the tokenizing, parsing and error handling of parsing input.  Each instance will implement their own
+      # +#tokenizer+ method
+      #
+      # @return [RaccLexer#] Each instance will define it's own +return_value+ with the result of parsing
+      # rubocop:disable Metrics/MethodLength
+      def parse(input)
+        return return_value unless anything_to_parse?(input)
+
+        tokenize(input)
+        do_parse
+        return_value
+      rescue ::Racc::ParseError => e
+        raise(
+          ::CSVPlusPlus::Error::FormulaSyntaxError.new(
+            "Error parsing #{parse_subject}",
+            bad_input: e.message,
+            wrapped_error: e
+          )
+        )
+      end
+      # rubocop:enable Metrics/MethodLength
+
+      protected
+
+      sig { abstract.params(input: ::String).returns(::T::Boolean) }
+      # Is the input even worth parsing? for example we don't want to parse cells unless they're a formula (start
+      # with '=')
+      #
+      # @param input [String]
+      #
+      # @return [Boolean]
+      def anything_to_parse?(input); end
+
+      sig { abstract.returns(::String) }
+      # Used for error messages, what is the thing being parsed? ("cell value", "modifier", "code section")
+      def parse_subject; end
+
+      sig { abstract.returns(::CSVPlusPlus::Lexer::RaccLexer::ReturnType) }
+      # The output of the parser
+      def return_value; end
+
+      sig { abstract.returns(::CSVPlusPlus::Lexer::Tokenizer) }
+      # Returns a +Lexer::Tokenizer+ configured for the given
+      def tokenizer; end
+
+      private
+
+      sig { params(input: ::String).void }
+      def tokenize(input)
+        return if input.nil?
+
+        t = tokenizer.scan(input)
+
+        until t.scanner.empty?
+          next if t.matches_ignore?
+
+          return if t.stop?
+
+          t.scan_tokens!
+          consume_token(t)
+        end
+
+        @tokens << %i[EOL EOL]
+      end
+
+      sig { params(tokenizer: ::CSVPlusPlus::Lexer::Tokenizer).void }
+      # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+      def consume_token(tokenizer)
+        if tokenizer.last_token&.token && tokenizer.last_match
+          @tokens << [::T.must(tokenizer.last_token).token, ::T.must(tokenizer.last_match)]
+        elsif tokenizer.scan_catchall
+          @tokens << [::T.must(tokenizer.last_match), ::T.must(tokenizer.last_match)]
+        # TODO: checking the +parse_subject+ like this is a little hacky... but we need to know if we're parsing
+        # modifiers or code_section (or formulas in a cell)
+        elsif parse_subject == 'modifier'
+          raise(
+            ::CSVPlusPlus::Error::ModifierSyntaxError.new(
+              "Unable to parse #{parse_subject} starting at",
+              bad_input: tokenizer.peek
+            )
+          )
+        else
+          raise(
+            ::CSVPlusPlus::Error::FormulaSyntaxError.new(
+              "Unable to parse #{parse_subject} starting at",
+              bad_input: tokenizer.peek
+            )
+          )
+        end
+      end
+      # rubocop:enable Metrics/AbcSize, Metrics/MethodLength
+    end
+  end
+end

data/lib/csv_plus_plus/lexer/tokenizer.rb

@@ -1,19 +1,33 @@
+# typed: strict
 # frozen_string_literal: true
 
-require 'strscan'
-
 module CSVPlusPlus
   module Lexer
     # A class that contains the use-case-specific regexes for parsing
     #
-    # @attr_reader last_token [String] The last token that's been matched.
-    # @attr_reader scanner [StringScanner] The StringScanner instance that's parsing the input.
+    # @attr_reader last_token [String, nil] The last token that's been matched.
     class Tokenizer
-      attr_reader :last_token, :scanner
+      extend ::T::Sig
+
+      sig { returns(::T.nilable(::CSVPlusPlus::Lexer::Token)) }
+      attr_reader :last_token
 
-      # @param input [String]
+      sig do
+        params(
+          tokens: ::T::Enumerable[::CSVPlusPlus::Lexer::Token],
+          catchall: ::T.nilable(::Regexp),
+          ignore: ::T.nilable(::Regexp),
+          alter_matches: ::T::Hash[::Symbol, ::T.proc.params(s: ::String).returns(::String)],
+          stop_fn: ::T.nilable(::T.proc.params(s: ::StringScanner).returns(::T::Boolean))
+        ).void
+      end
+      # @param tokens [Array<Regexp, String>] The list of tokens to scan
+      # @param catchall [Regexp] A final regexp to try if nothing else matches
+      # @param ignore [Regexp] Ignore anything matching this regexp
+      # @param alter_matches [Object] A map of matches to alter
+      # @param stop_fn [Proc] Stop parsing when this is true
       def initialize(tokens:, catchall: nil, ignore: nil, alter_matches: {}, stop_fn: nil)
-        @last_token = nil
+        @last_token = ::T.let(nil, ::T.nilable(::CSVPlusPlus::Lexer::Token))
 
         @catchall = catchall
         @ignore = ignore
@@ -22,67 +36,94 @@ module CSVPlusPlus
         @alter_matches = alter_matches
       end
 
+      sig { params(input: ::String).returns(::T.self_type) }
       # Initializers a scanner for the given input to be parsed
       #
       # @param input The input to be tokenized
+      #
       # @return [Tokenizer]
       def scan(input)
-        @scanner = ::StringScanner.new(input.strip)
+        @scanner = ::T.let(::StringScanner.new(input.strip), ::T.nilable(::StringScanner))
         self
       end
 
+      sig { returns(::StringScanner) }
+      # Returns the currently initialized +StringScanner+.  You must call +#scan+ first or else this will throw an
+      # exception.
+      #
+      # @return [StringScanner]
+      def scanner
+        # The caller needs to initialize this class with a call to #scan before we can do anything.  it sets up the
+        # +@scanner+ with it's necessary input.
+        unless @scanner
+          raise(::CSVPlusPlus::Error::CompilerError, 'Called Tokenizer#scanner without calling #scan first')
+        end
+
+        @scanner
+      end
+
+      sig { void }
       # Scan tokens and set +@last_token+ if any match
       #
       # @return [String, nil]
       def scan_tokens!
-        m = @tokens.find { |t| @scanner.scan(t.first) }
-        @last_token = m ? m[1] : nil
+        @last_token = @tokens.find { |t| scanner.scan(t.regexp) }
       end
 
+      sig { returns(::T.nilable(::String)) }
       # Scan input against the catchall pattern
       #
       # @return [String, nil]
       def scan_catchall
-        @scanner.scan(@catchall) if @catchall
+        scanner.scan(@catchall) if @catchall
       end
 
+      sig { returns(::T.nilable(::String)) }
       # Scan input against the ignore pattern
       #
       # @return [boolean]
       def matches_ignore?
-        @scanner.scan(@ignore) if @ignore
+        scanner.scan(@ignore) if @ignore
       end
 
+      sig { returns(::T.nilable(::String)) }
       # The value of the last token matched
       #
       # @return [String, nil]
       def last_match
-        return @alter_matches[@last_token].call(@scanner.matched) if @alter_matches.key?(@last_token)
+        # rubocop:disable Style/MissingElse
+        if @last_token && @alter_matches.key?(@last_token.token.to_sym)
+          # rubocop:enable Style/MissingElse
+          return ::T.must(@alter_matches[@last_token.token.to_sym]).call(scanner.matched)
+        end
 
-        @scanner.matched
+        scanner.matched
       end
 
+      sig { params(peek_characters: ::Integer).returns(::String) }
       # Read the input but don't consume it
       #
       # @param peek_characters [Integer]
       #
       # @return [String]
       def peek(peek_characters: 100)
-        @scanner.peek(peek_characters)
+        scanner.peek(peek_characters)
       end
 
+      sig { returns(::T::Boolean) }
       # Scan for our stop token (if there is one - some parsers stop early and some don't)
       #
       # @return [boolean]
       def stop?
-        @stop_fn ? @stop_fn.call(@scanner) : false
+        @stop_fn ? @stop_fn.call(scanner) : false
       end
 
+      sig { returns(::String) }
       # The rest of the un-parsed input.  The tokenizer might not need to parse the entire input
       #
       # @return [String]
       def rest
-        @scanner.rest
+        scanner.rest
       end
     end
   end

data/lib/csv_plus_plus/lexer.rb

@@ -1,14 +1,77 @@
+# typed: strict
 # frozen_string_literal: true
 
-require_relative './lexer/lexer'
+require_relative './lexer/racc_lexer'
 require_relative './lexer/tokenizer'
 
 module CSVPlusPlus
+  # Code for tokenizing a csvpp file
   module Lexer
+    extend ::T::Sig
+
+    # A token that's matched by +regexp+ and presented with +token+
+    class Token < ::T::Struct
+      const :regexp, ::Regexp
+      const :token, ::T.any(::String, ::Symbol)
+    end
+
     END_OF_CODE_SECTION = '---'
     public_constant :END_OF_CODE_SECTION
 
     VARIABLE_REF = '$$'
     public_constant :VARIABLE_REF
+
+    # @see https://github.com/ruby/racc/blob/master/lib/racc/parser.rb#L121
+    TOKEN_LIBRARY = ::T.let(
+      {
+        # A1_NOTATION: ::CSVPlusPlus::Lexer::Token.new(
+        # regexp: ::CSVPlusPlus::A1Reference::A1_NOTATION_REGEXP, token: :A1_NOTATION
+        # ),
+        FALSE: ::CSVPlusPlus::Lexer::Token.new(regexp: /false/i, token: :FALSE),
+        HEX_COLOR: ::CSVPlusPlus::Lexer::Token.new(regexp: ::CSVPlusPlus::Color::HEX_STRING_REGEXP, token: :HEX_COLOR),
+        INFIX_OP: ::CSVPlusPlus::Lexer::Token.new(regexp: %r{\^|\+|-|\*|/|&|<|>|<=|>=|<>}, token: :INFIX_OP),
+        NUMBER: ::CSVPlusPlus::Lexer::Token.new(regexp: /-?[\d.]+/, token: :NUMBER),
+        REF: ::CSVPlusPlus::Lexer::Token.new(regexp: /[$!\w:]+/, token: :REF),
+        STRING: ::CSVPlusPlus::Lexer::Token.new(
+          regexp: %r{"(?:[^"\\]|\\(?:["\\/bfnrt]|u[0-9a-fA-F]{4}))*"},
+          token: :STRING
+        ),
+        TRUE: ::CSVPlusPlus::Lexer::Token.new(regexp: /true/i, token: :TRUE),
+        VAR_REF: ::CSVPlusPlus::Lexer::Token.new(regexp: /\$\$/, token: :VAR_REF)
+      }.freeze,
+      ::T::Hash[::Symbol, ::CSVPlusPlus::Lexer::Token]
+    )
+    public_constant :TOKEN_LIBRARY
+
+    sig { params(str: ::String).returns(::String) }
+    # Run any transformations to the input before going into the CSV parser
+    #
+    # The CSV parser in particular does not like if there is whitespace after a double quote and before the next comma
+    #
+    # @param str [String]
+    # @return [String]
+    def self.preprocess(str)
+      str.gsub(/"\s*,/, '",')
+    end
+
+    sig { params(str: ::String).returns(::String) }
+    # When parsing a modifier with a quoted string field, we need a way to unescape.  Some examples of quoted and
+    # unquoted results:
+    #
+    # * "just a string" => "just a string"
+    # * "' this is a string'" => "this is a string"
+    # * "won\'t this work?" => "won't this work"
+    #
+    # @param str [::String]
+    #
+    # @return [::String]
+    def self.unquote(str)
+      # could probably do this with one regex but we do it in 3 steps:
+      #
+      # 1. remove leading and trailing spaces and '
+      # 2. remove any backslashes that are by themselves (none on either side)
+      # 3. turn double backslashes into singles
+      str.gsub(/^\s*'?|'?\s*$/, '').gsub(/([^\\]+)\\([^\\]+)/, '\1\2').gsub(/\\\\/, '\\')
+    end
   end
 end

data/lib/csv_plus_plus/modifier/conditional_formatting.rb

@@ -1,3 +1,4 @@
+# typed: true
 # frozen_string_literal: true
 
 module CSVPlusPlus

data/lib/csv_plus_plus/modifier/data_validation.rb

@@ -0,0 +1,138 @@
+# typed: true
+# frozen_string_literal: true
+
+module CSVPlusPlus
+  module Modifier
+    # A validation on a cell value.  Used to support the `validate=` modifier directive.  This is mostly based on the
+    # Google Sheets API spec:
+    #
+    # @see https://developers.google.com/sheets/api/reference/rest/v4/spreadsheets/other#ConditionType
+    #
+    # @attr_reader arguments [Array<::String>] The parsed arguments as required by the condition.
+    # @attr_reader condition [Symbol] The condition (:blank, :text_eq, :date_before, etc.)
+    # @attr_reader invalid_reason [::String, nil] If set, the reason why this modifier is not valid.
+    class DataValidation
+      attr_reader :arguments, :condition, :invalid_reason
+
+      # @param value [::String] The value to parse as a data validation
+      def initialize(value)
+        condition, args = value.split(/\s*:\s*/)
+        @arguments = ::CSVPlusPlus::Lexer.unquote(args || '').split(/\s+/)
+        @condition = condition.to_sym
+
+        validate!
+      end
+
+      # Each data validation (represented by +@condition+) has their own requirements for which arguments are valid.
+      # If this object is invalid, you can see the reason in +@invalid_reason+.
+      #
+      # @return [boolean]
+      def valid?
+        @invalid_reason.nil?
+      end
+
+      protected
+
+      def invalid!(reason)
+        @invalid_reason = reason
+      end
+
+      def a_number(arg)
+        Float(arg)
+      rescue ::ArgumentError
+        invalid!("Requires a number but given: #{arg}")
+      end
+
+      def a1_notation(arg)
+        return arg if ::CSVPlusPlus::A1Reference.valid_cell_reference?(arg)
+      end
+
+      def a_date(arg, allow_relative_date: false)
+        return arg if ::CSVPlusPlus::Entities::Date.valid_date?(arg)
+
+        if allow_relative_date
+          a_relative_date(arg)
+        else
+          invalid!("Requires a date but given: #{arg}")
+        end
+      end
+
+      def a_relative_date(arg)
+        return arg if %w[past_month past_week past_year yesterday today tomorrow].include?(arg.downcase)
+
+        invalid!('Requires a relative date: past_month, past_week, past_year, yesterday, today or tomorrow')
+      end
+
+      def no_args
+        return if @arguments.empty?
+
+        invalid!("Requires no arguments but #{@arguments.length} given: #{@arguments}")
+      end
+
+      def one_arg
+        return @arguments[0] if @arguments.length == 1
+
+        invalid!("Requires only one argument but #{@arguments.length} given: #{@arguments}")
+      end
+
+      def one_arg_or_more
+        return @arguments if @arguments.length.positive?
+
+        invalid!("Requires at least one argument but #{@arguments.length} given: #{@arguments}")
+      end
+
+      def two_dates
+        return @arguments if @arguments.length == 2 && a_date(@arguments[0]) && a_date(@arguments[1])
+
+        invalid!("Requires exactly two dates but given: #{@arguments}")
+      end
+
+      def two_numbers
+        return @arguments if @arguments.length == 2 && a_number(@arguments[0]) && a_number(@arguments[1])
+
+        invalid!("Requires exactly two numbers but given: #{@arguments}")
+      end
+
+      # validate_boolean is a weird one because it can have 0, 1 or 2 @arguments - all of them must be (true | false)
+      def validate_boolean
+        return @arguments if @arguments.empty?
+
+        converted_args = @arguments.map(&:strip).map(&:downcase)
+        return @arguments if [1, 2].include?(@arguments.length) && converted_args.all? do |arg|
+                               %w[true false].include?(arg)
+                             end
+
+        invalid!("Requires 0, 1 or 2 arguments and they all must be either 'true' or 'false'. Received: #{arguments}")
+      end
+
+      # rubocop:disable Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/AbcSize
+      def validate!
+        case condition.to_sym
+        when :blank, :date_is_valid, :not_blank, :text_is_email, :text_is_url
+          no_args
+        when :text_contains, :text_ends_with, :text_eq, :text_not_contains, :text_starts_with
+          one_arg
+        when :date_after, :date_before, :date_on_or_after, :date_on_or_before
+          a_date(one_arg, allow_relative_date: true)
+        when :date_eq, :date_not_eq
+          a_date(one_arg)
+        when :date_between, :date_not_between
+          two_dates
+        when :one_of_range
+          a1_notation(one_arg)
+        when :custom_formula, :one_of_list, :text_not_eq
+          one_arg_or_more
+        when :number_eq, :number_greater, :number_greater_than_eq, :number_less, :number_less_than_eq, :number_not_eq
+          a_number(one_arg)
+        when :number_between, :number_not_between
+          two_numbers
+        when :boolean
+          validate_boolean
+        else
+          invalid!('Not a recognized data validation directive')
+        end
+      end
+      # rubocop:enable Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/AbcSize
+    end
+  end
+end

data/lib/csv_plus_plus/modifier/expand.rb

@@ -0,0 +1,78 @@
+# typed: strict
+# frozen_string_literal: true
+
+module CSVPlusPlus
+  module Modifier
+    # The logic for how a row can expand
+    #
+    # @attr_reader ends_at [Integer, nil] Once the row has been expanded, where it ends at.
+    # @attr_reader repetitions [Integer, nil] How many times the row repeats/expands.
+    # @attr_reader starts_at [Integer, nil] Once the row has been expanded, where it starts at.
+    class Expand
+      extend ::T::Sig
+
+      sig { returns(::T.nilable(::Integer)) }
+      attr_reader :ends_at
+
+      sig { returns(::T.nilable(::Integer)) }
+      attr_reader :repetitions
+
+      sig { returns(::T.nilable(::Integer)) }
+      attr_reader :starts_at
+
+      sig { params(repetitions: ::T.nilable(::Integer), starts_at: ::T.nilable(::Integer)).void }
+      # @param repetitions [Integer, nil] How many times this expand repeats.  If it's +nil+ it will expand infinitely
+      #   (for the rest of the worksheet.)
+      # @param starts_at [Integer, nil] The final location where the +Expand+ will start.  It's important to note that
+      #   this can't be derived until all rows are expanded, because each expand modifier will push down the ones below
+      #   it.  So typically this param will not be passed in the initializer but instead set later.
+      def initialize(repetitions: nil, starts_at: nil)
+        @repetitions = ::T.let(repetitions, ::T.nilable(::Integer))
+        @starts_at = ::T.let(starts_at, ::T.nilable(::Integer)) unless starts_at.nil?
+        @ends_at = ::T.let(nil, ::T.nilable(::Integer))
+      end
+
+      sig { returns(::T::Boolean) }
+      # Has the row been expanded?
+      #
+      # @return [boolean]
+      def expanded?
+        !@starts_at.nil?
+      end
+
+      sig { returns(::T::Boolean) }
+      # Does this infinitely expand?
+      #
+      # @return [boolean]
+      def infinite?
+        repetitions.nil?
+      end
+
+      sig { params(row_index: ::Integer).void }
+      # Mark the start of the row once it's been expanded, as well as where it +ends_at+.  When expanding rows each one
+      # adds rows to the worksheet and if there are multiple `expand` modifiers in the worksheet, we don't know the
+      # final +row_index+ until we're in the phase of expanding all the rows out.
+      def starts_at=(row_index)
+        @starts_at = row_index
+        @ends_at = row_index + @repetitions unless @repetitions.nil?
+      end
+
+      sig { params(position: ::CSVPlusPlus::Runtime::Position).returns(::T::Boolean) }
+      # Does the given +position+ fall within this expand?
+      #
+      # @param position [Runtime::Position]
+      #
+      # @return [boolean]
+      def position_within?(position)
+        unless starts_at
+          raise(
+            ::CSVPlusPlus::Error::CompilerError,
+            'Must call Template.expand_rows! before checking the scope of expands.'
+          )
+        end
+
+        position.row_index >= ::T.must(starts_at) && (ends_at.nil? || position.row_index <= ::T.must(ends_at))
+      end
+    end
+  end
+end