nebulous 0.0.2 → 0.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: e16fcc8e895cd4078d1ea39ab8553ee3246cd5fa
-  data.tar.gz: 46b62d5133aff6cea1cec0370af683834f16fe28
+  metadata.gz: 15734d97663b72356a0c85a85a07e5bddc191a0f
+  data.tar.gz: a2fdc1e48d2835bc8a6e509733cbbb5a09ed9bdd
 SHA512:
-  metadata.gz: 7dd3e249cb5b64e7477c85bd0e34f44750b14d02ae0fafaeb6a33e93decab821c6798b66bfccfd87a10811c1115a008a94f4786c34ab1e72fd87c448aeeab705
-  data.tar.gz: 8791db72af0fd20b251382fd88982abd0af2a60c6bee83890efdfc8ac68576789783c9a6c1395ebf4c234a14f3db8f3a5dfe6d6d697b2c2d24e34aa0b82ab0b6
+  metadata.gz: d98c33418a5c0c497027d57491e8e3698357bd35762b9c11efe67e8d97dce7458693b4129d52516d155e498ff3023b825c6c04e224e679e5f681fa17886dfd3f
+  data.tar.gz: 43845b83b364c806fa803affe6b5c8981af2286a68e0ca157d321c63328b0c27ddfb4e7d4f451ba82e5bb5dff7ba617513282c1ef8aa1258f9b106c2bd885862

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    nebulous (0.0.1)
+    nebulous (0.0.3)
       activesupport
       cocaine (~> 0.5)
 

data/README.md

@@ -18,7 +18,7 @@ Or install it yourself as:
 
 ## Usage
 
-TBD
+
 
 ## Contributing
 

data/lib/nebulous.rb

@@ -3,10 +3,14 @@ require 'ostruct'
 require 'cocaine'
 require 'active_support/all'
 require 'nebulous/version'
-require 'nebulous/parser'
+require 'nebulous/delimiter_detector'
 require 'nebulous/row'
 require 'nebulous/chunk'
-require 'nebulous/delimiter_detector'
+require 'nebulous/input'
+require 'nebulous/input/reader'
+require 'nebulous/input/parsing'
+require 'nebulous/input/delimiters'
+require 'nebulous/parser'
 
 module Nebulous
   def self.process(file, *args, &block)

data/lib/nebulous/input.rb

@@ -0,0 +1,4 @@
+module Nebulous
+  module Input
+  end
+end

data/lib/nebulous/input/delimiters.rb

@@ -0,0 +1,16 @@
+module Nebulous
+  module Input
+    module Delimiters
+      def delimiters
+        @delimiters ||= Nebulous::DelimiterDetector.new(file.path).detect
+      end
+
+      private
+
+      def merge_delimiters
+        options.row_sep ||= delimiters[:row_sep]
+        options.col_sep ||= delimiters[:col_sep]
+      end
+    end
+  end
+end

data/lib/nebulous/input/parsing.rb

@@ -0,0 +1,56 @@
+module Nebulous
+  module Input
+    module Parsing
+      def parse_row
+        sequence
+        Row.parse(read_complete_line, options).to_numeric.merge(@headers)
+      end
+
+      def read_headers
+        @headers ||= Row.headers(readline, options) if options[:headers]
+      end
+
+      def chunk
+        @chunk ||= Chunk.new chunk_options
+      end
+
+      def sequence
+        @index += 1
+      end
+
+      def limit?
+        options.limit && options.limit == @index
+      end
+
+      def yield_chunk(chunk, &_block)
+        if chunk.full? || file.eof?
+          yield chunk.map(&:to_a)
+          @chunk = nil
+        end
+      end
+
+      def iterate(&block)
+        while !file.eof?
+          break if limit?
+          chunk << replace_keys(parse_row)
+          yield_chunk(chunk, &block) if block_given? && options.chunk
+        end
+
+        @chunk.to_a
+      end
+
+      def replace_keys(row)
+        return row unless options.mapping
+        row.map do |key, value|
+          [options.mapping[key], value] if options.mapping.has_key?(key)
+        end.compact.to_h
+      end
+
+      def chunk_options
+        Hash.new.tap do |attrs|
+          attrs[:size] = options.chunk.to_i if options.chunk
+        end
+      end
+    end
+  end
+end

data/lib/nebulous/input/reader.rb

@@ -0,0 +1,29 @@
+module Nebulous
+  module Input
+    module Reader
+      def read_input(input)
+        input.respond_to?(:readline) ? input : File.open(input, "r:#{encoding}")
+      end
+
+      def read_complete_line
+        ln = readline
+        while ln.count(options.quote_char) % 2 == 1
+          ln += readline
+        end
+        ln
+      end
+
+      def readline
+        file.readline(line_terminator).encode(encoding, invalid: :replace).chomp
+      end
+
+      def line_terminator
+        options.row_sep
+      end
+
+      def encoding
+        options.encoding
+      end
+    end
+  end
+end

data/lib/nebulous/parser.rb

@@ -1,5 +1,9 @@
 module Nebulous
   class Parser
+    include Nebulous::Input::Reader
+    include Nebulous::Input::Parsing
+    include Nebulous::Input::Delimiters
+
     DEFAULT_OPTIONS = {
       col_sep: nil,
       row_sep: nil,
@@ -9,7 +13,6 @@ module Nebulous
       headers: true,
       mapping: nil,
       limit: false,
-      remove_empty_values: true,
       encoding: Encoding::UTF_8.to_s
     }
 
@@ -34,10 +37,6 @@ module Nebulous
       file.rewind
     end
 
-    def delimiters
-      @delimiters ||= DelimiterDetector.new(file.path).detect
-    end
-
     private
 
     def reset
@@ -45,85 +44,5 @@ module Nebulous
       @headers = nil
       @chunk = nil
     end
-
-    def chunk
-      @chunk ||= Chunk.new chunk_options
-    end
-
-    def read_headers
-      @headers ||= Row.headers(readline, options) if options[:headers]
-    end
-
-    def iterate(&block)
-      while !file.eof?
-        break if limit?
-        chunk << replace_keys(parse_row.merge(@headers))
-        yield_chunk(chunk, &block) if block_given? && options.chunk
-      end
-
-      @chunk.to_a
-    end
-
-    def sequence
-      @index += 1
-    end
-
-    def limit?
-      options.limit && options.limit == @index
-    end
-
-    def parse_row
-      sequence
-      Row.parse(read_complete_line, options)
-    end
-
-    def yield_chunk(chunk, &_block)
-      if chunk.full? || file.eof?
-        yield chunk.map(&:to_a)
-        @chunk = nil
-      end
-    end
-
-    def read_input(input)
-      input.respond_to?(:readline) ? input : File.open(input, "r:#{encoding}")
-    end
-
-    def read_complete_line
-      ln = readline
-      while ln.count(options.quote_char) % 2 == 1
-        ln += readline
-      end
-      ln
-    end
-
-    def readline
-      file.readline(line_terminator).encode(encoding, invalid: :replace).chomp
-    end
-
-    def encoding
-      options.encoding
-    end
-
-    def merge_delimiters
-      options.row_sep ||= delimiters[:row_sep]
-      options.col_sep ||= delimiters[:col_sep]
-    end
-
-    def line_terminator
-      options.row_sep
-    end
-
-    def chunk_options
-      Hash.new.tap do |attrs|
-        attrs[:size] = options.chunk.to_i if options.chunk
-      end
-    end
-
-    def replace_keys(row)
-      return row unless options.mapping
-      row.map do |key, value|
-        [options.mapping[key], value] if options.mapping.has_key?(key)
-      end.compact.to_h
-    end
   end
 end

data/lib/nebulous/row.rb

@@ -9,26 +9,27 @@ module Nebulous
     end
 
     def self.parse(str, opts)
-      str.gsub!(opts.comment_exp, '')
+      opts = opts.to_h
+      str.gsub!(opts[:comment_exp], '')
       str.chomp!
 
       begin
-        args = opts.to_h.slice(:col_sep, :row_sep, :quote_char)
+        args = opts.slice(:col_sep, :row_sep, :quote_char)
         data = CSV.parse_line str, args
       rescue CSV::MalformedCSVError
-        exp = /(#{opts.col_sep})(?=(?:[^"]|"[^"]*")*$)/
+        exp = /(#{opts[:col_sep]})(?=(?:[^"]|"[^"]*")*$)/
         data = str.gsub(exp, "\0").split(/\0/)
       end
 
       data.map!(&:strip)
-      new(data).to_numeric
+      new(data)
     end
 
     def to_numeric
       arr = map do |val|
         case val
         when /^[+-]?\d+\.\d+$/
-          val.to_i
+          val.to_f
         when /^[+-]?\d+$/
           val.to_i
         else

data/lib/nebulous/version.rb

@@ -1,3 +1,3 @@
 module Nebulous
-  VERSION = '0.0.2'.freeze
+  VERSION = '0.0.4'.freeze
 end

data/spec/nebulous/chunk_spec.rb

@@ -1,6 +1,30 @@
 require 'spec_helper'
 
 describe Nebulous::Chunk do
-  context 'around batches of csv data' do
+  context 'around chunk of csv data' do
+    subject { Nebulous::Chunk }
+
+    context '#full?' do
+      let(:chunk) { subject.new size: size }
+
+      before do
+        chunk << ['row']
+        chunk << ['row']
+      end
+
+      context 'when not full' do
+        let(:size) { 3 }
+        it 'returns expected value' do
+          expect(chunk.full?).to be_falsy
+        end
+      end
+
+      context 'when full' do
+        let(:size) { 2 }
+        it 'returns expected value' do
+          expect(chunk.full?).to be_truthy
+        end
+      end
+    end
   end
 end

data/spec/nebulous/row_spec.rb

@@ -2,5 +2,70 @@ require 'spec_helper'
 
 describe Nebulous::Row do
   context 'around reading csv rows' do
+    subject { Nebulous::Row }
+
+    let(:col_sep) { ',' }
+    let(:row_sep) { "\n" }
+    let(:options) do
+      { col_sep: col_sep, row_sep: row_sep, quote_char: '"', comment_exp: /^#/ }
+    end
+
+    context '::headers' do
+      it 'parses and normalizes a csv string as headers' do
+        headers = subject.headers("First name, last-name, guests", options)
+        expect(headers).to eq(
+          {first_name: :first_name, last_name: :last_name, guests: :guests}
+        )
+      end
+    end
+
+    context '::parse' do
+      context 'with valid csv' do
+        it 'returns expected parsed result' do
+          row = subject.parse "raw denim, Austin,selvage,artisan", options
+          expect(row).to eq ["raw denim", "Austin", "selvage", "artisan"]
+        end
+      end
+
+      context 'with valid tsv' do
+        let(:col_sep) { "\t" }
+        it 'returns expected parsed result' do
+          row = subject.parse "raw denim\tAustin\t selvage\tartisan", options
+          expect(row).to eq ["raw denim", "Austin", "selvage", "artisan"]
+        end
+      end
+
+      context 'with malformed csv' do
+        it 'returns expected parsed result' do
+          row = subject.parse 'raw denim, Austin "TX, US", artisan', options
+          expect(row).to eq ["raw denim", "Austin \"TX, US\"", "artisan"]
+        end
+      end
+
+      context 'with malformed tsv' do
+        let(:col_sep) { "\t" }
+        it 'returns expected parsed result' do
+          row = subject.parse "raw denim\t Austin \"TX, US\"\t artisan", options
+          expect(row).to eq ["raw denim", "Austin \"TX, US\"", "artisan"]
+        end
+      end
+    end
+
+    context '#to_numeric' do
+      it 'converts numeric values to ints/floats' do
+        row = subject.new ["1", "two", "3", "4.5"]
+        expect(row.to_numeric).to eq [1, "two", 3, 4.5]
+      end
+    end
+
+    context '#merge' do
+      it 'zips a row with provided headers' do
+        headers = subject.headers "first name, last name", options
+        row = subject.new ["bob", "barker"]
+        expect(row.merge(headers)).to eq(
+          { first_name: "bob", last_name: "barker" }
+        )
+      end
+    end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: nebulous
 version: !ruby/object:Gem::Version
-  version: 0.0.2
+  version: 0.0.4
 platform: ruby
 authors:
 - Zach Graves
@@ -112,6 +112,10 @@ files:
 - lib/nebulous.rb
 - lib/nebulous/chunk.rb
 - lib/nebulous/delimiter_detector.rb
+- lib/nebulous/input.rb
+- lib/nebulous/input/delimiters.rb
+- lib/nebulous/input/parsing.rb
+- lib/nebulous/input/reader.rb
 - lib/nebulous/parser.rb
 - lib/nebulous/row.rb
 - lib/nebulous/version.rb