masticate 0.2.1 → 0.2.2

data/lib/masticate/base.rb

@@ -6,7 +6,7 @@ class Masticate::Base
 
   def initialize(args)
     case args
-    when String
+    when String, nil
       @filename = args
     when Hash
       configure(args)

data/lib/masticate/csvify.rb

@@ -13,7 +13,10 @@ class Masticate::Csvify < Masticate::Base
     with_input do |input|
       while line = get
         row = CSV.parse_line(line, csv_options)
-        emit(row.to_csv) if row
+        if row
+          row = row.map(&:strip)
+          emit(row)
+        end
       end
     end
     @output.close if opts[:output]

data/lib/masticate/datify.rb

@@ -14,7 +14,11 @@ class Masticate::Datify < Masticate::Base
 
   def crunch(row)
     if !@index
-      @index = row.index(@field) or raise "Unable to find column '#{@field}'"
+      if @field.is_a?(Fixnum) || @field =~ /^\d+/
+        @index = @field.to_i
+      else
+        @index = row.index(@field) or raise "Unable to find column '#{@field}'"
+      end
     elsif row
       ts = DateTime.strptime(row[@index], @format).to_time
       row[@index] = ts.to_i rescue nil

data/lib/masticate/exclude.rb

@@ -0,0 +1,32 @@
+# exclude rows based on field = value
+
+class Masticate::Exclude < Masticate::Base
+  def configure(opts)
+    standard_options(opts)
+
+    @field = opts[:field] or raise "missing field to exclude"
+    @value = opts[:value] or raise "missing value to exclude"
+
+    # row-loading automatically strips leading & trailing whitespace and converts blanks to nils,
+    # so when looking for blanks need to compare to nil instead of ''
+    @value = nil if @value.empty?
+  end
+
+  def exclude(opts)
+    execute(opts)
+  end
+
+  def crunch(row)
+    if !@headers
+      @headers = row
+      @index = @headers.index(@field) or raise "Unable to find column '#{@field}' in headers"
+      row
+    elsif row
+      if row[@index] == @value
+        # exclude
+      else
+        row
+      end
+    end
+  end
+end

data/lib/masticate/gsubber.rb

@@ -1,5 +1,4 @@
-# extract subset of columns from CSV
-require "csv"
+# apply regex transformation to a field
 
 class Masticate::Gsubber < Masticate::Base
   def configure(opts)

data/lib/masticate/myoptparse.rb

@@ -37,6 +37,10 @@ class Masticate::MyOptionParser
         @options[:field] = f
       end
 
+      opts.on("--value VALUE", String, "(*exclude* only) Value to compare field to for exclusion") do |s|
+        @options[:value] = s
+      end
+
       opts.on("--snip DIRECTIVE", String, "Specify header fields to snip: first N, or by name") do |f|
         @options[:snip] = f.to_i
       end
@@ -85,7 +89,8 @@ class Masticate::MyOptionParser
       'datify' => Masticate::Datify,
       'maxrows' => Masticate::MaxRows,
       'relabel' => Masticate::Relabel,
-      'pluck' => Masticate::Plucker
+      'pluck' => Masticate::Plucker,
+      'exclude' => Masticate::Exclude
     }
 
     klass = klasses[command]
@@ -145,6 +150,10 @@ EOT
       results = Masticate.cook(filename, options)
       logmessage(command, options, results)
 
+    when 'exclude'
+      results = Masticate.exclude(filename, options)
+      logmessage(command, options, results)
+
     else
       raise "unknown command #{command}"
     end

data/lib/masticate/plucker.rb

@@ -17,16 +17,15 @@ class Masticate::Plucker < Masticate::Base
       @headers = row
       @indexes = @fields.map do |f|
         case f
-        when String
-          row.index(f) or raise "Unable to find column '#{f}'"
-        when Fixnum
+        when Fixnum, /^\d+/
+          f = f.to_i
           if f > row.count
             raise "Cannot pluck column #{f}, there are only #{row.count} fields"
           else
             f-1
           end
         else
-          raise "Invalid field descriptor '#{f}'"
+          row.index(f) or raise "Unable to find column '#{f}'"
         end
       end
       @indexes.map {|i| row[i]}

data/lib/masticate/version.rb

@@ -1,3 +1,3 @@
 module Masticate
-  VERSION = "0.2.1"
+  VERSION = "0.2.2"
 end

data/lib/masticate.rb

@@ -14,6 +14,8 @@ require_relative "masticate/gsubber"
 require_relative "masticate/max_rows"
 require_relative "masticate/concat"
 require_relative "masticate/relabel"
+require_relative "masticate/exclude"
+
 require_relative "masticate/cook"
 
 module Masticate
@@ -53,6 +55,10 @@ module Masticate
     Relabel.new(filename).relabel(opts)
   end
 
+  def self.exclude(filename, opts)
+    Exclude.new(filename).exclude(opts)
+  end
+
   def self.cook(filename, opts)
     Cook.new(filename).cook(opts)
   end

data/spec/data/exclude_input.csv

@@ -0,0 +1,7 @@
+ID,DATE,VALUE
+,,594787
+1,201201060826,594823
+2,201201060521,594790
+7,201201060429,594780
+,201201060446,594786
+8,201201061011,594857

data/spec/data/exclude_results.csv

@@ -0,0 +1,5 @@
+ID,DATE,VALUE
+1,201201060826,594823
+2,201201060521,594790
+7,201201060429,594780
+8,201201061011,594857

data/spec/lib/exclude_spec.rb

@@ -0,0 +1,15 @@
+# spec for row exclusion
+
+require "spec_helper"
+
+describe "exclude" do
+  it "should be able to ignore rows with blank fields" do
+    filename = File.dirname(__FILE__) + "/../data/exclude_input.csv"
+    tmp = Tempfile.new('exclude')
+    results = Masticate.exclude(filename, :output => tmp, :field => 'ID', :value => '')
+    output = File.read(tmp)
+    correct_output = File.read(File.dirname(__FILE__) + "/../data/exclude_results.csv")
+
+    output.should == correct_output
+  end
+end

data/spec/lib/plucker_spec.rb

@@ -7,7 +7,7 @@ describe "plucker" do
   it "should pull named columns" do
     filename = File.dirname(__FILE__) + "/../data/namedcols.csv"
     tmp = Tempfile.new('plucker')
-    results = Masticate.pluck(filename, :output => tmp, :fields => ['three', 'five'])
+    results = Masticate.pluck(filename, :output => tmp, :fields => ['three', '5'])
     output = File.read(tmp)
     correct_output = File.read(File.dirname(__FILE__) + "/../data/namedcols.csv.output")
     tmp.unlink

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: masticate
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.2.2
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-04-23 00:00:00.000000000 Z
+date: 2012-04-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rspec
-  requirement: &2157087600 !ruby/object:Gem::Requirement
+  requirement: &2160837420 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -21,10 +21,10 @@ dependencies:
         version: 2.9.0
   type: :development
   prerelease: false
-  version_requirements: *2157087600
+  version_requirements: *2160837420
 - !ruby/object:Gem::Dependency
   name: guard-rspec
-  requirement: &2157087060 !ruby/object:Gem::Requirement
+  requirement: &2160836340 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -32,10 +32,10 @@ dependencies:
         version: 0.7.0
   type: :development
   prerelease: false
-  version_requirements: *2157087060
+  version_requirements: *2160836340
 - !ruby/object:Gem::Dependency
   name: ruby_gntp
-  requirement: &2157086580 !ruby/object:Gem::Requirement
+  requirement: &2160827420 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -43,7 +43,7 @@ dependencies:
         version: 0.3.4
   type: :development
   prerelease: false
-  version_requirements: *2157086580
+  version_requirements: *2160827420
 description: Data file crunching
 email:
 - jmay@pobox.com
@@ -65,6 +65,7 @@ files:
 - lib/masticate/cook.rb
 - lib/masticate/csvify.rb
 - lib/masticate/datify.rb
+- lib/masticate/exclude.rb
 - lib/masticate/gsubber.rb
 - lib/masticate/max_rows.rb
 - lib/masticate/mender.rb
@@ -82,6 +83,8 @@ files:
 - spec/data/datify_input.txt
 - spec/data/events.csv
 - spec/data/events_reduced.csv
+- spec/data/exclude_input.csv
+- spec/data/exclude_results.csv
 - spec/data/inlined_headers.csv
 - spec/data/inlined_headers.csv.output
 - spec/data/junk_header.csv
@@ -98,6 +101,7 @@ files:
 - spec/lib/cook_spec.rb
 - spec/lib/csvify_spec.rb
 - spec/lib/datify_spec.rb
+- spec/lib/exclude_spec.rb
 - spec/lib/gsub_spec.rb
 - spec/lib/maxrow_spec.rb
 - spec/lib/mender_spec.rb
@@ -138,6 +142,8 @@ test_files:
 - spec/data/datify_input.txt
 - spec/data/events.csv
 - spec/data/events_reduced.csv
+- spec/data/exclude_input.csv
+- spec/data/exclude_results.csv
 - spec/data/inlined_headers.csv
 - spec/data/inlined_headers.csv.output
 - spec/data/junk_header.csv
@@ -154,6 +160,7 @@ test_files:
 - spec/lib/cook_spec.rb
 - spec/lib/csvify_spec.rb
 - spec/lib/datify_spec.rb
+- spec/lib/exclude_spec.rb
 - spec/lib/gsub_spec.rb
 - spec/lib/maxrow_spec.rb
 - spec/lib/mender_spec.rb