build processors for reprojection of shapefiles

.travis.yml

@@ -8,4 +8,9 @@ env:
   global:
   - NOKOGIRI_USE_SYSTEM_LIBRARIES=true
 before_script:
-  - jdk_switcher use oraclejdk8
+  - jdk_switcher use oraclejdk8
+  - gdalinfo --version
+addons:
+  apt:
+    packages:
+    - gdal-bin

lib/geo_concerns/processors/gdal_projection.rb

@@ -0,0 +1,44 @@
+require 'open-uri'
+
+module GeoConcerns::Processors
+  class GdalProjection < Hydra::Derivatives::Processors::Processor
+    include Hydra::Derivatives::Processors::ShellBasedProcessor
+    attr_reader :source_path, :directives
+
+    def self.process(source_path, directives)
+      new(source_path, directives).reproject # XXX: clean up after ourselves
+    end
+
+    def initialize(source_path, directives)
+      fail ArgumentError unless source_path.ends_with?('.tif')
+
+      @source_path = source_path
+      @directives = directives
+
+      @output_filename = output_file(File.extname(source_path).gsub(/^\./, ''))
+      @output_directory = File.dirname(@output_filename)
+
+      @srid ||= directives.fetch(:srid, 4326)
+
+      FileUtils.mkdir_p(@output_directory) unless File.directory?(@output_directory)
+    end
+
+    def reproject
+      gdalwarp
+    end
+
+    # reproject via GDAL, @see http://www.gdal.org/gdalwarp.html
+    def gdalwarp(resample = 'bilinear')
+      tempfn = File.join(@output_directory, File.basename(source_path))
+      # reproject with gdalwarp (must uncompress here to prevent bloat)
+      self.class.execute "gdalwarp -q -r #{resample} -t_srs EPSG:#{@srid} #{source_path} #{tempfn} -co 'COMPRESS=NONE'"
+      fail "gdalwarp failed to create #{tempfn}" unless File.size?(tempfn)
+
+      # compress tempfn with gdal_translate
+      self.class.execute "gdal_translate -q -a_srs EPSG:#{@srid} #{tempfn} #{@output_filename} -co 'COMPRESS=LZW'"
+      fail "gdal_translate failed to create #{@output_filename}" unless File.size?(@output_filename)
+    ensure
+      FileUtils.rm_f(tempfn)
+    end
+  end
+end

lib/geo_concerns/processors/ogr_projection.rb

@@ -0,0 +1,46 @@
+require 'open-uri'
+
+module GeoConcerns::Processors
+  class OgrProjection < Hydra::Derivatives::Processors::Processor
+    include Hydra::Derivatives::Processors::ShellBasedProcessor
+    attr_reader :source_path, :directives
+
+    def self.process(source_path, directives)
+      new(source_path, directives).reproject # XXX: clean up after ourselves
+    end
+
+    def initialize(source_path, directives)
+      fail ArgumentError unless source_path.ends_with?('.shp')
+
+      @source_path = source_path
+      @directives = directives
+
+      @output_filename = output_file(File.extname(source_path).gsub(/^\./, ''))
+      @output_directory = File.dirname(@output_filename)
+
+      FileUtils.mkdir_p(@output_directory) unless File.directory?(@output_directory)
+    end
+
+    def reproject
+      ogr2ogr
+      normalize_prj_file
+    end
+
+    def prettywkt
+      @srid ||= directives.fetch(:srid, 4326)
+      @prettywkt ||= open("http://spatialreference.org/ref/epsg/#{@srid}/prettywkt/").read
+    end
+
+    # reproject via OGR, @see http://www.gdal.org/ogr2ogr.html
+    # prevent recoding using SHAPE_ENCODING environment variable
+    def ogr2ogr
+      cmd = "env SHAPE_ENCODING= ogr2ogr -q -t_srs '#{prettywkt}' '#{@output_filename}' '#{source_path}'"
+      self.class.execute(cmd)
+    end
+
+    # normalize prj file with the prettywkt
+    def normalize_prj_file
+      File.open(@output_filename.gsub(/\.shp$/, '.prj'), 'w') { |f| f.write(prettywkt) }
+    end
+  end
+end

lib/geo_concerns/processors/project_data.rb

@@ -0,0 +1,15 @@
+module GeoConcerns::Processors
+  class ProjectData < Hydra::Derivatives::Processors::Processor
+    def process
+      code = directives.fetch(:format)
+      case code
+      when 'ESRI Shapefile'
+        OgrProjection
+      when 'GTiff'
+        GdalProjection
+      else
+        fail NotImplementedError, "Unknown driver code: #{code}"
+      end.process(source_path, directives)
+    end
+  end
+end

lib/geo_concerns/runners/normalize_data_derivatives.rb

@@ -0,0 +1,7 @@
+module GeoConcerns::Runners
+  class NormalizeDataDerivatives < Hydra::Derivatives::Runner
+    def self.processor_class
+      GeoConcerns::Processors::NormalizeData
+    end
+  end
+end

spec/fixtures/McKay/S_566_1914_lines.prj

@@ -0,0 +1 @@
+PROJCS["NAD_1983_10TM_AEP_Forest",GEOGCS["GCS_North_American_1983",DATUM["D_North_American_1983",SPHEROID["GRS_1980",6378137.0,298.257222101]],PRIMEM["Greenwich",0.0],UNIT["Degree",0.0174532925199433]],PROJECTION["Transverse_Mercator"],PARAMETER["False_Easting",500000.0],PARAMETER["False_Northing",0.0],PARAMETER["Central_Meridian",-115.0],PARAMETER["Scale_Factor",0.9992],PARAMETER["Latitude_Of_Origin",0.0],UNIT["Meter",1.0]]

spec/processors/project_data_spec.rb

@@ -0,0 +1,39 @@
+require 'spec_helper'
+
+describe GeoConcerns::Processors::ProjectData do
+  let(:output_file) { double }
+  let(:file_name) { double }
+
+  subject { described_class.new(file_name, directives) }
+
+  before do
+    allow(subject).to receive(:output_file).with(file_name).and_return(output_file)
+  end
+
+  describe "for Shapefiles" do
+    let(:directives) { { format: 'ESRI Shapefile' } }
+    let(:file_name) { 'spec/fixtures/McKay/S_566_1914_lines.shp' }
+
+    it "will fire a Shapefile reprojection" do
+      expect { subject.process }.not_to raise_error
+    end
+  end
+
+  describe "for GeoTIFF" do
+    let(:directives) { { format: 'GTiff' } }
+    let(:file_name) { 'spec/fixtures/BackscatterA_8101_OffshoreBodegaHead.tif' }
+
+    it "will fire a GeoTIFF reprojection" do
+      expect { subject.process }.not_to raise_error
+    end
+  end
+
+  describe "for unknown formats" do
+    let(:directives) { { format: 'UNKNOWN' } }
+    let(:file_name) { 'noname' }
+
+    it "will throw an error for unknown formats" do
+      expect { subject.process }.to raise_error(NotImplementedError)
+    end
+  end
+end