geo_works-derivatives 0.4.0

data/lib/geo_works/derivatives/processors/raster/base.rb

@@ -0,0 +1,44 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Raster
+        class Base < Hydra::Derivatives::Processors::Processor
+          include Hydra::Derivatives::Processors::ShellBasedProcessor
+          include GeoWorks::Derivatives::Processors::BaseGeoProcessor
+          include GeoWorks::Derivatives::Processors::Image
+          include GeoWorks::Derivatives::Processors::Gdal
+
+          def self.encode(path, options, output_file)
+            case options[:label]
+            when :thumbnail
+              encode_raster(path, output_file, options)
+            when :display_raster
+              reproject_raster(path, output_file, options)
+            end
+          end
+
+          # Set of commands to run to encode the raster thumbnail.
+          # @return [Array] set of command name symbols
+          def self.encode_queue
+            [:translate, :convert, :trim, :center]
+          end
+
+          # Set of commands to run to reproject the raster.
+          # @return [Array] set of command name symbols
+          def self.reproject_queue
+            [:warp, :cloud_optimized_geotiff]
+          end
+
+          def self.encode_raster(in_path, out_path, options)
+            run_commands(in_path, out_path, encode_queue, options)
+          end
+
+          def self.reproject_raster(in_path, out_path, options)
+            run_commands(in_path, out_path, reproject_queue, options)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/raster/dem.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Raster
+        class Dem < GeoWorks::Derivatives::Processors::Raster::Base
+          # Set of commands to run to encode the DEM thumbnail.
+          # @return [Array] set of command name symbols
+          def self.encode_queue
+            [:hillshade, :convert, :trim, :center]
+          end
+
+          # Set of commands to run to reproject the DEM.
+          # @return [Array] set of command name symbols
+          def self.reproject_queue
+            [:hillshade, :warp, :compress]
+          end
+
+          # Executes a gdal hillshade command. Calculates hillshade
+          # on a raster that contains elevation data.
+          # @param in_path [String] file input path
+          # @param out_path [String] processor output file path
+          # @param options [Hash] creation options
+          def self.hillshade(in_path, out_path, _options)
+            execute "gdaldem hillshade -q "\
+                      "-of GTiff \"#{in_path}\" #{out_path}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/raster/info.rb

@@ -0,0 +1,69 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Raster
+        class Info
+          attr_accessor :doc
+          attr_writer :min_max, :size
+
+          def initialize(path)
+            @doc = gdalinfo(path)
+          end
+
+          # Returns the gdal driver name
+          # @return [String] driver name
+          def driver
+            @driver = driver_name
+          end
+
+          # Returns the min and max values for a raster.
+          # @return [String] computed min and max values
+          def min_max
+            @min_max ||= raster_min_max
+          end
+
+          # Returns the raster size.
+          # @return [Array] raster size
+          def size
+            @size ||= raster_size
+          end
+
+          private
+
+            # Given an output string from the gdalinfo command, returns
+            # the gdal driver used to read dataset.
+            # @return [String] gdal driver name
+            def driver_name
+              match = /(?<=Driver:\s).*?(?=\s)/.match(doc)
+              match ? match[0] : ''
+            end
+
+            # Runs the gdalinfo command and returns the result as a string.
+            # @param path [String] path to raster file
+            # @return [String] output of gdalinfo
+            def gdalinfo(path)
+              stdout, _stderr, _status = Open3.capture3("gdalinfo -mm #{path}")
+              stdout
+            end
+
+            # Given an output string from the gdalinfo command, returns
+            # a formatted string for the computed min and max values.
+            # @return [String] computed min and max values
+            def raster_min_max
+              match = %r{(?<=Computed Min/Max=).*?(?=\s)}.match(doc)
+              match ? match[0].tr(',', ' ') : ''
+            end
+
+            # Given an output string from the gdalinfo command, returns
+            # an array containing the raster width and height as strings.
+            # @return [String] raster size
+            def raster_size
+              match = /(?<=Size is ).*/.match(doc)
+              match ? match[0].tr(',', '') : ''
+            end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/raster.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Raster
+        require 'geo_works/derivatives/processors/raster/base'
+        require 'geo_works/derivatives/processors/raster/aig'
+        require 'geo_works/derivatives/processors/raster/dem'
+        require 'geo_works/derivatives/processors/raster/info'
+
+        class Processor < Hydra::Derivatives::Processors::Processor
+          def process
+            raster_processor_class.new(source_path,
+                                       directives,
+                                       output_file_service: output_file_service).process
+          end
+
+          # Returns a raster processor class based on mime type passed in the directives object.
+          # @return raster processing class
+          def raster_processor_class
+            case directives.fetch(:input_format)
+            when 'text/plain; gdal-format=USGSDEM'
+              GeoWorks::Derivatives::Processors::Raster::Dem
+            when 'application/octet-stream; gdal-format=AIG'
+              GeoWorks::Derivatives::Processors::Raster::Aig
+            else
+              GeoWorks::Derivatives::Processors::Raster::Base
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/rendering.rb

@@ -0,0 +1,82 @@
+# frozen_string_literal: true
+require 'simpler_tiles'
+
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Rendering
+        extend ActiveSupport::Concern
+        included do
+          # Renders a thumbnail from a vector dataset.
+          # @param in_path [String] file input path
+          # @param out_path [String] processor output file path
+          # @param options [Hash] creation options
+          def self.vector_thumbnail(in_path, out_path, options)
+            map = GeoWorks::Derivatives::Processors::Rendering.simple_tiles_map(in_path, options)
+            File.open(out_path, 'wb') { |f| f.write map.to_png }
+          end
+        end
+
+        class << self
+          # Builds a simple tiles map from a shapefile.
+          # @param in_path [String] file input path
+          # @param options [Hash] creation
+          # @return [SimplerTiles::Map] simple tiles map
+          def simple_tiles_map(in_path, options)
+            assign_rendering_options(in_path, options)
+            size = rendering_size(options)
+            SimplerTiles::Map.new do |m|
+              m.srs     = 'EPSG:4326'
+              m.bgcolor = GeoWorks::Derivatives::Config.rendering_config.bg_color
+              m.width   = size[0]
+              m.height  = size[1]
+              m.set_bounds(*simple_tiles_bounds(options))
+              add_shapefile_layer(in_path, m)
+            end
+          end
+
+          private
+
+            # Adds a shapefile layer to a simple tiles map.
+            # @param in_path [String] file input path
+            # @param options [Hash] creation options
+            def add_shapefile_layer(in_path, map)
+              Dir.glob("#{in_path}/*.shp").each do |shp|
+                map.layer shp do |l|
+                  l.query %(select * from "#{File.basename shp, '.shp'}") do |q|
+                    q.styles GeoWorks::Derivatives::Config.rendering_config.to_h
+                  end
+                end
+              end
+            end
+
+            # Re-orders options bounds for use with a simple tiles map.
+            # @param options [Hash] creation options
+            # @return [Array] simple tiles map bounds
+            def simple_tiles_bounds(options)
+              [options[:bounds][:east],
+               options[:bounds][:north],
+               options[:bounds][:west],
+               options[:bounds][:south]]
+            end
+
+            # Transforms the size directive into an array.
+            # @param options [Hash] creation options
+            # @return [Array] derivative size
+            def rendering_size(options)
+              options[:output_size].split(' ').map(&:to_i)
+            end
+
+            # Assigns new values from the vector info command to the creation options hash.
+            # @param in_path [String] file input path
+            # @param options [Hash] creation options
+            def assign_rendering_options(in_path, options)
+              vector_info = GeoWorks::Derivatives::Processors::Vector::Info.new(in_path)
+              options[:name] = vector_info.name
+              options[:bounds] = vector_info.bounds
+            end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/vector/base.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Vector
+        class Base < Hydra::Derivatives::Processors::Processor
+          include Hydra::Derivatives::Processors::ShellBasedProcessor
+          include GeoWorks::Derivatives::Processors::BaseGeoProcessor
+          include GeoWorks::Derivatives::Processors::Image
+          include GeoWorks::Derivatives::Processors::Ogr
+          include GeoWorks::Derivatives::Processors::Gdal
+          include GeoWorks::Derivatives::Processors::Rendering
+          include GeoWorks::Derivatives::Processors::Zip
+
+          def self.encode(path, options, output_file)
+            case options[:label]
+            when :thumbnail
+              encode_vector(path, output_file, options)
+            when :display_vector
+              reproject_vector(path, output_file, options)
+            end
+          end
+
+          # Set of commands to run to encode the vector thumbnail.
+          # @return [Array] set of command name symbols
+          def self.encode_queue
+            [:reproject, :vector_thumbnail, :trim, :center]
+          end
+
+          # Set of commands to run to reproject the vector.
+          # @return [Array] set of command name symbols
+          def self.reproject_queue
+            [:reproject, :zip]
+          end
+
+          def self.encode_vector(in_path, out_path, options)
+            run_commands(in_path, out_path, encode_queue, options)
+          end
+
+          def self.reproject_vector(in_path, out_path, options)
+            run_commands(in_path, out_path, reproject_queue, options)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/vector/info.rb

@@ -0,0 +1,89 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Vector
+        class Info
+          attr_accessor :doc
+          attr_writer :name, :driver
+
+          def initialize(path)
+            @doc = ogrinfo(path)
+          end
+
+          # Returns the vector dataset name
+          # @return [String] dataset name
+          def name
+            @name = vector_name
+          end
+
+          # Returns the ogr driver name
+          # @return [String] driver name
+          def driver
+            @driver = driver_name
+          end
+
+          # Returns vector geometry type
+          # @return [String] geom
+          def geom
+            @geom = vector_geom
+          end
+
+          # Returns vector bounds
+          # @return [String] bounds
+          def bounds
+            @bounds = vector_bounds
+          end
+
+          private
+
+            # Runs the ogrinfo command and returns the result as a string.
+            # @param path [String] path to vector file or shapefile directory
+            # @return [String] output of ogrinfo
+            def ogrinfo(path)
+              stdout, _stderr, _status = Open3.capture3("ogrinfo -ro -so -al #{path}")
+              stdout
+            end
+
+            # Given an output string from the ogrinfo command, returns
+            # the vector dataset name.
+            # @return [String] vector dataset name
+            def vector_name
+              match = /(?<=Layer name:\s).*?(?=\n)/.match(doc)
+              match ? match[0] : ''
+            end
+
+            # Given an output string from the ogrinfo command, returns
+            # the ogr driver used to read dataset.
+            # @return [String] ogr driver name
+            def driver_name
+              match = /(?<=driver\s`).*?(?=')/.match(doc)
+              match ? match[0] : ''
+            end
+
+            # Given an output string from the ogrinfo command, returns
+            # the vector geometry type.
+            # @return [String] vector geom
+            def vector_geom
+              match = /(?<=Geometry:\s).*?(?=\n)/.match(doc)
+              geom = match ? match[0] : ''
+              # Transform OGR-style 'Line String' into GeoJSON 'Line'
+              geom == 'Line String' ? 'Line' : geom
+            end
+
+            # Given an output string from the ogrinfo command, returns
+            # the vector bounding box.
+            # @return [Hash] vector bounds
+            def vector_bounds
+              match = /(?<=Extent:\s).*?(?=\n)/.match(doc)
+              extent = match ? match[0] : ''
+
+              # remove parens and spaces, split into array, and assign elements to variables
+              w, s, e, n = extent.delete(' ').gsub(')-(', ',').delete('(').delete(')').split(',')
+              { north: n.to_f, east: e.to_f, south: s.to_f, west: w.to_f }
+            end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/vector/shapefile.rb

@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Vector
+        class Shapefile < GeoWorks::Derivatives::Processors::Vector::Base
+          include GeoWorks::Derivatives::Processors::Zip
+
+          def self.encode(path, options, output_file)
+            unzip(path, output_file) do |zip_path|
+              case options[:label]
+              when :thumbnail
+                encode_vector(zip_path, output_file, options)
+              when :display_vector
+                reproject_vector(zip_path, output_file, options)
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/vector.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Vector
+        require 'geo_works/derivatives/processors/vector/base'
+        require 'geo_works/derivatives/processors/vector/info'
+        require 'geo_works/derivatives/processors/vector/shapefile'
+
+        class Processor < Hydra::Derivatives::Processors::Processor
+          def process
+            vector_processor_class.new(source_path,
+                                       directives,
+                                       output_file_service: output_file_service).process
+          end
+
+          # Returns a vector processor class based on mime type passed in the directives object.
+          # @return vector processing class
+          def vector_processor_class
+            case directives.fetch(:input_format)
+            when 'application/zip; ogr-format="ESRI Shapefile"'
+              GeoWorks::Derivatives::Processors::Vector::Shapefile
+            else
+              GeoWorks::Derivatives::Processors::Vector::Base
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors/zip.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      module Zip
+        extend ActiveSupport::Concern
+
+        included do
+          # Unzips a file, invokes a block, and then deletes the unzipped file(s).
+          # Use to wrap processor methods for geo file formats that
+          # are zipped before uploading.
+          # @param in_path [String] file input path
+          # @param output_file [String] processor output file path
+          def self.unzip(in_path, output_file, _options = {})
+            basename = File.basename(output_file, File.extname(output_file))
+            zip_out_path = "#{File.dirname(output_file)}/#{basename}_out"
+            execute "unzip -qq -j -d \"#{zip_out_path}\" \"#{in_path}\""
+            yield zip_out_path
+            FileUtils.rm_rf(zip_out_path)
+          end
+
+          # Zips a file or directory.
+          # @param in_path [String] file input path
+          # @param output_file [String] output zip file
+          def self.zip(in_path, output_file, _options = {})
+            execute "zip -j -qq -r \"#{output_file}\" \"#{in_path}\""
+          end
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/processors.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Processors
+      require 'geo_works/derivatives/processors/base_geo_processor'
+      require 'geo_works/derivatives/processors/gdal'
+      require 'geo_works/derivatives/processors/image'
+      require 'geo_works/derivatives/processors/ogr'
+      require 'geo_works/derivatives/processors/rendering'
+      require 'geo_works/derivatives/processors/zip'
+      require 'geo_works/derivatives/processors/raster'
+      require 'geo_works/derivatives/processors/vector'
+    end
+  end
+end

data/lib/geo_works/derivatives/runners/raster_derivatives.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Runners
+      class RasterDerivatives < Hydra::Derivatives::Runner
+        def self.processor_class
+          GeoWorks::Derivatives::Processors::Raster::Processor
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/runners/vector_derivatives.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Runners
+      class VectorDerivatives < Hydra::Derivatives::Runner
+        def self.processor_class
+          GeoWorks::Derivatives::Processors::Vector::Processor
+        end
+      end
+    end
+  end
+end

data/lib/geo_works/derivatives/runners.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    module Runners
+      require 'geo_works/derivatives/runners/raster_derivatives'
+      require 'geo_works/derivatives/runners/vector_derivatives'
+    end
+  end
+end

data/lib/geo_works/derivatives/version.rb

@@ -0,0 +1,6 @@
+# frozen_string_literal: true
+module GeoWorks
+  module Derivatives
+    VERSION = "0.4.0"
+  end
+end

data/lib/geo_works/derivatives.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+require 'geo_works/derivatives/version'
+require 'hydra/derivatives'
+require 'mime/types'
+
+module GeoWorks
+  module Derivatives
+    require 'geo_works/derivatives/config'
+    require 'geo_works/derivatives/processors'
+    require 'geo_works/derivatives/runners'
+  end
+end