Use orginfo json formatted output

app/derivative_services/geo_derivatives/processors/vector/info.rb

@@ -3,11 +3,21 @@ module GeoDerivatives
   module Processors
     module Vector
       class Info
-        attr_accessor :doc
+        attr_accessor :path
         attr_writer :name, :driver
 
         def initialize(path)
-          @doc = ogrinfo(path)
+          @path = path
+        end
+
+        # Parsed json out from ogrinfo
+        # @return [Hash]
+        def doc
+          @doc ||= begin
+                     JSON.parse(ogrinfo(path))
+                   rescue
+                     {}
+                   end
         end
 
         # Returns the vector dataset name
@@ -40,7 +50,7 @@ def bounds
           # @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.to_s)
+            stdout, stderr, status = Open3.capture3("ogrinfo", "-json", "-ro", "-so", "-al", path.to_s)
             raise(GeoDerivatives::OgrError, stderr) unless status.success?
             stdout
           end
@@ -49,27 +59,25 @@ def ogrinfo(path)
           # the vector dataset name.
           # @return [String] vector dataset name
           def vector_name
-            match = /(?<=Layer name:\s).*?(?=\n)/.match(doc)
-            match ? match[0] : ""
+            doc.dig("layers", 0, "name") || ""
           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] : ""
+            doc.fetch("driverShortName", "")
           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] : ""
+            geom = doc.dig("layers", 0, "geometryFields", 0, "type") || ""
+
             # Transform OGR-style 'Line String' into GeoJSON 'Line'
             # Transform OGR-style '3D Multi Polygon' into GeoJSON 'Polygon'
-            if geom == "Line String"
+            if geom == "LineString"
               geom = "Line"
             elsif geom == "3D Multi Polygon"
               geom = "Polygon"
@@ -82,12 +90,9 @@ def vector_geom
           # the vector bounding box.
           # @return [Hash] vector bounds
           def vector_bounds
-            match = /(?<=Extent:\s).*?(?=\n)/.match(doc)
-            extent = match ? match[0] : ""
+            extent = doc.dig("layers", 0, "geometryFields", 0, "extent").map { |c| c.truncate(6) }
 
-            # 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 }
+            { north: extent[3].to_f, east: extent[2].to_f, south: extent[1].to_f, west: extent[0].to_f }
           end
       end
     end

spec/characterization_services/gdal_characterization_service/vector_spec.rb

@@ -24,7 +24,7 @@
       file_set = valid_file_set
       new_file_set = described_class.new(file_set: file_set, persister: persister).characterize(save: false)
       expect(new_file_set.original_file.mime_type).to eq ["application/vnd.geo+json"]
-      expect(new_file_set.original_file.geometry).to eq ["Multi Polygon"]
+      expect(new_file_set.original_file.geometry).to eq ["MultiPolygon"]
     end
   end
 
@@ -36,7 +36,7 @@
       file_set = valid_file_set
       new_file_set = described_class.new(file_set: file_set, persister: persister).characterize(save: false)
       expect(new_file_set.original_file.mime_type).to eq ["application/vnd.geo+json"]
-      expect(new_file_set.original_file.geometry).to eq ["Multi Polygon"]
+      expect(new_file_set.original_file.geometry).to eq ["MultiPolygon"]
     end
   end
 

spec/derivative_services/geo_derivatives/processors/vector/info_spec.rb

@@ -4,9 +4,9 @@
 
 RSpec.describe GeoDerivatives::Processors::Vector::Info do
   let(:processor) { described_class.new(path) }
-  let(:path) { "test.tif" }
-  let(:polygon_info_doc) { file_fixture("files/gdal/ogrinfo_polygon.txt").read }
-  let(:line_info_doc) { file_fixture("files/gdal/ogrinfo_line.txt").read }
+  let(:path) { "test.geojson" }
+  let(:polygon_info_doc) { file_fixture("files/gdal/ogrinfo_polygon.json").read }
+  let(:line_info_doc) { file_fixture("files/gdal/ogrinfo_line.json").read }
   let(:status) { double(success?: true) }
 
   before do
@@ -15,14 +15,14 @@
 
   context "when initializing a new info class" do
     it "shells out to ogrinfo and sets the doc variable to the output string" do
-      expect(processor.doc).to eq(polygon_info_doc)
-      expect(Open3).to have_received(:capture3).with("ogrinfo", "-ro", "-so", "-al", path.to_s)
+      expect(processor.doc).to eq(JSON.parse(polygon_info_doc))
+      expect(Open3).to have_received(:capture3).with("ogrinfo", "-json", "-ro", "-so", "-al", path.to_s)
     end
   end
 
   context "with a polygon vector" do
     before do
-      allow(processor).to receive(:doc).and_return(polygon_info_doc)
+      allow(Open3).to receive(:capture3).and_return([polygon_info_doc, "", status])
     end
 
     describe "#name" do
@@ -46,7 +46,7 @@
     describe "#bounds" do
       it "returns bounds hash" do
         expect(processor.bounds).to eq(north: 42.408249,
-                                       east: -71.052853,
+                                       east: -71.052852,
                                        south: 42.347654,
                                        west: -71.163867)
       end
@@ -55,7 +55,7 @@
 
   context "with a line vector" do
     before do
-      allow(processor).to receive(:doc).and_return(line_info_doc)
+      allow(Open3).to receive(:capture3).and_return([line_info_doc, "", status])
     end
 
     describe "#geom" do