geohydra 0.3.1 → 0.3.3

data/bin/assemble_data.rb

@@ -3,64 +3,67 @@
 require File.expand_path(File.dirname(__FILE__) + '/../config/boot')
 require 'optparse'
 
-def assemble(path, flags)
-  ap({:path => path, :flags => flags}) if flags[:debug]
-  File.umask(002)
-  Dir.glob(File.join(path, '**', '*.shp')) do |shp|
-    raise ArgumentError, shp unless GeoHydra::Utils.shapefile?(shp)
+class AssembleData < GeoHydra::Process
+  def assemble(path, flags)
+    ap({:path => path, :flags => flags}) if flags[:debug]
+    File.umask(002)
+    Dir.glob(File.join(path, '**', '*.shp')) do |shp|
+      raise ArgumentError, "SyntaxError: Not a shapefile #{shp}" unless GeoHydra::Utils.shapefile?(shp)
 
-    ap({:shp => shp}) if flags[:debug]
-    geometry_type = GeoHydra::Transform.geometry_type(shp)
-    ap({:geometry_type => geometry_type}) if flags[:debug]
-    puts ['Scanned', File.basename(shp), geometry_type].join("\t") if flags[:verbose]
-    
-    basename = File.basename(shp, '.shp')
-    zipfn = File.join(path, 'content', 'data.zip')
-    puts "Compressing #{basename} into #{zipfn}" if flags[:verbose]
-    fns = Dir.glob(File.join(File.dirname(shp), "#{basename}.*")).select do |fn|
-      fn !~ /\.zip$/
-    end
-    metadata_fns = []
-    Dir.glob(File.join(File.dirname(shp), "#{basename}-iso19139*.xml")).each do |fn|
-      metadata_fns << fn
+      ap({:shp => shp}) if flags[:debug]
+      geometry_type = GeoHydra::Transform.geometry_type(shp)
+      ap({:geometry_type => geometry_type}) if flags[:debug]
+      puts ['Scanned', File.basename(shp), geometry_type].join("\t") if flags[:verbose]
+
+      basename = File.basename(shp, '.shp')
+      zipfn = File.join(path, 'content', 'data.zip')
+      puts "Compressing #{basename} into #{zipfn}" if flags[:verbose]
+      fns = Dir.glob(File.join(File.dirname(shp), "#{basename}.*")).select do |fn|
+        fn !~ /\.zip$/
+      end
+      metadata_fns = []
+      Dir.glob(File.join(File.dirname(shp), "#{basename}-iso19*.xml")).each do |fn|
+        metadata_fns << fn
+      end
+      system "mkdir -p #{File.dirname(zipfn)}" unless File.directory?(File.dirname(zipfn))
+      cmd =  "zip -vj '#{zipfn}' #{fns.join(' ')} #{metadata_fns.join(' ')}"
+      ap({:cmd => cmd, :fns => fns}) if flags[:debug]
+      system cmd
+      fns.each {|fn| FileUtils.rm(fn)} unless flags[:debug]
     end
-    cmd =  "zip -vj '#{zipfn}' #{fns.join(' ')} #{metadata_fns.join(' ')}"
-    ap({:cmd => cmd, :fns => fns}) if flags[:debug]
-    system cmd
-    fns.each {|fn| FileUtils.rm(fn)} unless flags[:debug]
   end
-end
 
-# __MAIN__
-begin
-  flags = {
-    :debug => false,
-    :verbose => false,
-    :srcdir => '/var/geomdtk/current/stage'
-  }
+  def run(args)
+    flags = {
+      :debug => false,
+      :verbose => false,
+      :srcdir => GeoHydra::Config.geohydra.stage || '/var/geomdtk/current/stage'
+    }
 
-  OptionParser.new do |opts|
-    opts.banner = <<EOM
-Usage: #{File.basename(__FILE__)} [options] [srcdir]
+    OptionParser.new do |opts|
+      opts.banner = <<EOM
+  Usage: #{File.basename(__FILE__)} [options] [srcdir]
 EOM
-    opts.on("-v", "--verbose", "Run verbosely") do |v|
-      flags[:debug] = true if flags[:verbose]
-      flags[:verbose] = true
-    end
-  end.parse!
+      opts.on("-v", "--verbose", "Run verbosely") do |v|
+        flags[:debug] = true if flags[:verbose]
+        flags[:verbose] = true
+      end
+    end.parse!(args)
 
-  
-  flags[:srcdir] = ARGV.pop unless File.directory?(flags[:srcdir])
-  raise ArgumentError, "Missing directory #{flags[:srcdir]}" unless flags[:srcdir] and File.directory?(flags[:srcdir])
 
-  puts "Searching for druid folders in #{flags[:srcdir]}..." if flags[:verbose]
-  n = 0
-  Dir.glob(File.join(flags[:srcdir], DruidTools::Druid.glob + '/')).sort.collect do |p|
-    assemble p, flags
-    n = n + 1
+    flags[:srcdir] = args.pop if args.size > 0
+    raise ArgumentError, "Missing directory #{flags[:srcdir]}" unless flags[:srcdir] and File.directory?(flags[:srcdir])
+
+    puts "Searching for druid folders in #{flags[:srcdir]}..." if flags[:verbose]
+    n = 0
+    Dir.glob(File.join(flags[:srcdir], DruidTools::Druid.glob + '/')).sort.collect do |p|
+      assemble p, flags
+      n = n + 1
+    end
+    puts "Processed #{n} folders."
   end
-  puts "Processed #{n} folders."
-rescue SystemCallError => e
-  $stderr.puts "ERROR: #{e.message}"
-  exit(-1)
 end
+
+
+# __MAIN__
+AssembleData.new.run(ARGV)

data/bin/assemble_placenames.rb

@@ -5,106 +5,106 @@ require 'druid-tools'
 require 'optparse'
 require 'json'
 
-@@g = GeoHydra::Gazetteer.new
-
-#
-# Resolves placenames using local gazetteer
-#
-#   * Changes subject/geographic with GeoNames as authority to have the correct valueURI
-#   * Adds correct rdf:resource to geo extension
-#   * Adds a LCSH or LCNAF keyword if needed
-#
-def resolve_placenames(modsFn, flags)
-  puts "Processing #{modsFn}" if flags[:verbose]
-  mods = Nokogiri::XML(File.open(modsFn, 'rb'))
-  r = mods.xpath('//mods:geographic', { 'mods' => 'http://www.loc.gov/mods/v3' })
-  r.each do |i|
-    ap({:i => i}) if flags[:debug]
-    k = i.content 
+class AssemblePlacenames < GeoHydra::Process
+  #
+  # Resolves placenames using local gazetteer
+  #
+  #   * Changes subject/geographic with GeoNames as authority to have the correct valueURI
+  #   * Adds correct rdf:resource to geo extension
+  #   * Adds a LCSH or LCNAF keyword if needed
+  #
+  def resolve_placenames(g, modsFn, flags)
+    puts "Processing #{modsFn}" if flags[:verbose]
+    mods = Nokogiri::XML(File.open(modsFn, 'rb'))
+    r = mods.xpath('//mods:geographic', { 'mods' => 'http://www.loc.gov/mods/v3' })
+    r.each do |i|
+      ap({:i => i}) if flags[:debug]
+      k = i.content 
     
-    # Verify Gazetteer keyword
-    uri = @@g.find_uri_by_keyword(k)
-    if uri.nil?
-      puts "WARNING: Missing gazetteer entry for '#{k}'" if flags[:verbose]
-      next
-    end
+      # Verify Gazetteer keyword
+      uri = g.find_placename_uri(k)
+      if uri.nil?
+        puts "WARNING: Missing gazetteer entry for '#{k}'" if flags[:verbose]
+        next
+      end
 
-    # Ensure correct valueURI for subject/geographic for GeoNames
-    i['valueURI'] = uri
-    i['authority'] = 'geonames'
-    i['authorityURI'] = 'http://www.geonames.org/ontology#'
+      # Ensure correct valueURI for subject/geographic for GeoNames
+      i['valueURI'] = uri
+      i['authority'] = 'geonames'
+      i['authorityURI'] = 'http://www.geonames.org/ontology#'
 
-    # Correct any linkages for placenames in the geo extension
-    coverages = mods.xpath('//mods:extension//dc:coverage', { 'mods' => 'http://www.loc.gov/mods/v3', 'dc' => 'http://purl.org/dc/elements/1.1/' })
-    coverages.each do |j|
-      if j['dc:title'] == k
-        puts "Correcting dc:coverage@rdf:resource for #{k}" if flags[:debug]
-        j['rdf:resource'] = uri
+      # Correct any linkages for placenames in the geo extension
+      coverages = mods.xpath('//mods:extension//dc:coverage', { 'mods' => 'http://www.loc.gov/mods/v3', 'dc' => 'http://purl.org/dc/elements/1.1/' })
+      coverages.each do |j|
+        if j['dc:title'] == k
+          puts "Correcting dc:coverage@rdf:resource for #{k}" if flags[:debug]
+          j['rdf:resource'] = uri + 'about.rdf'
+        end
       end
-    end
     
-    # Add a LC heading if needed
-    lc = @@g.find_lc_by_keyword(k)
-    ap({:lc => lc}) if flags[:debug]
-    unless lc.nil? or k == lc
-      puts "Adding Library of Congress entry to end of MODS record" if flags[:verbose]
-      lcauth = @@g.find_lcauth_by_keyword(k)
-      unless lcauth.nil?
-        lcuri = @@g.find_lcuri_by_keyword(k)
-        unless lcuri.nil?
-          lcuri = " valueURI='#{lcuri}'"
+      # Add a LC heading if needed
+      lc = g.find_loc_keyword(k)
+      ap({:lc => lc}) if flags[:debug]
+      unless lc.nil? or k == lc
+        puts "Adding Library of Congress entry to end of MODS record" if flags[:verbose]
+        lcauth = g.find_loc_authority(k)
+        unless lcauth.nil?
+          lcuri = g.find_loc_authority(k)
+          unless lcuri.nil?
+            lcuri = " valueURI='#{lcuri}'"
+          end
+          i.parent.parent << Nokogiri::XML("
+  <subject>
+    <geographic authority='#{lcauth}'#{lcuri}>#{lc}</geographic>
+  </subject>
+  ").root
         end
-        i.parent.parent << Nokogiri::XML("
-<subject>
-  <geographic authority='#{lcauth}'#{lcuri}>#{lc}</geographic>
-</subject>
-").root
       end
+      ap({:i => i}) if flags[:debug]
     end
-    ap({:i => i}) if flags[:debug]
-  end
   
-  # Save XML tree
-  mods.write_to(File.open(modsFn, 'wb'), :encoding => 'UTF-8', :indent => 2)
-end
+    # Save XML tree
+    mods.write_to(File.open(modsFn, 'wb'), :encoding => 'UTF-8', :indent => 2)
+  end
 
-def main(flags)
-  File.umask(002)
-  puts "Searching for MODS records..." if flags[:verbose]
-  Dir.glob(flags[:workspacedir] + '/**/' + DruidTools::Druid.glob + '/metadata/descMetadata.xml') do |modsFn|
-    resolve_placenames(modsFn, flags)
+  def main(flags)
+    g = GeoHydra::Gazetteer.new
+    File.umask(002)
+    puts "Searching for MODS records..." if flags[:verbose]
+    Dir.glob(flags[:workspacedir] + '/**/' + DruidTools::Druid.glob + '/metadata/descMetadata.xml') do |modsFn|
+      resolve_placenames(g, modsFn, flags)
+    end
   end
-end
 
-# __MAIN__
-begin
-  flags = {
-    :debug => false,
-    :verbose => false,
-    :workspacedir => GeoHydra::Config.geohydra.workspace || 'workspace'
-  }
+  def run(args)
+    flags = {
+      :debug => false,
+      :verbose => false,
+      :workspacedir => GeoHydra::Config.geohydra.workspace || 'workspace'
+    }
 
-  OptionParser.new do |opts|
-    opts.banner = <<EOM
-Usage: #{File.basename(__FILE__)} [options]
+    OptionParser.new do |opts|
+      opts.banner = <<EOM
+  Usage: #{File.basename(__FILE__)} [options]
 EOM
-    opts.on('-v', '--verbose', 'Run verbosely') do |v|
-      flags[:debug] = true if flags[:verbose]
-      flags[:verbose] = true
-    end
-    opts.on('--workspace DIR', "Workspace directory for assembly (default: #{flags[:workspacedir]})") do |v|
-      flags[:workspacedir] = v
+      opts.on('-v', '--verbose', 'Run verbosely') do |v|
+        flags[:debug] = true if flags[:verbose]
+        flags[:verbose] = true
+      end
+      opts.on('--workspace DIR', "Workspace directory for assembly (default: #{flags[:workspacedir]})") do |v|
+        flags[:workspacedir] = v
+      end
+    end.parse!(args)
+
+    %w{workspacedir}.each do |k|
+      d = flags[k.to_sym]
+      raise ArgumentError, "Missing directory #{d}" unless d.nil? or File.directory? d
     end
-  end.parse!
 
-  %w{workspacedir}.each do |k|
-    d = flags[k.to_sym]
-    raise ArgumentError, "Missing directory #{d}" unless d.nil? or File.directory? d
+    ap({:flags => flags}) if flags[:debug]
+    main flags
   end
-
-  ap({:flags => flags}) if flags[:debug]
-  main flags
-rescue SystemCallError => e
-  $stderr.puts "ERROR: #{e.message}"
-  exit(-1)
 end
+  
+# __MAIN__  
+AssemblePlacenames.new.run(ARGV)

data/bin/build_stage_options.rb

@@ -3,25 +3,31 @@
 require 'geohydra'
 require 'json'
 
-def doit(shp)
-  r = {}
-  r['druid'] = File.basename(File.dirname(File.dirname(shp)))
-  raise ArgumentError unless GeoHydra::Utils.shapefile?(shp)
-  r['geometryType'] = GeoHydra::Transform.geometry_type(shp)
-  r['filename'] = File.basename(shp)
-  File.open(File.join(File.dirname(shp), 'geoOptions.json'), 'w') do |f|
-    f.puts r.to_json.to_s
+class BuildStageOptions < GeoHydra::Process
+  # @param [String] shp assumes foo/bar/aa111bb1111/temp/shapefile.shp
+  def doit(shp)
+    r = {}
+    r['druid'] = File.basename(File.dirname(File.dirname(shp)))
+    raise ArgumentError, "SyntaxError: Not a shapefile: #{shp}" unless GeoHydra::Utils.shapefile?(shp)
+    r['geometryType'] = GeoHydra::Transform.geometry_type(shp)
+    r['filename'] = File.basename(shp)
+    File.open(File.join(File.dirname(shp), 'geoOptions.json'), 'w') do |f|
+      f.puts r.to_json.to_s
+    end
   end
-  
-end
 
-if ARGV.empty?
-  Dir.glob('/var/geomdtk/current/stage/**/temp/*.shp') do |shp|
-    doit(shp)
-  end
-else
-  ARGV.each do |shp|
-    doit(shp)
+  def run(args)
+    if args.empty?
+      Dir.glob('/var/geomdtk/current/stage/**/temp/*.shp') do |shp|
+        doit(shp)
+      end
+    else
+      args.each do |shp|
+        doit(shp)
+      end
+    end
   end
-  
 end
+
+# __MAIN__
+BuildStageOptions.new.run(ARGV)

data/bin/derive_wgs84.rb

@@ -4,82 +4,81 @@ require 'druid-tools'
 require 'fileutils'
 require 'optparse'
 
+class DeriveWGS84 < GeoHydra::Process
+  def run(args)
+    File.umask(002)
+    flags = {
+      :overwrite_prj => true,
+      :verbose => false,
+      :workspacedir => GeoHydra::Config.geohydra.workspace || 'workspace',
+      :tmpdir => GeoHydra::Config.geohydra.tmpdir || 'tmp',
+      # ogr2ogr is using a different WKT than GeoServer -- this one is from GeoServer 2.3.1.
+      # As implemented by EPSG database on HSQL:
+      #  http://docs.geotools.org/latest/userguide/library/referencing/hsql.html
+      # Also see:
+      #  http://spatialreference.org/ref/epsg/4326/prettywkt/
+      :wkt => {
+        '4326' => %Q{
+        GEOGCS["WGS 84",
+            DATUM["WGS_1984",
+                SPHEROID["WGS 84",6378137,298.257223563,
+                    AUTHORITY["EPSG","7030"]],
+                AUTHORITY["EPSG","6326"]],
+            PRIMEM["Greenwich",0,
+                AUTHORITY["EPSG","8901"]],
+            UNIT["degree",0.01745329251994328,
+                AUTHORITY["EPSG","9122"]],
+            AUTHORITY["EPSG","4326"]]
+        }.split.join.freeze
+      }
 
+    }  
 
-# __MAIN__
-begin
-  File.umask(002)
-  flags = {
-    :overwrite_prj => true,
-    :verbose => false,
-    :workspacedir => GeoHydra::Config.geohydra.workspace || 'workspace',
-    :tmpdir => GeoHydra::Config.geohydra.tmpdir || 'tmp',
-    # ogr2ogr is using a different WKT than GeoServer -- this one is from GeoServer 2.3.1.
-    # As implemented by EPSG database on HSQL:
-    #  http://docs.geotools.org/latest/userguide/library/referencing/hsql.html
-    # Also see:
-    #  http://spatialreference.org/ref/epsg/4326/prettywkt/
-    :wkt => {
-      '4326' => %Q{
-      GEOGCS["WGS 84",
-          DATUM["WGS_1984",
-              SPHEROID["WGS 84",6378137,298.257223563,
-                  AUTHORITY["EPSG","7030"]],
-              AUTHORITY["EPSG","6326"]],
-          PRIMEM["Greenwich",0,
-              AUTHORITY["EPSG","8901"]],
-          UNIT["degree",0.01745329251994328,
-              AUTHORITY["EPSG","9122"]],
-          AUTHORITY["EPSG","4326"]]
-      }.split.join.freeze
-    }
-    
-  }  
-  
-  OptionParser.new do |opts|
-    opts.banner = <<EOM
-Usage: #{File.basename(__FILE__)} [options] [druid...]
+    OptionParser.new do |opts|
+      opts.banner = <<EOM
+  Usage: #{File.basename(__FILE__)} [options] [druid...]
 EOM
-    opts.on("-v", "--verbose", "Run verbosely") do |v|
-      flags[:verbose] = true
-    end
-    opts.on("--workspace DIR", "Workspace directory for assembly (default: #{flags[:workspacedir]})") do |v|
-      flags[:workspacedir] = v
-    end
-    opts.on("--tmpdir DIR", "Temporary directory for assembly (default: #{flags[:tmpdir]})") do |v|
-      flags[:tmpdir] = v
-    end
-    opts.on("--wkt SRID FILE", "Read WKT for SRID from FILE") do |srid, f|
-      flags[:wkt][srid.to_s] = File.read(f).split.join.freeze
-    end
-  end.parse!
-  
-  [flags[:tmpdir], flags[:workspacedir]].each do |d|
-    raise ArgumentError, "Missing directory #{d}" unless File.directory? d
-  end
-
-  if ARGV.empty?
-    # matches druid workspace structure
-    Dir.glob(flags[:workspacedir] + '/??/???/??/????/???????????/content/*.zip').each do |fn| 
-      id = File.basename(File.dirname(File.dirname(fn)))
-      druid = DruidTools::Druid.new(id, flags[:workspacedir])
-      unless fn =~ %r{_EPSG_}i
-        puts "Processing #{druid.id} #{fn}"
-        GeoHydra::Transform.reproject druid, fn, flags 
+      opts.on("-v", "--verbose", "Run verbosely") do |v|
+        flags[:verbose] = true
+      end
+      opts.on("--workspace DIR", "Workspace directory for assembly (default: #{flags[:workspacedir]})") do |v|
+        flags[:workspacedir] = v
       end
+      opts.on("--tmpdir DIR", "Temporary directory for assembly (default: #{flags[:tmpdir]})") do |v|
+        flags[:tmpdir] = v
+      end
+      opts.on("--wkt SRID FILE", "Read WKT for SRID from FILE") do |srid, f|
+        flags[:wkt][srid.to_s] = File.read(f).split.join.freeze
+      end
+    end.parse!(args)
+
+    [flags[:tmpdir], flags[:workspacedir]].each do |d|
+      raise ArgumentError, "Missing directory #{d}" unless File.directory? d
     end
-  else
-    ARGV.each do |id|
-      druid = DruidTools::Druid.new(id, flags[:workspacedir])
-      Dir.glob(druid.content_dir + '/*.zip').each do |fn|
+
+    if args.empty?
+      # matches druid workspace structure
+      Dir.glob(flags[:workspacedir] + '/??/???/??/????/???????????/content/data.zip').each do |fn| 
+        id = File.basename(File.dirname(File.dirname(fn)))
+        druid = DruidTools::Druid.new(id, flags[:workspacedir])
         unless fn =~ %r{_EPSG_}i
           puts "Processing #{druid.id} #{fn}"
           GeoHydra::Transform.reproject druid, fn, flags 
         end
       end
+    else
+      args.each do |id|
+        druid = DruidTools::Druid.new(id, flags[:workspacedir])
+        Dir.glob(druid.content_dir + '/data.zip').each do |fn|
+          unless fn =~ %r{_EPSG_}i
+            puts "Processing #{druid.id} #{fn}"
+            GeoHydra::Transform.reproject druid, fn, flags 
+          end
+        end
+      end
     end
   end
-rescue SystemCallError => e
-  $stderr.puts "ERROR: #{e.message}"
-  exit(-1)
 end
+
+# __MAIN__
+DeriveWGS84.new.run(ARGV)