mspire 0.7.18 → 0.8.0

data/lib/mspire/mzml/run.rb

@@ -1,4 +1,7 @@
 require 'mspire/cv/paramable'
+require 'mspire/mzml/io_index'
+require 'mspire/mzml/spectrum_list'
+require 'mspire/mzml/chromatogram_list'
 
 module Mspire
   class Mzml
@@ -27,11 +30,13 @@ module Mspire
       attr_accessor :chromatogram_list
 
       # yields self if given a block
-      def initialize(id, default_instrument_configuration, opts={params: []}, &block)
+      def initialize(id, default_instrument_configuration)
         @id = id
         @default_instrument_configuration = default_instrument_configuration
-        super(opts)
-        block.call(self) if block
+        params_init
+        if block_given?
+          yield self
+        end
       end
 
       def to_xml(builder)
@@ -41,14 +46,66 @@ module Mspire
         atts[:defaultSourceFileRef] = @default_source_file.id if @default_source_file
         atts[:sampleRef] = @sample.id if @sample
         atts[:startTimeStamp] = @start_time_stamp if @start_time_stamp
+
+        default_ids = { instrument_configuration: @default_instrument_configuration.id }
         
         builder.run(atts) do |run_n|
           super(run_n)
-          spectrum_list.to_xml(run_n) if spectrum_list
-          chromatogram_list.to_xml(run_n) if chromatogram_list
+          spectrum_list.to_xml(run_n, default_ids) if spectrum_list
+          chromatogram_list.to_xml(run_n, default_ids) if chromatogram_list
         end
         builder
       end
+
+      # expects link to have the following keys:
+      #
+      #     :ref_hash
+      #     :instrument_config_hash
+      #     :source_file_hash
+      #     :sample_hash
+      #     :data_processing_hash
+      #     :spectrum_default_data_processing
+      #     :chromatogram_default_data_processing
+      #     :index_list
+      def self.from_xml(io, xml, link)
+
+        # expects that the DataProcessing objects to link to have *already* been
+        # parsed (parse the defaultDataProcessingRef's after grabbing the
+        # index, then grab the DataProcessing object associated with that id).
+        
+        obj = self.new(xml[:id], 
+          link[:instrument_configuration_hash][xml[:defaultInstrumentConfigurationRef]]
+        )
+        obj.start_time_stamp = xml[:startTimeStamp]
+
+        link[:default_instrument_configuration] = obj.default_instrument_configuration
+
+        # two optional object refs
+        if def_source_ref=xml[:defaultSourceFileRef]
+          obj.default_source_file = link[:source_file_hash][def_source_ref]
+        end
+        if sample_ref=xml[:sampleRef]
+          obj.sample = link[:sample_hash][sample_ref]
+        end
+
+        obj.describe_from_xml!(xml, link[:ref_hash])
+
+        index_list = link[:index_list]
+        [:spectrum, :chromatogram].each do |list_type|
+          next unless (byte_index = index_list[list_type])
+
+          io_index = IOIndex.new(io, byte_index, link)
+
+          list_obj = Mspire::Mzml.const_get(list_type.to_s.capitalize + "List")
+            .new(link["#{list_type}_default_data_processing".to_sym], 
+                 io_index, 
+                 Hash[byte_index.ids.each_with_index.map.to_a]
+                )
+
+          obj.send(list_type.to_s + "_list=", list_obj)
+        end
+        obj
+      end
     end
   end
 end

data/lib/mspire/mzml/sample.rb

@@ -5,13 +5,18 @@ module Mspire
   class Mzml
     class Sample
       include Mspire::CV::Paramable
+      extend Mspire::Mzml::List
 
-      attr_accessor :id, :name
+      # A unique identifier across the samples with which to reference this sample description.
+      attr_accessor :id
 
-      def initialize(id, name, opts={params: []}, &block)
-        @id, @name = id, name
-        super(opts)
-        block.call(self) if block
+      # An optional name for the sample description, mostly intended as a quick mnemonic.
+      attr_accessor :name
+
+      def initialize(id)
+        @id = id
+        params_init
+        yield(self) if block_given?
       end
 
       def to_xml(builder)
@@ -21,7 +26,12 @@ module Mspire
         builder
       end
 
-      extend(Mspire::Mzml::List)
+      def self.from_xml(xml, link)
+        obj = self.new(xml[:id])
+        obj.name = xml[:name]
+        obj.describe_self_from_xml!(xml, link[:ref_hash])
+      end
+
     end
   end
 end

data/lib/mspire/mzml/scan.rb

@@ -1,4 +1,5 @@
 require 'mspire/cv/paramable'
+require 'mspire/cv/paramable'
 require 'mspire/mzml/scan_window'
 
 module Mspire
@@ -14,31 +15,41 @@ module Mspire
       # a boolean indicating the spectrum is from an external source file
       attr_accessor :from_external_source_file
 
-      # an InstrumentConfiguration object
+      # an InstrumentConfiguration object (optional).
       attr_accessor :instrument_configuration
 
       # ScanWindow objects
       attr_accessor :scan_windows
 
-      def initialize(opts={params: []}, &block)
-        super(opts)
-        block.call(self) if block
+      def initialize
+        params_init
+        yield(self) if block_given?
       end
 
-      # takes a nokogiri node
-      #def self.from_xml(xml)
-      #end
-
-      def self.from_xml(xml)
-        scan = self.new
-        [:cvParam, :userParam].each {|v| scan.describe! xml.xpath("./#{v}") }
-        scan.scan_windows = xml.xpath('./scanWindowList/scanWindow').map do |scan_window_n|
-          Mspire::Mzml::ScanWindow.from_xml(scan_window_n)
+      # link should have:
+      #
+      #     :ref_hash
+      #     :default_instrument_configuration
+      #     :instrument_configuration_hash
+      def self.from_xml(xml, link)
+        ref_hash = link[:ref_hash]
+        obj = self.new
+        obj.instrument_configuration =
+          if icf = xml[:instrumentConfigurationRef]
+            link[:instrument_configuration_hash][icf]
+          else
+            link[:default_instrument_configuration]
+          end
+        scan_window_list_n = obj.describe_from_xml!(xml, ref_hash)
+        if scan_window_list_n
+          obj.scan_windows = scan_window_list_n.children.map do |scan_window_n|
+            Mspire::Mzml::ScanWindow.new.describe_self_from_xml!(scan_window_n, ref_hash)
+          end
         end
-        scan
+        obj
       end
 
-      def to_xml(builder)
+      def to_xml(builder, default_ids)
         atts = {}
         if @from_external_source_file
           atts[:sourceFileRef] = @spectrum.source_file.id
@@ -46,7 +57,11 @@ module Mspire
         else
           atts[:spectrumRef] = @spectrum.id if @spectrum
         end
-        atts[:instrumentConfigurationRef] = @instrument_configuration.id if @instrument_configuration
+        if @instrument_configuration
+          unless @instrument_configuration.id == default_ids[:instrument_configuration]
+            atts[:instrumentConfigurationRef] = @instrument_configuration.id 
+          end
+        end
         builder.scan(atts) do |prec_n|
           super(prec_n) # description
           ScanWindow.list_xml(@scan_windows, prec_n) if @scan_windows

data/lib/mspire/mzml/scan_list.rb

@@ -1,4 +1,5 @@
 require 'mspire/cv/paramable'
+require 'mspire/mzml/scan'
 
 module Mspire
   class Mzml
@@ -12,21 +13,33 @@ module Mspire
     class ScanList < Array
       include Mspire::CV::Paramable
 
-      def initialize(opts={params: []}, &block)
-        super(opts)
-        block.call(self) if block
+      def initialize
+        params_init
+        yield(self) if block_given?
       end
 
-      def to_xml(builder)
+      def to_xml(builder, default_ids)
         builder.scanList(count: self.size) do |sl_n|
           super(sl_n)
           self.each do |scan|
-            scan.to_xml(sl_n)
+            scan.to_xml(sl_n, default_ids)
           end
         end
         builder
       end
 
+      def self.from_xml(xml, link)
+        scan_list = self.new
+        scan_n = scan_list.describe_from_xml!(xml, link[:ref_hash])
+        if scan_n
+          loop do
+            scan_list << Mspire::Mzml::Scan.from_xml(scan_n, link)
+            break unless scan_n = scan_n.next
+          end
+        end
+        scan_list
+      end
+
       alias_method :list_xml, :to_xml
     end
   end

data/lib/mspire/mzml/scan_settings.rb

@@ -5,13 +5,14 @@ module Mspire
   class Mzml
     class ScanSettings
       include Mspire::CV::Paramable
+      extend Mspire::Mzml::List
 
       attr_accessor :id
 
-      def initialize(id, opts={params: []}, &block)
+      def initialize(id)
         @id = id
-        super(opts)
-        block.call(self) if block
+        params_init
+        yield(self) if block_given?
       end
 
       def to_xml(builder)
@@ -21,8 +22,6 @@ module Mspire
         builder
       end
 
-      extend(Mspire::Mzml::List)
-
     end
   end
 end

data/lib/mspire/mzml/scan_window.rb

@@ -8,12 +8,6 @@ module Mspire
     #     accession="MS:1000500" name="scan window upper limit" value="1800"
     class ScanWindow
       include Mspire::CV::Paramable
-
-      def self.from_xml(xml)
-        obj = self.new
-        [:cvParam, :userParam].each {|v| obj.describe! xml.xpath("./#{v}") }
-        obj
-      end
     end
   end
 end

data/lib/mspire/mzml/selected_ion.rb

@@ -1,7 +1,6 @@
 require 'mspire/mzml/list'
 require 'mspire/cv/paramable'
 
-
 module Mspire
   class Mzml
     # MUST supply a *child* term of MS:1000455 (ion selection attribute) one or more times
@@ -12,13 +11,7 @@ module Mspire
     #     e.g.: MS:1000744 (selected ion m/z)
     class SelectedIon
       include Mspire::CV::Paramable
-      extend(Mspire::Mzml::List)
-      
-      def self.from_xml(xml)
-        obj = self.new
-        [:cvParam, :userParam].each {|v| obj.describe! xml.xpath("./#{v}") }
-        obj
-      end
+      extend Mspire::Mzml::List
     end
   end
 end

data/lib/mspire/mzml/software.rb

@@ -6,13 +6,14 @@ module Mspire
   class Mzml
     class Software
       include Mspire::CV::Paramable
+      extend Mspire::Mzml::List
 
       attr_accessor :id, :version
 
-      def initialize(id='mspire', version=Mspire::VERSION, opts={params: []}, &block)
+      def initialize(id='mspire', version=Mspire::VERSION)
         @id, @version = id, version
-        super(opts)
-        block.call(self) if block
+        params_init
+        yield(self) if block_given?
       end
 
       def to_xml(builder)
@@ -22,7 +23,11 @@ module Mspire
         builder
       end
 
-      extend(Mspire::Mzml::List)
+      def self.from_xml(xml, link)
+        obj = self.new(xml[:id], xml[:version])
+        obj.describe_self_from_xml!(xml, link[:ref_hash])
+      end
+
     end
   end
 end

data/lib/mspire/mzml/source_file.rb

@@ -6,6 +6,7 @@ module Mspire
   class Mzml
     class SourceFile
       include Mspire::CV::Paramable
+      extend Mspire::Mzml::List
 
       DEFAULT_SOURCEFILE_ID = 'sourcefile1'
 
@@ -31,10 +32,10 @@ module Mspire
         [pathname.basename, 'file://'+ dir]
       end
 
-      def initialize(id="sourcefile1", name="mspire-simulated", location='file://', opts={params: []}, &block)
+      def initialize(id="sourcefile1", name="mspire-simulated", location='file://')
         @id, @name, @location = id, name, location
-        super(opts)
-        block.call(self) if block
+        params_init
+        yield(self) if block_given?
       end
 
       def to_xml(builder)
@@ -44,7 +45,10 @@ module Mspire
         builder
       end
 
-      extend(Mspire::Mzml::List)
+      def self.from_xml(xml, ref_hash)
+        self.new(xml[:id], xml[:name], xml[:location]).describe_self_from_xml!(xml, ref_hash)
+      end
+
     end
   end
 end

data/lib/mspire/mzml/spectrum.rb

@@ -1,6 +1,7 @@
-require 'mspire/mzml/data_array_container_like'
+
 require 'mspire/spectrum_like'
 require 'mspire/mzml/data_array'
+require 'mspire/mzml/data_array_container_like'
 require 'mspire/mzml/scan_list'
 require 'mspire/mzml/precursor'
 require 'mspire/mzml/product'
@@ -48,6 +49,8 @@ module Mspire
       # (optional) an Mspire::Mzml::SourceFile object
       attr_accessor :source_file
 
+      # data_processing is included with DataArrayContainerLike
+
       # (optional) The identifier for the spot from which this spectrum was derived, if a
       # MALDI or similar run.
       attr_accessor :spot_id
@@ -107,58 +110,80 @@ module Mspire
       end
 
       # takes a Nokogiri node and sets relevant properties
-      def self.from_xml(xml)
-        spec = Mspire::Mzml::Spectrum.new(xml[:id])
-
-        spec.spot_id = xml[:spotID]
+      # 
+      # link is a hash that should contain the following keys and associated
+      # objects (some are required by downstream objects, like Scan):
+      #
+      #     :ref_hash                         id -> ReferenceableParamGroup
+      #     :spectrum_list                    SpectrumList object
+      #     :data_processing_hash             id -> DataProcessing object
+      #     :default_data_processing          DataProcessing object
+      #     :instrument_configuration_hash    id -> InstrumentConfiguration object
+      #     :default_instrument_configuration InstrumentConfiguration object
+      #     :source_file_hash                 id -> SourceFile object
+      def self.from_xml(xml, link)
+        obj = self.new(xml[:id])
+        obj.spot_id = xml[:spotID]
+
+        obj.data_processing = link[:data_processing_hash][xml[:dataProcessingRef]] || link[:spectrum_default_data_processing]
+
+        if source_file_ref=xml[:sourceFileRef]
+          obj.source_file = link[:source_file_hash][source_file_ref]
+        end
 
-        [:cvParam, :userParam].each {|v| spec.describe! xml.xpath("./#{v}") }
+        xml_n = obj.describe_from_xml!(xml, link[:ref_hash])
+        return obj unless xml_n
 
-        scan_list = Mspire::Mzml::ScanList.new
-        xml.xpath('./scanList/scan').each do |scan_n|
-          scan_list << Mspire::Mzml::Scan.from_xml(scan_n)
+        loop do
+          case xml_n.name
+          when 'scanList'
+            obj.scan_list = Mspire::Mzml::ScanList.from_xml(xml_n, link)
+          when 'precursorList'
+            obj.precursors = xml_n.children.map do |prec_n|
+              Mspire::Mzml::Precursor.from_xml(prec_n, link)
+            end
+          when 'productList'
+            obj.products = xml_n.children.map do |product_n|
+              Mspire::Mzml::Product.from_xml(product_n, link)
+            end
+          when 'binaryDataArrayList'
+            obj.data_arrays = Mspire::Mzml::DataArray.data_arrays_from_xml(xml_n, link)
+          end
+          break unless xml_n = xml_n.next
         end
-        spec.scan_list = scan_list
-
-        spec.precursors = xml.xpath('./precursorList/precursor').map do |prec_n|
-          Mspire::Mzml::Precursor.from_xml(prec_n)
+        unless obj.data_arrays
+          obj.data_arrays = Mspire::Mzml::DataArray.empty_data_arrays
         end
 
-        data_arrays = xml.xpath('./binaryDataArrayList/binaryDataArray').map do |binary_data_array_n|
-          accessions = binary_data_array_n.xpath('./cvParam').map {|node| node['accession'] }
-          base64 = binary_data_array_n.xpath('./binary').text
-          Mspire::Mzml::DataArray.from_binary(base64, accessions)
-        end
 
-        # if there is no spectrum, we will still return a spectrum object, it
-        # just has no mzs or intensities
-        data_arrays = [Mspire::Mzml::DataArray.new, Mspire::Mzml::DataArray.new] if data_arrays.size == 0
-        spec.data_arrays = data_arrays
-        spec
+
+        obj
       end
 
       # the most common param to pass in would be ms level: 'MS:1000511'
       #
-      # This would generate a spectrum of ms_level=2 :
+      # This would generate a spectrum of ms_level 2 :
       #
-      #     Mspire::Mzml::Spectrum.new(0, "scan=1", 'MS:1000511')
+      #     spec = Mspire::Mzml::Spectrum.new("scan=1").describe!('MS:1000511')
       #
-      def initialize(id, opts={params: []}, &block)
+      def initialize(id)
         @id = id
-        params_initialize(opts)
-        block.call(self) if block
+        params_initialize
+        yield(self) if block_given?
       end
 
-
       # see SpectrumList for generating the entire list
-      def to_xml(builder)
-        atts = {}
+      def to_xml(builder, default_ids)
+        atts = data_array_xml_atts(default_ids)
         atts[:sourceFileRef] = @source_file.id if @source_file
         atts[:spotID] = @spot_id if @spot_id
-        super(builder, atts) do |node|
-          @scan_list.list_xml( node ) if @scan_list
-          Mspire::Mzml::Precursor.list_xml(@precursors, node) if @precursors
-          Mspire::Mzml::Product.list_xml(@products, node) if @products
+
+        builder.spectrum(atts) do |sp_n|
+          super(sp_n)
+          @scan_list.list_xml( sp_n, default_ids ) if @scan_list
+          Mspire::Mzml::Precursor.list_xml(@precursors, sp_n) if @precursors
+          Mspire::Mzml::Product.list_xml(@products, sp_n) if @products
+          Mspire::Mzml::DataArray.list_xml(@data_arrays, sp_n) if @data_arrays
         end
       end
 

data/lib/mspire/mzml/spectrum_list.rb

@@ -1,43 +1,14 @@
-require 'mspire/mzml/spectrum'
+require 'mspire/mzml/io_indexable_list'
 
 module Mspire
   class Mzml
-    class SpectrumList < Array
+    class SpectrumList < IOIndexableList
 
-      # a DataProcessing object
-      attr_reader :default_data_processing
+      attr_accessor :scan_to_index
 
-      def initialize(default_data_processing, spectra=[])
-        @default_data_processing = default_data_processing
-        super(spectra)
+      def fetch_by_scan_num(scan_num)
+        __getobj__[@scan_to_index[scan_num]]
       end
-
-      # This method takes an object responding to :data, creates a new
-      # Mspire::Mzml::Spectrum object with that data and puts it in the internal
-      # list
-      def add_ms_spectrum!(spectrum, id)
-        mzml_spec = Mspire::Mzml::Spectrum.new(id)
-        mzml_spec.data = spectrum.data
-        self << mzml_spec
-      end
-
-      # takes an array of spectra and performs add_spectrum on each
-      # returns self
-      def add_ms_spectra!(spectra, ids=[])
-        spectra.zip(ids).each_with_index {|(spec,id),i| add_spectrum(spec, "spectrum=#{i+1}") }
-        self
-      end
-
-      def to_xml(builder)
-        builder.spectrumList(count: self.size, defaultDataProcessingRef: @default_data_processing.id) do |spl_n|
-          self.each_with_index do |spectrum,i|
-            spectrum.index = i unless spectrum.index
-            spectrum.to_xml(spl_n)
-          end
-        end
-        builder
-      end
-
     end
   end
 end