genevalidatorapp 1.4.13 → 1.5.0

data/lib/{GeneValidatorApp → genevalidatorapp}/config.rb

@@ -36,6 +36,7 @@ module GeneValidatorApp
     # Write config data to config file.
     def write_config_file
       return unless config_file
+
       File.open(config_file, 'w') do |f|
         f.puts(data.delete_if { |_, v| v.nil? }.to_yaml)
       end
@@ -71,10 +72,11 @@ module GeneValidatorApp
     # Default configuration data.
     def defaults
       {
-        :num_threads  => 1,
-        :port         => 4567,
-        :host         => '0.0.0.0',
-        :web_dir      => Dir.pwd
+        :num_threads    => 1,
+        :port           => 4567,
+        :host           => '0.0.0.0',
+        :gv_app_dir     => '~/.genevalidatorapp/',
+        :max_characters => 'undefined'
       }
     end
 
@@ -82,5 +84,4 @@ module GeneValidatorApp
       '~/.genevalidatorapp.conf'
     end
   end
-
-end
+end

data/lib/{GeneValidatorApp → genevalidatorapp}/database.rb

@@ -3,15 +3,7 @@ require 'digest/md5'
 require 'forwardable'
 
 module GeneValidatorApp
-  # Captures a directory containing FASTA files and BLAST databases.
-  #
-  # It is important that formatted BLAST database files have the same dirname and
-  # basename as the source FASTA for GeneValidatorApp to be able to tell formatted
-  # FASTA from unformatted. And that FASTA files be formatted with `parse_seqids`
-  # option of `makeblastdb` for sequence retrieval to work.
-  #
-  # GeneValidatorApp will always place BLAST database files alongside input FASTA,
-  # and use `parse_seqids` option of `makeblastdb` to format databases.
+  # class on the BLAST databases
   class Database < Struct.new(:name, :title, :type)
     class << self
       extend Forwardable
@@ -77,16 +69,19 @@ module GeneValidatorApp
       # Recurisvely scan `database_dir` for blast databases.
       def scan_databases_dir
         database_dir = config[:database_dir]
-        list = `blastdbcmd -recursive -list #{database_dir} -list_outfmt "%p	%f	%t" 2>&1`
+        cmd  = "blastdbcmd -recursive -list #{database_dir} -list_outfmt" \
+               ' "%p::%f::%t"'
+        list = `#{cmd} 2>&1`
         list.each_line do |line|
-          type, name, title =  line.split('	')
+          type, name, title =  line.split('::', 3)
           next if multipart_database_name?(name)
-          next unless type.downcase == 'protein' # to ensure we only have protein dbs
+          next unless type.downcase == 'protein'
           self << Database.new(name, title, type)
         end
       end
 
-      # Returns true if the database name appears to be a multi-part database name.
+      # Returns true if the database name appears to be a multi-part database
+      # name.
       #
       # e.g.
       # /home/ben/pd.ben/sequenceserver/db/nr.00 => yes

data/lib/genevalidatorapp/exceptions.rb

@@ -0,0 +1,162 @@
+# This file defines all possible exceptions that can be thrown by
+# GeneValidatorApp on startup.
+#
+# Exceptions only ever inform another entity (downstream code or users) of an
+# issue. Exceptions may or may not be recoverable.
+#
+# Error classes should be seen as: the error code (class name), human readable
+# message (to_s method), and necessary attributes to act on the error.
+#
+# We define as many error classes as needed to be precise about the issue, thus
+# making it easy for downstream code (bin/genevalidatorapp or config.ru) to act
+# on them.
+
+module GeneValidatorApp
+  # Error in config file.
+  class CONFIG_FILE_ERROR < StandardError
+    def initialize(ent, err)
+      @ent = ent
+      @err = err
+    end
+
+    attr_reader :ent, :err
+
+    def to_s
+      <<MSG
+Error reading config file: #{ent}.
+#{err}
+MSG
+    end
+  end
+
+  ## ENOENT ##
+
+  # Name borrowed from standard Errno::ENOENT, this class serves as a template
+  # for defining errors that mean "expected to find <entity> at <path>, but
+  # didn't".
+  #
+  # ENOENT is raised if and only if an entity was set, either using CLI or
+  # config file. For instance, it's compulsory to set database_dir. But ENOENT
+  # is not raised if database_dir is not set. ENOENT is raised if database_dir
+  # was set, but does not exist.
+  class ENOENT < StandardError
+    def initialize(des, ent)
+      @des = des
+      @ent = ent
+    end
+
+    attr_reader :des, :ent
+
+    def to_s
+      "Could not find #{des}: #{ent}"
+    end
+  end
+
+  # Raised if bin dir set, but does not exist.
+  class BIN_DIR_NOT_FOUND < ENOENT
+    def initialize(ent)
+      super 'bin dir', ent
+    end
+  end
+
+  # Raised if database dir set, but does not exist.
+  class DATABASE_DIR_NOT_FOUND < ENOENT
+    def initialize(ent)
+      super 'database dir', ent
+    end
+  end
+
+  # Raised if extension file set, but does not exist.
+  class EXTENSION_FILE_NOT_FOUND < ENOENT
+    def initialize(ent)
+      super 'extension file', ent
+    end
+  end
+
+  ## NUM THREADS ##
+
+  # Raised if num_threads set by the user is incorrect.
+  class NUM_THREADS_INCORRECT < StandardError
+    def to_s
+      'Number of threads should be a number greater than or equal to 1.'
+    end
+  end
+
+  ## BLAST NOT INSTALLED OR NOT COMPATIBLE ##
+
+  # Raised if SequenceServer could not locate NCBI BLAST+ installation on
+  # user's system.
+  class BLAST_NOT_INSTALLED < StandardError
+    def to_s
+      'Could not locate BLAST+ binaries.'
+    end
+  end
+
+  # Raised if SequenceServer could not successfully execute 'blastp -version'
+  # on user's system (see #141).
+  class BLAST_NOT_EXECUTABLE < StandardError
+    def to_s
+      'Error executing BLAST+ binaries.'
+    end
+  end
+
+  # Raised if SequenceServer determined NCBI BLAST+ present on the user's
+  # system but not meeting SequenceServer's minimum version requirement.
+  class BLAST_NOT_COMPATIBLE < StandardError
+    def initialize(version)
+      @version  = version
+    end
+
+    attr_reader :version
+
+    def to_s
+      <<MSG
+Your BLAST+ version #{version} is outdated.
+SequenceServer needs NCBI BLAST+ version #{MINIMUM_BLAST_VERSION} or higher.
+MSG
+    end
+  end
+
+  ## BLAST+ DATABASE RELATED ##
+
+  # Raised if 'database_dir' not set.
+  class DATABASE_DIR_NOT_SET < StandardError
+    def to_s
+      'Database dir not set.'
+    end
+  end
+
+  # Raised if not even one BLAST+ database was found in database_dir.
+  class NO_BLAST_DATABASE_FOUND < StandardError
+    def initialize(database_dir)
+      @database_dir = database_dir
+    end
+
+    attr_reader :database_dir
+
+    def to_s
+      "Could not find BLAST+ databases in: #{database_dir}."
+    end
+  end
+
+  # Raised if there was an error determining BLAST+ databases in database_dir.
+  class BLAST_DATABASE_ERROR < StandardError
+    def initialize(cmd, out)
+      @cmd = cmd
+      @out = out
+    end
+
+    attr_reader :cmd, :out
+
+    def to_s
+      <<MSG
+Error obtaining BLAST databases.
+Tried: #{cmd}
+Error:
+#{out.strip}
+
+Please could you report this to 'https://groups.google.com/forum/#!forum/sequenceserver'?
+MSG
+    end
+  end
+end

data/lib/{GeneValidatorApp → genevalidatorapp}/genevalidator.rb

@@ -2,6 +2,7 @@ require 'forwardable'
 require 'bio'
 require 'fileutils'
 require 'genevalidator'
+require 'json'
 
 module GeneValidatorApp
   # Module that runs GeneValidator
@@ -24,7 +25,7 @@ module GeneValidatorApp
     class << self
       extend Forwardable
 
-      def_delegators GeneValidatorApp, :config, :logger
+      def_delegators GeneValidatorApp, :config, :logger, :public_dir
 
       attr_reader :gv_dir, :tmp_gv_dir, :input_file, :xml_file, :raw_seq,
                   :unique_id, :params
@@ -33,7 +34,7 @@ module GeneValidatorApp
       def init(url, params)
         create_unique_id
         create_subdir_in_main_tmpdir
-        create_soft_link_from_tmpdir_to_GV_dir
+        create_soft_link_from_tmpdir_to_gv_dir
         @params = params
         validate_params
         obtain_db_path
@@ -45,7 +46,9 @@ module GeneValidatorApp
       def run
         write_seq_to_file
         run_genevalidator
-        (@params[:result_link]) ? @url : produce_table_html
+        copy_json_folder
+        (@params[:result_link]) ? @url : output_json_file_path
+        parse_output_json
       end
 
       private
@@ -53,7 +56,7 @@ module GeneValidatorApp
       # Creates a unique run ID (based on time),
       def create_unique_id
         @unique_id = Time.new.strftime('%Y-%m-%d_%H-%M-%S_%L-%N')
-        @gv_tmpdir = GeneValidatorApp.tempdir + unique_id
+        @gv_tmpdir = File.join(GeneValidatorApp.temp_dir, unique_id)
         ensure_unique_id
       end
 
@@ -62,7 +65,7 @@ module GeneValidatorApp
       def ensure_unique_id
         while File.exist?(@gv_tmpdir)
           @unique_id = create_unique_id
-          @gv_tmpdir = GeneValidatorApp.tempdir + @unique_id
+          @gv_tmpdir = File.join(GeneValidatorApp.temp_dir, unique_id)
         end
         logger.debug("Unique ID = #{@unique_id}")
       end
@@ -74,8 +77,9 @@ module GeneValidatorApp
       end
 
       # Create the Tmp Dir and the create a soft link to it.
-      def create_soft_link_from_tmpdir_to_GV_dir
-        @gv_dir = GeneValidatorApp.public_dir + 'GeneValidator' + @unique_id
+      def create_soft_link_from_tmpdir_to_gv_dir
+        @gv_dir = File.join(GeneValidatorApp.public_dir, 'GeneValidator',
+                            @unique_id)
         logger.debug("Local GV dir = #{@gv_dir}")
         FileUtils.ln_s "#{@gv_tmpdir}", "#{@gv_dir}"
       end
@@ -84,6 +88,7 @@ module GeneValidatorApp
       #  Only important if POST request is sent via API - Web APP also validates
       #  all params via Javascript.
       def validate_params
+        logger.debug("Input Paramaters: #{@params}")
         check_seq_param_present
         check_seq_length
         check_validations_param_present
@@ -92,23 +97,20 @@ module GeneValidatorApp
 
       # Simply asserts whether that the seq param is present
       def check_seq_param_present
-        unless @params[:seq]
-          fail ArgumentError, 'No input sequence provided.'
-        end
+        return if @params[:seq]
+        fail ArgumentError, 'No input sequence provided.'
       end
 
       def check_seq_length
-        return unless config[:max_characters]
-        unless @params[:seq].length < config[:max_characters]
-          fail ArgumentError, 'The input sequence is too long.'
-        end
+        return unless config[:max_characters] != 'undefined'
+        return if @params[:seq].length < config[:max_characters]
+        fail ArgumentError, 'The input sequence is too long.'
       end
 
       # Asserts whether the validations param are specified
       def check_validations_param_present
-        unless @params[:validations]
-          fail ArgumentError, 'No validations specified'
-        end
+        return if @params[:validations]
+        fail ArgumentError, 'No validations specified'
       end
 
       # Asserts whether the database parameter is present
@@ -124,7 +126,7 @@ module GeneValidatorApp
 
       # Writes the input sequences to a file with the sub_dir in the temp_dir
       def write_seq_to_file
-        @input_fasta_file = @gv_tmpdir + 'input_file.fa'
+        @input_fasta_file = File.join(@gv_tmpdir, 'input_file.fa')
         logger.debug("Writing input seqs to: '#{@input_fasta_file}'")
         ensure_unix_line_ending
         ensure_fasta_valid
@@ -145,7 +147,8 @@ module GeneValidatorApp
         unique_queries = {}
         sequence       = @params[:seq].lstrip
         if sequence[0] != '>'
-          sequence.insert(0, ">Submitted:#{Time.now.strftime('%H:%M-%B_%d_%Y')}\n")
+          sequence.insert(0, '>Submitted:'\
+                             "#{Time.now.strftime('%H:%M-%B_%d_%Y')}\n")
         end
         sequence.gsub!(/^\>(\S+)/) do |s|
           if unique_queries.key?(s)
@@ -163,8 +166,7 @@ module GeneValidatorApp
       #  empty
       def assert_input_file_present
         unless File.exist?(@input_fasta_file) || File.zero?(@input_fasta_file)
-          fail RuntimeError, 'GeneValidatorApp was unable to create the input' \
-                             ' file.'
+          fail 'GeneValidatorApp was unable to create the input file.'
         end
       end
 
@@ -180,62 +182,54 @@ module GeneValidatorApp
 
       # Runs GeneValidator
       def run_genevalidator
-        opts = set_up_gv_opts
-        logger.debug("Running GeneValidator with options: #{opts}")
         create_gv_log_file
         run_gv
-        assert_table_output_file_produced
+        assert_json_output_file_produced
       rescue SystemExit
-        raise RuntimeError, 'GeneValidator failed to run properly'
+        raise 'GeneValidator failed to run properly'
       end
 
       def run_gv
-        original_stdout = $stdout.clone unless logger.debug?
-        $stdout.reopen(@gv_log_file, 'w') unless logger.debug?
-        (GeneValidator::Validation.new(opts, 1, true, true)).run
-        $stdout = original_stdout unless logger.debug?
-      end
-
-      def set_up_gv_opts
-        {
-          validations: @params[:validations],
-          db: @db,
-          num_threads: config[:num_threads],
-          fast: true,
-          input_fasta_file: @input_fasta_file.to_s
-        }
+        cmd = "genevalidator -v '#{@params[:validations].join(', ')}'" \
+              " -d #{@db} -n #{config[:num_threads]} #{@input_fasta_file}"
+        logger.debug("GV command: $ #{cmd}")
+        log_file = (logger.debug?) ? '' : "> #{@gv_log_file} 2>&1"
+        `#{cmd} #{log_file}`
       end
 
       def create_gv_log_file
-        @gv_log_file = (@gv_tmpdir + 'log_file.txt').to_s
+        @gv_log_file = File.join(@gv_tmpdir, 'log_file.txt')
         logger.debug("Log file: #{@gv_log_file}")
       end
 
       # Assets whether the results file is produced by GeneValidator.
-      def assert_table_output_file_produced
-        @table_file = @gv_dir + 'input_file.fa.html/files/table.html'
-        unless File.exist?(@table_file)
-          fail RuntimeError, 'GeneValidator did not produce the required' \
-                             ' output file.'
-        end
-      end
-
-      # Reads the GV output table file.
-      # Updates links to the plots with relative links to plot jsons.
-      def produce_table_html
-        orig_plots_dir  = 'files/json/input_file.fa_'
-        local_plots_dir = Pathname.new('GeneValidator') + @unique_id +
-                          'input_file.fa.html/files/json/input_file.fa_'
-        full_html = IO.binread(@table_file)
-        full_html.gsub(/#{orig_plots_dir}/, local_plots_dir.to_s).gsub(
-                  '#Place_external_results_link_here', @url)
+      def assert_json_output_file_produced
+        @json_file = File.join(@gv_dir, 'input_file.fa.json')
+        return if File.exist?(@json_file)
+        fail 'GeneValidator did not produce the required output file.'
       end
 
       # Reuturns the URL of the results page.
       def produce_result_url_link(url)
-        url.gsub(/input/, '').gsub(/\/*$/, '') +
+        url.gsub(/input/, '').gsub(%r{/*$}, '') +
           "/GeneValidator/#{@unique_id}/input_file.fa.html/results.html"
       end
+
+      def parse_output_json
+        json_contents = File.read(output_json_file_path)
+        JSON.parse(json_contents)
+      end
+
+      def output_json_file_path
+        "#{@input_fasta_file}.json"
+      end
+
+      def copy_json_folder
+        json_dir = File.join("#{@input_fasta_file}.html", 'files/json')
+        web_dir_json = File.join(public_dir, 'web_files/json')
+        logger.debug("Moving JSON files from #{json_dir} to #{web_dir_json}")
+        FileUtils.cp_r(json_dir, web_dir_json)
+      end
     end
   end
 end