rbbt 1.1.4 → 1.1.5

data/lib/rbbt/ner/regexpNER.rb

@@ -51,6 +51,7 @@ class RegExpNER
   end
 
   def match_hash(text)
+    return {} if text.nil? || text == ""
     matches = {}
     @index.each{|code, re|
       RegExpNER.match_re(text, re).each{|match|

data/lib/rbbt/sources/organism.rb

@@ -133,13 +133,13 @@ module Organism
       if i == 0
         i += 1
         next unless l=~/^\s*#/
-          formats  = l.chomp.sub(/^[\s#]+/,'').split(/\t/).collect{|n| n.strip}
+          formats  = Open.fields(l.sub(/^[\s#]+/,'')).collect{|n| n.strip}
         return formats unless examples
         next
       end
 
-      if l.chomp.split(/\t/).select{|name| name && name =~ /\w/}.length > examples.length
-        examples = l.chomp.split(/\t/).collect{|name| name.split(/\|/).first}
+      if Open.fields(l).select{|name| name && name =~ /\w/}.length > examples.length
+        examples = Open.fields(l).collect{|name| name.split(/\|/).first}
       end
       i += 1
     }
@@ -163,9 +163,10 @@ module Organism
     end
 
     lines.each{|l|
-      ids_per_type = l.split(/\t/)
+      ids_per_type = Open.fields(l)
       formats.zip(ids_per_type).each{|p|
         format = p[0]
+        p[1] ||= ""
         ids = p[1].split(/\|/)
         ids.each{|id|
           next if id.nil? || id == ""

data/lib/rbbt/sources/polysearch.rb

@@ -18,8 +18,18 @@ module Polysearch
 
   @@indexes = {}
   def self.type_index(type) #:nodoc:
-    @@indexes[type] ||= RegExpNER.new(File.join(Rbbt.datadir,'dbs','polysearch',type.to_s + '.txt'))
-    #@@indexes[type] ||= DictionaryNER.new(File.join(Rbbt.datadir,'dbs','polysearch',type + '.txt'))
+    if $stopwords
+      stopwords = $stopwords
+    else
+      stopwords = []
+    end
+
+    case type.to_sym
+    when :disease
+      stopwords << 'use'
+    end
+
+    @@indexes[type] ||= RegExpNER.new(File.join(Rbbt.datadir,'dbs','polysearch',type.to_s + '.txt'), :stopwords => stopwords)
   end
 
   # Find matches in a string of text, the types array specifies which thesauri

data/lib/rbbt/util/index.rb

@@ -36,6 +36,9 @@ module Index
       class << index; self; end.instance_eval{
         alias_method :old_get, :[]
         define_method(:[], proc{|key| old_get(key.to_s.downcase)})
+
+        alias_method :old_values_at, :values_at
+        define_method(:values_at, proc{|*keys| old_values_at(*keys.collect{|key| key.to_s.downcase }) })
       }
     end
 

data/lib/rbbt/util/misc.rb

@@ -4,7 +4,7 @@ require 'rbbt/util/open'
 class String
   CONSONANTS = []
   if File.exists? File.join(Rbbt.datadir, 'wordlists/consonants')
-    Open.read(File.join(Rbbt.datadir, 'wordlists/consonants')).each_line{|l| CONSONANTS << l.chomp}
+    Object::Open.read(File.join(Rbbt.datadir, 'wordlists/consonants')).each_line{|l| CONSONANTS << l.chomp}
   end
 
   # Uses heuristics to checks if a string seems like a special word, like a gene name.

data/lib/rbbt/util/open.rb

@@ -1,11 +1,19 @@
-require 'rbbt/util/tmpfile'
 require 'rbbt'
+require 'rbbt/util/tmpfile'
 
 
 # Provides with a few helper functions to read and write files, as well # as
 # for accessing remote files. It supports caching the files.
 module Open
 
+  def self.fields(line, sep = "\t")
+    chunks = line.chomp.split(/(#{sep})/).select{|c| c !~ /^#{sep}$/ }
+    if line =~ /#{sep}$/
+      chunks << ""
+    end
+    chunks
+  end
+
   class DirectoryNotFoundError < StandardError; end
 
   private 
@@ -165,19 +173,19 @@ module Open
       l = fix.call(l) if fix
       next if exclude and exclude.call(l)
 
-      parts = l.chomp.split(/#{sep}/)
-      id = parts[native]
+      row_fields = self.fields(l, sep)
+      id = row_fields[native]
       next if id.nil? || id == ""
 
       data[id] ||= []
       if extra
         fields = extra
       else
-        fields = (0..(parts.length - 1)).to_a - [native] 
+        fields = (0..(row_fields.length - 1)).to_a - [native] 
       end
       fields.each_with_index{|pos,i|
         data[id][i] ||= []
-        data[id][i] << parts[pos]
+        data[id][i] << row_fields[pos]
       }
     }
 

data/lib/rbbt/util/rake.rb

@@ -1,18 +1,67 @@
-def step(name, previous = nil, &block)
-  re = Regexp.new(/(?:^|\/)#{name}\/.*$/)
+require 'rake'
 
-  $step_descriptions ||= {}
-  if Rake.application.last_description 
-    $step_descriptions[re] = {:step => name, :message => Rake.application.last_description}
-    Rake.application.last_description = nil
+module Rake::Pipeline
+
+  def step_def(*args)
+    Rake::Pipeline::Step.step_def(*args)
+  end
+
+  def step_descriptions
+    Rake::Pipeline::Step.step_descriptions
+  end
+
+  def infile(t, &block)
+    File.open(t.prerequisites.first)do |f|
+      block.call(f)
+    end
   end
 
-  if previous.nil?
-    deps = []
-  else
-    deps = lambda{|filename| filename.sub(name.to_s, previous.to_s)}
+  def outfile(t, &block)
+    File.open(t.name, 'w')do |f|
+      block.call(f)
+    end
+  end
+
+
+end
+
+module Rake::Pipeline::Step
+
+  class << self
+    @@step_descriptions = {}
+    def step_descriptions
+      @@step_descriptions
+    end
+
+    def add_description(re, step, message)
+      @@step_descriptions[re] = "#{ step }: #{ message }"                       
+    end
+
+    def step_def(name, dependencies = nil)
+
+      re = Regexp.new(/(?:^|\/)#{name}\/.*$/)
+      re = Regexp.new(/#{name}\/.*/)
+
+        # Take the last_description and associate it with the name
+        if Rake.application.last_description
+          add_description(re, name, Rake.application.last_description)
+        end
+
+
+      # Generate the Hash definition
+      case 
+      when dependencies.nil?
+        re
+      when String === dependencies || Symbol === dependencies
+        {re => lambda{|filename| filename.sub(name.to_s,dependencies.to_s) }}
+      when Array === dependencies
+        {re => lambda{|filename| dependencies.collect{|dep| filename.sub(name.to_s, dep.to_s) } }}
+      when Proc === dependencies
+        {re => dependencies}
+      end
+    end
   end
 
-  Rake.application.create_rule(re => deps, &block) 
 end
 
+

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: rbbt
 version: !ruby/object:Gem::Version 
-  version: 1.1.4
+  version: 1.1.5
 platform: ruby
 authors: 
 - Miguel Vazquez
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-12-08 00:00:00 +01:00
+date: 2009-12-16 00:00:00 +01:00
 default_executable: rbbt_config
 dependencies: 
 - !ruby/object:Gem::Dependency