corenlp 0.0.3 → 0.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 072b1b153bb4591c16e6242713e9b4431ba003da
-  data.tar.gz: 0e71dd5289c128e0245f082ace874d29b51cd92f
+  metadata.gz: a88d19d7dc8eae9e7df59d4fe9b1e0c492aa194f
+  data.tar.gz: 5c6a32994f6720210b7a909c5b839503530793b3
 SHA512:
-  metadata.gz: 7969ddc18c42ca6c832c06bf677df56212f4ec54bc35bebbd4d9e4925425804015ef2e333cbc0463a88af76204ed46f814c355c10c703586761c9a7db501442d
-  data.tar.gz: 646eb3e03f42182e5a957fe6d52db3a7767cd52c9cf48bb4bd4bc36d0e07c03f1f4890c652addad58ccaf845f9a430c5d877143902b3824efc60c42e44757f31
+  metadata.gz: b5f185cda3feb604e97e5682440a01f763773e80631bf5a68aa19a1a35e2874999770dd8707af2d74cb9baf630491219fdba498284f37a784813510d48c6549f
+  data.tar.gz: 99cfc0054a47e92c517b6ba9025e063d6fb316a2788acf9255079ab828a7f046ff82057fee9b2b1a154127f7825ce6abcef0eae52a08e59175c0d94bcb58233b

data/lib/corenlp.rb

@@ -0,0 +1,111 @@
+require "nokogiri"
+require "bundler"
+Bundler.require
+
+module Corenlp
+  class Treebank
+    attr_accessor :raw_text, :filenames, :output_directory, :summary_file, :threads_to_use, :java_max_memory, :sentences
+
+    def initialize(attrs = {})
+      self.raw_text = attrs[:raw_text] || ""
+      self.filenames = []
+      self.output_directory = attrs[:output_directory] || "./tmp/language_processing"
+      self.summary_file = "#{output_directory}/summary_file_#{object_id}_#{Time.now.to_i}.txt"
+      self.filenames = []
+      self.threads_to_use = attrs[:threads_to_use] || 4
+      self.java_max_memory = attrs[:java_max_memory] || "-Xmx3g"
+      self.sentences = []
+    end
+
+    def write_output_file_and_summary_file
+      input_file = File.join(output_directory, "text_#{object_id}_#{Time.now.to_i}.txt")
+      filenames << input_file
+      File.open(input_file, "w"){|f| f.write(raw_text)}
+      File.open(summary_file, "w"){|f| f.write(filenames.join("\n"))}
+    end
+
+    def process_files_with_stanford_corenlp
+      deps = "./lib/ext" # dependencies directory: JARs, model files, taggers, etc.
+      classpath = "#{deps}/stanford-corenlp-3.4.jar:#{deps}/stanford-corenlp-3.4-models.jar:#{deps}/xom.jar:#{deps}/joda-time.jar:#{deps}/jollyday.jar:#{deps}/ejml-0.23.jar"
+      stanford_bin = "edu.stanford.nlp.pipeline.StanfordCoreNLP"
+      annotators = "tokenize,ssplit,pos,lemma,parse,ner"
+
+      options = []
+      options << ["-cp", classpath]
+      options << [java_max_memory, stanford_bin]
+      options << ["-annotators", annotators]
+      options << ["-ner.useSUTime", 0] # turn this off
+      #options << ["-sutime.binders", 0]
+      options << ["-outputDirectory", output_directory]
+      options << ["-nthreads", threads_to_use]
+      options << ["-filelist", summary_file] # a file with one zone file per line
+
+      command = "java #{options.map{|x| x.join(" ")}.join(" ")}"
+      puts "Running command: \n\n#{command}\n\n"
+      `#{command}`
+    end
+
+    def build_treebank
+      filenames.each do |filename|
+        xml_file = "#{filename}.xml"
+        doc = Nokogiri.XML(File.open(xml_file))
+        doc.xpath("//sentences/sentence").each_with_index do |sentence_node, idx|
+          sentence = Sentence.new(index: idx)
+          self.sentences << sentence
+          sentence_node.xpath(".//token").each_with_index do |token_node, index|
+            text = token_node.children.at('word').text
+            text = Token.clean_stanford_text(text)
+            cleaned_stanford_lemma = Token.clean_stanford_text(token_node.children.at('lemma').text)
+            token_attrs = {
+              index: index,
+              text: text,
+              penn_treebank_tag: token_node.children.at('POS').text,
+              stanford_lemma: cleaned_stanford_lemma,
+              type: Token.token_subclass_from_text(text),
+              ner: token_node.children.at('NER').text
+            }
+            token = Token.token_subclass_from_text(text).new(token_attrs)
+            sentence.tokens << token
+          end
+          sentence_node.xpath(".//dependencies[@type='collapsed-dependencies']/dep").each do |dep_node|
+            dependent_index = dep_node.children.at('dependent').attr('idx').to_i - 1
+            governor_index = dep_node.children.at('governor').attr('idx').to_i - 1
+            if dependent_index >= 0 && governor_index >= 0
+              dependent = sentence.get_dependency_token_by_index(dependent_index),
+              governor  = sentence.get_dependency_token_by_index(governor_index),
+              relation = dep_node.attr('type')
+              if dependent && governor && relation
+                token_dep = TokenDependency.new({
+                  dependent: sentence.get_dependency_token_by_index(dependent_index),
+                  governor: sentence.get_dependency_token_by_index(governor_index),
+                  relation: dep_node.attr('type')
+                })
+                sentence.token_dependencies << token_dep
+              end
+            end
+          end
+          sentence_node.xpath(".//parse").each do |parse_node|
+            sentence.parse_tree_raw = parse_node.text
+          end
+        end
+      end
+    end
+
+    def parse
+      write_output_file_and_summary_file
+      process_files_with_stanford_corenlp
+      build_treebank
+      self
+    end
+  end
+end
+
+require "corenlp/version"
+require "corenlp/sentence"
+require "corenlp/token"
+require "corenlp/token_dependency"
+require "corenlp/enclitic"
+require "corenlp/word"
+require "corenlp/punctuation"
+require "corenlp/number"
+require "corenlp/downloader"

data/lib/corenlp/downloader.rb

@@ -1,7 +1,9 @@
+require 'bundler'
+Bundler.require
 require 'net/http'
-require 'zip/zip'
 require 'fileutils'
 require 'uri'
+require 'zip'
 
 module Corenlp
   class Downloader
@@ -14,7 +16,7 @@ module Corenlp
 
     def extract
       puts "extracting file..."
-      Zip::ZipFile.open(local_file) do |zip_file|
+      Zip::File.open(local_file) do |zip_file|
         zip_file.each do |file|
           file_path = File.join(destination, file.name)
           zip_file.extract(file, file_path) unless File.exist?(file_path)
@@ -38,7 +40,7 @@ module Corenlp
 
     def download
       return unless url
-      puts "downloading zip file from url #{url} to #{destination}..."
+      puts "downloading zip file from url #{url}. Extracting files to #{destination}..."
       self.local_file = File.basename(url)
       uri = URI.parse(url)
       if local_file && uri

data/lib/corenlp/version.rb

@@ -1,3 +1,3 @@
 module Corenlp
-  VERSION = "0.0.3"
+  VERSION = "0.0.4"
 end

data/lib/tasks/downloader.rake

@@ -0,0 +1,10 @@
+require 'corenlp'
+
+desc "download Stanford CoreNLP dependencies files"
+namespace :corenlp do
+  task :download_deps do
+    zip_file_url = ENV['CORENLP_DOWNLOAD_URL'] || "http://nlp.stanford.edu/software/stanford-corenlp-full-2014-06-16.zip"
+    destination = File.join(ENV['CORENLP_DEPS_DIR'] || './lib/ext/')
+    Corenlp::Downloader.new(zip_file_url, destination).download
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: corenlp
 version: !ruby/object:Gem::Version
-  version: 0.0.3
+  version: 0.0.4
 platform: ruby
 authors:
 - Lengio Corporation
@@ -102,6 +102,7 @@ executables: []
 extensions: []
 extra_rdoc_files: []
 files:
+- lib/corenlp.rb
 - lib/corenlp/downloader.rb
 - lib/corenlp/enclitic.rb
 - lib/corenlp/number.rb
@@ -111,6 +112,7 @@ files:
 - lib/corenlp/token_dependency.rb
 - lib/corenlp/version.rb
 - lib/corenlp/word.rb
+- lib/tasks/downloader.rake
 - test/downloader_test.rb
 - test/enclitic_test.rb
 - test/number_test.rb