rbbt-text 1.3.11 → 1.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 587b7971fd0f8fec2602323ea480521445ca417c3a407d057cf95df1f3a36216
-  data.tar.gz: 4f66306f80e838a0a27299705d79e5856b38dd936005d2b18004539bbb192431
+  metadata.gz: d969cb752d1e7bc80458663d989ca4f58a2e134b7f708748dcf8383ca44f01d3
+  data.tar.gz: b0df4c7e9bb43f47b6031965b70a21ea9e3c7a12e012747d38776ad785f580e3
 SHA512:
-  metadata.gz: 4aa191aa4e5cb5e3f7d4a49b30beb1eb7259a34074a7521a10b9951cd1cc7a097a06ba6a97d9f4f4e100b2058de3f94f5199cb069a030f93a3f69bf1ecec09ff
-  data.tar.gz: f3d5eb11d12f8a9d951d1073abd7e6cb5ace99bd075e7dc897f0aa715ae1552271019b4eb6849a172529d830f30bbc09ce40dd351fd81f9c06f338b075523e36
+  metadata.gz: c7436bae2f407303bb81b812b586ec09bb97a70b2272f386f6bd574b257cde6a22789b362a22dd89546761c38147214423e2e53e58ff73ae6553630e34e2f6d2
+  data.tar.gz: f60cfc48e60112b2639d182eda684bed33f9249e1aeabdef220b77af05f4b77eb07f6f55d9e980c8372f2030e126847369c5a80dded711217b3fd78520a00db6

data/lib/rbbt/document/corpus/pubmed.rb

@@ -20,13 +20,25 @@ module Document::Corpus
                  elsif type.to_sym == :title
                    Document.setup(article.title || "", PUBMED_NAMESPACE, pmid, type.to_sym, self)
                  elsif type.to_sym == :title_and_abstract
-                   Document.setup((article.title || "") + "  " + (article.abstract || ""), PUBMED_NAMESPACE, pmid, type.to_sym, self)
+                   title = article.title
+                   abstract = article.abstract
+
+                   if title.nil? || title == ""
+                     text = article.abstract
+                     text = "" if text.nil?
+                   else
+                     title = title + "." unless title.end_with?(".")
+
+                     text = title + " " + abstract if abstract && ! abstract.empty?
+                   end
+
+                   Document.setup(text, PUBMED_NAMESPACE, pmid, type.to_sym, self)
                  else
                    raise "No FullText available for #{ pmid }" if article.full_text.nil?
                    Document.setup(article.full_text, PUBMED_NAMESPACE, pmid, :fulltext, self, :corpus => self)
                  end
       Log.debug "Loading pmid #{pmid}"
-      add_document(document)
+      add_document(document) if document
       document
     end
 

data/lib/rbbt/ner/g_norm_plus.rb

@@ -66,7 +66,7 @@ EOF
         end
 
         Open.write('config', CONFIG)
-        mem = Rbbt::Config.get(:java_mem, :GNormPlus, :g_norm_plus, :gnormplus, :gnp, :default => "2G")
+        mem = Rbbt::Config.get(:java_mem, :GNormPlus, :g_norm_plus, :gnormplus, :gnp, :default => "8G")
         CMD.cmd_log("java -Xmx#{mem} -Xms#{mem}  -jar '#{Rbbt.software.opt.GNormPlus.produce.find}/GNormPlus.jar' 'input' 'output' 'config'")
 
         if texts.respond_to? :key_field

data/lib/rbbt/segment.rb

@@ -172,10 +172,24 @@ module Segment
     end
   end
 
+  def self.relocate(segment, original, target, pad = 20)
+    if segment != target[segment.range]
+      start_pad = [pad, segment.offset].min
+      end_pad = [pad, original.length - segment.end].min
+      start = segment.offset - start_pad
+      eend = segment.end + end_pad
+
+      context = original[start..eend].gsub(/\s/,' ')
+      target = target.gsub(/\s/, ' ')
+      i = target.index context
+      raise "Context not found in original text" if i.nil?
+      segment.offset = i + start_pad
+    end
+  end
+
   def self.index(*args)
     Segment::RangeIndex.index(*args)
   end
-
 end
 
 require 'rbbt/segment/range_index'

data/share/install/software/OpenNLP

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 name='OpenNLP'
-url="http://apache.rediris.es/opennlp/opennlp-1.9.4/apache-opennlp-1.9.4-bin.tar.gz"
+url="http://apache.rediris.es/opennlp/opennlp-2.3.0/apache-opennlp-2.3.0-bin.tar.gz"
 
 install_src $name $url
 (cd $OPT_DIR/jars; ln -s $OPT_DIR/$name/lib/*.jar .)

data/test/rbbt/ner/test_g_norm_plus.rb

@@ -3,7 +3,7 @@ require 'rbbt/ner/g_norm_plus'
 
 Log.severity = 0
 class TestGNormPlus < Test::Unit::TestCase
-  def test_match
+  def _test_match
     text =<<-EOF
 
 Introduction
@@ -12,7 +12,7 @@ We found that TP53 is regulated by MDM2 in Homo
 sapiens
     EOF
 
-    Rbbt::Config.add_entry :java_mem, "2G", :gnp
+    Rbbt::Config.add_entry :java_mem, "16G", :gnp
     mentions = GNormPlus.process({:file => text})
 
     assert_equal 1, mentions.length
@@ -24,7 +24,7 @@ sapiens
 We found that TP53 is regulated by MDM2 in Homo sapiens
     EOF
 
-    Rbbt::Config.add_entry :java_mem, "2G", :gnp
+    Rbbt::Config.add_entry :java_mem, "16G", :gnp
     mentions = GNormPlus.entities({:file => text})
     assert mentions["file"].include?("TP53")
     mentions["file"].each do |mention|

data/test/rbbt/test_segment.rb

@@ -134,6 +134,25 @@ Atypical teratoid/rhabdoid tumors (AT/RTs) are highly aggressive brain tumors of
     assert_equal parts.first.docid, text.docid
   end
 
+  def test_relocate
+    original =<<-EOF
+This sentences contains
+a mention to gene TP53
+This is a followup sentence
+    EOF
+
+    target = <<-EOF
+    This sentence is added before
+This sentences contains a mention to gene TP53
+This is a followup sentence
+    EOF
+
+    segment = Segment.setup("TP53")
+    Segment.align(original, [segment])
+    Segment.relocate(segment, original, target)
+    assert_equal segment, target[segment.range]
+  end
+
   def test_segment_index
     text = "This sentence mentions the TP53 gene and the CDK5R1 protein"
     Document.setup(text, "TEST", "test_doc1", nil)

data/test/test_helper.rb

@@ -15,7 +15,11 @@ class Test::Unit::TestCase
 
   def setup
     FileUtils.mkdir_p Rbbt.tmp.test.persistence.find(:user)
-    Persist.cachedir = Rbbt.tmp.test.persistence.find :user
+    begin
+      Persist.cachedir = Rbbt.tmp.test.persistence.find :user
+    rescue
+      Persist.cache_dir = Rbbt.tmp.test.persistence.find :user
+    end
   end
 
   def teardown

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-text
 version: !ruby/object:Gem::Version
-  version: 1.3.11
+  version: 1.5.0
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-04-11 00:00:00.000000000 Z
+date: 2024-02-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util
@@ -175,45 +175,45 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.4.8
+rubygems_version: 3.5.0.dev
 signing_key: 
 specification_version: 4
 summary: Text mining tools for the Ruby Bioinformatics Toolkit (rbbt)
 test_files:
-- test/rbbt/nlp/test_nlp.rb
-- test/rbbt/nlp/open_nlp/test_sentence_splitter.rb
-- test/rbbt/nlp/genia/test_sentence_splitter.rb
 - test/rbbt/bow/test_bow.rb
-- test/rbbt/bow/test_misc.rb
 - test/rbbt/bow/test_dictionary.rb
-- test/rbbt/test_document.rb
-- test/rbbt/document/test_annotation.rb
+- test/rbbt/bow/test_misc.rb
 - test/rbbt/document/corpus/test_pubmed.rb
+- test/rbbt/document/test_annotation.rb
 - test/rbbt/document/test_corpus.rb
 - test/rbbt/entity/test_document.rb
-- test/rbbt/ner/test_patterns.rb
+- test/rbbt/ner/rnorm/test_tokens.rb
 - test/rbbt/ner/test_NER.rb
 - test/rbbt/ner/test_abner.rb
-- test/rbbt/ner/rnorm/test_tokens.rb
-- test/rbbt/ner/test_rnorm.rb
-- test/rbbt/ner/test_regexpNER.rb
-- test/rbbt/ner/test_ngram_prefix_dictionary.rb
+- test/rbbt/ner/test_banner.rb
 - test/rbbt/ner/test_brat.rb
-- test/rbbt/ner/test_g_norm_plus.rb
 - test/rbbt/ner/test_chemical_tagger.rb
-- test/rbbt/ner/test_banner.rb
-- test/rbbt/ner/test_token_trieNER.rb
 - test/rbbt/ner/test_finder.rb
-- test/rbbt/ner/test_rner.rb
+- test/rbbt/ner/test_g_norm_plus.rb
 - test/rbbt/ner/test_linnaeus.rb
+- test/rbbt/ner/test_ngram_prefix_dictionary.rb
 - test/rbbt/ner/test_oscar4.rb
-- test/rbbt/test_segment.rb
-- test/rbbt/segment/test_transformed.rb
-- test/rbbt/segment/test_overlaps.rb
+- test/rbbt/ner/test_patterns.rb
+- test/rbbt/ner/test_regexpNER.rb
+- test/rbbt/ner/test_rner.rb
+- test/rbbt/ner/test_rnorm.rb
+- test/rbbt/ner/test_token_trieNER.rb
+- test/rbbt/nlp/genia/test_sentence_splitter.rb
+- test/rbbt/nlp/open_nlp/test_sentence_splitter.rb
+- test/rbbt/nlp/test_nlp.rb
 - test/rbbt/segment/test_annotation.rb
-- test/rbbt/segment/test_named_entity.rb
+- test/rbbt/segment/test_corpus.rb
 - test/rbbt/segment/test_encoding.rb
+- test/rbbt/segment/test_named_entity.rb
+- test/rbbt/segment/test_overlaps.rb
 - test/rbbt/segment/test_range_index.rb
-- test/rbbt/segment/test_corpus.rb
-- test/test_spaCy.rb
+- test/rbbt/segment/test_transformed.rb
+- test/rbbt/test_document.rb
+- test/rbbt/test_segment.rb
 - test/test_helper.rb
+- test/test_spaCy.rb