biblicit 2.2.3 → 2.3.0

data/lib/biblicit/parscit.rb

@@ -17,14 +17,31 @@ module ParsCit
     attr_reader :result
 
     def initialize(in_txt, opts={})
-      mode = (opts.fetch :include_citations, false) ? 'extract_all' : 'extract_header'
-
+      parse_modes = if (opts.fetch :include_citations, false)
+                      { extract_all: [:citeseer, :parshed, :citations] }
+                    else
+                      {
+                        extract_header_svm_only: [:citeseer],
+                        extract_header_crf_only: [:parshed]
+                      }
+                    end
+
+      ENV['SVM_LIGHT_HOME'] ||= "#{File.dirname(`which svm_classify`)}"
       ENV['CRFPP_HOME'] ||= "#{File.dirname(`which crf_test`)}/../"
       ENV['PARSCIT_TMPDIR'] ||= "/tmp/"
 
-      output = `#{PERL_DIR}/bin/citeExtract.pl -q -m #{mode} #{in_txt.path}`
-
-      @result = parse(Nokogiri::XML output)
+      @result = {}
+
+      parse_modes.map do |mode, keys|
+        outf = Tempfile.new mode.to_s
+        pid = spawn("#{PERL_DIR}/bin/citeExtract.pl -q -m #{mode} #{in_txt.path}", out: outf.path)
+        [pid, outf, keys]
+      end.each do |pid, outf, keys|
+        Process.wait pid
+        output = File.read outf
+        outf.unlink
+        @result.merge! parse(Nokogiri::XML output).slice(*keys)
+      end
     end
 
   private

data/lib/biblicit/version.rb

@@ -2,6 +2,6 @@
 
 module Biblicit
 
-  VERSION = '2.2.3'
+  VERSION = '2.3.0'
 
 end

data/lib/biblicit.rb

@@ -2,5 +2,6 @@
 
 require 'active_support/core_ext/object'
 require 'active_support/core_ext/string'
+require 'active_support/core_ext/hash'
 
 require 'biblicit/extractor'

data/parscit/bin/citeExtract.pl

@@ -86,7 +86,7 @@ sub Help
 	print STDERR "\t-q\tQuiet Mode (don't echo license)\n";
 
 	# Thang v100401: add new mode (extract_section), and -i <inputType>
-	print STDERR "\t-m <mode>	   \tMode (extract_citations, extract_header, extract_section, extract_meta, extract_all, default: extract_citations)\n";
+	print STDERR "\t-m <mode>	   \tMode (extract_citations, extract_header, extract_header_svm_only, extract_header_crf_only, extract_section, extract_meta, extract_all, default: extract_citations)\n";
 	print STDERR "\t-i <inputType> \tType (raw, xml, default: raw)\n";
 	print STDERR "\t-e <exportType>\tExport citations into multiple types (ads|bib|end|isi|ris|wordbib). Multiple types could be specified by contatenating with \"-\" e.g., bib-end-ris. Output files will be named as outfile.exportFormat, with outfile being the input argument, and exportFormat being each individual format supplied by -e option.\n";
 	print STDERR "\t-t\tUse token level model instead\n";
@@ -281,6 +281,14 @@ sub ParseMode
 	{
 		return ($PARSHED | $SVM);
 	} 
+  elsif ($arg eq "extract_header_svm_only")
+  {
+    return $SVM;
+  }
+  elsif ($arg eq "extract_header_crf_only")
+  {
+    return $PARSHED;
+  }
 	elsif ($arg eq "extract_citations") 
 	{
 		return $PARSCIT;

metadata

@@ -2,25 +2,25 @@
 name: biblicit
 version: !ruby/object:Gem::Version
   prerelease: 
-  version: 2.2.3
+  version: 2.3.0
 platform: ruby
 authors:
 - David Judd
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-05-13 00:00:00.000000000 Z
+date: 2013-06-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
-  prerelease: false
-  name: activesupport
-  type: :runtime
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
       - !ruby/object:Gem::Version
         version: '0'
     none: false
+  name: activesupport
+  type: :runtime
+  prerelease: false
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
@@ -28,15 +28,15 @@ dependencies:
         version: '0'
     none: false
 - !ruby/object:Gem::Dependency
-  prerelease: false
-  name: nokogiri
-  type: :runtime
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
       - !ruby/object:Gem::Version
         version: '0'
     none: false
+  name: nokogiri
+  type: :runtime
+  prerelease: false
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
@@ -44,15 +44,15 @@ dependencies:
         version: '0'
     none: false
 - !ruby/object:Gem::Dependency
-  prerelease: false
-  name: rake
-  type: :development
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
       - !ruby/object:Gem::Version
         version: '0'
     none: false
+  name: rake
+  type: :development
+  prerelease: false
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
@@ -60,15 +60,15 @@ dependencies:
         version: '0'
     none: false
 - !ruby/object:Gem::Dependency
-  prerelease: false
-  name: rspec
-  type: :development
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
       - !ruby/object:Gem::Version
         version: '0'
     none: false
+  name: rspec
+  type: :development
+  prerelease: false
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
@@ -76,15 +76,15 @@ dependencies:
         version: '0'
     none: false
 - !ruby/object:Gem::Dependency
-  prerelease: false
-  name: pry
-  type: :development
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
       - !ruby/object:Gem::Version
         version: '0'
     none: false
+  name: pry
+  type: :development
+  prerelease: false
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - ! '>='
@@ -432,18 +432,12 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
   - - ! '>='
     - !ruby/object:Gem::Version
-      segments:
-      - 0
-      hash: 3256160240819012702
       version: '0'
   none: false
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ! '>='
     - !ruby/object:Gem::Version
-      segments:
-      - 0
-      hash: 3256160240819012702
       version: '0'
   none: false
 requirements:
@@ -469,3 +463,4 @@ test_files:
 - spec/fixtures/txt/sample1.txt
 - spec/fixtures/txt/sample2.txt
 - spec/spec_helper.rb
+has_rdoc: