tweaks to deal with changes in the pdfbox 2.0 version of tabula-java

tabulapdf · Apr 14, 2017 · b340017 · b340017
1 parent a356a6c
commit b340017
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 2 deletions.
diff --git a/...rs/tabula-0.9.2-jar-with-dependencies.jar → ...-1.0.0-SNAPSHOT-jar-with-dependencies.jar b/...rs/tabula-0.9.2-jar-with-dependencies.jar → ...-1.0.0-SNAPSHOT-jar-with-dependencies.jar
diff --git a/lib/tabula_java_wrapper.rb b/lib/tabula_java_wrapper.rb
@@ -67,12 +67,13 @@ module Extraction
 
     def Extraction.openPDF(pdf_filename, password='')
       raise Errno::ENOENT unless File.exists?(pdf_filename)
-      document = PDDocument.load(pdf_filename)
+      document = PDDocument.load(java.io.File.new(pdf_filename))
       #document = PDDocument.loadNonSeq(java.io.File.new(pdf_filename), nil, password)
       document
     end
 
     class ObjectExtractor < Java::TechnologyTabula.ObjectExtractor
+      field_accessor :pdfDocument
 
       alias_method :close!, :close
 
@@ -85,6 +86,10 @@ def initialize(pdf_filename, pages=[1], password='', options={})
 
         super(document)
       end
+
+      def page_count
+        self.pdfDocument.get_number_of_pages
+      end
     end
 
     class PagesInfoExtractor < ObjectExtractor

diff --git a/webapp/tabula_web.rb b/webapp/tabula_web.rb
@@ -9,7 +9,7 @@
 require 'fileutils'
 require 'securerandom'
 
-require_relative '../lib/jars/tabula-0.9.2-jar-with-dependencies.jar'
+require_relative '../lib/jars/tabula-1.0.0-SNAPSHOT-jar-with-dependencies.jar'
 
 require_relative '../lib/tabula_java_wrapper.rb'
 java_import 'java.io.ByteArrayOutputStream'