diff --git a/lib/jars/tabula-1.0.0-SNAPSHOT-jar-with-dependencies.jar b/lib/jars/tabula-1.0.0-SNAPSHOT-jar-with-dependencies.jar
index 8eabe30b..c5e67fbe 100644
Binary files a/lib/jars/tabula-1.0.0-SNAPSHOT-jar-with-dependencies.jar and b/lib/jars/tabula-1.0.0-SNAPSHOT-jar-with-dependencies.jar differ
diff --git a/lib/tabula_job_executor/jobs/detect_tables.rb b/lib/tabula_job_executor/jobs/detect_tables.rb
index 48ff3938..ff0fc06f 100644
--- a/lib/tabula_job_executor/jobs/detect_tables.rb
+++ b/lib/tabula_job_executor/jobs/detect_tables.rb
@@ -5,10 +5,8 @@
class DetectTablesJob < Tabula::Background::Job
include Observable
def perform
- filepath = options[:filepath]
+ filepath = options[:filepath]
output_dir = options[:output_dir]
-
-
page_areas_by_page = []
begin
@@ -39,6 +37,6 @@ def perform
end
at(100, 100, "complete")
- return nil
+ return nil
end
-end
+end
\ No newline at end of file
diff --git a/lib/tabula_job_executor/jobs/string_search.rb b/lib/tabula_job_executor/jobs/string_search.rb
new file mode 100644
index 00000000..163581fa
--- /dev/null
+++ b/lib/tabula_job_executor/jobs/string_search.rb
@@ -0,0 +1,35 @@
+require 'java'
+
+class StringSearchJob
+
+ def performString(output_dir, boundariesArray)
+ page_areas_by_page = []
+ begin
+ extractor = Tabula::Extraction::ObjectExtractor.new(File.join(output_dir, 'document.pdf'), :all)
+ page_count = extractor.page_count
+ rda = Java::TechnologyTabulaDetectors::StringSearch.new
+ extractor.extract.each do |page|
+ areas = rda.detect(page, boundariesArray)
+ page_areas_by_page << areas.map { |rect|
+ [ rect.getLeft,
+ rect.getTop,
+ rect.getWidth,
+ rect.getHeight ]
+ }
+ end
+
+ rescue Java::JavaLang::Exception => e
+ warn("String bounds detect failed. You may need to select tables manually.")
+ end
+
+ File.open(output_dir + "/string.json", 'w') do |f|
+ f.puts page_areas_by_page.to_json
+ end
+
+ File.open(output_dir + "/string_list.json", 'a') do |f|
+ f.puts boundariesArray[0] + "," + boundariesArray[1] + "," + boundariesArray[2] + "," + boundariesArray[3] + "\n"
+ end
+
+ return page_areas_by_page
+ end
+end
\ No newline at end of file
diff --git a/webapp/index.html b/webapp/index.html
index ca67c155..204f69bb 100644
--- a/webapp/index.html
+++ b/webapp/index.html
@@ -104,8 +104,45 @@
Copy to Clipboard
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+