An Apache PDFBox extension for extracting symbol labels and bounding boxes from born-digital PDF files. Implemented in Java.
Copyright (c) 2018, 2019 Ritvik Joshi, Parag Mali, Puneeth Kukkadapu, Mahshad Mahdavi, and Richard Zanibbi
Developed at the Document and Pattern Recognition Laboratory
(Rochester Institute of Technology, USA)
Dependencies: Make sure that both Maven (3.6.0 for us - other versions may work fine) and Java 1.8 are installed on your system. Perl is also used to generate the bash scripts used to run the program more easily.
Operating System: While this code should work on Windows systems as well, we have only tested on Linux and MacOS X so far.
From the main directory, issue make. This will run the Maven build system, download dependencies, etc., compile source files and generate .jar files in ./target. Finally, a bash script bin/sscraper is generated, so that the program can be easily be used in different directories.
The pom.xml file can be modified to change the Maven build parameters.
The sscraper bash script for processing directories of PDF files can be invoked using:
sscraper [-b] inDir outDir
where inDir and outDir are directories containing PDFs to process and place XML output files respectively. Use the '-b' flag to also generate copies of each PDF with bounding boxes in the output directory.
To process a single file, in this first version you will need to use Java directly (see below).
Note: Add the bin/ directory to your execution path (the PATH variable in bash shell) in order to use the command anywhere on your system.
From Java, SymbolScraper is invoked using these parameters:
java -jar target/SymbolScraper-0.1.0-jar-with-dependencies.jar <mode> [op1] [op2] <input dir/file> [output dir]
Parameters:
<mode>: single file input (default)
-batch for input directory containing PDF files
*requires an output directory* ([output dir])
[op1] write modified pdf files
-b add bounding boxes in pdf file output
-f filter english words (from a fixed dictionary) in pdf output file
[op2] display symbol data in XML format
-d display on standard output
-p print to file
<input dir/file path> - input directory *or* PDF file path
<output dir> - output directory (required in batch mode)
An example of processing a single file, writing an XML file to inputDir/Output:
java -jar target/SymbolScraper-0.1.0-jar-with-dependencies.jar -p inputDir/file.pdf
Please direct questions or concerns to Richard Zanibbi rxzvcs@rit.edu, the director of the Document and Pattern Recognition Lab.