Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

TrapRange: a Method to Extract Table Content in PDF Files

DZone's Guide to

TrapRange: a Method to Extract Table Content in PDF Files

· Java Zone
Free Resource

Microservices! They are everywhere, or at least, the term is. When should you use a microservice architecture? What factors should be considered when making that decision? Do the benefits outweigh the costs? Why is everyone so excited about them, anyway?  Brought to you in partnership with IBM.

Introduction

Table data structure is one of the most important data structures in document, especially when exporting data from systems, data is usually in table format. There are several data file formats are often used to store data including csv, text, and pdf. For the first two formats, it is definitely easy to extract data content by opening files, loop through lines and split cells depending on its cell separator, and of course we already had many libraries that help us to do that work automatically. It's normal work. On the other hand, regarding pdf file, it's not a familiar format to read and process directly from InputStream because it is a complicating file format that can contain not only text data, font, content style, but also image, audio and video [1]. In this post, we are going to describe our solution used to extract table data in pdf files. Our solution was implemented, experimented and adapted with pdf files having high density of table content. The advantages and disadvantages of our solution are also discussed.

How to recognize a table

After some investigation, I realized that:

  • Column: text content in cells of the same column lies on a rectangular space that does not overlap with other rectangular spaces of another column. For example, in the following image, red rectangle and blue rectangle are separated spaces
  • Row: words in same horizontal alignment are in the same row. But this is just sufficient condition because a cell in a row may be a multi-line cell. For example, the fourth cell in the yellow rectangle has two lines, phrases “FK to this customer’s record in” and "Ledgers table" are not int same horizontal alignment but they are still considered in the same row. In our solution, we simply assume that content in a cell only is single-line content. Different lines in a cell are considered to belong to different rows. Therefore the content in the yellow rectangle contains two rows: 1. {"Ledger_ID", "|" , "Sales Ledger Account" , "FK to this customer's record to"} 2. {NULL , NULL , NULL , "Ledgers table"}

recognize a table

PDFBox API

My work is based on the data returned by PDFBox API, an open source project. To extract text from a pdf file, PDFBox API provides 4 classes:

  • PDDocument: contains information of entire pdf file. In order to load a pdf file, we use method PDDocument.load(stream: InputStream)
  • PDPage: represents for each page in pdf document. We possibly archive a specific page content by passing the index of the page with this method: document.getDocumentCatalog().getAllPages().get(pageIdx: int)
  • TextPosition: represents an individual word or character in the document. We can fetch all TextPosition objects of a PDPage by overriding method processTextPosition(text: TextPosition) in class PDTextStripper. A TextPosition object has methods getX(), getY(), getWidth(), getHeight() that returns its bound in page and method getCharacter() to get its content.

In my work, I process text chunks directly by using TextPosition objects. For each text chunk in PDF file, it returns a text element with following attributes:

  • x: horizontal distance from the left of the page
  • y: vertical distance from the top border of the page
  • maxX: equals x + width of the text chunk
  • maxY: equals y+ height of the text chunk

textposition rectangle

Trap ranges

The most important point of my method is identifying the bound of each row and column in the table, because if we know the bound of a row or a column, we can retrieve all texts in this row or column so that we can extract all content of the table. We name these bounds are trap-ranges. TrapRange has two attributes:

  • lowerBound: contains the lower endpoint of this range
  • upperBound: contains the upper endpoint of this range To calculate values of trap-ranges, we loop through all texts of the page and project range of each text onto horizontal and vertical axis, get the result and join them together. After looping through all texts of page, we will calculate trap-ranges and use them to identify cell data of the table.

join sample

Algorithm 1: calculating trap-ranges for each pdf page:

columnTrapRanges <-- []
rowTrapRanges <-- []
for each text in page
begin
     columnTrapRanges <-- join(columnTrapRanges, {text.x, text.x + text.width} )
     rowTrapRanges <-- join(rowTrapRanges, {text.y, text.y + text.height} )
end


After calculating trap-ranges for the table, we loop through all texts again and classify them into correct cells of the table.

Algorithm 2: classifying text chunks into correct cells:

table <-- new Table()
for each text in page
begin
     rowIdx <-- in rowTrapRanges, get index of the range that containts this text
     columnIdx <-- in columnTrapRanges, get index of the range that contains this text
     table.addText(text, rowIdx, columnIdx)
end

Design and implement

traprange class diagram

The above is class diagram that describes main classes in our projects:

  • TrapRangeBuilder: method build() in this class will calculate and return all trap-ranges corresponding to added ranges
  • Table, TableRow and TableCell: contain data of extracted table
  • PDFTableExtractor is the most important class. It contains methods to initialize and extract table data from an input pdf file. Builder pattern was applied for this interface. Following is some important methods in this class:
    • setSource: set source of the pdf file. It has three versions of this method, including setSource(InputStream), setSource(File) and setSource(String)
    • addPage: to determine which pages will be extracted table content. If not set, all pages will be extracted
    • exceptPage: skip a page
    • exceptLine: this method is used to tell the algorithm that which page contains noisy data instead of table content. All texts in these lines will be avoided.
    • extract: do extraction and return result

Example

PDFTableExtractor extractor = new PDFTableExtractor();
List<Table> tables = extractor.setSource(“table.pdf”)
    .addPage(0)
    .addPage(1)
    .exceptLine(0) //the first line in each page
    .exceptLine(1) //the second line in each page
    .exceptLine(-1)//the last line in each page
    .extract();
String html = tables.get(0).toHtml();//table in html format
String csv = tables.get(0).toString();//table in csv format using semicolon as a delimiter 

Following are some sample results (check out and run the test file TestExtractor.java):

Evaluation

In experimentation, we used pdf files having high density of table data. The results show that our implementation realizes table data better than other open source tools: pdftotext, pdftohtml, pdf2table. With documents having multi tables or too much noisy data, our method does not work or works incorrectly. When a table row has a cell overlaps the space of beside columns, this row will be discarded.

Conclusion

TrapRange method works best with pdf files having high density of table data. With documents have multi-tables or too much noisy data, TrapRange is not a good choice. Our method also can be implemented in other programming language by replacing PDFBox by a corresponding pdf library or using command-line tool pdftohtml to extract text chunks and using these data as input data for algorithm 1, 2.

Visit and fork my project at: https://github.com/thoqbk/traprange

References

  1. http://en.wikipedia.org/wiki/Portable_Document_Format
  2. http://pdfbox.apache.org
  3. http://ieg.ifs.tuwien.ac.at/pub/yildiz_iicai_2005.pdf
  4. http://www.foolabs.com/xpdf/
  5. http://ieg.ifs.tuwien.ac.at/projects/pdf2table/

Discover how the Watson team is further developing SDKs in Java, Node.js, Python, iOS, and Android to access these services and make programming easy. Brought to you in partnership with IBM.

Topics:
java ,pdf ,pdfbox ,table

Published at DZone with permission of Tho Q Luong. See the original article here.

Opinions expressed by DZone contributors are their own.

THE DZONE NEWSLETTER

Dev Resources & Solutions Straight to Your Inbox

Thanks for subscribing!

Awesome! Check your inbox to verify your email so you can start receiving the latest in tech news and resources.

X

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}