If you want to test formatted text, the only way to do it is to render
a PDF page and compare the result with an image of the correctly formatted
content. Section
3.17: “Layout - Entire PDF Pages”
describes layout-tests using rendered pages.
And the utility RenderPdfToImages
renders a PDF document
page by page into PNG files.
:: :: Render PDF into image files. Each page as a file. :: @echo off setlocal set CLASSPATH=./lib/pdfunit-2015.10/*;%CLASSPATH% set CLASSPATH=./lib/jpedal/*;%CLASSPATH% set CLASSPATH=./lib/bouncycastle-jdk15on-150/*;%CLASSPATH% set TOOL=com.pdfunit.tools.RenderPdfToImages set OUT_DIR=./tmp set IN_FILE=compareToMaster.pdf set PASSWD= java %TOOL% %IN_FILE% %OUT_DIR% %PASSWD% endlocal
The input file compareToMaster.pdf
consists of 4 pages with
different images and text. The PDF Reader “SumatraPDF”
(http://code.google.com/p/sumatrapdf)
shows the first page:
After running the rendering program 4 files are created:
.\tmp\_rendered_compareToMaster_page1 .\tmp\_rendered_compareToMaster_page2 .\tmp\_rendered_compareToMaster_page3 .\tmp\_rendered_compareToMaster_page4
The first of these files looks the same as seen with the PDF Reader.
Internally, PDFUnit uses the same algorithm to render the pages as the rendering program does. So any difference found by a test is due to a change in the PDF document.
PDFUnit uses the class org.jpedal.PdfDecoder
from the project “jPedal”
(http://www.idrsolutions.com/).
Thanks to the developers.