As reported by PDFBox jira PDFBOX-790 (https://issues.apache.org/jira/browse/PDFBOX-790), exception occurs when indexing MS Word PDF files. One can repeat the problem with the attached pdf file. Fixed for 1.3.1.
The most current release of PDFBox project is 1.5.0, and Search is on 1.0.0 now. Need to upgrade it to higher version.