[TIKA-742] PDF2XHTML fails to insert <p> nor space around page marker

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.0
• Component/s: parser
• Assignee: Michael McCandless
• Reporter: Michael McCandless
• Created: 04/Oct/11 10:30
• Updated: 05/Oct/11 10:43
• Resolved: 05/Oct/11 10:43

Description

I have a test document (unfortunately not committable) whose page
numbers are rendered with no separator (<p> nor space) before the next
word. So I have words like:

• 1Massachusetts
• 2Course
• 3also
• 4The

But then when I ran the ExtractText -html command-line from PDFBox, I
can see that <p> is inserted after these page numbers (spookily, not
closing the previous <p>; I opened PDFBOX-1130 for that).

So I made a simple change to Tika's PDF2XHTML, to have it override the
writeStart/EndParagraph, and call handler.start/EndElement("p"), ie to
preserve the paragraph structure that PDFBOX detects out to the
resulting XHTML handler, and this fixes the issue (I now see the page
number as a separate paragraph, rendered w/ newline in "text" mode
from TikaCLI).

Note that this test document is the same document from PDFBOX-1129
(there are some quote characters that are not extracted correctly).