[jira] [Updated] (PDFBOX-1804) PDFTextStripper Issue related to word positions not correctly being parsed

["Joe Hosteny (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/PDFBOX-1804?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Joe Hosteny updated PDFBOX-1804:
--------------------------------

    Attachment: PDFBOX-1804.patch

I can confirm seeing this issue as well, and that the proposed fix works. I've attached a patch for the issue.

> PDFTextStripper Issue related to word positions not correctly being parsed
> --------------------------------------------------------------------------
>
>                 Key: PDFBOX-1804
>                 URL: https://issues.apache.org/jira/browse/PDFBOX-1804
>             Project: PDFBox
>          Issue Type: Bug
>          Components: Text extraction
>    Affects Versions: 1.8.3
>            Reporter: Andy Phillips
>         Attachments: PDFBOX-1804.patch
>
>
> I found in a PDF I was pulling text from by using a custom PDFTextStripper subclass that overrides writeString(String text, List<TextPosition> textPositions) that i was getting the wrong textPositions that were not lined up with the text.   I found that the test position of all “words” in a line always come over as the “last” text positions of the last word in the line.   I found the issue in the PDFTextStripper class
> So here is the Code Issue:
>     /**
>      * Used within {@link #normalize(List, boolean, boolean)} to handle a {@link TextPosition}.
>      * @return The StringBuilder that must be used when calling this method.
>      */
>     private StringBuilder normalizeAdd(LinkedList<WordWithTextPositions> normalized,
>             StringBuilder lineBuilder, List<TextPosition> wordPositions, TextPosition text)
>     {
>         if (text instanceof WordSeparator) 
>         {
>             normalized.add(createWord(lineBuilder.toString(), wordPositions));
>             lineBuilder = new StringBuilder();
>             wordPositions.clear();
>         }
>         else 
>         {
>             lineBuilder.append(text.getCharacter());
>             wordPositions.add(text);
>         }
>         return lineBuilder;
>     }
> When the normalizeAdd method, you create a new word passing the wordPositions.   A reference to the wordPositions is stored in the new WordWithTextPositions in the normalized linked list, but in the next line, you clear().   Since the last wordPositions was passed as a reference, the wordPositions is cleared in the WordWithTextPositions you just created.
> Soo, i would suggest you do the following:
> /**
>      * Used within {@link #normalize(List, boolean, boolean)} to handle a {@link TextPosition}.
>      * @return The StringBuilder that must be used when calling this method.
>      */
>     private StringBuilder normalizeAdd(LinkedList<WordWithTextPositions> normalized,
>             StringBuilder lineBuilder, List<TextPosition> wordPositions, TextPosition text)
>     {
>         if (text instanceof WordSeparator) 
>         {
>             normalized.add(createWord(lineBuilder.toString(), new ArrayList<TextPosition>(wordPositions)));
>             lineBuilder = new StringBuilder();
>             wordPositions.clear();
>         }
>         else 
>         {
>             lineBuilder.append(text.getCharacter());
>             wordPositions.add(text);
>         }
>         return lineBuilder;
>     }



--
This message was sent by Atlassian JIRA
(v6.1.4#6159)