[
https://issues.apache.org/jira/browse/PDFBOX-5327?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17446965#comment-17446965
]
Michael Klink commented on PDFBOX-5327:
---------------------------------------
As [~tilman] has indicated, the problem is that the {{PDFTextStripper}}
forwards text to {{writeString}} in chunks it considers to be on the same text
line. The code that attempts to determine text lines is in {{{}writePage{}}}.
It mostly recognizes a text line by its shared *y* position and mostly ignores
large *x* gaps (except for adding a space there).
If your PDF had been properly tagged, the tagging structure could have been
used as a guide. If it had been properly organized into articles and article
beads, that structure could have been used as a guide. As neither of these
options is used in your PDF, there only are heuristics at work, and the
heuristics applied in {{writePage}} apparently have been developed mostly with
single column documents in mind or to test with.
> Parse text from two rectangles to one rectangle
> -----------------------------------------------
>
> Key: PDFBOX-5327
> URL: https://issues.apache.org/jira/browse/PDFBOX-5327
> Project: PDFBox
> Issue Type: Bug
> Components: Parsing
> Affects Versions: 2.0.24
> Reporter: zzz
> Priority: Major
> Attachments: 072B006805-P32939I-(2)(1).pdf, 1-1.png, 2-1.png, 3-1.png
>
>
> Rectangle(CONSIGN TO) and Rectangle(PACKING LIST) regard as one
> Rectangle(CONSIGN TO PACKING LIST)
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
