[
https://issues.apache.org/jira/browse/PDFBOX-3062?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15029077#comment-15029077
]
Tilman Hausherr commented on PDFBOX-3062:
-----------------------------------------
I ran a test on about 2500 files. The change results in more splitting, i.e.
more stuff will be on different lines. This applies mostly for subscript and
superscript (where it happens often before the change too), but it will result
in better text extraction re: looking for tokens. And it will allow me to
resolve this issue.
One existing test file is modified: PDFBOX-3038-001033-p2.pdf. The "1" and "2"
are now on different lines than the text after the number. Two other
superscript numbers are on the same line - there is no universal solution for
this.
I am also adding some new test files to the repository, and adding the ones
that are possibly copyrighted to my own test set. If anyone is interested, I
can temporarly upload my complete test set to my website.
In case this change goes too far, one could try to set a different rule than
{{capHeight < glyphHeight}}, e.g. {{capHeight < glyphHeight * 0.75}}.
> Text extraction and height different in 2.0
> -------------------------------------------
>
> Key: PDFBOX-3062
> URL: https://issues.apache.org/jira/browse/PDFBOX-3062
> Project: PDFBox
> Issue Type: Sub-task
> Components: Text extraction
> Affects Versions: 2.0.0
> Reporter: Tilman Hausherr
> Assignee: Tilman Hausherr
> Attachments: 005021-reduced.pdf,
> PDFBOX-3062-H6NIYQXHLPGD3GI6SNIYINRAZBCDHUCB-reduced-marked-1.png,
> PDFBOX-3062-H6NIYQXHLPGD3GI6SNIYINRAZBCDHUCB-reduced.pdf,
> PDFBOX-3062-H6NIYQXHLPGD3GI6SNIYINRAZBCDHUCB.pdf,
> PDFBOX-3062-N2MOQ7YZICIYGTPLQJAWJ4HLN6CCEMHZ-reduced.pdf, garbled text 2.pdf
>
>
> AR:
> {code}
> WITH THE increasing complexity of optical modules,
> {code}
> 1.8:
> {code}
> WITH THE increasing complexity of optical modules,
> String[39.6,399.6 fs=1.0 xscale=29.888 height=20.114626 space=7.472
> width=28.214272]W
> String[69.488,386.16 fs=1.0 xscale=9.963 height=6.5955067 space=2.49075
> width=3.3176804]I
> String[72.80568,386.16 fs=1.0 xscale=9.963 height=6.5955067 space=2.49075
> width=6.0873947]T
> String[78.893074,386.16 fs=1.0 xscale=9.963 height=6.5955067 space=2.49075
> width=7.1932907]H
> String[90.71916,386.16 fs=1.0 xscale=9.963 height=6.5955067 space=2.49075
> width=6.0873947]T
> String[96.80656,386.16 fs=1.0 xscale=9.963 height=6.5955067 space=2.49075
> width=7.1932907]H
> {code}
> 2.0:
> {code}
> W
> ITH THE increasing complexity of optical modules,
> String[39.6,399.6 fs=1.0 xscale=29.888 height=9.584274 space=7.472
> width=28.209717]W
> String[69.488,386.16 fs=1.0 xscale=9.963 height=3.194865 space=2.49075
> width=3.3177567]I
> String[72.805756,386.16 fs=1.0 xscale=9.963 height=3.194865 space=2.49075
> width=6.0858]T
> String[78.891556,386.16 fs=1.0 xscale=9.963 height=3.194865 space=2.49075
> width=7.1949615]H
> String[90.719315,386.16 fs=1.0 xscale=9.963 height=3.194865 space=2.49075
> width=6.0858]T
> String[96.805115,386.16 fs=1.0 xscale=9.963 height=3.194865 space=2.49075
> width=7.1949615]H
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
