You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tika.apache.org by Chris Mattmann <ma...@apache.org> on 2020/08/29 18:49:16 UTC
Re: [EXTERNAL] I have some questions about tika-python
Thanks for reaching out Aditya and for using Tika Python. This issue is
best solved upstream in dev@tika.apache.org so I am copying that list
and making it the reply to.
The issue likely lies in the PDFBox algorithm. There are PDFBox folks on
this list. They can help you. Hopefully there is a simple config setting
to help out.
Cheers,
Chris
From: Aditya Sardesai <ad...@persistent.com>
Date: Thursday, August 27, 2020 at 11:44 PM
To: "Mattmann, Chris A (US 1740)" <ch...@jpl.nasa.gov>
Subject: [EXTERNAL] I have some questions about tika-python
Greetings Chris,
We had a requirement for our project which required parsing PDF files and extracting the text for some verification. I tried a number of other python packages but they all had issues recognizing text consistently across the file.
The most common issue which we faced was text not dumped the correct sequence. This was until we found Tika. We are very impressed by the recognition of text sequencing. It is exactly how we want.
However, we're facing an issue with vertically aligned text. There are two examples of vertically aligned text which I can show. In one instance the text is parsed correctly but not in the other.
Ex1.
In this the word values is read as,
V
al
ue
s
Ex2.
In this, the date is parsed correctly as,
2020-07-16 00:30
Can you please help us understand if there are some specifics about the tika algorithm, we should be aware of? Any suggestions on how we can better use the tool?
Please let me know if I need to connect with any other contributor for this.
Looking forward to your valuable comments.
Regards,
__
Aditya Sardesai
Lead Quality Engineer
aditya_sardesai@persistent.com
Connect with me on: LinkedIn
See Beyond, Rise Above
Re: [EXTERNAL] I have some questions about tika-python
Posted by Tilman Hausherr <TH...@t-online.de>.
The "detectAngles" option in the config.xml should do what you want,
this was done here: https://issues.apache.org/jira/browse/TIKA-2779
The config.xml should be somewhat like this:
<properties>
<parsers>
<parser class="org.apache.tika.parser.DefaultParser">
<parser-exclude class="org.apache.tika.parser.pdf.PDFParser"/>
</parser>
<parser class="org.apache.tika.parser.pdf.PDFParser">
<params>
<param name="detectAngles" type="bool">true</param>
</params>
</parser>
</parsers>
</properties>
Tilman
Am 29.08.2020 um 20:49 schrieb Chris Mattmann:
>
> Thanks for reaching out Aditya and for using Tika Python. This issue is
> best solved upstream in dev@tika.apache.org
> <ma...@tika.apache.org> so I am copying that list
> and making it the reply to.
>
> The issue likely lies in the PDFBox algorithm. There are PDFBox folks on
> this list. They can help you. Hopefully there is a simple config setting
> to help out.
>
> Cheers,
>
> Chris
>
> *From: *Aditya Sardesai <ad...@persistent.com>
> *Date: *Thursday, August 27, 2020 at 11:44 PM
> *To: *"Mattmann, Chris A (US 1740)" <ch...@jpl.nasa.gov>
> *Subject: *[EXTERNAL] I have some questions about tika-python
>
> Greetings Chris,
>
> We had a requirement for our project which required parsing PDF files
> and extracting the text for some verification. I tried a number of
> other python packages but they all had issues recognizing text
> consistently across the file.
>
> The most common issue which we faced was text not dumped the correct
> sequence. This was until we found Tika. We are very impressed by the
> recognition of text sequencing. It is exactly how we want.
>
> However, we're facing an issue with vertically aligned text. There are
> two examples of vertically aligned text which I can show. In one
> instance the text is parsed correctly but not in the other.
>
> *Ex1.*
>
> In this the word values is read as,
>
> V
>
> al
>
> ue
>
> s
>
> *Ex2.*
>
> In this, the date is parsed correctly as,
>
> 2020-07-16 00:30
>
> Can you please help us understand if there are some specifics about
> the tika algorithm, we should be aware of? Any suggestions on how we
> can better use the tool?
>
> Please let me know if I need to connect with any other contributor for
> this.
>
> Looking forward to your valuable comments.
>
> Regards,
>
> *__*
>
> **
>
> *Aditya Sardesai*
>
> Lead Quality Engineer
>
>
> aditya_sardesai@persistent.com <ma...@persistent.com>
>
> Connect with me on: LinkedIn
> <https://urldefense.us/v3/__https:/www.linkedin.com/in/aditya-sardesai-95774372/__;!!PvBDto6Hs4WbVuu7!f_18QJjKT8ui7bVboDUUhsPnSM1ozLuHT0Z0wxbSmaR8_QJLUArBMPqTiQbLYCAmlzjmAd6lBg$>
>
>
> <https://urldefense.us/v3/__http:/www.persistent.com/__;!!PvBDto6Hs4WbVuu7!f_18QJjKT8ui7bVboDUUhsPnSM1ozLuHT0Z0wxbSmaR8_QJLUArBMPqTiQbLYCAmlzjSngnzAA$>
> See Beyond, Rise Above
>