Re: 2.0.22 vs 2.0.23
Am 12.03.2021 um 14:05 schrieb Tim Allison: Many, many thanks to Tilman for running the regression tests! The 2 new exceptions are caused by PDFBOX-5127. I'm baffled that we haven't seen these before, but they do require some rare circumstances. Latest run: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_6.tar.xz the two files are gone. (Yeah this does not formally prove that we fixed it, but lets just believe it does) Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Am Freitag, dem 12.03.2021 um 08:15 -0500 schrieb Tim Allison: > > would it make sense to add that support? If yes could we get samples > > of > > various schema to support that development? Could look into that if > > we > > think that's worth the effort > > I think I can find some XMPs if they'd be of any use! :D That would be great - maybe together with expected extraction results - so I can start with proper unit tests. If you could add to https://issues.apache.org/jira/browse/PDFBOX-5128 that would be great. BR > > - > To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org > For additional commands, e-mail: dev-h...@pdfbox.apache.org > -- -- Maruan Sahyoun - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
> would it make sense to add that support? If yes could we get samples of > various schema to support that development? Could look into that if we > think that's worth the effort I think I can find some XMPs if they'd be of any use! :D - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Many, many thanks to Tilman for running the regression tests! The 2 new exceptions are caused by PDFBOX-5127. I'm baffled that we haven't seen these before, but they do require some rare circumstances. The 1 new Tika exception is a zero-byte file exception. This is my fault because I changed the files between Tilman's runs. As for XMPBox, Tilman is right that when I tried to use it many years ago, it did not have the flexibility needed for PDFs in the wild. See: https://lucene.472066.n3.nabble.com/DISCUSS-options-for-XMP-parsing-td4262520.html 2016 me: "I found that it fails on roughly 40% of XMPs I pulled out of PDFs from govdocs1/commoncrawl" Cheers, Tim On Thu, Mar 11, 2021 at 1:34 PM Tilman Hausherr wrote: > > Am 11.03.2021 um 09:00 schrieb sahy...@fileaffairs.de: > >> The three new exceptions weren't in earlier reports. > >> > >> IIRC the reason Tika uses Jempbox is because Xmpbox fails when there > >> is > >> a non standard schema. > > would it make sense to add that support? If yes could we get samles of > > various schema to support that development? Could look into that if we > > think that's worth the effort > > > Here's an example: > > https://issues.apache.org/jira/browse/PDFBOX-3440 > > > Tilman > > > > > > > Maruan > > > > > >> Tilman > >> > >> > >> - > >> To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org > >> For additional commands, e-mail: dev-h...@pdfbox.apache.org > >> > > > - > To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org > For additional commands, e-mail: dev-h...@pdfbox.apache.org > - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Am 11.03.2021 um 09:00 schrieb sahy...@fileaffairs.de: The three new exceptions weren't in earlier reports. IIRC the reason Tika uses Jempbox is because Xmpbox fails when there is a non standard schema. would it make sense to add that support? If yes could we get samles of various schema to support that development? Could look into that if we think that's worth the effort Here's an example: https://issues.apache.org/jira/browse/PDFBOX-3440 Tilman Maruan Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Am Donnerstag, dem 11.03.2021 um 07:56 +0100 schrieb Tilman Hausherr: > Am 11.03.2021 um 07:46 schrieb Andreas Lehmkuehler: > > Am 11.03.21 um 07:24 schrieb Tilman Hausherr: > > > new report > > > http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_5.tar.xz > > > > > > The content differences part is now the smallest ever, likely due > > > to > > > my change in tika-eval (TIKA-3314) and restoring a PDFBox code > > > segment I accidentally deleted (PDFBOX-5115). > > Cool!! > > > > > There are three new exceptions. Two are in jempbox and one is in > > > tika > > > itself so I suspect PDFBox isn't to blame. I'll look at it too if > > > I > > > have the time. > > As far as I remember the jempbox issue isn't new, Tim mentioned it > > some time ago. Just out of curiosity does it make sense to use an > > old > > lib to extract metadata? Is there anything missing in xmpbox but > > available in jempbox? > > > The three new exceptions weren't in earlier reports. > > IIRC the reason Tika uses Jempbox is because Xmpbox fails when there > is > a non standard schema. would it make sense to add that support? If yes could we get samles of various schema to support that development? Could look into that if we think that's worth the effort Maruan > > Tilman > > > - > To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org > For additional commands, e-mail: dev-h...@pdfbox.apache.org > -- -- Maruan Sahyoun FileAffairs GmbH Josef-Schappe-Straße 21 40882 Ratingen Tel: +49 (2102) 89497 88 Fax: +49 (2102) 89497 91 sahy...@fileaffairs.de www.fileaffairs.de Geschäftsführer: Maruan Sahyoun Handelsregister: AG Düsseldorf, HRB 53837 UST.-ID: DE248275827 - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Am 11.03.2021 um 07:46 schrieb Andreas Lehmkuehler: Am 11.03.21 um 07:24 schrieb Tilman Hausherr: new report http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_5.tar.xz The content differences part is now the smallest ever, likely due to my change in tika-eval (TIKA-3314) and restoring a PDFBox code segment I accidentally deleted (PDFBOX-5115). Cool!! There are three new exceptions. Two are in jempbox and one is in tika itself so I suspect PDFBox isn't to blame. I'll look at it too if I have the time. As far as I remember the jempbox issue isn't new, Tim mentioned it some time ago. Just out of curiosity does it make sense to use an old lib to extract metadata? Is there anything missing in xmpbox but available in jempbox? The three new exceptions weren't in earlier reports. IIRC the reason Tika uses Jempbox is because Xmpbox fails when there is a non standard schema. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Am 11.03.21 um 07:24 schrieb Tilman Hausherr: new report http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_5.tar.xz The content differences part is now the smallest ever, likely due to my change in tika-eval (TIKA-3314) and restoring a PDFBox code segment I accidentally deleted (PDFBOX-5115). Cool!! There are three new exceptions. Two are in jempbox and one is in tika itself so I suspect PDFBox isn't to blame. I'll look at it too if I have the time. As far as I remember the jempbox issue isn't new, Tim mentioned it some time ago. Just out of curiosity does it make sense to use an old lib to extract metadata? Is there anything missing in xmpbox but available in jempbox? Andreas Tilman Am 08.03.2021 um 11:17 schrieb Tilman Hausherr: new report: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_3.tar.xz Tilman Am 08.03.2021 um 10:35 schrieb Tilman Hausherr: I think we're good (despite the differences, most of which are because of the soft hyphen), but I'm now experimenting with a modified version of tika-eval to see what happens. Tilman Am 07.03.2021 um 19:47 schrieb Tilman Hausherr: new report at http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_2.tar.xz Tilman Am 07.03.2021 um 11:43 schrieb Tilman Hausherr: Am 07.03.2021 um 06:04 schrieb Tilman Hausherr: Report is here: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23.tar.xz There's not much changed. No new exceptions. Re content, the changes that seem important are all related to "soft hyphen". https://issues.apache.org/jira/browse/PDFBOX-5115 I am currently fixing this, and then I'll run the tests again. The text extraction differences will likely stay. It's possible that a change in tika-eval is needed too. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
new report http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_5.tar.xz The content differences part is now the smallest ever, likely due to my change in tika-eval (TIKA-3314) and restoring a PDFBox code segment I accidentally deleted (PDFBOX-5115). There are three new exceptions. Two are in jempbox and one is in tika itself so I suspect PDFBox isn't to blame. I'll look at it too if I have the time. Tilman Am 08.03.2021 um 11:17 schrieb Tilman Hausherr: new report: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_3.tar.xz Tilman Am 08.03.2021 um 10:35 schrieb Tilman Hausherr: I think we're good (despite the differences, most of which are because of the soft hyphen), but I'm now experimenting with a modified version of tika-eval to see what happens. Tilman Am 07.03.2021 um 19:47 schrieb Tilman Hausherr: new report at http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_2.tar.xz Tilman Am 07.03.2021 um 11:43 schrieb Tilman Hausherr: Am 07.03.2021 um 06:04 schrieb Tilman Hausherr: Report is here: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23.tar.xz There's not much changed. No new exceptions. Re content, the changes that seem important are all related to "soft hyphen". https://issues.apache.org/jira/browse/PDFBOX-5115 I am currently fixing this, and then I'll run the tests again. The text extraction differences will likely stay. It's possible that a change in tika-eval is needed too. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Hi, @Tilman Thanks for running all those tests. The results are looking good to me, although I've to admit that I don't understand every bit of those sheets especially those about the content ;-) However, I'm planning to cut the 2.0.23 release next Monday. Andreas Am 08.03.21 um 11:17 schrieb Tilman Hausherr: new report: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_3.tar.xz Tilman Am 08.03.2021 um 10:35 schrieb Tilman Hausherr: I think we're good (despite the differences, most of which are because of the soft hyphen), but I'm now experimenting with a modified version of tika-eval to see what happens. Tilman Am 07.03.2021 um 19:47 schrieb Tilman Hausherr: new report at http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_2.tar.xz Tilman Am 07.03.2021 um 11:43 schrieb Tilman Hausherr: Am 07.03.2021 um 06:04 schrieb Tilman Hausherr: Report is here: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23.tar.xz There's not much changed. No new exceptions. Re content, the changes that seem important are all related to "soft hyphen". https://issues.apache.org/jira/browse/PDFBOX-5115 I am currently fixing this, and then I'll run the tests again. The text extraction differences will likely stay. It's possible that a change in tika-eval is needed too. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
new report: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_3.tar.xz Tilman Am 08.03.2021 um 10:35 schrieb Tilman Hausherr: I think we're good (despite the differences, most of which are because of the soft hyphen), but I'm now experimenting with a modified version of tika-eval to see what happens. Tilman Am 07.03.2021 um 19:47 schrieb Tilman Hausherr: new report at http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_2.tar.xz Tilman Am 07.03.2021 um 11:43 schrieb Tilman Hausherr: Am 07.03.2021 um 06:04 schrieb Tilman Hausherr: Report is here: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23.tar.xz There's not much changed. No new exceptions. Re content, the changes that seem important are all related to "soft hyphen". https://issues.apache.org/jira/browse/PDFBOX-5115 I am currently fixing this, and then I'll run the tests again. The text extraction differences will likely stay. It's possible that a change in tika-eval is needed too. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
I think we're good (despite the differences, most of which are because of the soft hyphen), but I'm now experimenting with a modified version of tika-eval to see what happens. Tilman Am 07.03.2021 um 19:47 schrieb Tilman Hausherr: new report at http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_2.tar.xz Tilman Am 07.03.2021 um 11:43 schrieb Tilman Hausherr: Am 07.03.2021 um 06:04 schrieb Tilman Hausherr: Report is here: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23.tar.xz There's not much changed. No new exceptions. Re content, the changes that seem important are all related to "soft hyphen". https://issues.apache.org/jira/browse/PDFBOX-5115 I am currently fixing this, and then I'll run the tests again. The text extraction differences will likely stay. It's possible that a change in tika-eval is needed too. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
new report at http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23_2.tar.xz Tilman Am 07.03.2021 um 11:43 schrieb Tilman Hausherr: Am 07.03.2021 um 06:04 schrieb Tilman Hausherr: Report is here: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23.tar.xz There's not much changed. No new exceptions. Re content, the changes that seem important are all related to "soft hyphen". https://issues.apache.org/jira/browse/PDFBOX-5115 I am currently fixing this, and then I'll run the tests again. The text extraction differences will likely stay. It's possible that a change in tika-eval is needed too. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org
Re: 2.0.22 vs 2.0.23
Am 07.03.2021 um 06:04 schrieb Tilman Hausherr: Report is here: http://home.snafu.de/tilman/tmp/reports_pdfbox_2.0.22_vs_2.0.23.tar.xz There's not much changed. No new exceptions. Re content, the changes that seem important are all related to "soft hyphen". https://issues.apache.org/jira/browse/PDFBOX-5115 I am currently fixing this, and then I'll run the tests again. The text extraction differences will likely stay. It's possible that a change in tika-eval is needed too. Tilman - To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org For additional commands, e-mail: dev-h...@pdfbox.apache.org