Calibre Crashes on PDF->Epub Conversion (0.9.4.3)

Merischino · 08-11-2013, 07:12 PM

Hi there...

having a lot of trouble trying to convert PDFs to epub... have had lots of crashes over the last 24-48 hours after installing 0.9.4.3.

Notice that before, I could do conversions quite quickly on a bulky chunk of books (100+)
Since losing all my hardware and, having replaced it, am trying to re-instate my 3000+ title calibre library, I'm doing a lot of metadata downloading and file conversion (trying to embed metadata in the file this go-round, which I omitted last time.)

Not sure why, but for the first time I'm getting 3-4 hours to convert 20 pdfs or
just a complete crash/hang of my whole computer (20 or less files, more than 50 files, number of files not apparently a factor) requiring a hard reboot of the computer

Also never had this before, but starting to get errors on specific pdf's instructing me to turn on heuristic processing. Maybe that is why it's taking so long where before I could convert a complete series, grab some coffee, and begin work on the next one?

Or is this maybe all a function of something wrong in 0.9.4.3?

Would appreciate any help... thanks!

M-

kovidgoyal · 08-11-2013, 11:07 PM

There have been no changes to PDF conversion in calibre in a very long time. If you have an individual pdf that is causing problems feel free to attach it to a bug report and I will look at it.

Merischino · 08-12-2013, 03:48 PM

This is the debug output from one of 17 problem pdf files I've been trying to convert:
(bug processor put all this into one directory, named "input")

Spoiler:

theducks · 08-12-2013, 08:50 PM

Quote:

Originally Posted by Merischino

This is the debug output from one of 17 problem pdf files I've been trying to convert:
(bug processor put all this into one directory, named "input")

Spoiler:

Those are all images

Merischino · 08-12-2013, 09:04 PM

Quote:

Originally Posted by theducks

Those are all images

images are all that populated the bug output directory (called "input")
I didn't get any other results than that from the bug reporter.

BetterRed · 08-12-2013, 09:04 PM

@Merischino - is there any possibility that you are running multiple anti-virus programs on your new hardware

Doing that has been known to really slow things down, because the two programs are fighting against one another and potentially creating an impasse, where A wants what B has, and B wants what A has. Time outs will eventually break the impasse (deadlock, deadly embrace) - meantime everything is blocked.

BR

Merischino · 08-12-2013, 09:07 PM

Quote:

Originally Posted by BetterRed

@Merischino - is there any possibility that you are running multiple anti-virus programs on your new hardware

Actually, at the moment I'm not running any virus software. (when I do, it's implemented at the router. Nothing on the local machine anti-virus wise. But no anti-virus is running at all at the moment.)

BetterRed · 08-12-2013, 09:37 PM

Quote:

Originally Posted by Merischino

Actually, at the moment I'm not running any virus software. (when I do, it's implemented at the router. Nothing on the local machine anti-virus wise. But no anti-virus is running at all at the moment.)

Have you tried reverting to the version of Calibre you were using on the old hardware - in the download page for the different OS's there's link to where you can get earlier versions.

BR

Merischino · 08-12-2013, 10:40 PM

Quote:

How to ask a question about conversion problems
Conversion of ebooks from one format to another is complex. A conversion is the result of the interaction of thousands of different inputs and subsystems. In order to ask a question in a way that is likely to get help, you must provide at least the following information:

The file you are trying to convert. If the file is copyrighted then create an extract/sample that shows the problem, or use the calibre bug report system to create a bug with the file attached.
The conversion log from calibre. You get this by clicking the rotating spinner in the bottom right of the main calibre window and using the show job details button.
The output file, if any.
A description of what exactly the problem is. If there is an error message when converting then the full error message, which you can get by clicking the show details button on the error message.
A description of any conversion options you changed from the default

It is particularly important that you provide us with a way to reproduce your problem. That makes it much easier to answer your question. That means you provide at least the input file and a description of the problem as well as any non default conversion settings you use.

ok so in order,
the file I've chosen to report is one of dozens, but here it is.
Title: Mysterious Creatures_ A Guide to Cryptozoology, 2 Volume Set - George Eberhart.pdf
As it's copyrighted, I've uploaded only an excerpt:
File Attached: png image of one page with a graphic: "Mysterious Creature (page).png

create a bug with the file attached in the calibre bug system ... as far as I can tell, there isn't a calibre bug system in calibre, but outside the application there is a website at Launchpad where you can report new bugs etc. I think it's premature to use that here, so I think there must be some bug plug or something that I do not have installed that I'm meant to be using. I can't find it.

The conversion log from calibre:
very large. saved it as a text file:
conversion log 1
conversion log 2
Note: i noticed a boat load (many many pages of detailed log history) of hyphenation changes, so I changed my conversion specs to not include heuristic processing or hyphenation correction, reran it. (I had only started using heuristic processing after researching what was going on with my pdf conversion as somewhere I found a post recommending adding heursitic, changing the "split" amount to something higher, and something else I don't remember).

conversion output:
two sets, both included only collections of images which calibre's bug doohickey put into a directory called "input"
Both of them are very lengthy collections of images extracted from the pdf. some examples are attached.

Both conversions I ran tonight timed out at 10 minutes. I can't find how/where to change the timeout limit, but 10 minutes for one file does seem rather long...

I don't have conversion logs for the initial failures of my several hundreds of pdfs, however before I started messing around trying to fix this problem, the resulting failures were not time-out failures, but actual failures. I am sorry I did not figure out the log stuff until after I'd made so many changes.

My latest conversion options are as simple as possible:
default convert to epub with these exceptions:
a) confirmed no heuristic processing or special requests for hypenation, line changes, etc but does include a check box for the cover aspect ratio retention
b) The first conversion specified "tablet" (I think, that's what it's been on), the second specified "ipad" output, both did request insertion of updated metadata jacket and cover
c) input set as "generic"
I think that about does it.

BetterRed · 08-12-2013, 10:46 PM

Bug reporting goes here https://bugs.launchpad.net/calibre

To protect copyright content you can mark reports as Private so that only the developers can read

BR

PeterT · 08-12-2013, 11:46 PM

I know it might be heresy here, but, calibre is not the be-all and end-all for PDF conversions.

As an experiment I located a copy of the book on the web (all 772 pages of it in a PDF) and tried a conversion in Nitro PDF Professional 6.2.044 (an older version). In less than 4 minutes it had done a conversion to a Word DOC file; and from looking at it, a pretty good conversion.

It might well be worth while to find a PDF editor you are happy with (many of them have trial versions) and use that to do a conversion to a Word document.

BetterRed · 08-13-2013, 12:20 AM

Quote:

Originally Posted by PeterT

As an experiment I located a copy of the book on the web (all 772 pages of it in a PDF)

@PeterT it wasn't hard to find was it

I converted it to a PRC with MobiCreator, then to epub with Calibre - not perfect, but what is

Took a few minutes to download and do the 2 conversions - less time that its taken me to type this, MobiCreator only runs on Windows, but I think its based on one of the well know PDF2???? open source converters, so...

With PDF conversions I always start with MobiCreator - only if it produces really bad results do I try anything else. Calibre always does pretty good prc->epub conversions

BR

kovidgoyal · 08-13-2013, 01:50 AM

@Merishchino: I'm going to need an actual PDF file.

DoctorOhh · 08-14-2013, 12:46 AM

Quote:

Originally Posted by Merischino

Both conversions I ran tonight timed out at 10 minutes. I can't find how/where to change the timeout limit, but 10 minutes for one file does seem rather long...

The 10 minute time limit setting can be found by going to Preferences - Miscellaneous. Change it to Never abort and try again. Running longer than 10 minutes isn't unusual especially for PDF conversions using Heuristics.

@PeterT and @BetterRed are correct in pointing out that other options are available.

Quote:

Originally Posted by PeterT

I know it might be heresy here, but, calibre is not the be-all and end-all for PDF conversions.

I'm surprised to hear you say this. You have been here long enough (forever?

) to know that we are up front about the limits of PDF conversions with calibre and often refer folks to other solutions.

Moderator Notice
Moved to Conversion sub-forum.

Merischino · 08-14-2013, 02:29 PM

Quote:

Originally Posted by kovidgoyal

@Merishchino: I'm going to need an actual PDF file.

I didn't see where to mark my pdf upload as private. hoping since others found and downloaded it off the internet that i'm not going to be the reason the author doesn't get money.

Moderator Notice
removed attachment. Not posted to Calibre Bug reporting site

08-11-2013, 07:12 PM	#1
Merischino Groupie Posts: 199 Karma: 436584 Join Date: Jul 2010 Location: the beautiful Pacific Northwest Device: Kindle Oasis, Pixel 2, Kindle Fire HD 8	Calibre Crashes on PDF->Epub Conversion (0.9.4.3) Hi there... having a lot of trouble trying to convert PDFs to epub... have had lots of crashes over the last 24-48 hours after installing 0.9.4.3. Notice that before, I could do conversions quite quickly on a bulky chunk of books (100+) Since losing all my hardware and, having replaced it, am trying to re-instate my 3000+ title calibre library, I'm doing a lot of metadata downloading and file conversion (trying to embed metadata in the file this go-round, which I omitted last time.) Not sure why, but for the first time I'm getting 3-4 hours to convert 20 pdfs or just a complete crash/hang of my whole computer (20 or less files, more than 50 files, number of files not apparently a factor) requiring a hard reboot of the computer Also never had this before, but starting to get errors on specific pdf's instructing me to turn on heuristic processing. Maybe that is why it's taking so long where before I could convert a complete series, grab some coffee, and begin work on the next one? Or is this maybe all a function of something wrong in 0.9.4.3? Would appreciate any help... thanks! M-

08-11-2013, 11:07 PM	#2
kovidgoyal creator of calibre Posts: 45,196 Karma: 27110894 Join Date: Oct 2006 Location: Mumbai, India Device: Various	There have been no changes to PDF conversion in calibre in a very long time. If you have an individual pdf that is causing problems feel free to attach it to a bug report and I will look at it. Last edited by kovidgoyal; 08-11-2013 at 11:12 PM.

08-12-2013, 03:48 PM	#3
Merischino Groupie Posts: 199 Karma: 436584 Join Date: Jul 2010 Location: the beautiful Pacific Northwest Device: Kindle Oasis, Pixel 2, Kindle Fire HD 8	This is the debug output from one of 17 problem pdf files I've been trying to convert: (bug processor put all this into one directory, named "input") Spoiler: index-1_1.jpg index-37_1.png index-38_1.png index-61_1.png index-67_1.png index-76_1.png index-82_1.png index-83_1.png index-89_1.png index-90_1.png index-92_1.png index-94_1.png index-96_1.png index-101_1.png index-104_1.png index-109_1.png index-118_1.png index-122_1.png index-134_1.png index-135_1.png index-136_1.png index-141_1.png index-145_1.png index-147_1.png index-156_1.png index-172_1.png index-179_1.png index-187_1.png index-190_1.png index-192_1.png index-196_1.png index-202_1.png index-204_1.png index-213_1.png index-219_1.png index-234_1.png index-238_1.png index-239_1.png index-254_1.png index-264_1.png index-265_1.png index-273_1.png index-279_1.png index-300_1.png index-302_1.png index-306_1.png index-308_1.png index-314_1.png index-316_1.png index-318_1.png index-319_1.png index-325_1.png index-329_1.png index-332_1.png index-341_1.png index-350_1.png index-360_1.png index-379_1.png index-380_1.png index-392_1.png index-395_1.png index-398_1.png index-399_1.png index-404_1.png index-407_1.png index-410_1.png index-425_1.png index-426_1.png index-427_1.png index-429_1.png index-440_1.png index-441_1.png index-449_1.png index-450_1.png index-468_1.png index-471_1.png index-472_1.png index-473_1.png index-482_1.png index-488_1.png index-496_1.png index-502_1.png index-510_1.png index-512_1.png index-519_1.png index-520_1.png index-526_1.png index-528_1.png index-529_1.png index-531_1.png index-534_1.png index-543_1.png index-551_1.png index-555_1.png index-562_1.png index-563_1.png index-565_1.png index-569_1.png index-574_1.png index-578_1.png index-580_1.png index-581_1.png index-582_1.png index-593_1.png index-596_1.png index-597_1.png index-598_1.png index-606_1.png index-612_1.png index-617_1.png index-639_1.png index-641_1.png index-654_1.png index-658_1.png index-660_1.png index-666_1.png index-681_1.png index-682_1.png index-684_1.png index-694_1.png index-696_1.png index-700_1.png index-708_1.png index-719_1.png index.html metadata.opf Last edited by theducks; 08-12-2013 at 08:50 PM. Reason: Wrapped long paste in Spoiler

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Conversion from ePub to mobi crashes Calibre	click22	Conversion	4	02-22-2013 10:56 AM
HTML to EPUB conversion crashes	jakev383	Conversion	2	03-23-2011 11:00 AM
ePub conversion crashes reader	shmish	Calibre	1	03-07-2010 12:16 PM
epub to pdf conversion using calibre	rblearn	Calibre	0	02-23-2010 04:57 PM
Calibre 0.51 Crashes on EPUB conversion	fogus	Calibre	2	03-14-2009 09:47 PM

08-12-2013, 09:04 PM	#6
BetterRed null operator (he/him) Posts: 21,630 Karma: 29710510 Join Date: Mar 2012 Location: Sydney Australia Device: none	@Merischino - is there any possibility that you are running multiple anti-virus programs on your new hardware Doing that has been known to really slow things down, because the two programs are fighting against one another and potentially creating an impasse, where A wants what B has, and B wants what A has. Time outs will eventually break the impasse (deadlock, deadly embrace) - meantime everything is blocked. BR

08-12-2013, 10:46 PM	#10
BetterRed null operator (he/him) Posts: 21,630 Karma: 29710510 Join Date: Mar 2012 Location: Sydney Australia Device: none	Bug reporting goes here https://bugs.launchpad.net/calibre To protect copyright content you can mark reports as Private so that only the developers can read BR

08-12-2013, 11:46 PM	#11
PeterT Grand Sorcerer Posts: 13,306 Karma: 78876004 Join Date: Nov 2007 Location: Toronto Device: Libra H2O, Libra Colour	I know it might be heresy here, but, calibre is not the be-all and end-all for PDF conversions. As an experiment I located a copy of the book on the web (all 772 pages of it in a PDF) and tried a conversion in Nitro PDF Professional 6.2.044 (an older version). In less than 4 minutes it had done a conversion to a Word DOC file; and from looking at it, a pretty good conversion. It might well be worth while to find a PDF editor you are happy with (many of them have trial versions) and use that to do a conversion to a Word document.

08-13-2013, 01:50 AM	#13
kovidgoyal creator of calibre Posts: 45,196 Karma: 27110894 Join Date: Oct 2006 Location: Mumbai, India Device: Various	@Merishchino: I'm going to need an actual PDF file.

Advert

Advert