Issue 501: images with same name in different directories collide in final epub file

[KevinH]

Hi Actually,

This is a bug. Ignore the epub he created and all of the pieces. If you look in his html folder inside of book1/ you will find 3 very correct .xhtml files that properly show the images.

If he right clicks on Text in Sigil (lhs panel) and imports the chapter1.xhtml, all of the images from ch1 are properly imported.

If he again right clicks on Text in Sigil (lhs panel) and adds existing file chapter2.xhtml, his images are not properly grabbed from the ch2 folder (or they overwrite the last set?).

If he again right clicks on Text in Sigil (lhs panel) and adds existing file chapter3.xhtml, his images are again either not properly grabbed from ch3 folder or they overwrite images of the same name.

Since the underlying html is correct and works (including links to images), then I think this process should work in Sigil as well. It does not. I am so used to importing just one large xhtml file with images inside of it and splitting it into chapters, that I have never tried importing/adding existing xhtml files.

Perhaps there is a better way (try importing the images first?) but I do think as long as the underlying html files are proper, the order of how you import them should not impact the image files (subdirectories) should be seen and the directory structure should be recreated in the Images subdirectory.

Again, I could be all wet here but it does look like an issue that needs to be addressed for people that try and import multiple xhtml files into one project with subdirectories used to house different images all with the same name.

I assume just renaming the images files would work here but ...

My 2 cents,

KevinH

[Ray Tayek]

Originally Posted by JSWolf

In your ePub, you've messed things up big time. You do not have the directories for the images for chapter 2 & chapter 3. So the XML code is looking for the images where they do not exist. In fact, they aren't even in the ePub.

yes, that's true.

but mobi creator and calibre can eat this just fine. all of the diagrams are there.

so i thought sigil was broken.

thanks

[Ray Tayek]

Originally Posted by KevinH

Hi Actually,

This is a bug. Ignore the epub he created and all of the pieces. If you look in his html folder inside of book1/ you will find 3 very correct .xhtml files that properly show the images.

If he right clicks on Text in Sigil (lhs panel) and imports the chapter1.xhtml, all of the images from ch1 are properly imported.

If he again right clicks on Text in Sigil (lhs panel) and adds existing file chapter2.xhtml, his images are not properly grabbed ...

Since the underlying html is correct and works (including links to images), then I think this process should work in Sigil as well. It does not. ...

Perhaps there is a better way (try importing the images first?) ...

I assume just renaming the images files would work here but ...

KevinH

sigil will not let me import a folder.

the images have the same name in many cases. renaming 500 images would be a pain. i have instructed the author to not insert the images into word (or anything) but to insert a link instead.

i could prefix the folder name to each image, but that sorta feels wrong.

calibre and creator handle this problem.

maybe http://code.google.com/p/sigil/sourc...l?r=255694a4c8 did not fix all of the problems.

thanks

[KevinH]

Hi,

Just as a test, I created a chapter0.xhtml that includes links to all of the other chapters and links to ALL of the images.

When I import that file into Sigil, it recognizes that name clash and properly creates image001.gif through image009.gif and no collision happens. Later importing of the other html files still ends up with broken links.

I guess the key issue here is that you can not easily import multiple html files to form your ebook and expect the images (even is stored in different directories) to not collide.

I think there is a python script that can be run to merge multiple html files with links into one big html which could then be properly imported into Sigil and split into chapters.

Perhaps, importing an opf file with links to everything might work in Sigil. I have not tried it. I know that calibre requires that there be some file that names all of the other files and images being used upon import of an htmlz archive. Perhaps something like that might work with Sigil right now, but I have never done it.

Sorry I can't be more help here.

KevinH

[Ray Tayek]

Originally Posted by KevinH

Hi,

Just as a test, I created a chapter0.xhtml that includes links to all of the other chapters and links to ALL of the images.

When I import that file into Sigil, it recognizes that name clash and properly creates image001.gif through image009.gif and no collision happens. Later importing of the other html files still ends up with broken links.

I think there is a python script that can be run to merge multiple html files ...

Perhaps, importing an opf file with links to everything might work in Sigil. Sorry I can't be more help here.

KevinH

i tried htmlcat, but iirc it did not like my xhtml or something.

adding an opt to sigil just puts in the Misc folder.

your posts are helpful to me.

thanks

[Hitch]

JSWolf:

Maybe you can explain it better than I. I keep running into a communication issue; one second we're talking mobi, the next Calibre, then ePUB, html & xhtml...I don't know where this started and how it got here, but my explanation (here and at Amazon) about the paths seems to be problematic.

Thx,

Hitch

[JSWolf]

Originally Posted by Hitch

JSWolf:

Maybe you can explain it better than I. I keep running into a communication issue; one second we're talking mobi, the next Calibre, then ePUB, html & xhtml...I don't know where this started and how it got here, but my explanation (here and at Amazon) about the paths seems to be problematic.

Thx,

Hitch

let's just ignore anything that's not the ePub or XML. The HTML directory is fine. Everything seems to work as is. The ePub is missing two directories that contain the graphics. That's it in a nutshell.

I'll take the HTML/XML and toss it into Sigil and see what I can get. I have to update Sigil first.

[JSWolf]

Originally Posted by Valloric

Attach the HTML files as a comment to that issue, with an explanation of where the problem occurs.

I did just fine one problem. I have a ToC file. That links to separate XML files. All of these are in the same directory ad the links are fine. Now, I load the ToC file and Sigil does not load the rest of the files that go with the ToC.

[Hitch]

I admit, it's confusing me. I do this all the time--add xhtml files that I have here or there, or add a section that I've made previously, from an exploded epub, and even when I've had duplicate images (or stylesheets, for that matter), I've never had a collision issue like this; Sigil merely renames the files and keeps truckin'. My bigger problem is having the same image multiple times with different names and having to go back and delete the dupes and fix the paths so I don't have a boatload of excess unneeded baggage.

Actually, from a testing standpoint: Ray, if you rename the images folders "Images," (and obviously s&r the linking text inside the xhtml files), what happens if you upload them then? Do you still get collision or overwriting?

Ray: didn't you say, somewhere in this mess, either here or on DTP, that you had this in mobi? Did you make it with MBP C? So you should have one big (honkin' long) html file, with all the image calls, right? What about simply taking the html from inside MBP and whomping it into Sigil--why wouldn't that work? Or, if it didn't, why didn't it, or what was the symptom of what went wrong?

Hitch

[Valloric]

I've been following this thread and I do believe it's a corner-case bug, but I currently don't have the time to fix/troubleshoot this or other Sigil issues. I'm up to my neck with university work.

I'll take a closer look at it when things settle down on my end.

[Hitch]

Ray:

Over at DTP, you said:

Originally Posted by Ray Tayek

mobi creator and calibre don't seem to have the problem that sigil does.

we did start with word 2007 docs and docx's. these were all saved as filtered web pages.

then i use mobi to make an opf.

using dreamweaver and other tools, i have removed all of the style, class, name and id stuff from the html except for the authors headings and links.

i used dreamweaver to convert the html files into xhtml 1.1. these all validate at w3c validator.

if i make a prc with mobi, it has all of the diagrams.

if i add a toc to calibre and conver to epub, it has all of the diagrams.

if i put the epub from calibre through epubcheck, all the errors are ones that were introduced by calibre.

so that's why i though that sigil might be broken.

thanks

Okay...something's pinging around on the inside of my brain that I missed, here: if you take the html from inside of MobiPocket, and slap that bad boy into Sigil...what happens? I kind of see what you did; I used to flail around at these different things when I started out, and I spent a lot of time doing it.

IF you actually stuck all that filtered html into MobiPocket, I'd start there. I realize you have an html TOC issue that you're trying to address, but you could shortcut it a bit by making the TOC in MBP Creator, and then using all that code it shoves out, in html, the toc.ncx and the OPF in Sigil--but most importantly, for THIS five minutes, you should have one big long html file, with non-repetitive image pointers, to shove into Sigil. (n.b. - you could even "cheat" and just copy & paste the html toc code right out of MBP Creator or preview it and steal the source and then make a new "page" in Sigil and the links should all be right, I think...this is a bit backwards than how I work, which is ePUB first, mobi second.)

(Me--I'd edit the html first, and place the Sigil Chapter Markers before every Chapter heading, so that you don't have to do THAT one at a time and you can split that sucker ASAP. But I'm a serious NoteTab Pro Junkie, so...that might just be me, but I'm a total regex Ho.)

Jam that 50 chapters and 500 images into Sigil (go get a cup of coffee). Then save as ePUB; split; save again; and see if THAT dog hunts for you. I think, but am not sure, that might work--but it should either work or not fast enough to be worth trying.

I'm actually interested in this. Can you keep us posted?

Hitch

[Ray Tayek]

> Okay...something's pinging around on the inside of my brain that I
> missed, here: if you take the html from inside of MobiPocket, and slap > > that bad boy into Sigil...what happens? ...

not sure what you mean by "take the html from inside of MobiPocket". i don't know how to look into a prc file.

> IF you actually stuck all that filtered html into MobiPocket, I'd start
> there.

i added the 60 or so html files to mobi creator.

> I realize you have an html TOC issue that you're trying to address, but
> you could shortcut it a bit by making the TOC in MBP Creator,

that is how i currently get a toc.

> and then using all that code it shoves out, in html, the toc.ncx and the OPF in Sigil--but most importantly, for THIS five minutes, you should have one big long html file, with non-repetitive image pointers, to shove into Sigil. (n.b. - you could even "cheat" and just copy & paste the html toc code right out of MBP Creator or preview it and steal the source and then make a new "page" in Sigil and the links should all be right, I think...this is a bit backwards than how I work, which is ePUB first, mobi second.)

i get the impression that you know how to look inside a prc and get the html inside there and put that into sigil? if you know how to look inside a prc, please tell me.

>(Me--I'd edit the html first, and place the Sigil Chapter Markers before
> every Chapter heading, so that you don't have to do THAT one at a time > and you can split that sucker ASAP.

each chapter starts with a heading 1. i have not seen any need for chapter markers (yet).

But I'm a serious NoteTab Pro Junkie, so...that might just be me, but I'm a total regex Ho.)

grep is my friend. i have a trial copy of dreamweaver that is turning out to be quite useful also.

> Jam that 50 chapters and 500 images into Sigil (go get a cup of coffee). > Then save as ePUB; split; save again; and see if THAT dog hunts for
> you. I think, but am not sure, that might work--but it should either > work or not fast enough to be worth trying.

i was using sigil to get a free ncx on the previous book (which was all one html file with 700 diagrams) the using kindlegen (i think).

this time i tried to do the same thing, but ran into issue 501.

> I'm actually interested in this. Can you keep us posted?

sure. we have split the current book into two volumnes, and are close to finishing it. the mobi stuff looks good. so i will be trying the same trick to get an ncx out of sigil and running kindlegen to put it on the kindle in the next few days.

we also want to put it on the ipad, so i am trying out calibre to get an epub, so i can run epubcjeck. i have flightcrew, but haven't tried it yet.

i did get started on a parser, but that is going to be a bit harder than i thought.

i'd love to look at the stuff inside the prc as i don't like all of the stuff that calibre does. i a hoping that calibre has some options to turn that stuff off.

thanks

thanks

[Hitch]

Okay, I'm just going to quick-reply this: if you have a mobi file, you can put it into Calibre (yeah, yeah, I know, I am not crazy about that either) and try to convert it to something else (anything) and set the debug option, which should deposit one big long honking html file in your Calibre Debug folder. It ain't pretty, and I wouldn't suggest it to most folks, but if you GREP, you could fix that pretty quick. Which would give you ONE file to slam into Sigil, methinks. Try that, and see if it works. You should be able to regex all the flattened css back into "real" css in your ss's.

Also, gotta tell ya--have never seen a Calibre ePUB ePUBcheck, ever. I like Kovid; I like his work, and Calibre is, in its own way, Mighty; but they simply don't pass, long, short or otherwise.

HTH,

Hitch

[Ray Tayek]

> if you have a mobi file, you can put it into Calibre ...
> try to convert it to something else (anything)
> set the debug option, which should deposit one big long honking html

oh my! - looks like there are 4 copies of the html in variois stages in various folders with tocs and ncxs galore.

> Which would give you ONE file to slam into Sigil, methinks. ...
> You should be able to regex all the flattened css back into "real" css ...

when trying any of the big html files, epubcheck gets an an unfinished element in the opf. flightcrew complains about missing language and title.

> Also, gotta tell ya--have never seen a Calibre ePUB ePUBcheck, ever. ...

if i put the epub made by calibre (from the prc made by mobi), i only get a date warning and the warnings from the ul's that are improperly nested.

thanks

[Ray Tayek]

> if i put the epub made by calibre (from the prc made by mobi),
> i only get a date warning and the warnings from the ul's

i changed all of my files from xhtml 1.1 to html 4.01 strict.

putting the prc into calibre and converting to epub gets the same date warning and ul warnings. flightcrew complains about the ul's.

extracting the single html file from the calibre epub and putting it into sigil and making an epub and running epubcheck has a single warning:

ERROR: sigil2.1.epub/OEBPS/content.opf(6): unfinished element

Check finished with warnings or errors!

the content.opf looks like:

<?xml version='1.0' encoding='utf-8'?>
<package xmlns="http://www.idpf.org/2007/opf" version="2.0" unique-identifier="uuid_id">
<metadata xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:opf="http://www.idpf.org/2007/opf" xmlns:dcterms="http://purl.org/dc/terms/" xmlns:calibre="http://calibre.kovidgoyal.net/2009/metadata" xmlns:dc="http://purl.org/dc/elements/1.1/">
<dc:publisher>Robert J. Terry</dc:publisher>
<meta name="calibre:series_index" content="1"/>
<dc:language>en</dc:language>

so it looks like it's complaining about a missing </meta>. maybe it really needs the space before the /.

flightcrew complains about: The <language> element is missing. and: The <title> element is missing.

if i fool arounf with the unzipped epubs, is there a special way i need to zip them back up?

thanks