epubcheck plugin for Sigil

[philja]

Using Sigil 2.3.1 with epubcheck 0.4.7.1

I converted a 'no fault found' epub2 to epub3 using the plugin epub3-itizer v0.5.8 and then ran epubcheck again and was horrified to see vast numbers of pink errors, 25 of which concerned images :

OEBPS/Text/chapter10.xhtml,18,1118,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter10.xhtml,30,2338,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter10.xhtml,68,3887,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter10.xhtml,86,4610,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter10.xhtml,118,6146,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter10.xhtml,132,7188,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter10.xhtml,140,7419,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter11.xhtml,16,705,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter4.xhtml,77,3766,Col: 89: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter4.xhtml,118,8009,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter5.xhtml,17,839,Col: 89: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter5.xhtml,22,1098,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter5.xhtml,32,2127,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter6.xhtml,20,1345,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter6.xhtml,36,2282,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter6.xhtml,46,2917,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter7.xhtml,60,3084,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter7.xhtml,82,4029,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter8.xhtml,20,1171,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter8.xhtml,34,2041,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter8.xhtml,58,3444,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter8.xhtml,68,4162,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter8.xhtml,90,4786,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter8.xhtml,102,5810,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer
OEBPS/Text/chapter9.xhtml,24,2111,Col: 68: ERROR(RSC-005): Error while parsing file: value of attribute ""width"" is invalid; must be an integer

All concerned epub2 code like this :

<img alt="Image 7.1" src="../Images/image017.gif" width="50%"/>

My immediate concern was about the other 50 images in the book. The rest were all in chapters 11 - 21. They all had the same type of code so why weren't they flagged up?

Subsequent tests showed that epubcheck only reports 25 of these errors at a time. Once I had corrected these 25, I ran the epubcheck plugin again and got another 25.

I imagine this is a function of the EPUBCheck v5.1.0 rather than the plugin. Could the plugin capture and display all instances of an error on the first run through?

[Turtle91]

In the meantime, you could just do a regex search and replace to get all of the same errors:

Find: <img (.*?) width="(.*?)%"/>
Replace: <img \1 width="\2"/>


That should fix them all. The error comes from the % sign is no longer a thing for inline image styling.

[DNSB]

As far as I know, epubcheck will only return 25 errors of a type. For me, that is more than enough since I tend to hit regex to correct the error and that will clean up all the instances of that error.

[KevinH]

But I doubt you want to convert width="50%" to width="50" as 50 pixels is much different from 50% of the screen.

Much more likely, is to remove the width attribute and replace it with either a new "class" attribute or even an inline style attribute such as style="width:50%;" or something else that sets the limit for the width of the resulting image.

[philja]

Originally Posted by KevinH

But I doubt you want to convert width="50%" to width="50" as 50 pixels is much different from 50% of the screen.

Much more likely, is to remove the width attribute and replace it with either a new "class" attribute or even an inline style attribute such as style="width:50%;" or something else that sets the limit for the width of the resulting image.

Which is essentially what I did. I hacked together a regex FIND:

<img\s+(alt="Image .+?")\s+(src="../Images/image\d+\.gif")\s+width="(\d+)%"\s*/>

[ this also worked <img\s+(alt="[^"]*")\s+(src="[^"]*")\s+width="(\d+)%"\s*/> ]

with a REPLACE:

<img class="w\3" \2 \1/>

[This put the alt entry at the end keeping it out of the way visually when checking images later.]

Since many of the images had width=50%, I started with an additional CSS class in the style sheet :

.w50 {height: auto;width: 50%}

and then did a find/replace one at a time. This was a little tedious but pretty quick. When I came to an image with a different width %, if I then decided it would be better to keep it at that width, I just created an additional CSS style entry.

Afterwards, I was curious to find out why all the image entries were not displayed as errors by the epubcheck plugin. So I redid the epub3 conversion and investigated. I also had a look at the log and saw that it had a line:

 "message" : "Error while parsing file: value of attribute \"width\" is invalid; must be an integer",
    "additionalLocations" : 50,

I think that if some indication of the total number of errors would be displayed by the epubcheck plugin, it would be helpful to those authors who don't know how to access the log.

[Doitsu]

Originally Posted by philja

Subsequent tests showed that epubcheck only reports 25 of these errors at a time. Once I had corrected these 25, I ran the epubcheck plugin again and got another 25. [...] Could the plugin capture and display all instances of an error on the first run through?

The plugin runs EPUBCheck in JSON output mode, and in that mode only the first 25 occurrences are shown. I.e., there's nothing I can do about it.

[KevinH]

And since the user typically reruns the plugin repeatedly until no more errors are reported, that works just fine. As others have pointed out, you do not need to have an error "reported" to "fix" things once an ealier error shows you a repeated general error class. That is one of the primary uses for regular expression find and replace.

[JSWolf]

Originally Posted by philja

Which is essentially what I did. I hacked together a regex FIND:

<img\s+(alt="Image .+?")\s+(src="../Images/image\d+\.gif")\s+width="(\d+)%"\s*/>

[ this also worked <img\s+(alt="[^"]*")\s+(src="[^"]*")\s+width="(\d+)%"\s*/> ]

with a REPLACE:

<img class="w\3" \2 \1/>

[This put the alt entry at the end keeping it out of the way visually when checking images later.]

Since many of the images had width=50%, I started with an additional CSS class in the style sheet :

.w50 {height: auto;width: 50%}

and then did a find/replace one at a time. This was a little tedious but pretty quick. When I came to an image with a different width %, if I then decided it would be better to keep it at that width, I just created an additional CSS style entry.

Afterwards, I was curious to find out why all the image entries were not displayed as errors by the epubcheck plugin. So I redid the epub3 conversion and investigated. I also had a look at the log and saw that it had a line:

 "message" : "Error while parsing file: value of attribute \"width\" is invalid; must be an integer",
    "additionalLocations" : 50,

I think that if some indication of the total number of errors would be displayed by the epubcheck plugin, it would be helpful to those authors who don't know how to access the log.

You would solve the IMG problem by removing the width and replacing it with a class. Then all you would need to do is fix that one classes as needed. Plus, converting to ePub3 would work.

[philja]

Originally Posted by JSWolf

You would solve the IMG problem by removing the width and replacing it with a class. Then all you would need to do is fix that one classes as needed. Plus, converting to ePub3 would work.

Hey Jon - that's what I did and what is shown in my earlier post that you quote in yours. Maybe I phrased it badly? I rationalised some widths and ended up with 4 additions to the classes in the styles.css sheet: .w50, .w70, .w80, .w95.

Had I been a bit more competent, I'd have found a way to automate the creation of each new class. As it was, I wrote the first one, (.w50) and copy / pasted 3 copies which just needed the number changing in the name and width.

[Koen_new]

After a complete reinstall on my PC, I now get this error when I try to run the epubcheck plugin:

Status: failed

Running EPUBCheck 5.2.1... please wait.
EPUBCheck Java error.
java.lang.StackOverflowError
at com.thaiopensource.relaxng.pattern.DuplicateAttrib uteDetector.startChoice(DuplicateAttributeDetector .java:42) etc...

sigil version: 2.4.2
java version: 1.8.0.441

other plugins work OK, so I don't understand what I', doing wrong.

Any suggestions?

[KevinH]

You are probably using a 32 bit version of Java instead of a 64 bit version of java, resulting in very small stacks which overflowed.

Try removing that 32 bit version of java and install a 64 bit version of java.

Alternatively there are extra options to tell the 32 bit version of java to use larger stacks, but everything is 64 bits now so upgrading to 64 bit should be the better solution.

[Doitsu]

Originally Posted by Koen_new

After a complete reinstall on my PC, I now get this error when I try to run the epubcheck plugin:

As KevinH has already pointed out, this error is usually displayed if a 32bit Java version is installed.

You either change the following setting:

  "is32bit": false

to true:

  "is32bit": true

(for more information see the first post in this post)

or you could simply uninstall the current Java version and install the latest 64bit Java version.

[Koen_new]

Originally Posted by KevinH

You are probably using a 32 bit version of Java instead of a 64 bit version of java, resulting in very small stacks which overflowed.

Try removing that 32 bit version of java and install a 64 bit version of java.

Alternatively there are extra options to tell the 32 bit version of java to use larger stacks, but everything is 64 bits now so upgrading to 64 bit should be the better solution.

changing to the 64 bit version of java worked, thanks!

[DNSB]

I was noticing that epubcheck in Sigil does not offer me the update to 5.2.1 which fixes the regression that was flagging mailto: links as errors. I manually installed it back in January and didn't think about it until now when I did a new Sigil install on a virtual machine and got a mailto: error on the first epub I checked. calibre's editor did suggest the update.

Update check is true in the prefs file and last_time_checked shows as about 5 minutes ago when I last ran epubcheck.

[Doitsu]

Originally Posted by DNSB

I was noticing that epubcheck in Sigil does not offer me the update to 5.2.1

I wasn't able to reproduce this problem.

I just replaced the EPUBCheck files in the plugin folder with the EPUBCheck v5.1.0 files and changed the year of last_time_checked to:

  "last_time_checked": "2024-04-19 14:28:46.465902",

and when I ran the plugin, it downloaded and installed 5.2.1.

BTW, you can always force an upgrade by deleting the lib folder in the plugin folder.