[jsword-devel] Verse markup in OSIS modules

[DM Smith]

I see that there are two basic ways that the fix could go. Currently, 
PassageAbstractData constructs an OSIS document containing a title and a 
list of verses and it attaches to the verses the results of each call to 
getFilter().toOSIS(verseElement, rawdata);
(Actually, it is inside the toOSIS that the transformed data is attached 
to verseElement).

I see two different ways to solve it and I would like feedback as to 
which is better.
1) Modify the OSISFilter to strip out the verse if it is present. This 
will require iterating over the list of elements to see if any are verse 
elements and removing it and saving its children in its place.

2) Modify PassageAbstractBook to not create the verse element if it is 
present in the results of toOSIS. This would entail creating a dummy 
element and then analyzing the descendants of that element to see if any 
have an osisID containing the expected verse reference. If it is not 
found then that dummy element is replaced/transformed into a verse 
element and then appended to the div element. If it is found then all 
the children of the dummy element are appended to the div element.

The reason to look for an osisID is that <verse> is not the only way to 
specify a verse according to OSIS 2.0. The various filters should not 
care whether the data came from a Sword module or from some other system.

I noticed that in bugs.txt that having a pass-through for OSIS is listed 
as a post 1.0 activity. I'm not sure what this means exactly but it 
seems that it allows for any valid OSIS to be returned as is. This would 
be good for books that have verse markers between verses, that have 
non-verse text before verses, that have multiple verses for a single 
verse, that split verses into parts and have multiple markers per verse, 
.... All of which are allowed in OSIS.

Anyway, I am inclined to do #2 as I think that the call to toOSIS should 
not be lossy. This will allow for a pass-through mode. And I think that 
it will allow for multiple verses to be gotten at the same time from a 
fully marked up OSIS module.

I did some more reading of the sword-devel archives and it seems that 
there is a bit of disagreement on whether the stored text should contain 
verse markup.

If WLS is any indication on the final disposition of that disagreement, 
then the OSIS modules should be fully marked up.



DM Smith wrote:
> So as not to bury the lead, I am putting my question at the beginning of 
> the e-mail: How should I determine whether an OSIS module has verse 
> markup or not? I am cross posting this (please let me know if this is 
> not appropriate)
> 
> I have read some posts on sword-devel for information on verse markup 
> and I noted some discussion on how verses should be marked.
> 
> I am working on J-Sword (the java port of the Sword Engine) and I loaded 
> WLS, the BHS replacement, and J-Sword started to display the verse 
> numbers twice.
> 
> I looked into the module and the verses are all marked up in the 
> following fashion (This is English markup around Hebrew text and the 
> Hebrew may not survive the posting):
> <verse osisID="Gen.1.1">בְּרֵאשִׁ֖ית בָּרָ֣א אֱלֹהִ֑ים אֵ֥ת הַשָּׁמַ֖יִם וְאֵ֥ת הָאָֽרֶץ׃</verse>
> 
> In KJV w/ Strongs it is not surrounded with <verse> markup:
> <w lemma="x-Strongs:H07225">In the beginning</w> <w 
> lemma="x-Strongs:H0430">God</w> <w morph="x-StrongsMorph:TH8804" 
> lemma="x-Strongs:H0853|x-Strongs:H01254">created</w> <w 
> lemma="x-Strongs:H08064">the heaven</w> <w 
> lemma="x-Strongs:H0853">and</w> <w lemma="x-Strongs:H0776">the earth</w>.
> 
> At first I thought that the indexing of KJV was a bit lossy and did not 
> include the <verse> markup in the indexing, but I decompressed the 
> entirety of Genesis and found that there were no verse markers.
> 
> Anyway JSword assumes that the module does not contain verse markers and 
> adds them. When it comes to WLS, it is in there twice.
> 
> Chris Little wrote in a thread on sword-devel:
>  > Just check the OSISVersion entry in the .conf.  The new modules are
>  > tagged with OSISVersion=2.0.1.  Other OSIS modules won't have this tag
>  > at all and most of them would be tagged as OSISVersion=1.1.1, if we
>  > went back and added the metadata to them.  So you can assume that if
>  > OSISVersion >= 2 then ALL tags from the OSIS document are supposed to
>  > be in the Sword module, including <verse> tags.
> 
> I don't see that attribute in the conf for WLS. I saw that Sword handles 
>  WLS and KJV correctly.
> 
> With J-Sword, we build OSIS and render it with XSLT into HTML. I guess 
> we could parse the text to see if it contains verse markup and not add 
> it if it does. Any suggestions?
> 
> _______________________________________________
> jsword-devel mailing list
> jsword-devel at crosswire.org
> http://www.crosswire.org/mailman/listinfo/jsword-devel
>