[sword-devel] TEI filters - improvement work

Peter von Kaehne refdoc at gmx.net
Mon Mar 23 02:43:12 MST 2015

Following feedback from Karl, Troy, Nic and Greg offlist I have done the
following on teixhtml.cpp:

All TEI tags I could find as in use by us are rendered as simple HTML
with their TEI tag as class info added.

Simple HTML stands for - most ended in a span, rendering tags were
transformed into their equivalent in HTML, the reference links all work
(they did before) as HTML links. Both those to Bible modules and those
to internal links within the dictionary. 

I also transferred "type" and "rend" attributes as I found them in TEI
into HTML as a "type"/"rend" attribute.

The CSS is accessible via the TEIXHTML::getRenderHeader() method. Xiphos
already uses this, so you can see what I have in all its g(l)ory(ness)
by looking at the Webster 1913 in Xiphos.

I have in principle tried to achieve with CSS the following:

- Main sections as I perceived them are rendered as a block, rest

- TEI Rendering attributes have been transformed into CSS instructions.
Not completely, because they keep turning up. But some. More can follow
as desirable.

- I used :before and :after CSS content addition to add descriptive
headings to each section. This basically is the semantic info which is
in the TEI tags - stuff like Pronounciation, Usage etc. You may like it
or not. The good thing is that you can override it in your frontend.

I used mostly two Webster1913 entries to model what I wanted - Grace and
Abash. The filter renders now 30% more of the TEI tags than it did
before. I am sure though there are more tags hiding. Webster1913 is a
huge thing - and so is TEI. Also, I found Webster1913 is quite variable
in terms of its way of expressing itself, so there might be some
unevenness in presentation. 

This has been a bug bear of mine for a long time. 


On Sat, 2015-03-14 at 20:57 +0000, Peter von Kaehne wrote:
> I would like to solicit opinions re the current TEI filter set.
> I am heavily writing filters just now for LaTeX so have gained some
> level of insight in how they work. 
> At the current rate the teixhtml filter produces for the vast bulk of
> TEI tags either <i>s or <b>s. With little distinction and no graphical
> enhancement. 
> And yet TEI is very featureful + modules written in TEI could be with
> relatively little effort be quite pretty. 
> A father garish attempt from long ago wrt handling of TEI with CSS is
> here:
> http://www.crosswire.org/~refdoc/private/abash.html
> This is the TEI entry from Webster1913 for "abash", unaltered, slapped
> between a start and end html tag, with some css thrown across it. I do
> not say it is pretty, but it shows what amount of info comes along with
> TEI.
> So, my thought is that the engine might be coaxed to give a little bit
> more info - either by differentiating a bit the xhtml used or by adding
> class attributes to the existing output.
> Any thoughts - particularly from frontend developers? What would be your
> preferred way forward? more and different html? more css class
> information, both? 
> I will put no promise of a timescale onto completion - just that I can
> do this whenever I am able to come round to it. 
> _______________________________________________
> sword-devel mailing list: sword-devel at crosswire.org
> http://www.crosswire.org/mailman/listinfo/sword-devel
> Instructions to unsubscribe/change your settings at above page

More information about the sword-devel mailing list