incorporate the capitains sparrow selector into the input form

[balmas]

the goal is to add the capitains sparrow cts selector and workflow plugins to the alignment editor input form so that users can select from texts available in various CTS repositories as the contents of their alignment.

We can use the work done on the treebank editor input form as a guide.

The starting point is the current input form (https://github.com/alpheios-project/alignment-editor/blob/master/db/app/align-entersentence-perseids.xhtml and https://github.com/alpheios-project/alignment-editor/blob/master/db/script/alph-align-enter-ext.js)

Here are the steps I think are necessary:

1. add 2 selectors to the form - one for L1 and one for L2
2. EnterSentence javascript function currently calls an xquery service, as defined in this metatag <meta name="templateurl" content="../xq/align-maketemplate.xq"/>. We want to replace the use of this xquery service with the client side calls to the LLT segtok service to segment/tokenize the text and then send it to a client side transform using XSLT to create the alignment template. This will be similar to the segtok_to_tb.xsl but create an alignment xml file instead of a treebank one.
3. this output is then what gets submitted to Perseids to create the document etc.

Hopefully this is enough of a guide to get this started, but please please ask if anything isn't clear.

[balmas]

And @PonteIneptique will help...

[PonteIneptique]

God, this name is so nice when you receive notifications... (But on a more serious note, I'll help)

[PonteIneptique]

Adding documentation for the CTS.XSLT service is my next priority then

[balmas]

I don't know if looking at this will make the task easier or harder, but here is what I currently use when creating an alignment template within Perseids from the CTS references created in the annotation editor. It's a bit convoluted and I'm at a bit of a loss right now to say why I did it this way...

This transform is applied to the output of the tokenizer:
https://github.com/sosol/sosol/blob/perseus_shibboleth/data/xslt/cite/tokens_to_align.xsl

And then I construct the sentence xml in Ruby for some godforsaken reason:
https://github.com/sosol/sosol/blob/perseus_shibboleth/app/models/alignment_cite_identifier.rb#L104

[PonteIneptique]

 $(".targetedElement1, .targetedElement2").on("cts-passage:retrieved", function() {
  var instance = $(this).data("_cts_typeahead");
  var xml = instance.text.getXml(); // The XML when retrieved
  var textarea = $(this.settings.retrieve);

  //Then do stuff to get only the string
  })

[balmas]

ok, I know why I did it that way, because we need to merge the results of two separate tokenizations and transforms, one for each language/text, and dealing with that is a bit complicated. This is also why I only deal with single sentences right now, because it's not really worth doing more until we have an interface that allows the user to change which sentence(s) get matched with which for the word-by-word alignment, we will be wrong most of the time.

But we can at least improve the current functionality through use of the selector to allow the user to specific small chunks of text to align from the available CTS texts, as an alternative to the sentence-by-sentence selection option we have through the annotation editor.

I've created segtok_to_align.xsl

This will take the output of the segtok service for a single text and create a complete alignment template (wrapped in an OA wrapper), but with only one language in it. the crucial input parameters are:

e_lang
e_dir
e_lnum
e_docuri

an option e_includepunc can be set to false if you want to ignore punctuation (default is true)

The retrieved text for each language can be passed to this transform.

And then, probably via javascript, we just want to merge the two results, taking the language and wds element from l2 transform and plugging it into the appropriate spot in the first, much along the lines of the ruby code linked to above.

This is then what would get posted to perseids.