Re: Force text on line without using \textLengthOn

[Urs Liska]

Hi Harm,


Am 25.10.2018 um 23:38 schrieb Thomas Morley:
> Hi Urs,
>
> Am Do., 25. Okt. 2018 um 17:49 Uhr schrieb Urs Liska <address@hidden>:
> > ...
> >
> >
> > > Hi Urs,
> >
> > > from your description it sounds you need sort of a spanner with
> > > adjustable length.
> > > Attached some coding hacking TextSpanner.
> > > What do you think?
> >
> >
> > I didn't understand the code in detail, but am I right that
> >
> > the code basically prints the texts as the start of a TextSpanner
> Yes
>
> > \endTextSpanners parses a music expression and adds a \stopTextSpan 
> > automatically for any \startTextSpan
> Not exactly, the first event with \startTextSpan is disregarded, all
> other events with \startTextSpan get a \stopTextSpan added. The last
> note/rest/skip/eventchord gets a \stopTextSpan as well.

OK. I don't really understand the handling of first and last events 
here, but I think I don't need this approach anyway.

> > \testII creates such a \startTextSpan to be processed by  \endTextSpanners
> Yes
>
> > ?
> >
> > If so this means that all the music must be wrapped in an \endTextSpanners 
> > expression, isn't it?
> Not exactly.
> The proposed code/example does so, yes.
> Though, \endTextSpanners is more for conveniance., you could drop
> \endTextSpanners entirely and set \stopTextSpan manually.

I had a try that *seemed* to be successful, see below

> > If this is true then I would be quite a stretch to integrate this solution into 
> > my environment, because the input files are part of the edition in this case. 
> > They will be made available within a TEI file and therefore have to be 
> > extremely clean. So far there is exactly one \include and then either a single 
> > music expression (for a single-voice music example) or several music variables 
> > plus a custom \loadTemplate command that loads a file with a \score block to 
> > make use of the music variables.
> >
> > For the first case (simple music expression) I can achieve the result by calling 
> > \endTextSpanners at the very end of the include file so the music expression will be 
> > used as the argument for the function. But for the other cases I'm not sure how I could 
> > arrange my files. In addition I only need this functionality in a subset of my >560 
> > files, but such a solution could probably not be optional in the way I'm doing it 
> > elsewhere in the project. (The main include file loads a <filename>-include.ily 
> > file if that is present on disk. Within that optional file I can place a layout block, 
> > edition-engraver mods and an arbitrary number of \loadTool commands. While this latter 
> > function probably isn't necessary for efficiency reasons (it would probably be easier to 
> > simply include all the available code for each compiled file), it is a nice way to 
> > document the specific needs of an example.)
>
> Well, I'm not familiar with TEI, the edition-engraver, etc.
> Sorry.

That's not necessary here. All I wanted to point out is that I need 
"clean" input files and a generic structure where I can't wrap the music 
variables in an arbitrary function call. The reference to the TEI is 
only that the primary input file's content will be included in the 
edition and should be stripped of any code that is only needed for 
rendering purposes. Eventually the project will be publicly available, 
and I'll probably write about it so it'll become clearer what this is 
about ...

> > I think I'll do some more experiments with the code itself before trying to 
> > integrate this.
> >
> > Hm, as said I don't fully understand the code. But if I'm not mistaken 
> > \endTextSpanners works by first identifying a music expression that holds a 
> > \startTextSpan event and then adding a \stopTextSpan event to the note, rest or 
> > chord. Would it be possible to achieve the same thing directly from a 
> > music-function or event-function? Maybe if \testII wouldn't be an event 
> > function but a music-function taking the note/rest/chord as an argument, like
> >
> >    \testII ##t "John Doe" c'
> >
> > where the TextSpanner is added to the c' ?
> Well, you can't start _and_ end a TextSpanner on the same event. See
> the warning for { c'1\startTextSpan\stopTextSpan }.
> Thus the sophisticated endTextSpanner. Ofcourse endTextSpanner can't
> end a textSpanner which is started at the last note of a file as well.
>
> It's thinkable to create a music-functions taking a musical argument
> and doing something like
> { \myFunction c'1 }
> -> { c'1\startTextSpan <>\stopTextSpan }
> This will return a warning if at happens at the end of a file, though.

I copied all relevant code to a different structure, and while it 
compiles and produces the spanners as desired there are still problems.

%%%
\version "2.19.82"

#(define (minimum-length-from-stencil process-string?)
   ;; Set the minimum-length of a grob, which is supposed to be a TextSpanner,
   ;; to the length of the stencil of bound-details.left.text plus 0.1
   (lambda (grob)
     (let* ((bound-details (ly:grob-property grob 'bound-details))
            (lft (assoc-get 'left bound-details))
            (txt (assoc-get 'text lft))
            (txt-stil
             (if (markup? txt)
                 (grob-interpret-markup
                  grob
                  (if process-string?
                      (markup->string txt)
                      txt))
                 #f))
            (txt-x-ext (if txt-stil (ly:stencil-extent txt-stil X) #f))
            (txt-x-length
             (if (and txt-x-ext (interval-sane? txt-x-ext))
                 (- (cdr txt-x-ext) (car txt-x-ext))
                 0)))
       (+ txt-x-length 0.1))))


startSingleTextSpan =
#(define-event-function (text-length-on txt)((boolean? #f) markup?)
   #{
     -\tweak style #'none
     -\tweak staff-padding 3
     -\tweak springs-and-rods #ly:spanner::set-spacing-rods
     -\tweak minimum-length  #(minimum-length-from-stencil #f)
     -\tweak minimum-length-after-break 0
     -\tweak font-shape #'upright
     -\tweak bound-details.left.text
     #(if text-length-on
          txt
          (make-with-dimensions-markup empty-interval empty-interval txt))
     -\tweak bound-details.left-broken.text ##f
     \startTextSpan
   #})

spanner =
#(define-music-function (text-length-on text music)((boolean?) markup? 
ly:music?)
   (let*
    ((is-single-music #t))
    (define (append-script-at! prop)
      (set! (ly:music-property music prop)
            (append (ly:music-property music prop)
              (list (startSingleTextSpan text-length-on text)))))

    (case (ly:music-property music 'name)
      ((EventChord)
       (append-script-at! 'elements))
      ((NoteEvent RestEvent SkipEvent)
       (append-script-at! 'articulations))
      (else (set! is-single-music #f)))
    
    (if is-single-music
        #{
          #music
          <>\stopTextSpan
        #}
        music)))

my-upbow =
#(define-music-function (text-length-on music)((boolean?) ly:music?)
   (spanner text-length-on "hin." music))

my-downbow =
#(define-music-function (text-length-on music)((boolean?) ly:music?)
   (spanner text-length-on "her." music))

\relative {
  \my-upbow #f
  c'16 ( d e f )
  \my-downbow #f g (
  \my-upbow #f a g a )
  \my-upbow #f \time 2/4
}
%%%

I've reversed the logic of zero-dimensions? to a text-length-on argument 
(#t meaning: do use the space). This doesn't affect the outcome, and the 
wrong behaviour was the same before I changed this.

The problem is that with this text-length-on being true the spanner 
pushes the next *note*, even if that doesn't have any spanner attached 
(first attached image) while with it being false the spanners will 
collide (second attachment). Additionally (but this is already present 
in your code) when using make-with-dimensions-markup (where is this 
defined? I don't find this with git grep???) the spanner will have no 
vertical extent as well, which can cause it to be cut off from the page 
(or collide with the previous system).

I don't see why the horizontal behaviour is any different in my version, 
and I can only speculate that it *does* have something to do with the 
way \endTextSpanners handles the events.

Any ideas?

PS:
I assume it is *not* possible to write a function that can be written as 
post-event that will add the \startTextSpan event to the current note 
*and* adds the <>\stopTextSpan as the next note? I'm asking because I 
would actually like to have this function as a drop-in replacement for 
\upbow and \downbow.

> Cheers,
>    Harm

document.png
Description: PNG image

document.png
Description: PNG image