|
|
|
|
@ -3,6 +3,7 @@
|
|
|
|
|
racket/hash
|
|
|
|
|
racket/match
|
|
|
|
|
racket/list
|
|
|
|
|
racket/struct
|
|
|
|
|
txexpr
|
|
|
|
|
racket/function
|
|
|
|
|
"quad.rkt"
|
|
|
|
|
@ -65,16 +66,9 @@
|
|
|
|
|
(for/list ([elem (in-list (merge-adjacent-strings elems 'isolate-white))])
|
|
|
|
|
(match elem
|
|
|
|
|
[(? string?)
|
|
|
|
|
;; 190116 caveat: all quads with strings as elements will be atomized.
|
|
|
|
|
;; however, if the starting quad has a struct subtype of quad,
|
|
|
|
|
;; this subtype will be lost.
|
|
|
|
|
;; IOW, all atomized quads are of the base `quad` type.
|
|
|
|
|
;; this is because we can't get access to any subtype constructors here.
|
|
|
|
|
;; corollary: quads that need to keep their types should not have any strings as elements.
|
|
|
|
|
;; also, they will not have any run keys embedded
|
|
|
|
|
;; (but they shouldn't need it because they're not part of text runs)
|
|
|
|
|
;; overall I am persuaded that `atomize` is very texty and needs a name befitting that role.
|
|
|
|
|
(list ((quad-copier x) x next-attrs (list elem)))]
|
|
|
|
|
(define-values (xtype _) (struct-info x))
|
|
|
|
|
(define x-constructor (struct-type-make-constructor xtype))
|
|
|
|
|
(list (apply x-constructor (list* next-attrs (list elem) (cddr (struct->list x)))))]
|
|
|
|
|
[_ (loop elem next-attrs next-key)])))]
|
|
|
|
|
[_ (list x)])))
|
|
|
|
|
|
|
|
|
|
|
