The lread-ns-kw-map
branch lgtm (for the superficial judgement I can make)
I noticed that the following forms don't have test coverage (?):
^::foo
^::foo/bar
^{::foo ...}
^{::foo/bar ...}
...although they appear to just work :)
---
I constructed those with base/of-string
but I'm not entirely sure of how one would build them with rewrite-cljc.node.meta
which I think would be a more frequent use case for me.
e.g., let's say I want to replace x
with ^::foo x
, how would that be done? The doc might be scarce yet on that front
(note that ^::foo x
is textually different from ^{::foo true} x
, even if functionally equivalent. One should be able to specify one or the other resulting format)Very interesting @vemv thanks for sharing! This will definitely result in improvements, at the very least to the docs! Let’s start with: > let’s say I want to replace `x` with `^::foo x`, how would that be done? The doc might be scarce yet on that front Here is one way:
(require '[rewrite-cljc.zip :as z]
'[rewrite-cljc.node :as n])
(let [zloc (z/of-string "x")]
(-> zloc
(z/replace (n/meta-node [(n/keyword-node :foo true)
(n/whitespace-node " ")
(z/node zloc)]))
z/root-string))
;; => "^::foo x"
As for: > (note that `^::foo x` is textually different from `^{::foo true} x`, even if functionally equivalent. One should be able to specify one or the other resulting format) Here’s one way to create the nodes from the node API:
(require '[rewrite-cljc.node :as n])
(def n1 (n/meta-node (n/keyword-node :foo true) (n/token-node 'x)))
(n/string n1)
;; => "^::foo x"
(def n2 (n/meta-node (n/map-node [(n/keyword-node :foo true)
(n/whitespace-node " ")
(n/token-node 'true)])
(n/token-node 'x)))
(n/string n2)
;; => "^{::foo true} x"
(meta (n/sexpr n1))
;; => #:user{:foo true}
(meta (n/sexpr n2))
;; => #:user{:foo true}
(= (meta (n/sexpr n1)) (meta (n/sexpr n2)))
;; => true
Fantastic, thank you! Tried out. I'd be happy if these were added to the test suite as I'm writing a program that would depend on this feature
Great stuff! Yeah, I think that adding some tests makes good sense. The above examples, or something similar, would be good to add to the introduction document. The node creation fn docstrings could use some love/examples too. I certainly found them mysterious when first looking at them.
> The node creation fn docstrings could use some love/examples too. I certainly found them mysterious when first looking at them.
yeah this was my impression when using xsc/rewrite-clj
- I wasn't all sure if rewrite-clj.node.meta
was a ns supposed to be used.
especially when there was a ^:no-doc :)
That maybe more import-vars confusion… the node creation functions are exposed via the rewrite-clj.node
namespace.
I was thinking of maybe adjusting the tests to hit the public APIs rather than internal namespaces. That might help folks who refer to tests for usage, to hit the public APIs.