ludus/janet/validate.janet

662 lines
17 KiB
Plaintext
Raw Normal View History

2024-05-10 19:02:37 +00:00
### A validator for a Ludus AST
(comment
Tracking here, before I start writing this code, the kinds of validation we're hoping to accomplish:
* [ ] ensure called keywords are only called w/ one arg
2024-05-14 00:55:36 +00:00
* [ ] first-level property access with pkg, e.g. `Foo :bar`--bar must be on Foo
2024-05-14 22:56:18 +00:00
- [ ] accept pkg-kws
* [x] validate dict patterns
2024-05-18 21:05:47 +00:00
* [x] compile string-patterns
2024-05-14 22:41:21 +00:00
* [x] `loop` form arity checking
* [x] arity checking of explicit named function calls
2024-05-14 22:52:11 +00:00
* [x] flag tail calls
2024-05-14 22:56:18 +00:00
* [x] no re-bound names
* [x] no unbound names
* [x] no unbound names with `use` forms
* [x] recur in tail position in `loop` forms
* [x] recur not called outside of `loop` forms
* [x] splats come at the end of list, tuple, and dict patterns
Deferred until a later iteration of Ludus:
2024-05-14 00:55:36 +00:00
* [ ] no circular imports DEFERRED
* [ ] correct imports DEFERRED
* [ ] validate `with` forms
2024-05-10 19:02:37 +00:00
)
2024-05-14 00:55:36 +00:00
2024-05-15 00:44:43 +00:00
(try (os/cd "janet") ([_] nil))
2024-05-14 00:55:36 +00:00
(import ./scanner :as s)
2024-05-15 00:44:43 +00:00
(import ./parser :as p)
2024-05-14 00:55:36 +00:00
(defn- new-validator [parser]
(def ast (parser :ast))
@{:ast ast
:errors @[]
:ctx @{}
:status @{}}
)
(var validate nil)
2024-05-14 22:41:21 +00:00
(def terminals [:number :keyword :string :bool :nil :placeholder])
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(def simple-colls [:list :tuple :set :args])
2024-05-14 00:55:36 +00:00
(defn- simple-coll [validator]
(def ast (validator :ast))
(def data (ast :data))
(each node data
(set (validator :ast) node)
(validate validator))
validator)
(defn- iff [validator]
(def ast (validator :ast))
(def data (ast :data))
(each node data
(set (validator :ast) node)
(validate validator))
validator)
(defn- script [validator]
(def ast (validator :ast))
(def data (ast :data))
(def status (validator :status))
(set (status :toplevel) true)
(each node data
(set (validator :ast) node)
(validate validator))
validator)
(defn- block [validator]
(def ast (validator :ast))
(def data (ast :data))
(def status (validator :status))
(set (status :toplevel) nil)
2024-05-14 22:41:21 +00:00
(def tail? (status :tail))
(set (status :tail) false)
2024-05-14 00:55:36 +00:00
(def parent (validator :ctx))
(def ctx @{:^parent parent})
(set (validator :ctx) ctx)
2024-05-14 22:41:21 +00:00
(for i 0 (-> data length dec)
(set (validator :ast) (data i))
2024-05-14 00:55:36 +00:00
(validate validator))
2024-05-14 22:41:21 +00:00
(set (status :tail) tail?)
(set (validator :ast) (last data))
(validate validator)
2024-05-14 00:55:36 +00:00
(set (validator :ctx) parent)
validator)
(defn- resolve-name [ctx name]
(when (nil? ctx) (break nil))
(def node (get ctx name))
(if node node (resolve-name (get ctx :^parent) name)))
(defn- word [validator]
(def ast (validator :ast))
(def name (ast :data))
(def ctx (validator :ctx))
(def resolved (resolve-name ctx name))
(when (not resolved)
(array/push (validator :errors)
{:node ast :msg "unbound name"}))
validator)
### patterns
(var pattern nil)
(defn- lett [validator]
(def ast (validator :ast))
2024-05-14 22:41:21 +00:00
(def [lhs rhs] (ast :data))
2024-05-14 00:55:36 +00:00
# evaluate the expression first
# otherwise lhs names will appear bound
2024-05-14 22:41:21 +00:00
(set (validator :ast) rhs)
2024-05-14 00:55:36 +00:00
(validate validator)
2024-05-14 22:41:21 +00:00
(set (validator :ast) lhs)
2024-05-14 00:55:36 +00:00
(pattern validator)
validator)
(defn- splattern [validator]
(def ast (validator :ast))
(def status (validator :status))
(when (not (status :last))
(array/push (validator :errors)
{:node ast :msg "splats may only come last in collection patterns"}))
(def data (ast :data))
(when data
(set (validator :ast) data)
(pattern validator))
validator)
(defn- simple-coll-pattern [validator]
(def ast (validator :ast))
(def data (ast :data))
(when (empty? data) (break validator))
(def status (validator :status))
(for i 0 (-> data length dec)
(set (validator :ast) (get data i))
(pattern validator))
(set (status :last) true)
(set (validator :ast) (last data))
(pattern validator)
(set (status :last) nil)
validator)
(defn- word-pattern [validator]
(def ast (validator :ast))
(def name (ast :data))
(def ctx (validator :ctx))
(when (has-key? ctx name)
(def {:line line :input input} (get-in ctx [name :token]))
(array/push (validator :errors)
{:node ast :msg (string "name is already bound on line "
line " of " input)}))
(set (ctx name) ast)
2024-05-14 22:41:21 +00:00
(pp ctx)
2024-05-14 00:55:36 +00:00
validator)
(def types [
:nil
2024-05-14 22:41:21 +00:00
:bool
:number
:keyword
:string
:set
:tuple
:dict
:list
:fn
:ref
:pkg
])
2024-05-14 22:41:21 +00:00
(defn typed [validator]
(def ast (validator :ast))
(def [kw-type word] (ast :data))
(def type (kw-type :data))
(when (not (has-value? types type))
(array/push (validator :errors)
{:node kw-type :msg "unknown type"}))
(set (validator :ast) word)
2024-05-15 16:33:34 +00:00
(pattern validator))
2024-05-14 22:41:21 +00:00
(defn- str-pattern [validator]
(def ast (validator :ast))
(def data (ast :data))
2024-05-18 21:43:21 +00:00
(def last-term (-> data array/pop string))
(def grammar @{})
(def bindings @[])
(var current 0)
(each node data
(when (not (buffer? node))
(set (validator :ast) node)
(pattern validator))
(if (buffer? node)
(set (grammar (keyword current)) (string node))
(do
(set (grammar (keyword current))
~(<- (to ,(keyword (inc current)))))
(array/push bindings (node :data))))
(set current (inc current)))
2024-05-18 21:43:21 +00:00
(set (grammar (keyword current)) ~(* ,last-term -1))
(def rules (map keyword (range (length grammar))))
2024-05-18 21:43:21 +00:00
(set (grammar :main) ~(* ,;rules))
(set (ast :grammar) grammar)
(set (ast :compiled) (peg/compile grammar))
(set (ast :bindings) bindings))
(defn- pair [validator]
(def ast (validator :ast))
(def [_ patt] (ast :data))
(set (validator :ast) patt)
(pattern validator))
2024-05-14 00:55:36 +00:00
(defn- pattern* [validator]
2024-05-14 22:41:21 +00:00
(print "PATTERN*")
2024-05-14 00:55:36 +00:00
(def ast (validator :ast))
(def type (ast :type))
(print "validating pattern " type)
(cond
(has-value? terminals type) validator
(case type
:word (word-pattern validator)
:placeholder validator
:ignored validator
:word (word-pattern validator)
:list (simple-coll-pattern validator)
:tuple (simple-coll-pattern validator)
:dict (simple-coll-pattern validator)
2024-05-14 22:41:21 +00:00
:splat (splattern validator)
:typed (typed validator)
:interpolated (str-pattern validator)
:pair (pair validator)
2024-05-14 22:41:21 +00:00
)))
(set pattern pattern*)
# XXX: ensure guard includes only allowable names
# XXX: what to include here? (cf Elixir)
(defn- guard [validator])
2024-05-14 00:55:36 +00:00
(defn- match-clauses [validator clauses]
(print "validating clauses in match-clauses")
2024-05-14 00:55:36 +00:00
(each clause clauses
(def parent (validator :ctx))
(def ctx @{:^parent parent})
(set (validator :ctx) ctx)
(def [lhs guard rhs] clause)
(set (validator :ast) lhs)
(pattern validator)
(pp (validator :ctx))
(pp (validator :ctx))
(when guard
(set (validator :ast) guard)
(validate validator))
(set (validator :ast) rhs)
(validate validator)
(set (validator :ctx) parent)))
(defn- matchh [validator]
(print "validating in matchh")
2024-05-14 00:55:36 +00:00
(def ast (validator :ast))
(def [to-match clauses] (ast :data))
(print "validating expression:")
(pp to-match)
(set (validator :ast) to-match)
(validate validator)
(print "validating clauses")
(match-clauses validator clauses)
2024-05-14 00:55:36 +00:00
validator)
(defn- fnn [validator]
(def ast (validator :ast))
(def name (ast :name))
2024-05-14 22:41:21 +00:00
(print "function name: " name)
(def status (validator :status))
(def tail? (status :tail))
(set (status :tail) true)
2024-05-14 00:55:36 +00:00
(when name
(def ctx (validator :ctx))
(def resolved (resolve-name ctx name))
(when resolved
(def {:line line :input input} (get-in ctx [name :token]))
(array/push (validator :errors)
{:node ast :msg (string "name is already bound on line " line " of " input)}))
(set (ctx name) ast))
(match-clauses validator (ast :data))
2024-05-14 22:41:21 +00:00
(set (status :tail) tail?)
(def clauses (ast :data))
(def rest-arities @{})
(def arities @{:rest rest-arities})
(each clause clauses
(print "CLAUSE:")
(pp clause)
(def patt (first clause))
(def params (patt :data))
(def arity (length params))
(print "checking clause with arity " arity)
(def rest-param? (and (> arity 0) (= :splat ((last params) :type))))
(if rest-param?
(set (rest-arities arity) true)
(set (arities arity) true)))
(pp arities)
(set (ast :arities) arities)
2024-05-14 00:55:36 +00:00
validator)
(defn- ref [validator]
(def ast (validator :ast))
(def ctx (validator :ctx))
(def expr (ast :data))
(set (validator :ast) expr)
(validate validator)
(def name (ast :name))
(def resolved (resolve-name ctx name))
(when resolved
(def {:line line :input input} (get-in ctx [name :token]))
(array/push (validator :errors)
{:node ast :msg (string "name is already bound on line " line " of " input)}))
(set (ctx name) ast)
validator)
(defn- interpolated [validator]
(def ast (validator :ast))
(def data (ast :data))
(each node data
(when (not (buffer? node))
(set (validator :ast) node)
(validate validator))))
### TODO:
# * [ ] ensure properties are on pkgs (if *only* pkgs from root)
2024-05-14 22:41:21 +00:00
(defn- pkg-root [validator])
2024-05-14 22:52:11 +00:00
(defn- tail-call [validator]
(def ast (validator :ast))
(when (ast :partial) (break validator))
(def status (validator :status))
(when (not (status :tail)) (break validator))
(def data (ast :data))
(def args (last data))
(set (args :tail-call) true))
2024-05-14 22:41:21 +00:00
(defn- check-arity [validator]
(def ast (validator :ast))
# (when (ast :partial) (break validator))
2024-05-14 22:41:21 +00:00
(def ctx (validator :ctx))
(def data (ast :data))
(def fn-word (first data))
(def the-fn (resolve-name ctx (fn-word :data)))
(when (not the-fn) (break validator))
(when (= :function (type the-fn)) (break validator))
(when (= :cfunction (type the-fn) (break validator)))
(when (not= :fn (the-fn :type)) (break validator))
2024-05-14 22:41:21 +00:00
(print "fn name: " (the-fn :name))
(def arities (the-fn :arities))
(print "arities: ")
(pp arities)
(def args (get data 1))
(def num-args (length (args :data)))
(print "called with #args " num-args)
(pp (get (validator :ctx) "bar"))
2024-05-14 22:41:21 +00:00
(when (has-key? arities num-args) (break validator))
(def rest-arities (keys (arities :rest)))
(when (empty? rest-arities)
(array/push (validator :errors)
{:node ast :msg "mismatched arity"})
(break validator))
(def rest-min (min ;rest-arities))
(when (< num-args rest-min)
(array/push (validator :errors)
{:node ast :msg "mismatched arity"}))
validator)
(defn- kw-root [validator]
(def ast (validator :ast))
(def data (ast :data))
(def [_ args] data)
(when (not= :args (args :type))
(break (array/push (validator :errors)
{:node args :msg "called keyword expects an argument"})))
(when (not= 1 (length (args :data)))
(array/push (validator :errors)
{:node args :msg "called keywords take one argument"})))
2024-05-14 00:55:36 +00:00
(defn- synthetic [validator]
(def ast (validator :ast))
(def data (ast :data))
(each node data
(set (validator :ast) node)
2024-05-14 22:41:21 +00:00
(validate validator))
(set (validator :ast) ast)
(def ftype ((first data) :type))
(def stype ((get data 1) :type))
(def ltype ((last data) :type))
(print "ftype " ftype)
(print "stype " stype)
(print "ltype " ltype)
(when (= ftype :pkg-name) (pkg-root validator))
(when (= ftype :keyword) (kw-root validator))
# (when (= ltype :args) (tail-call validator))
2024-05-14 22:41:21 +00:00
(when (and (= ftype :word) (= stype :args))
(check-arity validator))
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- pair [validator]
(def ast (validator :ast))
(def [k v] (ast :data))
(set (validator :ast) k)
(validate validator)
(set (validator :ast) v)
(validate validator))
(defn- splat [validator]
(def ast (validator :ast))
(when (get-in validator [:status :pkg])
(array/push (validator :errors)
{:node ast :msg "splats are not allowed in pkgs"})
(break validator))
(def data (ast :data))
(when data
(set (validator :ast) data)
(validate validator))
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- dict [validator]
(def ast (validator :ast))
(def data (ast :data))
(each node data
(set (validator :ast) node)
(validate validator))
validator)
(defn- whenn [validator]
(def ast (validator :ast))
(def data (ast :data))
(each node data
(def [lhs rhs] node)
(set (validator :ast) lhs)
(validate validator)
(set (validator :ast) rhs)
(validate validator))
validator)
2024-05-14 00:55:36 +00:00
# XXX: do this!
2024-05-14 00:55:36 +00:00
(defn- withh [validator])
2024-05-14 22:41:21 +00:00
# XXX: tail calls in last position
(defn- doo [validator]
(def ast (validator :ast))
(def data (ast :data))
(each node data
(set (validator :ast) node)
(validate validator))
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- usee [validator]
(def ast (validator :ast))
(def data (ast :data))
(set (validator :ast) data)
(validate validator)
(def name (data :data))
(def ctx (validator :ctx))
(def pkg (get-in ctx [name :pkg] @{}))
(loop [[k v] :pairs pkg]
(set (ctx (string k)) v))
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- pkg-entry [validator pkg]
(def ast (validator :ast))
(def [key value] (ast :data))
(print "PKG ENTRY***")
(pp key)
(pp value)
(set (validator :ast) key)
(validate validator)
(set (validator :ast) value)
(validate validator)
(def kw (key :data))
(pp kw)
(set (pkg kw) value)
(pp pkg)
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- pkg [validator]
(def ast (validator :ast))
(def data (ast :data))
(def name (ast :name))
(def pkg @{})
(each node data
(set (validator :ast) node)
(pkg-entry validator pkg))
(set (ast :pkg) pkg)
(print "THE PACKAGE")
(pp pkg)
(def ctx (validator :ctx))
(set (ctx name) ast)
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- ns [validator]
(def ast (validator :ast))
(def data (ast :data))
(def name (ast :name))
(def parent (validator :ctx))
(def ctx @{:^parent parent})
(def block (data :data))
(each node block
(set (validator :ast) node)
(validate validator))
(set (ast :pkg) ctx)
(set (parent name) ast)
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- loopp [validator]
(def ast (validator :ast))
(def status (validator :status))
(def data (ast :data))
(def input (first data))
(print "LOOP INPUT")
(pp input)
(def clauses (get data 1))
(def input-arity (length (input :data)))
(set (ast :arity) input-arity)
(print "input arity to loop " input-arity)
(set (validator :ast) input)
(validate validator)
# harmonize arities
(def rest-arities @{})
(each clause clauses
(print "CLAUSE:")
(pp clause)
(def patt (first clause))
(def params (patt :data))
(def clause-arity (length params))
(print "checking clause with arity " clause-arity)
(def rest-param? (= :splat (get (last params) :type)))
(when (and
(not rest-param?) (not= clause-arity input-arity))
(array/push (validator :errors)
{:node patt :msg "arity mismatch"}))
(when rest-param?
(set (rest-arities clause-arity) patt)))
(pp rest-arities)
(loop [[arity patt] :pairs rest-arities]
(when (< input-arity arity)
(array/push (validator :errors)
{:node patt :msg "arity mismatch"})))
(def loop? (status :loop))
(set (status :loop) input-arity)
(def tail? (status :tail))
(set (status :tail) true)
(match-clauses validator clauses)
(set (status :loop) loop?)
(set (status :tail) tail?)
validator)
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- recur [validator]
(def ast (validator :ast))
(def status (validator :status))
(def loop-arity (status :loop))
(when (not loop-arity)
(array/push (validator :errors)
{:node ast :msg "recur may only be used inside a loop"})
(break validator))
(def called-with (get-in ast [:data :data]))
(def recur-arity (length called-with))
(print "loop arity " loop-arity)
(print "recur arity" recur-arity)
(when (not= recur-arity loop-arity)
(array/push (validator :errors)
{:node ast :msg "recur must have the same number of args as its loop"}))
(when (not (status :tail))
(array/push (validator :errors)
{:node ast :msg "recur must be in tail position"}))
(set (validator :ast) (ast :data))
(validate validator))
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- repeatt [validator]
(def ast (validator :ast))
(def [times body] (ast :data))
(set (validator :ast) times)
(validate validator)
(set (validator :ast) body)
(validate validator))
2024-05-14 00:55:36 +00:00
2024-05-14 22:41:21 +00:00
(defn- panic [validator]
(def ast (validator :ast))
(def data (ast :data))
(set (validator :ast) data)
(validate validator))
(defn- testt [validator]
(def ast (validator :ast))
(def [_ body] (ast :data))
(set (validator :ast) body)
(validate validator))
(defn- pkg-name [validator]
(def ast (validator :ast))
(def name (ast :data))
(def ctx (validator :ctx))
(def pkg (resolve-name ctx name))
(when (not pkg)
(array/push (validator :errors)
{:node ast :msg "unbound name"}))
validator)
2024-05-14 00:55:36 +00:00
(defn- validate* [validator]
(def ast (validator :ast))
(def type (ast :type))
(print "validating node " type)
(cond
(has-value? terminals type) true
(has-value? simple-colls type) (simple-coll validator)
(case type
:if (iff validator)
:let (lett validator)
:script (script validator)
:block (block validator)
:word (word validator)
:fn (fnn validator)
:match (matchh validator)
:interpolated (interpolated validator)
:synthetic (synthetic validator)
2024-05-14 22:41:21 +00:00
:do (doo validator)
:dict (dict validator)
:test (testt validator)
:panic (panic validator)
:repeat (repeatt validator)
:when (whenn validator)
:splat (splat validator)
:pair (pair validator)
:ns (ns validator)
:pkg (pkg validator)
:pkg-name (pkg-name validator)
:use (usee validator)
:loop (loopp validator)
:recur (recur validator)
2024-05-19 18:54:37 +00:00
:ref (ref validator)
2024-05-14 00:55:36 +00:00
(error (string "unknown node type " type)))))
(set validate validate*)
(defn valid [ast &opt ctx]
(default ctx @{})
2024-05-15 00:44:43 +00:00
(def validator (new-validator ast))
(def base-ctx @{:^parent ctx})
2024-05-19 22:23:34 +00:00
(set (validator :ctx) base-ctx)
2024-05-15 00:44:43 +00:00
(validate validator))
(import ./base :as b)
# (do
(comment
2024-05-14 00:55:36 +00:00
(def source `
:foo (1, 2)
2024-05-14 00:55:36 +00:00
`)
(def scanned (s/scan source))
(def parsed (p/parse scanned))
(valid parsed b/ctx)
2024-05-18 21:05:14 +00:00
)