Parses (Kraken, not the parser ;) )
This commit is contained in:
108
fungll.krak
108
fungll.krak
@@ -1,3 +1,10 @@
|
||||
import vec:*
|
||||
import vec_literals:*
|
||||
import map:*
|
||||
import set:*
|
||||
import util:*
|
||||
import str:*
|
||||
|
||||
// nonterminals are negative, terminals are positive
|
||||
obj Grammer (Object) {
|
||||
var nonterminals: vec<vec<int>>
|
||||
@@ -125,6 +132,9 @@ obj BS (Object) {
|
||||
destruct()
|
||||
copy_construct(&other)
|
||||
}
|
||||
fun to_string(): str {
|
||||
return str() + nonterminal + " " + rule_idx + " " + idx_into_rule + " " + left + " " + pivot + " " + right
|
||||
}
|
||||
}
|
||||
fun bs(nonterminal: int, rule_idx: int, idx_into_rule: int, left: int, pivot: int, right: int): BS {
|
||||
var to_ret.construct(nonterminal, rule_idx, idx_into_rule, left, pivot, right): BS
|
||||
@@ -137,13 +147,13 @@ fun bs(nonterminal: int, rule_idx: int, idx_into_rule: int, left: int, pivot: in
|
||||
/*fun fungll(grammer: ref Grammer, input: ref str): <, set<BS>> {*/
|
||||
fun fungll(grammer: ref Grammer, input: ref str): set<BS> {
|
||||
/*loop(grammer, input, decend(grammer, 0 [>start_symbol<], 0), set(), set(), set(), set())*/
|
||||
loop(grammer, input, decend(grammer, 0 /*start_symbol*/, 0), set(), map(), set(), set())
|
||||
loop(grammer, input, decend(grammer, 0 /*start_symbol*/, 0), set(), map<pair<int, int>, vec<Pending>>(), set(), set<BS>())
|
||||
}
|
||||
fun descend(grammer: ref Grammer, symbol: int, l: int): set<Descriptor> {
|
||||
/*grammer.get_nonterminal_rules(symbol).map(fun(rhs): descriptor { <X::=.rhs,l,l> })*/
|
||||
grammer.get_nonterminal_rules(symbol).map(fun(rhs): Descriptor { return descriptor(symbol, rhs, 0, l, l); })
|
||||
return grammer.get_nonterminal_rules(symbol).map(fun(rhs: int): Descriptor { return descriptor(symbol, rhs, 0, l, l); })
|
||||
}
|
||||
fun loop(grammer: ref Grammer, input: ref str, R: set<Descriptor>, U: set<Descriptor>, G, P, Y: set<BS>): set<BS> {
|
||||
fun loop(grammer: ref Grammer, input: ref str, R: set<Descriptor>, U: set<Descriptor>, G: map<pair<int, int>, vec<Pending>>, P: map<pair<int,int>, set<int>>, Y: set<BS>): set<BS> {
|
||||
/*if R.size() == 0 {*/
|
||||
/*return <U,Y>*/
|
||||
/*} else {*/
|
||||
@@ -155,9 +165,15 @@ fun loop(grammer: ref Grammer, input: ref str, R: set<Descriptor>, U: set<Descri
|
||||
/*}*/
|
||||
while R.size() != 0 {
|
||||
var d = R.pop()
|
||||
var <<Rp,Yp>,Gp,Pp> = process(grammer, input, d, G, P)
|
||||
/*var <<Rp,Yp>,Gp,Pp> = process(grammer, input, d, G, P)*/
|
||||
var it = process(grammer, input, d, G, P)
|
||||
var Rp = it.first.first
|
||||
var Yp = it.first.second
|
||||
var Gp = it.second
|
||||
var Pp = it.third
|
||||
|
||||
var U_with_d = U.union(set(d));
|
||||
var nextR = (R.union(Rp) - U_with_d
|
||||
var nextR = R.union(Rp) - U_with_d
|
||||
R = nextR
|
||||
U = U_with_d
|
||||
G += Gp
|
||||
@@ -167,7 +183,7 @@ fun loop(grammer: ref Grammer, input: ref str, R: set<Descriptor>, U: set<Descri
|
||||
return Y
|
||||
}
|
||||
/*fun process(grammer: ref Grammer, input: ref str, <X:=a.b,l,k>, G, P): {*/
|
||||
fun process(grammer: ref Grammer, input: ref str, discript: Descriptor, G, P): {
|
||||
fun process(grammer: ref Grammer, input: ref str, discript: Descriptor, G: map<pair<int, int>, vec<Pending>>, P: map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, Pending>, map<pair<int,int>, set<int>>> {
|
||||
/*if is_e(b) {*/
|
||||
if descript.idx_into_rule == grammer.get_nonterminal_rules(discript.nonterminal)[descript.rule_idx].size {
|
||||
/*return process_e(<X::=a.,l,k>, G, P)*/
|
||||
@@ -178,7 +194,7 @@ fun process(grammer: ref Grammer, input: ref str, discript: Descriptor, G, P): {
|
||||
}
|
||||
}
|
||||
/*fun process_e(<X::=a.,l,k>, G, P) {*/
|
||||
fun process_e(descript: Descriptor, G, P) {
|
||||
fun process_e(descript: Descriptor, G: ref map<pair<int, int>, vec<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, Pending>, map<pair<int,int>, set<int>>> {
|
||||
var nonterminal: int
|
||||
var rule_idx: int
|
||||
var left: int
|
||||
@@ -186,55 +202,87 @@ fun process_e(descript: Descriptor, G, P) {
|
||||
var X = descript.nonterminal
|
||||
var l = descript.left;
|
||||
/*var K = { <g,l'> | <<X,l>,<g,l'>> in G }*/
|
||||
var K = G.get(<X,l>)
|
||||
var <R,Y> = ascend(l,K,k)
|
||||
var K = G.get(make_pair(X,l))
|
||||
/*var <R,Y> = ascend(l,K,k)*/
|
||||
var it = ascend(l,K,k)
|
||||
var R = it.first
|
||||
var Y = it.second
|
||||
/*var Y' = { <X::=.,l,l,l> | is_e(a) }*/
|
||||
if grammer.get_nonterminal_rules(X)[descript.rule_idx].size == 0 {
|
||||
Y.add(bs(X,descript.rule_idx, 0, l, l, l))
|
||||
}
|
||||
/*return <<R,Y.union(Y')>,set(),set(<<X,l>,k>)>*/
|
||||
return <<R,Y>,set(),set(<<X,l>,k>)>
|
||||
return make_triple(make_pair(R,Y), map<pair<int, int>, Pending>(), map(make_pair(X,l), set(k)))
|
||||
}
|
||||
/*fun process_symbol(grammer: ref Grammer, input: ref str, <X::=a.sb',l,k>, G, P) {*/
|
||||
fun process_symbol(grammer: ref Grammer, input: ref str, descript: Descriptor, G, P) {
|
||||
fun process_symbol(grammer: ref Grammer, input: ref str, descript: Descriptor, G: ref map<pair<int, int>, vec<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, Pending>, map<pair<int,int>, set<int>>> {
|
||||
var s = grammer.get_nonterminal_rules(descript.nonterminal)[descript.rule_idx][descript.idx_into_rule]
|
||||
var k = descript.pivot
|
||||
/*R = { r | <<s,k>,r> in P }*/
|
||||
R = P.get(<s,k>)
|
||||
R = P.get(make_pair(s,k))
|
||||
/*G' = {<<s,k>, <X::=as.b',l>>}*/
|
||||
Gp = map(<s,k>, pending(descript.nonterminal, descript.rule_idx, descript.idx_into_rule, descript.left)
|
||||
var Gp = map(make_pair(s,k), pending(descript.nonterminal, descript.rule_idx, descript.idx_into_rule, descript.left))
|
||||
if grammer.is_terminal(s) {
|
||||
/*return <match(input,<X::=a.sb',l,k>),set(),set()>*/
|
||||
return <match(input,descript),set(),set()>
|
||||
/*return make_triple(match(input,descript), map<pair<int,int>, vec<Pending>>(), map<pair<int,int>, set<int>>())*/
|
||||
return make_triple(matc(input,descript), map<pair<int,int>, vec<Pending>>(), map<pair<int,int>, set<int>>())
|
||||
} else if R.size() == 0 { // s in N
|
||||
/*return <<descend(grammer,s,k), set()>, G', set()>*/
|
||||
return <<descend(grammer,s,k), set()>, Gp, set()>
|
||||
return make_triple(make_pair(descend(grammer,s,k), set<BS>()), Gp, map<pair<int,int>, set<int>>())
|
||||
} else { // s in N and R != set()
|
||||
/*return <skip(k,<X::=as.b',l> R), G', set()>*/
|
||||
return <skip(k,pending(descript.nonterminal, descript.rule_idx, descript.idx_into_rule, descript.left), R), Gp, set()>
|
||||
return make_triple(skip(k,pending(descript.nonterminal, descript.rule_idx, descript.idx_into_rule, descript.left), R), Gp, map<pair<int,int>, set<int>>())
|
||||
}
|
||||
}
|
||||
/*fun match(input: ref str, <X::=a.sb',l,k>): <discriptor_set, set<BS>> {*/
|
||||
fun match(input: ref str, descript: Descriptor): <discriptor_set, set<BS>> {
|
||||
if input[k] == grammer.get_terminal(s) {
|
||||
fun matc(input: ref str, descript: Descriptor): pair<set<Descriptor>, set<BS>> {
|
||||
/*if input[k] == grammer.get_terminal(s) {*/
|
||||
if input[descript.piviot] == grammer.get_terminal(s) {
|
||||
/*return <set(X::=as.b',l,k+1>), set(<X::=as.b',l,k,k+1)>*/
|
||||
return <set(descriptor(descript.nonterminal, descript.rule_idx, descript.idx_into_rule+1, descript.left, descript.pivot+1)), set(bs(descript.nonterminal, descript.rule_idx, descript.idx_into_rule+1, descript.left, descript.pivot, descript.pivot+1))>
|
||||
return make_pair(set(descriptor(descript.nonterminal, descript.rule_idx, descript.idx_into_rule+1, descript.left, descript.pivot+1)), set(bs(descript.nonterminal, descript.rule_idx, descript.idx_into_rule+1, descript.left, descript.pivot, descript.pivot+1)))
|
||||
} else {
|
||||
return <set(), set()>
|
||||
return make_pair(set<Descriptor>(), set<BS>())
|
||||
}
|
||||
}
|
||||
fun skip(k:int,c,R) { return nmatch(k, set(c), R); }
|
||||
fun ascend(k:int, K, r) { return nmatch(k,K,set(r)); }
|
||||
fun nmatch(k:int, K, R) {
|
||||
fun skip(k: int, c: Pending, R: ref set<int>): pair<set<Descriptor>, set<BS>> { return nmatch(k, set(c), R); }
|
||||
fun ascend(k:int, K: ref set<Pending>, r: int): pair<set<Descriptor>, set<BS>> { return nmatch(k, K, set(r)); }
|
||||
fun nmatch(k:int, K: ref set<Pending>, R: ref set<int>): pair<set<Descriptor>, set<BS>> {
|
||||
/*var R = { <g,l,r> | <g,l> in K, r in R }*/
|
||||
/*var Y = { <g,l,k,r> | <g,l> in K, r in R }*/
|
||||
var R = set()
|
||||
var Y = set()
|
||||
for pending in K {
|
||||
for r in R {
|
||||
R.push(descriptor(pending.nonterminal, pending.rule_idx, pending.idx_into_rule, pending.left, r)
|
||||
Y.push(bs(pending.nonterminal, pending.rule_idx, pending.idx_into_rule, pending.left, k, r)
|
||||
var R = set<Descriptor>()
|
||||
var Y = set<BS>()
|
||||
/*for pending in K {*/
|
||||
for (var i = 0; i < K.data.size; i++;) {
|
||||
var pending = K.data[i]
|
||||
/*for r in R {*/
|
||||
for (var j = 0; j < R.data.size; j++;) {
|
||||
var r = R.data[j]
|
||||
R.push(descriptor(pending.nonterminal, pending.rule_idx, pending.idx_into_rule, pending.left, r))
|
||||
Y.push(bs(pending.nonterminal, pending.rule_idx, pending.idx_into_rule, pending.left, k, r))
|
||||
}
|
||||
}
|
||||
return <R,Y>
|
||||
return make_pair(R,Y)
|
||||
}
|
||||
|
||||
fun main(argc: int, argv: **char): int {
|
||||
var grammer.construct(): Grammer
|
||||
/*add_new_nonterminal(rule: ref vec<int>): int {*/
|
||||
/*add_to_nonterminal(nonterminal: int, rule: ref vec<int>) {*/
|
||||
/*add_terminal(c: char): int {*/
|
||||
|
||||
/*fungll(grammer: ref Grammer, input: ref str): set<BS> {*/
|
||||
|
||||
|
||||
var one = grammer.add_terminal('1')
|
||||
var E = grammer.add_new_nonterminal(vec<int>())
|
||||
grammer.add_to_nonterminal(E, vec(one))
|
||||
grammer.add_to_nonterminal(E, vec(E,E,E))
|
||||
|
||||
var BSR = fungll(grammer, str("1"))
|
||||
for (var i = 0; i < BSR.data.size; i++;) {
|
||||
var BS = BSR.data[i]
|
||||
println(str() + i + ": " + BSR.data[i].to_string())
|
||||
}
|
||||
|
||||
return 0
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user