miloignis/kraken
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Use the new parser (now with reduction functions) for mal

Browse Source
This commit is contained in:
Nathan Braswell
2020-03-29 20:42:51 -04:00
parent a9bb011bd9
commit ce7d228358
2 changed files with 301 additions and 86 deletions
Download Patch File Download Diff File Expand all files Collapse all files

123
fungll.krak
View File

@@ -7,46 +7,57 @@ import str:*
import regex:*
// nonterminals are negative, terminals are positive
obj Grammer (Object) {
obj Grammer<T> (Object) {
var nonterminals: vec<vec<vec<int>>>
var nonterminal_names: vec<str>
var nonterminal_funs: vec<vec<fun(ref vec<T>): T>>
var terminals: vec<regex>
var terminal_funs: vec<fun(ref str,int,int): T>
var start_symbol: int
fun construct(): *Grammer {
nonterminals.construct()
nonterminal_names.construct()
nonterminal_funs.construct()
terminals.construct()
terminal_funs.construct()
start_symbol = 0
return this
}
fun copy_construct(old: *Grammer): void {
nonterminals.copy_construct(&old->nonterminals)
nonterminal_names.copy_construct(&old->nonterminal_names)
nonterminal_funs.copy_construct(&old->nonterminal_funs)
terminals.copy_construct(&old->terminals)
terminal_funs.copy_construct(&old->terminal_funs)
start_symbol = old->start_symbol
}
fun destruct(): void {
nonterminals.destruct()
nonterminal_names.destruct()
nonterminal_funs.destruct()
terminals.destruct()
terminal_funs.destruct()
}
fun operator=(other:ref Grammer):void {
destruct()
copy_construct(&other)
}
fun add_new_nonterminal(name: *char, rule: ref vec<int>): int {
return add_new_nonterminal(str(name), rule)
fun add_new_nonterminal(name: *char, rule: ref vec<int>, f: fun(ref vec<T>): T): int {
return add_new_nonterminal(str(name), rule, f)
}
fun add_new_nonterminal(name: ref str, rule: ref vec<int>): int {
fun add_new_nonterminal(name: ref str, rule: ref vec<int>, f: fun(ref vec<T>): T): int {
nonterminals.add(vec(rule))
nonterminal_names.add(name)
nonterminal_funs.add(vec(f))
return -1*nonterminals.size
}
fun add_to_nonterminal(nonterminal: int, rule: ref vec<int>) {
fun add_to_nonterminal(nonterminal: int, rule: ref vec<int>, f: fun(ref vec<T>): T) {
nonterminals[(-1*nonterminal)-1].add(rule)
nonterminal_funs[(-1*nonterminal)-1].add(f)
}
fun add_terminal(c: *char): int {
fun add_terminal(c: *char, f: fun(ref str,int,int): T): int {
terminals.add(regex(c))
terminal_funs.add(f)
return terminals.size
}
fun get_nonterminal_rules(nonterminal: int): ref vec<vec<int>> {
@@ -69,6 +80,9 @@ obj Grammer (Object) {
}
var erminal = nonterminals[(-1*it.nonterminal)-1][it.rule_idx][i]
rule_str += to_string(erminal)
if i < nonterminals[(-1*it.nonterminal)-1][it.rule_idx].size-1 {
rule_str += " "
}
}
if it.idx_into_rule == nonterminals[(-1*it.nonterminal)-1][it.rule_idx].size {
rule_str += "*"
@@ -82,6 +96,60 @@ obj Grammer (Object) {
return terminals[erminal-1].regexString
}
}
fun eval_BSR(input: ref str, BSR: ref set<BS>): T {
var top = -1
for (var i = 0; i < BSR.data.size; i++;) {
if BSR.data[i].nonterminal == start_symbol && BSR.data[i].idx_into_rule == nonterminals[(-1*BSR.data[i].nonterminal)-1][BSR.data[i].rule_idx].size && BSR.data[i].left == 0 && BSR.data[i].right == input.length() {
top = i
break
}
}
if top == -1 {
error("Could not find top")
}
return eval_BSR(input, BSR, top)
}
fun eval_BSR(input: ref str, BSR: ref set<BS>, c: int): T {
var bs = BSR.data[c]
var nonterminal = (-1*bs.nonterminal)-1
if bs.idx_into_rule != nonterminals[nonterminal][bs.rule_idx].size {
error("Evaluating BSR from not the end!")
}
var params = vec<T>()
for (var i = bs.idx_into_rule-1; i >= 0; i--;) {
var erminal = nonterminals[nonterminal][bs.rule_idx][i]
if is_terminal(erminal) {
var right_value = terminal_funs[erminal-1](input, bs.pivot, bs.right)
params.add(right_value)
} else {
/*var right = find_comp(erminal, bs.pivot, bs.right)*/
var right = -1
var sub_nonterminal_idx = (-1*erminal)-1
for (var j = 0; j < BSR.data.size; j++;) {
if BSR.data[j].nonterminal == erminal && BSR.data[j].idx_into_rule == nonterminals[sub_nonterminal_idx][BSR.data[j].rule_idx].size && BSR.data[j].left == bs.pivot && BSR.data[j].right == bs.right {
right = j
break
}
}
var right_value = eval_BSR(input, BSR, right)
params.add(right_value)
}
// get the new left bs
if i != 0 {
/*var new_bs_idx = find_mid(bs.nonterminal, bs.rule_idx, i, bs.left, bs.pivot)*/
var new_bs_idx = -1
for (var j = 0; j < BSR.data.size; j++;) {
if BSR.data[j].nonterminal == bs.nonterminal && BSR.data[j].rule_idx == bs.rule_idx && BSR.data[j].idx_into_rule == i && BSR.data[j].left == bs.left && BSR.data[j].right == bs.pivot {
new_bs_idx = j
break
}
}
bs = BSR.data[new_bs_idx]
}
}
var to_ret = nonterminal_funs[nonterminal][bs.rule_idx](params.reverse())
return to_ret
}
}
obj Pending (Object) {
var nonterminal: int
@@ -197,7 +265,7 @@ fun bs(nonterminal: int, rule_idx: int, idx_into_rule: int, left: int, pivot: in
var to_ret.construct(nonterminal, rule_idx, idx_into_rule, left, pivot, right): BS
return to_ret
}
fun fungll(grammer: ref Grammer, input: ref str): set<BS> {
fun fungll<T>(grammer: ref Grammer<T>, input: ref str): set<BS> {
var R = descend(grammer, grammer.start_symbol, 0)
var U = set<Descriptor>()
var G = map<pair<int, int>, set<Pending>>()
@@ -211,10 +279,10 @@ fun fungll(grammer: ref Grammer, input: ref str): set<BS> {
var Gp = it.second
var Pp = it.third
var U_with_d = U.union(set(d));
var nextR = R.union(Rp) - U_with_d
U.add(d)
var nextR = R.union(Rp) - U
R = nextR
U = U_with_d
for (var i = 0; i < Gp.keys.size; i++;) {
if G.contains_key(Gp.keys[i]) {
G[Gp.keys[i]].add(Gp.values[i])
@@ -233,13 +301,13 @@ fun fungll(grammer: ref Grammer, input: ref str): set<BS> {
}
return Y
}
fun descend(grammer: ref Grammer, symbol: int, l: int): set<Descriptor> {
fun descend<T>(grammer: ref Grammer<T>, symbol: int, l: int): set<Descriptor> {
var to_ret = set<Descriptor>()
for (var rhs = 0; rhs < grammer.get_nonterminal_rules(symbol).size; rhs++;)
to_ret.add(descriptor(symbol, rhs, 0, l, l))
return to_ret
}
fun process(grammer: ref Grammer, input: ref str, descript: Descriptor, G: ref map<pair<int, int>, set<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, set<Pending>>, map<pair<int,int>, set<int>>> {
fun process<T>(grammer: ref Grammer<T>, input: ref str, descript: Descriptor, G: ref map<pair<int, int>, set<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, set<Pending>>, map<pair<int,int>, set<int>>> {
// if at end / end is emptystr
if descript.idx_into_rule == grammer.get_nonterminal_rules(descript.nonterminal)[descript.rule_idx].size {
return process_e(grammer, descript, G, P)
@@ -247,7 +315,7 @@ fun process(grammer: ref Grammer, input: ref str, descript: Descriptor, G: ref m
return process_symbol(grammer, input, descript, G, P)
}
}
fun process_e(grammer: ref Grammer, descript: Descriptor, G: ref map<pair<int, int>, set<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, set<Pending>>, map<pair<int,int>, set<int>>> {
fun process_e<T>(grammer: ref Grammer<T>, descript: Descriptor, G: ref map<pair<int, int>, set<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, set<Pending>>, map<pair<int,int>, set<int>>> {
var nonterminal: int
var rule_idx: int
var left: int
@@ -264,7 +332,7 @@ fun process_e(grammer: ref Grammer, descript: Descriptor, G: ref map<pair<int, i
}
return make_triple(make_pair(R,Y), map<pair<int, int>, set<Pending>>(), map(make_pair(X,l), set(k)))
}
fun process_symbol(grammer: ref Grammer, input: ref str, descript: Descriptor, G: ref map<pair<int, int>, set<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, set<Pending>>, map<pair<int,int>, set<int>>> {
fun process_symbol<T>(grammer: ref Grammer<T>, input: ref str, descript: Descriptor, G: ref map<pair<int, int>, set<Pending>>, P: ref map<pair<int,int>, set<int>>): triple<pair<set<Descriptor>, set<BS>>, map<pair<int, int>, set<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
var R = P.get_with_default(make_pair(s,k), set<int>())
@@ -277,7 +345,7 @@ fun process_symbol(grammer: ref Grammer, input: ref str, descript: Descriptor, G
return make_triple(skip(k,pending(descript.nonterminal, descript.rule_idx, descript.idx_into_rule+1, descript.left), R), Gp, map<pair<int,int>, set<int>>())
}
}
fun matc(grammer: ref Grammer, input: ref str, descript: Descriptor): pair<set<Descriptor>, set<BS>> {
fun matc<T>(grammer: ref Grammer<T>, input: ref str, descript: Descriptor): pair<set<Descriptor>, set<BS>> {
/*println("trying to match " + grammer.to_string(grammer.get_nonterminal_rules(descript.nonterminal)[descript.rule_idx][descript.idx_into_rule]))*/
var match_length = grammer.match_terminal(grammer.get_nonterminal_rules(descript.nonterminal)[descript.rule_idx][descript.idx_into_rule], input, descript.pivot)
if match_length > 0 {
@@ -303,18 +371,31 @@ fun nmatch(k:int, K: ref set<Pending>, R: ref set<int>): pair<set<Descriptor>, s
return make_pair(Rp,Y)
}
/*fun main(argc: int, argv: **char): int {*/
/*var grammer.construct(): Grammer*/
/*var one = grammer.add_terminal("12")*/
/*var E = grammer.add_new_nonterminal("E", vec<int>())*/
/*grammer.add_to_nonterminal(E, vec(one))*/
/*grammer.add_to_nonterminal(E, vec(E,E,E))*/
/*var grammer.construct(): Grammer<int>*/
/*var Number = grammer.add_new_nonterminal("Number", vec(grammer.add_terminal("[0-9]+", fun(input: ref str, l: int, r: int): int { return string_to_num<int>(input.slice(l,r)); })), fun(i: ref vec<int>): int { return i[0]; })*/
/*var mult = grammer.add_terminal("\\*", fun(input: ref str, l: int, r: int): int { return 1; })*/
/*var Factor = grammer.add_new_nonterminal("Factor", vec(Number), fun(i: ref vec<int>): int { return i[0]; })*/
/*grammer.add_to_nonterminal(Factor, vec(Factor, mult, Number), fun(i: ref vec<int>): int { return i[0]*i[2]; })*/
/*var add = grammer.add_terminal("\\+", fun(input: ref str, l: int, r: int): int { return 1; })*/
/*var Term = grammer.add_new_nonterminal("Term", vec(Factor), fun(i: ref vec<int>): int { return i[0]; })*/
/*grammer.add_to_nonterminal(Term, vec(Term, add, Factor), fun(i: ref vec<int>): int { return i[0]+i[2]; })*/
/*grammer.set_start_symbol(Term)*/
/*var input = str("1+23*44")*/
/*var BSR = fungll(grammer, input)*/
/*var BSR = fungll(grammer, str("1212"))*/
/*println(str("length of BSR is: ") + BSR.size())*/
/*for (var i = 0; i < BSR.data.size; i++;) {*/
/*var BS = BSR.data[i]*/
/*println(str() + i + ": " + grammer.to_string(BSR.data[i]))*/
/*}*/
/*var res = grammer.eval_BSR(input, BSR)*/
/*println(str("result of grammer.eval_BSR(fungll(grammer, ") + input + ")) = " + res)*/
/*return 0*/
/*}*/