some tests failing because things have been made reference in vector, but grammer actually generates the state set for the real grammer in 2 minutes or so after a day of profiling and bugfixing, so this is gonna be committed.

This commit is contained in:
Nathan Braswell
2015-08-05 03:43:34 -04:00
parent e1dbe08c0a
commit dec9b7d0bd
13 changed files with 151 additions and 60 deletions

View File

@@ -1,6 +1,7 @@
import string
import vector
import set
import stack
import map
import symbol
import regex
@@ -176,90 +177,112 @@ obj grammer (Object) {
fun calculate_state_automaton() {
state_automata.items = vector::vector(rules[0].with_lookahead(set::set(symbol::eof_symbol())))
io::println("pre first closure")
state_automata = closure(state_automata)
var states = set::set(state_automata)
var newItem = true
while (newItem) {
newItem = false
states.for_each(fun(I: state) {
var possGoto = set::set<symbol::symbol>()
I.items.for_each(fun(r: rule) {
if (!r.at_end())
possGoto.add(r.next())
})
possGoto.for_each(fun(X: symbol::symbol) {
var goneState = goto(I, X)
if (goneState.items.size && !states.contains(goneState)) {
states.add(goneState)
newItem = true
}
})
io::println("post first closure")
var states = vector::vector(state_automata) // vector instead of set because we need to iterate by index
var newItems = stack::stack(0) // 0 is the index of the first and only item in states
var count = 0
while (newItems.size()) {
if (count%200 == 0) {
io::print("calculate_state_automaton while")
io::println(count)
}
count++
var I = newItems.pop()
var possGoto = set::set<symbol::symbol>()
states[I].items.for_each(fun(r: ref rule) {
if (!r.at_end())
possGoto.add(r.next())
})
possGoto.for_each(fun(X: ref symbol::symbol) {
var goneState = goto(states[I], X)
if (goneState.items.size && !states.contains(goneState)) {
newItems.push(states.size)
states.add(goneState)
}
})
}
io::println("ALL STATES:\n")
states.for_each(fun(i: state) {
states.for_each(fun(i: ref state) {
io::println("STATE:\n")
i.items.for_each(fun(r: rule) {
i.items.for_each(fun(r: ref rule) {
io::println(string::string("\t") + r.to_string())
})
})
io::println(" there were : states")
io::println(states.size)
}
fun closure(initial: state): state {
fun closure(initial: ref state): state {
initial.items = closure(initial.items)
return initial
}
fun closure(initial: vector::vector<rule>): vector::vector<rule> {
fun closure(initial: ref vector::vector<rule>): vector::vector<rule> {
var continueIt = true
var count = 0
//var count = 0
while (continueIt) {
io::print("while")
io::println(count)
count++
//io::print("closure while")
//io::println(count)
//count++
continueIt = false
initial.for_each(fun(i: rule) {
if (i.at_end()) {
return; // continue the for-each
for (var i = 0; i < initial.size; i++;) {
if (initial[i].at_end()) {
continue
}
rules.for_each(fun(r: rule) {
rules.for_each(fun(r: ref rule) {
// if i is |a::=c . Bb, a|, we're doing each B::=... in rules
if (r.lhs != i.next())
return // continuing rule-for_each
if (r.lhs != initial[i].next())
return // continue the for-each
// add r with lookahead
var newLookahead = first_vector(i.after_next())
var newLookahead = first_vector(initial[i].after_next())
if (newLookahead.contains(symbol::null_symbol())) {
newLookahead.remove(symbol::null_symbol())
newLookahead.add(i.lookahead)
newLookahead.add(initial[i].lookahead)
}
var alreadyInInSomeForm = false
for (var index = 0; index < initial.size; index++;) {
if (initial[index].equals_but_lookahead(r) && !initial[index].lookahead.contains(newLookahead)) {
//io::println(initial[index].to_string())
//io::println("and")
//io::println(r.to_string())
//io::println("are the same with different lookaheads")
initial[index].lookahead += newLookahead
continueIt = true
//io::println("contineu because equal_but_different")
return // continuing rule-for_each
if (initial[index].equals_but_lookahead(r)) {
alreadyInInSomeForm = true
if (!initial[index].lookahead.contains(newLookahead)) {
//io::println("\n\n\n")
//io::println(initial[index].to_string())
//io::println("and")
//io::println(r.to_string())
//io::println("with")
//var result = string::string("|lookahead {")
//newLookahead.for_each(fun(i: symbol::symbol) {
//result += i.to_string()
//})
//io::println(result)
//io::println("are the same with different lookaheads")
initial[index].lookahead += newLookahead
//io::println("so now it's")
//io::println(initial[index].to_string())
//io::println("contineu because equal_but_different")
continueIt = true
return // continue the rules for-each
}
}
}
var newRule = r.with_lookahead(newLookahead)
if (!initial.contains(newRule)) {
if (!alreadyInInSomeForm) {
continueIt = true
//io::println("\n\n\n")
//io::println("contineu because not contains")
initial.add(newRule)
//io::println(newRule.to_string())
initial.add(r.with_lookahead(newLookahead))
}
})
})
}
}
return initial
}
fun goto(I: state, X: symbol::symbol): state {
fun goto(I: ref state, X: ref symbol::symbol): state {
// loop through i, find all that have thing::= something . X more,
// add thing ::= something X . more
var jPrime = vector::vector<rule>()
I.items.for_each(fun(i: rule) {
I.items.for_each(fun(i: ref rule) {
if (!i.at_end() && i.next() == X)
jPrime.add(i.advanced())
})
@@ -323,7 +346,7 @@ obj rule (Object) {
lookahead.destruct()
}
fun next(): symbol::symbol {
fun next(): ref symbol::symbol {
return rhs[position]
}
fun after_next(): vector::vector<symbol::symbol> {

View File

@@ -71,6 +71,9 @@ obj set<T> (Object) {
}
data.remove(idx)
}
fun for_each(func: fun(ref T):void) {
data.for_each(func)
}
fun for_each(func: fun(T):void) {
data.for_each(func)
}

43
stdlib/stack.krak Normal file
View File

@@ -0,0 +1,43 @@
import vector
fun stack<T>():stack<T> {
var out.construct():stack<T>
return out
}
fun stack<T>(in:T):stack<T> {
var out.construct():stack<T>
out.push(in)
return out
}
obj stack<T> (Object) {
var data: vector::vector<T>
fun construct(): *stack<T> {
data.construct()
return this
}
fun copy_construct(other: *stack<T>) {
data.copy_construct(&other->data)
}
fun destruct() {
data.destruct()
}
fun operator=(other: ref stack<T>) {
data = other.data
}
fun push(it: ref T) {
data.addEnd(it)
}
fun pop(): T {
var toRet = data[data.size-1]
data.remove(data.size-1)
return toRet
}
fun top(): T {
return data[data.size-1]
}
fun size(): int {
return data.size
}
}

View File

@@ -121,19 +121,19 @@ obj vector<T> (Object) {
// on an object that we want to put in a vector. In this way we avoid the problem
// by not generating this function unless it's called - we also get the ability to
// do a find index using a different type, which could be fun.
fun find<U>(value: U): int {
fun find<U>(value: ref U): int {
for (var i = 0; i < size; i++;)
if (data[i] == value)
return i;
return -1;
}
// ditto
fun contains<U>(item: U): bool {
fun contains<U>(item: ref U): bool {
return find(item) != -1
}
// yep
fun operator==<U>(other:vector<U>):bool {
fun operator==<U>(other: ref vector<U>):bool {
if (size != other.size)
return false
for (var i = 0; i < size; i++;)
@@ -147,7 +147,7 @@ obj vector<T> (Object) {
return;
data[index] = dataIn;
}
fun add_all(dataIn: vector<T>): void {
fun add_all(dataIn: ref vector<T>): void {
for (var i = 0; i < dataIn.size; i++;)
addEnd(dataIn[i]);
}
@@ -156,8 +156,8 @@ obj vector<T> (Object) {
if (!contains(dataIn))
addEnd(dataIn)
}
fun add(dataIn: T): void { addEnd(dataIn); }
fun addEnd(dataIn: T): void {
fun add(dataIn: ref T): void { addEnd(dataIn); }
fun addEnd(dataIn: ref T): void {
if (size+1 >= available)
resize((size+1)*2);
maybe_copy_construct(&data[size], &dataIn);
@@ -173,6 +173,10 @@ obj vector<T> (Object) {
size--
}
fun for_each(func: fun(ref T):void):void {
for (var i = 0; i < size; i++;)
func(data[i])
}
fun for_each(func: fun(T):void):void {
for (var i = 0; i < size; i++;)
func(data[i])