1
0
mirror of https://github.com/DanilaFe/abacus synced 2024-12-22 15:30:09 -08:00

Implement a tentative pattern class that can be compiled from a string.

This commit is contained in:
Danila Fedorin 2017-07-24 17:42:52 -07:00
parent 2374c167a4
commit 6200381016
2 changed files with 202 additions and 0 deletions

View File

@ -0,0 +1,162 @@
package org.nwapw.abacus.lexing.pattern;
import java.util.Collection;
import java.util.HashMap;
import java.util.Stack;
import java.util.function.Function;
public class Pattern<T> {
private T id;
private PatternNode<T> head;
private String source;
private int index;
private HashMap<Character, Function<PatternChain<T>, PatternChain<T>>> operations =
new HashMap<Character, Function<PatternChain<T>, PatternChain<T>>>() {{
put('+', Pattern.this::transformPlus);
put('*', Pattern.this::transformStar);
put('?', Pattern.this::transformQuestion);
}};
private PatternChain<T> transformPlus(PatternChain<T> chain){
chain.tail.outputStates.add(chain.head);
return chain;
}
private PatternChain<T> transformStar(PatternChain<T> chain){
LinkNode<T> newTail = new LinkNode<>();
LinkNode<T> newHead = new LinkNode<>();
newHead.outputStates.add(chain.head);
newHead.outputStates.add(newTail);
chain.tail.outputStates.add(newTail);
newTail.outputStates.add(newHead);
chain.head = newHead;
chain.tail = newTail;
return chain;
}
private PatternChain<T> transformQuestion(PatternChain<T> chain){
LinkNode<T> newTail = new LinkNode<>();
LinkNode<T> newHead = new LinkNode<>();
newHead.outputStates.add(chain.head);
newHead.outputStates.add(newTail);
chain.tail.outputStates.add(newTail);
chain.head = newHead;
chain.tail = newTail;
return chain;
}
private PatternChain<T> combineChains(Collection<PatternChain<T>> collection){
LinkNode<T> head = new LinkNode<>();
LinkNode<T> tail = new LinkNode<>();
PatternChain<T> newChain = new PatternChain<>(head, tail);
for(PatternChain<T> chain : collection){
head.outputStates.add(chain.head);
chain.tail.outputStates.add(tail);
}
return newChain;
}
private PatternChain<T> parseValue(){
if(index >= source.length()) return null;
if(source.charAt(index) == '\\'){
if(++index >= source.length()) return null;
}
return new PatternChain<>(new ValueNode<>(source.charAt(index++)));
}
private PatternChain<T> parseOr(){
Stack<PatternChain<T>> orStack = new Stack<>();
while(index < source.length() && source.charAt(index) != ']'){
if(source.charAt(index) == '-'){
index++;
if(orStack.empty() || orStack.peek().tail.range() == '\0') return null;
PatternChain<T> bottomRange = orStack.pop();
PatternChain<T> topRange = parseValue();
if(topRange == null || topRange.tail.range() == '\0') return null;
orStack.push(new PatternChain<>(new RangeNode<>(bottomRange.tail.range(), topRange.tail.range())));
} else {
PatternChain<T> newChain = parseValue();
if(newChain == null) return null;
orStack.push(newChain);
}
index++;
}
if(index >= source.length()) return null;
return combineChains(orStack);
}
private PatternChain<T> parseSegment(boolean isSubsegment){
if(index >= source.length() || ((source.charAt(index) != '(') && isSubsegment)) return null;
if(isSubsegment) index++;
Stack<PatternChain<T>> orChain = new Stack<>();
PatternChain<T> fullChain = new PatternChain<>();
PatternChain<T> currentChain = null;
while (index < source.length() && source.charAt(index) != ')'){
char currentChar = source.charAt(index);
if(operations.containsKey(currentChar)){
if(currentChain == null) return null;
currentChain = operations.get(currentChar).apply(currentChain);
fullChain.append(currentChain);
currentChain = null;
index++;
} else if(currentChar == '|'){
if(currentChain == null) return null;
fullChain.append(currentChain);
orChain.push(fullChain);
currentChain = null;
fullChain = new PatternChain<>();
} else if(currentChar == '('){
if(currentChain != null) {
fullChain.append(currentChain);
}
currentChain = parseSegment(true);
if(currentChain == null) return null;
} else if(currentChar == '['){
if(currentChain != null){
fullChain.append(currentChain);
}
currentChain = parseOr();
if(currentChain == null) return null;
} else {
if(currentChain != null){
fullChain.append(currentChain);
}
currentChain = parseValue();
if(currentChain == null) return null;
}
}
if(!(!isSubsegment || (index < source.length() && source.charAt(index) == ')'))) return null;
if(isSubsegment) index++;
if(currentChain != null) fullChain.append(currentChain);
if(!orChain.empty()){
orChain.push(fullChain);
fullChain = combineChains(orChain);
}
return fullChain;
}
public Pattern(String from, T id){
this.id = id;
index = 0;
source = from;
PatternChain<T> chain = parseSegment(false);
if(chain == null) {
head = null;
} else {
chain.append(new EndNode<>(id));
head = chain.head;
}
}
}

View File

@ -0,0 +1,40 @@
package org.nwapw.abacus.lexing.pattern;
public class PatternChain<T> {
public PatternNode<T> head;
public PatternNode<T> tail;
public PatternChain(PatternNode<T> head, PatternNode<T> tail){
this.head = head;
this.tail = tail;
}
public PatternChain(PatternNode<T> node){
this(node, node);
}
public PatternChain(){
this(null);
}
public void append(PatternChain<T> other){
if(other.head == null || tail == null) {
this.head = other.head;
this.tail = other.tail;
} else {
tail.outputStates.add(other.head);
tail = other.tail;
}
}
public void append(PatternNode<T> node){
if(tail == null){
head = tail = node;
} else {
tail.outputStates.add(node);
tail = node;
}
}
}