All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.oracle.truffle.api.instrumentation.DefaultNearestNodeSearch Maven / Gradle / Ivy

Go to download

Truffle is a multi-language framework for executing dynamic languages that achieves high performance when combined with Graal.

There is a newer version: 24.1.1
Show newest version
/*
 * Copyright (c) 2018, 2019, Oracle and/or its affiliates. All rights reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * The Universal Permissive License (UPL), Version 1.0
 *
 * Subject to the condition set forth below, permission is hereby granted to any
 * person obtaining a copy of this software, associated documentation and/or
 * data (collectively the "Software"), free of charge and under any and all
 * copyright rights in the Software, and any and all patent rights owned or
 * freely licensable by each licensor hereunder covering either (i) the
 * unmodified Software as contributed to or provided by such licensor, or (ii)
 * the Larger Works (as defined below), to deal in both
 *
 * (a) the Software, and
 *
 * (b) any piece of software and/or hardware listed in the lrgrwrks.txt file if
 * one is included with the Software each a "Larger Work" to which the Software
 * is contributed by such licensors),
 *
 * without restriction, including without limitation the rights to copy, create
 * derivative works of, display, perform, and distribute the Software and make,
 * use, sell, offer for sale, import, export, have made, and have sold the
 * Software and the Larger Work(s), and to sublicense the foregoing rights on
 * either these or other terms.
 *
 * This license is subject to the following condition:
 *
 * The above copyright notice and either this complete permission notice or at a
 * minimum a reference to the UPL must be included in all copies or substantial
 * portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */
package com.oracle.truffle.api.instrumentation;

import java.util.Arrays;
import java.util.List;
import java.util.Set;

import com.oracle.truffle.api.instrumentation.InstrumentableNode.WrapperNode;
import com.oracle.truffle.api.nodes.Node;
import com.oracle.truffle.api.nodes.NodeUtil;
import com.oracle.truffle.api.nodes.NodeVisitor;
import com.oracle.truffle.api.source.SourceSection;

/**
 * Implements the default logic of searching for the nearest {@link Node node} to the given source
 * character index according to the execution flow, that is tagged with the given tag.
 * 

* This code assumes that the Truffle Node hierarchy corresponds with the logical guest language * AST structure. *

* If this is not the case for a particular guest language, * {@link InstrumentableNode#findNearestNodeAt(int, Set)} needs to be implemented, possibly with * logic in this class applied to the language logical AST nodes. */ class DefaultNearestNodeSearch { /** * We have a source offset (the character index) and a context node given. The offset is in * node's source. The context node is an instrumentable node, but may not have any particular * tags. We're searching for an instrumentable node tagged with at least one of the given tags, * which is nearest to the given offset according to the guest language execution flow. We find * the nearest tagged {@link Node node} in following steps: *

    *
  • When the offset is inside the context node, we search for the first tagged parent node * and call {@link #findChildTaggedNode(Node, int, Set, boolean)} to find a tagged child. If we * find it, we return it, if not, we continue to search recursively for a tagged child of node's * parent. If we encounter the first tagged parent node, return it, otherwise return tagged * child found during the recursive process, if any.
  • *
  • When the offset is behind the context node, we return the last tagged child of the * context node.
  • *
  • When the offset is before the context node, we return the first tagged child of the * context node.
  • *
*/ static Node findNearestNodeAt(int offset, Node contextNode, Set> tags) { Node node = (Node) ((InstrumentableNode) contextNode).materializeInstrumentableNodes(tags); SourceSection section = node.getSourceSection(); int startIndex = section.getCharIndex(); int endIndex = getCharEndIndex(section); if (startIndex <= offset && offset <= endIndex) { Node ch; while ((ch = findChildTaggedNode(node, offset, tags)) == null) { node = node.getParent(); if (node == null) { break; } } return ch; } else if (endIndex < offset) { return findLastNode(node, tags); } else { // offset < startIndex return findFirstNode(node, tags); } } private static void forEachInstrumentableChild(Node parent, NodeVisitor visitor, Set> tags) { NodeUtil.forEachChild(parent, new NodeVisitor() { private boolean keepVisiting = true; @Override public boolean visit(Node childNode) { Node ch = childNode; if (ch instanceof WrapperNode) { ch = ((WrapperNode) ch).getDelegateNode(); } if (ch instanceof InstrumentableNode && ((InstrumentableNode) ch).isInstrumentable()) { ch = (Node) ((InstrumentableNode) ch).materializeInstrumentableNodes(tags); } else { // An unknown node, process its children on the same level. NodeUtil.forEachChild(ch, this); return keepVisiting; } return (keepVisiting = visitor.visit(ch)); } }); } /** * Finds the nearest tagged {@link Node node}. See the algorithm description at * {@link InstrumentableNode#findNearestNodeAt(int, Set)}. */ private static Node findChildTaggedNode(Node node, int offset, Set> tags) { // Nodes with lower offset that we search for, inspected from the highest one. SortedNodes lowerNodes = new SortedNodes(); // Nodes with higher offset that we search for, inspected from the smallest one. SortedNodes higherNodes = new SortedNodes(); final Node[] foundNode = new Node[]{null}; forEachInstrumentableChild(node, new NodeVisitor() { int highestLowerTaggedNodeStart = -1; // The nearest tagged node with lower offset int highestLowerTaggedNodeEnd = -1; int lowestHigherTaggedNodeStart = -1; // The nearest tagged node with higher offset int lowestHigherTaggedNodeEnd = -1; @Override public boolean visit(Node ch) { SourceSection ss = ch.getSourceSection(); if (ss == null || !ss.isAvailable()) { return true; } boolean isTagged = isTaggedWith((InstrumentableNode) ch, tags); int i1 = ss.getCharIndex(); int i2 = getCharEndIndex(ss); if (isTagged && offset == i1) { // We're at it foundNode[0] = ch; return false; } if (i1 <= offset && offset <= i2) { // In an encapsulating source section Node taggedNode = findChildTaggedNode(ch, offset, tags); if (taggedNode != null) { foundNode[0] = taggedNode; return false; } if (isTagged) { // If nothing in and is tagged, return it foundNode[0] = ch; return false; } } if (offset < i1 && !(lowestHigherTaggedNodeStart <= i1 && i2 <= lowestHigherTaggedNodeEnd)) { // We're after the offset if (lowestHigherTaggedNodeStart == -1 || lowestHigherTaggedNodeStart > i1) { higherNodes.add(ch, i1); if (isTagged) { lowestHigherTaggedNodeStart = i1; lowestHigherTaggedNodeEnd = i2; higherNodes.cutHigherThan(i1); } } } if (i2 < offset && !(highestLowerTaggedNodeStart <= i1 && i2 <= highestLowerTaggedNodeEnd)) { // We're before the offset if (highestLowerTaggedNodeStart == -1 || (highestLowerTaggedNodeStart < i1)) { lowerNodes.add(ch, i1); if (isTagged) { highestLowerTaggedNodeStart = i1; highestLowerTaggedNodeEnd = i2; lowerNodes.cutLowerThan(i1); } } } return true; } }, tags); if (foundNode[0] != null) { return foundNode[0]; } Node taggedNode = findChildTaggedNode(higherNodes.nodes, higherNodes.size, offset, tags, false); if (taggedNode == null) { taggedNode = findChildTaggedNode(lowerNodes.nodes, lowerNodes.size, offset, tags, true); } return taggedNode; } private static Node findChildTaggedNode(Node[] nodes, int size, int offset, Set> tags, boolean reverse) { if (nodes == null) { return null; } for (int i = (reverse ? size - 1 : 0); reverse ? i >= 0 : i < size; i = (reverse) ? i - 1 : i + 1) { Node node = nodes[i]; if (isTaggedWith(node, tags)) { return node; } Node taggedNode = findChildTaggedNode(node, offset, tags); if (taggedNode != null) { return taggedNode; } } return null; } /** Get the last character index (inclusive). */ private static int getCharEndIndex(SourceSection ss) { if (ss.getCharLength() > 0) { return ss.getCharEndIndex() - 1; } else { return ss.getCharIndex(); } } private static Node findFirstNode(Node contextNode, Set> tags) { Node[] first = new Node[]{null}; contextNode.accept(new NodeVisitor() { @Override public boolean visit(Node node) { if (isTaggedWith(node, tags)) { first[0] = node; return false; } return true; } }); return first[0]; } private static Node findLastNode(Node contextNode, Set> tags) { if (isTaggedWith(contextNode, tags)) { return contextNode; } List children = NodeUtil.findNodeChildren(contextNode); for (int i = children.size() - 1; i >= 0; i--) { Node ch = children.get(i); if (ch instanceof WrapperNode) { ch = ((WrapperNode) ch).getDelegateNode(); } Node last = findLastNode(ch, tags); if (last != null) { return last; } } return null; } private static boolean isTaggedWith(Node node, Set> tags) { if (node instanceof InstrumentableNode && ((InstrumentableNode) node).isInstrumentable()) { InstrumentableNode inode = ((InstrumentableNode) node).materializeInstrumentableNodes(tags); return isTaggedWith(inode, tags); } return false; } private static boolean isTaggedWith(InstrumentableNode inode, Set> tags) { for (Class tag : tags) { if (inode.hasTag(tag)) { return true; } } return false; } private static final class SortedNodes { private static final int DEFAULT_SIZE = 10; private Node[] nodes = null; int[] nodeOffsets = null; int size = 0; void add(Node node, int offset) { if (nodes == null) { nodes = new Node[DEFAULT_SIZE]; nodeOffsets = new int[DEFAULT_SIZE]; nodes[0] = node; nodeOffsets[0] = offset; size++; } else { ensureCapacity(size + 1); if (nodeOffsets[size - 1] < offset) { // common case nodes[size] = node; nodeOffsets[size] = offset; } else { int index = Arrays.binarySearch(nodeOffsets, 0, size, offset); if (index < 0) { index = -index - 1; } System.arraycopy(nodes, index, nodes, index + 1, size - index); nodes[index] = node; System.arraycopy(nodeOffsets, index, nodeOffsets, index + 1, size - index); nodeOffsets[index] = offset; } size++; } } void ensureCapacity(int capacity) { if (nodes.length < capacity) { int newCapacity = capacity + (capacity >> 1); if (newCapacity < capacity) { newCapacity = capacity; } nodes = Arrays.copyOf(nodes, newCapacity); nodeOffsets = Arrays.copyOf(nodeOffsets, newCapacity); } } void cutHigherThan(int offset) { int index = Arrays.binarySearch(nodeOffsets, 0, size, offset); if (index < 0) { index = -index - 1; } else { index++; } if (index < size) { size = index; } } void cutLowerThan(int offset) { int index = Arrays.binarySearch(nodeOffsets, 0, size, offset); if (index < 0) { index = -index - 1; } if (index > 0) { System.arraycopy(nodes, index, nodes, 0, size - index); System.arraycopy(nodeOffsets, index, nodeOffsets, 0, size - index); size -= index; } } } }




© 2015 - 2024 Weber Informatics LLC | Privacy Policy