问为什么添加到我的ANTLR4语法中会破坏一个无关的规则选择？
EN

Stack Overflow用户

提问于 2014-10-01 05:42:13

回答 1查看 865关注 0票数 0

我正在使用ANTLR4解析魔术卡(org.antlr:antlr4 4-运行时:4.0)。此最小化版本能够解析能力，如Enchant creature、Enchant black creature、.以及Protection from creatures、Protection from legendary creatures、.

现在我正在尝试启用Protection from black，这给我带来了问题。

我有以下的lexer和解析器语法，以及一个测试类：

莱克瑟：

lexer grammar OracleLexer;

Black: 'black';
Creature: 'creature';
Creatures: 'creatures';
Enchant: 'enchant';
From: 'from';
Legendary: 'legendary';
Protection: 'protection';

WS: [ \t\f\r\n]+ -> skip;

解析器：

parser grammar OracleParser;

options {
  tokenVocab = OracleLexer;
}

line:
  keywordAbility EOF;

keywordAbility:
  Enchant singularObject #Enchant |
    Protection From pluralObject #Protection;                   // (A)
//  Protection From (pluralObject | objectQuality) #Protection; // (B)



singularObject:
  objectQuality? Creature;

pluralObject:
  objectQuality? Creatures;



objectQuality:
  cardtypeQuality+? |
  supertypeQuality+ cardtypeQuality*? |
  colorQuality+ supertypeQuality* cardtypeQuality*?;

colorQuality:
  Black;

supertypeQuality:
  Legendary;

cardtypeQuality:
  Creature;

考试班：

import static java.util.Arrays.*;
import static java.util.Collections.*;

import java.util.List;

import org.antlr.v4.runtime.*;


public class OracleParserTest {
    private static final List<String> ruleNames = unmodifiableList(asList(OracleParser.ruleNames));

    public static void main(String[] args) throws Exception {
        parse("Enchant creature");
        parse("Enchant black creature");
        parse("Protection from black");
        parse("Protection from black creatures");
    }

    private static void parse(String ability) throws RecognitionException {
        OracleLexer lexer = new OracleLexer(new ANTLRInputStream(ability.toLowerCase()));
        lexer.removeErrorListeners();
        lexer.addErrorListener(new BailErrorListener());

        OracleParser parser = new OracleParser(new CommonTokenStream(lexer));
        parser.removeErrorListeners();
        parser.setErrorHandler(new BailErrorStrategy());

        RuleContext ctx = parser.line();

        System.out.println(ctx.toStringTree(ruleNames));
        ctx.inspect(ruleNames);
    }
}

在解析器中，有一个keywordAbility规则，在该规则中，我将替代(A)更改为(B)，突然之间，我再也无法与Enchant black creature匹配了。以下是文本形式的所有解析树，如果有帮助的话：

 original strings
Enchant creature
Enchant black creature
Protection from black
Protection from black creatures

 parse trees variant (A)
(line (keywordAbility enchant (singularObject creature)) <EOF>)
(line (keywordAbility enchant (singularObject (objectQuality (colorQuality black)) creature)) <EOF>)
 does not parse - expected
(line (keywordAbility protection from (pluralObject (objectQuality (colorQuality black)) creatures)) <EOF>)

 parse trees variant (B)
(line (keywordAbility enchant (singularObject creature)) <EOF>)
 does not parse - problem!
(line (keywordAbility protection from (objectQuality (colorQuality black))) <EOF>)
(line (keywordAbility protection from (pluralObject (objectQuality (colorQuality black)) creatures)) <EOF>)

下面是Enchant black creature的堆栈跟踪

Exception in thread "main" org.antlr.v4.runtime.misc.ParseCancellationException
    at org.antlr.v4.runtime.BailErrorStrategy.recover(BailErrorStrategy.java:51)
    at net.slightlymagic.laterna.oracle.grammar.OracleParser.objectQuality(OracleParser.java:462)
    at net.slightlymagic.laterna.oracle.grammar.OracleParser.singularObject(OracleParser.java:235)
    at net.slightlymagic.laterna.oracle.grammar.OracleParser.keywordAbility(OracleParser.java:161)
    at net.slightlymagic.laterna.oracle.grammar.OracleParser.line(OracleParser.java:79)
    at net.slightlymagic.laterna.oracle.grammar.OracleParserTest.parse(OracleParserTest.java:49)
    at net.slightlymagic.laterna.oracle.grammar.OracleParserTest.main(OracleParserTest.java:35)
Caused by: org.antlr.v4.runtime.NoViableAltException
    at org.antlr.v4.runtime.atn.ParserATNSimulator.noViableAlt(ParserATNSimulator.java:1532)
    at org.antlr.v4.runtime.atn.ParserATNSimulator.execATNWithFullContext(ParserATNSimulator.java:816)
    at org.antlr.v4.runtime.atn.ParserATNSimulator.execATN(ParserATNSimulator.java:701)
    at org.antlr.v4.runtime.atn.ParserATNSimulator.predictATN(ParserATNSimulator.java:389)
    at org.antlr.v4.runtime.atn.ParserATNSimulator.adaptivePredict(ParserATNSimulator.java:346)
    at net.slightlymagic.laterna.oracle.grammar.OracleParser.objectQuality(OracleParser.java:440)
    ... 5 more

我如何获得语法来解析这种能力(同样如此)，为什么它不能正常工作呢？

(我尝试删除BailErrorStrategy，这使它工作，但也导致90%的魔术能力解析为假阳性，并没有解释为什么这一能力以前起作用。BailErrorStrategy是否完全干扰回溯？)

antlr

antlr4

java

parsing

在线客服搭建方案

搭建集AI机器人、在线客服、智能外呼等多功能服务的客服系统，支持快速接入及自搭建客服系统，从而让顾客快速得到解答

回答 1

Stack Overflow用户

回答已采纳

发布于 2014-10-01 06:07:12

ANTLR 4不使用回溯(运行时库实际上不包含任何形式的此类功能代码)。

我无法在这条消息的末尾使用测试来重现这个结果。我的猜测是以下原因之一造成了这个问题：

你把BailErrorStrategy添加到你的雷克萨斯。它从未打算以这种方式使用。相反，按照my other answer中的建议将语法错误完全推迟到解析器。
您可能正在使用旧版本的ANTLR，该版本的错误已经修复。
只有在使用单独的lexer和解析器语法(测试使用组合语法)时，问题才会出现。

测试用例：

@Test
public void testCardParsing() throws Exception {
    String grammar =
        "grammar Oracle;\n" +
        "\n" +
        "line @init{setErrorHandler(new BailErrorStrategy());} @after {System.out.println($ctx.toStringTree(this));} :\n" +
        "  keywordAbility EOF;\n" +
        "\n" +
        "keywordAbility:\n" +
        "  Enchant singularObject #Enchant |\n" +
        "//  Protection From pluralObject #Protection;                   // (A)\n" +
        "    Protection From (pluralObject | objectQuality) #Protection; // (B)\n" +
        "\n" +
        "singularObject:\n" +
        "  objectQuality? Creature;\n" +
        "\n" +
        "pluralObject:\n" +
        "  objectQuality? Creatures;\n" +
        "\n" +
        "objectQuality:\n" +
        "  cardtypeQuality+? |\n" +
        "  supertypeQuality+ cardtypeQuality*? |\n" +
        "  colorQuality+ supertypeQuality* cardtypeQuality*?;\n" +
        "\n" +
        "colorQuality:\n" +
        "  Black;\n" +
        "\n" +
        "supertypeQuality:\n" +
        "  Legendary;\n" +
        "\n" +
        "cardtypeQuality:\n" +
        "  Creature;\n" +
        "\n" +
        "Black: 'black';\n" +
        "Creature: 'creature';\n" +
        "Creatures: 'creatures';\n" +
        "Enchant: 'enchant';\n" +
        "From: 'from';\n" +
        "Legendary: 'legendary';\n" +
        "Protection: 'protection';\n" +
        "\n" +
        "WS: [ \\t\\f\\r\\n]+ -> skip;";

    String input = "enchant creature";
    String found = execParser("Oracle.g4", grammar, "OracleParser", "OracleLexer", "line", input, true);
    assertEquals("(line (keywordAbility enchant (singularObject creature)) <EOF>)\n", found);
    assertNull(stderrDuringParse);

    input = "enchant black creature";
    found = execParser("Oracle.g4", grammar, "OracleParser", "OracleLexer", "line", input, false);
    assertEquals("(line (keywordAbility enchant (singularObject (objectQuality (colorQuality black)) creature)) <EOF>)\n", found);
    assertNull(stderrDuringParse);

    input = "protection from black";
    found = execParser("Oracle.g4", grammar, "OracleParser", "OracleLexer", "line", input, false);
    assertEquals("(line (keywordAbility protection from (objectQuality (colorQuality black))) <EOF>)\n", found);
    assertNull(stderrDuringParse);

    input = "protection from black creatures";
    found = execParser("Oracle.g4", grammar, "OracleParser", "OracleLexer", "line", input, false);
    assertEquals("(line (keywordAbility protection from (pluralObject (objectQuality (colorQuality black)) creatures)) <EOF>)\n", found);
    assertNull(stderrDuringParse);
}