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

org.opendaylight.yangtools.yang.ir.IRArgument Maven / Gradle / Ivy

There is a newer version: 14.0.4
Show newest version
/*
 * Copyright (c) 2020 PANTHEON.tech, s.r.o. and others.  All rights reserved.
 *
 * This program and the accompanying materials are made available under the
 * terms of the Eclipse Public License v1.0 which accompanies this distribution,
 * and is available at http://www.eclipse.org/legal/epl-v10.html
 */
package org.opendaylight.yangtools.yang.ir;

import static java.util.Objects.requireNonNull;

import com.google.common.annotations.Beta;
import com.google.common.collect.ImmutableList;
import java.util.List;
import org.eclipse.jdt.annotation.NonNull;

/**
 * An argument to a YANG statement, as defined by section 6.1.3 of both
 * RFC6020 and
 * RFC7950. An argument is effectively any old string,
 * except it can be defined in a number of ways:
 * 
    *
  • it can be a simple unquoted string, or
  • *
  • it can be a single-quoted string, with its contents being completely preserved, or
  • *
  • it can be a double-quoted string, which defines some escaping and whitespace-stripping rules, or
  • *
  • it can be a concatenation of any number of single- or double-quoted strings
  • *
* *

* The first three cases as covered by {@link Single} subclass, which exposes appropriate methods to infer how its * string literal is to be interpreted. The last case is handled by {@link Concatenation} subclass, which exposes * the constituent parts as {@link Single} items. * *

* Please note that parser implementations producing these argument representations are NOT required to retain * the format of the original definition. They are free to perform quoting and concatenation transformations as long as * they maintain semantic equivalence. As a matter of example, these transformations are explicitly allowed: *

    *
  • elimination of unneeded quotes, for example turning {@code "foo"} into {@code foo}
  • *
  • transformation of quotes, for example turning {@code "foo\nbar"} into {@code 'foo bar'}
  • *
  • concatenation processing, for example turning {@code 'foo' + 'bar'} into {@code foobar}
  • *
*/ @Beta public abstract sealed class IRArgument extends AbstractIRObject { /** * An argument composed of multiple concatenated parts. */ public static final class Concatenation extends IRArgument { private final @NonNull ImmutableList parts; private Concatenation(final List parts) { this.parts = ImmutableList.copyOf(parts); } /** * Return the argument parts that need to be concatenated. * * @return Argument parts. */ public @NonNull List parts() { return parts; } @Override public int hashCode() { return parts.hashCode(); } @Override public boolean equals(final Object obj) { return this == obj || obj instanceof Concatenation other && parts.equals(other.parts); } @Override StringBuilder toYangFragment(final StringBuilder sb) { final var it = parts.iterator(); it.next().toYangFragment(sb); while (it.hasNext()) { it.next().toYangFragment(sb.append(" + ")); } return sb; } } /** * An argument composed of a single string. This string may need further validation and processing, as it may not * actually conform to the specification as requested by {@code yang-version}. */ /* * This is the public footprint which is served by three final subclasses: DoublyQuoted, SingleQuoted, Unquoted. * Those classes must never be exposed, as they are a manifestation of current implementation in StatementFactory. * As noted in the interface contract of IRArgument, we have very much free reign on syntactic transformations, * StatementFactory is just not taking advantage of those at this point. * * The subclasses may very much change, in terms of both naming and function, to support whatever StatementFactory * ends up doing. */ public abstract static sealed class Single extends IRArgument { private final @NonNull String string; private Single(final String string) { this.string = requireNonNull(string); } /** * Significant portion of this argument. For unquoted and single-quoted strings this is the unquoted string * literal. For double-quoted strings this is the unquoted string, after whitespace trimming as defined by * RFC6020/RFC7950 section 6.1.3, but before escape substitution. * * @return Significant portion of this argument. */ public final @NonNull String string() { return string; } /** * Imprecise check if this argument needs further unescape operation (which is version-specific) to arrive at * the literal string value. This is false for unquoted and single-quoted strings, which do not support any sort * of escaping. This may be true for double-quoted strings, as they may need to be further processed in * version-dependent ways to arrive at the correct literal value. * *

* This method is allowed to err on the false-positive side -- i.e. it may report any double-quoted string as * needing further processing, even when the actual content could be determined to not need further processing. * * @return False if the value of {@link #string} can be used as-is. */ public final boolean needUnescape() { return this instanceof DoubleQuoted; } /** * Imprecise check if this argument needs an additional content check for compliance. This is false if the * string was explicitly quoted and therefore cannot contain stray single- or double-quotes, or if the content * has already been checked to not contain them. * *

* The content check is needed to ascertain RFC7950 compliance, because RFC6020 allows constructs like *

abc"def
in unquoted strings, while RFC7950 explicitly forbids them. * *

* This method is allowed to err on the false-positive side -- i.e. it may report any unquoted string as * needing this check, even when the actual content could be determined to not contain quotes. * * @return True if this argument requires a version-specific check for quote content. */ public final boolean needQuoteCheck() { return this instanceof Unquoted; } /** * Imprecise check if this argument complies with the {@code identifier} YANG specification. * *

* This method is allowed to err on the false-negative side -- i.e. it may report any string as not being * compliant with {@code identifier}, even when the actual content could be determined to be compliant. * * @return True if this argument is known to be directly usable in contexts where YANG requires the use of */ public final boolean isValidIdentifier() { return this instanceof Identifier; } @Override public final int hashCode() { return string.hashCode(); } @Override public final boolean equals(final Object obj) { return this == obj || obj != null && getClass().equals(obj.getClass()) && string.equals(((Single) obj).string); } @Override StringBuilder toYangFragment(final StringBuilder sb) { return sb.append(string); } } private static final class DoubleQuoted extends Single { private DoubleQuoted(final String string) { super(string); } @Override StringBuilder toYangFragment(final StringBuilder sb) { // Note this is just an approximation. We do not have enough state knowledge to restore any whitespace we // may have trimmed. return super.toYangFragment(sb.append('"')).append('"'); } } private static final class SingleQuoted extends Single { static final @NonNull SingleQuoted EMPTY = new SingleQuoted(""); private SingleQuoted(final String string) { super(string); } @Override StringBuilder toYangFragment(final StringBuilder sb) { return super.toYangFragment(sb.append('\'')).append('\''); } } private static final class Identifier extends Single { private Identifier(final String string) { super(string); } } private static final class Unquoted extends Single { private Unquoted(final String string) { super(string); } } private IRArgument() { // Hidden on purpose } public static @NonNull Single empty() { return SingleQuoted.EMPTY; } public static @NonNull Single identifier(final String string) { return new Identifier(string); } public static @NonNull Single singleQuoted(final String string) { return new SingleQuoted(string); } public static @NonNull Single doubleQuoted(final String string) { return new DoubleQuoted(string); } public static @NonNull Single unquoted(final String string) { return new Unquoted(string); } public static @NonNull IRArgument of(final List parts) { return switch (parts.size()) { // A concatenation of empty strings, fall back to a single unquoted string case 0 -> empty(); // A single string concatenated with empty string(s), use just the significant portion case 1 -> parts.get(0); // TODO: perform concatenation of single-quoted strings. For double-quoted strings this may not be as nice, // but for single-quoted strings we do not need further validation in in the reactor and can use them // as raw literals. This saves some indirection overhead (on memory side) and can slightly improve // execution speed when we process the same IR multiple times. default -> new Concatenation(parts); }; } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy