Adds apostrophe entity selection

Author	DaveJarvis <email>
Date	2024-10-09 16:58:44 GMT-0700
Commit	1fae3e3e90038c3b4032e60b0e2d980dcf590854
Parent	a94a338

build.gradle

		}

		-tasks.withType( JavaCompile ).configureEach {
		- options.compilerArgs += "--enable-preview"
		-}
		-
		-tasks.withType( JavaExec ).configureEach {
		- jvmArgs += "--enable-preview"
		-}
		-
		-tasks.withType( Test ).configureEach {
		- jvmArgs += "--enable-preview"
		-}
		-

src/main/java/com/whitemagicsoftware/keenquotes/app/KeenQuotes.java

		-/* Copyright 2021 White Magic Software, Ltd. -- All rights reserved. */
		+/* Copyright 2021-2024 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: BSD-2-Clause
		+ */
		package com.whitemagicsoftware.keenquotes.app;


		contractions,
		settings.filterXml() ? FILTER_XML : FILTER_PLAIN,
		- settings.entities()
		+ settings.apostrophe()
		);

src/main/java/com/whitemagicsoftware/keenquotes/app/Settings.java

		-/* Copyright 2021 White Magic Software, Ltd. -- All rights reserved. */
		+/* Copyright 2021-2024 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: BSD-2-Clause
		+ */
		package com.whitemagicsoftware.keenquotes.app;

		import com.whitemagicsoftware.keenquotes.lex.XmlFilter;
		+import com.whitemagicsoftware.keenquotes.parser.Apostrophe;
		import picocli.CommandLine;

		import java.util.List;
		import java.util.concurrent.Callable;

		+import static com.whitemagicsoftware.keenquotes.parser.Apostrophe.fromType;
		import static java.util.Arrays.asList;
		import static java.util.Collections.emptyList;


		/**
		- * Encode quotation marks using HTML entities.
		+ * Set the type of entity to use when encoding apostrophes.
		*/
		@CommandLine.Option(
		- names = {"-e", "--entities"},
		- description = "Encode quotation marks using HTML entities"
		+ names = {"-a", "--apostrophe"},
		+ description = "Converted apostrophe entity (regular, modifier, hex, entity)"
		)
		- private boolean mEntities;
		+ private String mApostrophe;

		/**

		* @return {@code true} to encode quotation marks using HTML entities.
		*/
		- boolean entities() { return mEntities; }
		+ Apostrophe apostrophe() { return fromType( mApostrophe ); }

		List<String> getBeganUnambiguous() {

src/main/java/com/whitemagicsoftware/keenquotes/parser/Apostrophe.java

		+/* Copyright 2024 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: BSD-2-Clause
		+ */
		+package com.whitemagicsoftware.keenquotes.parser;
		+
		+/**
		+ * When converting quotation marks, these values are used to indicate what
		+ * type of entity to use in the conversion.
		+ */
		+public enum Apostrophe {
		+ /**
		+ * No conversion is performed.
		+ */
		+ CONVERT_REGULAR( "'", "regular" ),
		+ /**
		+ * Apostrophes become MODIFIER LETTER APOSTROPHE ({@code ʼ}).
		+ */
		+ CONVERT_MODIFIER( "ʼ", "modifier" ),
		+ /**
		+ * Apostrophes become APOSTROPHE ({@code '}).
		+ */
		+ CONVERT_APOS_HEX( "'", "hex" ),
		+ /**
		+ * Apostrophes become XML APOSTROPHE ({@code '}).
		+ */
		+ CONVERT_APOS_ENTITY( "'", "entity" );
		+
		+ private final String mCode;
		+ private final String mType;
		+
		+ Apostrophe( final String code, final String type ) {
		+ mCode = code;
		+ mType = type;
		+ }
		+
		+ public boolean isType( final String type ) {
		+ return mType.equalsIgnoreCase( type );
		+ }
		+
		+ public String toString() {
		+ return mCode;
		+ }
		+
		+ /**
		+ * Returns the instance that matches the given type.
		+ *
		+ * @param type The type of apostrophe entity conversion to use.
		+ * @return The {@link Apostrophe} to use when converting entities.
		+ */
		+ public static Apostrophe fromType( final String type ) {
		+ for( final var apostrophe : Apostrophe.values() ) {
		+ if( apostrophe.isType( type ) ) {
		+ return apostrophe;
		+ }
		+ }
		+
		+ return Apostrophe.CONVERT_REGULAR;
		+ }
		+}

src/main/java/com/whitemagicsoftware/keenquotes/parser/Contractions.java

		-/* Copyright 2021 White Magic Software, Ltd. -- All rights reserved. */
		+/* Copyright 2021-2024 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: BSD-2-Clause
		+ */
		package com.whitemagicsoftware.keenquotes.parser;


		* Placeholder for various types of contractions.
		*/
		-@SuppressWarnings( { "SpellCheckingInspection", "GrazieInspection" } )
		+@SuppressWarnings( "SpellCheckingInspection" )
		public class Contractions {


		// Ensure that 'n' isn't matched for ambiguity by enforcing length, yet
		// allow o' to match because 'a sentence can end with the letter o'.
		- return getEndedAmbiguous().contains( check ) \|\|
		- check.endsWith( "s" ) \|\| check.endsWith( "z" ) \|\|
		- check.endsWith( "x" ) \|\| (check.length() > 1 && check.endsWith( "n" ));
		+ return
		+ getEndedAmbiguous().contains( check ) \|\|
		+ check.endsWith( "s" ) \|\|
		+ check.endsWith( "z" ) \|\|
		+ check.endsWith( "x" ) \|\|
		+ (check.length() > 1 && check.endsWith( "n" ));
		}


		return
		toString( getBeganEndedUnambiguous(), "Unambiguous Began/Ended", "'%s" ) +
		- toString( getBeganUnambiguous(), "Unambiguous Began", "'%s" ) +
		- toString( getEndedUnambiguous(), "Unambiguous Ended", "%s'" ) +
		- toString( getBeganAmbiguous(), "Ambiguous Began", "'%s" ) +
		- toString( getEndedAmbiguous(), "Ambiguous Ended", "%s'" );
		+ toString( getBeganUnambiguous(), "Unambiguous Began", "'%s" ) +
		+ toString( getEndedUnambiguous(), "Unambiguous Ended", "%s'" ) +
		+ toString( getBeganAmbiguous(), "Ambiguous Began", "'%s" ) +
		+ toString( getEndedAmbiguous(), "Ambiguous Ended", "%s'" );
		}

src/main/java/com/whitemagicsoftware/keenquotes/parser/Curler.java

		private final Contractions mContractions;
		private final LexerFilter mFilter;
		- private final boolean mEntities;
		+ private final Apostrophe mApostrophe;

		/**
		* Maps quotes to curled character equivalents.
		*
		- * @param c Contractions listings.
		- * @param entities {@code true} to convert quotation marks to HTML entities.
		+ * @param c Contractions listings.
		+ * @param apostrophe How to convert quotation marks to HTML entities.
		*/
		public Curler(
		final Contractions c,
		final FilterType filterType,
		- final boolean entities
		+ final Apostrophe apostrophe
		) {
		assert c != null;

		mContractions = c;
		- mEntities = entities;
		mFilter = filterType.filter();
		+ mApostrophe = apostrophe;
		}


		text,
		mContractions,
		- replace( output, offset, mEntities ),
		+ replace( output, offset, mApostrophe ),
		mFilter
		);

		return output.toString();
		}

		/**
		* Replaces non-ambiguous tokens with their equivalent string representation.
		*
		- * @param output Continuously updated result document.
		- * @param offset Accumulating index where {@link Token} is replaced.
		- * @param entities {@code true} to convert quotation marks to HTML entities.
		+ * @param output Continuously updated result document.
		+ * @param offset Accumulating index where {@link Token} is replaced.
		+ * @param apostrophe How to convert quotation marks to HTML entities.
		* @return Instructions to replace a {@link Token} in the result document.
		*/
		public static Consumer<Token> replace(
		final StringBuilder output,
		final AtomicInteger offset,
		- final boolean entities
		+ final Apostrophe apostrophe
		) {
		return token -> {
		if( !token.isAmbiguous() ) {
		- final var text = token.toString( entities );
		+ final var text = token.toString( apostrophe );

		output.replace(

src/main/java/com/whitemagicsoftware/keenquotes/parser/Token.java

		-/* Copyright 2021 White Magic Software, Ltd. -- All rights reserved. */
		+/* Copyright 2021-2024 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: BSD-2-Clause
		+ */
		package com.whitemagicsoftware.keenquotes.parser;

		import com.whitemagicsoftware.keenquotes.lex.Lexeme;
		import com.whitemagicsoftware.keenquotes.lex.LexemeGlyph;

		import java.util.EnumMap;
		import java.util.Map;

		import static com.whitemagicsoftware.keenquotes.lex.LexemeGlyph.*;
		+import static com.whitemagicsoftware.keenquotes.parser.Apostrophe.*;
		import static com.whitemagicsoftware.keenquotes.parser.TokenType.*;


		ENTITIES.put( QUOTE_STRAIGHT_SINGLE, "'" );
		ENTITIES.put( QUOTE_STRAIGHT_DOUBLE, "\"" );
		- ENTITIES.put( QUOTE_APOSTROPHE, "'" );
		ENTITIES.put( QUOTE_PRIME_SINGLE, "′" );
		ENTITIES.put( QUOTE_PRIME_DOUBLE, "″" );

		boolean isAmbiguous() {
		return mTokenType == QUOTE_AMBIGUOUS_SINGLE \|\|
		- mTokenType == QUOTE_AMBIGUOUS_DOUBLE \|\|
		- mTokenType == QUOTE_AMBIGUOUS_LEADING \|\|
		- mTokenType == QUOTE_AMBIGUOUS_LAGGING;
		+ mTokenType == QUOTE_AMBIGUOUS_DOUBLE \|\|
		+ mTokenType == QUOTE_AMBIGUOUS_LEADING \|\|
		+ mTokenType == QUOTE_AMBIGUOUS_LAGGING;
		}


		public String toXml() {
		return "<" +
		- mTokenType +
		- " type='" + getType().name() + "'" +
		- " began='" + began() + "'" +
		- " ended='" + ended() + "' />";
		+ mTokenType +
		+ " type='" + getType().name() + "'" +
		+ " began='" + began() + "'" +
		+ " ended='" + ended() + "' />";
		}

		/**
		* Converts this token to its string representation, which will either be
		* an HTML entity or a character.
		*
		- * @param entities {@code true} to convert quotation marks to HTML entities.
		+ * @param apostrophe How to convert quotation marks to HTML entities.
		* @return A plain quotation mark character or an HTML entity.
		*/
		- public String toString( final boolean entities ) {
		+ public String toString( final Apostrophe apostrophe ) {
		final var glyph = mLexeme.getType().glyph();
		+ final var tokenType = getType();

		- return entities
		- ? I18N_ENTITIES.getOrDefault( glyph, ENTITIES.get( getType() ) )
		- : CHARS.getOrDefault( getType(), glyph.text() );
		+ // Retrieves the base glyph, unless curling was requested. When curling
		+ // apostrophes, this will determine the user-selected type of apostrophe
		+ // entity to use.
		+ return apostrophe == CONVERT_REGULAR
		+ ? CHARS.getOrDefault( tokenType, glyph.text() )
		+ : I18N_ENTITIES.getOrDefault( glyph, convert( tokenType, apostrophe ) );
		+ }
		+
		+ private String convert(
		+ final TokenType tokenType,
		+ final Apostrophe apostrophe ) {
		+ return tokenType == QUOTE_APOSTROPHE
		+ ? apostrophe.toString()
		+ : ENTITIES.get( tokenType );
		}

		@Override
		public String toString() {
		return getClass().getSimpleName() + '[' +
		- "mType=" + getType() +
		- ", mBegan=" + began() +
		- ", mEnded=" + ended() +
		- ']';
		+ "mType=" + getType() +
		+ ", mBegan=" + began() +
		+ ", mEnded=" + ended() +
		+ ']';
		}
		}

src/test/java/com/whitemagicsoftware/keenquotes/parser/AmbiguityResolverTest.java

		-/* Copyright 2022 White Magic Software, Ltd. -- All rights reserved.
		+/* Copyright 2022-2024 White Magic Software, Ltd. -- All rights reserved.
		*
		* SPDX-License-Identifier: BSD-2-Clause

		import java.util.concurrent.atomic.AtomicInteger;

		+import static com.whitemagicsoftware.keenquotes.parser.Apostrophe.*;
		import static com.whitemagicsoftware.keenquotes.parser.Curler.*;
		import static com.whitemagicsoftware.keenquotes.texts.TestResource.readPairs;

		input,
		CONTRACTIONS,
		- replace( output, offset, true ),
		- filter -> false
		+ replace( output, offset, CONVERT_APOS_ENTITY ),
		+ _ -> false
		);

src/test/java/com/whitemagicsoftware/keenquotes/parser/CurlerTest.java

		-/* Copyright 2022 White Magic Software, Ltd. -- All rights reserved.
		+/* Copyright 2022-2024 White Magic Software, Ltd. -- All rights reserved.
		*
		* SPDX-License-Identifier: BSD-2-Clause

		import static com.whitemagicsoftware.keenquotes.lex.FilterType.FILTER_PLAIN;
		import static com.whitemagicsoftware.keenquotes.lex.FilterType.FILTER_XML;
		+import static com.whitemagicsoftware.keenquotes.parser.Apostrophe.*;
		import static com.whitemagicsoftware.keenquotes.texts.TestResource.open;
		import static com.whitemagicsoftware.keenquotes.texts.TestResource.readPairs;

		@Test
		public void test_Parse_UncurledQuotes1_CurlyQuotes() throws IOException {
		- testCurler( createCurler( FILTER_PLAIN, true ), "unambiguous-1-pass.txt" );
		+ testCurler(
		+ createCurler( FILTER_PLAIN, CONVERT_APOS_ENTITY ),
		+ "unambiguous-1-pass.txt"
		+ );
		}

		@Test
		public void test_Parse_UncurledQuotes2_CurlyQuotes() throws IOException {
		- testCurler( createCurler( FILTER_PLAIN, true ), "unambiguous-2-pass.txt" );
		+ testCurler(
		+ createCurler( FILTER_PLAIN, CONVERT_APOS_ENTITY ),
		+ "unambiguous-2-pass.txt"
		+ );
		}

		@Disabled
		@SuppressWarnings( "unused" )
		public void test_Parse_AmbiguousQuotes_PartiallyCurled() throws IOException {
		- testCurler( createCurler( FILTER_PLAIN, false ), "ambiguous-n-pass.txt" );
		+ testCurler(
		+ createCurler( FILTER_PLAIN, CONVERT_REGULAR ), "ambiguous-n-pass.txt"
		+ );
		}

		@Test
		public void test_Parse_UncurledQuotesXml_CurlyQuotes() throws IOException {
		- testCurler( createCurler( FILTER_XML, true ), "xml.txt" );
		+ testCurler(
		+ createCurler( FILTER_XML, CONVERT_APOS_ENTITY ), "xml.txt"
		+ );
		}

		@Test
		public void test_Parse_UncurledQuotesI11l_CurlyQuotes() throws IOException {
		- testCurler( createCurler( FILTER_PLAIN, true ), "i18n.txt" );
		+ testCurler(
		+ createCurler( FILTER_PLAIN, CONVERT_APOS_ENTITY ), "i18n.txt"
		+ );
		}


		*/
		@ParameterizedTest
		- @ValueSource( strings = {"autonoma"} )
		+ @ValueSource( strings = { "autonoma" } )
		@Disabled
		void test_Parse_Story_Converted( final String filename ) throws IOException {
		final var sb = new StringBuilder( 2 ^ 20 );
		+ final var name = String.format( "%s%s", filename, ".html" );

		- try( final var reader = open( STR."\{filename}.html" ) ) {
		+ try( final var reader = open( name ) ) {
		String line;

		while( (line = reader.readLine()) != null ) {
		sb.append( line ).append( SEP );
		}
		}

		- final var curler = createCurler( FILTER_XML, true );
		+ final var curler = createCurler( FILTER_XML, CONVERT_APOS_ENTITY );
		System.out.println( curler.apply( sb.toString() ) );
		}

		private Function<String, String> createCurler(
		final FilterType filterType,
		- final boolean entities ) {
		- return new Curler( createContractions(), filterType, entities );
		+ final Apostrophe apostrophe ) {
		+ return new Curler( createContractions(), filterType, apostrophe );
		}

src/test/java/com/whitemagicsoftware/keenquotes/parser/QuoteEmitterTest.java

		-/* Copyright 2022 White Magic Software, Ltd. -- All rights reserved.
		+/* Copyright 2022-2024 White Magic Software, Ltd. -- All rights reserved.
		*
		* SPDX-License-Identifier: BSD-2-Clause

		import java.util.concurrent.atomic.AtomicInteger;

		+import static com.whitemagicsoftware.keenquotes.parser.Apostrophe.CONVERT_APOS_ENTITY;
		import static com.whitemagicsoftware.keenquotes.parser.Curler.replace;
		import static com.whitemagicsoftware.keenquotes.texts.TestResource.readPairs;

		input,
		CONTRACTIONS,
		- replace( output, offset, true ),
		- filter -> false
		+ replace( output, offset, CONVERT_APOS_ENTITY ),
		+ _ -> false
		);

src/test/resources/com/whitemagicsoftware/keenquotes/texts/unambiguous-1-pass.txt

		Computer says, “‘It is mysteries---’”

		-He goes 'long with it.
		-He goes 'long with it.
		-
		-The 'long and short' of it.
		-The ‘long and short’ of it.
		-

src/test/resources/com/whitemagicsoftware/keenquotes/texts/unambiguous-2-pass.txt

		'Twas and 'tis whate'er lay 'twixt dawn and dusk 'n River Styx.

		+'He goes 'long with it.'
		+‘He goes 'long with it.’
		+
		# ########################################################################
		# Possessives

Delta	171 lines added, 81 lines removed, 90-line increase