Dave Jarvis' Repositories

Home Files Commits Tags Clone

Repository:

git clone https://repo.autonoma.ca/repo/keenwrite.git

Adds unit test, isolates xref parsing

Author	DaveJarvis <email>
Date	2023-11-20 23:13:29 GMT-0800
Commit	c334bfecbc671ec3443ad04b318e9b8c4f315617
Parent	b7c3eb9
Delta	291 lines added, 158 lines removed, 133-line increase

src/test/java/com/keenwrite/processors/markdown/extensions/references/CrossReferencesExtensionTest.java

		import com.vladsch.flexmark.html.HtmlRenderer;
		import com.vladsch.flexmark.parser.Parser;
		-import org.junit.jupiter.api.BeforeEach;
		-import org.junit.jupiter.api.Test;
		+import org.junit.jupiter.params.ParameterizedTest;
		+import org.junit.jupiter.params.provider.Arguments;
		+import org.junit.jupiter.params.provider.MethodSource;

		import java.util.List;
		+import java.util.stream.Stream;

		import static org.junit.jupiter.api.Assertions.assertEquals;

		+@SuppressWarnings( "SpellCheckingInspection" )
		public class CrossReferencesExtensionTest {
		- private Parser mParser;
		- private HtmlRenderer mRenderer;
		-
		- @BeforeEach
		- public void setup() {
		+ @ParameterizedTest
		+ @MethodSource( "testDocuments" )
		+ public void test_References_Documents_Html(
		+ final String input,
		+ final String expected ) {
		final var extension = new CrossReferenceExtension();
		final var pBuilder = Parser.builder();
		final var hBuilder = HtmlRenderer.builder();
		final var extensions = List.of( extension );

		pBuilder.extensions( extensions );
		hBuilder.extensions( extensions );
		-
		- mParser = pBuilder.build();
		- mRenderer = hBuilder.build();
		- }

		- @Test
		- public void test_References_SingularReferences_ReferencesAndPointers() {
		- final var document = mParser.parse(
		- """
		- {#fig:cats} [@fig:cats]
		- {#table:dogs} [@table:dogs]
		- {#life:dolphins} [@life:dolphins]
		- """
		- );
		+ final var parser = pBuilder.build();
		+ final var renderer = hBuilder.build();

		- final String expected =
		- """
		- <p><a data-type="fig" name="cats" /> <a data-type="fig" href="#cats" />
		- <a data-type="table" name="dogs" /> <a data-type="table" href="#dogs" />
		- <a data-type="life" name="dolphins" /> <a data-type="life" href="#dolphins" /></p>
		- """;
		- final var actual = mRenderer.render( document );
		+ final var document = parser.parse( input );
		+ final var actual = renderer.render( document );

		assertEquals( expected, actual );
		}

		- @Test
		- @SuppressWarnings( "SpellCheckingInspection" )
		- public void test_References_DocumentReferences_ReferencesAndPointers() {
		- final var document = mParser.parse(
		- """
		- Lorem ipsum dolor sit amet, consectetur adipiscing elit.
		- {#fig:cats} Sed do eiusmod tempor incididunt ut
		- labore et dolore magna aliqua. Ut enim ad minim veniam,
		- quis nostrud exercitation ullamco laboris nisi ut aliquip
		- ex ea commodo consequat. [@fig:cats]
		+ private static Stream<Arguments> testDocuments() {
		+ return Stream.of(
		+ Arguments.of(
		+ """
		+ {#fig:cats} [@fig:cats]
		+ {#table:dogs} [@table:dogs]
		+ {#life:dolphins} [@life:dolphins]
		+ """,
		+ """
		+ <p><a data-type="fig" name="cats" /> <a data-type="fig" href="#cats" />
		+ <a data-type="table" name="dogs" /> <a data-type="table" href="#dogs" />
		+ <a data-type="life" name="dolphins" /> <a data-type="life" href="#dolphins" /></p>
		+ """ ),
		+ Arguments.of(
		+ """
		+ Lorem ipsum dolor sit amet, consectetur adipiscing elit.
		+ {#fig:cats} Sed do eiusmod tempor incididunt ut
		+ labore et dolore magna aliqua. Ut enim ad minim veniam,
		+ quis nostrud exercitation ullamco laboris nisi ut aliquip
		+ ex ea commodo consequat. [@fig:cats]
		+ """,
		+ """
		+ <p>Lorem ipsum dolor sit amet, consectetur adipiscing elit.
		+ <a data-type="fig" name="cats" /> Sed do eiusmod tempor incididunt ut
		+ labore et dolore magna aliqua. Ut enim ad minim veniam,
		+ quis nostrud exercitation ullamco laboris nisi ut aliquip
		+ ex ea commodo consequat. <a data-type="fig" href="#cats" /></p>
		+ """
		+ ),
		+ Arguments.of(
		+ """
		+ {#日本:w0mbatß}
		+ [@日本:w0mbatß]
		+ [@日本:w0mbatß]
		+ [@日本:w0mbatß]
		+ """,
		+ """
		+ <p><a data-type="日本" name="w0mbatß" />
		+ <a data-type="日本" href="#w0mbatß" />
		+ <a data-type="日本" href="#w0mbatß" />
		+ <a data-type="日本" href="#w0mbatß" /></p>
		+ """
		+ ),
		+ Arguments.of(
		+ """
		+ {#note:advancement}
		+
		+ Advancement isn't measured by the ingenuity of inventions, but
		+ by humanity's ability to anticipate and forfend dire aftermaths
		+ before using them.
		+
		+ [@note:advancement]
		+
		+ To what end?
		+ """,
		"""
		+ <p><a data-type="note" name="advancement" /></p>
		+ <p>Advancement isn't measured by the ingenuity of inventions, but
		+ by humanity's ability to anticipate and forfend dire aftermaths
		+ <em>before</em> using them.</p>
		+ <p><a data-type="note" href="#advancement" /></p>
		+ <p>To what end?</p>
		+ """
		+ )
		);
		-
		- final var expected =
		- """
		- <p>Lorem ipsum dolor sit amet, consectetur adipiscing elit.
		- <a data-type="fig" name="cats" /> Sed do eiusmod tempor incididunt ut
		- labore et dolore magna aliqua. Ut enim ad minim veniam,
		- quis nostrud exercitation ullamco laboris nisi ut aliquip
		- ex ea commodo consequat. <a data-type="fig" href="#cats" /></p>
		- """;
		-
		- final var actual = mRenderer.render( document );
		-
		- assertEquals( expected, actual );
		}
		}

src/main/java/com/keenwrite/processors/markdown/extensions/references/AnchorNameNode.java


		import com.vladsch.flexmark.ast.DelimitedNodeImpl;
		-import com.vladsch.flexmark.parser.core.delimiter.Delimiter;
		import com.vladsch.flexmark.util.sequence.BasedSequence;
		import org.jetbrains.annotations.NotNull;
		-
		-import java.util.regex.Pattern;
		-
		-import static java.util.regex.Pattern.compile;

		/**
		* Responsible for writing HTML anchor names in the form
		* {@code <a data-type="..." name="name" />}, where {@code name} can be
		* referred to by a cross-reference.
		*
		* @see AnchorXrefNode
		*/
		class AnchorNameNode extends DelimitedNodeImpl implements CrossReferenceNode {
		- private static final String REGEX_ANCHOR = "#(\\w+):(\\w+)";
		- private static final Pattern PATTERN_ANCHOR = compile( REGEX_ANCHOR );

		private BasedSequence mOpeningMarker = BasedSequence.EMPTY;
		private BasedSequence mClosingMarker = BasedSequence.EMPTY;

		- private String mTypeName = "";
		- private String mIdName = "";
		+ private BasedSequenceNameParser mParser;

		- public AnchorNameNode( final Delimiter opener, final Delimiter closer ) {}
		+ public AnchorNameNode() {}

		@Override
		public String getTypeName() {
		- return mTypeName;
		+ return mParser.getTypeName();
		}

		@Override
		public String getIdName() {
		- return mIdName;
		+ return mParser.getIdName();
		}


		@Override
		public void setText( final BasedSequence text ) {
		- final var matcher = PATTERN_ANCHOR.matcher( text.toString() );
		-
		- if( matcher.find() ) {
		- mTypeName = matcher.group( 1 );
		- mIdName = matcher.group( 2 );
		- }
		+ mParser = BasedSequenceNameParser.parse( text );
		}

src/main/java/com/keenwrite/processors/markdown/extensions/references/AnchorXrefInlineParser.java

		-/* Copyright 2023 White Magic Software, Ltd. -- All rights reserved.
		- *
		- * SPDX-License-Identifier: MIT
		- */
		-package com.keenwrite.processors.markdown.extensions.references;
		-
		-import com.vladsch.flexmark.parser.InlineParserExtensionFactory;
		-import com.vladsch.flexmark.parser.delimiter.DelimiterProcessor;
		-import com.vladsch.flexmark.parser.internal.InlineParserImpl;
		-import com.vladsch.flexmark.parser.internal.LinkRefProcessorData;
		-import com.vladsch.flexmark.util.data.DataHolder;
		-
		-import java.util.BitSet;
		-import java.util.List;
		-import java.util.Map;
		-import java.util.regex.Pattern;
		-
		-import static java.util.regex.Pattern.compile;
		-
		-final class AnchorXrefInlineParser extends InlineParserImpl {
		-
		- private static final String REGEX_REFERENCE = "@(\\w+):(\\w+)";
		- private static final Pattern PATTERN_REFERENCE = compile( REGEX_REFERENCE );
		-
		- private boolean mFoundReference;
		-
		- AnchorXrefInlineParser(
		- final DataHolder options,
		- final BitSet specialCharacters,
		- final BitSet delimiterCharacters,
		- final Map<Character, DelimiterProcessor> delimiterProcessors,
		- final LinkRefProcessorData referenceLinkProcessors,
		- final List<InlineParserExtensionFactory> inlineParserExtensions ) {
		- super(
		- options,
		- specialCharacters,
		- delimiterCharacters,
		- delimiterProcessors,
		- referenceLinkProcessors,
		- inlineParserExtensions
		- );
		- }
		-
		- @Override
		- public boolean parseOpenBracket() {
		- mFoundReference = peek( 1 ) == '@';
		-
		- return super.parseOpenBracket();
		- }
		-
		- @Override
		- public boolean parseCloseBracket() {
		- final boolean foundBracket = super.parseCloseBracket();
		-
		- if( mFoundReference ) {
		- final var blockNode = getBlock();
		- final var linkRef = blockNode.getLastChild();
		-
		- if( linkRef != null ) {
		- final var text = linkRef.getChildChars();
		- final var matcher = PATTERN_REFERENCE.matcher( text );
		-
		- if( matcher.find() ) {
		- final var typeName = matcher.group( 1 );
		- final var idName = matcher.group( 2 );
		- final var xref = new AnchorXrefNode( typeName, idName );
		-
		- linkRef.unlink();
		- blockNode.appendChild( xref );
		- }
		- }
		- }
		-
		- return foundBracket;
		- }
		-}

src/main/java/com/keenwrite/processors/markdown/extensions/references/AnchorXrefProcessorFactory.java

		+/* Copyright 2023 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: MIT
		+ */
		+package com.keenwrite.processors.markdown.extensions.references;
		+
		+import com.vladsch.flexmark.parser.LinkRefProcessor;
		+import com.vladsch.flexmark.parser.LinkRefProcessorFactory;
		+import com.vladsch.flexmark.util.ast.Document;
		+import com.vladsch.flexmark.util.ast.Node;
		+import com.vladsch.flexmark.util.data.DataHolder;
		+import com.vladsch.flexmark.util.sequence.BasedSequence;
		+import org.jetbrains.annotations.NotNull;
		+
		+/**
		+ * Responsible for processing {@code [@type:id]} anchors.
		+ */
		+class AnchorXrefProcessorFactory implements LinkRefProcessorFactory {
		+ private final LinkRefProcessor mProcessor = new AnchorLinkRefProcessor();
		+
		+ @Override
		+ public boolean getWantExclamationPrefix( @NotNull final DataHolder options ) {
		+ return false;
		+ }
		+
		+ @Override
		+ public int getBracketNestingLevel( @NotNull final DataHolder options ) {
		+ return 0;
		+ }
		+
		+ @NotNull
		+ @Override
		+ public LinkRefProcessor apply( @NotNull final Document document ) {
		+ return mProcessor;
		+ }
		+
		+ private static class AnchorLinkRefProcessor implements LinkRefProcessor {
		+
		+ @Override
		+ public boolean getWantExclamationPrefix() {
		+ return false;
		+ }
		+
		+ @Override
		+ public int getBracketNestingLevel() {
		+ return 0;
		+ }
		+
		+ @Override
		+ public boolean isMatch( @NotNull final BasedSequence nodeChars ) {
		+ return nodeChars.indexOf( '@' ) == 1;
		+ }
		+
		+ @NotNull
		+ @Override
		+ public Node createNode( @NotNull final BasedSequence nodeChars ) {
		+ return BasedSequenceXrefParser.parse( nodeChars ).toNode();
		+ }
		+
		+ @NotNull
		+ @Override
		+ public BasedSequence adjustInlineText(
		+ @NotNull final Document document,
		+ @NotNull final Node node ) {
		+ return BasedSequence.EMPTY;
		+ }
		+
		+ @Override
		+ public boolean allowDelimiters(
		+ @NotNull final BasedSequence chars,
		+ @NotNull final Document document,
		+ @NotNull final Node node ) {
		+ return false;
		+ }
		+
		+ @Override
		+ public void updateNodeElements(
		+ @NotNull final Document document,
		+ @NotNull final Node node ) {}
		+ }
		+}

src/main/java/com/keenwrite/processors/markdown/extensions/references/BasedSequenceNameParser.java

		+/* Copyright 2023 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: MIT
		+ */
		+package com.keenwrite.processors.markdown.extensions.references;
		+
		+import com.vladsch.flexmark.util.sequence.BasedSequence;
		+
		+import java.util.regex.Matcher;
		+import java.util.regex.Pattern;
		+
		+import static java.util.regex.Pattern.UNICODE_CHARACTER_CLASS;
		+import static java.util.regex.Pattern.compile;
		+
		+class BasedSequenceNameParser extends BasedSequenceParser {
		+ private static final String REGEX = STR. "#\{ REGEX_INNER }" ;
		+ private static final Pattern PATTERN = compile(
		+ REGEX, UNICODE_CHARACTER_CLASS
		+ );
		+
		+ private BasedSequenceNameParser( final String text ) {
		+ super( text );
		+ }
		+
		+ @Override
		+ Matcher createMatcher( final String text ) {
		+ return PATTERN.matcher( text );
		+ }
		+
		+ static BasedSequenceNameParser parse( final BasedSequence chars ) {
		+ return new BasedSequenceNameParser( chars.toString() );
		+ }
		+}

src/main/java/com/keenwrite/processors/markdown/extensions/references/BasedSequenceParser.java

		+/* Copyright 2023 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: MIT
		+ */
		+package com.keenwrite.processors.markdown.extensions.references;
		+
		+import java.util.regex.Matcher;
		+
		+abstract class BasedSequenceParser {
		+ /**
		+ * Shared syntax between subclasses.
		+ */
		+ static final String REGEX_INNER = "(\\p{Alnum}+):(\\p{Alnum}+)";
		+
		+ private final String mTypeName;
		+ private final String mIdName;
		+
		+ BasedSequenceParser( final String text ) {
		+ final var matcher = createMatcher( text );
		+
		+ if( matcher.find() ) {
		+ mTypeName = matcher.group( 1 );
		+ mIdName = matcher.group( 2 );
		+ }
		+ else {
		+ mTypeName = null;
		+ mIdName = null;
		+ }
		+ }
		+
		+ abstract Matcher createMatcher( final String text );
		+
		+ String getTypeName() {
		+ return mTypeName;
		+ }
		+
		+ String getIdName() {
		+ return mIdName;
		+ }
		+}

src/main/java/com/keenwrite/processors/markdown/extensions/references/BasedSequenceXrefParser.java

		+/* Copyright 2023 White Magic Software, Ltd. -- All rights reserved.
		+ *
		+ * SPDX-License-Identifier: MIT
		+ */
		+package com.keenwrite.processors.markdown.extensions.references;
		+
		+import com.vladsch.flexmark.util.ast.Node;
		+import com.vladsch.flexmark.util.sequence.BasedSequence;
		+
		+import java.util.regex.Matcher;
		+import java.util.regex.Pattern;
		+
		+import static com.keenwrite.processors.markdown.extensions.EmptyNode.EMPTY_NODE;
		+import static java.util.regex.Pattern.UNICODE_CHARACTER_CLASS;
		+import static java.util.regex.Pattern.compile;
		+
		+class BasedSequenceXrefParser extends BasedSequenceParser {
		+ private static final String REGEX = STR. "\\[@\{ REGEX_INNER }]" ;
		+ private static final Pattern PATTERN = compile(
		+ REGEX, UNICODE_CHARACTER_CLASS
		+ );
		+
		+ BasedSequenceXrefParser( final String text ) {
		+ super( text );
		+ }
		+
		+ @Override
		+ Matcher createMatcher( final String text ) {
		+ return PATTERN.matcher( text );
		+ }
		+
		+ static BasedSequenceXrefParser parse( final BasedSequence chars ) {
		+ return new BasedSequenceXrefParser( chars.toString() );
		+ }
		+
		+ Node toNode() {
		+ final var typeName = getTypeName();
		+ final var idName = getIdName();
		+
		+ return typeName == null \|\| idName == null
		+ ? EMPTY_NODE
		+ : new AnchorXrefNode( typeName, idName );
		+ }
		+}

src/main/java/com/keenwrite/processors/markdown/extensions/references/CrossReferenceExtension.java

		@Override
		public void extend( final Parser.Builder builder ) {
		- builder.customInlineParserFactory( AnchorXrefInlineParser::new );
		+ builder.linkRefProcessorFactory( new AnchorXrefProcessorFactory() );
		builder.customDelimiterProcessor( new AnchorNameDelimiterProcessor() );
		}

src/main/java/com/keenwrite/processors/markdown/extensions/references/CrossReferenceNode.java

		String getRefAttrName();

		- default String toHtml() {
		- final String typeName = getTypeName();
		- final String idName = getIdName();
		-
		- return toHtml( typeName, idName );
		- }
		-
		- default String toHtml( final String type, final String id ) {
		- return STR.
		- "<a data-type=\"\{ type }\" \{ getRefAttrName() }=\"\{ id }\" />" ;
		- }
		-
		+ /**
		+ * Writes the HTML representation for this cross-reference node.
		+ *
		+ * @param html The HTML tag is written to the {@link HtmlWriter}.
		+ */
		default void write( final HtmlWriter html ) {
		- html.raw( toHtml() );
		+ final var type = getTypeName();
		+ final var id = getIdName();
		+ final var attr = getRefAttrName();
		+
		+ html.raw( STR. "<a data-type=\"\{ type }\" \{ attr }=\"\{ id }\" />" );
		}
		}