Match double primes

Author	Dave Jarvis <email>
Date	2021-06-01 21:30:57 GMT-0700
Commit	100fc760579caadf026caeef0e93ec5619cb4d3e
Parent	2a21d6d

lib/src/main/java/com/keenwrite/quotes/Parser.java

		}

		+ for( final var lex : mLexemes ) {
		+ parse( lex, consumer );
		+ }
		+
		// Create/convert a list of all unambiguous quotations.
		// Let TERM ::= (, \| ; \| ! \| ? \| .)

		if( lex2.isType( QUOTE_SINGLE ) && lex3.isType( WORD ) &&
		lex1.anyType( WORD, PERIOD, NUMBER ) ) {
		+ // Examples: y'all, Ph.D.'ll, 20's
		consumer.accept( new Token( QUOTE_APOSTROPHE, lex2 ) );
		flush( mLexemes );
		}
		else if( lex1.isType( QUOTE_SINGLE ) && lex3.isType( QUOTE_SINGLE ) &&
		"n".equalsIgnoreCase( lex2.toString( mText ) ) ) {
		+ // I.e., 'n'
		consumer.accept( new Token( QUOTE_APOSTROPHE, lex1 ) );
		consumer.accept( new Token( QUOTE_APOSTROPHE, lex3 ) );
		}
		else if( lex1.isType( NUMBER ) && lex2.isType( QUOTE_SINGLE ) ) {
		- consumer.accept( new Token( QUOTE_PRIME_SINGLE, lex2 ) );
		+ if( lex3.isType( QUOTE_SINGLE ) ) {
		+ // E.g., 2''
		+ consumer.accept(
		+ new Token( QUOTE_PRIME_DOUBLE, lex2.began(), lex3.ended() ) );
		+ }
		+ else {
		+ // E.g., 2'
		+ consumer.accept( new Token( QUOTE_PRIME_SINGLE, lex2 ) );
		+ }
		}
		else if( lex1.isType( NUMBER ) && lex2.isType( QUOTE_DOUBLE ) ) {
		+ // E.g., 2"
		consumer.accept( new Token( QUOTE_PRIME_DOUBLE, lex2 ) );
		}
		else if( lex1.isType( QUOTE_SINGLE ) && lex2.isType( WORD ) &&
		+ // E.g., 'contraction
		beginsUnambiguously( lex2.toString( mText ) ) ) {
		consumer.accept( new Token( QUOTE_APOSTROPHE, lex1 ) );
		}
		else if( lex1.isType( ESC_SINGLE ) ) {
		+ // E.g., \'
		consumer.accept( new Token( QUOTE_STRAIGHT_SINGLE, lex1 ) );
		}
		else if( lex1.isType( ESC_DOUBLE ) ) {
		- consumer.accept( new Token( QUOTE_STRAIGHT_DOUBLE, lex2 ) );
		+ // E.g., \"
		+ consumer.accept( new Token( QUOTE_STRAIGHT_DOUBLE, lex1 ) );
		}
		else if( lex1.anyType( QUOTE_SINGLE, QUOTE_DOUBLE ) ) {

lib/src/main/java/com/keenwrite/quotes/Token.java

		public class Token implements Comparable<Token> {
		private final TokenType mType;
		- private final Lexeme mLexeme;
		+ final int mBegan;
		+ final int mEnded;

		- public Token( final TokenType type, final Lexeme lexeme ) {
		+ /**
		+ * Convenience constructor to create a token that uses the lexeme's offsets.
		+ *
		+ * @param type The type of {@link Token} to create.
		+ * @param lexeme Container for beginning and ending text offsets.
		+ */
		+ Token( final TokenType type, final Lexeme lexeme ) {
		+ this( type, lexeme.began(), lexeme.ended() );
		+ }
		+
		+ Token( final TokenType type, final int began, final int ended ) {
		assert type != null;
		- assert lexeme != null;
		+ assert began >= 0;
		+ assert ended > began;

		mType = type;
		- mLexeme = lexeme;
		+ mBegan = began;
		+ mEnded = ended;
		}

		TokenType getType() {
		return mType;
		}

		int began() {
		- return mLexeme.began();
		+ return mBegan;
		}

		int ended() {
		- return mLexeme.ended();
		+ return mEnded;
		}

		@Override
		public int compareTo( final Token that ) {
		- return mLexeme.compareTo( that.mLexeme );
		- }
		-
		- public String toString( final String text ) {
		- return mLexeme.toString( text );
		+ return this.mBegan - that.mBegan;
		}

		@Override
		public String toString() {
		return getClass().getSimpleName() + "{" +
		"mType=" + mType +
		- ", mLexeme=" + mLexeme +
		+ ", mBegan=" + mBegan +
		+ ", mEnded=" + mEnded +
		'}';
		}

lib/src/test/java/com/keenwrite/quotes/LexerTest.java


		/**
		- * Tests lexing words, numbers, punctuation, spaces, and newlines.
		+ * Tests lexing words, numbers, punctuation, spaces, newlines, etc.
		*/
		class LexerTest {

lib/src/test/resources/com/keenwrite/quotes/smartypants.txt

		+# ########################################################################
		+# Escaped quotes
		+# ########################################################################
		+She stood 5\'7\".
		+She stood 5'7".
		+
		+# ########################################################################
		+# Primes (single, double)
		+# ########################################################################
		+
		+# No space after the feet sign.
		+It's 4'11" away.
		+It's 4′11″ away.
		+
		+Alice's friend is 6'3" tall.
		+Alice's friend is 6′3″ tall.
		+
		+Bob's table is 5'' × 4''.
		+Bob's table is 5″ × 4″.
		+
		+Bob's table is 5'×4'.
		+Bob's table is 5′×4′.
		+
		+What's this -5.5'' all about?
		+What's this -5.5″ all about?
		+
		++7.9'' is weird.
		++7.9″ is weird.
		+
		+Use 11.5"x14.25" virtual paper!
		+Use 11.5″x14.25″ virtual paper!
		+
		+Angular measure 3° 5' 30" means 3 degs, 5 arcmins, & 30 arcsecs.
		+Angular measure 3° 5′ 30″ means 3 degs, 5 arcmins, & 30 arcsecs.
		+
		+# ########################################################################
		+# Inside contractions (no leading/trailing apostrophes)
		+# ########################################################################
		+I don't like it: I love's it!
		+I don't like it: I love's it!
		+
		+We'd've thought that pancakes'll be sweeter there.
		+We'd've thought that pancakes'll be sweeter there.
		+
		+She'd be coming o'er when the horse'd gone to pasture...
		+She'd be coming o'er when the horse'd gone to pasture...
		+
		+# ########################################################################
		+# Beginning contractions (leading apostrophes)
		+# ########################################################################
		+'Twas and 'tis whate'er lay 'twixt dawn and dusk 'n River Styx.
		+'Twas and 'tis whate'er lay 'twixt dawn and dusk 'n River Styx.
		+
		+# ########################################################################
		+# Outside contractions (leading and trailing, no middle)
		+# ########################################################################
		+Salt 'n' vinegar, fish-'n'-chips, sugar 'n' spice!
		+Salt 'n' vinegar, fish-'n'-chips, sugar 'n' spice!
		+
		+# ########################################################################
		+# Ending contractions (trailing apostrophes)
		+# ########################################################################
		+Didn' get th' message.
		+Didn' get th' message.
		+
		+Namsayin', y'know what I'ma sayin'?
		+Namsayin', y'know what I'ma sayin'?
		+
		+# ########################################################################
		+# Decades
		+# ########################################################################
		+The Roaring '20s had the best music, no?
		+The Roaring '20s had the best music, no?
		+
		+Took place in '04, yes'm!
		+Took place in '04, yes'm!
		+
		+# ########################################################################
		+# Backticks (left and right double quotes)
		+# ########################################################################
		+``I am Sam''
		+“I am Sam”
		+
		+``Sam's away today''
		+“Sam's away today”
		+
		+``Sam's gone!
		+“Sam's gone!
		+
		+``5'10" tall 'e was!''
		+“5′10″ tall 'e was!”
		+
		+# ########################################################################
		+# Consecutive quotes
		+# ########################################################################
		+"'I'm trouble.'"
		+“‘I'm trouble.’”
		+
		+'"Trouble's my name."'
		+‘“Trouble's my name.“‘
		+
		+# ########################################################################
		+# Escaped quotes
		+# ########################################################################
		+\"What?\"
		+“What?”
		+
		+# ########################################################################
		+# Double quotes
		+# ########################################################################
		+"I am Sam"
		+“I am Sam”
		+
		+"...even better!"
		+“...even better!”
		+
		+"It was so," said he.
		+“It was so,” said he.
		+
		+"She said, 'Llamas'll languish, they'll--
		+“She said, ‘Llamas'll languish, they'll--
		+
		+With "air quotes" in the middle.
		+With “air quotes” in the middle.
		+
		+With--"air quotes"--and dashes.
		+With--“air quotes”--and dashes.
		+
		+"Not "quite" what you expected?"
		+“Not “quite” what you expected?”
		+
		+# ########################################################################
		+# Nested quotations
		+# ########################################################################
		+"'Here I am,' said Sam"
		+“‘Here I am,’ said Sam”
		+
		+'"Here I am," said Sam'
		+‘“Here I am,”, said Sam’
		+
		+'Hello, "Dr. Brown," what's your real name?'
		+‘Hello, “Dr. Brown,” what's your real name?’
		+
		+"'Twas, t'wasn't thy name, 'twas it?" said Jim "the Barber" Brown.
		+“'Twas, t'wasn't thy name, 'twas it?” said Jim “the Barber” Brown.
		+
		+# ########################################################################
		+# Single quotes
		+# ########################################################################
		+'I am Sam'
		+‘I am Sam’
		+
		+'It was so,' said he.
		+‘It was so,’ said he.
		+
		+'...even better!'
		+‘...even better!’
		+
		+With 'quotes' in the middle.
		+With ‘quotes’ in the middle.
		+
		+With--'imaginary'--dashes.
		+With--‘imaginary’--dashes.
		+
		+'Not 'quite' what you expected?'
		+‘Not ‘quite’ what you expected?’
		+
		+''Cause I don't like it, 's why,' said Pat.
		+‘'Cause I don't like it, 's why,’ said Pat.
		+
		+'It's a beautiful day!'
		+‘It's a beautiful day!’
		+
		+'He said, "Thinkin'."'
		+‘He said, “Thinkin’.”’
		+
		+# ########################################################################
		+# Possessives
		+# ########################################################################
		+Sam's Sams' and the Ross's roses' thorns were prickly.
		+Sam's Sams' and the Ross's roses' thorns were prickly.
		+
		+# ########################################################################
		+# Mixed
		+# ########################################################################
		+"I heard she said, 'That's Sam's'," said the Sams' cat.
		+“I heard she said, ‘That's Sam's’,” said the Sams' cat.
		+
		+"'Janes' said, ''E'll be spooky, Sam's son with the jack-o'-lantern!'" said the O'Mally twins'---y'know---ghosts in unison.
		+“‘Janes' said, ‘'E'll be spooky, Sam's son with the jack-o'-lantern!’” said the O'Mally twins'---y'know---ghosts in unison.
		+
		+'He's at Sams'
		+‘He' at Sams’
		+
		+\"Hello!\"
		+“Hello!”
		+
		+ma'am
		+ma'am
		+
		+'Twas midnight
		+'Twas midnight
		+
		+\"Hello,\" said the spider. \"'Shelob' is my name.\"
		+“Hello,” said the spider. “‘Shelob’ is my name.”
		+
		+'A', 'B', and 'C' are letters.
		+‘A’ ‘B’ and ‘C’ are letters.
		+
		+'Oak,' 'elm,' and 'beech' are names of trees. So is 'pine.'
		+‘Oak,’ ‘elm,’ and ‘beech’ are names of trees. So is ‘pine.’
		+
		+'He said, \"I want to go.\"' Were you alive in the 70's?
		+‘He said, “I want to go.”’ Were you alive in the 70's?
		+
		+\"That's a 'magic' sock.\"
		+“That's a ‘magic’ sock.”
		+
		+Website! Company Name, Inc. (\"Company Name\" or \"Company\") recommends reading the following terms and conditions, carefully:
		+Website! Company Name, Inc. (“Company Name” or “Company”) recommends reading the following terms and conditions, carefully:
		+
		+Website! Company Name, Inc. ('Company Name' or 'Company') recommends reading the following terms and conditions, carefully:
		+Website! Company Name, Inc. (‘Company Name’ or ‘Company’) recommends reading the following terms and conditions, carefully:
		+
		+Workin' hard
		+Workin' hard
		+
		+'70s are my favorite numbers,' she said.
		+‘70s are my favorite numbers,’ she said.
		+
		+'70s fashion was weird.
		+'70s fashion was weird.
		+
		+12\" record, 5'10\" height
		+12″ record, 5′10″ height
		+
		+Model \"T2000\"
		+Model “T2000”
		+
		+iPad 3's battery life is not great.
		+iPad 3's battery life is not great.
		+
		+Book 'em, Danno. Rock 'n' roll. 'Cause 'twas the season.
		+Book 'em, Danno. Rock 'n' roll. 'Cause 'twas the season.
		+
		+'85 was a good year. (The entire '80s were.)
		+'85 was a good year. (The entire '80s were.)

Delta	290 lines added, 15 lines removed, 275-line increase