From 949b5c240e28bb41b554193c09c7f4b6465ef78c Mon Sep 17 00:00:00 2001
From: Nick Dumas <nobodys@business.com>
Date: Sun, 25 Jun 2023 10:05:27 -0400
Subject: [PATCH] I believe I now have a working wikilink lexer

---
 itemtype_string.go |  12 ++--
 lexer.go           |   7 ++-
 states.go          | 146 ++++++++++++++++++++++++---------------------
 3 files changed, 91 insertions(+), 74 deletions(-)

diff --git a/itemtype_string.go b/itemtype_string.go
index 4209011..26df3bb 100644
--- a/itemtype_string.go
+++ b/itemtype_string.go
@@ -10,16 +10,18 @@ func _() {
 	var x [1]struct{}
 	_ = x[ItemError-0]
 	_ = x[ItemEOF-1]
-	_ = x[ItemText-2]
+	_ = x[ItemIdent-2]
 	_ = x[ItemOpenLink-3]
 	_ = x[ItemCloseLink-4]
-	_ = x[ItemFragment-5]
-	_ = x[ItemAlias-6]
+	_ = x[ItemHeading-5]
+	_ = x[ItemBlockRef-6]
+	_ = x[ItemAlias-7]
+	_ = x[ItemText-8]
 }
 
-const _ItemType_name = "ItemErrorItemEOFItemTextItemOpenLinkItemCloseLinkItemFragmentItemAlias"
+const _ItemType_name = "ItemErrorItemEOFItemIdentItemOpenLinkItemCloseLinkItemHeadingItemBlockRefItemAliasItemText"
 
-var _ItemType_index = [...]uint8{0, 9, 16, 24, 36, 49, 61, 70}
+var _ItemType_index = [...]uint8{0, 9, 16, 25, 37, 50, 61, 73, 82, 90}
 
 func (i ItemType) String() string {
 	if i < 0 || i >= ItemType(len(_ItemType_index)-1) {
diff --git a/lexer.go b/lexer.go
index 012259a..b12e53d 100644
--- a/lexer.go
+++ b/lexer.go
@@ -12,11 +12,13 @@ import (
 const (
 	ItemError ItemType = iota
 	ItemEOF
-	ItemText
+	ItemIdent
 	ItemOpenLink
 	ItemCloseLink
-	ItemFragment
+	ItemHeading
+	ItemBlockRef
 	ItemAlias
+	ItemText
 )
 
 const (
@@ -27,6 +29,7 @@ const (
 	OpenLink  = "[["
 	CloseLink = "]]"
 	Alias     = "|"
+	Heading   = "#"
 	BlockRef  = "#^"
 )
 
diff --git a/states.go b/states.go
index c8938b5..dec93ce 100644
--- a/states.go
+++ b/states.go
@@ -1,55 +1,103 @@
 package wikilink
 
 import (
-	"log"
 	"strings"
-	"unicode"
 )
 
-func lexFragment(l *Lexer) stateFn {
-	log.Println("entering lexFragment")
-	for {
-		if strings.HasPrefix(l.input[l.pos:], CloseLink) {
-			return lexCloseLink
-		}
+func isOpenLink(s string) bool {
+	return strings.HasPrefix(s, OpenLink)
+}
 
-		if l.peek() == '^' {
-			l.next()
-			l.emit(ItemFragment)
-			l.acceptRun("abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789 -.,")
+func isCloseLink(s string) bool {
+	return strings.HasPrefix(s, CloseLink)
+}
 
-			return lexInsideLink
-		}
+func isAlias(s string) bool {
+	return strings.HasPrefix(s, Alias)
+}
 
-		return lexInsideLink
-	}
+func isHeading(s string) bool {
+	return strings.HasPrefix(s, Heading)
 }
 
-func lexAlias(l *Lexer) stateFn {
-	// l.accept
-	log.Println("entering lexAlias")
+func isBlockRef(s string) bool {
+	return strings.HasPrefix(s, BlockRef)
+}
+
+func lexIdent(l *Lexer) stateFn {
 	for {
-		if strings.HasPrefix(l.input[l.pos:], CloseLink) {
+		r := l.next()
+		s := l.input[l.pos:]
+		if r == '\\' { // i think this will handle escape characters?
+			break
+		}
+		switch {
+		case isBlockRef(s):
+			l.emit(ItemIdent)
+			return lexBlockRef
+		case isAlias(s):
+			l.emit(ItemIdent)
+			return lexAlias
+		case isCloseLink(s):
+			l.emit(ItemIdent)
 			return lexCloseLink
+		case isHeading(s):
+			l.emit(ItemIdent)
+			return lexHeading
 		}
-		r := l.next()
-
 	}
+	return nil
+}
+
+func lexHeading(l *Lexer) stateFn {
+	l.pos += len(Heading)
+	l.emit(ItemHeading)
+
+	return lexIdent
+}
+
+func lexBlockRef(l *Lexer) stateFn {
+	l.pos += len(BlockRef)
+	l.emit(ItemBlockRef)
+
+	return lexIdent
+}
+
+func lexAlias(l *Lexer) stateFn {
+	l.pos += len(Alias)
 	l.emit(ItemAlias)
 
-	return lexInsideLink
+	return lexIdent
 }
 
-func lexIdent(l *Lexer) stateFn {
+func lexText(l *Lexer) stateFn {
 	for {
-		if strings.HasPrefix(l.input[l.pos:], CloseLink) {
-			return lexCloseLink
+		if isOpenLink(l.input[l.pos:]) {
+			return lexOpenLink
+		}
+		r := l.next()
+		switch {
+		case r == EOF || r == '\n':
+			return l.errorf("wikilink terminated incorrectly")
 		}
-		// r := l.next()
 	}
-	return lexInsideLink
 }
 
+func lexOpenLink(l *Lexer) stateFn {
+	l.pos += len(OpenLink)
+	l.emit(ItemOpenLink)
+
+	return lexIdent
+}
+
+func lexCloseLink(l *Lexer) stateFn {
+	l.pos += len(CloseLink)
+	l.emit(ItemCloseLink)
+
+	return lexText
+}
+
+/*
 func lexInsideLink(l *Lexer) stateFn {
 	log.Println("entering lexInsideLink")
 
@@ -67,8 +115,9 @@ func lexInsideLink(l *Lexer) stateFn {
 			return l.errorf("unclosed link")
 		case r == '#':
 			l.emit(ItemText)
+			if l.peek() = '^' { return lexBlockRef }
 
-			return lexFragment
+			return lexHeading
 		case r == '|':
 			l.emit(ItemText)
 
@@ -77,42 +126,5 @@ func lexInsideLink(l *Lexer) stateFn {
 	}
 }
 
-func lexOpenLink(l *Lexer) stateFn {
-	log.Println("entering lexOpenLink")
-	l.pos += len(OpenLink)
-	l.emit(ItemOpenLink)
-
-	return lexInsideLink
-}
 
-func lexCloseLink(l *Lexer) stateFn {
-	log.Println("entering lexCloseLink")
-	l.pos += len(CloseLink)
-	l.emit(ItemCloseLink)
-
-	return lexText
-}
-
-func lexText(l *Lexer) stateFn {
-	log.Println("entering lexText")
-
-	for {
-		if strings.HasPrefix(l.input[l.pos:], OpenLink) {
-			if l.pos > l.start {
-				l.emit(ItemText)
-			}
-
-			return lexOpenLink
-		}
-
-		if l.next() == EOF {
-			break
-		}
-		if l.pos > l.start {
-			l.emit(ItemText)
-		}
-		l.emit(ItemEOF)
-		return nil
-	}
-	return nil
-}
+*/