feat(markup): nested lists

markup/lexer.go

@@ -31,6 +31,7 @@ type GemLexerState struct {
 	// Temporaries
 	img   *Img
 	table *Table
+	list  *List
 }
 
 type Line struct {
@@ -64,12 +65,6 @@ func lineToAST(line string, state *GemLexerState, ast *[]Line) {
 	// Process empty lines depending on the current state
 	if "" == strings.TrimSpace(line) {
 		switch state.where {
-		case "list":
-			state.where = ""
-			addLine(state.buf + "</ul>")
-		case "number":
-			state.where = ""
-			addLine(state.buf + "</ol>")
 		case "pre":
 			state.buf += "\n"
 		case "launchpad":
@@ -95,12 +90,10 @@ func lineToAST(line string, state *GemLexerState, ast *[]Line) {
 		goto imgState
 	case "table":
 		goto tableState
-	case "pre":
-		goto preformattedState
 	case "list":
 		goto listState
-	case "number":
-		goto numberState
+	case "pre":
+		goto preformattedState
 	case "launchpad":
 		goto launchpadState
 	default: // "p" or ""
@@ -121,6 +114,14 @@ tableState:
 	}
 	return
 
+listState:
+	if done := state.list.Parse(line); done {
+		state.list.Finalize()
+		state.where = ""
+		goto normalState
+	}
+	return
+
 preformattedState:
 	switch {
 	case startsWith("```"):
@@ -133,38 +134,6 @@ preformattedState:
 	}
 	return
 
-listState:
-	switch {
-	case startsWith("* "):
-		state.buf += fmt.Sprintf("\t<li>%s</li>\n", ParagraphToHtml(state.name, line[2:]))
-	case startsWith("```"):
-		state.where = "pre"
-		addLine(state.buf + "</ul>")
-		state.id++
-		state.buf = fmt.Sprintf("<pre id='%d' alt='%s' class='codeblock'><code>", state.id, strings.TrimPrefix(line, "```"))
-	default:
-		state.where = ""
-		addLine(state.buf + "</ul>")
-		goto normalState
-	}
-	return
-
-numberState:
-	switch {
-	case startsWith("*. "):
-		state.buf += fmt.Sprintf("\t<li>%s</li>\n", ParagraphToHtml(state.name, line[3:]))
-	case startsWith("```"):
-		state.where = "pre"
-		addLine(state.buf + "</ol>")
-		state.id++
-		state.buf = fmt.Sprintf("<pre id='%d' alt='%s' class='codeblock'><code>", state.id, strings.TrimPrefix(line, "```"))
-	default:
-		state.where = ""
-		addLine(state.buf + "</ol>")
-		goto normalState
-	}
-	return
-
 launchpadState:
 	switch {
 	case startsWith("=>"):
@@ -190,16 +159,6 @@ normalState:
 		addParagraphIfNeeded()
 		state.where = "pre"
 		state.buf = fmt.Sprintf("<pre id='%d' alt='%s' class='codeblock'><code>", state.id, strings.TrimPrefix(line, "```"))
-	case startsWith("* "):
-		addParagraphIfNeeded()
-		state.where = "list"
-		state.buf = fmt.Sprintf("<ul id='%d'>\n", state.id)
-		goto listState
-	case startsWith("*. "):
-		addParagraphIfNeeded()
-		state.where = "number"
-		state.buf = fmt.Sprintf("<ol id='%d'>\n", state.id)
-		goto numberState
 
 	case startsWith("###### "):
 		addParagraphIfNeeded()
@@ -241,6 +200,12 @@ normalState:
 	case MatchesHorizontalLine(line):
 		addParagraphIfNeeded()
 		*ast = append(*ast, Line{id: -1, contents: "<hr/>"})
+	case MatchesList(line):
+		addParagraphIfNeeded()
+		list, _ := NewList(line, state.name)
+		state.where = "list"
+		state.list = list
+		addLine(state.list)
 	case MatchesImg(line):
 		addParagraphIfNeeded()
 		img, shouldGoBackToNormal := ImgFromFirstLine(line, state.name)

markup/list.go

@@ -0,0 +1,171 @@
+package markup
+
+import (
+	"errors"
+	"strings"
+)
+
+func parseListItem(line string) (level int, offset int, ordered bool, err error) {
+	for line[level] == '*' {
+		level++
+	}
+
+	if line[level] == '.' {
+		ordered = true
+		offset = level + 2
+	} else {
+		ordered = false
+		offset = level + 1
+	}
+
+	if line[offset-1] != ' ' || len(line) < offset+2 || level < 1 || level > 6 {
+		err = errors.New("ill-formatted list item")
+	}
+	return
+}
+
+func MatchesList(line string) bool {
+	level, _, _, err := parseListItem(line)
+	return err == nil && level == 1
+}
+
+type listItem struct {
+	content  string
+	parent   *listItem
+	children []*listItem
+	depth    int
+}
+
+func newListItem(parent *listItem) *listItem {
+	depth := 0
+	if parent != nil {
+		depth = parent.depth + 1
+	}
+	return &listItem{
+		parent:   parent,
+		children: make([]*listItem, 0),
+		depth:    depth,
+	}
+}
+
+func (item *listItem) renderAsHtmlTo(b *strings.Builder, hyphaName string, ordered bool) {
+	if len(item.content) > 0 {
+		b.WriteString("<li>")
+		b.WriteString(ParagraphToHtml(hyphaName, item.content))
+	}
+
+	if len(item.children) > 0 {
+		if ordered {
+			b.WriteString("<ol>")
+		} else {
+			b.WriteString("<ul>")
+		}
+
+		for _, child := range item.children {
+			child.renderAsHtmlTo(b, hyphaName, ordered)
+		}
+
+		if ordered {
+			b.WriteString("</ol>")
+		} else {
+			b.WriteString("</ul>")
+		}
+	}
+
+	if len(item.content) > 0 {
+		b.WriteString("</li>")
+	}
+}
+
+// A structure representing ordered and unordered lists in the AST.
+type List struct {
+	curr      *listItem
+	hyphaName string
+	ordered   bool
+	finalized bool
+}
+
+func NewList(line, hyphaName string) (*List, bool) {
+	list := &List{
+		hyphaName: hyphaName,
+		curr:      newListItem(nil),
+	}
+	return list, list.Parse(line)
+}
+
+func (list *List) pushItem() {
+	item := newListItem(list.curr)
+	list.curr.children = append(list.curr.children, item)
+	list.curr = item
+}
+
+func (list *List) popItem() {
+	if list.curr == nil {
+		return
+	}
+	list.curr = list.curr.parent
+}
+
+func (list *List) balance(level int) {
+	for level > list.curr.depth {
+		list.pushItem()
+	}
+
+	for level < list.curr.depth {
+		list.popItem()
+	}
+}
+
+func (list *List) Parse(line string) (done bool) {
+	level, offset, ordered, err := parseListItem(line)
+	if err != nil {
+		list.Finalize()
+		return true
+	}
+
+	// update ordered flag if the current node is the root one
+	// (i.e. no parsing has been done yet)
+	if list.curr.parent == nil {
+		list.ordered = ordered
+	}
+
+	// if list type has suddenly changed (ill-formatted list), quit
+	if ordered != list.ordered {
+		list.Finalize()
+		return true
+	}
+
+	list.balance(level)
+
+	// if the current node already has content, create a new one
+	// to prevent overwriting existing content (effectively creating
+	// a new sibling node)
+	if len(list.curr.content) > 0 {
+		list.popItem()
+		list.pushItem()
+	}
+
+	list.curr.content = line[offset:]
+
+	return false
+}
+
+func (list *List) Finalize() {
+	if !list.finalized {
+		// close all opened nodes, effectively going up to the root node
+		list.balance(0)
+		list.finalized = true
+	}
+}
+
+func (list *List) RenderAsHtml() (html string) {
+	// for a good measure
+	list.Finalize()
+
+	b := &strings.Builder{}
+
+	// fire up recursive render process
+	list.curr.renderAsHtmlTo(b, list.hyphaName, list.ordered)
+
+	return b.String()
+}

markup/parser.go

@@ -15,6 +15,8 @@ func Parse(ast []Line, from, to int, recursionLevel int) (html string) {
 				html += v.ToHtml()
 			case Table:
 				html += v.asHtml()
+			case *List:
+				html += v.RenderAsHtml()
 			case string:
 				html += v
 			default: