Add table pretty printing & alignment

Also dismissed implementing colgroups for now - had it but didn't like the
added complexity for a very questionable benefit - i've actually never used
that feature of org tables...

README.org

@@ -8,8 +8,6 @@ A basic org-mode parser in go
 - support more keywords: https://orgmode.org/manual/In_002dbuffer-settings.html
   - #+LINK
   - #+INCLUDE
-- table colgroups https://orgmode.org/worg/org-tutorials/tables.html
-- table pretty printing
 *** TODO [[https://github.com/chaseadamsio/goorgeous/issues/10][#10]]: Support noexport
 *** TODO [[https://github.com/chaseadamsio/goorgeous/issues/31][#31]]: Support #+INCLUDE
 - see https://orgmode.org/manual/Include-files.html

etc/style.css

@@ -104,3 +104,7 @@ figcaption {
   background-color: #ccc; }
 .footnote-definition .footnote-body p:only-child {
   margin: 0.2em 0; }
+
+.align-left   { text-align: left;   }
+.align-center { text-align: center; }
+.align-right  { text-align: right;  }

org/html.go

@@ -84,12 +84,6 @@ func (w *HTMLWriter) writeNodes(ns ...Node) {
 
 		case Table:
 			w.writeTable(n)
-		case TableHeader:
-			w.writeTableHeader(n)
-		case TableRow:
-			w.writeTableRow(n)
-		case TableSeparator:
-			w.writeTableSeparator(n)
 
 		case Paragraph:
 			w.writeParagraph(n)
@@ -283,34 +277,39 @@ func (w *HTMLWriter) writeNodeWithMeta(n NodeWithMeta) {
 
 func (w *HTMLWriter) writeTable(t Table) {
 	w.WriteString("<table>\n")
-	w.writeNodes(t.Header)
+	beforeFirstContentRow := true
+	for i, row := range t.Rows {
+		if row.IsSpecial || len(row.Columns) == 0 {
+			continue
+		}
+		if beforeFirstContentRow {
+			beforeFirstContentRow = false
+			if i+1 < len(t.Rows) && len(t.Rows[i+1].Columns) == 0 {
+				w.WriteString("<thead>\n")
+				w.writeTableColumns(row.Columns, "th")
+				w.WriteString("</thead>\n<tbody>\n")
+				continue
+			} else {
 				w.WriteString("<tbody>\n")
-	w.writeNodes(t.Rows...)
+			}
+		}
+		w.writeTableColumns(row.Columns, "td")
+	}
 	w.WriteString("</tbody>\n</table>\n")
 }
 
-func (w *HTMLWriter) writeTableRow(t TableRow) {
+func (w *HTMLWriter) writeTableColumns(columns []Column, tag string) {
 	w.WriteString("<tr>\n")
-	for _, column := range t.Columns {
-		w.WriteString("<td>")
-		w.writeNodes(column...)
-		w.WriteString("</td>")
+	for _, column := range columns {
+		if column.Align == "" {
+			w.WriteString(fmt.Sprintf("<%s>", tag))
+		} else {
+			w.WriteString(fmt.Sprintf(`<%s class="align-%s">`, tag, column.Align))
 		}
-	w.WriteString("\n</tr>\n")
-}
-
-func (w *HTMLWriter) writeTableHeader(t TableHeader) {
-	w.WriteString("<thead>\n")
-	for _, column := range t.Columns {
-		w.WriteString("<th>")
-		w.writeNodes(column...)
-		w.WriteString("</th>")
+		w.writeNodes(column.Children...)
+		w.WriteString(fmt.Sprintf("</%s>\n", tag))
 	}
-	w.WriteString("\n</thead>\n")
-}
-
-func (w *HTMLWriter) writeTableSeparator(t TableSeparator) {
-	w.WriteString("<tr></tr>\n")
+	w.WriteString("</tr>\n")
 }
 
 func withHTMLAttributes(input string, kvs ...string) string {

org/org.go

@@ -41,6 +41,12 @@ func (w *OrgWriter) emptyClone() *OrgWriter {
 	return &wcopy
 }
 
+func (w *OrgWriter) nodesAsString(nodes ...Node) string {
+	tmp := w.emptyClone()
+	tmp.writeNodes(nodes...)
+	return tmp.String()
+}
+
 func (w *OrgWriter) writeNodes(ns ...Node) {
 	for _, n := range ns {
 		switch n := n.(type) {
@@ -65,12 +71,6 @@ func (w *OrgWriter) writeNodes(ns ...Node) {
 
 		case Table:
 			w.writeTable(n)
-		case TableHeader:
-			w.writeTableHeader(n)
-		case TableRow:
-			w.writeTableRow(n)
-		case TableSeparator:
-			w.writeTableSeparator(n)
 
 		case Paragraph:
 			w.writeParagraph(n)
@@ -206,34 +206,45 @@ func (w *OrgWriter) writeListItem(li ListItem) {
 }
 
 func (w *OrgWriter) writeTable(t Table) {
-	w.writeNodes(t.Header)
-	w.writeNodes(t.Rows...)
-}
+	for _, row := range t.Rows {
+		w.WriteString(w.indent)
+		if len(row.Columns) == 0 {
+			w.WriteString(`|`)
+			for i := 0; i < len(t.ColumnInfos); i++ {
+				w.WriteString(strings.Repeat("-", t.ColumnInfos[i].Len+2))
+				if i < len(t.ColumnInfos)-1 {
+					w.WriteString("+")
+				}
+			}
+			w.WriteString(`|`)
 
-func (w *OrgWriter) writeTableHeader(th TableHeader) {
-	w.writeNodes(th.SeparatorBefore)
-	w.writeTableColumns(th.Columns)
-	w.writeNodes(th.SeparatorAfter)
-}
-
-func (w *OrgWriter) writeTableRow(tr TableRow) {
-	w.writeTableColumns(tr.Columns)
-}
-
-func (w *OrgWriter) writeTableSeparator(ts TableSeparator) {
-	w.WriteString(w.indent + ts.Content + "\n")
-}
-
-func (w *OrgWriter) writeTableColumns(columns [][]Node) {
-	w.WriteString(w.indent + "| ")
-	for i, columnNodes := range columns {
-		w.writeNodes(columnNodes...)
-		w.WriteString(" |")
-		if i < len(columns)-1 {
+		} else {
+			w.WriteString(`|`)
+			for _, column := range row.Columns {
+				w.WriteString(` `)
+				content := w.nodesAsString(column.Children...)
+				if content == "" {
+					content = " "
+				}
+				n := column.Len - len(content)
+				if n < 0 {
+					n = 0
+				}
+				if column.Align == "center" {
+					if n%2 != 0 {
 						w.WriteString(" ")
 					}
+					w.WriteString(strings.Repeat(" ", n/2) + content + strings.Repeat(" ", n/2))
+				} else if column.Align == "right" {
+					w.WriteString(strings.Repeat(" ", n) + content)
+				} else {
+					w.WriteString(content + strings.Repeat(" ", n))
+				}
+				w.WriteString(` |`)
+			}
 		}
 		w.WriteString("\n")
+	}
 }
 
 func (w *OrgWriter) writeHorizontalRule(hr HorizontalRule) {

org/table.go

@@ -2,27 +2,35 @@ package org
 
 import (
 	"regexp"
+	"strconv"
 	"strings"
 )
 
 type Table struct {
-	Header Node
-	Rows   []Node
+	Rows        []Row
+	ColumnInfos []ColumnInfo
 }
 
-type TableSeparator struct{ Content string }
-
-type TableHeader struct {
-	SeparatorBefore Node
-	Columns         [][]Node
-	SeparatorAfter  Node
+type Row struct {
+	Columns   []Column
+	IsSpecial bool
 }
 
-type TableRow struct{ Columns [][]Node }
+type Column struct {
+	Children []Node
+	*ColumnInfo
+}
+
+type ColumnInfo struct {
+	Align string
+	Len   int
+}
 
 var tableSeparatorRegexp = regexp.MustCompile(`^(\s*)(\|[+-|]*)\s*$`)
 var tableRowRegexp = regexp.MustCompile(`^(\s*)(\|.*)`)
 
+var columnAlignRegexp = regexp.MustCompile(`^<(l|c|r)>$`)
+
 func lexTable(line string) (token, bool) {
 	if m := tableSeparatorRegexp.FindStringSubmatch(line); m != nil {
 		return token{"tableSeparator", len(m[1]), m[2], m}, true
@@ -33,43 +41,87 @@ func lexTable(line string) (token, bool) {
 }
 
 func (d *Document) parseTable(i int, parentStop stopFn) (int, Node) {
-	rows, start := []Node{}, i
-	for !parentStop(d, i) && (d.tokens[i].kind == "tableRow" || d.tokens[i].kind == "tableSeparator") {
-		consumed, row := d.parseTableRowOrSeparator(i, parentStop)
-		i += consumed
-		rows = append(rows, row)
-	}
-
-	consumed := i - start
-
-	if len(rows) >= 2 {
-		if row, ok := rows[0].(TableRow); ok {
-			if separator, ok := rows[1].(TableSeparator); ok {
-				return consumed, Table{TableHeader{nil, row.Columns, separator}, rows[2:]}
-			}
-		}
-	}
-	if len(rows) >= 3 {
-		if separatorBefore, ok := rows[0].(TableSeparator); ok {
-			if row, ok := rows[1].(TableRow); ok {
-				if separatorAfter, ok := rows[2].(TableSeparator); ok {
-					return consumed, Table{TableHeader{separatorBefore, row.Columns, separatorAfter}, rows[3:]}
-				}
+	rawRows, start := [][]string{}, i
+	for ; !parentStop(d, i); i++ {
+		if t := d.tokens[i]; t.kind == "tableRow" {
+			rawRow := strings.FieldsFunc(d.tokens[i].content, func(r rune) bool { return r == '|' })
+			for i := range rawRow {
+				rawRow[i] = strings.TrimSpace(rawRow[i])
 			}
+			rawRows = append(rawRows, rawRow)
+		} else if t.kind == "tableSeparator" {
+			rawRows = append(rawRows, nil)
+		} else {
+			break
 		}
 	}
 
-	return consumed, Table{nil, rows}
+	table := Table{nil, getColumnInfos(rawRows)}
+	for _, rawColumns := range rawRows {
+		row := Row{nil, isSpecialRow(rawColumns)}
+		if len(rawColumns) != 0 {
+			for i := range table.ColumnInfos {
+				column := Column{nil, &table.ColumnInfos[i]}
+				if i < len(rawColumns) {
+					column.Children = d.parseInline(rawColumns[i])
+				}
+				row.Columns = append(row.Columns, column)
+			}
+		}
+		table.Rows = append(table.Rows, row)
+	}
+	return i - start, table
 }
 
-func (d *Document) parseTableRowOrSeparator(i int, _ stopFn) (int, Node) {
-	if d.tokens[i].kind == "tableSeparator" {
-		return 1, TableSeparator{d.tokens[i].content}
+func getColumnInfos(rows [][]string) []ColumnInfo {
+	columnCount := 0
+	for _, columns := range rows {
+		if n := len(columns); n > columnCount {
+			columnCount = n
 		}
-	fields := strings.FieldsFunc(d.tokens[i].content, func(r rune) bool { return r == '|' })
-	row := TableRow{}
-	for _, field := range fields {
-		row.Columns = append(row.Columns, d.parseInline(strings.TrimSpace(field)))
 	}
-	return 1, row
+
+	columnInfos := make([]ColumnInfo, columnCount)
+	for i := 0; i < columnCount; i++ {
+		countNumeric, countNonNumeric := 0, 0
+		for _, columns := range rows {
+			if !(i < len(columns)) {
+				continue
+			}
+
+			if len(columns[i]) > columnInfos[i].Len {
+				columnInfos[i].Len = len(columns[i])
+			}
+
+			if m := columnAlignRegexp.FindStringSubmatch(columns[i]); m != nil && isSpecialRow(columns) {
+				switch m[1] {
+				case "l":
+					columnInfos[i].Align = "left"
+				case "c":
+					columnInfos[i].Align = "center"
+				case "r":
+					columnInfos[i].Align = "right"
+				}
+			} else if _, err := strconv.ParseFloat(columns[i], 32); err == nil {
+				countNumeric++
+			} else if strings.TrimSpace(columns[i]) != "" {
+				countNonNumeric++
+			}
+		}
+
+		if columnInfos[i].Align == "" && countNumeric >= countNonNumeric {
+			columnInfos[i].Align = "right"
+		}
+	}
+	return columnInfos
+}
+
+func isSpecialRow(rawColumns []string) bool {
+	isAlignRow := true
+	for _, rawColumn := range rawColumns {
+		if !columnAlignRegexp.MatchString(rawColumn) && rawColumn != "" {
+			isAlignRow = false
+		}
+	}
+	return isAlignRow
 }

org/testdata/footnotes.html

@@ -53,13 +53,16 @@ and tables
 <table>
 <tbody>
 <tr>
-<td>1</td><td>a</td>
+<td class="align-right">1</td>
+<td>a</td>
 </tr>
 <tr>
-<td>2</td><td>b</td>
+<td class="align-right">2</td>
+<td>b</td>
 </tr>
 <tr>
-<td>3</td><td>c</td>
+<td class="align-right">3</td>
+<td>c</td>
 </tr>
 </tbody>
 </table>

org/testdata/lists.html

@@ -60,11 +60,17 @@ and another one with a table
 </p>
 <table>
 <thead>
-<th>a</th><th>b</th><th>c</th>
+<tr>
+<th class="align-right">a</th>
+<th class="align-right">b</th>
+<th class="align-right">c</th>
+</tr>
 </thead>
 <tbody>
 <tr>
-<td>1</td><td>2</td><td>3</td>
+<td class="align-right">1</td>
+<td class="align-right">2</td>
+<td class="align-right">3</td>
 </tr>
 </tbody>
 </table>

org/testdata/misc.html

@@ -20,10 +20,12 @@ crazy ain't it?
 <table>
 <tbody>
 <tr>
-<td><strong>foo</strong></td><td>foo</td>
+<td><strong>foo</strong></td>
+<td>foo</td>
 </tr>
 <tr>
-<td><strong>bar</strong></td><td>bar</td>
+<td><strong>bar</strong></td>
+<td>bar</td>
 </tr>
 </tbody>
 </table>

org/testdata/tables.html

@@ -1,11 +1,17 @@
 <figure>
 <table>
 <thead>
-<th>a</th><th>b</th><th>c</th>
+<tr>
+<th class="align-right">a</th>
+<th class="align-right">b</th>
+<th class="align-right">c</th>
+</tr>
 </thead>
 <tbody>
 <tr>
-<td>1</td><td>2</td><td>3</td>
+<td class="align-right">1</td>
+<td class="align-right">2</td>
+<td class="align-right">3</td>
 </tr>
 </tbody>
 </table>
@@ -16,11 +22,17 @@ table with separator before and after header
 <figure>
 <table>
 <thead>
-<th>a</th><th>b</th><th>c</th>
+<tr>
+<th class="align-right">a</th>
+<th class="align-right">b</th>
+<th class="align-right">c</th>
+</tr>
 </thead>
 <tbody>
 <tr>
-<td>1</td><td>2</td><td>3</td>
+<td class="align-right">1</td>
+<td class="align-right">2</td>
+<td class="align-right">3</td>
 </tr>
 </tbody>
 </table>
@@ -31,9 +43,10 @@ table with separator after header
 <figure>
 <table>
 <tbody>
-<tr></tr>
 <tr>
-<td>1</td><td>2</td><td>3</td>
+<td class="align-right">1</td>
+<td class="align-right">2</td>
+<td class="align-right">3</td>
 </tr>
 </tbody>
 </table>
@@ -45,7 +58,9 @@ table without header (but separator before)
 <table>
 <tbody>
 <tr>
-<td>1</td><td>2</td><td>3</td>
+<td class="align-right">1</td>
+<td class="align-right">2</td>
+<td class="align-right">3</td>
 </tr>
 </tbody>
 </table>
@@ -53,3 +68,50 @@ table without header (but separator before)
 table without header
 </figcaption>
 </figure>
+<figure>
+<table>
+<thead>
+<tr>
+<th class="align-left">left aligned</th>
+<th class="align-right">right aligned</th>
+<th class="align-center">center aligned</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td class="align-left">42</td>
+<td class="align-right">42</td>
+<td class="align-center">42</td>
+</tr>
+<tr>
+<td class="align-left">foobar</td>
+<td class="align-right">foobar</td>
+<td class="align-center">foobar</td>
+</tr>
+</tbody>
+</table>
+<figcaption>
+table with aligned columns
+</figcaption>
+</figure>
+<figure>
+<table>
+<thead>
+<tr>
+<th class="align-right">long column a</th>
+<th class="align-right">long column b</th>
+<th class="align-right">long column c</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td class="align-right">1</td>
+<td class="align-right">2</td>
+<td class="align-right">3</td>
+</tr>
+</tbody>
+</table>
+<figcaption>
+table with right aligned columns (because numbers)
+</figcaption>
+</figure>

org/testdata/tables.org

@@ -16,3 +16,14 @@
 #+CAPTION: table without header
 | 1 | 2 | 3 |
 
+#+CAPTION: table with aligned columns
+| left aligned | right aligned | center aligned |
+|--------------+---------------+----------------|
+| <l>          |           <r> |       <c>      |
+| 42           |            42 |       42       |
+| foobar       |        foobar |     foobar     |
+
+#+CAPTION: table with right aligned columns (because numbers)
+| long column a | long column b | long column c |
+|---------------+---------------+---------------|
+|             1 |             2 |             3 |