Add simple table parser

README.md

@@ -0,0 +1,19 @@
+# Go Markdown Parser
+
+Markdown parser engine written in go.
+
+## Supports
+
+* H1 (#)
+* H2 (##)
+* H3 (###)
+* H4 (####)
+* H5 (#####)
+* H6 (######)
+* Code block starting with ``` and ~~~
+* Paragraph
+
+## To Do
+
+* Table
+* Blockquote

element.go

@@ -2,6 +2,7 @@ package parser
 
 import (
 	"regexp"
+	"strings"
 )
 
 // ElementHierarchy provides hierachical structure
@@ -117,3 +118,64 @@ func tryCode(block string) (string, bool) {
 	}
 	return "", false
 }
+
+func tryTable(block string) ([][]string, bool) {
+	output := [][]string{}
+
+	lines := strings.Split(block, "\n")
+	if len(lines) < 3 {
+		return nil, false
+	}
+
+	// check header separator
+	// header separator for one column
+	patSep := "^|\\s+:?---+?\\s+|$"
+	reSep := regexp.MustCompile(patSep)
+	if !reSep.MatchString(lines[1]) {
+		// header separator does not present
+		return nil, false
+	}
+
+	colCount := columnCount(lines[1])
+
+	if colCount != columnCount(lines[0]) {
+		// header and column count does not match
+		return nil, false
+	}
+
+	row := []string{}
+	for i := 0; i < colCount; i++ {
+		row = append(row, getCellValue(i, lines[0]))
+	}
+	output = append(output, row)
+
+	for i, n := 2, len(lines); i < n; i++ {
+		row = []string{}
+		for j := 0; j < colCount; j++ {
+			row = append(row, getCellValue(j, lines[i]))
+		}
+		output = append(output, row)
+	}
+
+	return output, true
+}
+
+func getCellValue(index int, line string) string {
+	cols := strings.Split(line, "|")
+	if cols[0] == "" { // detect left boundary pipe
+		cols = cols[1:]
+	}
+	if cols[len(cols)-1] == "" { // detect right boundary pipe
+		cols = cols[0 : len(cols)-1]
+	}
+
+	return strings.TrimSpace(cols[index])
+}
+
+func columnCount(line string) int {
+	colCount := strings.Count(line, "|") - 1
+	if line[0] != '|' || line[len(line)-1] != '|' {
+		colCount++
+	}
+	return colCount
+}

element_test.go

@@ -108,3 +108,19 @@ func TestTryAlternativeCodeBlock(t *testing.T) {
 	assert.True(t, success)
 	assert.Equal(t, "Content\nWith\nNew Line", text)
 }
+
+func TestTry1ColumnTable(t *testing.T) {
+	content := "| Header |\n| --- |\n| Body |"
+
+	text, success := tryTable(content)
+
+	assert.True(t, success)
+	assert.Equal(t, [][]string{
+		[]string{
+			"Header",
+		},
+		[]string{
+			"Body",
+		},
+	}, text)
+}

tokenizer_test.go

@@ -125,3 +125,13 @@ func TestTokenizeSimpleDataWith1AlternativeCodeBlock(t *testing.T) {
 
 	assert.Equal(t, expected, result)
 }
+
+func TestTokenizeTableData(t *testing.T) {
+	content := "| Header |\n| - |\n| Body |"
+	expected := []string{"| Header |\n| - |\n| Body |"}
+	tokenizer := NewTokenizer()
+
+	result := tokenizer.Tokenize(content)
+
+	assert.Equal(t, expected, result)
+}