add cmd/printast and code to pretty-print ast

ast/node.go

@@ -1,7 +1,6 @@
 package ast
 
 import (
-	"bytes"
 	"fmt"
 )
 
@@ -462,37 +461,3 @@ func WalkFunc(n Node, f NodeVisitorFunc) {
 	visitor := NodeVisitorFunc(f)
 	Walk(n, visitor)
 }
-
-func dump(ast Node) {
-	fmt.Println(dumpString(ast))
-}
-
-func getContent(node Node) []byte {
-	if c := node.AsContainer(); c != nil {
-		if c.Literal != nil {
-			return c.Literal
-		}
-		return c.Content
-	}
-	return nil
-}
-
-func dumpR(ast Node, depth int) string {
-	if ast == nil {
-		return ""
-	}
-	indent := bytes.Repeat([]byte("\t"), depth)
-	content := ast.AsContainer().Literal
-	if content == nil {
-		content = ast.AsContainer().Content
-	}
-	result := fmt.Sprintf("%s%T(%q)\n", indent, ast, content)
-	for _, child := range ast.GetChildren() {
-		result += dumpR(child, depth+1)
-	}
-	return result
-}
-
-func dumpString(ast Node) string {
-	return dumpR(ast, 0)
-}

ast/print.go

@@ -0,0 +1,90 @@
+package ast
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"strings"
+)
+
+// Print is for debugging. It prints a string representation of parsed
+// markdown doc (result of parser.Parse()) to dst.
+//
+// To make output readable, it shortens text output.
+func Print(dst io.Writer, doc Node) {
+	PrintWithPrefix(dst, doc, "  ")
+}
+
+// PrintWithPrefix is like Pring but allows customizing prefix used for
+// indentation. By default it's 2 spaces. You can change it to e.g. tab
+// by passing "\t"
+func PrintWithPrefix(w io.Writer, doc Node, prefix string) {
+	printRecur(w, doc, prefix, 0)
+}
+
+// ToString is like Dump but returns result as a string
+func ToString(doc Node) string {
+	var buf bytes.Buffer
+	Print(&buf, doc)
+	return buf.String()
+}
+
+func contentToString(d1 []byte, d2 []byte) string {
+	if d1 != nil {
+		return string(d1)
+	}
+	if d2 != nil {
+		return string(d2)
+	}
+	return ""
+}
+
+func getContent(node Node) string {
+	if c := node.AsContainer(); c != nil {
+		return contentToString(c.Literal, c.Content)
+	}
+	leaf := node.AsLeaf()
+	return contentToString(leaf.Literal, leaf.Content)
+}
+
+func shortenString(s string, maxLen int) string {
+	// for cleaner, one-line ouput, replace some white-space chars
+	// with their escaped version
+	s = strings.Replace(s, "\n", `\n`, -1)
+	s = strings.Replace(s, "\r", `\r`, -1)
+	s = strings.Replace(s, "\t", `\t`, -1)
+	if maxLen < 0 {
+		return s
+	}
+	if len(s) < maxLen {
+		return s
+	}
+	// add "..." to indicate truncation
+	return s[:maxLen-3] + "..."
+}
+
+// get a short name of the type of v which excludes package name
+// and strips "()" from the end
+func getNodeType(node Node) string {
+	s := fmt.Sprintf("%T", node)
+	s = strings.TrimSuffix(s, "()")
+	if idx := strings.Index(s, "."); idx != -1 {
+		return s[idx+1:]
+	}
+	return s
+}
+
+func printRecur(w io.Writer, node Node, prefix string, depth int) {
+	if node == nil {
+		return
+	}
+	indent := strings.Repeat(prefix, depth)
+	io.WriteString(w, indent)
+
+	content := shortenString(getContent(node), 40)
+	typeName := getNodeType(node)
+	fmt.Fprintf(w, "%s%s '%s'\n", indent, typeName, content)
+	for _, child := range node.GetChildren() {
+		printRecur(w, child, prefix, depth+1)
+	}
+}

cmd/printast/main.go

@@ -0,0 +1,37 @@
+package main
+
+import (
+	"fmt"
+	"io/ioutil"
+	"os"
+
+	"github.com/gomarkdown/markdown"
+	"github.com/gomarkdown/markdown/ast"
+)
+
+// This prints AST of parsed markdown document.
+// Usage: printast <markdown-file>
+
+func usageAndExit() {
+	fmt.Printf("Usage: printast <markdown-file>\n")
+	os.Exit(1)
+}
+
+func main() {
+	nFiles := len(os.Args) - 1
+	if nFiles < 1 {
+		usageAndExit()
+	}
+	for i := 0; i < nFiles; i++ {
+		fileName := os.Args[i+1]
+		d, err := ioutil.ReadFile(fileName)
+		if err != nil {
+			fmt.Fprintf(os.Stderr, "Couldn't open '%s', error: '%s'\n", fileName, err)
+			continue
+		}
+		doc := markdown.Parse(d, nil)
+		fmt.Printf("Ast of file '%s':\n", fileName)
+		ast.Print(os.Stdout, doc)
+		fmt.Print("\n")
+	}
+}