Merge pull request #32 from EricChiang/0.3.8

0.3.8

display.go

@@ -6,10 +6,10 @@ import (
 	"regexp"
 	"strings"
 
-	"golang.org/x/net/html"
-	"golang.org/x/net/html/atom"
 	"github.com/fatih/color"
 	"github.com/mattn/go-colorable"
+	"golang.org/x/net/html"
+	"golang.org/x/net/html/atom"
 )
 
 func init() {
@@ -70,7 +70,13 @@ func (t TreeDisplayer) Display(nodes []*html.Node) {
 func (t TreeDisplayer) printNode(n *html.Node, level int) {
 	switch n.Type {
 	case html.TextNode:
-		s := html.EscapeString(n.Data)
+		s := n.Data
+		if pupEscapeHTML {
+			// don't escape javascript
+			if n.Parent == nil || n.Parent.DataAtom != atom.Script {
+				s = html.EscapeString(s)
+			}
+		}
 		s = strings.TrimSpace(s)
 		if s != "" {
 			t.printIndent(level)
@@ -85,7 +91,10 @@ func (t TreeDisplayer) printNode(n *html.Node, level int) {
 			fmt.Printf("<%s", n.Data)
 		}
 		for _, a := range n.Attr {
-			val := html.EscapeString(a.Val)
+			val := a.Val
+			if pupEscapeHTML {
+				val = html.EscapeString(val)
+			}
 			if pupPrintColor {
 				fmt.Print(" ")
 				attrKeyColor.Printf("%s", a.Key)
@@ -113,10 +122,14 @@ func (t TreeDisplayer) printNode(n *html.Node, level int) {
 		}
 	case html.CommentNode:
 		t.printIndent(level)
+		data := n.Data
+		if pupEscapeHTML {
+			data = html.EscapeString(data)
+		}
 		if pupPrintColor {
-			commentColor.Printf("<!--%s-->\n", n.Data)
+			commentColor.Printf("<!--%s-->\n", data)
 		} else {
-			fmt.Printf("<!--%s-->\n", n.Data)
+			fmt.Printf("<!--%s-->\n", data)
 		}
 		t.printChildren(n, level)
 	case html.DoctypeNode, html.DocumentNode:
@@ -151,7 +164,14 @@ type TextDisplayer struct{}
 func (t TextDisplayer) Display(nodes []*html.Node) {
 	for _, node := range nodes {
 		if node.Type == html.TextNode {
-			fmt.Println(node.Data)
+			data := node.Data
+			if pupEscapeHTML {
+				// don't escape javascript
+				if node.Parent == nil || node.Parent.DataAtom != atom.Script {
+					data = html.EscapeString(data)
+				}
+			}
+			fmt.Println(data)
 		}
 		children := []*html.Node{}
 		child := node.FirstChild
@@ -173,7 +193,10 @@ func (a AttrDisplayer) Display(nodes []*html.Node) {
 		attributes := node.Attr
 		for _, attr := range attributes {
 			if attr.Key == a.Attr {
-				val := html.EscapeString(attr.Val)
+				val := attr.Val
+				if pupEscapeHTML {
+					val = html.EscapeString(val)
+				}
 				fmt.Printf("%s\n", val)
 			}
 		}
@@ -188,7 +211,11 @@ func jsonify(node *html.Node) map[string]interface{} {
 	vals := map[string]interface{}{}
 	if len(node.Attr) > 0 {
 		for _, attr := range node.Attr {
-			vals[attr.Key] = html.EscapeString(attr.Val)
+			if pupEscapeHTML {
+				vals[attr.Key] = html.EscapeString(attr.Val)
+			} else {
+				vals[attr.Key] = attr.Val
+			}
 		}
 	}
 	vals["tag"] = node.DataAtom.String()
@@ -200,6 +227,12 @@ func jsonify(node *html.Node) map[string]interface{} {
 		case html.TextNode:
 			text := strings.TrimSpace(child.Data)
 			if text != "" {
+				if pupEscapeHTML {
+					// don't escape javascript
+					if node.DataAtom != atom.Script {
+						text = html.EscapeString(text)
+					}
+				}
 				// if there is already text we'll append it
 				currText, ok := vals["text"]
 				if ok {
@@ -209,6 +242,9 @@ func jsonify(node *html.Node) map[string]interface{} {
 			}
 		case html.CommentNode:
 			comment := strings.TrimSpace(child.Data)
+			if pupEscapeHTML {
+				comment = html.EscapeString(comment)
+			}
 			currComment, ok := vals["comment"]
 			if ok {
 				comment = fmt.Sprintf("%s %s", currComment, comment)
@@ -252,3 +288,10 @@ func (j JSONDisplayer) Display(nodes []*html.Node) {
 	}
 	fmt.Printf("%s", data)
 }
+
+// Print the number of features returned
+type NumDisplayer struct{}
+
+func (d NumDisplayer) Display(nodes []*html.Node) {
+	fmt.Println(len(nodes))
+}

parse.go

@@ -17,6 +17,7 @@ var (
 	pupCharset       string        = ""
 	pupMaxPrintLevel int           = -1
 	pupPrintColor    bool          = false
+	pupEscapeHTML    bool          = true
 	pupIndentString  string        = " "
 	pupDisplayer     Displayer     = TreeDisplayer{}
 )
@@ -53,6 +54,7 @@ Flags
     -i --indent        number of spaces to use for indent or character
     -n --number        print number of elements selected
     -l --limit         restrict number of levels printed
+    -p --plain         don't escape html
     --charset          specify the charset for pup to use
     --version          display version
 `
@@ -83,6 +85,8 @@ func ProcessFlags(cmds []string) (nonFlagCmds []string, err error) {
 		switch cmd {
 		case "-c", "--color":
 			pupPrintColor = true
+		case "-p", "--plain":
+			pupEscapeHTML = false
 		case "-f", "--file":
 			filename := cmds[i+1]
 			pupIn, err = os.Open(filename)
@@ -113,6 +117,8 @@ func ProcessFlags(cmds []string) (nonFlagCmds []string, err error) {
 		case "--version":
 			fmt.Println(VERSION)
 			os.Exit(0)
+		case "-n", "--number":
+			pupDisplayer = NumDisplayer{}
 		default:
 			if cmd[0] == '-' {
 				return []string{}, fmt.Errorf("Unrecognized flag '%s'", cmd)

pup.go

@@ -16,7 +16,7 @@ import (
 //       |/ \_( # |"
 //      C/ ,--___/
 
-var VERSION string = "0.3.7"
+var VERSION string = "0.3.8"
 
 func main() {
 	// process flags and arguments

tests/cmds.txt

@@ -45,3 +45,5 @@ link , a
 link , a sup
 link , a:parent-of(sup)
 link , a:parent-of(sup) sup
+li --number
+li -n

tests/expected_output.txt

@@ -10,8 +10,8 @@ a92e50c09cd56970625ac3b74efbddb83b2731bb table li
 66950e746590d7f4e9cfe3d1adef42cd0addcf1d table li:last-of-type
 0a37d612cd4c67a42bd147b1edc5a1128456b017 table a[title="The Practice of Programming"]
 0d3918d54f868f13110262ffbb88cbb0b083057d table a[title="The Practice of Programming"] text{}
-48a00be8203baea11bf9672bf10b183cca98c3b2 json{}
-74fa5ddf9687041ece8aed54913cc2d1a6e7101c text{}
+ecb542a30fc75c71a0c6380692cbbc4266ccbce4 json{}
+95ef88ded9dab22ee3206cca47b9c3a376274bda text{}
 e4f7358fbb7bb1748a296fa2a7e815fa7de0a08b .after-portlet
 da39a3ee5e6b4b0d3255bfef95601890afd80709 .after
 5b3020ba03fb43f7cdbcb3924546532b6ec9bd71 :empty
@@ -26,7 +26,7 @@ d314e83b059bb876b0e5ee76aa92d54987961f9a .navbox-list li:nth-last-child(1)
 0b20c98650efa5df39d380fea8d5b43f3a08cb66 .navbox-list li:nth-child(3n+1)
 52e726f0993d2660f0fb3ea85156f6fbcc1cfeee .navbox-list li:nth-last-child(n+1)
 972973fe1e8f63e4481c8641d6169c638a528a6e .navbox-list li:nth-last-child(3n+1)
-a55bb21d37fbbd3f8f1b551126795fbc733451fe :only-child
+6c45ee6bca361b8a9baee50a15f575fc6ac73adc :only-child
 44c99f6ad37b65dc0893cdcb1c60235d827ee73e .navbox-list li:only-child
 641037814e358487d1938fc080e08f72a3846ef8 .summary
 641037814e358487d1938fc080e08f72a3846ef8 [class=summary]
@@ -45,3 +45,5 @@ da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li > li
 0d1f66765d1632c70f8608947890524e78459362 link , a sup
 b6a3d6cccd305fcc3e8bf2743c443743bdaaa02b link , a:parent-of(sup)
 0d1f66765d1632c70f8608947890524e78459362 link , a:parent-of(sup) sup
+da39a3ee5e6b4b0d3255bfef95601890afd80709 li --number
+da39a3ee5e6b4b0d3255bfef95601890afd80709 li -n