Implement --invert to remove tags matching selector.

parse.go

@@ -19,6 +19,7 @@ var (
 	pupPreformatted  bool          = false
 	pupPrintColor    bool          = false
 	pupEscapeHTML    bool          = true
+	pupInvertSelect  bool          = false
 	pupIndentString  string        = " "
 	pupDisplayer     Displayer     = TreeDisplayer{}
 )
@@ -56,6 +57,7 @@ Flags
     -n --number        print number of elements selected
     -l --limit         restrict number of levels printed
     -p --plain         don't escape html
+    -v --invert        invert selection (remove matching elements)
     --pre              preserve preformatted text
     --charset          specify the charset for pup to use
 `
@@ -88,6 +90,8 @@ func ProcessFlags(cmds []string) (nonFlagCmds []string, err error) {
 			pupPrintColor = true
 		case "-p", "--plain":
 			pupEscapeHTML = false
+		case "-v", "--invert":
+			pupInvertSelect = true
 		case "--pre":
 			pupPreformatted = true
 		case "-f", "--file":

pup.go

@@ -3,6 +3,8 @@ package main
 import (
 	"fmt"
 	"os"
+
+	"golang.org/x/net/html"
 )
 
 //      _=,_
@@ -34,4 +36,28 @@ func main() {
 		fmt.Fprintf(os.Stderr, "%s\n", err.Error())
 		os.Exit(2)
 	}
+
+}
+
+func RemoveInverseMatches(root *html.Node, selectedNodes []*html.Node) {
+	var remove []*html.Node
+	for node := root.FirstChild; node != nil; node = node.NextSibling {
+		if HtmlNodeInList(node, selectedNodes) {
+			remove = append(remove, node)
+		} else {
+			RemoveInverseMatches(node, selectedNodes)
+		}
+	}
+	for _, rm := range remove {
+		root.RemoveChild(rm)
+	}
+}
+
+func HtmlNodeInList(a *html.Node, list []*html.Node) bool {
+	for _, b := range list {
+		if b == a {
+			return true
+		}
+	}
+	return false
 }

pup_test.go

@@ -96,6 +96,7 @@ func TestMain(t *testing.T) {
 		50: {[]string{"table.infobox tr > th:first-child json{}"}},
 		51: {[]string{"table.infobox tr > *:first-child json{}"}},
 		52: {[]string{"frew text{}"}},
+		53: {[]string{"li", "-v"}},
 	}
 
 	for i, tt := range tests {

runselectors.go

@@ -49,7 +49,12 @@ func runSelectors(w io.Writer,  cmds []string, root *html.Node) error {
 		}
 	}
 	selectedNodes = append(selectedNodes, currNodes...)
-	pupDisplayer.Display(w, selectedNodes)
+	if pupInvertSelect {
+		RemoveInverseMatches(root, selectedNodes)
+		pupDisplayer.Display(w, []*html.Node{root})
+	} else {
+		pupDisplayer.Display(w, selectedNodes)
+	}
 
 	return nil
 }

tests/cmds.txt

@@ -0,0 +1,50 @@
+#footer
+#footer li
+#footer li + a
+#footer li + a attr{title}
+#footer li > li
+table li
+table li:first-child
+table li:first-of-type
+table li:last-child
+table li:last-of-type
+table a[title="The Practice of Programming"]
+table a[title="The Practice of Programming"] text{}
+json{}
+text{}
+.after-portlet
+.after
+:empty
+td:empty
+.navbox-list li:nth-child(1)
+.navbox-list li:nth-child(2)
+.navbox-list li:nth-child(3)
+.navbox-list li:nth-last-child(1)
+.navbox-list li:nth-last-child(2)
+.navbox-list li:nth-last-child(3)
+.navbox-list li:nth-child(n+1)
+.navbox-list li:nth-child(3n+1)
+.navbox-list li:nth-last-child(n+1)
+.navbox-list li:nth-last-child(3n+1)
+:only-child
+.navbox-list li:only-child
+.summary
+[class=summary]
+[class="summary"]
+#toc
+#toc li + a
+#toc li + a text{}
+#toc li + a json{}
+#toc li + a + span
+#toc li + span
+#toc li > li
+li a:not([rel])
+link, a
+link ,a
+link , a
+link , a sup
+link , a:parent-of(sup)
+link , a:parent-of(sup) sup
+li	--number
+li	-n
+li a	--invert

tests/expected_output.txt

@@ -0,0 +1,50 @@
+c00fef10d36c1166cb5ac886f9d25201b720e37e #footer
+a7bb8dbfdd638bacad0aa9dc3674126d396b74e2 #footer li
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #footer li + a
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #footer li + a attr{title}
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #footer li > li
+a92e50c09cd56970625ac3b74efbddb83b2731bb table li
+505c04a42e0084cd95560c233bd3a81b2c59352d table li:first-child
+505c04a42e0084cd95560c233bd3a81b2c59352d table li:first-of-type
+66950e746590d7f4e9cfe3d1adef42cd0addcf1d table li:last-child
+66950e746590d7f4e9cfe3d1adef42cd0addcf1d table li:last-of-type
+0a37d612cd4c67a42bd147b1edc5a1128456b017 table a[title="The Practice of Programming"]
+0d3918d54f868f13110262ffbb88cbb0b083057d table a[title="The Practice of Programming"] text{}
+199188dc8f1522426a628e41d96264bffb8beb0f json{}
+95ef88ded9dab22ee3206cca47b9c3a376274bda text{}
+e4f7358fbb7bb1748a296fa2a7e815fa7de0a08b .after-portlet
+da39a3ee5e6b4b0d3255bfef95601890afd80709 .after
+5b3020ba03fb43f7cdbcb3924546532b6ec9bd71 :empty
+3406ca0f548d66a7351af5411ce945cf67a2f849 td:empty
+30fff0af0b1209f216d6e9124e7396c0adfa0758 .navbox-list li:nth-child(1)
+a38e26949f047faab5ea7ba2acabff899349ce03 .navbox-list li:nth-child(2)
+d954831229a76b888e85149564727776e5a2b37a .navbox-list li:nth-child(3)
+d314e83b059bb876b0e5ee76aa92d54987961f9a .navbox-list li:nth-last-child(1)
+1f19496e239bca61a1109dbbb8b5e0ab3e302b50 .navbox-list li:nth-last-child(2)
+1ec9ebf14fc28c7d2b13e81241a6d2e1608589e8 .navbox-list li:nth-last-child(3)
+52e726f0993d2660f0fb3ea85156f6fbcc1cfeee .navbox-list li:nth-child(n+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)
+6c45ee6bca361b8a9baee50a15f575fc6ac73adc :only-child
+44c99f6ad37b65dc0893cdcb1c60235d827ee73e .navbox-list li:only-child
+641037814e358487d1938fc080e08f72a3846ef8 .summary
+641037814e358487d1938fc080e08f72a3846ef8 [class=summary]
+641037814e358487d1938fc080e08f72a3846ef8 [class="summary"]
+613bf65ac4042b6ee0a7a47f08732fdbe1b5b06b #toc
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a text{}
+cd0d4cc32346750408f7d4f5e78ec9a6e5b79a0d #toc li + a json{}
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a + span
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + span
+da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li > li
+87eee1189dd5296d6c010a1ad329fc53c6099d72 li a:not([rel])
+055f3c98e9160beb13f72f1009ad66b6252a9bba link, a
+055f3c98e9160beb13f72f1009ad66b6252a9bba link ,a
+055f3c98e9160beb13f72f1009ad66b6252a9bba link , a
+0d1f66765d1632c70f8608947890524e78459362 link , a sup
+b6a3d6cccd305fcc3e8bf2743c443743bdaaa02b link , a:parent-of(sup)
+0d1f66765d1632c70f8608947890524e78459362 link , a:parent-of(sup) sup
+d200756078ef79d77fb17848c5f15b85bcea7e00 li	--number
+d200756078ef79d77fb17848c5f15b85bcea7e00 li	-n
+1fa074385374bcd613394678b5e08ed1ad4d5d9d li a	--invert