Merge be5dee7868 into 5a57cf1113

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
     --version          display version
@@ -89,6 +91,8 @@ 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

@@ -76,5 +76,33 @@ func main() {
 		}
 	}
 	selectedNodes = append(selectedNodes, currNodes...)
-	pupDisplayer.Display(selectedNodes)
+	if pupInvertSelect {
+		RemoveInverseMatches(root, selectedNodes)
+		pupDisplayer.Display([]*html.Node{root})
+	} else {
+		pupDisplayer.Display(selectedNodes)
+	}
+}
+
+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
 }

tests/cmds.txt

@@ -45,5 +45,6 @@ link , a
 link , a sup
 link , a:parent-of(sup)
 link , a:parent-of(sup) sup
-li --number
-li -n
+li	--number
+li	-n
+li a	--invert

tests/expected_output.txt

@@ -10,7 +10,7 @@ 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{}
-ecb542a30fc75c71a0c6380692cbbc4266ccbce4 json{}
+199188dc8f1522426a628e41d96264bffb8beb0f json{}
 95ef88ded9dab22ee3206cca47b9c3a376274bda text{}
 e4f7358fbb7bb1748a296fa2a7e815fa7de0a08b .after-portlet
 da39a3ee5e6b4b0d3255bfef95601890afd80709 .after
@@ -34,7 +34,7 @@ d314e83b059bb876b0e5ee76aa92d54987961f9a .navbox-list li:nth-last-child(1)
 613bf65ac4042b6ee0a7a47f08732fdbe1b5b06b #toc
 da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a
 da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a text{}
-97d170e1550eee4afc0af065b78cda302a97674c #toc li + a json{}
+cd0d4cc32346750408f7d4f5e78ec9a6e5b79a0d #toc li + a json{}
 da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a + span
 da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + span
 da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li > li
@@ -45,5 +45,6 @@ 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
+d200756078ef79d77fb17848c5f15b85bcea7e00 li	--number
+d200756078ef79d77fb17848c5f15b85bcea7e00 li	-n
+1fa074385374bcd613394678b5e08ed1ad4d5d9d li a	--invert

tests/run.py

@@ -8,7 +8,7 @@ data = open("index.html", "r").read()
 
 for line in open("cmds.txt", "r"):
     line = line.strip()
-    p = Popen(['pup', line], stdout=PIPE, stdin=PIPE, stderr=PIPE)
+    p = Popen(['pup']+line.split('\t'), stdout=PIPE, stdin=PIPE, stderr=PIPE)
     h = sha1()
     h.update(p.communicate(input=data)[0])
     print("%s %s" % (h.hexdigest(), line))