- [+] implement excluding control. close #6

Author: suntong

html2md_test.go

@@ -12,6 +12,17 @@ const (
 
 	boldText   = "<strong>Bold Text</strong>"
 	boldEscape = "<strong>option src_ip</strong>"
+
+	excludingControlTest = `<div class="outter-class">
+        <h1 class="inner-class">
+        <div class="other-class3">
+            <h3>Some heading i don't need</h3>
+        </div>
+        The string I need
+        <span class="other-class" >Some value I don't need</span>
+        <span class="other-class2" title="sometitle"></span>
+        </h1>
+    </div>`
 )
 
 type testCase struct {
@@ -45,6 +56,20 @@ func TestExec(t *testing.T) {
 			"Only <del>blue ones</del> <s> left</s>",
 			[]string{"-i", "--plugin-strikethrough"},
 		},
+		{
+			"ExclChildren", "The string I need", excludingControlTest,
+			[]string{"-s", "h1", "--xc", "-i"},
+		},
+		{
+			"Excl1", `### Some heading i don't need
+
+ The string I need`, excludingControlTest,
+			[]string{"-s", "h1", "-xspan", "-i"},
+		},
+		{
+			"Excl2", "The string I need", excludingControlTest,
+			[]string{"-s", "h1", "-x", "span", "-xdiv", "-i"},
+		},
 		// {
 		// 	"", "", "", []string{"-i"},
 		// },

prop_html2md.go

@@ -35,6 +35,17 @@ func Html2md(ctx *cli.Context) error {
 	doc, err := goquery.NewDocumentFromReader(rootArgv.Filei)
 	clis.AbortOn("Reading file with goquery", err)
 	content := doc.Find(rootArgv.Sel)
+	if rootArgv.ExclChildren {
+		content = content.Children().Remove().End()
+		// h, _ := goquery.OuterHtml(content)
+		// clis.Verbose(3, "%#v\n", h)
+	} else if len(rootArgv.Excl) != 0 {
+		for _, ex := range rootArgv.Excl {
+			content = content.Find(ex).Remove().End()
+			h, _ := content.Html()
+			clis.Verbose(5, "%#v\n", h)
+		}
+	}
 
 	domain, url := rootArgv.Domain, rootArgv.Filei.Name()
 	if domain == "" && regexp.MustCompile(`(?i)^http`).MatchString(url) {