add optional list support

Mario Hros · Mario Hros · commit b9e0d6e2894c · 2023-05-10T22:21:32.000+02:00
diff --git a/html2text.go b/html2text.go
@@ -24,6 +24,7 @@ var numericEntityRE = regexp.MustCompile(`(?i)^#(x?[a-f0-9]+)$`)
 type options struct {
 	lbr            string
 	linksInnerText bool
+	listSupport    bool
 }
 
 func newOptions() *options {
@@ -51,6 +52,13 @@ func WithLinksInnerText() Option {
 	}
 }
 
+// WithListSupport formats <ul> and <li> lists with dashes
+func WithListSupport() Option {
+	return func(o *options) {
+		o.listSupport = true
+	}
+}
+
 func parseHTMLEntity(entName string) (string, bool) {
 	if r, ok := entity[entName]; ok {
 		return string(r), true
@@ -231,10 +239,14 @@ func HTML2TextWithOptions(html string, reqOpts ...Option) string {
 			tag := html[tagStart:i]
 			tagNameLowercase := strings.ToLower(tag)
 
-			if tagNameLowercase == "/ul" {
+			if tagNameLowercase == "/ul" || tagNameLowercase == "/ol" {
 				outBuf.WriteString(opts.lbr)
 			} else if tagNameLowercase == "li" || tagNameLowercase == "li/" {
-				outBuf.WriteString(opts.lbr)
+				if opts.listSupport {
+					outBuf.WriteString(opts.lbr + "- ")
+				} else {
+					outBuf.WriteString(opts.lbr)
+				}
 			} else if headersRE.MatchString(tagNameLowercase) {
 				if canPrintNewline {
 					outBuf.WriteString(opts.lbr + opts.lbr)
diff --git a/html2text_test.go b/html2text_test.go
@@ -102,6 +102,7 @@ func TestHTML2Text(t *testing.T) {
 			So(HTML2Text(``), ShouldEqual, "")
 			So(HTML2Text(`<html><head><title>Good</title></head><body>x</body>`), ShouldEqual, "x")
 			So(HTML2Text(`<html><head href="foo"><title>Good</title></head><body>x</body>`), ShouldEqual, "x")
+			So(HTML2Text(`<htMl><hEad><titLe>Good</Title></head><boDy>x</Body>`), ShouldEqual, "x")
 			So(HTML2Text(`we are not <script type="javascript"></script>interested in scripts`),
 				ShouldEqual, "we are not interested in scripts")
 		})
@@ -122,6 +123,15 @@ func TestHTML2Text(t *testing.T) {
 			So(HTML2TextWithOptions(`<p>two</p><p>paragraphs</p>`), ShouldEqual, "two\r\n\r\nparagraphs")
 		})
 
+		Convey("No list support by default (original behavior)", func() {
+			So(HTML2Text(`list of items<ul><li>One</li><li>Two</li><li>Three</li></ul>`), ShouldEqual, "list of items\r\nOne\r\nTwo\r\nThree\r\n")
+		})
+
+		Convey("Optional list support", func() {
+			So(HTML2TextWithOptions(`list of items<ul><li>One</li><li>Two</li><li>Three</li></ul>`, WithListSupport()), ShouldEqual, "list of items\r\n- One\r\n- Two\r\n- Three\r\n")
+			So(HTML2TextWithOptions(`list of items<ol><li>One</li><li>Two</li><li>Three</li></ol>`, WithListSupport()), ShouldEqual, "list of items\r\n- One\r\n- Two\r\n- Three\r\n")
+		})
+
 		Convey("Custom HTML Tags", func() {
 			So(HTML2Text(`<aa>hello</aa>`), ShouldEqual, "hello")
 			So(HTML2Text(`<aa >hello</aa>`), ShouldEqual, "hello")
