strip dots at the end of a URL

stapelberg · stapelberg · commit 74fb94d77a43 · 2022-04-22T08:41:56.000+02:00
fixes #140
diff --git a/internal/convert/convert.go b/internal/convert/convert.go
@@ -115,6 +115,15 @@ func findUrls(txt string) [][]int {
 		return end
 	}
 
+	// A dot at the end of a URL is most likely not part of the URL, but part of
+	// the text. https://github.com/Debian/debiman/issues/140
+	maybeStripDot := func(end int) int {
+		if txt[end-1] == '.' {
+			end--
+		}
+		return end
+	}
+
 Outer:
 	for i, r := range txt {
 		// As per https://stackoverflow.com/a/1547940/712014:
@@ -150,6 +159,7 @@ Outer:
 	}
 	if inUrl {
 		end := maybeStripParens(len(txt))
+		end = maybeStripDot(end)
 		results = append(results, []int{lastWordBoundary + 1, end})
 	}
 	return results
diff --git a/internal/convert/convert_test.go b/internal/convert/convert_test.go
@@ -450,3 +450,38 @@ func TestXrefHrefExclude(t *testing.T) {
 		t.Fatalf("Unexpected xref() HTML result: %v", err)
 	}
 }
+
+func TestXrefHrefExcludeDot(t *testing.T) {
+	input := &html.Node{
+		Type: html.TextNode,
+		Data: "the upstream website goes into more detail at http://debian.org/.",
+	}
+
+	a1 := &html.Node{
+		Type: html.ElementNode,
+		Data: "a",
+		Attr: []html.Attribute{
+			{Key: "href", Val: "http://debian.org/"},
+		},
+	}
+	a1.AppendChild(&html.Node{
+		Type: html.TextNode,
+		Data: "http://debian.org/",
+	})
+
+	want := []*html.Node{
+		&html.Node{
+			Type: html.TextNode,
+			Data: "the upstream website goes into more detail at ",
+		},
+		a1,
+		&html.Node{
+			Type: html.TextNode,
+			Data: ".",
+		},
+	}
+	got := xref(input.Data, func(ref string) string { return ref })
+	if err := cmpElems(input, got, want); err != nil {
+		t.Fatalf("Unexpected xref() HTML result: %v", err)
+	}
+}
