diff options
| author | Bjørn Erik Pedersen <bjorn.erik.pedersen@gmail.com> | 2023-02-06 17:29:12 +0100 |
|---|---|---|
| committer | Bjørn Erik Pedersen <bjorn.erik.pedersen@gmail.com> | 2023-02-07 09:11:02 +0100 |
| commit | d33a7ebcc16e804f1db0dc1f1edad4d9f9e816ef (patch) | |
| tree | 93953221b8e208b4522d58db18bfba6b6edc47f6 /publisher | |
| parent | 2a364cca6487e7ecc1982c8ade2927d18c592cee (diff) | |
Make the HTML collector parsing more robust
Most notably better handling self-closing elements
Closes #10698
Diffstat (limited to 'publisher')
| -rw-r--r-- | publisher/htmlElementsCollector.go | 17 | ||||
| -rw-r--r-- | publisher/htmlElementsCollector_test.go | 4 |
2 files changed, 17 insertions, 4 deletions
diff --git a/publisher/htmlElementsCollector.go b/publisher/htmlElementsCollector.go index 91e1237a9..c3b88c4cc 100644 --- a/publisher/htmlElementsCollector.go +++ b/publisher/htmlElementsCollector.go @@ -294,9 +294,10 @@ func htmlLexElementStart(w *htmlElementsCollectorWriter) htmlCollectorStateFunc } tagName := w.buff.Bytes()[1:] + isSelfClosing := tagName[len(tagName)-1] == '/' switch { - case skipInnerElementRe.Match(tagName): + case !isSelfClosing && skipInnerElementRe.Match(tagName): // pre, script etc. We collect classes etc. on the surrounding // element, but skip the inner content. w.backup() @@ -432,10 +433,18 @@ func parseStartTag(s string) string { }) if spaceIndex == -1 { - return s[1 : len(s)-1] + s = s[1 : len(s)-1] + } else { + s = s[1:spaceIndex] } - return s[1:spaceIndex] + if s[len(s)-1] == '/' { + // Self closing. + s = s[:len(s)-1] + } + + return s + } // isClosedByTag reports whether b ends with a closing tag for tagName. @@ -487,7 +496,7 @@ LOOP: } } - if state != 2 { + if state != 2 || lo >= hi { return false } diff --git a/publisher/htmlElementsCollector_test.go b/publisher/htmlElementsCollector_test.go index 11590e0a3..f9c9424cb 100644 --- a/publisher/htmlElementsCollector_test.go +++ b/publisher/htmlElementsCollector_test.go @@ -110,6 +110,9 @@ func TestClassCollector(t *testing.T) { {"DOCTYPE should beskipped", `<!DOCTYPE html>`, f("", "", "")}, {"Comments should be skipped", `<!-- example comment -->`, f("", "", "")}, {"Comments with elements before and after", `<div></div><!-- example comment --><span><span>`, f("div span", "", "")}, + {"Self closing tag", `<div><hr/></div>`, f("div hr", "", "")}, + // svg with self closing style tag. + {"SVG with self closing style tag", `<svg><style/><g><path class="foo"/></g></svg>`, f("g path style svg", "foo", "")}, // Issue #8530 {"Comment with single quote", `<!-- Hero Area Image d'accueil --><i class="foo">`, f("i", "foo", "")}, {"Uppercase tags", `<DIV></DIV>`, f("div", "", "")}, @@ -174,6 +177,7 @@ func TestEndsWithTag(t *testing.T) { {"match space", "foo< / div>", "div", true}, {"match space 2", "foo< / div \n>", "div", true}, {"match case", "foo</DIV>", "div", true}, + {"self closing", `</defs><g><g><path fill="#010101" d=asdf"/>`, "div", false}, } { c.Run(test.name, func(c *qt.C) { got := isClosedByTag([]byte(test.s), []byte(test.tagName)) |