diff options
| author | Ryan Martinsen <rmartinsen@about.com> | 2014-12-18 22:59:39 +0300 |
|---|---|---|
| committer | bep <bjorn.erik.pedersen@gmail.com> | 2014-12-26 18:24:28 +0300 |
| commit | 7dacc999f8c0bb2554ddae7a256928b4c499403b (patch) | |
| tree | 24607e8a66ba172ecd2206b188fe172a45afb2c7 /transform | |
| parent | f5a3fb149fff3a44c2b3be8a0c46d02594d755c8 (diff) | |
Add ability to canonify URLs in rendered XML output.
Diffstat (limited to 'transform')
| -rw-r--r-- | transform/absurl.go | 25 | ||||
| -rw-r--r-- | transform/chain_test.go | 14 |
2 files changed, 39 insertions, 0 deletions
diff --git a/transform/absurl.go b/transform/absurl.go index 6fb1cb4a1..0a0cd7239 100644 --- a/transform/absurl.go +++ b/transform/absurl.go @@ -31,6 +31,31 @@ func AbsURL(absURL string) (trs []link, err error) { return } +func AbsURLInXML(absURL string) (trs []link, err error) { + var baseURL *url.URL + + if baseURL, err = url.Parse(absURL); err != nil { + return + } + + base := strings.TrimRight(baseURL.String(), "/") + + var ( + srcedq = []byte(" src="" + base + "/") + hrefedq = []byte(" href="" + base + "/") + srcesq = []byte(" src='" + base + "/") + hrefesq = []byte(" href='" + base + "/") + ) + trs = append(trs, func(content []byte) []byte { + content = guardReplace(content, []byte(" src="//"), []byte(" src="/"), srcedq) + content = guardReplace(content, []byte(" src='//"), []byte(" src='/"), srcesq) + content = guardReplace(content, []byte(" href="//"), []byte(" href="/"), hrefedq) + content = guardReplace(content, []byte(" href='//"), []byte(" href='/"), hrefesq) + return content + }) + return +} + func guardReplace(content, guard, match, replace []byte) []byte { if !bytes.Contains(content, guard) { content = bytes.Replace(content, match, replace, -1) diff --git a/transform/chain_test.go b/transform/chain_test.go index 94135f6d7..936a6a1ad 100644 --- a/transform/chain_test.go +++ b/transform/chain_test.go @@ -9,10 +9,18 @@ const H5_JS_CONTENT_ABS_URL_WITH_NAV = "<!DOCTYPE html><html><head><script src=\ const CORRECT_OUTPUT_SRC_HREF_WITH_NAV = "<!DOCTYPE html><html><head><script src=\"http://two/foobar.js\"></script></head><body><nav><ul><li hugo-nav=\"section_0\"></li><li hugo-nav=\"section_1\"></li></ul></nav><article>content <a href=\"http://two/foobar\">foobar</a>. Follow up</article></body></html>" +const H5_XML_CONTENT_ABS_URL = "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"yes\" ?><feed xmlns=\"http://www.w3.org/2005/Atom\"><entry><content type=\"html\"><p><a href="/foobar">foobar</a></p> <p>A video: <iframe src='/foo'></iframe></p></content></entry></feed>" + +const CORRECT_OUTPUT_SRC_HREF_IN_XML = "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"yes\" ?><feed xmlns=\"http://www.w3.org/2005/Atom\"><entry><content type=\"html\"><p><a href="http://xml/foobar">foobar</a></p> <p>A video: <iframe src='http://xml/foo'></iframe></p></content></entry></feed>" + var two_chain_tests = []test{ {H5_JS_CONTENT_ABS_URL_WITH_NAV, CORRECT_OUTPUT_SRC_HREF_WITH_NAV}, } +var xml_abs_url_tests = []test{ + {H5_XML_CONTENT_ABS_URL, CORRECT_OUTPUT_SRC_HREF_IN_XML}, +} + func TestChainZeroTransformers(t *testing.T) { tr := NewChain() in := new(bytes.Buffer) @@ -31,3 +39,9 @@ func BenchmarkChain(b *testing.B) { apply(b.Errorf, tr, two_chain_tests) } } + +func TestXMLAbsUrl(t *testing.T) { + absURLInXML, _ := AbsURLInXML("http://xml") + tr := NewChain(absURLInXML...) + apply(t.Errorf, tr, xml_abs_url_tests) +} |