Tools: HTML document `<base href="...">` does not influence import scanning.
Discovered while working on polymer-bundler to implement <base> tag handling, I discovered that calling document.getByKind('import') on my analyzed document, the set returned did not include all the expected transitive dependencies. The point of disconnect was an imported document that included a <base> tag which modified the intent of its href and src attributes etc.
Propose extending the analyzer to make document "base url" a thing different but derived from the document url and properly identify such when <base> present in an HTML document.
usergenic
All 6 comments
+1
Do you have an example document easily available?
rictic
on 28 Jan 2017
My test is these four documents.
- Root (no base tag) https://github.com/Polymer/polymer-bundler/blob/master/test/html/base.html
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<link rel="import" href="imports/base.html">
<title></title>
</head>
<body>
</body>
</html>
- Import, has base tag https://github.com/Polymer/polymer-bundler/blob/master/test/html/imports/base.html
<base href="base-foo">
<link rel="import" href="sub-base.html">
- Nested Import, https://github.com/Polymer/polymer-bundler/blob/master/test/html/imports/base-foo/sub-base.html
<base href="../../imports">
<link rel="import" href="super-base.html">
<span href="hello">From Sub-Base!</span>
- Nested Nested Import, https://github.com/Polymer/polymer-bundler/blob/master/test/html/imports/super-base.html
<a href="sub-base/sub-base.html">Hello also from super base!</a>
Good catch.
Seems like in addition to those tests we need a simple root with a base tag.
justinfagnani
on 28 Jan 2017
Seems like, since multiple scanners will want this info, this would be a good thing to put on the ParsedHtmlDocument when we produce it
rictic
on 28 Jan 2017
Quickly, I think there are two changes to do. First, ParsedHtmlDocument, or possibly ParsedDocument, should get a baseUrl property and a resolveUrl() method. Second HtmlImportScanner and friends need to use that.
@rictic jinx
justinfagnani
on 28 Jan 2017
I think we fixed this a bunch. Closing.
usergenic
on 22 Mar 2017
Related issues
web-padawan
路
4Comments
lpellegr
路
4Comments
dilansankalpa
路
3Comments
manolo
路
4Comments
marcushellberg
路
3Comments
Most helpful comment
Quickly, I think there are two changes to do. First,
ParsedHtmlDocument, or possiblyParsedDocument, should get abaseUrlproperty and aresolveUrl()method. SecondHtmlImportScannerand friends need to use that.@rictic jinx