Strip bindings before calling Uri.parse

pkg/polymer/lib/src/build/import_inliner.dart

 // Copyright (c) 2013, the Dart project authors.  Please see the AUTHORS file
 // for details. All rights reserved. Use of this source code is governed by a
 // BSD-style license that can be found in the LICENSE file.
 
 /// Transfomer that inlines polymer-element definitions from html imports.
 library polymer.src.build.import_inliner;
 
 import 'dart:async';
 import 'dart:convert';
 
@@ skipping 244 matching lines @@
 ///
 ///     myPkgName|lib/foo/bar.html -> myPkgName.foo.bar_html
 ///     myPkgName|web/foo/bar.html -> myPkgName.web.foo.bar_html
 ///
 /// This should roughly match the recommended library name conventions.
 String _libraryNameFor(AssetId id, int suffix) {
   var name = '${path.withoutExtension(id.path)}_'
       '${path.extension(id.path).substring(1)}';
   if (name.startsWith('lib/')) name = name.substring(4);
   name = name.split('/').map((part) {
-    part = part.replaceAll(INVALID_LIB_CHARS_REGEX, '_');
-    if (part.startsWith(NUM_REGEX)) part = '_${part}';
+    part = part.replaceAll(_INVALID_LIB_CHARS_REGEX, '_');
+    if (part.startsWith(_NUM_REGEX)) part = '_${part}';
     return part;
   }).join(".");
   return '${id.package}.${name}_$suffix';
 }
 
 /// Parse [code] and determine whether it has a library directive.
 bool _hasLibraryDirective(String code) =>
     parseDirectives(code, suppressErrors: true)
         .directives.any((d) => d is LibraryDirective);
 
@@ skipping 53 matching lines @@
     return changed;
   }
 
   visitElement(Element node) {
     // TODO(jakemac): Support custom elements that extend html elements which
     // have url-like attributes. This probably means keeping a list of which
     // html elements support each url-like attribute.
     if (!isCustomTagName(node.localName)) {
       node.attributes.forEach((name, value) {
         if (_urlAttributes.contains(name)) {
-          if (value != '' && !value.trim().startsWith(_BINDINGS)) {
+          if (value != '' && !value.trim().startsWith(_BINDINGS_REGEX)) {
             node.attributes[name] = _newUrl(value, node.sourceSpan);
             changed = changed || value != node.attributes[name];
           }
         }
       });
     }
     if (node.localName == 'style') {
       node.text = visitCss(node.text);
       changed = true;
     } else if (node.localName == 'script' &&
         node.attributes['type'] == TYPE_DART &&
         !node.attributes.containsKey('src')) {
       // TODO(jmesserly): we might need to visit JS too to handle ES Harmony
       // modules.
       node.text = visitInlineDart(node.text);
       changed = true;
     }
     return super.visitElement(node);
   }
 
   static final _URL = new RegExp(r'url\(([^)]*)\)', multiLine: true);
   static final _QUOTE = new RegExp('["\']', multiLine: true);
-  static final _BINDINGS = new RegExp(r'({{)|(\[\[)');
 
   /// Visit the CSS text and replace any relative URLs so we can inline it.
   // Ported from:
   // https://github.com/Polymer/vulcanize/blob/c14f63696797cda18dc3d372b78aa3378acc691f/lib/vulcan.js#L149
   // TODO(jmesserly): use csslib here instead? Parsing with RegEx is sadness.
   // Maybe it's reliable enough for finding URLs in CSS? I'm not sure.
   String visitCss(String cssText) {
     var url = spanUrlFor(sourceId, transform);
     var src = new SourceFile.text(url, cssText);
     return cssText.replaceAllMapped(_URL, (match) {
@@ skipping 36 matching lines @@
     }
 
     if (!output.hasEdits) return code;
 
     // TODO(sigmund): emit source maps when barback supports it (see
     // dartbug.com/12340)
     return (output.commit()..build(file.url)).text;
   }
 
   String _newUrl(String href, Span span) {
-    // Uri.parse blows up on invalid characters (like {{). Encoding the uri
-    // allows it to be parsed, which does the correct thing in the general case.
-    // This uri not used to build the new uri, so it never needs to be decoded.
-    var uri = Uri.parse(Uri.encodeFull(href));
+    // Take all bindings out of the path and replace with [_BINDING_PLACEHOLDER]
+    // so [Uri.parse] will not fail.
+    var bindingMatches = _BINDING_REGEX.allMatches(href).map((m) => m.group(0));
+    href = href.replaceAll(_BINDING_REGEX, _BINDING_PLACEHOLDER);

[Siggi Cherem (dart-lang), 2014/07/31 21:54:22]

I wonder if we need to replace all, or if we should just find the first one. At
that point, we might be OK just splitting it and don't need a placeholder. For
example:

a/b/{{e}}/f/{{g}}/q.dart

->

prefix = a/b/
suffix = {{e}}/f/{{g}}/q.dart

Then later newPath = '${id.path}$suffix'

WDYT?

[jakemac, 2014/08/01 15:20:47]

This seems like a good idea, but I tried it out and there ends up being a lot of
weird edge cases. For instance, a path of just './' turns into '../web'. Also,
the trailing slashes always get removed, so if I just tack on other things I
have to add it back (but only if it had one before). Covering all of these cases
is going to overall make things more complicated unfortunately and I would
likely miss a few.

On 2014/07/31 21:54:22, Siggi Cherem (dart-lang) wrote:

[Siggi Cherem (dart-lang), 2014/08/01 16:02:51]

Interesting, here are a couple other ideas... it might be worth giving them a
shot.

You can do like you are doing with adding a %%BINDING%% but just for the first
one:


a/b/{{e}}/f/{{g}}/q.dart

->
prefix = a/b/%%BINDING%%
suffix = {{e}}/f/{{g}}/q.dart

what do you think?


If that doesn't work, I think we can still get rid of the loop that restores the
bindings (which would get rid of the buildingMatches) by just replacing the
suffix in the resulting path starting from the first binding position.

var uriSafeHref = href.replaceAll(_BINDING_REGEX, _BINDING_PLACEHOLDER);
...
var newPath = id.path;
var bindingIndex = newPath.indexOf(_BINDING_PLACEHOLDER);
if (bindingIndex >= 0) {
  var prefix = newPath.substring(0, bindingIndex);
  var suffix = href.substring(href.indexOf(_BINDING_REGEX));
  var newPath = '$prefix$suffix';
}

[jakemac, 2014/08/01 22:12:35]

Done.

+
+    var uri = Uri.parse(href);
     if (uri.isAbsolute) return href;

[Siggi Cherem (dart-lang), 2014/07/31 21:54:22]

shouldn't this return the original href without the placeholders?
(same below on most of the return statements)

[jakemac, 2014/08/01 15:20:47]

Good catch, I have updated it so I don't write over the original href varibale
any more, instead I am making a new one called uriSafeHref.

On 2014/07/31 21:54:22, Siggi Cherem (dart-lang) wrote:

     if (!uri.scheme.isEmpty) return href;
     if (!uri.host.isEmpty) return href;
     if (uri.path.isEmpty) return href;  // Implies standalone ? or # in URI.
     if (path.isAbsolute(href)) return href;
 
     var id = uriToAssetId(sourceId, href, transform.logger, span);
     if (id == null) return href;
     var primaryId = transform.primaryInput.id;
 
-    if (id.path.startsWith('lib/')) {
-      return '${topLevelPath}packages/${id.package}/${id.path.substring(4)}';
+    // Put the original bindings back into the path now that we have parsed it.
+    var newPath = id.path;
+    for (var bindingMatch in bindingMatches) {
+      newPath = newPath.replaceFirst(_BINDING_PLACEHOLDER, bindingMatch);
     }
 
-    if (id.path.startsWith('asset/')) {
-      return '${topLevelPath}assets/${id.package}/${id.path.substring(6)}';
+    if (newPath.startsWith('lib/')) {
+      return '${topLevelPath}packages/${id.package}/${newPath.substring(4)}';
+    }
+
+    if (newPath.startsWith('asset/')) {
+      return '${topLevelPath}assets/${id.package}/${newPath.substring(6)}';
     }
 
     if (primaryId.package != id.package) {
       // Techincally we shouldn't get there
       transform.logger.error("don't know how to include $id from $primaryId",
           span: span);
       return href;
     }
 
     var builder = path.url;
-    return builder.relative(builder.join('/', id.path),
+    return builder.relative(builder.join('/', newPath),
         from: builder.join('/', builder.dirname(primaryId.path)));
   }
 }
 
 /// HTML attributes that expect a URL value.
 /// <http://dev.w3.org/html5/spec/section-index.html#attributes-1>
 ///
 /// Every one of these attributes is a URL in every context where it is used in
 /// the DOM. The comments show every DOM element where an attribute can be used.
 const _urlAttributes = const [
   'action',     // in form
   'background', // in body
   'cite',       // in blockquote, del, ins, q
   'data',       // in object
   'formaction', // in button, input
   'href',       // in a, area, link, base, command
   'icon',       // in command
   'manifest',   // in html
   'poster',     // in video
   'src',        // in audio, embed, iframe, img, input, script, source, track,
                 //    video
 ];
 
 /// When inlining <link rel="stylesheet"> tags copy over all attributes to the
 /// style tag except these ones.
 const IGNORED_LINKED_STYLE_ATTRS =
     const ['charset', 'href', 'href-lang', 'rel', 'rev'];
 
-/// Global RegExp objects for validating generated library names.
-final INVALID_LIB_CHARS_REGEX = new RegExp('[^a-z0-9_]');
-final NUM_REGEX = new RegExp('[0-9]');
+/// Global RegExp objects.
+final _INVALID_LIB_CHARS_REGEX = new RegExp('[^a-z0-9_]');
+final _NUM_REGEX = new RegExp('[0-9]');
+final _BINDING_REGEX = new RegExp(r'(({{.*}})|(\[\[.*\]\]))');
+
+/// Placeholder for bindings in urls for [Uri.parse].
+final _BINDING_PLACEHOLDER = '%BINDING%';
 
 _getSpan(SourceFile file, AstNode node) => file.span(node.offset, node.end);