Re-enable and add new HTML support

pkg/analyzer/lib/src/task/html.dart

Index: pkg/analyzer/lib/src/task/html.dart
diff --git a/pkg/analyzer/lib/src/task/html.dart b/pkg/analyzer/lib/src/task/html.dart
index ab0b9b53b1f800d16e56ec5e2e1d82358f5fa85c..da124c410aebf2bf45a8923c5f541611438f3957 100644
--- a/pkg/analyzer/lib/src/task/html.dart
+++ b/pkg/analyzer/lib/src/task/html.dart
@@ -4,14 +4,80 @@
 
 library analyzer.src.task.html;
 
+import 'dart:collection';
+
 import 'package:analyzer/src/generated/engine.dart' hide AnalysisTask;
+import 'package:analyzer/src/generated/error.dart';
 import 'package:analyzer/src/generated/source.dart';
 import 'package:analyzer/src/task/general.dart';
 import 'package:analyzer/task/general.dart';
 import 'package:analyzer/task/html.dart';
 import 'package:analyzer/task/model.dart';
-//import 'package:html/dom.dart';
-//import 'package:html/parser.dart';
+import 'package:html/dom.dart';
+import 'package:html/parser.dart';
+import 'package:source_span/source_span.dart';
+
+/**
+ * The errors found while parsing an HTML file.
+ */
+final ListResultDescriptor<AnalysisError> DOCUMENT_ERRORS =

[Paul Berry, 2015/06/15 21:34:39]

I'm concerned about the use of the term "document" here too, for similar reasons
to what I stated before.  How about "HTML_DOCUMENT_ERRORS"?

[Brian Wilkerson, 2015/06/16 16:28:08]

Done

+    new ListResultDescriptor<AnalysisError>(
+        'DOCUMENT_ERRORS', AnalysisError.NO_ERRORS);
+
+/**
+ * A task that merges all of the errors for a single source into a single list
+ * of errors.
+ */
+class HtmlErrorsTask extends SourceBasedAnalysisTask {
+  /**
+   * The name of the [DOCUMENT_ERRORS] input.
+   */
+  static const String DOCUMENT_ERRORS_INPUT = 'DOCUMENT_ERRORS';
+
+  /**
+   * The task descriptor describing this kind of task.
+   */
+  static final TaskDescriptor DESCRIPTOR = new TaskDescriptor('HtmlErrorsTask',
+      createTask, buildInputs, <ResultDescriptor>[HTML_ERRORS]);
+
+  HtmlErrorsTask(InternalAnalysisContext context, AnalysisTarget target)
+      : super(context, target);
+
+  @override
+  TaskDescriptor get descriptor => DESCRIPTOR;
+
+  @override
+  void internalPerform() {
+    //
+    // Prepare inputs.
+    //
+    List<AnalysisError> errors = getRequiredInput(DOCUMENT_ERRORS_INPUT);
+    //
+    // Record outputs.
+    //
+    outputs[HTML_ERRORS] = errors;
+  }
+
+  /**
+   * Return a map from the names of the inputs of this kind of task to the task
+   * input descriptors describing those inputs for a task with the
+   * given [target].
+   */
+  static Map<String, TaskInput> buildInputs(Source target) {
+    return <String, TaskInput>{
+      DOCUMENT_ERRORS_INPUT: DOCUMENT_ERRORS.of(target)
+    };
+  }
+
+  /**
+   * Create an [HtmlErrorsTask] based on the given [target] in the given
+   * [context].
+   */
+  static HtmlErrorsTask createTask(
+      AnalysisContext context, AnalysisTarget target) {
+    return new HtmlErrorsTask(context, target);
+  }
+}
 
 /**
  * A task that scans the content of a file, producing a set of Dart tokens.
@@ -26,8 +92,7 @@ class ParseHtmlTask extends SourceBasedAnalysisTask {
    * The task descriptor describing this kind of task.
    */
   static final TaskDescriptor DESCRIPTOR = new TaskDescriptor('ParseHtmlTask',
-      createTask, buildInputs,
-      <ResultDescriptor>[/*DOCUMENT, DOCUMENT_ERRORS*/]);
+      createTask, buildInputs, <ResultDescriptor>[DOCUMENT, DOCUMENT_ERRORS]);
 
   /**
    * Initialize a newly created task to access the content of the source
@@ -41,14 +106,21 @@ class ParseHtmlTask extends SourceBasedAnalysisTask {
 
   @override
   void internalPerform() {
-//    String content = getRequiredInput(CONTENT_INPUT_NAME);
-//
-//    HtmlParser parser = new HtmlParser(content);
-//    Document document = parser.parse();
-//    List<ParseError> errors = parser.errors;
-//
-//    outputs[DOCUMENT] = document;
-//    outputs[DOCUMENT_ERRORS] = errors;
+    String content = getRequiredInput(CONTENT_INPUT_NAME);
+
+    HtmlParser parser = new HtmlParser(content);
+    parser.compatMode = 'quirks';
+    Document document = parser.parse();
+    List<ParseError> parseErrors = parser.errors;
+    List<AnalysisError> errors = <AnalysisError>[];
+    for (ParseError parseError in parseErrors) {
+      SourceSpan span = parseError.span;
+      errors.add(new AnalysisError(target.source, span.start.offset,
+          span.length, HtmlErrorCode.PARSE_ERROR, [parseError.message]));
+    }
+
+    outputs[DOCUMENT] = document;
+    outputs[DOCUMENT_ERRORS] = errors;
   }
 
   /**
@@ -68,3 +140,72 @@ class ParseHtmlTask extends SourceBasedAnalysisTask {
     return new ParseHtmlTask(context, target);
   }
 }
+
+/**
+ * A task that computes the Dart libraries that are referenced by an HTML file.
+ */
+class ReferencedLibrariesTask extends SourceBasedAnalysisTask {
+  /**
+   * The name of the [DOCUMENT] input.
+   */
+  static const String DOCUMENT_INPUT = 'DOCUMENT';
+
+  /**
+   * The task descriptor describing this kind of task.
+   */
+  static final TaskDescriptor DESCRIPTOR = new TaskDescriptor(
+      'ReferencedLibrariesTask', createTask, buildInputs,
+      <ResultDescriptor>[REFERENCED_LIBRARIES]);
+
+  ReferencedLibrariesTask(
+      InternalAnalysisContext context, AnalysisTarget target)
+      : super(context, target);
+
+  @override
+  TaskDescriptor get descriptor => DESCRIPTOR;
+
+  @override
+  void internalPerform() {
+    //
+    // Prepare inputs.
+    //
+    List<Source> libraries = <Source>[];
+    Document document = getRequiredInput(DOCUMENT_INPUT);
+    List<Element> scripts = document.getElementsByTagName('script');
+    for (Element script in scripts) {
+      LinkedHashMap<dynamic, String> attributes = script.attributes;
+      if (attributes['type'] == 'application/dart') {
+        String src = attributes['src'];
+        if (AnalysisEngine.isDartFileName(src)) {
+          Source source = context.sourceFactory.resolveUri(target.source, src);

[Paul Berry, 2015/06/15 21:34:39]

This presumes that all the rules for resolving URI's in HTML files are the same
as for resolving URI's in .dart files (e.g. "package:" and "dart:" URI's are
permitted, and relative URI's are resolved relative to the referencing file.) 
Is that in fact the case, or are URI's in HTML files more restricted?

[Brian Wilkerson, 2015/06/16 16:28:08]

I don't know whether package: or dart: URI's are allowed, but relative URI's
should be resolved relative to the referencing file. If we're handling more than
we should we can always come back and add an error message (while still
understanding the user's intent).

+          if (source != null) {
+            libraries.add(source);
+          }
+        }
+      }
+    }
+    //
+    // Record outputs.
+    //
+    outputs[REFERENCED_LIBRARIES] =
+        libraries.isEmpty ? Source.EMPTY_LIST : libraries;

[Paul Berry, 2015/06/15 21:34:39]

Again, it seems of dubious benefit to me to canonicalize the empty list.

[Brian Wilkerson, 2015/06/16 16:28:08]

I know that we're storing these lists, so the value depends on how many HTML
files we end up analyzing.

+  }
+
+  /**
+   * Return a map from the names of the inputs of this kind of task to the task
+   * input descriptors describing those inputs for a task with the
+   * given [target].
+   */
+  static Map<String, TaskInput> buildInputs(Source target) {
+    return <String, TaskInput>{DOCUMENT_INPUT: DOCUMENT.of(target)};
+  }
+
+  /**
+   * Create a [ReferencedLibrariesTask] based on the given [target] in the given
+   * [context].
+   */
+  static ReferencedLibrariesTask createTask(
+      AnalysisContext context, AnalysisTarget target) {
+    return new ReferencedLibrariesTask(context, target);
+  }
+}