Merge pull request scala#4307 from som-snytt/issue/4339-2.10-b

SI-4339 Backpatch event errors and attr fix

Author: gkossakowski

bincompat-forward.whitelist.conf

@@ -187,6 +187,14 @@ filter {
         {
             matchName="scala.xml.dtd.ElementValidator.scala$xml$dtd$ElementValidator$$find$2"
             problemName=IncompatibleMethTypeProblem
+        },
+        {
+            matchName="scala.xml.pull.ExceptionEvent"
+            problemName=MissingClassProblem
+        },
+        {
+            matchName="scala.xml.pull.ExceptionEvent$"
+            problemName=MissingClassProblem
         }
     ]
 }

src/compiler/scala/tools/nsc/ast/parser/Scanners.scala

@@ -367,7 +367,9 @@ trait Scanners extends ScannersCommon {
             val last = if (charOffset >= 2) buf(charOffset - 2) else ' '
             nextChar()
             last match {
-              case ' ' | '\t' | '\n' | '{' | '(' | '>' if isNameStart(ch) || ch == '!' || ch == '?' =>
+              // exclude valid xml names that happen to be Scala operator chars
+              case ' ' | '\t' | '\n' | '{' | '(' | '>' if (isNameStart(ch) && ch != ':' && !isSpecial(ch))
+                  || ch == '!' || ch == '?' =>
                 token = XMLSTART
               case _ =>
                 // Console.println("found '<', but last is '"+in.last+"'"); // DEBUG

src/library/scala/xml/parsing/MarkupParserCommon.scala

@@ -58,8 +58,9 @@ private[scala] trait MarkupParserCommon extends TokenTests {
    @param endCh either `'` or `"`
    */
   def xAttributeValue(endCh: Char): String = {
+    require(endCh == '\'' || endCh == '"', s"Expected single or double quote, found $endCh")
     val buf = new StringBuilder
-    while (ch != endCh) {
+    while (ch != endCh && !eof) {
       // well-formedness constraint
       if (ch == '<') return errorAndResult("'<' not allowed in attrib value", "")
       else if (ch == SU) truncatedError("")

src/library/scala/xml/parsing/TokenTests.scala

@@ -30,43 +30,25 @@ trait TokenTests {
   def isAlpha(c: Char) = (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z')
   def isAlphaDigit(c: Char) = isAlpha(c) || (c >= '0' && c <= '9')
 
-  /** {{{
-   *  NameChar ::= Letter | Digit | '.' | '-' | '_' | ':'
-   *             | CombiningChar | Extender
-   *  }}}
-   *  See [4] and Appendix B of XML 1.0 specification.
-  */
-  def isNameChar(ch: Char) = {
-    import java.lang.Character._
-    // The constants represent groups Mc, Me, Mn, Lm, and Nd.
-
-    isNameStart(ch) || (getType(ch).toByte match {
-      case COMBINING_SPACING_MARK |
-              ENCLOSING_MARK | NON_SPACING_MARK |
-              MODIFIER_LETTER | DECIMAL_DIGIT_NUMBER => true
-      case _                                         => ".-:" contains ch
-    })
-  }
-
-  /** {{{
-   *  NameStart ::= ( Letter | '_' )
-   *  }}}
-   *  where Letter means in one of the Unicode general
-   *  categories `{ Ll, Lu, Lo, Lt, Nl }`.
-   *
-   *  We do not allow a name to start with `:`.
-   *  See [3] and Appendix B of XML 1.0 specification
-   */
-  def isNameStart(ch: Char) = {
-    import java.lang.Character._
-
-    getType(ch).toByte match {
-      case LOWERCASE_LETTER |
-              UPPERCASE_LETTER | OTHER_LETTER |
-              TITLECASE_LETTER | LETTER_NUMBER => true
-      case _                                   => ch == '_'
-    }
-  }
+  def isNameChar(c: Char): Boolean = (
+    isNameStart(c) ||
+    (c >= '0' && c <= '9') ||
+    c == '-' ||
+    c == '.' ||
+    c == 0xB7 ||
+    (c >= 0x300 && c <= 0x36F) ||
+    (c >= 0x203F && c <= 0x2040)
+  )
+  def isNameStart(c: Char): Boolean = (
+         if (c < 0x00C0)  isAlpha(c) || c == ':' || c == '_'
+    else if (c < 0x0300)  c != 0xD7 && c != 0xF7
+    else if (c < 0x2000)  c >= 0x370 && c != 0x37E
+    else if (c < 0x3001)  c == 0x200C || c == 0x200D || (0x2070 to 0x218F contains c) ||
+                          (0x2C00 to 0x2FEF contains c)
+    else if (c < 0xD800)  true
+    else if (c < 0x10000) (0xF900 to 0xFDCF contains c) || (0xFDF0 to 0xFFFD contains c)
+    else                  false // codepoint < 0xF0000
+  )
 
   /** {{{
    *  Name ::= ( Letter | '_' ) (NameChar)*

src/library/scala/xml/pull/XMLEventReader.scala

@@ -91,12 +91,16 @@ extends scala.collection.AbstractIterator[XMLEvent]
 
     override def run() {
       curInput = input
-      interruptibly { this.initialize.document() }
-      setEvent(POISON)
+      try interruptibly { this.initialize.document() }
+      catch { case e:Exception => setEvent(ExceptionEvent(e)) }
+      finally setEvent(POISON)
     }
   }
 }
 
+// An internal class used to propagate exception from helper threads to API end users.
+private case class ExceptionEvent(exception:Exception) extends XMLEvent
+
 // An iterator designed for one or more producers to generate
 // elements, and a single consumer to iterate.  Iteration will continue
 // until closeIterator() is called, after which point producers
@@ -141,6 +145,7 @@ trait ProducerConsumerIterator[T >: Null] extends Iterator[T] {
   def next() = {
     if (eos) throw new NoSuchElementException("ProducerConsumerIterator")
     if (buffer == null) fillBuffer
+    if (buffer.isInstanceOf[ExceptionEvent]) throw buffer.asInstanceOf[ExceptionEvent].exception
 
     drainBuffer
   }

test/files/jvm/unittest_xml.scala

@@ -62,7 +62,9 @@ object Test {
   object UtilityTest {
     def run() {
       assert(Utility.isNameStart('b'))
-      assert(!Utility.isNameStart(':'))
+
+      // no longer: this was a convenience for the implementation, not to spec.
+      //assert(!Utility.isNameStart(':'))
 
       val x = <foo>
                  <toomuchws/>

test/files/run/t4339.check

@@ -0,0 +1,3 @@
+Saw failure: scala.xml.parsing.FatalError: expected closing tag of foo
+EvElemStart(null,foo, bar="baz/>",)
+Saw failure: scala.xml.parsing.FatalError: expected closing tag of foo

test/files/run/t4339.scala

@@ -0,0 +1,35 @@
+
+import scala.util.{ Try, Success, Failure }
+import xml._
+import scala.io.Source.fromString
+import java.io.PrintStream
+
+object Test extends App {
+
+  def quietSource(text: String) = new io.Source {
+    override protected val iter = io.Source fromString text
+    override def report(pos: Int, msg: String, out: PrintStream) = ()
+  }
+  def reading(text: String)(f: pull.XMLEventReader => Unit): Unit = {
+    val r = new pull.XMLEventReader(quietSource(text))
+    try f(r)
+    finally r.stop()
+  }
+  def trying(body: => Unit): Unit =
+    Try (body) match {
+      case Success(_) => Console println "Expected failure"
+      case Failure(e) => Console println s"Saw failure: $e"
+    }
+
+  val problematic = """<foo bar="baz/>"""
+
+  trying (
+    parsing.ConstructingParser.fromSource(quietSource(problematic), false).document.docElem
+  )
+
+  trying (
+    reading(problematic) { r =>
+      while (r.hasNext) println(r.next())
+    }
+  )
+}