Capture the DTD body while it is being parsed.

Change-Id: Ibef02ca759eb56a00f0f72f4063d129ef5350d21
http://b/3241492

3 files changed, 85 insertions, 23 deletions

diff --git a/luni/src/test/java/libcore/xml/PullParserDtdTest.java b/luni/src/test/java/libcore/xml/PullParserDtdTest.java
index 0101d7b..834bf00 100644
--- a/luni/src/test/java/libcore/xml/PullParserDtdTest.java
+++ b/luni/src/test/java/libcore/xml/PullParserDtdTest.java
@@ -168,6 +168,31 @@ public abstract class PullParserDtdTest extends TestCase {
         assertEquals(XmlPullParser.END_DOCUMENT, parser.next());
     }
 
+    public void testStructuredEntityAndNextToken() throws Exception {
+        String xml = "<!DOCTYPE foo [<!ENTITY bb \"<bar>baz<!--quux--></bar>\">]><foo>a&bb;c</foo>";
+        XmlPullParser parser = newPullParser(xml);
+        assertEquals(XmlPullParser.DOCDECL, parser.nextToken());
+        assertEquals(XmlPullParser.START_TAG, parser.nextToken());
+        assertEquals("foo", parser.getName());
+        assertEquals(XmlPullParser.TEXT, parser.nextToken());
+        assertEquals("a", parser.getText());
+        assertEquals(XmlPullParser.ENTITY_REF, parser.nextToken());
+        assertEquals("bb", parser.getName());
+        assertEquals("", parser.getText());
+        assertEquals(XmlPullParser.START_TAG, parser.nextToken());
+        assertEquals("bar", parser.getName());
+        assertEquals(XmlPullParser.TEXT, parser.nextToken());
+        assertEquals("baz", parser.getText());
+        assertEquals(XmlPullParser.COMMENT, parser.nextToken());
+        assertEquals("quux", parser.getText());
+        assertEquals(XmlPullParser.END_TAG, parser.nextToken());
+        assertEquals("bar", parser.getName());
+        assertEquals(XmlPullParser.TEXT, parser.nextToken());
+        assertEquals("c", parser.getText());
+        assertEquals(XmlPullParser.END_TAG, parser.next());
+        assertEquals(XmlPullParser.END_DOCUMENT, parser.next());
+    }
+
     /**
      * Android's Expat replaces external entities with the empty string.
      */
@@ -462,10 +487,6 @@ public abstract class PullParserDtdTest extends TestCase {
         }
     }
 
-    /**
-     * In honeycomb, KxmlParser's DTD handling was improved but no longer
-     * supports returning the full DTD text. http://b/3241492
-     */
     public void testDoctypeWithNextToken() throws Exception {
         String xml = "<!DOCTYPE foo [<!ENTITY bb \"bar baz\">]><foo>a&bb;c</foo>";
         XmlPullParser parser = newPullParser(xml);
@@ -473,13 +494,21 @@ public abstract class PullParserDtdTest extends TestCase {
         assertEquals(" foo [<!ENTITY bb \"bar baz\">]", parser.getText());
         assertNull(parser.getName());
         assertEquals(XmlPullParser.START_TAG, parser.nextToken());
-        assertEquals(XmlPullParser.TEXT, parser.nextToken());
-        assertEquals("a", parser.getText());
-        assertEquals(XmlPullParser.ENTITY_REF, parser.nextToken());
-        assertEquals("bb", parser.getName());
-        assertEquals("bar baz", parser.getText());
-        assertEquals(XmlPullParser.TEXT, parser.nextToken());
-        assertEquals("c", parser.getText());
+        assertEquals(XmlPullParser.TEXT, parser.next());
+        assertEquals("abar bazc", parser.getText());
+        assertEquals(XmlPullParser.END_TAG, parser.next());
+        assertEquals(XmlPullParser.END_DOCUMENT, parser.next());
+    }
+
+    public void testDoctypeSpansBuffers() throws Exception {
+        char[] doctypeChars = new char[READ_BUFFER_SIZE + 1];
+        Arrays.fill(doctypeChars, 'x');
+        String doctypeBody = " foo [<!--" + new String(doctypeChars) + "-->]";
+        String xml = "<!DOCTYPE" + doctypeBody + "><foo/>";
+        XmlPullParser parser = newPullParser(xml);
+        assertEquals(XmlPullParser.DOCDECL, parser.nextToken());
+        assertEquals(doctypeBody, parser.getText());
+        assertEquals(XmlPullParser.START_TAG, parser.next());
         assertEquals(XmlPullParser.END_TAG, parser.next());
         assertEquals(XmlPullParser.END_DOCUMENT, parser.next());
     }
diff --git a/luni/src/test/java/libcore/xml/PullParserTest.java b/luni/src/test/java/libcore/xml/PullParserTest.java
index 2862de0..aaa76a6 100644
--- a/luni/src/test/java/libcore/xml/PullParserTest.java
+++ b/luni/src/test/java/libcore/xml/PullParserTest.java
@@ -615,6 +615,15 @@ public abstract class PullParserTest extends TestCase {
         assertEquals(0, parser.getDepth());
     }
 
+    public void testNextAfterEndDocument() throws Exception {
+        XmlPullParser parser = newPullParser();
+        parser.setInput(new StringReader("<foo></foo>"));
+        assertEquals(XmlPullParser.START_TAG, parser.next());
+        assertEquals(XmlPullParser.END_TAG, parser.next());
+        assertEquals(XmlPullParser.END_DOCUMENT, parser.next());
+        assertEquals(XmlPullParser.END_DOCUMENT, parser.next());
+    }
+
     public void testNamespaces() throws Exception {
         String xml = "<one xmlns='ns:default' xmlns:n1='ns:1' a='b'>\n"
                 + "  <n1:two c='d' n1:e='f' xmlns:n2='ns:2'>text</n1:two>\n"
diff --git a/xml/src/main/java/org/kxml2/io/KXmlParser.java b/xml/src/main/java/org/kxml2/io/KXmlParser.java
index 9fb858b..f9d6461 100644
--- a/xml/src/main/java/org/kxml2/io/KXmlParser.java
+++ b/xml/src/main/java/org/kxml2/io/KXmlParser.java
@@ -110,6 +110,13 @@ public class KXmlParser implements XmlPullParser, Closeable {
     private boolean keepNamespaceAttributes;
 
     /**
+     * If non-null, the contents of the read buffer must be copied into this
+     * string builder before the read buffer is overwritten. This is used to
+     * capture the raw DTD text while parsing the DTD.
+     */
+    private StringBuilder bufferCapture;
+
+    /**
      * Entities defined in or for this document. This map is created lazily.
      */
     private Map<String, char[]> documentEntities;
@@ -408,10 +415,7 @@ public class KXmlParser implements XmlPullParser, Closeable {
                 }
                 break;
             case DOCDECL:
-                readDoctype();
-                if (justOneToken) {
-                    text = ""; // TODO: support capturing the doctype text
-                }
+                readDoctype(justOneToken);
                 break;
 
             default:
@@ -563,16 +567,32 @@ public class KXmlParser implements XmlPullParser, Closeable {
      * Read the document's DTD. Although this parser is non-validating, the DTD
      * must be parsed to capture entity values and default attribute values.
      */
-    private void readDoctype() throws IOException, XmlPullParserException {
+    private void readDoctype(boolean saveDtdText) throws IOException, XmlPullParserException {
         read(START_DOCTYPE);
-        skip();
-        rootElementName = readName();
-        readExternalId(true, true);
-        skip();
-        if (peekCharacter() == '[') {
-            readInternalSubset();
+
+        int startPosition = -1;
+        if (saveDtdText) {
+            bufferCapture = new StringBuilder();
+            startPosition = position;
         }
-        skip();
+        try {
+            skip();
+            rootElementName = readName();
+            readExternalId(true, true);
+            skip();
+            if (peekCharacter() == '[') {
+                readInternalSubset();
+            }
+            skip();
+        } finally {
+            if (saveDtdText) {
+                bufferCapture.append(buffer, 0, position);
+                bufferCapture.delete(0, startPosition);
+                text = bufferCapture.toString();
+                bufferCapture = null;
+            }
+        }
+
         read('>');
     }
 
@@ -1456,6 +1476,10 @@ public class KXmlParser implements XmlPullParser, Closeable {
             }
         }
 
+        if (bufferCapture != null) {
+            bufferCapture.append(buffer, 0, position);
+        }
+
         if (limit != position) {
             limit -= position;
             System.arraycopy(buffer, position, buffer, 0, limit);