Fix bug #50539 - Better fix for html-style br tags (invalid XML) inside XSSF documents

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1067217 13f79535-47bb-0310-9956-ffa450edef68

src/documentation/content/xdocs/status.xml

@@ -34,6 +34,7 @@
 
     <changes>
         <release version="3.8-beta1" date="2010-??-??">
+           <action dev="poi-developers" type="fix">50539 - Better fix for html-style br tags (invalid XML) inside XSSF documents</action>
            <action dev="poi-developers" type="add">49928 - allow overridden built-in formats in HSSFCellStyle</action>
            <action dev="POI-DEVELOPERS" type="add">50607 - Added implementation for CLEAN(), CHAR() and ADDRESS()</action>
            <action dev="poi-developers" type="add">50587 - Improved documentation on user-defined functions</action>

src/ooxml/java/org/apache/poi/xssf/util/EvilUnclosedBRFixingInputStream.java

@@ -54,16 +54,26 @@ public class EvilUnclosedBRFixingInputStream extends InputStream {
 
    @Override
    public int read(byte[] b, int off, int len) throws IOException {
-      if(spare != null) {
-         // This is risky, but spare is normally only a byte or two...
-         System.arraycopy(spare, 0, b, off, spare.length);
-         int ret = spare.length;
-         spare = null;
-         return ret;
+      // Grab any data left from last time
+      int readA = readFromSpare(b, off, len);
+
+      // Now read from the stream 
+      int readB = source.read(b, off+readA, len-readA);
+      
+      // Figure out how much we've done
+      int read;
+      if(readB == -1 || readB == 0) {
+         read = readA; 
+      } else {
+         read = readA + readB;
       }
       
-      int read = source.read(b, off, len);
+      // Fix up our data
+      if(read > 0) {
          read = fixUp(b, off, read);
+      }
+      
+      // All done
       return read;
    }
 
@@ -72,10 +82,71 @@ public class EvilUnclosedBRFixingInputStream extends InputStream {
       return this.read(b, 0, b.length);
    }
    
+   /**
+    * Reads into the buffer from the spare bytes
+    */
+   private int readFromSpare(byte[] b, int offset, int len) {
+      if(spare == null) return 0;
+      if(len == 0) throw new IllegalArgumentException("Asked to read 0 bytes");
+      
+      if(spare.length <= len) {
+         // All fits, good
+         System.arraycopy(spare, 0, b, offset, spare.length);
+         int read = spare.length;
+         spare = null;
+         return read;
+      } else {
+         // We have more spare than they can copy with...
+         byte[] newspare = new byte[spare.length-len];
+         System.arraycopy(spare, 0, b, offset, len);
+         System.arraycopy(spare, len, newspare, 0, newspare.length);
+         spare = newspare;
+         return len;
+      }
+   }
+   private void addToSpare(byte[] b, int offset, int len, boolean atTheEnd) {
+      if(spare == null) {
+         spare = new byte[len];
+         System.arraycopy(b, offset, spare, 0, len);
+      } else {
+         byte[] newspare = new byte[spare.length+len];
+         if(atTheEnd) {
+            System.arraycopy(spare, 0, newspare, 0, spare.length);
+            System.arraycopy(b, offset, newspare, spare.length, len);
+         } else {
+            System.arraycopy(b, offset, newspare, 0, len);
+            System.arraycopy(spare, 0, newspare, len, spare.length);
+         }
+         spare = newspare;
+      }
+   }
+
    private int fixUp(byte[] b, int offset, int read) {
+      // Do we have any potential overhanging ones?
+      for(int i=0; i<detect.length-1; i++) {
+         int base = offset+read-1-i;
+         if(base < 0) continue;
+            
+         boolean going = true;
+         for(int j=0; j<=i && going; j++) {
+            if(b[base+j] == detect[j]) {
+               // Matches
+            } else {
+               going = false;
+            }
+         }
+         if(going) {
+            // There could be a <br> handing over the end, eg <br|
+            addToSpare(b, base, i+1, true);
+            read -= 1;
+            read -= i;
+            break;
+         }
+      }
+      
       // Find places to fix
       ArrayList<Integer> fixAt = new ArrayList<Integer>();
-      for(int i=offset; i<offset+read-4; i++) {
+      for(int i=offset; i<=offset+read-detect.length; i++) {
          boolean going = true;
          for(int j=0; j<detect.length && going; j++) {
             if(b[i+j] != detect[j]) {
@@ -91,11 +162,22 @@ public class EvilUnclosedBRFixingInputStream extends InputStream {
          return read;
       }
       
-      // Save a bit, if needed to fit
-      int overshoot = offset+read+fixAt.size() - b.length;  
+      // If there isn't space in the buffer to contain
+      //  all the fixes, then save the overshoot for next time
+      int needed = offset+read+fixAt.size();
+      int overshoot = needed - b.length;  
       if(overshoot > 0) {
-         spare = new byte[overshoot];
-         System.arraycopy(b, b.length-overshoot, spare, 0, overshoot);
+         // Make sure we don't loose part of a <br>!
+         int fixes = 0;
+         for(int at : fixAt) {
+            if(at > offset+read-detect.length-overshoot-fixes) {
+               overshoot = needed - at - 1 - fixes;
+               break;
+            }
+            fixes++;
+         }
+
+         addToSpare(b, offset+read-overshoot, overshoot, false);
          read -= overshoot;
       }
       
@@ -103,6 +185,14 @@ public class EvilUnclosedBRFixingInputStream extends InputStream {
       //  positions are valid
       for(int j=fixAt.size()-1; j>=0; j--) {
          int i = fixAt.get(j);
+         if(i >= read+offset) {
+            // This one has moved into the overshoot
+            continue;
+         }
+         if(i > read-3) {
+            // This one has moved into the overshoot
+            continue;
+         }
 
          byte[] tmp = new byte[read-i-3];
          System.arraycopy(b, i+3, tmp, 0, tmp.length);

src/ooxml/testcases/org/apache/poi/xssf/util/TestEvilUnclosedBRFixingInputStream.java

@@ -70,6 +70,37 @@ public final class TestEvilUnclosedBRFixingInputStream extends TestCase {
       assertEquals(fixed, result);
    }
    
+   /**
+    * Checks that we can copy with br tags around the buffer boundaries
+    */
+   public void testBufferSize() throws Exception {
+      byte[] orig = "<p><div>Hello<br> <br>There!</div> <div>Tags!<br><br></div></p>".getBytes("UTF-8");
+      byte[] fixed = "<p><div>Hello<br/> <br/>There!</div> <div>Tags!<br/><br/></div></p>".getBytes("UTF-8");
+      
+      // Vary the buffer size, so that we can end up with the br in the
+      //  overflow or only part in the buffer
+      for(int i=5; i<orig.length; i++) {
+         EvilUnclosedBRFixingInputStream inp = new EvilUnclosedBRFixingInputStream(
+               new ByteArrayInputStream(orig)
+         );
+         
+         ByteArrayOutputStream bout = new ByteArrayOutputStream();
+         boolean going = true;
+         while(going) {
+            byte[] b = new byte[i];
+            int r = inp.read(b);
+            if(r > 0) {
+               bout.write(b, 0, r);
+            } else {
+               going = false;
+            }
+         }
+         
+         byte[] result = bout.toByteArray();
+         assertEquals(fixed, result);
+      }
+   }
+
    protected void assertEquals(byte[] a, byte[] b) {
       assertEquals(a.length, b.length);
       for(int i=0; i<a.length; i++) {