WMF fixes

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1722465 13f79535-47bb-0310-9956-ffa450edef68

7 files changed, 129 insertions, 130 deletions

diff --git a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmap16.java b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmap16.java
index c0293b4336..2ff0d118a8 100644
--- a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmap16.java
+++ b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmap16.java
@@ -76,33 +76,8 @@ public class HwmfBitmap16 {
         byte buf[] = new byte[bytes];

         leis.read(buf);

         

-//        FileOutputStream fos = new FileOutputStream("bla16.bmp");

-//        fos.write(buf);

-//        fos.close();

-        

-        

-//        BufferedImage bi = new BufferedImage(width, height, BufferedImage.TYPE_INT_RGB);

-//        

-//        int size2 = 0;

-//        byte buf[] = new byte[widthBytes];

-//        for (int h=0; h<height; h++) {

-//            leis.read(buf);

-//            size2 += widthBytes;

-//

-//            ImageInputStream iis = new MemoryCacheImageInputStream(new ByteArrayInputStream(buf));

-//

-//            for (int w=0; w<width; w++) {

-//                long bitsAtPixel = iis.readBits(bitsPixel);

-//                // TODO: is bitsPixel a multiple of 3 (r,g,b)

-//                // which colortable should be used for the various bit sizes???

-//                

-//            }

-//        }

-//        

-//        assert (bytes == size2);

-//

-//        size += size2;

-        

+        // TODO: this is not implemented ... please provide a sample, if it

+        // ever happens to you, to come here ...

         

         return size;

     }

diff --git a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmapDib.java b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmapDib.java
index 12f9b30a9f..0f720239f3 100644
--- a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmapDib.java
+++ b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfBitmapDib.java
@@ -22,6 +22,7 @@ import java.awt.image.BufferedImage;
 import java.io.BufferedInputStream;

 import java.io.ByteArrayInputStream;

 import java.io.IOException;

+import java.io.InputStream;

 import java.util.ArrayList;

 import java.util.List;

 

@@ -207,7 +208,7 @@ public class HwmfBitmapDib {
     @SuppressWarnings("unused")

     private Color colorTable[];

     @SuppressWarnings("unused")

-    private int colorMaskRed=0,colorMaskGreen=0,colorMaskBlue=0;

+    private int colorMaskR=0,colorMaskG=0,colorMaskB=0;

 

     // size of header and color table, for start of image data calculation

     private int introSize;

@@ -222,7 +223,7 @@ public class HwmfBitmapDib {
         introSize += readColors(leis);

         assert(introSize < 10000);

 

-        int fileSize = (headerImageSize != 0) ? (int)(introSize+headerImageSize) : recordSize;

+        int fileSize = (headerImageSize < headerSize) ? recordSize : (int)Math.min(introSize+headerImageSize,recordSize);

         

         imageData = new byte[fileSize];

         leis.reset();

@@ -316,41 +317,43 @@ public class HwmfBitmapDib {
             return 0;

         case BI_BITCOUNT_1:

             // 2 colors

-            return readRGBQuad(leis, 2);

+            return readRGBQuad(leis, (int)(headerColorUsed == 0 ? 2 : Math.min(headerColorUsed,2)));

         case BI_BITCOUNT_2:

             // 16 colors

-            return readRGBQuad(leis, 16);

+            return readRGBQuad(leis, (int)(headerColorUsed == 0 ? 16 : Math.min(headerColorUsed,16)));

         case BI_BITCOUNT_3:

             // 256 colors

-            return readRGBQuad(leis, (int)headerColorUsed);

-        case BI_BITCOUNT_5:

-            colorMaskRed=0xFF;

-            colorMaskGreen=0xFF;

-            colorMaskBlue=0xFF;

-            return 0;

+            return readRGBQuad(leis, (int)(headerColorUsed == 0 ? 256 : Math.min(headerColorUsed,256)));

         case BI_BITCOUNT_4:

-            if (headerCompression == Compression.BI_RGB) {

-                colorMaskBlue = 0x1F;

-                colorMaskGreen = 0x1F<<5;

-                colorMaskRed = 0x1F<<10;

+            switch (headerCompression) {

+            case BI_RGB:

+                colorMaskB = 0x1F;

+                colorMaskG = 0x1F<<5;

+                colorMaskR = 0x1F<<10;

                 return 0;

-            } else {

-                assert(headerCompression == Compression.BI_BITFIELDS);

-                colorMaskBlue = leis.readInt();

-                colorMaskGreen = leis.readInt();

-                colorMaskRed = leis.readInt();

+            case BI_BITFIELDS:

+                colorMaskB = leis.readInt();

+                colorMaskG = leis.readInt();

+                colorMaskR = leis.readInt();

                 return 3*LittleEndianConsts.INT_SIZE;

+            default:

+                throw new IOException("Invalid compression option ("+headerCompression+") for bitcount ("+headerBitCount+").");

             }

+        case BI_BITCOUNT_5:

         case BI_BITCOUNT_6:

-            if (headerCompression == Compression.BI_RGB) {

-                colorMaskBlue = colorMaskGreen = colorMaskRed = 0xFF;

+            switch (headerCompression) {

+            case BI_RGB:

+                colorMaskR=0xFF;

+                colorMaskG=0xFF;

+                colorMaskB=0xFF;

                 return 0;

-            } else {

-                assert(headerCompression == Compression.BI_BITFIELDS);

-                colorMaskBlue = leis.readInt();

-                colorMaskGreen = leis.readInt();

-                colorMaskRed = leis.readInt();

+            case BI_BITFIELDS:

+                colorMaskB = leis.readInt();

+                colorMaskG = leis.readInt();

+                colorMaskR = leis.readInt();

                 return 3*LittleEndianConsts.INT_SIZE;

+            default:

+                throw new IOException("Invalid compression option ("+headerCompression+") for bitcount ("+headerBitCount+").");

             }

         }

     }

@@ -372,30 +375,36 @@ public class HwmfBitmapDib {
         return size;

     }

 

-    public BufferedImage getImage() {

+    public InputStream getBMPStream() {

         if (imageData == null) {

             throw new RecordFormatException("bitmap not initialized ... need to call init() before");

         }

 

+        // sometimes there are missing bytes after the imageData which will be 0-filled

+        int imageSize = (int)Math.max(imageData.length, introSize+headerImageSize);

+        

         // create the image data and leave the parsing to the ImageIO api

-        byte buf[] = new byte[BMP_HEADER_SIZE+imageData.length];

+        byte buf[] = new byte[BMP_HEADER_SIZE+imageSize];

 

         // https://en.wikipedia.org/wiki/BMP_file_format #  Bitmap file header

         buf[0] = (byte)'B';

         buf[1] = (byte)'M';

         // the full size of the bmp

-        LittleEndian.putInt(buf, 2, (int)(BMP_HEADER_SIZE + introSize + headerImageSize));

+        LittleEndian.putInt(buf, 2, BMP_HEADER_SIZE+imageSize);

         // the next 4 bytes are unused

         LittleEndian.putInt(buf, 6, 0);

         // start of image = BMP header length + dib header length + color tables length

         LittleEndian.putInt(buf, 10, BMP_HEADER_SIZE + introSize);

-        

+        // fill the "known" image data

         System.arraycopy(imageData, 0, buf, BMP_HEADER_SIZE, imageData.length);

         

+        return new ByteArrayInputStream(buf);

+    }

+    

+    public BufferedImage getImage() {

         try {

-            return ImageIO.read(new ByteArrayInputStream(buf));

+            return ImageIO.read(getBMPStream());

         } catch (IOException e) {

-            // ... shouldn't happen

             throw new RecordFormatException("invalid bitmap data", e);

         }

     }

diff --git a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfFill.java b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfFill.java
index ee38c7b94e..4b01deb07d 100644
--- a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfFill.java
+++ b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfFill.java
@@ -614,7 +614,7 @@ public class HwmfFill {
         

         @Override

         public int init(LittleEndianInputStream leis, long recordSize, int recordFunction) throws IOException {

-            boolean hasBitmap = (recordSize > ((recordFunction >> 8) + 3));

+            boolean hasBitmap = (recordSize/2 != ((recordFunction >> 8) + 3));

 

             int size = 0;

             int rasterOpCode = leis.readUShort();

@@ -802,7 +802,7 @@ public class HwmfFill {
         

         @Override

         public int init(LittleEndianInputStream leis, long recordSize, int recordFunction) throws IOException {

-            boolean hasBitmap = (recordSize > ((recordFunction >> 8) + 3));

+            boolean hasBitmap = (recordSize/2 != ((recordFunction >> 8) + 3));

 

             int size = 0;

             int rasterOpCode = leis.readUShort();

@@ -840,7 +840,7 @@ public class HwmfFill {
 

         @Override

         public BufferedImage getImage() {

-            return target.getImage();

+            return (target == null) ? null : target.getImage();

         }

     }

 

diff --git a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfText.java b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfText.java
index 6d332607f3..5e6dde83a6 100644
--- a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfText.java
+++ b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfText.java
@@ -259,6 +259,9 @@ public class HwmfText {
         

         @Override

         public int init(LittleEndianInputStream leis, long recordSize, int recordFunction) throws IOException {

+            // -6 bytes of record function and length header

+            final int remainingRecordSize = (int)(recordSize-6);

+            

             y = leis.readShort();

             x = leis.readShort();

             stringLength = leis.readShort();

@@ -266,7 +269,7 @@ public class HwmfText {
             

             int size = 4*LittleEndianConsts.SHORT_SIZE;

             

-            if (fwOpts != 0) {

+            if (fwOpts != 0 && size+8<=remainingRecordSize) {

                 // the bounding rectangle is optional and only read when fwOpts are given

                 left = leis.readShort();

                 top = leis.readShort();

@@ -280,8 +283,6 @@ public class HwmfText {
             text = new String(buf, 0, stringLength, LocaleUtil.CHARSET_1252);

             size += buf.length;

             

-            // -6 bytes of record function and length header

-            int remainingRecordSize = (int)(recordSize-6);

             if (size < remainingRecordSize) {

                 if (size + stringLength*LittleEndianConsts.SHORT_SIZE < remainingRecordSize) {

                     throw new RecordFormatException("can't read Dx array - given recordSize doesn't contain enough values for string length "+stringLength);

diff --git a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfWindowing.java b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfWindowing.java
index 1cb19e005f..a6d7204583 100644
--- a/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfWindowing.java
+++ b/src/scratchpad/src/org/apache/poi/hwmf/record/HwmfWindowing.java
@@ -542,14 +542,17 @@ public class HwmfWindowing {
             count = leis.readUShort();

             top = leis.readUShort();

             bottom = leis.readUShort();

-            left_scanline = new int[count];

-            right_scanline = new int[count];

-            for (int i=0; i*2<count; i++) {

+            int size = 3*LittleEndianConsts.SHORT_SIZE;

+            left_scanline = new int[count/2];

+            right_scanline = new int[count/2];

+            for (int i=0; i<count/2; i++) {

                 left_scanline[i] = leis.readUShort();

                 right_scanline[i] = leis.readUShort();

+                size += 2*LittleEndianConsts.SHORT_SIZE;

             }

             count2 = leis.readUShort();

-            return 8 + count*4;

+            size += LittleEndianConsts.SHORT_SIZE;

+            return size;

         }

     }

 

@@ -618,26 +621,23 @@ public class HwmfWindowing {
         public int init(LittleEndianInputStream leis, long recordSize, int recordFunction) throws IOException {

             nextInChain = leis.readShort();

             objectType = leis.readShort();

-            objectCount = leis.readUShort();

+            objectCount = leis.readInt();

             regionSize = leis.readShort();

             scanCount = leis.readShort();

             maxScan = leis.readShort();

-            bottom = leis.readShort();

-            right = leis.readShort();

-            top = leis.readShort();

             left = leis.readShort();

+            top = leis.readShort();

+            right = leis.readShort();

+            bottom = leis.readShort();

+            

+            int size = 9*LittleEndianConsts.SHORT_SIZE+LittleEndianConsts.INT_SIZE;

+

+            scanObjects = new WmfScanObject[scanCount];

+            for (int i=0; i<scanCount; i++) {

+                size += (scanObjects[i] = new WmfScanObject()).init(leis);

+            }

 

-            List<WmfScanObject> soList = new ArrayList<WmfScanObject>();

-            int scanCountI = 0, size = 0;

-            do {

-                WmfScanObject so = new WmfScanObject();

-                size += so.init(leis);

-                scanCountI += so.count;

-                soList.add(so);

-            } while  (scanCountI < scanCount);

-            scanObjects = soList.toArray(new WmfScanObject[soList.size()]);

-

-            return 20 + size;

+            return size;

         }

 

         @Override

diff --git a/src/scratchpad/src/org/apache/poi/hwmf/usermodel/HwmfPicture.java b/src/scratchpad/src/org/apache/poi/hwmf/usermodel/HwmfPicture.java
index 3ebd23bf30..3bea50eaf6 100644
--- a/src/scratchpad/src/org/apache/poi/hwmf/usermodel/HwmfPicture.java
+++ b/src/scratchpad/src/org/apache/poi/hwmf/usermodel/HwmfPicture.java
@@ -70,12 +70,8 @@ public class HwmfPicture {
             int remainingSize = (int)(recordSize - consumedSize);

             assert(remainingSize >= 0);

             if (remainingSize > 0) {

-//                byte remaining[] = new byte[remainingSize];

-//                leis.read(remaining);

-//                FileOutputStream fos = new FileOutputStream("remaining.dat");

-//                fos.write(remaining);

-//                fos.close();

-                 leis.skip(remainingSize);

+            	// skip size in loops, because not always all bytes are skipped in one call 

+                for (int i=remainingSize; i>0; i-=leis.skip(i));

             }

         }

     }

diff --git a/src/scratchpad/testcases/org/apache/poi/hwmf/TestHwmfParsing.java b/src/scratchpad/testcases/org/apache/poi/hwmf/TestHwmfParsing.java
index a1f2cbf1d0..ad585eccbd 100644
--- a/src/scratchpad/testcases/org/apache/poi/hwmf/TestHwmfParsing.java
+++ b/src/scratchpad/testcases/org/apache/poi/hwmf/TestHwmfParsing.java
@@ -20,14 +20,17 @@ package org.apache.poi.hwmf;
 import static org.junit.Assert.assertEquals;

 

 import java.awt.image.BufferedImage;

-import java.io.ByteArrayInputStream;

 import java.io.File;

 import java.io.FileFilter;

 import java.io.FileInputStream;

 import java.io.FileOutputStream;

+import java.io.FilterInputStream;

 import java.io.IOException;

+import java.net.URL;

 import java.util.List;

 import java.util.Locale;

+import java.util.zip.ZipEntry;

+import java.util.zip.ZipInputStream;

 

 import javax.imageio.ImageIO;

 

@@ -55,55 +58,70 @@ public class TestHwmfParsing {
 

     @Test

     @Ignore

-    public void extract() throws IOException {

-        File dir = new File("test-data/slideshow");

-        File files[] = dir.listFiles(new FileFilter() {

-            public boolean accept(File pathname) {

-                return pathname.getName().matches("(?i).*\\.pptx?$");

-            }

-        });

-

-        boolean outputFiles = false;

-

+    public void fetchWmfFromGovdocs() throws IOException {

+        URL url = new URL("http://digitalcorpora.org/corpora/files/govdocs1/by_type/ppt.zip");

         File outdir = new File("build/ppt");

-        if (outputFiles) {

-            outdir.mkdirs();

-        }

-        int wmfIdx = 1;

-        for (File f : files) {

+        outdir.mkdirs();

+        ZipInputStream zis = new ZipInputStream(url.openStream());

+        ZipEntry ze;

+        while ((ze = zis.getNextEntry()) != null) {

+            String basename = ze.getName().replaceAll(".*?([^/]+)\\.wmf", "$1");

+            FilterInputStream fis = new FilterInputStream(zis){

+                public void close() throws IOException {}

+            };

             try {

-                SlideShow<?,?> ss = SlideShowFactory.create(f);

+                SlideShow<?,?> ss = SlideShowFactory.create(fis);

+                int wmfIdx = 1;

                 for (PictureData pd : ss.getPictureData()) {

                     if (pd.getType() != PictureType.WMF) continue;

                     byte wmfData[] = pd.getData();

-                    if (outputFiles) {

-                        String filename = String.format(Locale.ROOT, "pic%04d.wmf", wmfIdx);

-                        FileOutputStream fos = new FileOutputStream(new File(outdir, filename));

-                        fos.write(wmfData);

-                        fos.close();

-                    }

-

-                    HwmfPicture wmf = new HwmfPicture(new ByteArrayInputStream(wmfData));

-

-                    int bmpIndex = 1;

-                    for (HwmfRecord r : wmf.getRecords()) {

-                        if (r instanceof HwmfImageRecord) {

-                            BufferedImage bi = ((HwmfImageRecord)r).getImage();

-                            if (outputFiles) {

-                                String filename = String.format(Locale.ROOT, "pic%04d-%04d.png", wmfIdx, bmpIndex);

-                                ImageIO.write(bi, "PNG", new File(outdir, filename));

-                            }

-                            bmpIndex++;

-                        }

-                    }

-

+                    String filename = String.format(Locale.ROOT, "%s-%04d.wmf", basename, wmfIdx);

+                    FileOutputStream fos = new FileOutputStream(new File(outdir, filename));

+                    fos.write(wmfData);

+                    fos.close();

                     wmfIdx++;

                 }

                 ss.close();

             } catch (Exception e) {

-                System.out.println(f+" ignored.");

+                System.out.println(ze.getName()+" ignored.");

+            }

+        }

+    }

+    

+    @Test

+    @Ignore

+    public void parseWmfs() throws IOException {

+        boolean outputFiles = false;

+        File indir = new File("build/ppt"), outdir = indir;

+        final String startFile = "";

+        File files[] = indir.listFiles(new FileFilter() {

+            boolean foundStartFile = false;

+            public boolean accept(File pathname) {

+                foundStartFile |= startFile.isEmpty() || pathname.getName().contains(startFile);

+                return foundStartFile && pathname.getName().matches("(?i).*\\.wmf?$");

+            }

+        });

+        for (File f : files) {

+            try {

+                String basename = f.getName().replaceAll(".*?([^/]+)\\.wmf", "$1");

+                FileInputStream fis = new FileInputStream(f);

+                HwmfPicture wmf = new HwmfPicture(fis);

+                fis.close();

+                

+                int bmpIndex = 1;

+                for (HwmfRecord r : wmf.getRecords()) {

+                    if (r instanceof HwmfImageRecord) {

+                        BufferedImage bi = ((HwmfImageRecord)r).getImage();

+                        if (bi != null && outputFiles) {

+                            String filename = String.format(Locale.ROOT, "%s-%04d.png", basename, bmpIndex);

+                            ImageIO.write(bi, "PNG", new File(outdir, filename));

+                        }

+                        bmpIndex++;

+                    }

+                }

+            } catch (Exception e) {

+                System.out.println(f.getName()+" ignored.");                

             }

         }

     }

-

 }