moparisthebest/poi
1
0
Fork 0
Code Issues Pull Requests Releases 1 Wiki Activity

use last file extension if multiple extensions; close opened stream; extract String constants

Browse Source 
git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1776827 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Javen O'Neal 2017-01-01 06:57:42 +00:00
parent a4a0c1e91a
commit 11c8858f6e
1 changed files with 34 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
src/ooxml/java/org/apache/poi/ss/extractor/EmbeddedExtractor.java
View File

@ -17,6 +17,8 @@
package org.apache.poi.ss.extractor; package org.apache.poi.ss.extractor;
import static org.apache.poi.util.StringUtil.endsWithIgnoreCase;
import java.io.ByteArrayOutputStream; import java.io.ByteArrayOutputStream;
import java.io.IOException; import java.io.IOException;
import java.io.InputStream; import java.io.InputStream;
@ -50,6 +52,18 @@ import org.apache.poi.xssf.usermodel.XSSFObjectData;
public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> { public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
private static final POILogger LOG = POILogFactory.getLogger(EmbeddedExtractor.class); private static final POILogger LOG = POILogFactory.getLogger(EmbeddedExtractor.class);
// contentType
private static final String CONTENT_TYPE_BYTES = "binary/octet-stream";
private static final String CONTENT_TYPE_PDF = "application/pdf";
private static final String CONTENT_TYPE_DOC = "application/msword";
private static final String CONTENT_TYPE_XLS = "application/vnd.ms-excel";
// default file extension
private static final String PDF_EXT = ".pdf";
private static final String DOC_EXT = ".doc";
private static final String XLS_EXT = ".xls";
private static final String OLE_EXT = ".ole";
/** /**
* @return the list of known extractors, if you provide custom extractors, override this method * @return the list of known extractors, if you provide custom extractors, override this method
*/ */
@ -98,11 +112,10 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
if (od.hasDirectoryEntry()) { if (od.hasDirectoryEntry()) {
data = extractOne((DirectoryNode)od.getDirectory()); data = extractOne((DirectoryNode)od.getDirectory());
} else { } else {
String contentType = "binary/octet-stream";
if (od instanceof XSSFObjectData) { if (od instanceof XSSFObjectData) {
contentType = ((XSSFObjectData)od).getObjectPart().getContentType(); String contentType = ((XSSFObjectData)od).getObjectPart().getContentType();
} }
data = new EmbeddedData(od.getFileName(), od.getObjectData(), contentType); data = new EmbeddedData(od.getFileName(), od.getObjectData(), CONTENT_TYPE_BYTES);
} }
} catch (Exception e) { } catch (Exception e) {
LOG.log(POILogger.WARN, "Entry not found / readable - ignoring OLE embedding", e); LOG.log(POILogger.WARN, "Entry not found / readable - ignoring OLE embedding", e);
@ -119,7 +132,7 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
data.setShape(shape); data.setShape(shape);
String filename = data.getFilename(); String filename = data.getFilename();
String extension = (filename == null || filename.indexOf('.') == -1) ? ".bin" : filename.substring(filename.indexOf('.')); String extension = (filename == null || filename.lastIndexOf('.') == -1) ? ".bin" : filename.substring(filename.lastIndexOf('.'));
// try to find an alternative name // try to find an alternative name
if (filename == null || "".equals(filename) || filename.startsWith("MBD") || filename.startsWith("Root Entry")) { if (filename == null || "".equals(filename) || filename.startsWith("MBD") || filename.startsWith("Root Entry")) {
@ -157,7 +170,7 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
dest.writeFilesystem(bos); dest.writeFilesystem(bos);
dest.close(); dest.close();
return new EmbeddedData(dn.getName(), bos.toByteArray(), "binary/octet-stream"); return new EmbeddedData(dn.getName(), bos.toByteArray(), CONTENT_TYPE_BYTES);
} }
protected EmbeddedData extract(Picture source) throws IOException { protected EmbeddedData extract(Picture source) throws IOException {
@ -176,7 +189,7 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
try { try {
// TODO: inspect the CompObj record for more details, i.e. the content type // TODO: inspect the CompObj record for more details, i.e. the content type
Ole10Native ole10 = Ole10Native.createFromEmbeddedOleObject(dn); Ole10Native ole10 = Ole10Native.createFromEmbeddedOleObject(dn);
return new EmbeddedData(ole10.getFileName(), ole10.getDataBuffer(), "binary/octet-stream"); return new EmbeddedData(ole10.getFileName(), ole10.getDataBuffer(), CONTENT_TYPE_BYTES);
} catch (Ole10NativeException e) { } catch (Ole10NativeException e) {
throw new IOException(e); throw new IOException(e);
} }
@ -198,7 +211,7 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
InputStream is = dn.createDocumentInputStream("CONTENTS"); InputStream is = dn.createDocumentInputStream("CONTENTS");
IOUtils.copy(is, bos); IOUtils.copy(is, bos);
is.close(); is.close();
return new EmbeddedData(dn.getName()+".pdf", bos.toByteArray(), "application/pdf"); return new EmbeddedData(dn.getName() + PDF_EXT, bos.toByteArray(), CONTENT_TYPE_PDF);
} }
@Override @Override
@ -238,10 +251,10 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
byte[] pdfBytes = new byte[pictureBytesLen]; byte[] pdfBytes = new byte[pictureBytesLen];
System.arraycopy(pictureBytes, idxStart, pdfBytes, 0, pictureBytesLen); System.arraycopy(pictureBytes, idxStart, pdfBytes, 0, pictureBytesLen);
String filename = source.getShapeName().trim(); String filename = source.getShapeName().trim();
if (!filename.toLowerCase(Locale.ROOT).endsWith(".pdf")) { if (!endsWithIgnoreCase(filename, PDF_EXT)) {
filename += ".pdf"; filename += PDF_EXT;
} }
return new EmbeddedData(filename, pdfBytes, "application/pdf"); return new EmbeddedData(filename, pdfBytes, CONTENT_TYPE_PDF);
} }
@ -259,8 +272,8 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
@Override @Override
public EmbeddedData extract(DirectoryNode dn) throws IOException { public EmbeddedData extract(DirectoryNode dn) throws IOException {
EmbeddedData ed = super.extract(dn); EmbeddedData ed = super.extract(dn);
ed.setFilename(dn.getName()+".doc"); ed.setFilename(dn.getName() + DOC_EXT);
ed.setContentType("application/msword"); ed.setContentType(CONTENT_TYPE_DOC);
return ed; return ed;
} }
} }
@ -277,8 +290,8 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
@Override @Override
public EmbeddedData extract(DirectoryNode dn) throws IOException { public EmbeddedData extract(DirectoryNode dn) throws IOException {
EmbeddedData ed = super.extract(dn); EmbeddedData ed = super.extract(dn);
ed.setFilename(dn.getName()+".xls"); ed.setFilename(dn.getName() + XLS_EXT);
ed.setContentType("application/vnd.ms-excel"); ed.setContentType(CONTENT_TYPE_XLS);
return ed; return ed;
} }
} }
@ -291,7 +304,7 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
@Override @Override
public EmbeddedData extract(DirectoryNode dn) throws IOException { public EmbeddedData extract(DirectoryNode dn) throws IOException {
EmbeddedData ed = super.extract(dn); EmbeddedData ed = super.extract(dn);
ed.setFilename(dn.getName()+".ole"); ed.setFilename(dn.getName() + OLE_EXT);
// TODO: read the content type from CombObj stream // TODO: read the content type from CombObj stream
return ed; return ed;
} }
@ -306,11 +319,14 @@ public class EmbeddedExtractor implements Iterable<EmbeddedExtractor> {
copyNodes(srcDir, destDir); copyNodes(srcDir, destDir);
} else { } else {
InputStream is = src.createDocumentInputStream(e); InputStream is = src.createDocumentInputStream(e);
try {
dest.createDocument(e.getName(), is); dest.createDocument(e.getName(), is);
} finally {
is.close(); is.close();
} }
} }
} }
}