Rename the outlook extractor to be more consistent with other extractors
git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@897249 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
parent
cefe4e1d28
commit
98cea49eb5
@ -34,7 +34,7 @@
|
||||
|
||||
<changes>
|
||||
<release version="3.7-SNAPSHOT" date="2010-??-??">
|
||||
<action dev="POI-DEVELOPERS" type="fix">Add a text extractor to HSMF for simpler extraction of text from .msg files</action>
|
||||
<action dev="POI-DEVELOPERS" type="fix">Add a text extractor (OutlookTextExtractor) to HSMF for simpler extraction of text from .msg files</action>
|
||||
<action dev="POI-DEVELOPERS" type="fix">Some improvements to HSMF parsing of .msg files</action>
|
||||
<action dev="POI-DEVELOPERS" type="fix">Initialise the link type of HSSFHyperLink, so that getType() on it works</action>
|
||||
<action dev="POI-DEVELOPERS" type="fix">48425 - improved performance of DateUtil.isCellDateFormatted() </action>
|
||||
|
@ -31,7 +31,7 @@ import org.apache.poi.POIXMLDocument;
|
||||
import org.apache.poi.POIXMLTextExtractor;
|
||||
import org.apache.poi.hdgf.extractor.VisioTextExtractor;
|
||||
import org.apache.poi.hslf.extractor.PowerPointExtractor;
|
||||
import org.apache.poi.hsmf.extractor.HSMFTextExtactor;
|
||||
import org.apache.poi.hsmf.extractor.OutlookTextExtactor;
|
||||
import org.apache.poi.hssf.extractor.ExcelExtractor;
|
||||
import org.apache.poi.hwpf.extractor.WordExtractor;
|
||||
import org.apache.poi.openxml4j.exceptions.InvalidFormatException;
|
||||
@ -142,7 +142,7 @@ public class ExtractorFactory {
|
||||
if(entry.getName().equals("__substg1.0_1000001E") ||
|
||||
entry.getName().equals("__substg1.0_0047001E") ||
|
||||
entry.getName().equals("__substg1.0_0037001E")) {
|
||||
return new HSMFTextExtactor(poifsDir, fs);
|
||||
return new OutlookTextExtactor(poifsDir, fs);
|
||||
}
|
||||
}
|
||||
throw new IllegalArgumentException("No supported documents found in the OLE2 stream");
|
||||
|
@ -25,7 +25,7 @@ import org.apache.poi.POITextExtractor;
|
||||
import org.apache.poi.POIDataSamples;
|
||||
import org.apache.poi.hdgf.extractor.VisioTextExtractor;
|
||||
import org.apache.poi.hslf.extractor.PowerPointExtractor;
|
||||
import org.apache.poi.hsmf.extractor.HSMFTextExtactor;
|
||||
import org.apache.poi.hsmf.extractor.OutlookTextExtactor;
|
||||
import org.apache.poi.hssf.extractor.ExcelExtractor;
|
||||
import org.apache.poi.hwpf.extractor.WordExtractor;
|
||||
import org.apache.poi.poifs.filesystem.POIFSFileSystem;
|
||||
@ -169,7 +169,7 @@ public class TestExtractorFactory extends TestCase {
|
||||
// Outlook msg
|
||||
assertTrue(
|
||||
ExtractorFactory.createExtractor(msg)
|
||||
instanceof HSMFTextExtactor
|
||||
instanceof OutlookTextExtactor
|
||||
);
|
||||
assertTrue(
|
||||
ExtractorFactory.createExtractor(msg).getText().length() > 50
|
||||
@ -248,7 +248,7 @@ public class TestExtractorFactory extends TestCase {
|
||||
// Outlook msg
|
||||
assertTrue(
|
||||
ExtractorFactory.createExtractor(new FileInputStream(msg))
|
||||
instanceof HSMFTextExtactor
|
||||
instanceof OutlookTextExtactor
|
||||
);
|
||||
assertTrue(
|
||||
ExtractorFactory.createExtractor(new FileInputStream(msg)).getText().length() > 50
|
||||
@ -303,7 +303,7 @@ public class TestExtractorFactory extends TestCase {
|
||||
// Outlook msg
|
||||
assertTrue(
|
||||
ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(msg)))
|
||||
instanceof HSMFTextExtactor
|
||||
instanceof OutlookTextExtactor
|
||||
);
|
||||
assertTrue(
|
||||
ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(msg))).getText().length() > 50
|
||||
|
@ -26,17 +26,21 @@ import org.apache.poi.hsmf.exceptions.ChunkNotFoundException;
|
||||
import org.apache.poi.poifs.filesystem.DirectoryNode;
|
||||
import org.apache.poi.poifs.filesystem.POIFSFileSystem;
|
||||
|
||||
public class HSMFTextExtactor extends POIOLE2TextExtractor {
|
||||
public HSMFTextExtactor(MAPIMessage msg) {
|
||||
/**
|
||||
* A text extractor for HSMF (Outlook) .msg files.
|
||||
* Outputs in a format somewhat like a plain text email.
|
||||
*/
|
||||
public class OutlookTextExtactor extends POIOLE2TextExtractor {
|
||||
public OutlookTextExtactor(MAPIMessage msg) {
|
||||
super(msg);
|
||||
}
|
||||
public HSMFTextExtactor(DirectoryNode poifsDir, POIFSFileSystem fs) throws IOException {
|
||||
public OutlookTextExtactor(DirectoryNode poifsDir, POIFSFileSystem fs) throws IOException {
|
||||
this(new MAPIMessage(poifsDir, fs));
|
||||
}
|
||||
public HSMFTextExtactor(POIFSFileSystem fs) throws IOException {
|
||||
public OutlookTextExtactor(POIFSFileSystem fs) throws IOException {
|
||||
this(new MAPIMessage(fs));
|
||||
}
|
||||
public HSMFTextExtactor(InputStream inp) throws IOException {
|
||||
public OutlookTextExtactor(InputStream inp) throws IOException {
|
||||
this(new MAPIMessage(inp));
|
||||
}
|
||||
|
||||
@ -74,5 +78,4 @@ public class HSMFTextExtactor extends POIOLE2TextExtractor {
|
||||
|
||||
return s.toString();
|
||||
}
|
||||
|
||||
}
|
@ -29,10 +29,10 @@ import org.apache.poi.poifs.filesystem.POIFSFileSystem;
|
||||
/**
|
||||
* Tests to verify that the text extractor works
|
||||
*/
|
||||
public final class TestHSMFTextExtractor extends TestCase {
|
||||
public final class TestOutlookTextExtractor extends TestCase {
|
||||
private POIDataSamples samples;
|
||||
|
||||
public TestHSMFTextExtractor() throws IOException {
|
||||
public TestOutlookTextExtractor() throws IOException {
|
||||
samples = POIDataSamples.getHSMFInstance();
|
||||
}
|
||||
|
||||
@ -49,7 +49,7 @@ public final class TestHSMFTextExtractor extends TestCase {
|
||||
);
|
||||
MAPIMessage msg = new MAPIMessage(simple);
|
||||
|
||||
HSMFTextExtactor ext = new HSMFTextExtactor(msg);
|
||||
OutlookTextExtactor ext = new OutlookTextExtactor(msg);
|
||||
String text = ext.getText();
|
||||
|
||||
assertContains(text, "From: Kevin Roast\n");
|
||||
@ -66,7 +66,7 @@ public final class TestHSMFTextExtractor extends TestCase {
|
||||
new FileInputStream(samples.getFile("simple_test_msg.msg"))
|
||||
));
|
||||
|
||||
HSMFTextExtactor ext = new HSMFTextExtactor(msg);
|
||||
OutlookTextExtactor ext = new OutlookTextExtactor(msg);
|
||||
String text = ext.getText();
|
||||
|
||||
assertContains(text, "From: Travis Ferguson\n");
|
||||
@ -79,13 +79,13 @@ public final class TestHSMFTextExtractor extends TestCase {
|
||||
}
|
||||
|
||||
public void testConstructors() throws Exception {
|
||||
String inp = (new HSMFTextExtactor(new FileInputStream(
|
||||
String inp = (new OutlookTextExtactor(new FileInputStream(
|
||||
samples.getFile("simple_test_msg.msg")
|
||||
)).getText());
|
||||
String poifs = (new HSMFTextExtactor(new POIFSFileSystem(new FileInputStream(
|
||||
String poifs = (new OutlookTextExtactor(new POIFSFileSystem(new FileInputStream(
|
||||
samples.getFile("simple_test_msg.msg")
|
||||
))).getText());
|
||||
String mapi = (new HSMFTextExtactor(new MAPIMessage(new FileInputStream(
|
||||
String mapi = (new OutlookTextExtactor(new MAPIMessage(new FileInputStream(
|
||||
samples.getFile("simple_test_msg.msg")
|
||||
))).getText());
|
||||
|
Loading…
Reference in New Issue
Block a user