Add verifying test-case for bug 52420

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1649147 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Dominik Stadler 2015-01-02 22:38:52 +00:00
parent b930f0c46c
commit d909e0998c
2 changed files with 83 additions and 74 deletions

View File

@ -39,18 +39,18 @@ public class TestWordToHtmlConverter extends POITestCase
private static String getHtmlText( final String sampleFileName ) private static String getHtmlText( final String sampleFileName )
throws Exception throws Exception
{ {
return getHtmlText( sampleFileName, false ); return getHtmlText( sampleFileName, false);
} }
private static String getHtmlText( final String sampleFileName, private static String getHtmlText( final String sampleFileName,
boolean emulatePictureStorage ) throws Exception boolean emulatePictureStorage ) throws Exception
{ {
HWPFDocument hwpfDocument = new HWPFDocument( POIDataSamples HWPFDocument hwpfDocument = new HWPFDocument( POIDataSamples
.getDocumentInstance().openResourceAsStream( sampleFileName ) ); .getDocumentInstance().openResourceAsStream( sampleFileName ));
Document newDocument = XMLHelper.getDocumentBuilderFactory().newDocumentBuilder().newDocument(); Document newDocument = XMLHelper.getDocumentBuilderFactory().newDocumentBuilder().newDocument();
WordToHtmlConverter wordToHtmlConverter = new WordToHtmlConverter( WordToHtmlConverter wordToHtmlConverter = new WordToHtmlConverter(
newDocument ); newDocument);
if ( emulatePictureStorage ) if ( emulatePictureStorage )
{ {
@ -62,21 +62,21 @@ public class TestWordToHtmlConverter extends POITestCase
{ {
return suggestedName; return suggestedName;
} }
} ); });
} }
wordToHtmlConverter.processDocument( hwpfDocument ); wordToHtmlConverter.processDocument( hwpfDocument);
StringWriter stringWriter = new StringWriter(); StringWriter stringWriter = new StringWriter();
Transformer transformer = TransformerFactory.newInstance() Transformer transformer = TransformerFactory.newInstance()
.newTransformer(); .newTransformer();
transformer.setOutputProperty( OutputKeys.INDENT, "yes" ); transformer.setOutputProperty( OutputKeys.INDENT, "yes");
transformer.setOutputProperty( OutputKeys.ENCODING, "utf-8" ); transformer.setOutputProperty( OutputKeys.ENCODING, "utf-8");
transformer.setOutputProperty( OutputKeys.METHOD, "html" ); transformer.setOutputProperty( OutputKeys.METHOD, "html");
transformer.transform( transformer.transform(
new DOMSource( wordToHtmlConverter.getDocument() ), new DOMSource( wordToHtmlConverter.getDocument() ),
new StreamResult( stringWriter ) ); new StreamResult( stringWriter ));
String result = stringWriter.toString(); String result = stringWriter.toString();
return result; return result;
@ -84,172 +84,181 @@ public class TestWordToHtmlConverter extends POITestCase
public void testAIOOBTap() throws Exception public void testAIOOBTap() throws Exception
{ {
String result = getHtmlText( "AIOOB-Tap.doc" ); String result = getHtmlText( "AIOOB-Tap.doc");
assertContains( result.substring( 0, 6000 ), "<table class=\"t1\">" ); assertContains(result.substring( 0, 6000 ), "<table class=\"t1\">");
} }
public void testBug33519() throws Exception public void testBug33519() throws Exception
{ {
String result = getHtmlText( "Bug33519.doc" ); String result = getHtmlText( "Bug33519.doc");
assertContains( assertContains(
result, result,
"\u041F\u043B\u0430\u043D\u0438\u043D\u0441\u043A\u0438 \u0442\u0443\u0440\u043E\u0432\u0435" ); "\u041F\u043B\u0430\u043D\u0438\u043D\u0441\u043A\u0438 \u0442\u0443\u0440\u043E\u0432\u0435");
assertContains( result, assertContains(result,
"\u042F\u0432\u043E\u0440 \u0410\u0441\u0435\u043D\u043E\u0432" ); "\u042F\u0432\u043E\u0440 \u0410\u0441\u0435\u043D\u043E\u0432");
} }
public void testBug46610_2() throws Exception public void testBug46610_2() throws Exception
{ {
String result = getHtmlText( "Bug46610_2.doc" ); String result = getHtmlText( "Bug46610_2.doc");
assertContains( assertContains(
result, result,
"012345678911234567892123456789312345678941234567890123456789112345678921234567893123456789412345678" ); "012345678911234567892123456789312345678941234567890123456789112345678921234567893123456789412345678");
} }
public void testBug46817() throws Exception public void testBug46817() throws Exception
{ {
String result = getHtmlText( "Bug46817.doc" ); String result = getHtmlText( "Bug46817.doc");
final String substring = "<table class=\"t1\">"; final String substring = "<table class=\"t1\">";
assertContains( result, substring ); assertContains(result, substring);
} }
public void testBug47286() throws Exception public void testBug47286() throws Exception
{ {
String result = getHtmlText( "Bug47286.doc" ); String result = getHtmlText( "Bug47286.doc");
assertFalse( result.contains( "FORMTEXT" ) ); assertFalse(result.contains( "FORMTEXT" ));
assertContains( result, "color:#4f6228;" ); assertContains(result, "color:#4f6228;");
assertContains( result, "Passport No and the date of expire" ); assertContains(result, "Passport No and the date of expire");
assertContains( result, "mfa.gov.cy" ); assertContains(result, "mfa.gov.cy");
} }
public void testBug48075() throws Exception public void testBug48075() throws Exception
{ {
getHtmlText( "Bug48075.doc" ); getHtmlText( "Bug48075.doc");
} }
public void testBug52583() throws Exception public void testBug52583() throws Exception
{ {
String result = getHtmlText( "Bug52583.doc" ); String result = getHtmlText( "Bug52583.doc");
assertContains( assertContains(
result, result,
"<select><option selected>riri</option><option>fifi</option><option>loulou</option></select>" ); "<select><option selected>riri</option><option>fifi</option><option>loulou</option></select>");
} }
public void testBug53182() throws Exception public void testBug53182() throws Exception
{ {
String result = getHtmlText( "Bug53182.doc" ); String result = getHtmlText( "Bug53182.doc");
assertFalse( result.contains( "italic" ) ); assertFalse(result.contains( "italic" ));
} }
public void testDocumentProperties() throws Exception public void testDocumentProperties() throws Exception
{ {
String result = getHtmlText( "documentProperties.doc" ); String result = getHtmlText( "documentProperties.doc");
assertContains( result, "<title>This is document title</title>" ); assertContains(result, "<title>This is document title</title>");
assertContains( result, assertContains(result,
"<meta content=\"This is document keywords\" name=\"keywords\">" ); "<meta content=\"This is document keywords\" name=\"keywords\">");
} }
public void testEmailhyperlink() throws Exception public void testEmailhyperlink() throws Exception
{ {
String result = getHtmlText( "Bug47286.doc" ); String result = getHtmlText( "Bug47286.doc");
final String substring = "provisastpet@mfa.gov.cy"; final String substring = "provisastpet@mfa.gov.cy";
assertContains( result, substring ); assertContains(result, substring);
} }
public void testEndnote() throws Exception public void testEndnote() throws Exception
{ {
String result = getHtmlText( "endingnote.doc" ); String result = getHtmlText( "endingnote.doc");
assertContains( assertContains(
result, result,
"<a class=\"a1 endnoteanchor\" href=\"#endnote_1\" name=\"endnote_back_1\">1</a>" ); "<a class=\"a1 endnoteanchor\" href=\"#endnote_1\" name=\"endnote_back_1\">1</a>");
assertContains( assertContains(
result, result,
"<a class=\"a1 endnoteindex\" href=\"#endnote_back_1\" name=\"endnote_1\">1</a> <span" ); "<a class=\"a1 endnoteindex\" href=\"#endnote_back_1\" name=\"endnote_1\">1</a> <span");
assertContains( result, "Ending note text" ); assertContains(result, "Ending note text");
} }
public void testEquation() throws Exception public void testEquation() throws Exception
{ {
String result = getHtmlText( "equation.doc" ); String result = getHtmlText( "equation.doc");
assertContains( result, "<!--Image link to '0.emf' can be here-->" ); assertContains(result, "<!--Image link to '0.emf' can be here-->");
} }
public void testHyperlink() throws Exception public void testHyperlink() throws Exception
{ {
String result = getHtmlText( "hyperlink.doc" ); String result = getHtmlText( "hyperlink.doc");
assertContains( result, "<span>Before text; </span><a " ); assertContains(result, "<span>Before text; </span><a ");
assertContains( result, assertContains(result,
"<a href=\"http://testuri.org/\"><span class=\"s1\">Hyperlink text</span></a>" ); "<a href=\"http://testuri.org/\"><span class=\"s1\">Hyperlink text</span></a>");
assertContains( result, "</a><span>; after text</span>" ); assertContains(result, "</a><span>; after text</span>");
} }
public void testInnerTable() throws Exception public void testInnerTable() throws Exception
{ {
getHtmlText( "innertable.doc" ); getHtmlText( "innertable.doc");
} }
public void testListsMargins() throws Exception public void testListsMargins() throws Exception
{ {
String result = getHtmlText( "lists-margins.doc" ); String result = getHtmlText( "lists-margins.doc");
assertContains( result, assertContains(result,
".s1{display: inline-block; text-indent: 0; min-width: 0.4861111in;}" ); ".s1{display: inline-block; text-indent: 0; min-width: 0.4861111in;}");
assertContains( result, assertContains(result,
".s2{display: inline-block; text-indent: 0; min-width: 0.23055555in;}" ); ".s2{display: inline-block; text-indent: 0; min-width: 0.23055555in;}");
assertContains( result, assertContains(result,
".s3{display: inline-block; text-indent: 0; min-width: 0.28541666in;}" ); ".s3{display: inline-block; text-indent: 0; min-width: 0.28541666in;}");
assertContains( result, assertContains(result,
".s4{display: inline-block; text-indent: 0; min-width: 0.28333333in;}" ); ".s4{display: inline-block; text-indent: 0; min-width: 0.28333333in;}");
assertContains( result, assertContains(result,
".p4{text-indent:-0.59652776in;margin-left:-0.70069444in;" ); ".p4{text-indent:-0.59652776in;margin-left:-0.70069444in;");
} }
public void testO_kurs_doc() throws Exception public void testO_kurs_doc() throws Exception
{ {
getHtmlText( "o_kurs.doc" ); getHtmlText( "o_kurs.doc");
} }
public void testPageref() throws Exception public void testPageref() throws Exception
{ {
String result = getHtmlText( "pageref.doc" ); String result = getHtmlText( "pageref.doc");
assertContains( result, "<a href=\"#userref\">" ); assertContains(result, "<a href=\"#userref\">");
assertContains( result, "<a name=\"userref\">" ); assertContains(result, "<a name=\"userref\">");
assertContains( result, "1" ); assertContains(result, "1");
} }
public void testPicture() throws Exception public void testPicture() throws Exception
{ {
String result = getHtmlText( "picture.doc", true ); String result = getHtmlText( "picture.doc", true);
// picture // picture
assertContains( result, "src=\"0.emf\"" ); assertContains(result, "src=\"0.emf\"");
// visible size // visible size
assertContains( result, "width:3.1305554in;height:1.7250001in;" ); assertContains(result, "width:3.1305554in;height:1.7250001in;");
// shift due to crop // shift due to crop
assertContains( result, "left:-0.09375;top:-0.25694445;" ); assertContains(result, "left:-0.09375;top:-0.25694445;");
// size without crop // size without crop
assertContains( result, "width:3.4125in;height:2.325in;" ); assertContains(result, "width:3.4125in;height:2.325in;");
} }
public void testPicturesEscher() throws Exception public void testPicturesEscher() throws Exception
{ {
String result = getHtmlText( "pictures_escher.doc", true ); String result = getHtmlText( "pictures_escher.doc", true);
assertContains( result, "<img src=\"s0.PNG\">" ); assertContains(result, "<img src=\"s0.PNG\">");
assertContains( result, "<img src=\"s808.PNG\">" ); assertContains(result, "<img src=\"s808.PNG\">");
} }
public void testTableMerges() throws Exception public void testTableMerges() throws Exception
{ {
String result = getHtmlText( "table-merges.doc" ); String result = getHtmlText( "table-merges.doc");
assertContains( result, "<td class=\"td1\" colspan=\"3\">" ); assertContains(result, "<td class=\"td1\" colspan=\"3\">");
assertContains( result, "<td class=\"td2\" colspan=\"2\">" ); assertContains(result, "<td class=\"td2\" colspan=\"2\">");
}
public void testBug52420() throws Exception {
String result = getHtmlText( "52420.doc");
assertFalse(result.contains( "FORMTEXT" ));
assertContains(result, "\u0417\u0410\u0414\u0410\u041d\u0418\u0415");
assertContains(result, "\u041f\u0440\u0435\u043f\u043e\u0434\u0430\u0432\u0430\u0442\u0435\u043b\u044c");
} }
} }

Binary file not shown.