Special characters in XML files created with QDomDocument

[jim_kaiser]

So, sounds like you needs ASCII char 10 in your data. Did you try this?

@ QString contents = QString("Foo %1 Bar").arg((char)10); @

Remove the spaces around %1 .. seems code tag is eating it without the spaces..

[ Edit: No different from putting a \n ofcourse.. that doesn't work in excel?? ]

[ Edit2: Just use a string with \n while in Qt.. in the end when you save to an excel file... convert characters to excels format. That should work no? Okay my bad.. i see what you need.. the special chars from excel in Qt.. ]

[Chris Kawa]

When you use CDATA everything you put there will be used as is, without any parsing so no wonder it shows untransformed 

Maybe you tried it already, but I think the mix of those two solutions i.e. CDATA and \n should work.

[Edit.] Heh, just a wild idea just now if that doesn't work. You can combine CDATA and text like so:
@ <![CDATA[some text]]>&#10;<![CDATA[some more text]]> @

[goetz]

Guys! Get you some book about basic XML!

XML wise all this is equivalent:

@
&#10; == '\n' == QChar(10)
@

It is completely irrelevant if you put &#10; or '\n' into an XML parser! If it makes a difference, the parser is not standards compliant.

And just to prove: Make a hand crafted xlsx file an just put a newline in it. At least in OpenOffice it is displayed with a line break. I didn't have a "genuine" Excel at hand to test.

[ionwind]

jim_kaiser, just to mention that \n doesn't work in Excel. Anyway the last solution you posted would work of course. I take it you mean something like this:
@QFile exportFile(fileName);
if(exportFile.open(QIODevice::WriteOnly))
{
QTextStream TextStream(&exportFile);
QString docString = doc->toString().replace("&ampamp;", "&");
TextStream << docString;
exportFile.close();
}@
I could do it that way, yes. In my case the exported XML files have thousands of lines and the largest ones have "only" 30-50 thousand lines, so it's not a problem, even though searching and replacing bits of that large string just isn't programmatically good-looking ;P

Krzysztof, thanks for the suggestions but so far any combination of CDATA and any newline sequence doesn't seem to do the trick :\

[Chris Kawa]

@Volker It doesn't make a difference if it's in a text node, but id does if it's in the CDATA node. &#10; is a 4-letter string there and not a newline mark.
Other than that what you said is absolutely true - that's why I think \n (or QChar(10) which is not necessarily the same with windows 2-char newlines is it?) should work just fine inside CDATA node(used to have those & and <), but if ionwind says it doesn't than I don't really know..

[zester]

@Volker when ubuntu finishes there Qt Unity Interface and all those gtk developers turn into qt developers,
and you have 40,000 american ubuntu users swarming the qt forums wanting help learning qt.

There going to chew you up and spit you out my friend, lol If I was you I would work on my people skills.
and how do I know this because I did gtk development support for ubuntu at one time I had a single thread
about gtk programming. It received 142,000 hits and 3,000 comments and everyone wanted to be kindly spoon-feed.
If you want proof of my claims just ask I am sure I can dig that old thread up.

just saying ;)

Number of Ubuntu users = 12+ million :)
Ubuntu's current focus = Qt

[ionwind]

In case someone wants to try it out, here's a xml file that can be opened in Excel. On the first cell (that couldn't be made with Qt, so far) there's a newline that Excel understands, the other three are suggestions made here that can be produced by Qt but won't work in Excel.
Thanks for everyone anyway ;D
@<?xml version="1.0"?>
<?mso-application progid="Excel.Sheet"?>
<Workbook xmlns:ss="urn:schemas-microsoft-com:office:spreadsheet">
<Worksheet ss:Name="Newline">
<Table>
<Row>
<Cell><Data ss:Type="String">Foo&amp#10;Bar</Data></Cell>
<Cell><Data ss:Type="String">Foo&ampamp;#10;Bar</Data></Cell>
<Cell><Data ss:Type="String">Foo<![CDATA[&amp#10;]]>Bar</Data></Cell>
<Cell><Data ss:Type="String">Foo\nBar</Data></Cell>
</Row>
</Table>
</Worksheet>
</Workbook>@

[Edit] There should be xmlns="urn:schemas-microsoft-com:office:spreadsheet" in Workbook element too, but for some reason it won't appear in the code :|

[mlong]

You could always render your well-formed XML into a QString (or QByteArray, maybe?) and then use a QRegExp to malform it into something hideous that Excel can understand (i.e., replace all instances of &#10; with )

It's a dirty hack, and not without it's limitations, but it might work...

[goetz]

Hi,

the following example works for me.

It creates an xlsx file with two cells. The first contains a text without line breaks, the second cell has a line break after each word.

It's displayed correctly in OpenOffice here. As I'm in a train now and have no access to a windows box, I cannot check on genuine Microsoft Office's Excel - I'll do that on Tuesday. I'd be happy to get feedback though.

As you can see, it's all Qt DOM methods in use, no fancy search and replace magic.

@
#include <QApplication>
#include <QDebug>
#include <QtXml>
#include <QFile>
#include <QTextStream>

QDomElement appendDomNode(QDomNode &parent, const QString &name, const QString &content = QString())
{
QDomDocument doc = parent.ownerDocument();
QDomElement elem = doc.createElement(name);
if(!content.isNull()) {
QDomCharacterData pcdata = doc.createTextNode(content);
elem.appendChild(pcdata);
}
parent.appendChild(elem);
return elem;
}

int main(int argc, char *argv[])
{
QApplication a(argc, argv);

QDomDocument doc;
QDomProcessingInstruction xmlVers = doc.createProcessingInstruction("xml", "version=\"1.0\" encoding='utf-8'");
doc.appendChild(xmlVers);

QDomElement Workbook = doc.createElementNS("urn:schemas-microsoft-com:office:spreadsheet", "ss:Workbook");
doc.appendChild(Workbook);

QDomElement Styles = appendDomNode(Workbook, "ss:Styles");
QDomElement Style = appendDomNode(Styles, "ss:Style");
Style.setAttribute("ss:ID", "1");

QDomElement Worksheet = appendDomNode(Workbook, "ss:Worksheet");
Worksheet.setAttribute("ss:Name", "Sheet1");

QDomElement Table = appendDomNode(Worksheet, "ss:Table");

QDomElement Column_1 = appendDomNode(Table, "ss:Column");
Column_1.setAttribute("ss:Width", "80");

QDomElement Column_2 = appendDomNode(Table, "ss:Column");
Column_2.setAttribute("ss:Width", "80");

QDomElement Row = appendDomNode(Table, "ss:Row");
Row.setAttribute("ss:StyleID", "1");

QDomElement Cell_1 = appendDomNode(Row, "ss:Cell");
QDomElement Data_1 = appendDomNode(Cell_1, "ss:Data", "A string without line breaks");
Data_1.setAttribute("ss:Type", "String");

QDomElement Cell_2 = appendDomNode(Row, "ss:Cell");
QDomElement Data_2 = appendDomNode(Cell_2, "ss:Data", "A\nstring\nwith\nline\nbreaks");
Data_2.setAttribute("ss:Type", "String");

qDebug() << doc.toString();

QFile xlsxFile&#40;"/tmp/qttest.xlsx"&#41;;
xlsxFile.open(QIODevice::WriteOnly);
QTextStream ts(&xlsxFile);
doc.save(ts, 4);
xlsxFile.close();

return 0;

}
@

[goetz]

Here are the results of the checks with Microsoft's Office Excel:

• one has to rename the file from .xlsx to .xml to be able to import it
• using a literal newline character ('\n') in the strings does not work
• replacing the literal newline with '&#10;' (aka entity) does work
• one must put a suitable cell format into the xml to enable line breaks

So, the conclusion is, that the XML parser that is used by Excel to handle the file is definitely broken and does not follow the standards!

You can easily test this by using xmllint: Save a file from excel in XML format, have xmllint parse it and save the result (this replaces all '&#10;' with literal newlines) and try to open the file in Excel. It will fail.

Unfortunately I see no chance in getting some easy support for this case into Qt. You would have to patch method "static QString encodeText()" in qdom.cpp to escape newlines.