sep=";" statement breaks utf8 BOM in CSV file which is generated by XSL
csv, excel, xslt
Solution
You are right, there is no way in Excel 2007 to get it load both the encoding and the seperator correctly across different locales when someone double clicks a CSV file.
It seems like when you specify `sep=` after the BOM it forgets the BOM has told it that it is UTF-8.
You have to specify the BOM because in certain locales Excel does not detect the separator. For instance in Danish, the default separator is `;`. If you output tab or comma-separated text then it does not detect the separator and in other locales, if you separate with semi-colon it doesn't load. You can test this by changing the locale format in Windows settings - Excel then picks this up.
From this question: Is it possible to force Excel recognize UTF-8 CSV files automatically?
and the answers it seems the only way is to use UTF-16 LE encoding with BOM.
Note also that as per http://wiki.scn.sap.com/wiki/display/ABAP/CSV+tests+of+encoding+and+column+separator?original_fqdn=wiki.sdn.sap.com it seems that if you use utf16-le with tab separators then it works.
I've wondered if Excel reads `sep=;` and then re-calls the method to get the CSV text and loses the BOM - I've tried giving incorrect text and I can't find any workaround that tells Excel to take both the `sep` and the encoding.
Problem
I'm currently developing CSV export with XSLT. And CSV file will be used %99 percent with Excel in my case, so I have to consider Excel behavior. My first problem was German special characters in csv. Even fact that CSV encoding is UTF8, Excel cannot open properly CSV file with UTF8. The special characters are getting weird symbols. I found a solution for this problem. I just added 3 additional bytes(EF BB BF - a.k.a BOM Header) beginning of content bytes. Because UTF8 BOM is way to say that 'hey dude, it is UTF8, open it properly' to Excel. Problem solved! And my second problem was about separator. The default separator could be comma or semicolon depending on region. I think it is semicolon in Germany and comma in UK. So, in order to prevent this problem, I had to add the line in below: ``` <xsl:text>sep=;</xsl:text> ``` or ``` <xsl:text>sep=,</xsl:text> ``` (This separator was not implemented as hard-coded) But my problem which I cannot find any solution is that if you add "sep=;" or "sep=," beginning of the file while the CSV file is being generated with UT8-BOM, the BOM doesn't help for showing special characters properly anymore! And I'm sure that BOM bytes are always in the beginning of byte array. This screen shot is from MS Excel in Mac OS X: First 3 symbols belong to BOM header. Have you ever had like this problem or do you have any suggestions? Thank you. Edit: I share the printscreens. a. With BOM and `<xsl:text>sep=;</xsl:text>` b. Just with BOM The Java code: ``` // Write the bytes ServletOutputStream out = resp.getOutputStream(); if(contentType.toString().equals("CSV")) { // The additional bytes in below is prefix indicates that the content is in UTF-8. out.write(239); out.write(187); out.write(191); } out.write(bytes); // Content bytes, in this case XSL ``` The XSL code: ``` <?xml version="1.0" encoding="UTF-8"?> <xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"> <xsl:output method="text" version="1.0" encoding="UTF-8" indent="yes" /> <xsl:template match="/"> <xsl:text>sep=;</xsl:text> <table> ... </table> </xsl:template> ```