Hi Ben,
thank you for the information. The commands I mention in the previous email are terminal commands. Your file is kind of difficult because it contains:
a) two different character encodings (latin-1 and utf-8) for variable description and the actual data
b) <CR><LF> within the text elements inside the variable data section
Let me know if the procedure with perl and grep works.
Friedrich If you complete a survey that I designed for my thesis on this site: http://studentenforschung.de/ the results are gathered by the site in the form of these csv files. There is an alternative file that seems to be output to csv format in a different manner - the two are labelled "abc" and "012", but I really don't see much of a difference. I attached the alternative file with "012" I am horrified at the programming of these guys and recommend everyone to do their surveys somewhere else. From that discussion, this must be where the problem is at: "At least one survey website generates text files, intended for use with SPSS, whose lines end in only CRs (not LF or CR+LF). PSPP does not understand this format. SPSS's behavior is not well understood in this area. " I'll try to reproduce what you did in Emacs, resp. the commands you kindly provided. Are they Emacs commands or terminal commands? However I wonder why CR LF are not visible in my text editor (Kate from KDE)... Thanks, Ben
Am So, 2. Nov 2014, um 18:34, schrieb Friedrich Beckmann:
Hi Ben,
can you tell me how you produced the original csv file? There is some discussion about the handling of CR and LF here:
https://savannah.gnu.org/bugs/?40605
Friedrich
<ver40-012.20.csv>
|