Hi,

Say I have an Excel file c:\temp\test.xls containing:

TEST (Row 1, the variable name) ABC ABC<Alt-Enter>DEF ABC<Alt-Enter>DEF<Alt-Enter>GHI

So, in Excel, it looks like (pretend the horrid ASCII "art" are gridlines)

TEST ---- ABC ---- ABC DEF ---- ABC DEF GHI ----

If I import the XLS file like so:

proc import file="c:\temp\test.xls" out=test1 dbms=excel replace; getnames=yes; run;

proc print; format test $hex.; run;

I get three observations. The data still has the embedded Alt-Enter (hex code 0A). Which is what I want.

However, if I save the Excel worksheet as CSV, my CSV file looks like:

TEST ABC "ABC DEF" "ABC DEF GHI"

And if I import the CSV like so:

proc import file="c:\temp\test.csv" out=test dbms=csv replace; getnames=yes; guessingrows=32767; run;

proc print; run;

I get 6 observations, which mirror the CSV file above.

Is there a way to import the CSV file such that it mimics the import of the XLS file? I don't care if PROC IMPORT can't do it; I'm happy to write a "manual" data step.

Perhaps with some creative use of recfm=N so that "0A"x becomes just another character, instead of an end-of-record indicator?

I guess the logic would be:

* if this data is not quoted, read until end-of-record * if this data is quoted, read until terminating quotation mark, ignoring any end-of-record characters.

Free beer if it also performs well ;-)

Cheers, Scott