Re: st: insheet and dropping cases

[Phil Schumm <pschumm@uchicago.edu>]

On Feb 20, 2014, at 9:39 AM, Ben Hoen <bhoen@lbl.gov> wrote:
> Below is an example of four records that span such a scenario in that file (without the non-printable characters)
> 
> ***************************************************************************
> The last line it reads 
> ***************************************************************************

<snip>

> N 89 DEG 46'47" E 1308.19 FT TO WATSON RD; N/LY ALONG THE CURVE OF WATSON RD


It's the double-quote character here that is causing the problem; -insheet- sees it, and continues reading over multiple lines until it finds a matching close (double) quote or until EOF.  Prior to Stata 13, this would quickly exhaust the length limit of a string variable, but in Stata 13, that is no longer a limitation.

IIRC, -insheet- was very limited in how it dealt with double quotes.  Basically, it wanted to see them in pairs enclosing a single text value, but couldn't handle them otherwise (those who recall differently, please correct me if I'm wrong).

In Stata 13, -import delimited- is much more flexible WRT how double quotes are handled, providing the -bindquotes()- and -stripquotes()- options (see http://www.stata.com/manuals13/dimportdelimited.pdf for more information).  I just tried your example in Stata 13, and it works fine.

Thus, unless someone knows something undocumented about -insheet- that I don't (which is certainly possible), I'm afraid your only option is to upgrade (of course, there are a million other reasons to do so too).


-- Phil


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/