Re: st: Problem with -infix- with if qualifiers and strings?

[Robert Picard <picard@netbox.com>]

Here's how your code runs on my Mac using Stata 12.1. As you can see,
I don't lose an observation. I'm pretty sure that the last line of
your version of "test.txt" is missing a return character which causes
Stata to skip the last line. As to -infix-'s behavior, it must read in
values for -type- and -number- before it can try to evaluate -if
type==20-. That generates the errors you see. One solution is to read
the second number as as string and then convert it to float later, see
example below.

Robert

--------------- test.txt ------------------------
10ABC
20321
10ZYX
20654
-------------------------------------------------

. infix type 1-2 str text 3-5 if type==10 using test.txt
(2 observations read)

. list

     +-------------+
     | type   text |
     |-------------|
  1. |   10    ABC |
  2. |   10    ZYX |
     +-------------+

.
. clear

. infix type 1-2 number 3-5 if type==20 using test.txt
'ABC' cannot be read as a number for number[1]
'ZYX' cannot be read as a number for number[2]
(2 observations read)

. list

     +---------------+
     | type   number |
     |---------------|
  1. |   20      321 |
  2. |   20      654 |
     +---------------+

.
. clear

. infix type 1-2 str snumber 3-5 if type==20 using test.txt
(2 observations read)

. gen number = real(snumber)

. list

     +-------------------------+
     | type   snumber   number |
     |-------------------------|
  1. |   20       321      321 |
  2. |   20       654      654 |
     +-------------------------+

On Tue, Nov 12, 2013 at 5:13 PM, Jorge Eduardo Pérez Pérez
<jorge_perez@brown.edu> wrote:
> Dear Statalist.
>
> I have noticed weird behavior regarding -infix- and data where the
> variable type may change per line. -infix- is not reading the dataset
> properly.
>
> Here's an example. My dataset is:
>
> 10ABC
> 20321
> 10ZYX
> 20654
>
> If I try to read the lines that start with 10 and the remainder as a
> string, everything works:
>
> . clear
>
> . infix type 1-2 str text 3-5 if type==10 using test.txt
> (2 observations read)
>
> . list
>
>      +-------------+
>      | type   text |
>      |-------------|
>   1. |   10    ABC |
>   2. |   10    ZYX |
>      +-------------+
>
> But if I try to read the lines that start with 20, where the remainder
> is a number, Stata seems to be trying to read the lines that start
> with 10 as well, producing a "cannot be read" message and , worse,
> dropping observations from my data!
>
> . clear
>
> . infix type 1-2 number 3-5 if type==20 using test.txt
> 'ABC' cannot be read as a number for number[1]
> 'ZYX' cannot be read as a number for number[2]
> (1 observations read)
>
> . list
>
>      +---------------+
>      | type   number |
>      |---------------|
>   1. |   20      321 |
>      +---------------+
>
> I have replicated this in both Stata 12.1, Windows 7 and 13.1 on
> Windows 8. Can someone replicate to see if this is a bug, or am I
> missing something? In the meantime I will read my variables as strings
> and destring afterwards.
>
> Thanks!
>
> --------------------------------------------
> Jorge Eduardo Pérez Pérez
> Graduate Student
> Department of Economics
> Brown University
>
> *
> *   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/

*
*   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/