check whether string variable contains characters from another variable

[Tim Grünebaum]

Hi all,

one short question:
I want to check if - as in the title - a string variable (var1) contains characters (or the exact realization) from another string variable (var2).

Example:

var1	var2	var3
blabla AAA blabla	AAA	1
blabla blabla	AAA	0
bla bla	AAA	0
bla bla bla	BBB	0
bla bla BBB	BBB	1
bla BBB blabla	BBB	1
CCC bla bla	CCC	1

thus something like

gen var3=0
replace var3=1 if strpos(var1,"var2")

which of course does not work because s2 in strpos is not supposed to be a variable but a certain word.
The variable thing makes it complicated to me.

Thanks in advance

Tim

[Nick Cox]

s2 can certainly be a string variable, or indeed any string expression.

gen var3=strpos(var1, var2) > 0

[Tim Grünebaum]

Stupid me.
Thank you, Nick! Helpful as always!

[Straso Jovanovski]

I am just interested, is there a way to have a check like this, but from among any of the observations of a variable, not only direct matches. In other words, to be able to go line by line and take a portion of a string and look across all observations in the other variable to check whether that partial string is contained within any of the observations of the other variable.

[Nick Cox]

Some technique:

* Example generated by -dataex-. To install: ssc install dataex
clear
input str4(y1 y2)
"frog" "toad"
"frog" "toad"
"toad" "toad"
"newt" "newt"
"frog" "newt"
end

. levelsof y1 , local(levels)
`"frog"' `"newt"' `"toad"'

. foreach level of local levels {
  2. list y2 if strpos(y2, "`level'")
  3. }

     +------+
     |   y2 |
     |------|
  4. | newt |
  5. | newt |
     +------+

     +------+
     |   y2 |
     |------|
  1. | toad |
  2. | toad |
  3. | toad |
     +------+

. forval j = 1/`=_N
{ required
r(100);


. forval j = 1/`=_N' {
  2. di "`j'"
  3. list y2 if strpos(y2, y1[`j'])
  4. }
1
2
3

     +------+
     |   y2 |
     |------|
  1. | toad |
  2. | toad |
  3. | toad |
     +------+
4

     +------+
     |   y2 |
     |------|
  4. | newt |
  5. | newt |
     +------+
5

[Straso Jovanovski]

Thank you as always Nick Cox. Your solution worked very well. Now, if there would only be a way to condition the items returned by the -list- command based on another shared field. For example, what I am working with is a list of cities located within states. I would like to end up with a list of cities (var1) that are contained within the other variable (var2) only if they share the same state. That is, If there would be a way to introduce some sort of an if command/function in the code. I include below a sample of the data that I am working with. I want to check whether (and which ones) cities in variable "City" (_merge==2) are found as part of the variable containing a more complete version of place names (NAMELSAD) in _merge==1. I am really trying to check whether I could increase my overall match count by finding any places/cities not captured by using an exactly matching common ID. In this direction, I wish there was a way to use a more flexible ID option when merging two datasets - for example, being able to merge on a partial match (portion of a string) between the two ID fields, master and using.

. list

     +-------------------------------------------------------------------------------------+
     |           City                          NAMELSAD            State            _merge |
     |-------------------------------------------------------------------------------------|
  1. |                                     Tri-City CDP           Oregon   master only (1) |
  2. |                                 South Weldon CDP   North Carolina   master only (1) |
  3. |                                 Vaughnsville CDP             Ohio   master only (1) |
  4. |                                 Hunts Point town       Washington   master only (1) |
  5. |                              West Livingston CDP            Texas   master only (1) |
     |-------------------------------------------------------------------------------------|
  6. |                                 Freeland borough     Pennsylvania   master only (1) |
  7. |                                       Piedra CDP         Colorado   master only (1) |
  8. |                  Parcelas La Milagrosa comunidad      Puerto Rico   master only (1) |
  9. |                                      Upsala city        Minnesota   master only (1) |
 10. |                                     Savannah CDP            Texas   master only (1) |
     |-------------------------------------------------------------------------------------|
 11. |                                       Adams city        Wisconsin   master only (1) |
 12. |                                   Mansfield town       Washington   master only (1) |
 13. |     Rocky Hill                                        Connecticut    using only (2) |
 14. |    Mont Vernon                                      New Hampshire    using only (2) |
 15. |      Chikaming                                           Michigan    using only (2) |
     |-------------------------------------------------------------------------------------|
 16. |         Pelham                                      New Hampshire    using only (2) |
 17. | Sagamore Hills                                               Ohio    using only (2) |
 18. |         Covert                                           Michigan    using only (2) |
 19. |      Dartmouth                                      Massachusetts    using only (2) |
 20. |      Lexington                                           Kentucky    using only (2) |
     |-------------------------------------------------------------------------------------|
 21. |         Benton                                           Michigan    using only (2) |
 22. |            Lee                                      New Hampshire    using only (2) |
 23. |         Palmer                                      Massachusetts    using only (2) |
     +-------------------------------------------------------------------------------------+