Re: st: Extract identifier in the middle of a string variable

[Ghislaine Vantomme <ghislaine.vantomme@gmail.com>]

Dear Robert and Nick,

sorry for the delay to answer.

I finally implemented your two solutions. The one by Nick yields a "r"
error message in the loop but the moss solution by Robert work
perfectly. I was able to adapt it to subsequent query.

I also take this mail as an opportunity to wish all the Statlist users
a merry Christmas.

Thanks a lot !!
Best
Ghislaine

2012/12/14 Robert Picard <picard@netbox.com>:
> Here is how I would do it with -moss- (SSC). The pattern looks for
> strings that start with "NYSE:" and matches any characters that follow
> that are not a ";". Since some observations do not end with a ";", the
> final "*" makes the last semi-colon optional. All NYSE identifiers are
> matched. I assume that when there are multiple identifiers, they all
> start the same way so I then remove the period and what follows in the
> first match if there was a second match ("D.PRI" become "D" for the
> last observation).
>
> clear
>
> input str70 s
> "BMV:STP N;  DB:S9H;  NYSE:STP;  XTRA:S9H"
> "BMV:STP N;  DB:S9H; XTRA:S9H"
> "BASE:C;  BAYB:C;  BMV:C *;  BST:TRVC;  BVL:C;  DB:TRVC;  NYSE:C.PRH;"
> "NYSE:C;  NYSE:C.PRI;  SOMA:CTGP11B;  TSE:8710;  XTRA:TRV"
> "SOMA:CTGP11B;  TSE:8710;  XTRA:TRV;  NYSE:D.PRI;  NYSE:D;  "
> end
>
> moss s, match("NYSE:([^;]+);*") regex
>
> replace _match1 = regexr(_match1,"\..*","") if !mi(_match2)
> list _match1 s
>
>
> On Fri, Dec 14, 2012 at 5:19 AM, Nick Cox <njcoxstata@gmail.com> wrote:
>> I tried various solutions with -moss- (SSC) but kept messing up the
>> regular expression. Robert Picard is more fluent than I with regexp.
>>
>> Here is a more simple-minded approach:
>>
>> clear
>> input str80 myid
>> "BASE:C;  BAYB:C;  BMV:C *;  BST:TRVC;  BVL:C;  DB:TRVC;  NYSE:C.PRH;"
>> "NYSE:C;  NYSE:C.PRI;  SOMA:CTGP11B;  TSE:8710;  XTRA:TRV"
>> end
>> gen wc = wordcount(myid)
>> gen nyse = ""
>> su wc, meanonly
>> qui forval i = 1/`r(max)' {
>>         replace nyse = nyse + substr(word(myid, `i'), 6, .) + " " if
>> substr(word(myid, `i'), 1, 4) == "NYSE"
>> }
>> replace nyse = trim(subinstr(nyse, ";", "", .))
>> split nyse
>> local nsplit : word count `r(varlist)'
>> gen shortest = nyse1
>> qui forval i = 2/`nsplit' {
>>         replace shortest = nyse`i' if length(nyse`i') < length(shortest) &
>> length(nyse`i')
>> }
>>
>>
>> On Fri, Dec 14, 2012 at 12:46 PM, Ghislaine Vantomme
>> <ghislaine.vantomme@gmail.com> wrote:
>>> BASE:C;  BAYB:C;  BMV:C *;  BST:TRVC;  BVL:C;  DB:TRVC;  NYSE:C.PRH;
>>> NYSE:C;  NYSE:C.PRI;  SOMA:CTGP11B;  TSE:8710;  XTRA:TRV
>>>
>>>
>> *
>> *   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/
*
*   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/