Extracting multiple dates from string with varying date formats

[Girish Venkataraman]

Hello all: I have an issue with extracting multiple dates with varying formats as MM/DD/YYYY, M/DD/YY per observation (basically missing leading zeros for month and day randomly sometimes). Any help. Did not use dataex for this since this was a small usecase. Pardon that in advance. Tried moss install from a post by Nick but my data looks messier and I don't know how best to tweak the code.

var_finaldiagnosistext
Accession1. Node, biopsy (outside slides S21-334, 2/03/21): Hodgkin lymphoma, see microscopy
Accession2. A. Node, biopsy (outside slides dated SMS-20-44, dated 03/3/2020): Diffuse large B-cell lymphoma, see microscopic description. B. Bone marrow, biopsy (outside slides BMS-19-44, 01/01/19) : No evidence of lymphoma.

Desired outcome

date1 date2
02/03/2021
03/03/2020 01/01/2019

[Andrew Musau]

Increase the number of iterations in the loop to match the maximum number of dates to be extracted.

* Example generated by -dataex-. For more info, type help dataex
clear
input str225 var_finaldiagnosistext
"Accession1. Node, biopsy (outside slides S21-334, 2/03/21): Hodgkin lymphoma, see microscopy"                                                                                                                                    
"Accession2. A. Node, biopsy (outside slides dated SMS-20-44, dated 03/3/2020): Diffuse large B-cell lymphoma, see microscopic description. B. Bone marrow, biopsy (outside slides BMS-19-44, 01/01/19) : No evidence of lymphoma."
end

gen text=var_finaldiagnosistext
forval i=1/3{
    gen wanted`i'=daily(ustrregexra(text,".*\b([\d]+/[\d]+/[\d]+)\b.*", "$1"), "MDY", 2030)
    format wanted`i' %td
    replace text= subinstr(text, ustrregexra(text,".*\b([\d]+/[\d]+/[\d]+)\b.*", "$1"), "", 1)
}

Res.:

. l wanted*

     +---------------------------------+
     |   wanted1     wanted2   wanted3 |
     |---------------------------------|
  1. | 03feb2021           .         . |
  2. | 01jan2019   03mar2020         . |
     +---------------------------------+

[Girish Venkataraman]

Thanks so much, Andrew. Tried it in my Stata11 and its giving me an error as below. is there some module I need to install since its an old version? Guess this is an expression available only from Stata14. Will try in Stata17 which I will install and try shortly.

unknown function ustrregexra()
r(133);

[Andrew Musau]

The Unicode regular expression functions were added in Stata 14, so you need that version or above.