Feature Request: Port CRAN's fuzzyjoin Funtions to dplyr

[jtelleriar]

I would suggest porting some of CRAN's fuzzyjoin package functions to dplyr (With proper Contribution Acknowledgement):

https://cran.r-project.org/web/packages/fuzzyjoin/index.html

Of special interest is "fuzzy_join" function, which allows to do SQL queries such as these:

SELECT
    REPLACE(PATH, 'W:\\', '\\\\naswoav.usr.corp.gamesa.es\\WO_AV_DATA\\') AS PATH
  FROM
    dispoin.t_DISPOIN_DIR AS A
    INNER JOIN dispoin.t_scadas AS B
      ON
        A.PATH LIKE CONCAT('%', ScadaCode, '%')
        AND (A.PATH LIKE '%2016%' OR A.PATH LIKE '%2017%')
        -- También se puede usar  A.PATH REGEXP '2016|2017';
        AND A.PATH NOT LIKE '%Data_10M%'
        AND A.PATH NOT LIKE '%rejectAL%'
        AND A.PATH NOT LIKE '%REPROCESSED_FILES%'
        AND A.PATH NOT LIKE '%GA677_201308_AL.csv.s3542017.zip%';

[krlmlr]

Thanks. Are you suggesting to run the fuzzy join directly on the database?

[jtelleriar]

Either in database or in R itself.

However, when you query a SQL database I guess you could always do a cheat like this I guess:

t_DISPOIN_DIR <- tbl(con, "t_DISPOIN_DIR")

t_DISPOIN_DIR %>%
sql(
"
SELECT
    REPLACE(PATH, 'W:\\', '\\\\naswoav.usr.corp.gamesa.es\\WO_AV_DATA\\') AS PATH
  FROM
    dispoin.t_DISPOIN_DIR AS A
    INNER JOIN dispoin.t_scadas AS B
      ON
        A.PATH LIKE CONCAT('%', ScadaCode, '%')
        AND (A.PATH LIKE '%2016%' OR A.PATH LIKE '%2017%')
        -- También se puede usar  A.PATH REGEXP '2016|2017';
        AND A.PATH NOT LIKE '%Data_10M%'
        AND A.PATH NOT LIKE '%rejectAL%'
        AND A.PATH NOT LIKE '%REPROCESSED_FILES%'
        AND A.PATH NOT LIKE '%GA677_201308_AL.csv.s3542017.zip%';
")

[krlmlr]

fuzzyjoin seems to be removed from CRAN, are you aware of a successor?

[jtelleriar]

At the company where I work we have GitHub Webpage Blocked, and I cannot check properly this issue.

The point is that if the package has been removed from CRAN, there is even more reason to fork some of its functions (as they are) and move them to mainstream dplyr.

I checked them out in the past and they are properly written.

The fuzzyjoin GitHub Repository is at the following link:

https://github.com/dgrtwo/fuzzyjoin

I'll check it out when I arrive at home, though these days I am little busy learning the H2O.ai framework, and doing a couple of Cheatsheets for using its machine learning functions.

Juan

[jtelleriar]

The fuzzyjoin package indeed seems not to be longer actively maintained.

[batpigandme]

Just released a new version on CRAN today:
https://cran.r-project.org/web/packages/fuzzyjoin/index.html

[krlmlr]

Good to see it back. Will think about it in the context of #2240.

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/