Simplify strings with respect to substring - r

Question

I have a db with some repeated entries that reports (inconsistently) additional information. I would like to get rid of the information and keep the most simple version for each entry.
db <- data.frame(company=c("ENTRY_X","ENTRY_X COUNTY_1","COUNTY_2 ENTRY_X","ENTRY_Y"))
db_desiderata <- data.frame(company=c(rep("ENTRY_X",3),"ENTRY_Y"))
Entries are possibly lengthy strings (some with spaces). Some examples are: "General Motors Company" and "General Motors".
I manage to isolate all the entries that need to be substituted with their substring (in db$included).
I plan to run it recursively.
Attempted code (all works, I get stuck on how to proceed):
db$included <- lapply(db$company, function(x) c(grep(x,db$company,value=T)))
db$lenght <- lapply(db$included, function(x) length(unlist(x)))
db$included <- ifelse(db$lenght==1,NA,db$included)

Answer 1

The following should work if the data strictly conforms to these patterns:
The desired name must be the first in the sequence of alternative names
The desired name must be the shortest in the sequence of alternative names and can't be followed by a company name which is a shorter subset of the preceding company name.
I'll use a variation of Chuck P's data to illustrate how this works and the problems if the patterns aren't followed.
db <- data.frame(company = c("General Foods","More General Foods","General Foods Cereal Division","General Auto",
"General Motors Company", "General Motors", "European General Motors Company",
"General", "Asia General Toys") )
companies <- Reduce( f = function(y,x) {if(grepl(pattern = y, x=x)) y else x},
x=db$company, accumulate = TRUE)
which gives
companies
[1] General Foods General Foods General Foods General Auto General Motors Company
[6] General Motors General Motors General General

Answer 2

I think I understand your situation a little better after your comment but I still would be very wary of a fully automated solution, One slip or one term that was too general (pun-intended) and you're hosed...
I've taken your early work and done a little renaming. Think of your original length as more a measure of potential. I'd look at the potential column with a human eye and pick and choose the places to replace. I'd approach this with stringr::str_replace_all. If you use the named vector as I've shown below you should be able to handle a wide array of cases with cut and paste. "^.*General Motors.*$" just means if you find it anywhere in the string, front or back. You can work iteratively and just keep adding to the named vector until you have it cleaned.
library(dplyr)
library(stringr)
db <- data.frame(company = c("General Foods","More General Foods","General Foods Cereal Division","General", "General Auto", "General Motors Company", "General Motors", "European General Motors Company"))
db$similar_company <- sapply(db$company, function(x) c(grep(x, db$company, value=T)), simplify = TRUE)
db$potential <- sapply(db$similar_company, function(x) length(unlist(x)), simplify = TRUE)
glimpse(db)
#> Rows: 8
#> Columns: 3
#> $ company <chr> "General Foods", "More General Foods", "General Foods…
#> $ similar_company <named list> [<"General Foods", "More General Foods", "Gene…
#> $ potential <int> 3, 1, 1, 8, 1, 2, 3, 1
db %>% arrange(desc(potential)) %>% select(-similar_company)
#> company potential
#> 1 General 8
#> 2 General Foods 3
#> 3 General Motors 3
#> 4 General Motors Company 2
#> 5 More General Foods 1
#> 6 General Foods Cereal Division 1
#> 7 General Auto 1
#> 8 European General Motors Company 1
db$newcompany <-
str_replace_all(db$company, c("^.*General Foods.*$" = "General Foods",
"^.*General Motors.*$" = "General Motors"),
)
db %>% select(company, newcompany)
#> company newcompany
#> 1 General Foods General Foods
#> 2 More General Foods General Foods
#> 3 General Foods Cereal Division General Foods
#> 4 General General
#> 5 General Auto General Auto
#> 6 General Motors Company General Motors
#> 7 General Motors General Motors
#> 8 European General Motors Company General Motors
Created on 2020-05-08 by the reprex package (v0.3.0)