R function(): how to pass parameters which contain characters and regular expression - regex

Question

my data as follows:
>df2
id calmonth product
1 101 01 apple
2 102 01 apple&nokia&htc
3 103 01 htc
4 104 01 apple&htc
5 104 02 nokia
Now i wanna calculate the number of ids whose products contain both 'apple' and 'htc' when calmonth='01'. Because what i need is not only 'apple' and 'htc', also i need 'apple' and 'nokia',etc.
So i want to realize this by a function like this:
xandy=function(a,b) data.frame(product=paste(a,b,sep='&'),
csum=length(grep('a.*b',x=df2$product))
)
also, i make a parameters list like this:
para=c('apple','htc','nokia')
but the problem is here. When i pass parameters like
xandy(para[1],para[2])
the results is as follows:
product csum
1 apple&htc 0
What my expecting result should be
product csum calmonth
1 apple&htc 2 01
2 apple&htc 0 02
So where is wrong about the parameters passing?
and, how can i add the calmonth in to the function() xandy correctly?
FYI.This question stems from my another question before
What's the R statement responding to SQL's 'in' statement
EDIT AFTER COMMENT
My predictive result will be:
product csum calmonth
1 apple&htc 2 01
2 apple&htc 0 02

Answer 1

May answer is another way how to tackle your problem.
library(stringr)
The function contains will split up the elements of a string vector according to the split character and evaluate if all target words are contained.
contains <- function(x, target, split="&") {
l <- str_split(x, split)
sapply(l, function(x, y) all(y %in% x), y=target)
}
contains(d$product, c("apple", "htc"))
[1] FALSE TRUE FALSE TRUE FALSE
The rest is just subsetting and summarizing
get_data <- function(a, b) {
e <- subset(d, contains(product, c(a, b)))
e$product2 <- paste(a, b, sep="&")
ddply(e, .(calmonth, product2), summarise, csum=length(id))
}
Using the data below, order does not play a role now anymore (see comment below).
get_data("apple", "htc")
calmonth product2 csum
1 1 apple&htc 1
2 2 apple&htc 2
get_data("htc", "apple")
calmonth product2 csum
1 1 htc&apple 1
2 2 htc&apple 2
I know this is not a direct answer to your question but I find this approach quite clean.
EDIT AFTER COMMENT
The reason that you get csum=0 is simply that you are searching for the wrong regex pattern, i.e. a something in between b not for apple ... htc. You need to construct the correct regex pattern,i.e. paste0(a, ".*", b).
Here a complete solution. I would not call it beautiful code, but anyway (note that I change the data to show that it generalizes for months).
library(plyr)
df2 <- read.table(text="
id calmonth product
101 01 apple
102 01 apple&nokia&htc
103 01 htc
104 02 apple&htc
104 02 apple&htc", header=T)
xandy <- function(a, b) {
pattern <- paste0(a, ".*", b)
d1 <- df2[grep(pattern, df2$product), ]
d1$product <- paste0(a,"&", b)
ddply(d1, .(calmonth), summarise,
csum=length(calmonth),
product=unique(product))
}
xandy("apple", "htc")
calmonth csum product
1 1 1 apple&htc
2 2 2 apple&htc