Simpler logical subsetting by strings with `grep`

grep_operator.Rmd

```{r, echo=FALSE}
library(magrittr)
```

# `%grep%` operator

Using `R`'s built-in `grep` function is really inconvenient for interactive work.
There already exists a convenient `%in%` operator for testing membership in a sequence.
However, real data analysis rarely presents with well-defined sequences. 
Strings are much more common.

## Implementation

```{r}
`%grep%` <- function(pattern, x) grepl(pattern, x)
```

That's all.

## Usage

The `%grep%` operator functions the same way that `%in%` does, but the lookup is within a string.

Direct string matching:

```{r}
# Select 'setosa' entries
iris['setosa' %grep% iris$Species, ] %>% head
```

Regular expressions:

```{r}
# Select all entries where species ends with 'a'
X <- iris['a$' %grep% iris$Species, ]
unique(X$Species)
```

## More complicated example

This allows us to produce more complicated chained selections.
I always forget how to write a regular expression for  "`one` but not `two`".
In case of the `%grep%` operator, we can use `R`'s logical functions.

Let's create some sample data:

```{r}
# Create some silly example
X <- data.frame(
  value = rnorm(100),
  fb = paste(
    sample(c('foo', 'bar'), 100, replace=T),
    sample(c('foo', 'bar'), 100, replace=T)))
X %>% head
```

And apply `%grep%`:

```{r}
# Select foo without bar (same as `foo foo`)
X['foo' %grep% X$fb & !'bar' %grep% X$fb, ] %>% head
```