Glue strings to data with glue

String interpolation, evaluating a variable name to a value within a string, isa feature of many programming languages including Python, Julia, Javascript, Rust, and most Unix Shells. R's `sprintf()` and `paste()` functions provide some of this functionality, but have limitations which make them cumbersome to use. There are also some existing add on packages with similar functionality,however each has drawbacks.The glue package performs robust string interpolation for R. This includes evaluation of variables and arbitrary R code,with a clean and simple syntax. Because it is dependency-free, it is easy to incorporate into packages. In addition, glue provides an extensible interface to perform more complex transformations; such as `glue_sql()` to construct SQL queries with automatically quoted variables.This talk will show how to utilize glue to write beautiful code which is easy to read, write and maintain. We will also discuss ways to best use glue when performance is a concern. Finally we will create custom glue functions tailored towards specific use cases, such as JSON construction, colored messages, emoji interpolation and more.

Transcript

1. rstd.io/glue
   GLUE STRINGS TO DATA WITH
   GLUE STRINGS TO DATA WITH
   Jim Hester
   
   
   
   @jimhester_
   jimhester
   jim.hester@rstudio.com
   

   View Slide

2. GLUING STRINGS IS EVERYWHERE
   GLUING STRINGS IS EVERYWHERE
   EXCEL
   EXCEL
   =CONCATENATE(A1, " ", B1)
   BASH
   BASH
   dir=/tmp
   for file in $dir/*; do
   cat $dir/$file
   done
   PYTHON
   PYTHON
   apples = 4
   print("I have {a} apples".format(a=apples))
   # Python 3.6+
   print(f"I have {apples} apples")
   

   View Slide

3. GLUING STRINGS IN R IS PAINFUL
   GLUING STRINGS IN R IS PAINFUL
   apples <- 3
   bananas <- 2
   paste0("Inventory", "\n",
   " Apples: ", apples, "\n",
   " Bananas: ", bananas, "\n",
   "Total: ", apples + bananas)
   ## Inventory
   ## Apples: 3
   ## Bananas: 2
   ## Total: 5
   sprintf("Inventory\n Apples: %i\n Bananas: %i\nTotal: %i",
   apples, bananas, apples + bananas)
   ## Inventory
   ## Apples: 3
   ## Bananas: 2
   ## Total: 5
   if (!file.exists(file)) {
   stop("'", file, "' not found")
   }
   

   View Slide

4. GLUE
   GLUE
   GLUE
   GLUE MAKES GLUING STRINGS EASY!
   MAKES GLUING STRINGS EASY!
   apples <- 3
   bananas <- 2
   glue("
   Inventory
   Apples: {apples}
   Bananas: {bananas}
   Total: {apples + bananas}")
   ## Inventory
   ## Apples: 3
   ## Bananas: 2
   ## Total: 5
   

   View Slide

5. GLUE
   GLUE
   GLUE
   GLUE IS CONVIENENT
   IS CONVIENENT
   if (TRUE) {
   glue("
   You can indent naturally \\
   and break up long lines \\
   if needed.
   ")
   }
   ## You can indent naturally and break up long lines if needed.
   

   View Slide

6. GLUE
   GLUE
   GLUE
   GLUE IS SAFE
   IS SAFE
   glue("{1:3} will recycle with a length of 1 {letters[1]}")
   ## 1 will recycle with a length of 1 a
   ## 2 will recycle with a length of 1 a
   ## 3 will recycle with a length of 1 a
   glue("{1:3} will recycle with the same length {letters[1:3]}")
   ## 1 will recycle with the same length a
   ## 2 will recycle with the same length b
   ## 3 will recycle with the same length c
   glue("{1:3} will not recycle with inconsistent length {letters[1:2]}")
   ## Error: Variables must be length 1 or 3
   

   View Slide

7. GLUE
   GLUE
   GLUE
   GLUE HANDLES MISSINGS
   HANDLES MISSINGS
   footies %>% mutate(glue("{first} {middle} {last}")) %>% pull()
   ## Tim Filiga Cahill
   ## Harry NA Kewell
   ## Mark NA Schwarzer
   footies %>% mutate(glue("{first} {middle} {last}", .na = NULL)) %>%
   pull()
   ## Tim Filiga Cahill
   ## NA
   ## NA
   footies %>% mutate(glue("{first} {middle} {last}", .na = "")) %>%
   pull()
   ## Tim Filiga Cahill
   ## Harry Kewell
   ## Mark Schwarzer
   

   View Slide

8. GLUE
   GLUE
   GLUE
   GLUE IS FOR PACKAGES
   IS FOR PACKAGES
   Zero dependencies, tested to R 3.1
   Customizable
   Fast
   abort <- function(..., .sep = "", .envir = parent.frame()) {
   stop(glue(..., .envir = .envir), call. = FALSE)
   }
   if (actual != expected) {
   abort("
   Expected content-type:
   * {expected}
   Actual content-type:
   * {actual}
   ")
   }
   

   View Slide

9. GLUE
   GLUE
   GLUE
   GLUE IS CUSTOMIZABLE
   IS CUSTOMIZABLE
   

   View Slide

10. .OPEN
    .OPEN AND
    AND .CLOSE
    .CLOSE CHANGE DELIMITERS
    CHANGE DELIMITERS
    j_glue <- function(..., .envir = parent.frame()) {
    glue(..., .open = "<", .close = ">", .envir = .envir)
    }
    apples <- 1; bananas <- 2
    json <- j_glue('
    {
    "fruits": {
    "apples": ,
    "bananas":
    }
    }')
    jsonlite::fromJSON(json)
    ## $fruits
    ## $fruits$apples
    ## [1] 1
    ##
    ## $fruits$bananas
    ## [1] 2
    

    View Slide

11. .TRANSFORMER
    .TRANSFORMER ARE ROBUST
    ARE ROBUST
    shell_transformer <- function(code, envir) {
    shQuote(eval(parse(text = code), envir))
    }
    glue_sh <- function(..., .envir = parent.frame()) {
    glue(..., .envir = .envir, .transformer = shell_transformer)
    }
    filename <- "test"
    writeLines("hello!", filename)
    cmd <- glue_sh("cat {filename}")
    cmd
    ## cat 'test'
    

    View Slide

12. .TRANSFORMER
    .TRANSFORMER ARE USEFUL
    ARE USEFUL
    glue_fmt("π: {pi:.7}")
    ## π: 3.1415927
    ratio <- sum(mtcars$gear == 3) / nrow(mtcars)
    glue_fmt("{ratio * 100:.2}% of mtcars have 3 gears")
    ## 46.88% of mtcars have 3 gears
    

    View Slide

13. .TRANSFORMER
    .TRANSFORMER ARE FUN
    ARE FUN
    glue_ji("If life gives you :lemon: make :tropical_drink:")
    ## If life gives you make
    glue_ji("Many :hands*: make :light: :work:")
    ## Many make
    

    View Slide

14. GLUE
    GLUE
    GLUE
    GLUE IS FAST
    IS FAST
    

    View Slide

15. TYPES OF SPEED
    TYPES OF SPEED
    Speed of writing
    Speed of recall
    Speed of execution
    

    View Slide

16. GLUE
    GLUE
    GLUE
    GLUE IS FAST ON SINGLE STRINGS
    IS FAST ON SINGLE STRINGS
    bar <- "bar"
    glue("foo{bar}")
    

    View Slide

17. GLUE
    GLUE
    GLUE
    GLUE IS FAST ON LOTS OF STRINGS
    IS FAST ON LOTS OF STRINGS
    bar <- rep("bar", 100000)
    glue("foo{bar}")
    

    View Slide

18. GLUE
    GLUE
    GLUE
    GLUE IS FAST
    IS FAST
    { } parsing in C
    performance dominated by parse() eval() and
    paste0().
    still slow? Vectorize!
    https://glue.tidyverse.org/articles/speed.html
    

    View Slide

19. GLUE
    GLUE
    GLUE
    GLUE CONTAINS MORE THAN
    CONTAINS MORE THAN GLUE()
    GLUE()
    

    View Slide

20. GLUE_DATA()
    GLUE_DATA() BINDS STRINGS TO ANY LIST /
    BINDS STRINGS TO ANY LIST /
    ENVIRONMENT
    ENVIRONMENT
    library(magrittr)
    head(mtcars) %>%
    glue_data("{rownames(.)} has {cyl} cylinders and {hp}hp")
    ## Mazda RX4 has 6 cylinders and 110hp
    ## Mazda RX4 Wag has 6 cylinders and 110hp
    ## Datsun 710 has 4 cylinders and 93hp
    ## Hornet 4 Drive has 6 cylinders and 110hp
    ## Hornet Sportabout has 8 cylinders and 175hp
    ## Valiant has 6 cylinders and 105hp
    

    View Slide

21. GLUE_COLLAPSE()
    GLUE_COLLAPSE() COLLAPSES VECTORS
    COLLAPSES VECTORS
    glue_collapse(1:10, sep = " ")
    ## 1 2 3 4 5 6 7 8 9 10
    glue_collapse(1:10, sep = " ", width = 10)
    ## 1 2 3 4...
    glue_collapse(backtick(1:10), sep = ", ", last = " and ")
    ## `1`, `2`, `3`, `4`, `5`, `6`, `7`, `8`, `9` and `10`
    

    View Slide

22. GLUE_SQL()
    GLUE_SQL() QUOTES SQL STRINGS
    QUOTES SQL STRINGS
    con <- DBI::dbConnect(RSQLite::SQLite(), ":memory:")
    tbl <- DBI::Id(schema = "xyz", table = "iris")
    glue_sql(
    "SELECT * FROM {`tbl`} WHERE species IN ({vals*})",
    vals = c("setosa", "versicolor"), .con = con)
    ## SELECT * FROM `xyz`.`iris` WHERE species IN ('setosa', 'versicol
    

    View Slide

23. STR_GLUE()
    STR_GLUE() IS IN
    IS IN tidyverse
    www.rstudio.com
    library(tidyverse)
    str_glue("
    glue version: {packageVersion('glue')}
    tidyverse version: {packageVersion('tidyverse')}
    ")
    ## glue version: 1.2.0.9000
    ## tidyverse version: 1.2.1
    

    View Slide

24. GLUE
    GLUE
    GLUE
    GLUE IS POPULAR???
    IS POPULAR???
    

    View Slide

25. 60 reverse dependencies
    Monthly downloads downloads
    downloads 419K/month
    419K/month
    dplyr, stringr, tidyr dependency, maybe already installed
    

    View Slide

26. install.packages("glue")
    glue.tidyverse.org
    adjective <- c("luminous", "stylish", "super", "striking",
    "impressive", "fantastic")
    glue("Have a {sample(adjective, 1)} day!")
    ## Have a impressive day!
      
    @jimhester_ jimhester jim.hester@rstudio.com
    

    View Slide