- gguke is under construction! 🚧 Feedback welcome!
- use parsing to expand dataframe to have row per finger placement.
{gguke} will provide tools for visualizing ukelele chords with lyrics. This is for people that aren’t so good at ukelele; they may not really know the chords yet. It’s a chord-first approach to song charts, which might have only chord names or small finger position.
And because we make it with ggplot2, we can do a lot with customization. For example, fingers get assigned a color using the viridis palette.
This is currently a {knitrExtra} project; we’re writing in a kind-of stream of consciousness way, bundling up useful code into functions as we go in our README.rmd narrative, and then sending written-up functions to our R folder with knitr/rmd magic. Our commitment to keeping things as they are is low at this point, but it is a real package. Lifecycle is early experimental. See also {litr}.
Beyond the descriptions of our work, we interject comments on our hesitations 🤔 and areas that need some work 🚧, for your consideration marked with emoji.
First, something I’m very excited about, a new-chord ingestion… - it’s just text. Here’s a C major fingering, where first line is whether string is played or not. There are four strings across, the number indicates where on the fret board the specific finger should be placed:
CM <-
"pppp
----
----
---3"
GM <-
"pppp
----
-1-2
--3-"
chord_library <- data.frame(chord_name =
c("CM", "GM"),
fingering_str =
c(CM, GM))
chord_library
#> chord_name fingering_str
#> 1 CM pppp\n----\n----\n---3
#> 2 GM pppp\n----\n-1-2\n--3-# easy to write
Dm <-
"pppp
--1-
23--
----"
# re-writen to save space
GM <- "pppp\n----\n-1-2\n--3-"
E7M <- "pppp\n1---\n-2-3\n----"
AM <- "pppp\n-1--\n2---\n----"
DM <- "pppp\n----\n123-\n----"
Dbm <- "pppp\n12--\n----\n----"
`F#m` <- "pppp\n-1--\n2-3-\n----"
D7M <- "pppp\n----\n1111\n---2"
A7M <- "pppp\n-1--\n----\n----"
Am <- "pppp\n1---\n----\n----"
Am7 <- "pppp\n----\n----\n----"
Bm <- "pppp\n----\n-111\n----\n3---"
Em <- "pppp\n----\n---1\n----\n-3--"chord_library <- tibble(chord_name =
c("CM", "GM", "Dm", "E7M", "AM","DM", "Dbm", "F#m", "D7M", "A7M", "Am", "Am7", "Bm", "Em"),
fingering_str =
c(CM, GM, Dm, E7M, AM, DM, Dbm, `F#m`, D7M, A7M, Am, Am7, Bm, Em))
chord_library
#> # A tibble: 14 × 2
#> chord_name fingering_str
#> <chr> <chr>
#> 1 CM "pppp\n----\n----\n---3"
#> 2 GM "pppp\n----\n-1-2\n--3-"
#> 3 Dm "pppp\n--1-\n23--\n----"
#> 4 E7M "pppp\n1---\n-2-3\n----"
#> 5 AM "pppp\n-1--\n2---\n----"
#> 6 DM "pppp\n----\n123-\n----"
#> 7 Dbm "pppp\n12--\n----\n----"
#> 8 F#m "pppp\n-1--\n2-3-\n----"
#> 9 D7M "pppp\n----\n1111\n---2"
#> 10 A7M "pppp\n-1--\n----\n----"
#> 11 Am "pppp\n1---\n----\n----"
#> 12 Am7 "pppp\n----\n----\n----"
#> 13 Bm "pppp\n----\n-111\n----\n3---"
#> 14 Em "pppp\n----\n---1\n----\n-3--"Then translate that to data frame that records string, fret, and finger that will play each note. I’m just doing the first thing that came to mind - some string manipulation, but might move to read.delim…
CM |> stringr::str_split("") %>% .[[1]] %>% .[1:4]
#> [1] "p" "p" "p" "p"chart <- CM |> stringr::str_split("") %>% .[[1]] %>% .[5:length(.)]
num_frets <- length(chart)/5
string <- rep(1:4, num_frets)
fret <- sort(rep(1:num_frets, 4))
data.frame(finger = chart) %>%
dplyr::filter(finger != "\n") %>%
dplyr::mutate(fret = fret) %>%
dplyr::mutate(string = string) %>%
dplyr::mutate(finger = ifelse(.data$finger == "-", NA, finger) %>% as.numeric()) %>%
dplyr::filter(!is.na(finger))
#> finger fret string
#> 1 3 3 4🤔 maybe read.delim(delim = ““), would work better and feel more grokable that this string split business! :-)
parse_chord <- function(chord = CM){
play_TF <- chord |> stringr::str_split("") %>% .[[1]] %>% .[1:4]
chart <- chord |> stringr::str_split("") %>% .[[1]] %>% .[5:length(.)]
num_frets <- length(chart)/5
string <- rep(1:4, num_frets)
fret <- sort(rep(1:num_frets, 4))
data.frame(finger = chart) %>%
dplyr::filter(finger != "\n") %>%
dplyr::mutate(fret = fret) %>%
dplyr::mutate(string = string) %>%
dplyr::mutate(finger = ifelse(.data$finger == "-", NA, finger) %>% as.numeric()) %>%
dplyr::filter(!is.na(finger)) ->
fingering_df
list(play_tf = play_TF,
fingering_df = fingering_df)
}# play or not for each string
parse_chord()[[1]]
#> [1] "p" "p" "p" "p"# dataframe with finger placement info
parse_chord()[[2]]
#> finger fret string
#> 1 3 3 4parse_chord_fingering <- function(chord){
parse_chord(chord)[[2]]
}
chord_library_parse_chords <- function(chord_library){
parsed_chords <- list()
for (i in 1:nrow(chord_library)){
parsed_chords[[i]] <- parse_chord_fingering(chord_library$fingering_str[i])
}
chord_library$phrase_chord_id <- 1:nrow(chord_library)
chord_library$parsed_chords <- parsed_chords
chord_library %>%
unnest(parsed_chords)
}chord_library_parsed <- chord_library %>% chord_library_parse_chords()
chord_library_parsed
#> # A tibble: 33 × 6
#> chord_name fingering_str phrase_chord_id finger fret string
#> <chr> <chr> <int> <dbl> <int> <int>
#> 1 CM "pppp\n----\n----\n---3" 1 3 3 4
#> 2 GM "pppp\n----\n-1-2\n--3-" 2 1 2 2
#> 3 GM "pppp\n----\n-1-2\n--3-" 2 2 2 4
#> 4 GM "pppp\n----\n-1-2\n--3-" 2 3 3 3
#> 5 Dm "pppp\n--1-\n23--\n----" 3 1 1 3
#> 6 Dm "pppp\n--1-\n23--\n----" 3 2 2 1
#> 7 Dm "pppp\n--1-\n23--\n----" 3 3 2 2
#> 8 E7M "pppp\n1---\n-2-3\n----" 4 1 1 1
#> 9 E7M "pppp\n1---\n-2-3\n----" 4 2 2 2
#> 10 E7M "pppp\n1---\n-2-3\n----" 4 3 2 4
#> # ℹ 23 more rowsusethis::use_pipe()
#> ✔ Setting active project to '/Users/evangelinereynolds/Google
#> Drive/r_packages/gguke'knitrExtra:::chunk_to_r("parse_chord")
#> It seems you are currently knitting a Rmd/Qmd file. The parsing of the file will be done in a new R session.data.frame(lyric = c("hello", "goodbye"),
chord_name = c("CM", "GM")) %>%
inner_join(chord_library %>% chord_library_parse_chords()) %>%
ggplot() +
facet_wrap(~paste(phrase_chord_id, lyric, chord_name)) +
geom_point(size = 15,
aes(x = string, y = fret + .5),
color = "white"
) +
geom_point(size = 15, pch = 21, alpha = .6,
aes(x = string, y = fret + .5,
fill = finger),
) +
geom_text(size = 10,
aes(x = string, y = fret + .5, label = finger)
) +
scale_y_reverse() +
coord_equal(xlim = c(1,4), ylim = c(4, 1))
#> Joining with `by = join_by(chord_name)`
Make it more ggplot2 grammatical
chord_library %>%
chord_library_parse_chords() ->
chord_library_parsed
compute_group_uke_fingering <- function(data, scales, chord_library = chord_library_parsed){
data %>%
mutate(row = row_number()) %>%
left_join(chord_library_parsed %>%
rename(chord = chord_name)) %>%
mutate(x = string)
}StatUkefingers <- ggplot2::ggproto(
`_class` = "StatUkefingers",
`_inherit` = ggplot2::Stat,
required_aes = c("chord"),
compute_panel = compute_group_uke_fingering,
default_aes = ggplot2::aes(label = after_stat(finger),
# color = after_stat(finger),
fill = after_stat(finger),
y = after_stat(fret+.5)))library(statexpress)
stat_chord <- function(...){
qlayer(stat = StatUkefingers,
geom = qproto_update(GeomPoint,
aes(fill = "white",
size = 15,
shape = 21)), ...)
}
geom_chord <- stat_chord
geom_chord_text <- function(...){
qlayer(stat = StatUkefingers,
geom = qproto_update(GeomText,
aes(size = 10)),...)
}tribble(~lyric, ~chord_name,
"Mary had a little lamb,", "CM",
"little lamb,", "GM",
"little lamb.", "CM",
"Mary had a little lamb...", "CM") %>%
ggplot() +
scale_color_viridis_c() +
aes(chord = chord_name) +
geom_chord(fill = "white") +
geom_chord(alpha = .6, color = "black") +
geom_chord_text(color = "black", size = 10) +
facet_wrap(~fct_inorder(lyric)) +
scale_y_reverse() +
coord_equal(xlim = c(1,4), ylim = c(4, 1))
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
library(ggplot2)
ggplot() +
annotate(geom = "segment", x = 1:4, y = 0,
xend = 1:4, yend = 5, linewidth = 3,
lineend = "round") +
annotate(geom = "segment", y = 0:4,
yend = 0:4, x = 1, xend = 4, linewidth = 3) +
scale_y_reverse() +
scale_x_continuous(expand = expansion(.2)) +
# theme_void() +
coord_equal() +
scale_fill_viridis_c(limits = c(1,4), guide = F)
#> Warning: The `guide` argument in `scale_*()` cannot be `FALSE`. This was deprecated in
#> ggplot2 3.3.4.
#> ℹ Please use "none" instead.
#> This warning is displayed once every 8 hours.
#> Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
#> generated.
stamp_fretboard <- function(){
list(
annotate(geom = "segment", x = 1:4, y = 0,
xend = 1:4, yend = 5, linewidth = 3,
lineend = "round"),
annotate(geom = "segment", y = 0:4,
yend = 0:4, x = 1, xend = 4, linewidth = 3,
lineend = "butt"),
scale_y_reverse(),
scale_x_continuous(expand = expansion(.2)),
theme_void(),
coord_equal(),
scale_fill_viridis_c(limits = c(1,4), guide = F)
)
}
gguke <- function(data){
ggplot(data = data) + stamp_fretboard()
}tribble(~lyric, ~chord_name,
"Mary had a little lamb,", "CM",
"little lamb,", "GM",
"little lamb.", "CM",
"Mary had a little lamb...", "CM") %>%
gguke() +
scale_color_viridis_c() +
aes(chord = chord_name) +
geom_chord(fill = "white") +
geom_chord(alpha = .6, color = "black") +
geom_chord_text(color = "black", size = 10) +
facet_wrap(~fct_inorder(lyric))
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
#> Joining with `by = join_by(chord)`
🤔 maybe gguke() would be better 🤔 maybe a coord_uke could be a long-term goal.
knitrExtra:::chunk_to_r("uke_fretboard")
#> It seems you are currently knitting a Rmd/Qmd file. The parsing of the file will be done in a new R session.library(tidyverse)
library(gguke)
lyric_chord_df <- tibble::tribble(~lyric, ~chord_name,
"Come stop your cryin', it'll be all right", "CM",
"Just take my hand, hold it tight", "CM",
"I will protect you, from all around you", "FM",
"I will be here, don't you ","Dm",
"...cry" , "GM" ,
" ", "E7M",
"'Cause you'll be in my ... heart", "AM",
"... heart", "DM",
"Yes, you'll be in my heart", "E7M",
"... heart", "Dbm",
"From this day on", "AM",
"Now and forever... more", "DM",
"... more", "GM",
"...", "E7M",
"You'll be in my ...", "AM",
"... heart", "DM",
"No matter what they ...", "E7M",
"say", "Dbm",
"You'll be here in my... ", "AM",
" heart", "DM",
"Always", "GM")song <- readLines("Untitled.txt")
#> Warning in readLines("Untitled.txt"): incomplete final line found on
#> 'Untitled.txt'song_line <- sort(rep(1:(length(song)/2),2))
characters <- song %>% stringr::str_split("")
element <- rep(c("chord", "lyric"), length(song)/2)
rep_spaces <- function(x){
rep("", x)
}
combine <- function(x, y){
c(x, y)
}
tibble(song_line, characters, element) %>%
mutate(num_char = map_dbl(characters, length)) %>%
group_by(song_line) %>%
mutate(max_char = max(num_char)) %>%
ungroup() %>%
mutate(diff_char = max_char - num_char) %>%
mutate(spaces_to_add = map(diff_char, rep_spaces)) %>%
mutate(chars_evened_up = map2(characters, spaces_to_add, combine) ) %>%
pivot_wider(id_cols = song_line,
names_from = element,
values_from = chars_evened_up) %>%
unnest() %>%
mutate(ind_chord_space = chord == " ") %>%
mutate(ind_init_chord = !ind_chord_space & lag(ind_chord_space)) %>%
mutate(ind_init_chord = replace_na(ind_init_chord, FALSE) ) %>%
mutate(id_chord_phrase = cumsum(ind_init_chord)) %>%
group_by(id_chord_phrase) %>%
summarize(lyric = paste(lyric, collapse = ""),
chord_name = paste(chord, collapse = "") %>% str_trim()) ->
snow_from_txt
#> Warning: `cols` is now required when using `unnest()`.
#> ℹ Please use `cols = c(chord, lyric)`.txt_chord_lyrics_to_df <- function(path){
song <- readLines(path)
song_line <- sort(rep(1:(length(song)/2),2))
characters <- song %>% stringr::str_split("")
element <- rep(c("chord", "lyric"), length(song)/2)
rep_spaces <- function(x){
rep("", x)
}
combine <- function(x, y){
c(x, y)
}
tibble(song_line, characters, element) %>%
mutate(num_char = map_dbl(characters, length)) %>%
group_by(song_line) %>%
mutate(max_char = max(num_char)) %>%
ungroup() %>%
mutate(diff_char = max_char - num_char) %>%
mutate(spaces_to_add = map(diff_char, rep_spaces)) %>%
mutate(chars_evened_up = map2(characters, spaces_to_add, combine) ) %>%
pivot_wider(id_cols = song_line,
names_from = element,
values_from = chars_evened_up) %>%
unnest() %>%
mutate(ind_chord_space = chord == " ") %>%
mutate(ind_init_chord = !ind_chord_space & lag(ind_chord_space)) %>%
mutate(ind_init_chord = replace_na(ind_init_chord, FALSE) ) %>%
mutate(id_chord_phrase = cumsum(ind_init_chord)) %>%
group_by(id_chord_phrase) %>%
summarize(lyric = paste(lyric, collapse = ""),
chord_name = paste(chord, collapse = "") %>% str_trim())
}- There may have been a problem with phrases that don’t start with a new chord. I worked on it in the text file.
gershwin_ly_ch_df <- txt_chord_lyrics_to_df("gershwin.txt")
#> Warning in readLines(path): incomplete final line found on 'gershwin.txt'
#> Warning: `cols` is now required when using `unnest()`.
#> ℹ Please use `cols = c(chord, lyric)`.Test…
gershwin_ly_ch_df[1:4,] %>%
rename(chord = chord_name) %>%
compute_group_uke_fingering(chord_library = chord_library_parsed)
#> Joining with `by = join_by(chord)`
#> # A tibble: 6 × 10
#> id_chord_phrase lyric chord row fingering_str phrase_chord_id finger fret
#> <int> <chr> <chr> <int> <chr> <int> <dbl> <int>
#> 1 0 "Its v… D7 1 <NA> NA NA NA
#> 2 1 "Our l… FM 2 <NA> NA NA NA
#> 3 2 " here… Dm 3 "pppp\n--1-\… 3 1 1
#> 4 2 " here… Dm 3 "pppp\n--1-\… 3 2 2
#> 5 2 " here… Dm 3 "pppp\n--1-\… 3 3 2
#> 6 3 " stay" Am7 4 <NA> NA NA NA
#> # ℹ 2 more variables: string <int>, x <int>knitr::knit_exit()