link chords to lyrics

chords + lyrics

library(chorrrds)
library(tidyverse)
# Chords ------------------------------------------------------------------

notes     = c('A','B','C','D','E','F','G')
flats     = 'b'
sharps    = '#'
minor     = 'm'
all_notes = c(notes, 
              paste0(notes, flats),
              paste0(notes, sharps),
              paste0(notes, minor),
              paste0(notes, flats, sharps),
              paste0(notes, minor, sharps),
              paste0(notes, flats, minor),
              paste0(notes, flats, sharps, minor))


# Example -----------------------------------------------------------------


url <- get_songs("The Weeknd")
url_something <- url$url[2]

x <- xml2::read_html(paste0("https://www.cifraclub.com.br",
                            url_something))

chords_lyrics <- rvest::html_nodes(x, "pre") %>% 
  rvest::html_text()

chords_lyrics <- 
  chords_lyrics %>% 
  str_remove_all(pattern = "[0-9]|[-][0-9][-]|\\||[0-9][br]|~")

chords <- tibble(sapply(chords_lyrics, function(x) strsplit(x, "\n")[[1]], USE.NAMES=FALSE))

chords.dat <- as.data.frame(cbind(rep(NA, nrow(chords)/2),
                                  rep(NA, nrow(chords)/2)))

for (i in 1:nrow(chords)) {
  
  ifelse(any(unlist(str_split(as.character(chords[i, 1]), " ")) %in% all_notes) == TRUE,
         chords.dat[i, 1] <- as.character(chords[i, 1]),
         chords.dat[i, 2] <- as.character(chords[i, 1])
  )
  
}

chords.dat <- chords.dat[ grep("Intro:", chords.dat$V1, invert = TRUE), ]
chords.dat <- chords.dat[ grep("--", chords.dat$V2, invert = TRUE), ]
chords.dat[,1] <- sub("( *)(\\w+)", "\\2\\1", chords.dat[,1])                               # put all the first chords at the beginning of the verse 
# (keeping he same spaces between more than two chords)
chords.dat <- chords.dat %>%
  mutate_at(c("V2"), funs(lead), n = 1 )
chords.dat <- chords.dat[complete.cases(chords.dat), ]
rownames(chords.dat) <- NULL
rownames(chords.dat) <- as.numeric(rownames(chords.dat))

# Connect chords with lyrics ----------------------------------------------

# The idea is to have a data frame with two columns, one for chords and the other for the connected lyrics,
# such that each row contains one chord (col1) and the lyrics linked to that (col2)

chords.net <- as.data.frame(
  bind_cols(
    chord = rep(NA, sum(na.omit(str_count(chords.dat[, 1], "\\S+")))*2),                    # nrow = number of chords in the song
    lyric = rep(NA, sum(na.omit(str_count(chords.dat[, 1], "\\S+")))*2)
  ))


# chords

for(i in 1:nrow(chords.dat)) {
  
  if(str_count(chords.dat[i, 1], "\\S+") > 1){                                                                    # if there is more than 1 chord
    
    for (
      j in 1:str_count(substring(chords.dat[i, 1], 1, nchar(chords.dat[i, 2])), "\\S+")) {                   # for each chord in verse i
      
      chords.net[i+j, 1] <- substring(chords.dat[i, 1],                                                         # find chords in verse i
                                      sort(unique(na.omit(str_locate(chords.dat[i, 1], all_notes)[,1])))[j],    # from this position
                                      sort(unique(na.omit(str_locate(chords.dat[i, 1], all_notes)[,1])))[j]+1)  # until this position
      
    }
    } else { chords.net[i, 1] <- chords.dat[i, 1] }                                                                       # otherwise, put that chord in the row

  i
  }

# lyrics
for (i in 1:nrow(chords.dat)) {
  
  if(str_count(chords.dat[i, 1], "\\S+") > 1){                                                                    # if there is more than 1 chord
         
         for (j in 1:str_count(substring(chords.dat[i, 1], 1, nchar(chords.dat[i, 2])), "\\S+")) {                    # for each chord in verse i
           
           chords.net[i+j, 2] <- substring(chords.dat[i, 2],                                                          # find words in verse i linked to chord j
                                           sort(unique(na.omit(str_locate(chords.dat[i, 1], all_notes)[,1])))[j],     # from this position (position of chord j)
                                           sort(unique(na.omit(str_locate(chords.dat[i, 1], all_notes)[,1])))[j+1]-1) # until this position (position of next chord, j+1)
           
         }
  } else { chords.net[i, 2] <- chords.dat[i, 2]  }                                                                       # otherwise, all the words in the verse are linked to the single chord
  
}

Cumbersome solution for i+j:

chords.net[match(as.numeric(do.call(paste0, expand.grid(i,j))), sort(as.numeric(do.call(paste0, expand.grid(1:sum(na.omit(str_count(chords.dat[, 1], "\\S+"))),1:max(str_count(substring(chords.dat[, 1], 1, nchar(chords.dat[, 2])), "\\S+"))))))), 1]