My 2018 in Music

print("starts")

library("jsonlite")
library("tidyverse")
library("xml2")
library("RCurl")
library("scales")
library("purrrlyr")
plays <- fromJSON(lb)

stamp <- as.numeric(as.POSIXct("2018-01-01", format="%Y-%m-%d"))
recentPlays <- plays[plays$timestamp >= stamp, ]
recentPlays <- as_tibble(recentPlays[c("artist_name", "track_name", "release_name", "timestamp")])
nrow(recentPlays)

13226

recentPlays$date <- as.Date(as.POSIXct(recentPlays$timestamp, origin="1970-01-01"))
plot <- ggplot(recentPlays, aes(format(recentPlays$date, "%Y-%U"))) +
		geom_bar(stat = "count") +
		labs(x = "Week", title="Tracks streamed per week.") +
		theme(axis.text.x=element_text(angle = -90, hjust = 0),
					panel.border = element_blank(),
					legend.key = element_blank(),
					panel.background = element_blank(),
					plot.background = element_rect(fill = "transparent",colour = NA)
		)
ggsave(file=fname, plot=plot, width=7, height=4, dpi=300, bg="transparent")
fname

top_artists <-recentPlays %>%
		count(artist_name, sort=T)
top_artists %>% head()

recentPlays %>%
		count(artist_name, track_name, sort=T) %>%
		head(5)

top_songs <- recentPlays %>%
		group_by(artist_name, track_name) %>%
		count(sort=T) %>%
		ungroup() %>%
		distinct(artist_name, .keep_all=T) %>%
		head(5)

topAlbums <- recentPlays %>%
		group_by(artist_name, release_name) %>%
		count(sort=T)
topAlbums %>% head()

getAlbum <- function(row) {
		mburl <- sprintf(
				'https://beta.musicbrainz.org/ws/2/release/?query=artist:%s+release:%s+AND+status:official+AND+format:"Digital%%20Media"&inc=release-group&limit=1',
				curlEscape(row$artist_name),
				curlEscape(row$release_name)
		)
		print(mburl)
		Sys.sleep(0.25)
		groupData <- read_xml(mburl)
		xml_ns_strip(groupData)
		release <- xml_find_first(groupData, '//release[@ns2:score=100]')
		xml_ns_strip(release)
		# If it is empty
		if (class(release) == "xml_missing") {
				release <- xml_new_document() %>% xml_add_child("")
		}
		# Go with the earliest release date given.
		date <- xml_text(xml_find_first(release, "//date"))
		artistId <- xml_text(xml_find_first(release, "//artist/@id"))
		df <- data.frame(date, artistId, stringsAsFactors=FALSE)
		colnames(df) <- c("date", "artistId")
		return(df)
}

recentAlbums <- topAlbums %>% filter(n > 25) %>% by_row(..f=getAlbum, .to=".out") %>% unnest()

recentAlbums %>%
		filter(str_detect(date, "2018")) %>%
		select(artist_name, release_name, n, date) %>%
		filter(n > 75)

 getLengths <- function(row) {
song_stripped <- trimws(sub("\\(.*\\)", "", row$track_name))
mburl <- sprintf(
					'https://beta.musicbrainz.org/ws/2/recording/?query=artist:%s+AND+recording:%s&limit=2',
					curlEscape(row$artist_name),
					curlEscape(song_stripped)
)
# To comply with the rate limit.
Sys.sleep(0.5)
albumData <- read_xml(mburl)
xml_ns_strip(albumData)
length <- xml_integer(xml_find_first(albumData, "//length"))
return(length)
	}

set.seed(425368203)
len_sample <- recentPlays %>% sample_n(250) %>% by_row(..f=getLengths, .to="length") %>% unnest()

mean_len <- len_sample %>% dplyr::summarize(Mean=mean(length, na.rm=T))

mins <- nrow(recentPlays) * mean(as.numeric(mean_len)) / 60000

top_artist_ids <- recentAlbums %>%
		group_by(artistId) %>%
		filter(!is.na(artistId)) %>%
		summarize(Sum=sum(n)) %>%
		arrange(desc(Sum))
top_artist_ids %>%
		summarize(sum(Sum))

fetchGenres <- function(row) {
		mburl <- sprintf(
				"https://beta.musicbrainz.org/ws/2/artist/%s?inc=genres",
				row$artistId
		)
		print(mburl)
		Sys.sleep(0.25)
		groupData <- read_xml(mburl)
		xml_ns_strip(groupData)
		genres <- xml_text(xml_find_all(groupData, "//genre/name"))
		return(genres)
}

top_artist_ids <- top_artist_ids %>%
		by_row(..f=fetchGenres, .to="Genres") %>%
		unnest()

topGenres <- top_artist_ids %>%
		group_by(Genres) %>%
		summarize(Sum=sum(Sum)) %>%
		arrange(desc(Sum))
topGenres %>% head()

library("ggpubr")
library("png")
library("raster")

myTheme <- ttheme(colnames.style = colnames_style(color = "white",
																									fill = "#8cc257",
																									linewidth=0),
									tbody.style = tbody_style(color = "white", linewidth=0,
																						fill = "#8cc257"))

bgTheme <- theme(
		plot.background =
				element_rect(fill = "#8cc257", color="#8cc257"),
		panel.border = element_blank(),
		)

top_artist_names <- top_artists$artist_name %>%
		head()
artistTable <- ggtexttable(top_artist_names, rows = NULL,
													 theme = myTheme, cols=c("Top Artists")) + bgTheme
trackTable <- ggtexttable(top_songs$track_name, rows = NULL,
													theme = myTheme, cols=c("Top Songs")) + bgTheme
minutes <- as_ggplot(text_grob(
		paste("Minutes Listened",
					toString(round(mins)),
					"",
					"Top Genre",
					toString(topGenres[1,1]),
					sep="\n"),
		color="white")) + bgTheme
img <- readPNG("images/albums.png")
im_A <- ggplot() +
		background_image(img[1:250, 1:250, 1:3]) +
		theme(
				plot.margin = margin(t=.5, l=.5, r=.5, b=.5, unit = "cm"),
		) + bgTheme
p <- ggarrange(im_A, artistTable, minutes, trackTable, ncol=2, nrow=2)
ggsave(file=fname, plot=p, width=4.5, height=4.5, dpi=300)
fname