In mpadge/lwl-destinations:

library (magrittr)
dat <- readxl::read_xlsx ("./data-raw/Final_Antonia_Koloniale Objekte_Gesamt ohne Werl_04062021.xlsx") %>%
    data.frame ()
cols <- c ("...31", "...32")
for (i in cols) {
    dat [[i]] [!is.na (dat [[i]])] <- paste0 (" ", dat [[i]] [!is.na (dat [[i]])])
    dat [[i]] [is.na (dat [[i]])] <- ""
}
dat$Suchwort <- gsub ("^\\s+", "", paste0 (dat$Suchwort, dat$`...31`, dat$`...32`))
dat <- dat [, -match (cols, names (dat))]

latcols <- grep ("^Lat", names (dat))
names (dat) [latcols] <- c ("Lat", "Lat_origin")
loncols <- grep ("^Long", names (dat))
names (dat) [loncols] <- c ("Long", "Long_origin")

cols <- c ("Museumsname",
           "Standort",
           "Lat",
           "Long",
           "Objektbezeichnung",
           "Hersteller.Künstler",
           "Land.Ort",
           "Lat_origin",
           "Long_origin")

dat <- dat [, match (cols, names (dat))]
dat <- dat [which (!is.na (dat$Lat_origin) & !is.na (dat$Long_origin)), ]
names (dat) [names (dat) == "Objektbezeichnung"] <- "object"
sort (unique (dat$object))

obj <- dat$object

obj [grep ("teller|gefäß|schal|trink|becher|topf|tisch", obj, ignore.case = TRUE)] <- "domestic"
obj [grep ("wasser|kanne(\\s|$)|kännchen|flasche|löffel|schälchen", obj, ignore.case = TRUE)] <- "domestic"
obj [grep ("schüssel|besteck|(^|\\s)(ess|eß)|deckel|kamm|kerzen", obj, ignore.case = TRUE)] <- "domestic"
obj [grep ("spiegel", obj, ignore.case = TRUE)] <- "domestic"
obj [grep ("zigar|rauch|tabak|pfeife", obj, ignore.case = TRUE)] <- "smoking"
obj [grep ("teppich|tapete|lampe|laterne|korb$|regal|sockel", obj, ignore.case = TRUE)] <- "furniture"
obj [grep ("vase|bild|figur|figür|ton|tier|schmuck|pokal", obj, ignore.case = TRUE)] <- "decorative"
obj [grep ("^plastik|plastik($|\\s|\\))|model|maske|dose($|\\s)", obj, ignore.case = TRUE)] <- "decorative"
obj [grep ("schnitz|behälter|^holz|bemalte|madonna|gerahmte", obj, ignore.case = TRUE)] <- "decorative"
obj [grep ("gefächerte|verzier(u|t)|aquarell|buddha|beschlag", obj, ignore.case = TRUE)] <- "decorative"
obj [grep ("relief|sammel|beklebt|büste|gebets|beschriftet", obj, ignore.case = TRUE)] <- "decorative"
obj [grep ("behang", obj, ignore.case = TRUE)] <- "decorative"
obj [grep ("tee|tea$", obj, ignore.case = TRUE)] <- "tea"
obj [grep ("mokka|kaffee", obj, ignore.case = TRUE)] <- "tea"
obj [grep ("axt|speer|dolch|rüstung|schwert|pfeil|lanze$|(\\s|^)lanze", obj, ignore.case = TRUE)] <- "weaponry"
obj [grep ("kampf|weapon|panzer|säbel|keule|brust|bogen|helm|gewehr", obj, ignore.case = TRUE)] <- "weaponry"
obj [grep ("messer", obj, ignore.case = TRUE)] <- "domestic" # after weaponry, so kampf goes there first
obj [grep ("trommel|flöte|music|musik|gong|rassel|horn|instrument", obj, ignore.case = TRUE)] <- "music"
obj [grep ("buch|roman$|berich|mappe|album|koran|literatur", obj, ignore.case = TRUE)] <- "printed"
obj [grep ("karte|zeitung|manifest|(^|\\s)heft|umschlag|foto", obj, ignore.case = TRUE)] <- "printed"
obj [grep ("printed|diplom|biograph|druck|graphik", obj, ignore.case = TRUE)] <- "printed"
obj [grep ("hut$|poncho|sandalen|schuh|gürtel|tuch", obj, ignore.case = TRUE)] <- "clothing"
obj [grep ("geld|münze|dollar", obj, ignore.case = TRUE)] <- "money"
obj [grep ("amulet|kette(\\s|$)|reif(\\s|$)", obj, ignore.case = TRUE)] <- "jewellery"
obj [grep ("mokka|armband", obj, ignore.case = TRUE)] <- "jewellery"
obj [grep ("spiel", obj, ignore.case = TRUE)] <- "playtoys"

sort (unique (obj)); length (unique (obj))
dat$object <- obj

table (dat$Land.Ort)
dat$Land.Ort [dat$Land.Ort == "Chna"] <- "China"
dat$Land.Ort [dat$Land.Ort == "Naher Ostem"] <- "Naher Osten"
dat$Land.Ort [dat$Land.Ort == "Salomonen"] <- "Salomoninseln"
dat$Land.Ort [dat$Land.Ort == "Somali"] <- "Somalia"
dat$Land.Ort [dat$Land.Ort == "Westafika"] <- "Westafrika"
dat$Land.Ort [dat$Land.Ort == "Westafria"] <- "Westafrika"

write.table (dat, "data/LWL.tsv", sep = "\t", row.names = FALSE)

categories <- c ("domestic", "smoking", "furniture", "decorative", "tea", "weaponry",
                 "music", "printed", "clothing", "money", "jewellery", "playtoys")
dat$obj <- obj
dat <- dat [which (obj %in% categories), ]

817 / 960 objects in these categories

mpadge/lwl-destinations documentation built on Jan. 8, 2022, 4:27 a.m.