Some text sorting

SortText.r

library(stringr)

grav_roys <- read.csv("G:/New Folder (18)/rays_haug.csv", encoding = "UTF-8", sep = ",")
str(grav_roys)
myvars <- c("OBJECTID", "objid", "informasjo")
newdata <- grav_roys[myvars]

subset <- head(grav_roys)
subset
str(subset)

if (grepl("\\d", "nteionrasop") == TRUE) {
  print("True")
} else {
  print("Not True")
}

1:nrow(subset)
# Høyde
for (row in 1:nrow(newdata)) {
  orgText <- newdata$informasjo[row]
  newdata$hoyde[row] <- hoydeFunc(orgText)
}

# Mål
for (row in 1:nrow(newdata)) {
  orgText <- newdata$informasjo[row]
  newdata$maal[row] <- maalFunc(orgText)
}

# Diameter
for (row in 1:nrow(newdata)) {
  orgText <- newdata$informasjo[row]
  newdata$diameter[row] <- diamFunc(orgText)
}

# Type
for (row in 1:nrow(newdata)) {
  orgText <- newdata$informasjo[row]
  newdata$type[row] <- typeFunc(orgText)
}



hoydeFunc <- function(orgText) {
  pat <- '(?i)(^.*høyde.*?|h. *?)(\\d+(?:[\\.\\,]\\d{1,2})?).*'
  #hoyde <- gsub(pat, '\\2', orgText)
  #hoyde <- str_match(orgText, pat)[,3]
  #hoyde <- sapply(tmp, "[", c(3))
  #hoyde <- as.numeric(gsub(",", ".", gsub("\\.", "", hoyde)))
  tmp <- gsub("[\r\n]", " ", orgText) # Remove \n new lines
  hoyde <- as.numeric(gsub(",", ".", gsub("\\.", "", str_match(tmp, pat)[,3])))
  if (grepl('\\d', hoyde) == TRUE) {
    return(hoyde)
  } else if (is.na(hoyde) == TRUE) {
      return(NA)
    } else {
      return(NA)
    }
}

maalFunc <- function(orgText) {
  pat <- '(?i)^.*mål.*?(\\d+(?:[\\.\\,\\xx]\\d{1,2})?).*'
  maal <- gsub(pat, '\\1', orgText)
  if (orgText == maal) {
    return(NA)
  } else if (grepl('\\d', maal) == TRUE) {
    return(maal)
  } else {
    return(NA)
  }
}

diamFunc <- function(orgText) {
  pat <- '(?i)^.*diam.*?(\\d+(?:[\\.\\,\\xx]\\d{1,2})?).*'
  diameter <- gsub(pat, '\\1', orgText)
  if (orgText == diameter) {
    return(NA)
  } else if (grepl('\\d', diameter) == TRUE) {
    return(diameter)
  } else {
    return(NA)
  }
}

(?:s|rund.*)

typeFunc <- function(orgText) {
  pat <- '(?i)\\b(rundhaug|rundrøys|langhaug|rund haug)\\b'
  type <- str_extract(orgText, pat)
  if (is.na(type) == TRUE) {
    return(NA)
  } else if (grepl('\\d', type) == FALSE) {
    return(type)
  } else {
    return(NA)
  }
}

str(newdata)
newdata$hoyde <- as.numeric(gsub(",", ".", gsub("\\.", "", newdata$hoyde)))
newdata$diameter <- as.numeric(gsub(",", ".", gsub("\\.", "", newdata$diameter)))
#format(newdata$hoyde, decimal.mark = '.')
write.csv(newdata, file = 'G:/kulturminner/New Folder (18)/sorted.csv')

newdata2 <- newdata[ which(newdata$hoyde > 0.2), ]
newdata2$diameter <- as.numeric(gsub(",", ".", gsub("\\.", "", newdata2$diameter)))
write.csv(newdata2, file = 'G:/kulturminner/New Folder (18)/sorted2.csv')