Errors attempting to applying function to data frame rows in R

Question

I wrote a very simple function (which works well) that returns the timezone given a set of coordinates:

library(XML)
findTZ <- function(lon, lat, date=Sys.Date())
{ apiurl <- sprintf("https://maps.googleapis.com/maps/api/timezone/%s?location=%s,%s&timestamp=%d&sensor=%s", 
                    "xml", lat, lon, as.numeric(as.POSIXct(date)), "false")
  TZ <- xmlParse(readLines(apiurl))[["string(//time_zone_id)"]]
  return(TZ)
}

findTZ(-112.86, 53.61) # example

However, when I try to run the function on a list of coordinates in a dataframe I get an error: Error in file(con, "r") : invalid 'description' argument

Any hints at what I'm getting wrong here? It seems like it should be very simple.

Here's the very basic data I'm testing on:

DF <- data.frame(
  longitude = c(-122, -112, -102),
  latitude = c(54, 53, 52)
)

DF$timezone = findTZ(lon=DF$longitude, lat=DF$latitude)

Thank you for any pointers!

EDIT / ADDITION

After implementing the answer from @Floo0 I tried implementing the same solution with another function for calculating sunrise/set times using the same location data (and that I want to return in local time, hence the timezone function).

Here's the sunrise function:

    library(maptools)
SSun <- function(lon, lat, date, deg=0, dir, tzone)
{ # deg = solar depth: rise/set=0, civil=6, nautical=12, astronomical=18
  # dir = direction: sunrise="dawn", sunset="dusk"
  # tzone  = time zone of output, NOT of location
  siteX <- SpatialPoints(matrix(c(lon, lat), nrow=1), proj4string=CRS("+proj=longlat +datum=WGS84"))
  dateX <- as.POSIXct(date, tz=tzone)
  duskX <- crepuscule(siteX, dateX, solarDep=deg, direction=dir, POSIXct.out=TRUE)
  duskX <- duskX$time # keep only date and time, discard day_frac
  return(duskX)
}

SSun(-112.86, 53.61, "2016-09-25", deg=0, dir="dawn", tzone="America/Edmonton") # example

And the updated timezone function:

library(tidyverse); library(xml2)
findTZ <- function(lon, lat, date=Sys.Date()){ 
  apiurl <- sprintf("https://maps.googleapis.com/maps/api/timezone/%s?location=%s,%s&timestamp=%d&sensor=%s", 
                    "xml", lat, lon, as.numeric(as.POSIXct(date)), "false")
  read_xml(apiurl) %>% xml_find_first(".//time_zone_id") %>% xml_text
}

findTZ(-112.86, 53.61) # example

And the code I used to call both functions:

DF %>% mutate(date = as.POSIXct(date),
              TZ = map2_chr(longitude, latitude, findTZ),
              sunrise = SSun(longitude, latitude, date, deg=0, dir="dawn", tzone=TZ))

I feel like I must be misunderstanding how this works. Any insights?

Answer 1

You can do the following (using xml2 instead of XML as i find it easier to use)

require(xml2)
findTZ <- function(lon, lat, date=Sys.Date()){ 
  apiurl <- sprintf("https://maps.googleapis.com/maps/api/timezone/%s?location=%s,%s&timestamp=%d&sensor=%s", 
                    "xml", lat, lon, as.numeric(as.POSIXct(date)), "false")
  read_xml(apiurl) %>% xml_find_first(".//time_zone_id") %>% xml_text
}

To loop through your test data you can use:

require(tidyverse)
DF %>% mutate(TZ = map2_chr(longitude, latitude, findTZ))

Which gives you:

  longitude latitude                TZ
1      -122       54 America/Vancouver
2      -112       53  America/Edmonton
3      -102       52    America/Regina

As @Rich Scriven points out correctly you need to loop through the data somewhere. This loop is "hidden" in the map2_chr call.

Answer 2

考虑mapply将每对按元素值传递给函数以返回向量：

DF$timezones <- mapply(findTZ, DF$longitude, DF$latitude)