I have a kml file, the unpacked version of this. It has several thousand layers with the XML tag, many of which have duplicate layer names.
I'd like to load this into R using sf::st_read. The trick is st_read reads one layer at a time and requires a layer name. I would be happy to iterate through the layer names fetched with st_layers()if they were unique, but they are not.
Is there an alternative way to specify the desired layer or, perhaps, a way to batch rename all the layers with a unique Id?
Thanks.
adding some color based on the accepted answer below. Initially, I tried to use 'read_xml' to edit the <name> nodes but they didn't seem to be found.
I downloaded the KMZ file, loaded it into Google Earth, then saved it back out as a KML file ("Reports.kml"). This was my first mistake. The resulting KML is tab-delimited which confuses read_xml. It is valid XML but the tags are not recognized properly by read_xml even though the st_ functions work. Better to use unzip on the KMZ file. Here is what happens with the saved-with-Google-Earth version:
layers<-st_layers("reports.kml")
data_frame(name=layers$name, type=flatten_chr(layers$geomtype)) %>%
count(name, type, sort=TRUE)
# A tibble: 1,358 x 3
# name type n
# <chr> <chr> <int>
# 1 July 2006 25
# 2 October 2006 25
# 3 August 2008 20
# 4 July 2009 19
# 5 August 2005 18
# 6 August 2007 18
# 7 November 2006 18
# 8 October 2004 17
# 9 August 2000 16
#10 November 2012 16
# ... with 1,348 more rows
kml<-read_xml("reports.kml")
xml_find_all(kml, ".//Folder/name")
# {xml_nodeset (0)}
Nothing! But there is something there:
xml_children(kml)
# {xml_nodeset (1)}
# [1] <Folder>\n <name>Reports</name>\n <open>1</open>\n <Folder>\n
# <name>Class A</name>\n ...
Here is what happens with the unzipped KMZ :
download.file(url="http://www.bfro.net/app/AllReportsKMZ.aspx",
destfile = "AllBFROReports.kmz",
mode="wb")
unzip("AllBFROReports.kmz",junkpaths = TRUE) #creates "doc.kml"
layers <- st_layers("doc.kml")
data_frame(name=layers$name, type=flatten_chr(layers$geomtype)) %>%
count(name, type, sort=TRUE)
# # A tibble: 1,376 x 3
# name type n
# <chr> <chr> <int>
# 1 July 2006 25
# 2 October 2006 25
# 3 August 2008 20
# 4 July 2009 19
# 5 August 2005 18
# 6 August 2007 18
# 7 November 2006 18
# 8 October 2004 17
# 9 August 2000 16
# 10 November 2012 16
# # ... with 1,366 more rows
st_layers is the same, but now the nodes are properly found!
kml <- read_xml("doc.kml")
xml_find_all(kml, ".//Folder/name")
{xml_nodeset (3874)}
[1] <name>June 2000</name>
[2] <name> 1995</name>
[3] <name>February 2004</name>
[4] <name>June 2004</name>
[5] <name>February 2004</name>
[6] <name>April 2008</name>
[7] <name>July 2009</name>
[8] <name>September 1981 and 1982</name>
[9] <name>July 1999</name>
[10] <name>November 1983</name>
[11] <name>October 2000</name>
[12] <name>August 1993</name>
[13] <name> 79, 80, 99</name>
[14] <name> 1978</name>
[15] <name>November 1980</name>
[16] <name>January 1997</name>
[17] <name> 1990</name>
[18] <name>December 1996</name>
[19] <name> 2000</name>
[20] <name> 2001</name>
...
Now the answer provided below works like a charm!
A bit of XML surgery will do the trick.
First, show the problem:
library(sf)
library(xml2)
library(tidyverse)
layers <- st_layers("AllBFROReports.kml")
data_frame(name=layers$name, type=flatten_chr(layers$geomtype)) %>%
count(name, type, sort=TRUE)
## # A tibble: 1,376 x 3
## name type n
## <chr> <chr> <int>
## 1 July 2006 25
## 2 October 2006 25
## 3 August 2008 20
## 4 July 2009 19
## 5 August 2005 18
## 6 August 2007 18
## 7 November 2006 18
## 8 October 2004 17
## 9 August 2000 16
## 10 November 2012 16
## # ... with 1,366 more rows
ugh. A very mean person made that file.
Now, read it in "raw":
kml <- read_xml("AllBFROReports.kml")
Add a sequential index number to each layer name:
idx <- 0
xml_find_all(kml, ".//Folder/name") %>%
walk(~{
idx <<- idx + 1
xml_text(.x) <- sprintf("%s-%s", idx, xml_text(.x))
})
Make a new file:
write_xml(kml, "AllBFROReports-unique.kml")
Prove it worked:
layers2 <- st_layers("AllBFROReports-unique.kml")
data_frame(name=layers2$name, type=flatten_chr(layers2$geomtype)) %>%
count(name, type, sort=TRUE)
## # A tibble: 3,874 x 3
## name type n
## <chr> <chr> <int>
## 1 1-June 2000 1
## 2 10-November 1983 1
## 3 100-September 1992 1
## 4 1000-October 1987 1
## 5 1001-October 1987 1
## 6 1002-October 1979 1
## 7 1003-June 1993 3D Point 1
## 8 1004- 1982 3D Point 1
## 9 1005- 1982 3D Point 1
## 10 1006-August 1977 3D Point 1
## # ... with 3,864 more rows
Read one layer in with the new indexified-name:
st_read("AllBFROReports-unique.kml", layer = "10-November 1983")
## Reading layer `10-November 1983' from data source `/Users/bob/Desktop/AllBFROReports-unique.kml' using driver `KML'
## Simple feature collection with 2 features and 2 fields
## geometry type: GEOMETRY
## dimension: XYZ
## bbox: xmin: -86.4677 ymin: 34.9484 xmax: -86.4441 ymax: 34.9637
## epsg (SRID): 4326
## proj4string: +proj=longlat +datum=WGS84 +no_defs