anonymize ITS file

[tjmahr]

Two options for anonymizing: Deleting data and overwriting. Consult with wiki and python version.

Toybox demo of how to modify an XML document in R...

library("XML")
library("magrittr")

# Pretend this is an ITS file
tree <- '
  <xml_sucks>
    <xml_is_stupid amen="TRUE"/>
    <xml_is_lame lameness="1000">yes is it </xml_is_lame>
    <xml_is_lame lameness="1001">what am i even doing </xml_is_lame>
  </xml_sucks>
'

# Parse it
doc <- xmlParse(tree)

# Nuclear option: Delete all attributes
remove_attributes_in_xpath <- function(doc, path) {
  doc %>% 
    getNodeSet(path) %>% 
    lapply(removeAttributes)
}

# Use dots to specify the attribute to be updated and the new value
set_attribute_in_xpath <- function(doc, path, ...) {
  # Create named character vector from dots
  edits <- unlist(list(...))

  # One-off function to apply to each node
  set_this_attribute <- function(node) {
    xmlAttrs(node) <- edits
    node
  }

  # Find matching nodes and update
  doc %>% 
    getNodeSet(path) %>% 
    lapply(set_this_attribute)
}

# Drop all attributes from a node
remove_attributes_in_xpath(doc, "//xml_is_stupid")
#> [[1]]
#> <xml_is_stupid/>

# Overwrite an attribute
set_attribute_in_xpath(doc, "//xml_is_lame", lameness = 2000, verbosity = 10)
#> [[1]]
#> <xml_is_lame lameness="2000" verbosity="10">yes is it </xml_is_lame> 
#> 
#> [[2]]
#> <xml_is_lame lameness="2000" verbosity="10">what am i even doing </xml_is_lame>

# look at updated tree
doc
#> <?xml version="1.0"?>
#> <xml_sucks>
#>   <xml_is_stupid/>
#>   <xml_is_lame lameness="2000" verbosity="10">yes is it </xml_is_lame>
#>   <xml_is_lame lameness="2000" verbosity="10">what am i even doing </xml_is_lame>
#> </xml_sucks>
#> 

[Teebusch]

I suggest adding a function anonymize_its() that returns an anonymized its (xml) file. This way it can be used to anonymize its files in batch.
Then add a parameter anonymize to load_its() that, if TRUE, will call anonymize_its internally to anonymize during loading.

[tjmahr]

My main reason for not implementing this fully is that I wasn’t sure whether I had a comprehensive list of all the sensitive fields in the ITS file.