Convert Python dictionary to R data.frame

dict_to_df.R

py_dict = readLines('python_dictionary.txt')
# e.g.
#{"cat_name": "Ella", "dwell_status": "tree_dweller", "coat_color": "gray, white, orange", "is_from_hell": "Y"}
#{"cat_name": "Billie", "dwell_status": "bush_dweller", "coat_color": "gray, white", "is_from_hell": "N"}

dict_to_df = function(dict) {
  
  require(plyr)
  df = data.frame()
  df_temp = list()
  store = list()
  
  for (i in 1:length(dict)) {

    # Split up the dictionary entry
    split = unlist(strsplit(dict[i], '\",'))
    split = gsub('\\{', '', split)
    split = gsub('\\}', '', split)
    values = unlist(strsplit(split, ':'))

    # Parse out what will be the df headers
    headers = values[seq(1, length(values), 2)]       
    headers = gsub('\"', '', headers) # Remove quotes
    headers = gsub(' ', '', headers)  # and whitespace

    # Parse out what will be the df values
    row_values = values[seq(0, length(values), 2)]
    row_values = gsub('\"', '', row_values) # Remove quotes
    row_values = gsub(' ', '', row_values)  # and whitespace

    # Construct a dataframe with 1 row
    out = data.frame(t(row_values))
    colnames(out) = headers

    store[i] = list(out)

    if (i %% 1000 == 0) { print(round(i / length(dict), 2)) }

  }

    # rbind all the dataframes together into one dataframe 
    list_length = length(store)
    
    # If the dictionary is sufficiently large rbind will be slow
    # as all hell, so break the rbinding into multiple steps
    if (list_length >= 3000) {

      no_splits = round(list_length / 500)
      chunks = split(store, 1:no_splits)

      for (j in 1:no_splits) {

        df_temp[j] = list(rbind.fill(chunks[[j]]))
        
      }
      df = rbind.fill(df_temp)
      return(df)
    }

    else {

      df = rbind.fill(store)
      return(df)
    }

}

r_df = dict_to_df(py_dict)

r_df
#  cat_name dwell_status        coat_color is_from_hell
#1     Ella tree_dweller gray,white,orange            Y
#2   Billie bush_dweller        gray,white            N

# http://animal.discovery.com/tv-shows/my-cat-from-hell

Watch for big performance losses as your dictionary size increases: http://crowding.github.io/plyr/rbind-fill.html

Thanks for this script. I had to fiddle with it as my application required the ordering of dict to be retained, and currently that is lost where list_length>=3000 (due to call of interaction within store when applied to a list).