Processing JSON with Perl

If we have an array of JSON morphological analyses, like this:

[{"RDF":{"Annotation":{"about":"urn:TuftsMorphologyService:usu:morpheus","hasTarget":{"Description":{"about":"urn:word:usu"}},"hasBody":[{"resource":"urn:uuid:e8e23662-6141-4b39-8ae2-ff25322dcdba"},{"resource":"urn:uuid:4ff79224-742c-46ff-8871-9f48c16d7d54"}],"title":null,"creator":{"Agent":{"about":"org.perseus:tools:morpheus.v1"}},"created":"29\nDec\n2012\n20:40:04\nGMT","Body":[{"about":"urn:uuid:e8e23662-6141-4b39-8ae2-ff25322dcdba","type":{"resource":"cnt:ContentAsXML"},"rest":{"entry":{"uri":null,"dict":{"hdwd":{"lang":"lat","$":"usus#2"},"pofs":{"order":3,"$":"noun"},"decl":"4th","gend":"masculine"},"infl":{"term":{"lang":"lat","stem":"u_s","suff":"u_"},"pofs":{"order":3,"$":"noun"},"decl":"4th","case":{"order":3,"$":"ablative"},"gend":"masculine","num":"singular","stemtype":"us_us"}}}},{"about":"urn:uuid:4ff79224-742c-46ff-8871-9f48c16d7d54","type":{"resource":"cnt:ContentAsXML"},"rest":{"entry":{"uri":null,"dict":{"hdwd":{"lang":"lat","$":"utor"},"pofs":{"order":3,"$":"noun"},"gend":"neuter"},"infl":{"term":{"lang":"lat","stem":"us","suff":"u_"},"pofs":{"order":3,"$":"noun"},"case":{"order":5,"$":"dative"},"gend":"neuter","mood":"supine","num":"singular","stemtype":"pp4"}}}}]}}},
{"RDF":{"Annotation":{"about":"urn:TuftsMorphologyService:ut:morpheus","hasTarget":{"Description":{"about":"urn:word:ut"}},"hasBody":{"resource":"urn:uuid:72303a08-7a9e-434d-bc75-e170070cdcae"},"title":null,"creator":{"Agent":{"about":"org.perseus:tools:morpheus.v1"}},"created":"29\nDec\n2012\n20:40:04\nGMT","Body":{"about":"urn:uuid:72303a08-7a9e-434d-bc75-e170070cdcae","type":{"resource":"cnt:ContentAsXML"},"rest":{"entry":{"uri":null,"dict":{"hdwd":{"lang":"lat","$":"ut"},"pofs":{"order":7,"$":"adverb"}},"infl":{"term":{"lang":"lat","stem":"ut"},"pofs":{"order":7,"$":"adverb"},"stemtype":"adverb","morph":"indeclform"}}}}}}}]

… and we want to get at specific contents of JSON objects, e. g. to traverse the route to the word which was sent to the service:

'RDF'/'Annotation'/'hasTarget'/'Description'/'about'

or to the lemma of that word:

'RDF'/'Annotation'/'Body'/'rest'/'entry'/'dict'/'hdwd'/'$'

or to the part-of-speech identification:

'RDF'/'Annotation'/'Body'/'rest'/'entry'/'dict'/'pofs'/'$'

… how to do this? We have also to account for the possibilities that (1) a word won't be recognized, (2) that there'll be several probable lemmata.

Yesterday and today, internet, the Llama book, and much experimenting taught us how to do it in Perl. The following script, for whatever it's worth, actually works on three possible responses to a Latin word query (no identification, an unambiguous lemma, an ambiguous one), and on any amount of Morphology Service JSON.

#!/usr/bin/perl -w
# persjson.pl - access different field values of Perl JSON parsing
# usage: perl persjson.pl
use JSON qw( decode_json );     # From CPAN
use Data::Dumper;               # Perl core module
use strict;                     # Good practice
use warnings;                   # Good practice
use File::Slurp 'read_file';
 
# read in the file:
my $jsonfile = $ARGV[0];
my $json = read_file( $jsonfile ) ;
 
# a list variable to hold the result:
my @result;
 
# Decode the entire JSON:
my $decoded_json = decode_json $json;
 
# for testing purposes; comment this when done.
# print Dumper $decoded_json;
 
# JSON is not a hash, but a list:
 for my $item (@$decoded_json) {
# get the query:
 	my $val2 = $item->{'RDF'}{'Annotation'}{'hasTarget'}{'Description'}{'about'};
# is there a Body element? 	
    my $val = $item->{'RDF'}{'Annotation'}{'Body'};
# is Body an array (of hashes)?    
    if (ref($val) eq "ARRAY") {
# iterate over array:
	my $numberof = scalar(@{$val});
# make counter:
	for(my $index=0; $index < $numberof; $index++) {
		my $row = @$val[$index];
# result as CSV, fields: query, lemma, qualification:
push (@result, '"' . $val2 . '","' . $row->{'rest'}{'entry'}{'dict'}{'hdwd'}{'$'} . '","VERBUM AMBIGUUM"', "\n");
}
}
 
# is Body hash?
       elsif (ref($val) eq "HASH") {
   	my @values = keys %$val;
# result as CSV, fields: query, lemma, part of speech:
       push (@result, '"' . $val2 . '","' . $val->{'rest'}{'entry'}{'dict'}{'hdwd'}{'$'} . '","' . $val->{'rest'}{'entry'}{'dict'}{'pofs'}{'$'}, '"', "\n");
   }
# if there's no Body (unidentified):
   else {
      push(@result, '"' . $val2 . '","FORMA NON RECOGNITA",""', "\n");
}
}
# put it all together:
 print @result;