This test program works against your sample data, try running it against the complete file

#!perl
use strict;
use warnings;
use XML::Simple;
use Data::Dump 'pp';

my $blast = XMLin('BLAST1.XML');
my $hits = $blast->{BlastOutput_iterations}->{Iteration}->{Iteration_h
+its}->{Hit};
my $ret;
#push @ret, $_->{Hit_def} foreach (@{$hits});
foreach (@{$hits}) {
  push @{$ret},join '|', 
      $_->{Hit_def},
      $_->{Hit_num},
      $_->{Hit_hsps}->{Hsp}->{Hsp_identity};
}
pp $ret;
[download]

Ah, it's killing me. I tried your test program with my original XML file. Same error as before: 'Not a HASH reference at line 12' (which is: push @{$ret},join '|',).

I tried it however with the partial file that I sent you. Wow! It works perfectly! I ran again the original program with your modification on the partial XML file. Again, it works perfectly, I get exactly the results I hoped for.

So is it related to the input file? Maybe my XML file is somehow messed up. So for testing I generated a few more XML files with the appropriate software, but all of them caused this 'Not a HASH reference' error. I compared the complete XML files with the partial XML I sent you, went over and over them like a thousand times, but I couldn't find any difference, except for the number of 'Hit'-s of course, and consequently, the size. Oh, there was one other thing: In the complete XMLs the lines ended with a single newline character (\n), but in the partial XML the EOL was a carriage return and a newline (\r\n). So I replaced all the \n with \r\n, but I still got the error, so the EOL seems to be irrelevant. And with the partial XML the program still worked correctly even if I replaced every \r\n with \n.

I also tried to shamelessly hack into your code with my limited Perl knowledge, trying different ways to reference, but it only got worse (as had been expected :))

So all in all, I am totally clueless. I don't get why it should be a HASH reference in the first place; @{$ret} is an array, right? Not a hash. Then I don't get how the input file influences the reference. Especially that in line 12 there is nothing related to the input file, it only says that we will push values into the end of the empty @{$ret} array (and join some of them). And finally I don't get what is the key difference between the 'good' and 'bad' XML files. Why only the partial file is working? If the program runs properly for 2 hits, why it doesn't for 99 hits?

Mysterious. So much for today, tomorrow I will start removing the hits from a complete XML file one by one, to see if there is a size limit somewhere, or if it has any effect at all...

Thank you for your selfless help again!

Look in the XMl file for instances where you have multiple <Hit_hsps> tags within a <Hit> or multiple <Hsp> tags within a <Hit_hsps>.

This test data replicates your error

<?xml version="1.0"?>
<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "http://ww
+w.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd">
<BlastOutput>
<BlastOutput_iterations>
<Iteration>
<Iteration_hits>
<Hit>
  <Hit_num>1</Hit_num>
  <Hit_def>Uncultured Sulfuricurvum sp. RIFRC-1, complete genome</Hit_
+def>
  <Hit_hsps>
    <Hsp>
      <Hsp_identity>16</Hsp_identity>
    </Hsp>
  </Hit_hsps>
</Hit>
<Hit>
  <Hit_num>2</Hit_num>
  <Hit_def>Neosartorya fischeri NRRL 181 conserved hypothetical protei
+n (NFIA_106270) partial mRNA</Hit_def>
  <Hit_hsps>
    <Hsp>
      <Hsp_identity>16</Hsp_identity>
    </Hsp>
  </Hit_hsps>
  <Hit_hsps>
    <Hsp>
      <Hsp_identity>16a</Hsp_identity>
    </Hsp>
    <Hsp>
      <Hsp_identity>16b</Hsp_identity>
    </Hsp>
  </Hit_hsps>
</Hit>
</Iteration_hits>
</Iteration>
</BlastOutput_iterations>
</BlastOutput>
[download]

Update : Try this
Read more... (924 Bytes)
poj

Same error as before: 'Not a HASH reference at line 12' (which is: push @{$ret},join '|',).

The statement isn't over, keep following the comma

  push @{$ret},join '|', 
      $_->{Hit_def}, ## HASH ACCESS
      $_->{Hit_num}, ## HASH ACCESS
      $_->{Hit_hsps}->{Hsp}->{Hsp_identity};  ## HASH OF HASH OF HASH 
+ACCESS
[download]

So is it related to the input file?

Absolutely, sometimes a Hit_hsps doesn't have a Hsp_identity or an Hsp ... if the data isn't there its not there

You can cure death with judicious use of block eval and subroutines and Data::Diver :) See Re: Parsing SOAP::Lite results (with Data::Diver), Re: $var{el} += 1 vs $var{el} = $var{el} + 1, Re: Problem with counting the nodes in a hash created by XML::Simple (use XML::Rules, Data::Diver )

So is it related to the input file? Maybe my XML file is somehow messed up.

Yes, its because you're using XML::Simple when you want to use XML::Rules

XML::Rules does by default what folks expect from XML::Simple

See examples Re: XML::Simple Not getting the proper output. ( xml2XMLRules, Re^3: Parsing XML (XML::Rules xml2XMLRules )

The more files you give xml2XMLRules the better the rules it spits out for you to plug into XML::Rules, the more stable the resulting data structure

... ? ... ! ...

You can always use rehohy to visualize your data like Re^3: Parsing a JSON response, Re: A "simple" question regarding accessing a value in a complex data structure., Re^2: Array of structures within an array of structures.

See also references quick reference, hashes are curly
Re: How to traverse a two dimentional array in a constructor?, Re: Parsing SOAP::Lite results,, Re: Trouble accessing Hash of Arrays (troubleshooting basics),
Re^5: Combining 3 files,