Normally, PerlMonks is not a coding service, but this one happened to be interesting to me. It appears that the Excel file is encoded in one of the Mac formats, I'm guessing MacRoman. I think this does what you want:

use warnings;
use strict;
use Spreadsheet::ParseExcel ();
use Spreadsheet::Read 'ReadData';
use Encode 'decode';
use XML::LibXML;

my $INFILE = 'TestPGR.xls';
my $ENCODING = 'MacRoman';
my $OUTFILE = 'TestPGR.xml';
my %FIELDS = ( 1=>'docid', 2=>'title', 3=>'version', 4=>'revision',
    5=>'node_order', 6=>'description', 7=>'status', 8=>'type',
    9=>'expected_coverage', );

my $book = ReadData($INFILE, rc=>1, cells=>0);
my $sheet = $book->[1] or die "Book doesn't have a sheet 1";

my $doc = XML::LibXML::Document->createDocument('1.0', 'UTF-8');
my $reqs = $doc->createElement('requirements');
$doc->setDocumentElement($reqs);

for my $r ( $sheet->{minrow}+1 .. $sheet->{maxrow} ) {
    my $req = $doc->createElement('requirement');
    for my $c ( $sheet->{mincol} .. $sheet->{maxcol} ) {
        next unless exists $FIELDS{$c};
        my $val = decode($ENCODING, $sheet->{cell}[$c][$r],
            Encode::FB_CROAK);
        my $node = $doc->createElement($FIELDS{$c});
        $node->appendText($val);
        $req->appendChild($node);
    }
    $reqs->appendChild($req);
}

$doc->toFile($OUTFILE,1);
[download]

Output (a UTF-8 encoded file):

<?xml version="1.0" encoding="UTF-8"?>
<requirements>
  <requirement>
    <docid>PP10-RG-010</docid>
    <title>MASTER DATA</title>
    <version>1</version>
    <revision>1</revision>
    <node_order>1</node_order>
    <description>Le format et le contenu des 2 documents sont décrits dans la SFD XXX (JIRA 624).</description>
    <status>V</status>
    <type>3</type>
    <expected_coverage>1</expected_coverage>
  </requirement>
  <requirement>
    <docid>PP10-RG-020</docid>
    <title>MASTER DATA</title>
    <version>1</version>
    <revision>1</revision>
    <node_order>2</node_order>
    <description>éiùûôêçà</description>
    <status>V</status>
    <type>3</type>
    <expected_coverage>1</expected_coverage>
  </requirement>
  <requirement>
    <docid>PP10-RG-030</docid>
    <title>MASTER DATA</title>
    <version>1</version>
    <revision>1</revision>
    <node_order>2</node_order>
    <description>éiùûôêçà&lt;&gt;
aqwzsx</description>
    <status>V</status>
    <type>3</type>
    <expected_coverage>1</expected_coverage>
  </requirement>
</requirements>

Many thanks of your return
I would like to convert MacRoman in HTML é --> & eacute (I put a space that HTML is not interpreter).
I don't know how to do that

Many thank Haukex

it's strange if I do the same code in a XLSX file, I have an error on accent. I'm working to understand what it's wrong.

it's strange if I do the same code in a XLSX file, I have an error on accent.

XLSX files contain XML files, and assuming Excel is writing them correctly, these encoding issues should hopefully not arise. In other words, the strings you get should not need the extra decode step, and that line in my code could be changed to my $val = $sheet->{cell}[$c][$r];.

I would like to convert MacRoman in HTML é --> & eacute

As I said and as haj explained some more, this doesn't really make sense in the context of XML, only in HTML, but you haven't explained how you want to integrate the two. I think the best thing would be if you could show an example of exactly what the output file should look like. Note that even in HTML, if the encoding is declared correctly, you don't even need escapes like é.