Rewriting XML::DOM based module as XML::SAX

Cappadonna3030 has asked for the wisdom of the Perl Monks concerning the following question:

Hello: I recently wrote a script that will extract medical data from XML. I've been using XML::DOM. however, upon further testing. I realize that for my needs (this will be an automated script handling upto 7 20k XML documents at one time), that DOM isn't flexible enough.

Upon further research, realize that XML::SAX (or any general event driven XML parser) is probably my best bet for my requirements. However, I've never used SAX before and I have no idea how to begin to change over my code. I I've been pooring over web documents (including an online PDF from the PERL and XML books from O'reilly and New Rider, and I am still lost (or maybe its panic taking hold?)

Here is my original, XML::DOM code

package CathReport;
require XML::DOM;
use Date::Manip;
use POSIX qw(ceil floor);
use strict;


sub new 
{
        my $self  = {};
        $self->{CASE_HEADER}    = undef;
        $self->{PATIENT}    = undef;
        $self->{STAFF}        = undef;
          $self->{PROC}          = undef;
          $self->{LAB}          = undef;
        $self->{PROC}          = undef;
        $self->{ALDRETE}    = undef;
        $self->{MED}        = undef;
        $self->{CONDITIONS}    = undef;
        
        
        bless($self);           # but see below
        return $self;
}
    
    1;

## ł
sub CamPatData
{
    my $obj = shift;
    Date_Init("TZ=EST");

    
    if (@_)
    {
        my $file = shift;
        my $parser = XML::DOM::Parser->new();
        my $doc = $parser->parsefile($file);
        my (@patdemo, @header);
        my (%staff_set, %labs, %inv, %cond);
        my(%staff, %p1, %lab, %p2, %diag);
        my (%aldrete, %notes, %vital, %tools);
        my (%header);
        
        my ($cathno, $t_arrive, $t_end, $t_start, $t_event, $event_nam
+e);
        my %cathevents;
                
        my ($ln, $fn, $sex, $dob, $htu, $htn, $wtu, $wtn, $patnum, $pa
+tient, $dos, $stat, $proct, $endt);
        my ($s1, $s2, $s3, $workerbee);
        my ($chemical, $bc, $bu, $bv);
        my ($procname, $procsite, @sites, $amounts, $connum, $loc, $pi
+n);
        my ($cdate, $ctime, $csite, $chr1, $chr2);


        foreach my $cde ($doc->getElementsByTagName('CathDefinedEvents
+'))
        {
            $event_name = $cde->getAttribute('DefinedEventText');
            $t_event = $cde->getElementsByTagName('EventTime')->item(0
+)->getFirstChild->getNodeValue;                
            $cathevents{$event_name} = $t_event;
        }    
        
        #Get Patient Data
        foreach $patient($doc->getElementsByTagName('SA_PATIENT'))
        {
                    
            $ln = $patient->getElementsByTagName('MY_LAST_NAME')->item
+(0)->getFirstChild->getNodeValue;
            $fn = $patient->getElementsByTagName('MY_FIRST_NAME')->ite
+m(0)->getFirstChild->getNodeValue; 
            $dob = $patient->getElementsByTagName('MY_BIRTH_DATE')->it
+em(0)->getFirstChild->getNodeValue;
            $sex = $patient->getElementsByTagName('MY_GENDER')->item(0
+)->getFirstChild->getNodeValue; 
            $dob = $patient->getElementsByTagName('MY_BIRTH_DATE')->it
+em(0)->getFirstChild->getNodeValue;    
        }
            
            
            
            
        foreach $patient($doc->getElementsByTagName('CathStudy'))
        {
                    
            $htn = $patient->getElementsByTagName('PatHeight')->item(0
+)->getFirstChild->getNodeValue;
            $htu = $patient->getElementsByTagName('PatHeight')->item(0
+)->getAttribute('Units');
            $wtn = $patient->getElementsByTagName('PatWeight')->item(0
+)->getFirstChild->getNodeValue; 
            $wtu = $patient->getElementsByTagName('PatWeight')->item(0
+)->getAttribute('Units');
            $patnum = $patient->getElementsByTagName('StudyID')->item(
+0)->getFirstChild->getNodeValue;
            
        }        
        
        #Rounding Off 

        $wtn = ceil($wtn);
        $htn = ceil($htn);
        
        my $date = ParseDate($dob);    
        $dob = UnixDate($date,"%m/%d/%Y");
        
        @patdemo = ($ln, $fn, $dob, $sex, $htn, $htu, $wtn, $wtu, $pat
+num);
        
        foreach $workerbee($doc->getElementsByTagName('CathStaff'))
        {
            
            $s1 = $workerbee->getElementsByTagName('LastName')->item(0
+)->getFirstChild->getNodeValue;
            $s2 = $workerbee->getElementsByTagName('FirstName')->item(
+0)->getFirstChild->getNodeValue; 
            $s3 = $workerbee->getAttribute('Role');
            $staff_set{$s3} = "$s2 $s1";
            
        
            
        }
        
        foreach my $chemical($doc->getElementsByTagName('BloodComposit
+ion'))
        {
            my $bc = $chemical->getAttribute('BCMeasurement');
            my $bv = $chemical->getElementsByTagName('Value')->item(0)
+->getFirstChild->getNodeValue; 
            #$bv = floor($bv);
            $bv = sprintf("%.2f", $bv);
            $lab{$bc} = "$bv";
        }
        
        foreach $pin($doc->getElementsByTagName('CathProcedure'))
        {
            $procname = $pin->getAttribute('Procedure');
            $connum = $pin->getAttribute('ProcNum');

            #Check site
                $inv{$connum} = "$procname";        
                            
        }
        
            foreach my $dictation($doc->getElementsByTagName('CathLogE
+vent'))
            {
                my $timestamp = $dictation->getElementsByTagName('Even
+tTime')->item(0)->getFirstChild->getNodeValue;
                my $notetext = $dictation->getElementsByTagName('Event
+Text')->item(0)->getFirstChild->getNodeValue;
                $timestamp =~s/T/ /g;
                my $date = ParseDate($timestamp);    
                #$timestamp = UnixDate($date,"%m/%d/%Y");             
+   
                $timestamp = UnixDate($date,"%R");                
                
                $notes{$timestamp} = $notetext;
                                    
            }
            
             foreach my $oa($doc->getElementsByTagName('CathOA'))
             {
                 my $oatime = $oa->getElementsByTagName('OAtime')->ite
+m(0)->getFirstChild->getNodeValue; 
                  my $sysbp = $oa->getElementsByTagName('Systolic')->i
+tem(0)->getFirstChild->getNodeValue;
                  my $diabp = $oa->getElementsByTagName('Diastolic')->
+item(0)->getFirstChild->getNodeValue; 
                  my $hrt = $oa->getElementsByTagName('HeartRate')->it
+em(0)->getFirstChild->getNodeValue;
                  my $sao2 = $oa->getElementsByTagName('SaO2')->item(0
+)->getFirstChild->getNodeValue;                  
                  my $resp = $oa->getElementsByTagName('RespRate')->it
+em(0)->getFirstChild->getNodeValue; 
                  #my $oacom = $oa->getElementsByTagName('OAText')->it
+em(0)->getFirstChild->getNodeValue; 
                  my $oacom = "";
                  $oatime =~s/T/ /g;
                  $oatime = UnixDate($oatime,"%R");
                  
                  
                  $vital{$oatime} = [$sysbp, $diabp, $hrt, $sao2, $res
+p,$oacom];            
                              
            }
            
            ###Aldrete
            
            foreach my $ald ($doc->getElementsByTagName('CathAldrete')
+)
            {
                   my $aldtype = $ald->getAttribute('AldreteType'); 
                    my $aldact = $ald->getElementsByTagName('Activity'
+)->item(0)->getFirstChild->getNodeValue;
                    my $aldresp = $ald->getElementsByTagName('Respirat
+ion')->item(0)->getFirstChild->getNodeValue; 
                    my $aldcirc = $ald->getElementsByTagName('Circulat
+ion')->item(0)->getFirstChild->getNodeValue;
                    my $aldloc = $ald->getElementsByTagName('LOC')->it
+em(0)->getFirstChild->getNodeValue;                  
                    my $aldcolor = $ald->getElementsByTagName('Color')
+->item(0)->getFirstChild->getNodeValue; 
                    my $aldtot = $ald->getElementsByTagName('TotalScor
+e')->item(0)->getFirstChild->getNodeValue; 
                    $aldrete{$aldtype} = [$aldact, $aldresp, $aldloc, 
+$aldcolor, $aldtot, $aldcirc];            
            
                
            }


        #$obj->{CONDITIONS} = \%cond;        
        $obj->{PATIENT} = \@patdemo;
        $obj->{STAFF} = \%staff_set;
        $obj->{LAB} = \%lab;
        $obj->{PROC} = \%inv;
        $obj->{NOTES} = \%notes;
        $obj->{ONASS} = \%vital;
        $obj->{ALDRETE} = \%aldrete;
        
        
    
    }
    
    return $obj;
}
[download]

Any ideas would appreciate

Cappadonna

Readmore tags added by GrandFather

Comment on Rewriting XML::DOM based module as XML::SAX Download Code

Replies are listed 'Best First'.
Re: Rewriting XML::DOM based module as XML::SAX by planetscape (Chancellor) on Jun 10, 2006 at 17:59 UTC
It's entirely possible I'm shooting in the dark here, but I did find some interesting-looking hits by searching "dom2sax" perl and XML::DOM XML::SAX convert "dom2sax"; including this promising (Python) link: parser to generate SAX events from a DOM tree... More pyxml how-to here... I am equally unsure if the article mentioned here would be of any use... There's a bit of discussion here on SAX vs DOM, differences of which you are probably already aware... A much older writeup, but perhaps the links could still prove useful: Re: XM-Hell... And hopefully, someone far more knowledgeable than I will come along, see this, and have better answers... Good luck, Update: Even better search terms: "dom to sax" perl... Update 2: Here are two potentially helpful methods I have pieced together from my research: serialise the DOM to SAX events using XML::XPath; write out using XML::SAX::Writer XML::STX - DOM to SAX using DOMSource and SAXResult And one more link: SAX2 Chapter 3: Producing SAX2 Events planetscape	[reply]
Re: Rewriting XML::DOM based module as XML::SAX by GrandFather (Saint) on Jun 10, 2006 at 20:10 UTC
or for something only a little different it may be worth taking a look at XML::Twig. The code you present could be mapped easily into almost any XML parsing module. What are the new areas that you want to be able to handle that DOM is a pain for? I notice that at one point you have a for loop over 'SA_PATIENT' elements and retain data from the last element found which is later saved to @patdemo. Is it intended behaviour that you only retain the last element's data? By the way you could have trimed the code down to about a dozen lines and included some sample data (keep the data small too) for posting purposes - then people may feel more inclined to post sample solutions using other XML systems (I started, but couldn't be bothered - too much code all the same). DWIM is Perl's answer to Gödel	[reply]
Re^2: Rewriting XML::DOM based module as XML::SAX by Cappadonna3030 (Sexton) on Jun 12, 2006 at 14:09 UTC
Well, there are two really. First, every XML file will have the same general structure but not the same information. For example, the XML file A may have a social security number while B doesn't have one. Or, study C may contain EKG readings that don't show up in A or B. Using DOM, I would have to build a level of intelligence and redundancy to ensure that the script doesn't crap out if it doesn't see certain information. From what I've read, the event driven model is more appropriate, since I getting data fed from a another server. the other is speed. On my local workstation, the script runs fine. But this module will be part of a (slightly) larger automated script for a hospital in which multiple patients (XML files) will input at the same time. Like most patient care systems, I can't afford for this thing to crash b/c of memory hogging.	[reply]
Re^2: Rewriting XML::DOM based module as XML::SAX by Cappadonna3030 (Sexton) on Jun 12, 2006 at 16:44 UTC
Okay: Here is the code: sub CamPatData { my $obj = shift; Date_Init("TZ=EST"); if (@_) { my $file = shift; my $parser = XML::DOM::Parser->new(); my $doc = $parser->parsefile($file); my (@patdemo); my ($ln, $fn, $sex, $dob, $htu, $htn, $wtu, $wtn, $patnum, $pa +tient); #Get Patient Data foreach $patient($doc->getElementsByTagName('SA_PATIENT')) { $ln = $patient->getElementsByTagName('MY_LAST_NAME')->item +(0)->getFirstChild->getNodeValue; $fn = $patient->getElementsByTagName('MY_FIRST_NAME')->ite +m(0)->getFirstChild->getNodeValue; $dob = $patient->getElementsByTagName('MY_BIRTH_DATE')->it +em(0)->getFirstChild->getNodeValue; $sex = $patient->getElementsByTagName('MY_GENDER')->item(0 +)->getFirstChild->getNodeValue; $dob = $patient->getElementsByTagName('MY_BIRTH_DATE')->it +em(0)->getFirstChild->getNodeValue; } foreach $patient($doc->getElementsByTagName('CathStudy')) { $htn = $patient->getElementsByTagName('PatHeight')->item(0 +)->getFirstChild->getNodeValue; $htu = $patient->getElementsByTagName('PatHeight')->item(0 +)->getAttribute('Units'); $wtn = $patient->getElementsByTagName('PatWeight')->item(0 +)->getFirstChild->getNodeValue; $wtu = $patient->getElementsByTagName('PatWeight')->item(0 +)->getAttribute('Units'); $patnum = $patient->getElementsByTagName('StudyID')->item( +0)->getFirstChild->getNodeValue; } #Rounding Off $wtn = ceil($wtn); $htn = ceil($htn); my $date = ParseDate($dob); $dob = UnixDate($date,"%m/%d/%Y"); @patdemo = ($ln, $fn, $dob, $sex, $htn, $htu, $wtn, $wtu, $pat +num); } [download] Here is the Sample Data: `</CathStudy> <SA_PATIENT> <MY_PATIENT_ID>XXXXX</MY_PATIENT_ID> <MY_BIRTH_DATE>123-45-6789</MY_BIRTH_DATE> <MY_FIRST_NAME>BERT</MY_FIRST_NAME> <MY_GENDER>M</MY_GENDER> <MY_LAST_NAME>PUPPET</MY_LAST_NAME> <Address>123 SESAME ST</Address> <City>NEW YORK</City> <State>NY</State> <Zip>10110</Zip> <Race>Caucasian</Race> </SA_PATIENT>` [download] Any suggestions?	[reply] [d/l] [select]
Re^3: Rewriting XML::DOM based module as XML::SAX by GrandFather (Saint) on Jun 16, 2006 at 18:27 UTC
Looks like the simplest way to do this might be to slurp all the data into a hash using XML::Simple then either clean up the hash or extract the data from it into another hash as is appropriate: `use strict; use warnings; use Data::Dump::Streamer; use XML::Simple; my $str = do {local $/; <DATA>}; my %data = %{XMLin ($str)}; Dump (\%data);` [download] Read more... Data and output (2 kB)	[reply] [d/l] [select]