From 0dd2651f302ede2afb676a68211474f2b3a9b5c1 Mon Sep 17 00:00:00 2001
From: Arnaud Giersch <arnaud.giersch@iut-bm.univ-fcomte.fr>
Date: Tue, 18 Jan 2011 09:29:48 +0100
Subject: [PATCH] Add extract.pl.

A script to ease data extraction for graph drawing.  It prints data in a
format usable by gnuplot, or graph from the GNU plotutils.
---
 README     |  14 +++++++
 extract.pl | 106 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 120 insertions(+)
 create mode 100755 extract.pl

diff --git a/README b/README
index f6c136d..ff8c438 100644
--- a/README
+++ b/README
@@ -3,6 +3,7 @@ Contenu
 * Compilation de SimGrid
 * Compilation...
 * Utilisation
+* Tracé de courbes
 * Communications
 * Pour ajouter un nouvel algorithme d'équilibrage
 * Pour ajouter une nouvelle option au programme
@@ -45,6 +46,16 @@ Pour changer le niveau de détail des affichages :
 Pour plus de détail sur les options de logging :
     http://simgrid.gforge.inria.fr/doc/group__XBT__log.html#log_use
 
+Tracé de courbes
+================
+
+Le script extract.pl permet d'extraire les données à partir des traces
+de simulation et de le présenter sous un format acceptable par gnuplot
+ou par graph (plotutils).
+
+Exemple:
+        ./loba platform.xml 2>&1 | ./extract.pl | graph -CTX
+
 Communications
 ==============
 
@@ -178,6 +189,9 @@ Liste de fichiers
     colorized-loba              script pour exécuter loba en colorant les
                                 sorties
 
+    extract.pl                  outil d'extraction des données à partir des
+                                traces, pour tracer des courbes
+
     setlocalversion             calcule un numéro de version à partir du hash
                                 du dernier commit (git)
 
diff --git a/extract.pl b/extract.pl
new file mode 100755
index 0000000..fa393b9
--- /dev/null
+++ b/extract.pl
@@ -0,0 +1,106 @@
+#!/usr/bin/env perl
+
+use strict;
+use warnings;
+
+my $bookkeeping;
+my $flt = '[+-]?\d+(?:\.\d+)?(?:[eE][+-]?\d+)?';
+my $pflt = "($flt)";
+my $prefix = '^\[([^: ]+)(?::loba:\(\d+\))? ' . $pflt . '\] \[proc/INFO\] ';
+my $initmatch = $prefix . 'Initial load: ' . $pflt . '';
+my $finalmatch;
+my $plainmatch;
+
+my %alldata = ();
+
+while (<>) {
+    chomp;
+    if (s{^(?:\[0\.0+\] )?\[main/INFO\] \| bookkeeping\.*: }{}) {
+        $bookkeeping = $_ eq "on";
+        if ($bookkeeping) {
+            $finalmatch = $prefix .
+                'Final load after (\d+):(\d+) iterations: ' . $pflt .
+                ' ; expected: ' . $pflt;
+            $plainmatch = $prefix .
+                '\((\d+):(\d+)\) current load: ' . $pflt .
+                ' ; expected: ' . $pflt;
+        } else {
+            $finalmatch = $prefix .
+                'Final load after (\d+) iterations: ' . $pflt;
+            $plainmatch = $prefix . '\((\d+)\) current load: ' . $pflt;
+        }
+        if (0) {
+            print STDERR "BOOKKEEPING: \"$_\" ($bookkeeping)\n";
+            print STDERR "INITMATCH..: \"$initmatch\"\n";
+            print STDERR "PLAINMATCH.: \"$plainmatch\"\n";
+            print STDERR "FINALMATCH.: \"$finalmatch\"\n";
+        }
+    }
+    next if not defined $bookkeeping;
+    if (m{$plainmatch}) {
+        my $host = $1;
+        my $data;
+        if ($bookkeeping) {
+            $data = {
+                time     => $2,
+                lb       => $3,
+                comp     => $4,
+                load     => $5,
+                expected => $6,
+            };
+        } else {
+            $data = {
+                time     => $2,
+                lb       => $3,
+                comp     => $3,
+                load     => $4,
+                expected => $4,
+            };
+        }
+#        print STDERR "PUSH $host $data->{time} $data->{load} (plain)\n";
+        push @{$alldata{$host}}, $data;
+    } if (m{$initmatch}) {
+        my $host = $1;
+        my $data = {
+            time     => $2,
+            lb       => 0,
+            comp     => 0,
+            load     => $3,
+            expected => $3,
+        };
+#        print STDERR "PUSH $host $data->{time} $data->{load} (init)\n";
+        push @{$alldata{$host}}, $data;
+    } elsif (m{$finalmatch}) {
+        my $host = $1;
+        my $data;
+        if ($bookkeeping) {
+            $data = {
+                time     => $2,
+                lb       => $3,
+                comp     => $4,
+                load     => $5,
+                expected => $6,
+            };
+         } else {
+             $data = {
+                time     => $2,
+                lb       => $3,
+                comp     => $3,
+                load     => $4,
+                expected => $4,
+             };
+         }
+#        print STDERR "PUSH $host $data->{time} $data->{load} (final)\n";
+        push @{$alldata{$host}}, $data;
+    }
+}
+
+foreach my $host (sort(keys %alldata)) {
+#    print STDERR "GOT \"$host\"\n";
+    my $datalist = $alldata{$host};
+    print "# $host\n";
+    foreach my $data (@{$datalist}) {
+        print "$data->{time} $data->{load}\n";
+    }
+    print "\n"
+}
-- 
2.39.5