From 250f75706e4de1470163e0b4c791035bc9822fb6 Mon Sep 17 00:00:00 2001
From: graehl <graehl@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Tue, 3 Aug 2010 21:56:25 +0000
Subject: graph

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@470 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 report/prune_results/do.sh            |   7 +++
 report/prune_results/graph.sh         | 102 ++++++++++++++++++++++++++++++++++
 report/prune_results/merge-columns.pl |  20 +++++++
 3 files changed, 129 insertions(+)
 create mode 100755 report/prune_results/do.sh
 create mode 100755 report/prune_results/graph.sh
 create mode 100755 report/prune_results/merge-columns.pl

(limited to 'report')

diff --git a/report/prune_results/do.sh b/report/prune_results/do.sh
new file mode 100755
index 00000000..d541ee93
--- /dev/null
+++ b/report/prune_results/do.sh
@@ -0,0 +1,7 @@
+title="In multipass (TM -> TM+3gram), tuning for 1st pass -> BLEU (y) lost to pruning (x)." \
+    n4="tuned_unigram" \
+    n3="tuned_without_lm" \
+    n2="2ndpass_weights_without_lm" \
+    xlbl="post-pruning fraction of edges kept" \
+    obase=compare.1stpass.tune \
+    ./graph.sh space.2pass- hdt hdt0 hdt1
diff --git a/report/prune_results/graph.sh b/report/prune_results/graph.sh
new file mode 100755
index 00000000..ffd1dec8
--- /dev/null
+++ b/report/prune_results/graph.sh
@@ -0,0 +1,102 @@
+# see do.sh for usage
+d=$(dirname `readlink -f $0`)
+
+plboth() {
+    local o=$1
+    local oarg="-landscape"
+    [ "$portrait" ] && oarg=
+    shift
+    pl -png -o $o.png "$@"
+    pl -ps -o $o.ps $oarg "$@"
+    ps2pdf $o.ps $o.pdf
+}
+
+graph3() {
+    local y=$1
+    local ylbl="$2"
+    local y2=$3
+    local ylbl2="$4"
+    local y3=$5
+    local ylbl3="$6"
+    local ylbldistance=${7:-'0.7"'}
+    local name=${name:-$data}
+    local obase=${obase:-$opre$name.x_`filename_from $xlbl`.$y.`filename_from $ylbl`.$y2.`filename_from $ylbl2`.$y3.`filename_from $ylbl3`}
+    local of=$obase.png
+    local ops=$obase.ps
+        #yrange=0
+    #pointsym=none pointsym2=none
+    title=${title:-$ylbl $ylbl2 $ylbl3 vs. $xlbl}
+    xlbl=${xlbl:=x}
+    showvars_required obase xlbl ylbl ylbl2 ylbl3
+    require_files $data
+    plboth $obase -prefab lines data=$data x=1  y=$y name="$ylbl" y2=$y2 name2="$ylbl2" y3=$y3 name3="$ylbl3" ylbldistance=$ylbldistance xlbl="$xlbl" title="$title" ystubfmt '%4g' ystubdet="size=6" linedet2="style=1" linedet3="style=3" -scale ${scale:-1.4}
+    echo $of
+}
+
+pre() {
+    local pre=$1
+    shift
+    local f
+    for f in "$@"; do
+        echo $pre$f
+    done
+}
+
+main() {
+    files=( $(pre "$@") )
+    data=${data:-compare.$(filename_from $*)}
+    xlbl=${xlbl:-space}
+    showvars_required files data xlbl
+    require_files ${files[*]}
+    $d/merge-columns.pl ${files[*]} > $data
+    preview $data
+    graph3 2 "${n2:-$2}" 3 "${n3:-$3}" 4 "${n4:-$4}"
+    exit
+}
+
+echo2() {
+ echo "$@" 1>&2
+}
+
+errorq() {
+ echo2 ERROR: "$@"
+}
+
+error() {
+ errorq "$@"
+ return 2
+}
+
+showvars_required() {
+ echo2 $0 RUNNING WITH REQUIRED VARIABLES:
+    local k
+ for k in "$@"; do
+  eval local v=\$$k
+  echo2 $k=$v
+  if [ -z "$v" ] ; then
+    errorq "required (environment or shell) variable $k not defined!"
+    return 1
+  fi
+ done
+ echo2
+}
+
+require_files() {
+ local f
+ [ "$*" ] || error "require_files called with empty args list"
+ for f in "$@"; do
+    if ! have_file "$f" ; then
+        error "missing required file: $f"
+        return 1
+    fi
+ done
+ return 0
+}
+
+have_file() {
+    [ "$1" -a -f "$1" -a \( -z "$2" -o "$1" -nt "$2" \)  -a \( -z "$require_nonempty" -o -s "$1" \) ]
+}
+
+if ! [ "$nomain" ] ; then
+    main "$@";exit
+fi
diff --git a/report/prune_results/merge-columns.pl b/report/prune_results/merge-columns.pl
new file mode 100755
index 00000000..07b02a92
--- /dev/null
+++ b/report/prune_results/merge-columns.pl
@@ -0,0 +1,20 @@
+#!/usr/bin/perl -w
+use strict;
+my $T="\t";
+my %d;
+sub info {
+    local $,=' ';
+    print STDERR @_,"\n";
+}
+for my $n (0..$#ARGV) {
+    open F,'<',$ARGV[$n];
+    info($n,$ARGV[$n]);
+    while(<F>) {
+        my ($x,$f,$r)=split ' ',$_,3;
+        $d{$x}->[$n]=$f
+    }
+}
+for (sort keys %d) {
+    my @f=map { $T.(defined($_)?$_:'x') } @{$d{$_}};
+    print $_,@f,"\n";
+}
-- 
cgit v1.2.3