ftbench fixes

myisam/ftbench/Ecompare.pl:
  bug in multi-column comparison fixed
myisam/ftbench/Ereport.pl:
  bug fixed
  report at 0, 0.5, 1
myisam/ftbench/README:
  updated
myisam/ftbench/ft-test-run.sh:
  various fixes

myisam/ftbench/Ecompare.pl

@@ -73,14 +73,14 @@ for $file (split) {
 
   for (0..$#l1) {
     $pp[$_]+=$p[$_]; $mm[$_]+=$m[$_];
-    $total.=rep($file, ($#l1 ? $_ : undef), $p[$_], $m[$_]);
+    $total[$_].=rep($file, ($#l1 ? $_ : undef), $p[$_], $m[$_]);
   }
   close OUT1;
   close OUT2;
 }
 
 for (0..$#l1) {
-  rep($total, ($#l1 ? $_ : undef), $pp[$_], $mm[$_]);
+  rep($total[$_], ($#l1 ? $_ : undef), $pp[$_], $mm[$_]);
 }
 
 sub rep {

myisam/ftbench/Ecreate.pl

+#!/usr/bin/perl
+
+$test=shift || die "Usage $0 testname [option]";
+$option=shift;
+
+open(D, "<data/$test.d") || die "Cannot open(<data/$test.d): $!";
+open(Q, "<data/$test.q") || die "Cannot open(<data/$test.q): $!";
+
+$N=0;
+
+print <<__HEADER__;
+DROP TABLE IF EXISTS $test;
+CREATE TABLE $test (
+  id int(10) unsigned NOT NULL,
+  text text NOT NULL,
+  FULLTEXT KEY text (text)
+) TYPE=MyISAM CHARSET=latin1;
+
+ALTER TABLE $test DISABLE KEYS;
+__HEADER__
+
+while (<D>) { chomp;
+  s/'/\\'/g; ++$N;
+  print "INSERT $test VALUES ($N, '$_');\n";
+}
+
+print <<__PREP__;
+ALTER TABLE $test ENABLE KEYS;
+SELECT $N;
+__PREP__
+
+$N=0;
+
+while (<Q>) { chomp;
+  s/'/\\'/g; ++$N;
+  $_="MATCH text AGAINST ('$_' $option)";
+  print "SELECT $N, id, $_ FROM $test WHERE $_;\n";
+}
+
+print <<__FOOTER__;
+DROP TABLE $test;
+__FOOTER__
+
+

myisam/ftbench/Ereport.pl

@@ -29,7 +29,7 @@ sub Favg { my $a=shift; $Pavg*$Ravg ? 1/($a/$Pavg+(1-$a)/$Ravg) : 0; }
   # F0    : a=0                 -- ignore precision
   # F5    : a=0.5
   # F1    : a=1                 -- ignore recall
-  while($eout_str =~ /^$qid\s+(\d+)\s+(\d+\.\d+)/) {
+  while($eout_str =~ /^$qid\s+(\d+)\s+(\d+(?:\.\d+)?)/) {
     $B++;
     $AB++ if $dq{$1+0};
     $Ravg+=$AB;
@@ -41,7 +41,7 @@ sub Favg { my $a=shift; $Pavg*$Ravg ? 1/($a/$Pavg+(1-$a)/$Ravg) : 0; }
   $Ravg/=$B*$A if $B;
   $Pavg/=$B    if $B;
 
-  printf "%5d %1.12f\n", $qid, Favg(0.5);
+  printf "%5d %1.12f %1.12f %1.12f\n", $qid, Favg(0),Favg(0.5),Favg(1);
 }
 
 exit 0;

myisam/ftbench/README

 1. should be run from myisam/ftbench/
 2. myisam/ftdefs.h should NOT be locked (bk get, not bk edit!)
 3. there should be ./data/ subdir with test collections, files:
-     test1.test
-     test1.relj
-     test2.test
-     test2.relj
+     test1.d
+     test1.q
+     test1.r
+     test2.d
+     test2.q
+     test2.r
    where test1, test2, etc - are arbitrary test names
-   
-   *.test are SQL files of the structure:
-=====
-DROP TABLE IF EXISTS ft;
-CREATE TABLE ft (
-  id int(10) unsigned NOT NULL,
-  text text NOT NULL,
-  FULLTEXT KEY text (text)
-);
-
-INSERT INTO ft VALUES (1, 'doc1...'),(2, 'doc2...'),...
-...
-
-SELECT COUNT(*) FROM ft;
-SELECT 1, id, MATCH text AGAINST ('query1') FROM ft WHERE MATCH text AGAINST ('query1');
-SELECT 2, id, MATCH text AGAINST ('query2') FROM ft WHERE MATCH text AGAINST ('query2');
-...
-=====
-
-   *.relj files have the structure:
+
+  *.[dq] files contain documents/queries one item per line.
+
+   *.r files have the structure:
    1 16 .....blablabla
    1 09 .....blablabla
    2 116 .....blablabla
@@ -47,7 +33,8 @@ SELECT 2, id, MATCH text AGAINST ('query2') FROM ft WHERE MATCH text AGAINST ('q
 
   all test results are compared with BEST results.
   
-  test directories *must* contain ftdefs.h, and *may* contain my.cnf
+  test directories may contain ftdefs.h, my.cnf, ft_mode
+  (the last one is used as in ... MATCH ... AGAINST ("..." $ft_mode) ...)
   NOTE: all *.out files in test directories will NOT be overwritten!
   delete them to re-test
 

myisam/ftbench/ft-test-run.sh

@@ -15,12 +15,7 @@ SOCK=$DATA/mysql.sock
 PID=$DATA/mysql.pid
 H=../ftdefs.h
 OPTS="--no-defaults --socket=$SOCK --character-sets-dir=$ROOT/sql/share/charsets"
-
-#  --ft_min_word_len=#
-#  --ft_max_word_len=#
-#  --ft_max_word_len_for_sort=# 
-#  --ft_stopword_file=name 
-#  --key_buffer_size=#
+DELAY=10
 
 stop_myslqd()
 {
@@ -38,48 +33,66 @@ if [ -w $H ] ; then
   exit 1
 fi
 
-for batch in t/BEST t/* ; do
-  A=`ls $batch/*.out`
-  [ ! -d $batch -o -n "$A" ] && continue
+stop_myslqd
+rm -rf var > /dev/null 2>&1
+mkdir var
+mkdir var/test
+
+for batch in t/* ; do
+  [ ! -d $batch ] && continue
+  [ $batch -ef t/BEST -a $batch != t/BEST ] && continue
+
+  rm -rf var/test/* > /dev/null 2>&1
   rm -f $H
-  ln -s $BASE/$batch/ftdefs.h $H
-  touch $H
+  if [ -f $BASE/$batch/ftdefs.h ] ; then
+    cat $BASE/$batch/ftdefs.h > $H
+    chmod a-wx $H
+  else
+    bk get -q $H
+  fi
   OPTS="--defaults-file=$BASE/$batch/my.cnf --socket=$SOCK --character-sets-dir=$ROOT/sql/share/charsets"
   stop_myslqd
   rm -f $MYSQLD
-  (cd $ROOT; gmake)
+  echo "building $batch"
+  echo "============== $batch ===============" >> var/ft_test.log
+  (cd $ROOT; gmake) >> var/ft_test.log 2>&1
 
   for prog in $MYSQLD $MYSQL $MYSQLADMIN ; do
     if [ ! -x $prog ] ; then
-      echo "No $prog"
+      echo "build failed: no $prog"
       exit 1
     fi
   done
 
-  rm -rf var 2>&1 >/dev/null
-  mkdir var
-  mkdir var/test
-
+  echo "=====================================" >> var/ft_test.log
   $MYSQLD $OPTS --basedir=$BASE --skip-bdb --pid-file=$PID \
                 --language=$ROOT/sql/share/english \
                 --skip-grant-tables --skip-innodb \
-                --skip-networking --tmpdir=$DATA &
+                --skip-networking --tmpdir=$DATA >> var/ft_test.log 2>&1 &
 
-  sleep 60
+  sleep $DELAY
   $MYSQLADMIN $OPTS ping
   if [ $? != 0 ] ; then
     echo "$MYSQLD refused to start"
     exit 1
   fi
-  for test in `cd data; echo *.test|sed "s/\.test//g"` ; do
-    echo "test $batch/$test"
-    $MYSQL $OPTS --skip-column-names test <data/$test.test >var/$test.eval
-    echo "report $batch/$test"
-    ./Ereport.pl var/$test.eval data/$test.relj > $batch/$test.out || exit
+  for test in `cd data; echo *.r|sed "s/\.r//g"` ; do
+    if [ -f $batch/$test.out ] ; then
+      echo "skipping $batch/$test.out"
+      continue
+    fi
+    echo "testing $batch/$test"
+    FT_MODE=`cat $batch/ft_mode 2>/dev/null`
+    ./Ecreate.pl $test "$FT_MODE" | $MYSQL $OPTS --skip-column-names test >var/$test.eval
+    echo "reporting $batch/$test"
+    ./Ereport.pl var/$test.eval data/$test.r > $batch/$test.out || exit
   done
   stop_myslqd
   rm -f $H
-  echo "compare $batch"
-  [ $batch -ef t/BEST ] || ./Ecompare.pl t/BEST $batch >> t/BEST/report.txt
+  bk get -q $H
+  if [ ! $batch -ef t/BEST ] ; then
+    echo "comparing $batch"
+    ./Ecompare.pl t/BEST $batch >> t/BEST/report.txt
+  fi
 done