Menu
▾
▴
[Woc-commit] CVS: woc/data/wml splitwoclist.pl,NONE,1.1 syncwoclist.pl,NONE,1.1 woclist.pl,NONE,1.1 woclist.sh,NONE,1.1 woclist.xml,NONE,1.1 .cvsignore,1.1,1.2 install_source,1.1,NONE
|
From: Egon W. <eg...@us...> - 2004-01-14 13:02:04
|
Update of /cvsroot/woc/woc/data/wml
In directory sc8-pr-cvs1:/tmp/cvs-serv22668
Modified Files:
.cvsignore
Added Files:
splitwoclist.pl syncwoclist.pl woclist.pl woclist.sh
woclist.xml
Removed Files:
install_source
Log Message:
Added woclist: a single file wml source
--- NEW FILE: splitwoclist.pl ---
#! /usr/bin/perl
use strict;
use diagnostics;
# Author: Geert Josten
# hulp-script gebruikt door wocml.sh
my $infile;
my $outdir;
my $outfile;
if (@ARGV >= 2) {
$infile = shift;
$outdir = shift;
} else {
$infile = "woclist.xml";
$outdir = "tmp/";
}
open (INFILE, "$infile");
my $inheader = 1;
my $fileisopen = 0;
my $incml = 0;
my $header = "";
while (<INFILE>) {
my $file;
if (/DOCTYPE woclist/i) {
s#<!DOCTYPE woclist.*>#<!DOCTYPE ITEM PUBLIC \"-//www-woc.sci.kun.nl//DTD WOC Markup Language V1.0//EN\" \"http://www-woc.sci.kun.nl/wocml.dtd\">#i;
$header .= $_;
} elsif (/<woclist/i) {
$inheader = 0;
$fileisopen = 0;
} elsif (/<ITEM.*?FILE=\"(.*?)\"/i) {
# print OUTFILE "\n" if $fileisopen;
my $outfile = $outdir.$1.".xml";
s/FILE/CODE/i;
if (open (OUTFILE, ">$outfile")) {
$fileisopen = 1;
print OUTFILE $header;
print OUTFILE;
} else {
warn "Cannot write to file $outfile!\n";
$fileisopen = 0;
}
$inheader = 0;
} elsif (/<\/woclist/i) {
} elsif (/(.*?)<!-- \[(INSERT.*?)\](.*?)\[(\/INSERT)\] -->(.*)/i) {
print OUTFILE "$1<$2>$3<$4>$5\n";
$incml = 1;
} elsif (/<\/molecule/i) {
$incml = 0;
} elsif ($inheader) {
$header .= $_
} else {
print OUTFILE if $fileisopen && ! $incml;
}
}
# zoals in Convert.java
sub NAMELikeToANCHORLike {
# heeft 1 parameter
my $param = shift @_;
$param =~ s/^\s*(.*?)\s*$/$1/ig;
$param =~ s/\ /\_/g;
$param =~ s/[,|.|<|>]//g;
$param =~ tr/[A-Z]/[a-z]/;
return $param;
}
--- NEW FILE: syncwoclist.pl ---
#! /usr/bin/perl
use strict;
use diagnostics;
# Author: Geert Josten
# hulp-script gebruikt door wocml.sh
my $newdir;
my $orgdir;
if (@ARGV >= 2) {
$newdir = shift;
$orgdir = shift;
} else {
$newdir = "tmp/";
$orgdir = "../../../../share/woc/data/wml/";
}
my @newfiles = glob $newdir."*.xml";
foreach my $newfile (@newfiles) {
my $orgfile = $newfile;
$orgfile =~ s/^$newdir//;
my $subdir = (substr $orgfile, 0, 1)."/";
$subdir =~ s/[0-9]/123/;
$orgfile = $orgdir.$subdir.$orgfile;
if (-s $orgfile) {
my $diff = `diff $newfile $orgfile`;
if ($diff) {
warn "Original file different from $newfile!\n";
}
} else {
warn "Original file $orgfile of $newfile not found!\n";
}
}
--- NEW FILE: woclist.pl ---
#! /usr/bin/perl
use strict;
use diagnostics;
# Author: Geert Josten
# hulp-script gebruikt door wocml.sh
#my $cmldir="../../data/dadml/physicalprop/cml";
my $cmldir="../../../../share/woc/data/dadml/physicalprop/cml";
my $infile;
my $outfile;
if (@ARGV >= 2) {
$infile = shift;
$outfile = shift;
} else {
$infile = "woclist.tmp2.xml";
$outfile = "woclist.xml";
}
open (INFILE, "$infile");
open (OUTFILE, ">$outfile");
#print eerste regel alvast
$_ = <INFILE>; print OUTFILE;
while (<INFILE>) {
my $file;
if (/ITEM.*?CODE=/i) {
s/CODE/FILE/i;
print OUTFILE;
} elsif (/INSERT.*PhysicalProperties.*>(.*)</i) {
my @inserts = `awk '/<molecule/,/<\\/molecule/{print " ",\$0}' $cmldir/$1`;
s/<(INSERT.*?)>(.*?)<\/INSERT>/<!-- \[$1\]$2\[\/INSERT\] -->/i;
print OUTFILE;
foreach my $insert (@inserts) {
if ($insert =~ /^(.*Stoichiometry.*>)(.*)(<.*)$/i) {
my $pre = $1;
my $bruto = $2;
my $brutoID = $2;
my $post = $3;
$bruto =~ s/(\d+)/<sub>$1<\/sub>/ig;
print OUTFILE $pre,$bruto,$post,$/;
$pre =~ s/Stoichiometry/StoichiometryID/i;
$brutoID =~ s/(\d+)/{$1}/ig;
while ($brutoID =~ /\{(\d+?)\}/) {
my $number = sprintf "%04i", $1;
$brutoID =~ s/\{\d+?\}/$number/;
}
print OUTFILE $pre,$brutoID,$post,$/;
} else {
print OUTFILE $insert;
}
}
} elsif (/\<\?xml.*\?\>/i) {
s/\<\?xml.*\?\>//i;
print OUTFILE;
} else {
print OUTFILE;
}
}
# zoals in Convert.java
sub NAMELikeToANCHORLike {
# heeft 1 parameter
my $param = shift @_;
$param =~ s/^\s*(.*?)\s*$/$1/ig;
$param =~ s/\ /\_/g;
$param =~ s/[,|.|<|>]//g;
$param =~ tr/[A-Z]/[a-z]/;
return $param;
}
--- NEW FILE: woclist.sh ---
#! /usr/bin/sh -e
# Author: Geert Josten
NAME="woclist"
EXT=".xml"
TEMP=".tmp"
TEMP2=".tmp2"
#WMLDIR="../../data/wml"
WMLDIR="../../../../share/woc/data/wml"
# cmldir is in perl-stukje gedefinieerd
echo "" > $NAME$TEMP
cat `find $WMLDIR/123 -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/a -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/b -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/c -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/d -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/e -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/f -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/g -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/h -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/i -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/j -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/k -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/l -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/m -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/n -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/o -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/p -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/q -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/r -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/s -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/t -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/u -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/v -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/w -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/x -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/y -name "*.xml" | grep -v templates` >> $NAME$TEMP
cat `find $WMLDIR/z -name "*.xml" | grep -v templates` >> $NAME$TEMP
echo '<?xml version="1.0" encoding="ISO-8859-1"?>' > $NAME$TEMP2
echo '<!DOCTYPE '$NAME' SYSTEM "../../lib/xml/dtd/'$NAME'.dtd">' >> $NAME$TEMP2
echo '<'$NAME' xmlns:xsl="http://www.w3.org/1999/XSL/Transform"' >> $NAME$TEMP2
echo ' xmlns:import="http://www.sci.kun.nl/woc/import"' >> $NAME$TEMP2
echo ' xmlns:xlink="http://www.sci.kun.nl/woc/topicmap">' >> $NAME$TEMP2
awk "/<ITEM/,/<\/ITEM/{print}" $NAME$TEMP >> $NAME$TEMP2
echo '</'$NAME'>' >> $NAME$TEMP2
/usr/bin/perl $NAME.pl $NAME$TEMP2 $NAME$EXT
rm $NAME$TEMP $NAME$TEMP2
# chgrp wwwwoc $NAME$EXT
chmod 664 $NAME$EXT
--- NEW FILE: woclist.xml ---
<?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE woclist SYSTEM "../../lib/xml/dtd/woclist.dtd">
<woclist xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
xmlns:import="http://www.sci.kun.nl/woc/import"
xmlns:xlink="http://www.sci.kun.nl/woc/topicmap">
<ITEM NAME="1,1,1-trichloorethaan" ID="WOC00001542" CLASS="CHEMICAL" FILE="111trichloorethaan">
<DICT>
<NAME CLASS="IUPAC">
<LANG ID="EN">1,1,1-trichloorethane</LANG>
<LANG ID="NL">1,1,1-trichloorethaan</LANG>
</NAME>
<NAME CLASS="TRIVIAL">
<LANG ID="EN">methylchloroform</LANG>
</NAME>
</DICT>
<INDEX CLASS="CAS-NUMBER">25323-89-1</INDEX>
<GROUP>oplosmiddelen</GROUP>
</ITEM>
<ITEM NAME="1,11-difenylcyclopropaan" CLASS="CHEMICAL" ID="WOC00000001" FILE="11difenylcyclopropaan">
[...38534 lines suppressed...]
<NAME CLASS="TRIVIAL">
<LANG ID="NL">zwitterion</LANG>
<LANG ID="EN">zwitterion</LANG>
<LANG ID="FR">zwitterion</LANG>
</NAME>
</DICT>
<WOC>
<COMMENT>
<p>Een zwitterion heeft zowel een positieve als negatieve lading.
Moleculen met een zure en een basische groep zijn bij een bepaalde
pH zwitterion. Het
ion kan gevormd worden door een interne zuur-base reaktie. Een
aminozuur is een voorbeeld van een zwitterion.</p>
</COMMENT>
<tm:topicmap xmlns:tm="http://www.topic-maps.de/content/resources\tm.dtd"
xmlns:woctm="http://www.sci.kun.nl/woc/woctm">
</tm:topicmap>
</WOC>
</ITEM>
</woclist>
Index: .cvsignore
===================================================================
RCS file: /cvsroot/woc/woc/data/wml/.cvsignore,v
retrieving revision 1.1
retrieving revision 1.2
diff -C2 -r1.1 -r1.2
*** .cvsignore 15 Aug 2002 15:16:24 -0000 1.1
--- .cvsignore 14 Jan 2004 13:01:56 -0000 1.2
***************
*** 1,2 ****
--- 1,3 ----
Makefile
Makefile.in
+ tmp
--- install_source DELETED ---
|
