Author: moeller
Date: 2014-11-26 10:50:56 +0000 (Wed, 26 Nov 2014)
New Revision: 18517
Added:
trunk/community/edam/test/
trunk/community/edam/test/Makefile
trunk/community/edam/test/test_01.sh
Modified:
trunk/community/edam/debian2edam
Log:
Rough skeleton of a translator into EDAM ontology
I crafted a BASH script to help with the transition from
the Debian-typical YAML-like EDAM-description of our packages
into something that is closer to the expected JSON.
To play with it, please enter the test directory and type
'make'.
debian-med/community/edam/test $ make
./test_01.sh
Package: muscle
Description: Multiple alignment program of protein sequences
(
Package summary
Version 1:3.8.31
Description
Topic topic_0182
(function
(input data_2044)
(input format_1929)
(output data_1916)
W: found colon in ID of line ' output: format:1984 FASTA-aln' - transscribing
to underscore
(output format_1984)
(output format_1982)
(output format_1997)
)# openfunction(1) after loop
)# openscope(1) after loop
Modified: trunk/community/edam/debian2edam
===================================================================
--- trunk/community/edam/debian2edam 2014-11-25 17:44:08 UTC (rev 18516)
+++ trunk/community/edam/debian2edam 2014-11-26 10:50:56 UTC (rev 18517)
@@ -1,30 +1,219 @@
#!/bin/bash -e
+
+# A routine to facilitate the output to STDERR instead of the default STDIN
+function STDERR () {
+ cat - 1>&2
+}
+
+function echoindent () {
+ for i in $(seq 1 $1)
+ do
+ echo -n " "
+ done
+}
+
+# helper to properly close an open paranthesis
+function closeParenthesis () {
+ level=$((level-1))
+ echoindent $level
+ echo -n ")"
+ if [ -n "$1" ]; then
+ echo "# $1"
+ else
+ echo
+ fi
+}
+
+# Key argument indicating the debian directory from which to retrieve all the
+# information
pathToDebian=$1
+#verbose="yes"
+verbose=""
+# Variable keeping usage information
USAGE=<<EOUSAGE
debian2edam <path to 'debian' directory>
EOUSAGE
+filename=$(basename "$pathToDebian")
+if [ "edam" = "$filename" ]; then
+ pathToDebian=$(dirname "$pathToDebian") # upstream
+ pathToDebian=$(dirname "$pathToDebian") # debian
+fi
+
+if [ -z "$pathToDebian" ]; then
+ echo "$USAGE" | STDERR
+ echo "E: Please specify debian directory in which to find EDAM
annotation." | STDERR
+ exit -1
+fi
+
if [ ! -d "$pathToDebian" ]; then
- echo "$USAGE"
- echo "Could not find directory 'pathToDebian'"
+ echo "$USAGE" | STDERR
+ echo "E: Could not find directory '$pathToDebian'" | STDERR
exit -1
fi
+
+if [ ! -r "$pathToDebian/changelog" ]; then
+ echo "$USAGE" | STDERR
+ echo "E: Could not find a changelog file expected at
'$pathToDebian/changelog'" | STDERR
+ exit -1
+fi
+
cd $(dirname "$pathToDebian")
edamfile="debian/upstream/edam"
if [ ! -r "$edamfile" ]; then
- echo "$USAGE"
- echo "Could not access file '$edamfile' from $(pwd)"
+ echo "$USAGE" | STDERR
+ echo "E: Could not access file '$edamfile' from $(pwd)" | STDERR
exit -1
fi
-package=$(dpkg-parsechangelog |grep ^Source | sed -e 's/`^Source: //' )
+sourcepackage=$(dpkg-parsechangelog |grep ^Source | sed -e 's/`^Source: //' )
version=$(dpkg-parsechangelog |grep ^Version | cut -f2 -d\ | sed -e
's/-[^-][^-]*//' )
-echo "Package $package"
-echo "Version $version"
+declare -a descriptions
+#cat debian/control
+grep "^Package:" debian/control | \
+while read desc; do
+ d=$(echo "$desc"|sed -e 's/^[^:]*: *//')
+ echo Package: $d
+ packages[${#packages[*]}]="$d"
+done
+grep "^Description:" debian/control | \
+while read desc; do
+ d=$(echo "$desc"|sed -e 's/^[^:]*: *//')
+ echo Description: $d
+ descriptions[${#descriptions[*]}]="$d"
+done
+if [ ${#packagesp[*]} != ${#descriptions[*]} ]; then
+ echo "E: Internal error - expected same number of packages
(${#packagesp[*]}) as for their descriptions (${#descriptions[*]})" | STDERR
+ exit -1
+fi
+
+for packageno in $(seq 1 ${#descriptions[*]})
+do
+ echo Packages: ${packages[$packageno]}
+ echo Descriptions: ${descriptions[$packageno]}
+done
+prevstate="start";
+previndent=0
+currentscope=""
+currenttopic=""
+opentopic=0
+openfunction=0
+openscope=0
+indentlen=0
+level=0
+
+# Core part of the program
+# It reads every line of the EDAM file (see end of loop for the redirection)
+# and decides what to print to STDOUT.
+
+while IFS='' read -r line
+do
+ if [ -z "$line" ]; then
+ echo "Read empty line"
+ continue
+ fi
+
+ if [ -n "$verbose" ]; then
+ echo "line: '$line'" | STDERR
+ fi
+
+ # retrieve different parts of the description
+ blanks=$(echo "$line"|sed -e 's/^\( *\)\([^ :]\+\): *\([^ ]\+\).*$/\1/')
+ type=$(echo "$line"|sed -e 's/^\( *\)\([^ :]\+\): *\([^ ]\+\).*$/\2/')
+ val=$(echo "$line"|sed -e 's/^\( *\)\([^ :]\+\): *\([^ ]\+\).*$/\3/')
+
+ if echo "$val" | grep -q : ; then
+ echo "W: found colon in ID of line '$line' - transscribing to
underscore"|STDERR
+ val=$(echo "$val"|tr ":" "_")
+ fi
+
+ #echo "Indent='$blanks'"
+ #echo "Indentlength='$indentlen'"
+ #echo "Type='$type'"
+ #echo "Val='$val'"
+
+ if [ -n "$currentscope" -a "*" != "$currentscope" -a "summary" !=
"$currentscope" -a "scope" != "$type" ]; then
+ echo "I: Wrong scope ($currentscope) - ignored '$line'" | STDERR
+ continue
+ fi
+ indentlen=${#blanks}
+
+ if [ "scope" = "$type" ]; then
+ if [ $openfunction -gt 0 ]; then closeParenthesis
"openfunction($openfunction) in scope"; fi
+ currentscope="$val"
+ resourcename=$sourcepackage
+ if [ "*"!=$val -a "summary"!="$val" ];then
+ resourcename=$val
+ fi
+
+ if [ "summary" != "$val" -a "*" != "$val" ]; then
+ echo "I: treatment of multiple scopes not yet
implemented" | STDERR
+ else
+ echo "("
+ level=$((level+1))
+ echoindent
+ echo "Package $resourcename"
+ echoindent
+ echo "Version $version"
+ echoindent
+ echo "Description ${descriptions[1]}"
+ echoindent
+ echo "Topic $currenttopic"
+ openscope=1
+ fi
+ elif [ "topic" = "$type" ]; then
+ if [ $openfunction -gt 0 ]; then closeParenthesis
"openfunction($openfunction) in topic"; openfunction=0; fi
+ if [ $openscope -gt 0 ]; then closeParenthesis
"openscope($openscope) after loop"; openscope=0; fi
+ if [ "start" != "$prevstate" ]; then
+ closeParenthesis "topic with prior state - weird"
+ fi
+ currenttopic="$val"
+ # at some laterimplementation point, bits generated here would
be cached and then distributed
+ # to various lower-level scopes
+ elif [ "function" = "$type" ]; then
+ if [ $openfunction -gt 0 ]; then
+ closeParenthesis "openfunction($openfunction) in
function"
+ openfunction=0
+ fi
+ echoindent $level
+ echo "(function";
+ level=$((level+1))
+ openfunction=1
+ elif [ "input" = "$type" -o "output" = "$type" ]; then
+ echoindent $level
+ echo "($type $val)"
+ else
+ echo "W: unknown type '$type' - ignored" | STDERR
+ fi
+ prevstate=$type
+ #echo "indentlen='$indentlen'"
+done < $edamfile
+
+if [ $openfunction -gt 0 ]; then
+ closeParenthesis "openfunction($openfunction) after loop"
+ openfunction=0
+fi
+
+if [ $openscope -gt 0 ]; then
+ #echo "I: treatment of multiple scopes not yet implemented"|STDERR
+ closeParenthesis "openscope($openscope) after loop"
+ openscope=0
+fi
+
+#echo "indentlen='$indentlen'" | STDERR
+
+if [ $opentopic -gt 0 ]; then
+ opentopic=0
+fi
+
+#for i in $(seq $(($indentlen-$openfunction-$openscope-$opentopic)) -1 1)
+#do
+# closeParenthesis "indent $i"
+#done
Added: trunk/community/edam/test/Makefile
===================================================================
--- trunk/community/edam/test/Makefile (rev 0)
+++ trunk/community/edam/test/Makefile 2014-11-26 10:50:56 UTC (rev 18517)
@@ -0,0 +1,2 @@
+test:
+ ./test_01.sh
Added: trunk/community/edam/test/test_01.sh
===================================================================
--- trunk/community/edam/test/test_01.sh (rev 0)
+++ trunk/community/edam/test/test_01.sh 2014-11-26 10:50:56 UTC (rev
18517)
@@ -0,0 +1 @@
+../debian2edam ../../../packages/muscle/trunk/debian/upstream/edam
Property changes on: trunk/community/edam/test/test_01.sh
___________________________________________________________________
Added: svn:executable
+ *
_______________________________________________
debian-med-commit mailing list
debian-med-commit@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-med-commit
