[Swift-commit] r4225 - provenancedb/apps/oops
lgadelha at ci.uchicago.edu
lgadelha at ci.uchicago.edu
Sun Mar 27 18:55:29 CDT 2011
Author: lgadelha
Date: 2011-03-27 18:55:29 -0500 (Sun, 27 Mar 2011)
New Revision: 4225
Modified:
provenancedb/apps/oops/oops_extractor.sh
Log:
OOPS annotation extractor update to match new schema.
Modified: provenancedb/apps/oops/oops_extractor.sh
===================================================================
--- provenancedb/apps/oops/oops_extractor.sh 2011-03-27 21:54:23 UTC (rev 4224)
+++ provenancedb/apps/oops/oops_extractor.sh 2011-03-27 23:55:29 UTC (rev 4225)
@@ -37,12 +37,12 @@
OOPS_RUN_ID=$(echo $i | awk -F . '{print $3}')
cd $PROTESTS_HOME/$k/$i
LOG_FILENAME=$(ls *-*-*-*.log)
- WORKFLOW_ID=$(echo "select id from workflow where log_filename like '%$LOG_FILENAME%'" | $SQLCMD -t | awk '{print $1}')
+ WORKFLOW_ID=$(echo "select id from run where log_filename like '%$LOG_FILENAME%'" | $SQLCMD -t | awk '{print $1}')
cd $IMPORT_HOME/swift-logs
- echo "insert into annot_wf_txt (id, name, value) values ('$WORKFLOW_ID','oops_run_id','$OOPS_RUN_ID');" | $SQLCMD
+ echo "insert into annot_run_text (id, name, value) values ('$WORKFLOW_ID','oops_run_id','$OOPS_RUN_ID');" | $SQLCMD
#extracts scientific parameters given as input to the workflow in file *.params.
- echo "select file.id,file.filename from process, ds_usage, ds_containment, file where process.id=ds_usage.process_id and ds_usage.dataset_id=out_id and file.id=ds_containment.in_id and filename like '%.params' and process.name='PrepLoop' and process.workflow_id='$WORKFLOW_ID' and ds_usage.direction='O';" > query.sql;
+ echo "select file.id,file.filename from process, dataset_production, dataset_containment, file where process.id=dataset_production.process_id and dataset_usage.dataset_id=out_id and file.id=dataset_containment.in_id and filename like '%.params' and process.name='PrepLoop' and process.run_id='$WORKFLOW_ID';" > query.sql;
$SQLCMD -t -A -F " " -f query.sql -o result.txt
@@ -55,23 +55,23 @@
NAME=$(echo $line | awk 'BEGIN { FS = "=" }; {print $1}')
RIGHT=$(echo $line | awk 'BEGIN { FS = "=" }; {print $2}')
if [ "$NAME" = "SAMPLE RANGE" ]; then
- echo "insert into annot_ds_txt values ('$DATASET_ID', 'sample_range', '$RIGHT');" | $SQLCMD
+ echo "insert into annot_dataset_text values ('$DATASET_ID', 'sample_range', '$RIGHT');" | $SQLCMD
fi
if [ "$NAME" = "RESTRAIN DISTANCE" ]; then
VALUE1=$(echo $RIGHT | awk 'BEGIN { FS = "," }; {print $1}')
VALUE2=$(echo $line | awk 'BEGIN { FS = "=" }; {print $2}' | awk 'BEGIN { FS = "," }; {print $2}')
- echo "insert into annot_ds_num values ('$DATASET_ID', 'restrain_distance_1', $VALUE1);" | $SQLCMD
- echo "insert into annot_ds_num values ('$DATASET_ID', 'restrain_distance_2', $VALUE2);" | $SQLCMD
+ echo "insert into annot_dataset_numeric values ('$DATASET_ID', 'restrain_distance_1', $VALUE1);" | $SQLCMD
+ echo "insert into annot_dataset_numeric values ('$DATASET_ID', 'restrain_distance_2', $VALUE2);" | $SQLCMD
fi
if [ "$NAME" = "MAXIMUM NUMBER OF STEPS" ]; then
- echo "insert into annot_ds_num values ('$DATASET_ID', 'maximum_number_of_steps', $RIGHT);" | $SQLCMD
+ echo "insert into annot_dataset_numeric values ('$DATASET_ID', 'maximum_number_of_steps', $RIGHT);" | $SQLCMD
fi
done < $FILENAME
#extracts length of the fasta sequence given as input to the workflow in file *.fasta.
cd $IMPORT_HOME/swift-logs
- echo "select file.id,file.filename from process, ds_usage, ds_containment, file where process.id=ds_usage.process_id and ds_usage.dataset_id=out_id and file.id=ds_containment.in_id and filename like '%.fasta' and process.name='PrepLoop' and process.workflow_id='$WORKFLOW_ID' and ds_usage.direction='O';" > query.sql;
+ echo "select file.id,file.filename from process, dataset_production, dataset_containment, file where process.id=dataset_production.process_id and dataset_production.dataset_id=out_id and file.id=dataset_containment.in_id and filename like '%.fasta' and process.name='PrepLoop' and process.run_id='$WORKFLOW_ID';" > query.sql;
$SQLCMD -t -A -F " " -f query.sql -o result.txt
@@ -82,8 +82,9 @@
if [ -n "$FILENAME" ]; then
SEQLENGTH=$(awk '{if (NR==2) print $1}' $FILENAME | wc -c)
- echo "insert into annot_ds_num values ('$DATASET_ID', 'fasta_sequence_length', $SEQLENGTH);" | $SQLCMD
+ echo "insert into annot_dataset_numeric values ('$DATASET_ID', 'fasta_sequence_length', $SEQLENGTH);" | $SQLCMD
fi
+
# extracts scientific parameters given as output by the workflow in *.log.
# relevant lines:
# zone2 (Initial Energy: -21352.116911)
@@ -98,7 +99,7 @@
cd $IMPORT_HOME/swift-logs
- echo "select file.id,file.filename from process, ds_usage, ds_containment, file where process.id=ds_usage.process_id and ds_usage.dataset_id=out_id and file.id=ds_containment.in_id and filename like '%.log' and process.name='LoopModel' and process.workflow_id='$WORKFLOW_ID' and ds_usage.direction='O';" > query.sql;
+ echo "select file.id,file.filename from process, dataset_production, dataset_containment, file where process.id=dataset_production.process_id and dataset_production.dataset_id=out_id and file.id=dataset_containment.in_id and filename like '%.log' and process.name='LoopModel' and process.run_id='$WORKFLOW_ID';" > query.sql;
$SQLCMD -t -A -F " " -f query.sql -o result.txt
@@ -109,28 +110,28 @@
while read token1 token2 token3 token4; do
if [ "$token2" = "(Initial Energy:" ]; then
initialenergy=$(echo $token3 | awk 'BEGIN { FS = "\)" }; {print $1}')
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'initial_energy', $initialenergy);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'initial_energy', $initialenergy);" | $SQLCMD
fi
if [ "$token1" = "Total" ] && [ "$token2" = "Function" ] && [ "$token3" = "Evaluations:" ]; then
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'total_function_evaluations', $token4);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'total_function_evaluations', $token4);" | $SQLCMD
fi
if [ "$token1" = "Increasing" ] && [ "$token2" = "transitions:" ]; then
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'accepted_increasing_transitions', $token3);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'accepted_increasing_transitions', $token3);" | $SQLCMD
fi
if [ "$token1" = "Decreasing" ] && [ "$token2" = "transitions:" ]; then
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'accepted_decreasing_transitions', $token3);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'accepted_decreasing_transitions', $token3);" | $SQLCMD
fi
if [ "$token1" = "Rejected" ] && [ "$token2" = "transitions:" ]; then
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'rejected_transitions', $token3);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'rejected_transitions', $token3);" | $SQLCMD
fi
if [ "$token1" = "Final" ] && [ "$token2" = "Energy:" ]; then
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'final_energy', $token3);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'final_energy', $token3);" | $SQLCMD
fi
if [ "$token1" = "Final" ] && [ "$token2" = "Temp:" ]; then
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'final_temp', $token3);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'final_temp', $token3);" | $SQLCMD
fi
if [ "$token1" = "Total" ] && [ "$token2" = "Running" ] && [ "$token3" = "Time:" ]; then
- echo "insert into annot_ds_num (id, name, value) values ('$dataset', 'total_running_time', $token4);" | $SQLCMD
+ echo "insert into annot_dataset_numeric (id, name, value) values ('$dataset', 'total_running_time', $token4);" | $SQLCMD
fi
done < $FILENAME
done < $IMPORT_HOME/swift-logs/result.txt
More information about the Swift-commit
mailing list