Add another parser for local files with metainformation

[proteocache.git] / datadb / compbio / cassandra / CassandraCreate.java

diff --git a/datadb/compbio/cassandra/CassandraCreate.java b/datadb/compbio/cassandra/CassandraCreate.java
index 54fffac..ad1fcbe 100644 (file)
--- a/datadb/compbio/cassandra/CassandraCreate.java
+++ b/datadb/compbio/cassandra/CassandraCreate.java
@@ -26,7 +26,9 @@ public class CassandraCreate {
        StringSerializer ss = StringSerializer.get();
        LongSerializer ls = LongSerializer.get();
 
-       // connect to the cluster
+       /*
+        * connect to the cluster and look weather the dababase has any data inside
+        */
        public void Connection() {
                cluster = HFactory.getOrCreateCluster("Protein Cluster", "127.0.0.1:9160");
                KeyspaceDefinition keyspaceDef = cluster.describeKeyspace("ProteinKeyspace");
@@ -60,11 +62,12 @@ public class CassandraCreate {
        }
 
        /*
-        * parsing data from
-        * http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat
+        * parsing data source and filling the database
         */
-       public void Parsing() {
-               /* CF ProteinRow store protein and prediction */
+       public void Parsing(String source) {
+               /*
+                * CF ProteinRow store protein and prediction
+                */
                mutatorString = HFactory.createMutator(ksp, ss);
 
                /*
@@ -72,21 +75,38 @@ public class CassandraCreate {
                 */
                mutatorLog = HFactory.createMutator(ksp, ss);
 
-               /* CF ProteinData store id and protein per data */
+               /*
+                * CF ProteinData store id and protein per data
+                */
                mutatorLong = HFactory.createMutator(ksp, ls);
 
-               System.out.println("Parsing......");
-               String in = "http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat";
-               DataParsing datParsing = new DataParsing();
-               datParsing.Parsing(in, 4);
-               flushData();
+               if (source.equals("http")) {
+                       // get data from real Jpred production server
+                       System.out.println("Parsing web data source......");
+                       String datasrc = "http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat";
+                       String prefix = "http://www.compbio.dundee.ac.uk/www-jpred/results";
+                       JpredParserHTTP parser = new JpredParserHTTP(prefix);
+                       parser.setSource("");
+                       parser.Parsing(datasrc, 4);
+                       flushData();
+               } else if (source.equals("file")) {
+                       // get irtifical data generated for the DB stress tests
+                       System.out.println("Parsing local file data source......");
+                       String datasrc = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/data.dat";
+                       String prefix = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/Jpreddata";
+                       JpredParserLocalFile parser = new JpredParserLocalFile(prefix);
+                       parser.Parsing(datasrc, 365);
+                       flushData();
+               } else {
+                       System.out.println("Unknown data source......");
+               }
        }
 
        public void flushData() {
                mutatorString.execute();
                mutatorLong.execute();
                mutatorLog.execute();
-               //System.out.println("Flush new data...");
+               // System.out.println("Flush new data...");
        }
 
        public void Closing() {
@@ -94,7 +114,9 @@ public class CassandraCreate {
                System.out.println("Cassandra has been shut down");
        }
 
-       // check whether the job id exists in the DB
+       /*
+        * check whether the job id exists in the DB
+        */
        public boolean CheckID(String jobid) {
                SliceQuery<String, String, String> sliceQuery = HFactory.createSliceQuery(ksp, ss, ss, ss);
                sliceQuery.setColumnFamily("ProteinLog").setKey(jobid).setRange("", "", false, 100);
@@ -105,6 +127,9 @@ public class CassandraCreate {
                return false;
        }
 
+       /*
+        * prepare data for insertion into the db
+        */
        public void InsertData(long dataWork, String dataBegin, String dataEnd, String ip, String id, String statusEx, String statusFinal,
                        String protein, List<FastaSequence> jnetpred) {
                mutatorLog.addInsertion(id, "ProteinLog", HFactory.createColumn("ip", ip, ss, ss))