datadb/compbio/cassandra/JpredParserHTTP.java

   1 package compbio.cassandra;
   2
   3 import java.io.BufferedReader;
   4 import java.io.FileNotFoundException;
   5 import java.io.IOException;
   6 import java.io.InputStream;
   7 import java.io.InputStreamReader;
   8 import java.net.HttpURLConnection;
   9 import java.net.MalformedURLException;
  10 import java.net.URL;
  11 import java.net.URLConnection;
  12 import java.util.ArrayList;
  13 import java.util.Calendar;
  14 import java.util.Date;
  15 import java.util.List;
  16
  17 import compbio.cassandra.JpredParser;
  18 import compbio.data.sequence.FastaReader;
  19 import compbio.data.sequence.FastaSequence;
  20 import compbio.engine.JpredJob;
  21 import compbio.engine.ProteoCachePropertyHelperManager;
  22 import compbio.engine.archive.Archive;
  23 import compbio.engine.archive.ArchivedJob;
  24 import compbio.util.PropertyHelper;
  25 import compbio.util.Util;
  26
  27 public class JpredParserHTTP implements JpredParser {
  28         private CassandraWriter cw = new CassandraWriter();
  29         private static Archive archive;
  30         private String dirprefix;
  31         private List<FastaSequence> alignment;
  32         private List<FastaSequence> predictions;
  33         private int countNoData;
  34         private static boolean archiving = false;
  35         private static final PropertyHelper ph = ProteoCachePropertyHelperManager.getPropertyHelper();
  36
  37         public JpredParserHTTP() {
  38                 dirprefix = "http://www.compbio.dundee.ac.uk/www-jpred/results";
  39         }
  40
  41         public JpredParserHTTP(String sourceurl) {
  42                 dirprefix = sourceurl;
  43         }
  44
  45         public void setSource(String newsourceprefix) {
  46                 dirprefix = newsourceprefix;
  47         }
  48
  49         private boolean initBooleanValue(String key) {
  50                 assert key != null;
  51                 String status = ph.getProperty(key);
  52                 if (Util.isEmpty(status)) {
  53                         return false;
  54                 }
  55                 return new Boolean(status.trim()).booleanValue();
  56         }
  57
  58         public void Parsing(String source, int nDays) throws IOException {
  59                 Calendar cal = Calendar.getInstance();
  60                 cal.add(Calendar.DATE, -nDays);
  61                 archiving = initBooleanValue("archive.enable");
  62                 if (archiving) {
  63                         archive = new Archive();
  64                 }
  65                 for (int i = 0; i < nDays; ++i) {
  66                         cal.add(Calendar.DATE, 1);
  67                         String date = cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DATE);
  68                         ParsingOneDay(source, date);
  69                 }
  70         }
  71
  72         /*
  73          * The method parses the Jpred output concise file in the FASTA format If
  74          * there is a record with ID = QUERY or jobid, this a "one protein" job
  75          * otherwise this is an alignment job
  76          */
  77         private String parsePredictions(final InputStream stream, String jobid) throws FileNotFoundException {
  78                 final FastaReader fr = new FastaReader(stream);
  79                 String protein = "";
  80                 while (fr.hasNext()) {
  81                         final FastaSequence fs = fr.next();
  82                         String seqid = fs.getId();
  83                         String seq = fs.getSequence().replaceAll("\n", "");
  84                         if (seqid.equals("jnetpred") || seqid.equals("Lupas_21") || seqid.equals("Lupas_14") || seqid.equals("Lupas_28")
  85                                         || seqid.equals("JNETSOL25") || seqid.equals("JNETSOL5") || seqid.equals("JNETSOL0") || seqid.equals("JNETCONF")
  86                                         || seqid.equals("JNETHMM") || seqid.equals("JNETPSSM") || seqid.equals("JNETCONF")) {
  87                                 predictions.add(fs);
  88                         } else {
  89                                 alignment.add(fs);
  90                                 if (seqid.equals("QUERY") || seqid.equals(jobid))
  91                                         protein = seq;
  92                         }
  93                 }
  94                 return protein;
  95         }
  96         private String parseSeqFile(final InputStream stream, String jobid) throws FileNotFoundException {
  97                 final FastaReader fr = new FastaReader(stream);
  98                 String protein = "";
  99                 final FastaSequence fs = fr.next();
 100                 protein = fs.getSequence().replaceAll("\n", "");
 101                 if (fr.hasNext()) {
 102                         // this is an aligment job...
 103                         return "alignment";
 104                 }
 105                 return protein;
 106         }
 107
 108         private String parseLogFile(final InputStream stream) throws IOException {
 109                 String out = "";
 110                 BufferedReader buffer = new BufferedReader(new InputStreamReader(stream));
 111                 String line;
 112                 while (null != (line = buffer.readLine())) {
 113                         out += line;
 114                 }
 115                 return out;
 116         }
 117
 118         private int analyseJob(String[] jobinfo) throws IOException {
 119                 alignment = new ArrayList<FastaSequence>();
 120                 predictions = new ArrayList<FastaSequence>();
 121                 boolean running = true;
 122                 boolean ConcisefileExists = false;
 123                 boolean LogfileExists = false;
 124                 JpredJob job = new JpredJob(jobinfo[jobinfo.length - 1], jobinfo[0], jobinfo[1]);
 125                 job.setIP(jobinfo[2]);
 126                 Date currDate = new Date();
 127                 String maindir = dirprefix + "/" + job.getJobID() + "/";
 128
 129                 try {
 130                         URL dirurl = new URL(maindir);
 131                         HttpURLConnection httpConnection_dirurl = (HttpURLConnection) dirurl.openConnection();
 132                         if (httpConnection_dirurl.getResponseCode() < 199 || 300 <= httpConnection_dirurl.getResponseCode()) {
 133                                 return 0;
 134                         }
 135                         URL conciseurl = new URL(maindir + job.getJobID() + ".concise.fasta");
 136                         URL logurl = new URL(maindir + "LOG");
 137                         HttpURLConnection httpConnection_conciseurl = (HttpURLConnection) conciseurl.openConnection();
 138                         HttpURLConnection httpConnection_logurl = (HttpURLConnection) logurl.openConnection();
 139                         if (199 < httpConnection_conciseurl.getResponseCode() && httpConnection_conciseurl.getResponseCode() < 300) {
 140                                 ConcisefileExists = true;
 141                                 running = false;
 142                                 try {
 143                                         job.setProtein(parsePredictions(conciseurl.openStream(), job.getJobID()));
 144                                 } catch (IOException e) {
 145                                         e.printStackTrace();
 146                                 }
 147                         } else {
 148                                 // The job still can be running of failed...
 149                                 ++countNoData;
 150                         }
 151                         if (199 < httpConnection_logurl.getResponseCode() && httpConnection_logurl.getResponseCode() < 300) {
 152                                 LogfileExists = true;
 153                                 job.setLog(parseLogFile(logurl.openStream()));
 154                         } else {
 155                                 // The job has not been started at all...
 156                                 job.setExecutionStatus("FAIL");
 157                                 job.setFinalStatus("STOPPED");
 158                                 running = false;
 159                         }
 160                         if (job.getLog().matches("(.*)TIMEOUT\\syour\\sjob\\stimed\\sout(.*)")) {
 161                                 // blast job was too long (more than 3600 secs by default)...
 162                                 job.setExecutionStatus("FAIL");
 163                                 job.setFinalStatus("TIMEDOUT");
 164                                 running = false;
 165                         } else if (job.getLog().matches("(.*)Jpred\\serror:\\sDied\\sat(.*)")) {
 166                                 // an internal Jpred error...
 167                                 job.setExecutionStatus("FAIL");
 168                                 job.setFinalStatus("JPREDERROR");
 169                                 running = false;
 170                         } else if ((currDate.getTime() - job.getEndTime()) / 1000 > 3601 && LogfileExists && !ConcisefileExists) {
 171                                 // the job was stopped with unknown reason...
 172                                 job.setExecutionStatus("FAIL");
 173                                 job.setFinalStatus("STOPPED");
 174                                 running = false;
 175                         }
 176
 177                         httpConnection_conciseurl.disconnect();
 178                         httpConnection_logurl.disconnect();
 179                 } catch (MalformedURLException e) {
 180                         e.printStackTrace();
 181                 }
 182
 183                 if (!running) {
 184                         // logging the job
 185                         job.setAlignment(alignment);
 186                         job.setPredictions(predictions);
 187                         if (job.getExecutionStatus().equals("FAIL")) {
 188                                 URL sequrl = new URL(maindir + job.getJobID() + ".seq");
 189                                 HttpURLConnection httpConnection_sequrl = (HttpURLConnection) sequrl.openConnection();
 190                                 if (199 < httpConnection_sequrl.getResponseCode() && httpConnection_sequrl.getResponseCode() < 300) {
 191                                         try {
 192                                                 job.setProtein(parseSeqFile(sequrl.openStream(), job.getJobID()));
 193                                         } catch (IOException e) {
 194                                                 e.printStackTrace();
 195                                         }
 196                                 }
 197                         }
 198                         cw.FormQueryTables(job);
 199
 200                         // archiving the job
 201                         if (archiving) {
 202                                 ArchivedJob ajob = new ArchivedJob(job.getJobID());
 203                                 String arlink = archive.createJob(job.getJobID());
 204                                 if (job.getFinalStatus().equals("OK")) {
 205                                         ajob.setArchivePath(arlink);
 206                                         ajob.copyArchiveFromWeb(maindir + job.getJobID() + ".tar.gz");
 207                                         cw.ArchiveData(job, arlink);
 208                                 } else {
 209                                         cw.ArchiveData(job, "undefined");
 210                                 }
 211                         }
 212                         return 1;
 213                 }
 214
 215                 return 0;
 216         }
 217
 218         private void ParsingOneDay(String input, String date) {
 219                 int totalcount = 0;
 220                 int countinsertions = 0;
 221                 int countinserted = 0;
 222                 int countNotanalyzed = 0;
 223                 countNoData = 0;
 224
 225                 System.out.println("Inserting jobs for " + date);
 226                 try {
 227                         URL url = new URL(input);
 228                         URLConnection conn = url.openConnection();
 229                         BufferedReader alljobs = new BufferedReader(new InputStreamReader(conn.getInputStream()));
 230                         String line;
 231
 232                         while ((line = alljobs.readLine()) != null) {
 233                                 if (line.matches(date + ":(.*)jp_[^\\s]+")) {
 234                                         totalcount++;
 235                                         String[] job = line.split("\\s+");
 236                                         String jobid = job[job.length - 1];
 237                                         if (cw.JobisNotInsterted(jobid)) {
 238                                                 countinsertions += analyseJob(job);
 239                                         } else {
 240                                                 ++countinserted;
 241                                         }
 242                                 } else {
 243                                         ++countNotanalyzed;
 244                                 }
 245                         }
 246                         alljobs.close();
 247                         System.out.println("Total number of jobs = " + totalcount);
 248                         System.out.println("   " + countinserted + " jobs inserted already");
 249                         System.out.println("   " + countNotanalyzed + " not analysed jobs");
 250                         System.out.println("   " + countNoData + " jobs without *.concise.fasta file (RUNNING or FAILED)");
 251                         System.out.println("   " + countinsertions + " new job insertions\n");
 252                 } catch (MalformedURLException e) {
 253                         e.printStackTrace();
 254                 } catch (IOException e) {
 255                         e.printStackTrace();
 256                 }
 257                 ;
 258         }
 259 };