+/* Copyright (c) 2011 Peter Troshin\r
+ * \r
+ * JAva Bioinformatics Analysis Web Services (JABAWS) @version: 2.0 \r
+ * \r
+ * This library is free software; you can redistribute it and/or modify it under the terms of the\r
+ * Apache License version 2 as published by the Apache Software Foundation\r
+ * \r
+ * This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without\r
+ * even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Apache \r
+ * License for more details.\r
+ * \r
+ * A copy of the license is in apache_license.txt. It is also available here:\r
+ * @see: http://www.apache.org/licenses/LICENSE-2.0.txt\r
+ * \r
+ * Any republication or derived work distributed in source code form\r
+ * must include this copyright and license notice.\r
+ */\r
package compbio.stat.collector;\r
\r
import java.io.File;\r
import org.apache.log4j.Logger;\r
\r
import compbio.engine.client.Executable;\r
+import compbio.engine.client.PathValidator;\r
+import compbio.engine.client.SkeletalExecutable;\r
import compbio.metadata.JobStatus;\r
import compbio.util.FileUtil;\r
import compbio.ws.client.Services;\r
\r
static SimpleDateFormat DF = new SimpleDateFormat("dd/MM/yyyy hh:mm:ss");\r
\r
+ final private File workDirectory;\r
final private List<JobStat> stats;\r
/**\r
* Consider the job that has been working for longer than timeOutInHours\r
public ExecutionStatCollector(String workDirectory, int timeOutInHours) {\r
log.info("Starting stat collector for directory: " + workDirectory);\r
log.info("Maximum allowed runtime(h): " + timeOutInHours);\r
- File[] files = FileUtil.getFiles(workDirectory, directories);\r
+ if (!PathValidator.isValidDirectory(workDirectory)) {\r
+ throw new IllegalArgumentException("workDirectory '"\r
+ + workDirectory + "' does not exist!");\r
+ }\r
+ this.workDirectory = new File(workDirectory);\r
stats = new ArrayList<JobStat>();\r
- assert timeOutInHours > 0;\r
- this.timeOutInHours = timeOutInHours;\r
- for (File file : files) {\r
- JobDirectory jd = new JobDirectory(file);\r
- JobStat jstat = jd.getJobStat();\r
- // Do not record stats on the job that has not completed yet\r
- if (hasCompleted(jd)) {\r
- stats.add(jstat);\r
- } else {\r
- log.debug("Skipping the job: " + jstat);\r
- log.debug("As it has not completed yet");\r
- }\r
- // System.out.println(jd.getJobStat().getJobReportTabulated());\r
+ if (timeOutInHours <= 0) {\r
+ throw new IllegalArgumentException(\r
+ "Timeout value must be greater than 0! Given value: "\r
+ + timeOutInHours);\r
}\r
+ this.timeOutInHours = timeOutInHours;\r
}\r
\r
boolean hasCompleted(JobDirectory jd) {\r
return ((System.currentTimeMillis() - jd.jobdir.lastModified()) / (1000 * 60 * 60)) > timeOutInHours;\r
}\r
\r
- public StatProcessor getStats() {\r
+ /*\r
+ * Make sure that collectStatistics methods was called prior to calling\r
+ * this! TODO consider running collectStatistics from here on the first call\r
+ */\r
+ StatProcessor getStats() {\r
+ if (stats.isEmpty()) {\r
+ log.info("Please make sure collectStatistics method was called prior to calling getStats()!");\r
+ }\r
return new StatProcessor(stats);\r
}\r
\r
- public void writeStatToDB() throws SQLException {\r
+ void writeStatToDB() throws SQLException {\r
Set<JobStat> rjobs = new HashSet<JobStat>(stats);\r
StatDB statdb = new StatDB();\r
log.debug("Removing records that has already been recorded");\r
*/\r
\r
/**\r
- * \r
- * @param args\r
- * @throws IOException\r
- * @throws SQLException\r
+ * Not in use\r
*/\r
public static void main(String[] args) throws IOException, SQLException {\r
\r
File jobdir;\r
Map<String, File> files = new HashMap<String, File>();\r
\r
- public JobDirectory(File directory) {\r
+ JobDirectory(File directory) {\r
this.jobdir = directory;\r
for (File f : jobdir.listFiles()) {\r
files.put(f.getName(), f);\r
}\r
}\r
\r
- public boolean hasStatus(JobStatus status) {\r
+ boolean hasStatus(JobStatus status) {\r
return files.containsKey(status.toString());\r
}\r
\r
if (startfile == null) {\r
startfile = files.get(JobStatus.SUBMITTED.toString());\r
}\r
- if (startfile != null) {\r
+ try {\r
+ if (startfile != null) {\r
+ String start = FileUtil.readFileToString(startfile);\r
+ starttime = Long.parseLong(start.trim());\r
+ }\r
+ } catch (IOException ignore) {\r
+ log.warn(\r
+ "IOException while reading STARTED status file! Ignoring...",\r
+ ignore);\r
+ // fall back\r
+ starttime = startfile.lastModified();\r
+ } catch (NumberFormatException ignore) {\r
+ log.warn(\r
+ "NumberFormatException while reading STARTED status file! Ignoring...",\r
+ ignore);\r
+ // fall back\r
starttime = startfile.lastModified();\r
- /*\r
- * String start = FileUtil.readFileToString(startfile);\r
- * starttime = Long.parseLong(start.trim());\r
- */\r
}\r
+\r
return starttime;\r
}\r
\r
clustjobId = FileUtil.readFileToString(jobid);\r
}\r
} catch (IOException ioe) {\r
- ioe.printStackTrace();\r
- // TODO LOG\r
+ log.error(\r
+ "IO Exception while reading the content of JOBID file for job "\r
+ + jobid, ioe);\r
}\r
return clustjobId.trim();\r
}\r
long ftime = UNDEFINED;\r
File finished = files.get(JobStatus.FINISHED.toString());\r
if (finished != null) {\r
- ftime = finished.lastModified();\r
- /*\r
- * String start = FileUtil.readFileToString(finished); ftime =\r
- * Long.parseLong(start.trim());\r
- */\r
- // System.out.println("f " + ftime);\r
+ try {\r
+ if (finished != null) {\r
+ String start = FileUtil.readFileToString(finished);\r
+ ftime = Long.parseLong(start.trim());\r
+ }\r
+ } catch (IOException ignore) {\r
+ log.warn(\r
+ "IOException while reading FINISHED status file! Ignoring...",\r
+ ignore);\r
+ // fall back\r
+ ftime = finished.lastModified();\r
+ } catch (NumberFormatException ignore) {\r
+ log.warn(\r
+ "NumberFormatException while reading FINISHED status file! Ignoring...",\r
+ ignore);\r
+ // fall back\r
+ ftime = finished.lastModified();\r
+ }\r
}\r
- /*\r
- * } catch (IOException e) { log.log(Level.WARN,\r
- * "Cannot parse finished time: " + e.getMessage(), e); } catch\r
- * (NumberFormatException e) { log.log(Level.WARN,\r
- * "Cannot parse finished time: " + e.getMessage(), e); }\r
- */\r
return ftime;\r
}\r
\r
return Services.getServiceByJobDirectory(jobdir);\r
}\r
\r
- // Mafft, Muscle, Tcoffee, Clustal task:fasta.in result:fasta.out\r
- // Probcons task:fasta.in result:alignment.out\r
- /*\r
- * TODO replace with Universal names for WS!\r
- */\r
long getResultSize() {\r
Class<? extends Executable<?>> name = Services\r
.getRunnerByJobDirectory(jobdir);\r
\r
File f = null;\r
- if (name.getSimpleName().equalsIgnoreCase("Probcons")) {\r
- f = files.get("alignment.out");\r
- } else if (name.getSimpleName().equalsIgnoreCase("ClustalW")) {\r
- f = files.get("output.txt");\r
+ if (name.getSimpleName().equalsIgnoreCase("IUPred")) {\r
+ f = files.get("out.glob");\r
+ if (f == null)\r
+ f = files.get("out.short");\r
+ if (f == null)\r
+ f = files.get("out.long");\r
} else {\r
- f = files.get("fasta.out");\r
+ f = files.get(SkeletalExecutable.OUTPUT);\r
}\r
if (f != null) {\r
return f.length();\r
}\r
\r
long getInputSize() {\r
- File input = files.get("fasta.in");\r
+ Class<? extends Executable<?>> name = Services\r
+ .getRunnerByJobDirectory(jobdir);\r
+\r
+ File input = files.get(SkeletalExecutable.INPUT);\r
if (input != null) {\r
return input.length();\r
}\r
return false;\r
return true;\r
}\r
+ }\r
\r
+ void collectStatistics() {\r
+ File[] files = workDirectory.listFiles(directories);\r
+ for (File file : files) {\r
+ JobDirectory jd = new JobDirectory(file);\r
+ JobStat jstat = jd.getJobStat();\r
+ // Do not record stats on the job that has not completed yet\r
+ if (hasCompleted(jd)) {\r
+ stats.add(jstat);\r
+ } else {\r
+ log.debug("Skipping the job: " + jstat);\r
+ log.debug("As it has not completed yet");\r
+ }\r
+ // System.out.println(jd.getJobStat().getJobReportTabulated());\r
+ }\r
}\r
\r
@Override\r
public void run() {\r
log.info("Started updating statistics at " + new Date());\r
+ log.info("For directory: " + workDirectory.getAbsolutePath());\r
+\r
+ collectStatistics();\r
\r
StatProcessor local_stats = getStats();\r
log.info("Found " + local_stats.getJobNumber() + " jobs!");\r
}\r
log.info("Finished updating statistics at " + new Date());\r
}\r
-\r
}\r