print status of finished jobs to a log-file (towards #26)

cptools2/generate_scripts.py

@@ -162,9 +162,9 @@ def make_qsub_scripts(commands_location, commands_count_dict, logfile_location):
     analysis_script.loop_through_file(cmd_path["cp_commands"])
     analysis_loc = os.path.join(commands_location,
                                 "{}_analysis_script.sh".format(time_now))
+    analysis_script.template += make_logfile_text(logfile_location, job_hex)
     print("** saving analysis submission script at '{}'".format(analysis_loc))
     analysis_script.save(analysis_loc)
-
     destaging_script = BodgeScript(
         name="destaging_{}".format(job_hex),
         memory="1G",
@@ -183,6 +183,24 @@ def make_qsub_scripts(commands_location, commands_count_dict, logfile_location):
     utils.make_executable(submit_script)
 
 
+def make_logfile_text(logfile_location, job_file):
+    text = """
+    # get the exit code from the cellprofiler job
+    RETURN_VAL=$?
+
+    if [[ $RETURN_VAL == 0]]; then
+        RETURN_STATUS="Finished"
+    else
+        RETURN_STATUS="Failed with error code: $RETURN_VAL
+    fi
+
+    LOG_FILE_LOC={logfile_location}/{job_file}.log
+    echo "`date +"%Y%m%d %H:%M"`  $JOB_ID  $SGE_TASK_ID  $RETURN_STATUS" >> $LOG_FILE_LOC
+    """.format(logfile_location=logfile_location,
+               job_file=job_file)
+    return textwrap.dedent(text)
+
+
 def load_venv_store():
     """
     Load the virtual environment yaml file that details each user's path to
@@ -237,6 +255,13 @@ def make_submit_script(commands_location, job_date):
     # create text for a shell script that qsub's the scripts
     output = """
              #!/bin/sh
+             
+             # This script submits the staging, analysis and destaging
+             # scripts in the correct order
+
+             # NOTE: run this as a shell script, NOT a submission script
+             # so either call `./name_of_script` or `bash name_of_script`
+
              qsub {staging_script}
              qsub {analysis_script}
              qsub {destaging_script}

cptools2/job.py

@@ -146,6 +146,7 @@ def create_commands(self, pipeline, location, commands_location, job_size):
             file path to location in which to store the stage, analysis and
             destage commands.
         """
+        print("** creating image list")
         if self.has_loaddata is False:
             self._create_loaddata(job_size)
         cp_commands, rsync_commands, rm_commands = [], [], []

cptools2/loaddata.py

@@ -74,15 +74,15 @@ def cast_dataframe(dataframe, check_nan=True):
     # rename FileName columns from 1, 2... to FileName_W1, FileName_W2 ...
     columns = {}
     for i in range(1, n_channels+1):
-        columns[i] = "FileName_W" + str(i)
+        columns[i] = "FileName_W{0}".format(str(i))
     wide_df.rename(columns=columns, inplace=True)
     # duplicate PathName for each channel
     for i in range(1, n_channels+1):
         wide_df["PathName_W" + str(i)] = wide_df.path
     wide_df.drop(["path"], axis=1, inplace=True)
     if check_nan is True:
         if utils.any_nan_values(dataframe):
-            raise Warning("dataframe contains missing values")
+            raise LoadDataError("dataframe contains missing values")
     return wide_df