check for new filenames before merge/subset

python logic written in separate library
- looks for lines in datatype sheet which duplicate a previous datatype (new)
- ignore normal sample files (which are ignored by importer)
perform check from subset-impact-data.sh and merge.py and exit on noticing new filename patterns.

import-scripts/merge.py

@@ -7,6 +7,7 @@
 import shutil
 import re
 import csv
+import updated_filename_for_datatype_test
 
 # ------------------------------------------------------------------------------
 # globals
@@ -814,6 +815,15 @@ def organize_files(studies, file_types, merge_clinical):
             else:
                 file_types[SUPP_DATA].append(study_file)
 
+def exit_if_unsupported_filenames_detected(study_paths):
+    unsupported_filenames_detected = False
+    for study_path in study_paths:
+        if updated_filename_for_datatype_test.directory_uses_updated_filenames(study_path):
+            print("unsupported filenames present in study directory '%s' - unable to merge" % study_path)
+            unsupported_filenames_detected = True
+    if unsupported_filenames_detected:
+        sys.exit(2)
+
 def usage():
     print >> OUTPUT_FILE, 'merge.py --subset [/path/to/subset] --output-directory [/path/to/output] --study-id [study id] --cancer-type [cancer type] --merge-clinical [true/false] --exclude-supplemental-data [true/false] --excluded-samples [/path/to/exclude_list] <path/to/study path/to/study ...>'
 
@@ -919,6 +929,9 @@ def main():
     else:
         exclude_supp_data = True
 
+    # test input study paths for new filenames
+    exit_if_unsupported_filenames_detected(args)
+
     # get all the filenames
     organize_files(args, file_types, merge_clinical)
 

import-scripts/subset-impact-data.sh

@@ -55,6 +55,26 @@ if [ -z $PORTAL_SCRIPTS_DIRECTORY ]; then
 fi
 echo -e "\tPORTAL_SCRIPTS_DIRECTORY="$PORTAL_SCRIPTS_DIRECTORY
 
+# check for presence of new filename pattens (not yet supported) and fail if present
+
+PYTHON_FILENAME_CHECK_CALL="import sys
+import updated_filename_for_datatype_test
+sys.exit(updated_filename_for_datatype_test.directory_uses_updated_filenames(\"$INPUT_DIRECTORY\"))
+"
+if ! python -c "$PYTHON_FILENAME_CHECK_CALL" ; then
+    echo "unsupported filenames present - skipping subset" >&2
+    exit 1
+fi
+
+PYTHON_FILENAME_CHECK_CALL="import sys
+import updated_filename_for_datatype_test
+sys.exit(updated_filename_for_datatype_test.directory_uses_updated_filenames(\"$INPUT_DIRECTORY\"))
+"
+if ! python -c "$PYTHON_FILENAME_CHECK_CALL" ; then
+    echo "unsupported filenames present - skipping subset" >&2
+    exit 1
+fi
+
 # status flags
 GEN_SUBSET_LIST_FAILURE=0
 MERGE_SCRIPT_FAILURE=0