Compression options for AlignAndFocusPowderFromFiles

Framework/PythonInterface/plugins/algorithms/AlignAndFocusPowderFromFiles.py

@@ -78,6 +78,14 @@
 PROPS_FOR_PD_CHARACTER = ["FrequencyLogNames", "WaveLengthLogNames"]
 
 
+def determineCompression(filename, compression, chunking, absorption):
+    if compression == 0.0 or chunking > 0.0 or absorption:
+        return False
+    sizeGiB = os.path.getsize(filename) / 1024.0 / 1024.0 / 1024.0
+    if sizeGiB > compression:
+        return True
+
+
 def determineChunking(filename, chunkSize):
     # chunkSize=0 signifies that the user wants to read the whole file
     if chunkSize == 0.0:
@@ -147,6 +155,7 @@ def summary(self):
     def PyInit(self):
         self.declareProperty(MultipleFileProperty(name="Filename", extensions=EXTENSIONS_NXS), "Files to combine in reduction")
         self.declareProperty("MaxChunkSize", 0.0, "Specify maximum Gbytes of file to read in one chunk.  Default is whole file.")
+        self.declareProperty("MinSizeCompressOnLoad", 0.0, "Specify the file size in GB to use compression")
         self.declareProperty("FilterBadPulses", 0.0, doc="Filter out events measured while proton charge is more than 5% below average")
 
         self.declareProperty(
@@ -213,6 +222,11 @@ def __createLoader(self, filename, wkspname, progstart=None, progstop=None, skip
         loader.setPropertyValue("Filename", filename)
         loader.setPropertyValue("OutputWorkspace", wkspname)
 
+        if self.do_compression:
+            self.kwargs = self.__getAlignAndFocusArgs()
+            loader.setPropertyValue("CompressTolerance", str(self.kwargs["CompressTolerance"]))
+            loader.setPropertyValue("CompressBinningMode", self.kwargs["CompressBinningMode"])
+
         if skipLoadingLogs:
             if self.__loaderName != "LoadEventNexus":
                 raise RuntimeError("Cannot set LoadLogs=False in {}".format(self.__loaderName))
@@ -664,6 +678,7 @@ def PyExec(self):
         self.__loaderName = "Load"  # set the loader to be generic on first load
         self.filterBadPulses = self.getProperty("FilterBadPulses").value
         self.chunkSize = self.getProperty("MaxChunkSize").value
+        self.compression_threshold = self.getProperty("MinSizeCompressOnLoad").value
         self.absorption = self.getProperty("AbsorptionWorkspace").value
         self.charac = self.getProperty("Characterizations").value
         self.useCaching = len(self.getProperty("CacheDir").value) > 0
@@ -677,6 +692,10 @@ def PyExec(self):
         # empty string means it is not used
         finalunfocusname = self.getPropertyValue("UnfocussedWorkspace")
 
+        # determing compression
+        self.do_compression = determineCompression(
+            filename=self._filenames[0], compression=self.compression_threshold, chunking=self.chunkSize, absorption=self.absorption
+        )
         if self.useCaching:
             # unfocus check only matters if caching is requested
             if finalunfocusname != "":

Testing/SystemTests/tests/framework/AlignAndFocusPowderFromFilesTest.py

@@ -213,6 +213,32 @@ def validate(self):
         return ("with_chunks", "no_chunks")
 
 
+class CompressedCompare(systemtesting.MantidSystemTest):
+    # this test is very similar to SNAPRedux.Simple
+
+    def requiredMemoryMB(self):
+        return 24 * 1024  # GiB
+
+    def runTest(self):
+        GRP_WKSP = "SNAP_compress_params"
+
+        # 11MB file
+        kwargs = {"Filename": "SNAP_45874", "Params": (0.5, -0.004, 7), "GroupingWorkspace": GRP_WKSP}
+
+        # create grouping for two output spectra
+        CreateGroupingWorkspace(InstrumentFilename="SNAP_Definition.xml", GroupDetectorsBy="Group", OutputWorkspace=GRP_WKSP)
+
+        AlignAndFocusPowderFromFiles(
+            OutputWorkspace="compress1", MaxChunkSize=0.0, CompressTolerance=1e-2, MinSizeCompressOnLoad=1e-14, **kwargs
+        )
+
+    def validateMethod(self):
+        return None
+
+    def validate(self):
+        return None
+
+
 class UseCache(systemtesting.MantidSystemTest):
     cal_file = "PG3_FERNS_d4832_2011_08_24.cal"
     char_file = "PG3_characterization_2012_02_23-HR-ILL.txt"