pdr.utils.datamgmt.checksum_of(): provide buf size as opt. arg

python/nistoar/pdr/utils/datamgmt.py

@@ -79,15 +79,23 @@ def build_mime_type_map(filelist):
         update_mimetypes_from_file(out, file)
     return out
 
-def checksum_of(filepath):
+def checksum_of(filepath, bufsize: int=10240000):
     """
     return the checksum for the given file
+
+    :param str|Path filepath:  the path of the file to calculate the checksum for
+    :param int       bufsize:  the memory buffer size to use when reading the file.
+                               The default is 10 MB; multithreaded applications should 
+                               consider a smaller value.
     """
-    bfsz = 10240000   # 10 MB buffer
+    if not isinstance(bufsize, int):
+        raise TypeError("checksum_of(): bufsize arg must be an integer")
+    if bufsize < 1:
+        raise ValueError("checksum_of(): bufsize arg must be a positive integer")
     sum = hashlib.sha256()
     with open(filepath, mode='rb') as fd:
         while True:
-            buf = fd.read(bfsz)
+            buf = fd.read(bufsize)
             if not buf: break
             sum.update(buf)
     return sum.hexdigest()

python/tests/nistoar/pdr/utils/test_datamgmt.py

@@ -58,6 +58,23 @@ def test_checksum_of(self):
         self.assertEqual(utils.checksum_of(dfile), self.syssum(dfile))
         dfile = os.path.join(testdatadir2,"trial3/trial3a.json")
         self.assertEqual(utils.checksum_of(dfile), self.syssum(dfile))
+        dfile = os.path.join(testdatadir3,"3A1EE2F169DD3B8CE0531A570681DB5D1491.json")
+        self.assertEqual(utils.checksum_of(dfile), self.syssum(dfile))
+
+    def test_bufsize(self):
+        dfile = os.path.join(testdatadir3,"3A1EE2F169DD3B8CE0531A570681DB5D1491.json")
+        syssum = self.syssum(dfile)
+        self.assertEqual(utils.checksum_of(dfile, 10240), syssum)
+        self.assertEqual(utils.checksum_of(dfile, 1024), syssum)
+        self.assertEqual(utils.checksum_of(dfile, 10), syssum)
+        self.assertEqual(utils.checksum_of(dfile, 1), syssum)
+
+        with self.assertRaises(ValueError):
+            utils.checksum_of(dfile, 0)
+        with self.assertRaises(ValueError):
+            utils.checksum_of(dfile, -20)
+        with self.assertRaises(TypeError):
+            utils.checksum_of(dfile, "1024")
 
     def syssum(self, filepath):
         cmd = ["sha256sum", filepath]