switch to json encoder that provides a better storage size estimate

requirements.txt

@@ -4,7 +4,6 @@ scipy
 pandas >=2.2.0
 pint >=0.24.3
 pint-pandas ==0.6.2
-pymongo >= 4.7.3
 fireworks >=2.0.4
 pyyaml
 dill

setup.cfg

@@ -49,7 +49,6 @@ install_requires =
     pandas >=2.2.0
     pint >=0.24.3
     pint-pandas ==0.7.1
-    pymongo >= 4.7.3
     fireworks >=2.0.4
     pyyaml
     dill

src/virtmat/language/utilities/serializable.py

 """serialization/deserialization code"""
-from dataclasses import dataclass
 import typing
+from dataclasses import dataclass
+from json import JSONEncoder
+from itertools import islice
 import numpy
 import pandas
 import pint_pandas
-from bson import json_util
 from fireworks.utilities.fw_serializers import FWSerializable
 from fireworks.utilities.fw_serializers import serialize_fw
 from fireworks.utilities.fw_serializers import recursive_serialize
@@ -45,6 +46,18 @@ def versioned_serialize(func):
     return decorator
 
 
+def get_json_size(obj, max_size):
+    """compute JSON size in bytes of a JSON serializable object up to max_size"""
+    gen = JSONEncoder().iterencode(obj)
+    chunk_size = 1024
+    json_size = 0
+    next_chunk = len(''.join(islice(gen, chunk_size)).encode())
+    while next_chunk and json_size < max_size:
+        json_size += next_chunk
+        next_chunk = len(''.join(islice(gen, chunk_size)).encode())
+    return json_size
+
+
 @dataclass
 class FWDataObject(FWSerializable):
     """top-level FWSerializable dataclass to hold any FWSerializable objects"""
@@ -63,7 +76,7 @@ class FWDataObject(FWSerializable):
         if self.datastore is None:
             dct = recursive_dict(self.value)
             b_thres = ioops.DATASTORE_CONFIG['inline-threshold']
-            b_size = json_util.get_size(dct, b_thres)
+            b_size = get_json_size(dct, b_thres)
             logger.debug('%s: data type: %s', f_name, type(self.value))
             logger.debug('%s: data size [B]: %s', f_name, b_size)
             logger.debug('%s: inline-threshold [B]: %s', f_name, b_thres)