Add '--checksum' attribute to maintain the exported model data

readme.md

@@ -233,7 +233,7 @@ First of all have a quick view on the available arguments:
 $ python -m sklearn_porter [-h] --input <PICKLE_FILE> [--output <DEST_DIR>] \
                            [--class_name <CLASS_NAME>] [--method_name <METHOD_NAME>] \
                            [--c] [--java] [--js] [--go] [--php] [--ruby] \
-                           [--export] [--pipe]
+                           [--export] [--checksum] [--pipe]
 ```
 
 The following example shows how you can save an trained estimator to the [pickle format](http://scikit-learn.org/stable/modules/model_persistence.html#persistence-example):

sklearn_porter/__main__.py

@@ -47,6 +47,12 @@ def parse_args(args):
         default=False,
         action='store_true',
         help='Whether to export the model data or not.')
+    optional.add_argument(
+        '--checksum',
+        required=False,
+        default=False,
+        action='store_true',
+        help='Whether to append the checksum to the filename or not.')
     optional.add_argument(
         '--pipe', '-p',
         required=False,
@@ -106,13 +112,16 @@ def main():
 
     # Port estimator:
     try:
-        porter = Porter(estimator, language=language)
         class_name = args.get('class_name')
         method_name = args.get('method_name')
+        with_export = bool(args.get('export'))
+        with_checksum = bool(args.get('checksum'))
+        porter = Porter(estimator, language=language)
         output = porter.export(class_name=class_name,
                                method_name=method_name,
                                export_dir=dest_dir,
-                               export_data=bool(args.get('export')),
+                               export_data=with_export,
+                               export_append_checksum=with_checksum,
                                details=True)
     except Exception as e:
         sys.exit('Error: {}'.format(str(e)))

sklearn_porter/estimator/classifier/AdaBoostClassifier/__init__.py

@@ -85,7 +85,7 @@ def __init__(self, estimator, target_language='java',
 
     def export(self, class_name, method_name,
                export_data=False, export_dir='.', export_filename='data.json',
-               embed_data=True, **kwargs):
+               export_append_checksum=False, embed_data=True, **kwargs):
         """
         Port a trained estimator to the syntax of a chosen programming language.
 
@@ -101,6 +101,8 @@ def export(self, class_name, method_name,
             The directory where the model data should be saved.
         :param export_filename : string
             The filename of the exported model data.
+        :param export_append_checksum : bool
+            Whether to append the checksum to the filename or not.
         :param embed_data : bool
             Whether the model data should be embedded in the template or not.
 
@@ -128,7 +130,8 @@ def export(self, class_name, method_name,
         if self.target_method == 'predict':
             # Exported:
             if export_data and os.path.isdir(export_dir):
-                self.export_data(export_dir, export_filename)
+                self.export_data(export_dir, export_filename,
+                                 export_append_checksum)
                 return self.predict('exported')
             # Embedded:
             return self.predict('embedded')
@@ -158,7 +161,7 @@ def predict(self, temp_type):
             meth = self.create_embedded_meth()
             return self.create_embedded_class(meth)
 
-    def export_data(self, directory, filename):
+    def export_data(self, directory, filename, with_md5_hash=False):
         """
         Save model data in a JSON file.
 
@@ -168,6 +171,8 @@ def export_data(self, directory, filename):
             The directory.
         :param filename : string
             The filename.
+        :param with_md5_hash : bool
+            Whether to append the checksum to the filename or not.
         """
         model_data = []
         for est in self.estimators:
@@ -179,9 +184,14 @@ def export_data(self, directory, filename):
                 'indices': est.tree_.feature.tolist()
             })
         encoder.FLOAT_REPR = lambda o: self.repr(o)
+        json_data = json.dumps(model_data, sort_keys=True)
+        if with_md5_hash:
+            import hashlib
+            json_hash = hashlib.md5(json_data).hexdigest()
+            filename = filename.split('.json')[0] + '_' + json_hash + '.json'
         path = os.path.join(directory, filename)
         with open(path, 'w') as fp:
-            json.dump(model_data, fp)
+            fp.write(json_data)
 
     def create_branches(self, left_nodes, right_nodes, threshold,
                         value, features, node, depth, init=False):

sklearn_porter/estimator/classifier/BernoulliNB/__init__.py

@@ -59,7 +59,7 @@ def __init__(self, estimator, target_language='java',
 
     def export(self, class_name, method_name,
                export_data=False, export_dir='.', export_filename='data.json',
-               **kwargs):
+               export_append_checksum=False, **kwargs):
         """
         Port a trained estimator to the syntax of a chosen programming language.
 
@@ -75,6 +75,8 @@ def export(self, class_name, method_name,
             The directory where the model data should be saved.
         :param export_filename : string
             The filename of the exported model data.
+        :param export_append_checksum : bool
+            Whether to append the checksum to the filename or not.
 
         Returns
         -------
@@ -127,7 +129,8 @@ def export(self, class_name, method_name,
         if self.target_method == 'predict':
             # Exported:
             if export_data and os.path.isdir(export_dir):
-                self.export_data(export_dir, export_filename)
+                self.export_data(export_dir, export_filename,
+                                 export_append_checksum)
                 return self.predict('exported')
             # Separated:
             return self.predict('separated')
@@ -156,7 +159,7 @@ def predict(self, temp_type):
         method = self.create_method()
         return self.create_class(method)
 
-    def export_data(self, directory, filename):
+    def export_data(self, directory, filename, with_md5_hash=False):
         """
         Save model data in a JSON file.
 
@@ -166,6 +169,8 @@ def export_data(self, directory, filename):
             The directory.
         :param filename : string
             The filename.
+        :param with_md5_hash : bool
+            Whether to append the checksum to the filename or not.
         """
         neg_prob = np.log(1 - np.exp(self.estimator.feature_log_prob_))
         delta_probs = (self.estimator.feature_log_prob_ - neg_prob).T
@@ -175,9 +180,14 @@ def export_data(self, directory, filename):
             'delProbs': delta_probs.tolist()
         }
         encoder.FLOAT_REPR = lambda o: self.repr(o)
+        json_data = json.dumps(model_data, sort_keys=True)
+        if with_md5_hash:
+            import hashlib
+            json_hash = hashlib.md5(json_data).hexdigest()
+            filename = filename.split('.json')[0] + '_' + json_hash + '.json'
         path = os.path.join(directory, filename)
         with open(path, 'w') as fp:
-            json.dump(model_data, fp)
+            fp.write(json_data)
 
     def create_method(self):
         """

sklearn_porter/estimator/classifier/DecisionTreeClassifier/__init__.py

@@ -111,7 +111,7 @@ def __init__(self, estimator, target_language='java',
 
     def export(self, class_name, method_name,
                export_data=False, export_dir='.', export_filename='data.json',
-               embed_data=False, **kwargs):
+               export_append_checksum=False, embed_data=False, **kwargs):
         """
         Port a trained estimator to the syntax of a chosen programming language.
 
@@ -127,6 +127,8 @@ def export(self, class_name, method_name,
             The directory where the model data should be saved.
         :param export_filename : string
             The filename of the exported model data.
+        :param export_append_checksum : bool
+            Whether to append the checksum to the filename or not.
         :param embed_data : bool
             Whether the model data should be embedded in the template or not.
 
@@ -185,15 +187,16 @@ def export(self, class_name, method_name,
         if self.target_method == 'predict':
             # Exported:
             if export_data and os.path.isdir(export_dir):
-                self.export_data(export_dir, export_filename)
+                self.export_data(export_dir, export_filename,
+                                 export_append_checksum)
                 return self.predict('exported')
             # Embedded:
             if embed_data:
                 return self.predict('embedded')
             # Separated:
             return self.predict('separated')
 
-    def export_data(self, directory, filename):
+    def export_data(self, directory, filename, with_md5_hash=False):
         """
         Save model data in a JSON file.
 
@@ -203,6 +206,8 @@ def export_data(self, directory, filename):
             The directory.
         :param filename : string
             The filename.
+        :param with_md5_hash : bool
+            Whether to append the checksum to the filename or not.
         """
         model_data = {
             'leftChilds': self.estimator.tree_.children_left.tolist(),
@@ -212,9 +217,14 @@ def export_data(self, directory, filename):
             'classes': [c[0] for c in self.estimator.tree_.value.tolist()]
         }
         encoder.FLOAT_REPR = lambda o: self.repr(o)
+        json_data = json.dumps(model_data, sort_keys=True)
+        if with_md5_hash:
+            import hashlib
+            json_hash = hashlib.md5(json_data).hexdigest()
+            filename = filename.split('.json')[0] + '_' + json_hash + '.json'
         path = os.path.join(directory, filename)
         with open(path, 'w') as fp:
-            json.dump(model_data, fp)
+            fp.write(json_data)
 
     def predict(self, temp_type='separated'):
         """

sklearn_porter/estimator/classifier/GaussianNB/__init__.py

@@ -58,7 +58,7 @@ def __init__(self, estimator, target_language='java',
 
     def export(self, class_name, method_name,
                export_data=False, export_dir='.', export_filename='data.json',
-               **kwargs):
+               export_append_checksum=False, **kwargs):
         """
         Port a trained estimator to the syntax of a chosen programming language.
 
@@ -74,6 +74,8 @@ def export(self, class_name, method_name,
             The directory where the model data should be saved.
         :param export_filename : string
             The filename of the exported model data.
+        :param export_append_checksum : bool
+            Whether to append the checksum to the filename or not.
 
         Returns
         -------
@@ -124,7 +126,8 @@ def export(self, class_name, method_name,
         if self.target_method == 'predict':
             # Exported:
             if export_data and os.path.isdir(export_dir):
-                self.export_data(export_dir, export_filename)
+                self.export_data(export_dir, export_filename,
+                                 export_append_checksum)
                 return self.predict('exported')
             # Separated:
             return self.predict('separated')
@@ -152,7 +155,7 @@ def predict(self, temp_type):
         method = self.create_method()
         return self.create_class(method)
 
-    def export_data(self, directory, filename):
+    def export_data(self, directory, filename, with_md5_hash=False):
         """
         Save model data in a JSON file.
 
@@ -162,16 +165,23 @@ def export_data(self, directory, filename):
             The directory.
         :param filename : string
             The filename.
+        :param with_md5_hash : bool
+            Whether to append the checksum to the filename or not.
         """
         model_data = {
             'priors': self.estimator.class_prior_.tolist(),
             'sigmas': self.estimator.sigma_.tolist(),
             'thetas': self.estimator.theta_.tolist()
         }
         encoder.FLOAT_REPR = lambda o: self.repr(o)
+        json_data = json.dumps(model_data, sort_keys=True)
+        if with_md5_hash:
+            import hashlib
+            json_hash = hashlib.md5(json_data).hexdigest()
+            filename = filename.split('.json')[0] + '_' + json_hash + '.json'
         path = os.path.join(directory, filename)
         with open(path, 'w') as fp:
-            json.dump(model_data, fp)
+            fp.write(json_data)
 
     def create_method(self):
         """

sklearn_porter/estimator/classifier/KNeighborsClassifier/__init__.py

@@ -62,7 +62,7 @@ def __init__(self, estimator, target_language='java',
 
     def export(self, class_name, method_name,
                export_data=False, export_dir='.', export_filename='data.json',
-               **kwargs):
+               export_append_checksum=False, **kwargs):
         """
         Port a trained estimator to the syntax of a chosen programming language.
 
@@ -78,6 +78,8 @@ def export(self, class_name, method_name,
             The directory where the model data should be saved.
         :param export_filename : string
             The filename of the exported model data.
+        :param export_append_checksum : bool
+            Whether to append the checksum to the filename or not.
 
         Returns
         -------
@@ -110,12 +112,13 @@ def export(self, class_name, method_name,
         if self.target_method == 'predict':
             # Exported:
             if export_data and os.path.isdir(export_dir):
-                self.export_data(export_dir, export_filename)
+                self.export_data(export_dir, export_filename,
+                                 export_append_checksum)
                 return self.predict('exported')
             # Separated:
             return self.predict('separated')
 
-    def export_data(self, directory, filename):
+    def export_data(self, directory, filename, with_md5_hash=False):
         """
         Save model data in a JSON file.
 
@@ -125,6 +128,8 @@ def export_data(self, directory, filename):
             The directory.
         :param filename : string
             The filename.
+        :param with_md5_hash : bool
+            Whether to append the checksum to the filename or not.
         """
         model_data = {
             'X': self.estimator._fit_X.tolist(),  # pylint: disable=W0212
@@ -134,9 +139,14 @@ def export_data(self, directory, filename):
             'power': self.power_param
         }
         encoder.FLOAT_REPR = lambda o: self.repr(o)
+        json_data = json.dumps(model_data, sort_keys=True)
+        if with_md5_hash:
+            import hashlib
+            json_hash = hashlib.md5(json_data).hexdigest()
+            filename = filename.split('.json')[0] + '_' + json_hash + '.json'
         path = os.path.join(directory, filename)
         with open(path, 'w') as fp:
-            json.dump(model_data, fp)
+            fp.write(json_data)
 
     def predict(self, temp_type):
         """