added deserialization from json

2 files changed, 48 insertions, 19 deletions

diff --git a/sec_certs/certificate.py b/sec_certs/certificate.py
index 91b9fb97..9ec073c2 100644
--- a/sec_certs/certificate.py
+++ b/sec_certs/certificate.py
@@ -74,21 +74,29 @@ class CommonCriteriaCert(Certificate):
         def to_dict(self):
             return self.__dict__
 
+        @classmethod
+        def from_dict(cls, dct):
+            return cls(*tuple(dct.values()))
+
     @dataclass(eq=True, frozen=True)
     class ProtectionProfile:
         """
         Object for holding protection profiles.
         """
-        name: str
-        link: Optional[str]
+        pp_name: str
+        pp_link: Optional[str]
 
         def __post_init__(self):
-            super().__setattr__('name', helpers.sanitize_string(self.name))
-            super().__setattr__('link', helpers.sanitize_link(self.link))
+            super().__setattr__('pp_name', helpers.sanitize_string(self.pp_name))
+            super().__setattr__('pp_link', helpers.sanitize_link(self.pp_link))
 
         def to_dict(self):
             return self.__dict__
 
+        @classmethod
+        def from_dict(cls, dct):
+            return cls(*tuple(dct.values()))
+
     def __init__(self, category: str, name: str, manufacturer: str, scheme: str,
                  security_level: Union[str, set], not_valid_before: date,
                  not_valid_after: date, report_link: str, st_link: str, src: str, cert_link: Optional[str],
@@ -110,7 +118,7 @@ class CommonCriteriaCert(Certificate):
         self.cert_link = helpers.sanitize_link(cert_link)
         self.manufacturer_web = helpers.sanitize_link(manufacturer_web)
         self.protection_profiles = protection_profiles
-        self.maintainances = maintainance_updates
+        self.maintainance_updates = maintainance_updates
 
     @property
     def dgst(self) -> str:
@@ -133,10 +141,9 @@ class CommonCriteriaCert(Certificate):
                 setattr(self, att, getattr(other, att))
             elif self.src == 'csv' and other.src == 'html' and att == 'protection_profiles':
                 setattr(self, att, getattr(other, att))
-            elif att == 'maintainances':
-                # TODO Fix me: This is a simplification. Basically take the longer list of maintainances as a ground truth.
-                if len(getattr(self, att)) < len(getattr(other, att)):
-                    setattr(self, att, getattr(other, att))
+            elif self.src == 'csv' and other.src == 'html' and att == 'maintainance_updates':
+                # TODO Fix me: This is a simplification. At the moment html contains more reliable info
+                setattr(self, att, getattr(other, att))
             elif att == 'src':
                 pass  # This is expected
             else:
@@ -150,8 +157,12 @@ class CommonCriteriaCert(Certificate):
 
     @classmethod
     def from_dict(cls, dct: dict) -> 'CommonCriteriaCert':
-        # TODO: Implement me
-        pass
+        dct['maintainance_updates'] = set(dct['maintainance_updates'])
+        dct['protection_profiles'] = set(dct['protection_profiles'])
+
+        args = tuple(dct.values())
+
+        return cls(*args)
 
     @classmethod
     def from_html_row(cls, row: Tag, category: str) -> 'CommonCriteriaCert':
diff --git a/sec_certs/dataset.py b/sec_certs/dataset.py
index a0b8f030..b50285c5 100644
--- a/sec_certs/dataset.py
+++ b/sec_certs/dataset.py
@@ -20,25 +20,41 @@ class DatasetJSONEncoder(json.JSONEncoder):
             return list(obj)
         if isinstance(obj, date):
             return str(obj)
+        if isinstance(obj, Path):
+            return str(obj)
         if isinstance(obj, CommonCriteriaCert.ProtectionProfile):
             return obj.to_dict()
         if isinstance(obj, CommonCriteriaCert.MaintainanceReport):
             return obj.to_dict()
         if isinstance(obj, Dataset):
-            return list(obj.certs.values())
+            return obj.to_dict()
 
         return super().default(obj)
 
 
+class DatasetJSONDecoder(json.JSONDecoder):
+    def __init__(self, *args, **kwargs):
+        json.JSONDecoder.__init__(self, object_hook=self.object_hook, *args, **kwargs)
+
+    def object_hook(self, obj):
+        if 'root_dir' in obj:  # TODO: This is a heavy simplification
+            return CCDataset.from_dict(obj)
+        if 'pp_name' in obj and 'pp_link' in obj:
+            return CommonCriteriaCert.ProtectionProfile.from_dict(obj)
+        if 'maintainance_date' in obj and 'maintainance_title' in obj and 'maintainance_report_link' in obj and 'maintainance_st_link':
+            return CommonCriteriaCert.MaintainanceReport.from_dict(obj)
+        if 'category' in obj:  # TODO: This is heavy simplification.
+            return CommonCriteriaCert.from_dict(obj)
+
+
 class Dataset(ABC):
     def __init__(self, certs: dict, root_dir: Path, name: str = 'dataset name', description: str = 'dataset_description'):
-        self.certs = certs
         self.root_dir = root_dir
-
         self.timestamp = datetime.now()
         self.sha256_digest = 'not implemented'
         self.name = name
         self.description = description
+        self.certs = certs
 
     def __iter__(self):
         for cert in self.certs.values():
@@ -59,18 +75,20 @@ class Dataset(ABC):
     def __str__(self) -> str:
         return 'Not implemented'
 
-    def to_json(self):
-        pass
-
     def to_csv(self):
         pass
 
     def to_dataframe(self):
         pass
 
+    def to_dict(self):
+        return {'root_dir': self.root_dir, 'timestamp': self.timestamp, 'sha256_digest': self.sha256_digest, 'name': self.name,
+                'description': self.description, 'n_certs': len(self), 'certs': list(self.certs.values())}
+
     @classmethod
-    def from_json(cls):
-        pass
+    def from_dict(cls, dct: Dict):
+        certs = {x.dgst: x for x in dct['certs']}
+        return cls(certs, dct['root_dir'], dct['name'], dct['description'])
 
     @classmethod
     def from_csv(cls):