progress on class hierarchy

5 files changed, 111 insertions, 65 deletions

diff --git a/sec_certs/certificate.py b/sec_certs/certificate.py
index d00e1b05..14adf86a 100644
--- a/sec_certs/certificate.py
+++ b/sec_certs/certificate.py
@@ -1,47 +1,52 @@
 from typing import Type
 import json
+from abc import ABC, abstractmethod
+from . import constants as constants
 
 
-# TODO: Move me to appropriate place. I'm here just to demonstrate how we're going to serialize to json
 class CertificateJSONEncoder(json.JSONEncoder):
     def default(self, obj):
         if isinstance(obj, Certificate):
-            return obj.asdict()
+            return obj.to_dict()
         return super().default(obj)
 
 
-class Certificate:
+class Certificate(ABC):
     def __init__(self):
         self.sha256 = None
-        self.framework = None  # CC or FIPS or PP or whatever
-        self.product_name = None
-        self.vendor_name = None
-        self.pdf_path = None
 
     def __repr__(self):
-        return str(self.asdict())
+        return str(self.to_dict())
 
     def __str__(self):
-        return 'Not implemented'  # TODO: Introduce some meaningful representation of the certificate
+        return 'Not implemented'
 
-    def asdict(self):
-        return {'not': 'implemented'}  # TODO: Implement me, I'll be used for json serialization!
+    @abstractmethod
+    def to_dict(self):
+        raise NotImplementedError('Not meant to be implemented')
 
     def __eq__(self, other: 'Certificate') -> bool:
         return self.sha256 == other.sha256
 
     @classmethod
+    @abstractmethod
     def from_dict(cls, dct):
         raise NotImplementedError('Mot meant to be implemented')
 
 
 class CommonCriteriaCert(Certificate):
+    def to_dict(self):
+        pass
+
     @classmethod
     def from_dict(cls, dct):
         return CommonCriteriaCert()
 
 
 class FIPSCertificate(Certificate):
+    def to_dict(self):
+        pass
+
     @classmethod
     def from_dict(cls, dct):
         return FIPSCertificate()
diff --git a/sec_certs/constants.py b/sec_certs/constants.py
index 8d2e0657..d4da75bd 100644
--- a/sec_certs/constants.py
+++ b/sec_certs/constants.py
@@ -5,6 +5,12 @@ class CertFramework(Enum):
     CC = 'Common Criteria'
     FIPS = 'FIPS'
 
+cc_active_csv_url = 'https://www.commoncriteriaportal.org/products/certified_products.csv'
+cc_active_html_url = 'https://www.commoncriteriaportal.org/products/?expand&names'
+
+cc_archived_csv_url = 'https://www.commoncriteriaportal.org/products/certified_products-archived.csv'
+cc_archived_html_url = 'https://www.commoncriteriaportal.org/products/?archived=1&expand&names'
+
 TAG_MATCH_COUNTER = 'count'
 TAG_MATCH_MATCHES = 'matches'
 
@@ -31,4 +37,23 @@ TAG_PP_REGISTRATOR_SIMPLIFIED = 'pp_registrator_simplified'
 TAG_PP_SPONSOR = 'pp_sponsor'
 TAG_PP_EDITOR = 'pp_editor'
 TAG_PP_REVIEWER = 'pp_reviewer'
-TAG_KEYWORDS = 'keywords'
\ No newline at end of file
+TAG_KEYWORDS = 'keywords'
+
+cc_features = ['name', 'category', 'vendor', 'vendor_web', 'scheme', 'security_level', 'protection_profiles',
+               'not_valid_before', 'not_valid_after', 'cert_link', 'cert_report_link', 'security_target_link',
+               'maintenances']
+
+cc_categories = {'AC': 'Access Control Devices and Systems',
+                 'BP': 'Boundary Protection Devices and Systems',
+                 'DP': 'Data Protection',
+                 'DB': 'Databases',
+                 'DD': 'Detection Devices and Systems',
+                 'IC': 'ICs, Smart Cards and Smart Card-Related Devices and Systems',
+                 'KM': 'Key Management Systems',
+                 'MD': 'Mobility',
+                 'MF': 'Multi-Function Devices',
+                 'NS': 'Network and Network-Related Devices and Systems',
+                 'OS': 'Operating Systems',
+                 'OD': 'Other Devices and Systems',
+                 'DG': 'Products for Digital Signatures',
+                 'TC': 'Trusted Computing'}
diff --git a/sec_certs/dataset.py b/sec_certs/dataset.py
index 4c008ba1..a05c6dba 100644
--- a/sec_certs/dataset.py
+++ b/sec_certs/dataset.py
@@ -1,6 +1,6 @@
-from .meta_parser import CCMetaParser
 from .certificate import CommonCriteriaCert
 from . import constants
+from datetime import datetime
 
 
 class Dataset:
@@ -8,9 +8,9 @@ class Dataset:
         self.certs = certs
         self.framework = framework
         self.data_dir = data_dir
-
-        self.sha256_digest = 'not implemented'  # TODO: Implement as hash of all certs
-        self.name = name  # TODO: Allow for naming of the datasets
+        self.timestamp = datetime.now()
+        self.sha256_digest = 'not implemented'
+        self.name = name
         self.description = description  # Allow for descriptions of the dataset
 
     # The idea is to iterate over dataset as: `for cert in Dataset: ... `
@@ -29,38 +29,38 @@ class Dataset:
     def __len__(self):
         return len(self.certs)
 
-    def dump_to_json(self):
+    def to_json(self):
         pass
 
-    # Not sure if we're going to implement
-    def dump_to_csv(self):
+    def to_csv(self):
         pass
 
-    # Not sure if we want such behaviour
-    def __eq__(self, other):
-        return self.certs == other.certs
-
-    def __str__(self):
-        return 'Not implemented'  # TODO: Prepare some meaningful representation of the dataset
+    def to_dataframe(self):
+        pass
 
     @classmethod
-    def init_from_meta(cls, data_dir, name, description):
+    def from_json(cls):
         pass
 
+    @classmethod
+    def from_csv(cls):
+        pass
 
-class CCDataset(Dataset):
     @classmethod
-    def init_from_meta(cls, data_dir, name, description):
-        parser = CCMetaParser()
-        records = parser.parse_meta()
-        certs = {}
+    def from_dataframe(cls, df):
+        pass
+
+    def merge(self, other):
+        return self
+
+    # Not sure if we want such behaviour
+    def __eq__(self, other):
+        return self.certs == other.certs
+
+    def __str__(self):
+        return 'Not implemented'
+
 
-        for r in records:
-            cert = CommonCriteriaCert.from_dict(r)
-            certs[cert.sha256] = cert
 
-        return CCDataset(certs, constants.CertFramework.CC, data_dir, name, description)
 
 
-class FIPSDataset(Dataset):
-    pass
diff --git a/sec_certs/dataset_loader.py b/sec_certs/dataset_loader.py
new file mode 100644
index 00000000..e6094336
--- /dev/null
+++ b/sec_certs/dataset_loader.py
@@ -0,0 +1,44 @@
+from abc import ABC, abstractmethod
+from .dataset import Dataset
+import pandas as pd
+
+
+class CertLoader(ABC):
+    def __init__(self, download_pdfs, meta_dir, pdf_dir):
+        self.df = pd.DataFrame()
+        self.download_pdfs = download_pdfs
+        self.meta_dir = meta_dir
+        self.pdf_dir = pdf_dir
+        self.certs = {}
+
+    @abstractmethod
+    def load(self):
+        pass
+
+
+class CCCertLoader(CertLoader):
+
+    def load_csv(self):
+        pass
+
+    def load_html(self):
+        pass
+
+    def load(self):
+        self.load_csv()
+        self.load_html()
+        return self.certs
+
+
+class FIPSCertLoader(CertLoader):
+    def load_csv(self):
+        pass
+
+    def load_html(self):
+        pass
+
+    def load(self):
+        self.load_csv()
+        self.load_html()
+        return self.certs
+
diff --git a/sec_certs/meta_parser.py b/sec_certs/meta_parser.py
deleted file mode 100644
index bd72bae7..00000000
--- a/sec_certs/meta_parser.py
+++ /dev/null
@@ -1,28 +0,0 @@
-class MetaParser:
-    def parse_certs_from_html(self):
-        raise NotImplementedError('Meant to be provided by child classes')
-
-    def parse_certs_from_csv(self):
-        raise NotImplementedError('Meant to be provided by child classes')
-
-
-class CCMetaParser(MetaParser):
-    def __init__(self):
-        self.records = {}
-
-    def parse_meta(self):
-        self.parse_certs_from_csv()
-        self.parse_certs_from_html()
-
-        return self.records
-
-    def parse_certs_from_csv(self):
-        pass
-
-    def parse_certs_from_html(self):
-        pass
-
-
-class FIPSMetaParser(MetaParser):
-    def __init__(self):
-        pass
\ No newline at end of file