Merge branch 'dataset_classes' of github.com:petrs/sec-certs into dataset_classes

4 files changed, 64 insertions, 32 deletions

diff --git a/sec_certs/certificate.py b/sec_certs/certificate.py
index 14adf86a..a4ff8288 100644
--- a/sec_certs/certificate.py
+++ b/sec_certs/certificate.py
@@ -1,5 +1,6 @@
 from typing import Type
 import json
+
 from abc import ABC, abstractmethod
 from . import constants as constants
 
diff --git a/sec_certs/constants.py b/sec_certs/constants.py
index d4da75bd..8d2e0657 100644
--- a/sec_certs/constants.py
+++ b/sec_certs/constants.py
@@ -5,12 +5,6 @@ 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'
 
@@ -37,23 +31,4 @@ 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'
-
-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'}
+TAG_KEYWORDS = 'keywords'
\ No newline at end of file
diff --git a/sec_certs/dataset.py b/sec_certs/dataset.py
index a05c6dba..b77eb962 100644
--- a/sec_certs/dataset.py
+++ b/sec_certs/dataset.py
@@ -1,6 +1,9 @@
 from .certificate import CommonCriteriaCert
 from . import constants
 from datetime import datetime
+from .meta_parser import CCMetaParser
+from .certificate import CommonCriteriaCert
+from . import constants
 
 
 class Dataset:
@@ -8,12 +11,16 @@ class Dataset:
         self.certs = certs
         self.framework = framework
         self.data_dir = data_dir
+
         self.timestamp = datetime.now()
         self.sha256_digest = 'not implemented'
         self.name = name
+
+        self.sha256_digest = 'not implemented'  # TODO: Implement as hash of all certs
+        self.name = name  # TODO: Allow for naming of the datasets
         self.description = description  # Allow for descriptions of the dataset
 
-    # The idea is to iterate over dataset as: `for cert in Dataset: ... `
+    # The idea is to iterate over dataset as: `for cert in Dataset: ...
     def __iter__(self):
         for cert in self.certs.values():
             yield cert
@@ -29,6 +36,13 @@ class Dataset:
     def __len__(self):
         return len(self.certs)
 
+    # Not sure if we want such behaviour
+    def __eq__(self, other):
+        return self.certs == other.certs
+
+    def __str__(self):
+        return 'Not implemented'
+
     def to_json(self):
         pass
 
@@ -53,14 +67,28 @@ class Dataset:
     def merge(self, other):
         return self
 
-    # Not sure if we want such behaviour
-    def __eq__(self, other):
-        return self.certs == other.certs
+    def dump_to_json(self):
+        pass
 
-    def __str__(self):
-        return 'Not implemented'
+    @classmethod
+    def init_from_meta(cls, data_dir, name, description):
+        pass
+
+
+class CCDataset(Dataset):
+    @classmethod
+    def init_from_meta(cls, data_dir, name, description):
+        parser = CCMetaParser()
+        records = parser.parse_meta()
+        certs = {}
 
+        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/meta_parser.py b/sec_certs/meta_parser.py
new file mode 100644
index 00000000..bd72bae7
--- /dev/null
+++ b/sec_certs/meta_parser.py
@@ -0,0 +1,28 @@
+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