further advances on mockup classes, meta (csv, html) parsers introduced

3 files changed, 72 insertions, 9 deletions

diff --git a/sec_certs/certificate.py b/sec_certs/certificate.py
index 3e66f278..d00e1b05 100644
--- a/sec_certs/certificate.py
+++ b/sec_certs/certificate.py
@@ -29,3 +29,19 @@ class Certificate:
 
     def __eq__(self, other: 'Certificate') -> bool:
         return self.sha256 == other.sha256
+
+    @classmethod
+    def from_dict(cls, dct):
+        raise NotImplementedError('Mot meant to be implemented')
+
+
+class CommonCriteriaCert(Certificate):
+    @classmethod
+    def from_dict(cls, dct):
+        return CommonCriteriaCert()
+
+
+class FIPSCertificate(Certificate):
+    @classmethod
+    def from_dict(cls, dct):
+        return FIPSCertificate()
diff --git a/sec_certs/dataset.py b/sec_certs/dataset.py
index 278dfae9..4c008ba1 100644
--- a/sec_certs/dataset.py
+++ b/sec_certs/dataset.py
@@ -1,12 +1,17 @@
+from .meta_parser import CCMetaParser
+from .certificate import CommonCriteriaCert
+from . import constants
+
+
 class Dataset:
-    def __init__(self):
-        self.certs = {}
-        self.framework = None
-        self.data_dir = None
+    def __init__(self, certs, framework, data_dir, name='dataset name', description='dataset_description'):
+        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 = 'not implemented'  # TODO: Allow for naming of the datasets
-        self.description = 'not implemented'  # Allow for descriptions of the dataset
+        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: ... `
     def __iter__(self):
@@ -38,10 +43,24 @@ class Dataset:
     def __str__(self):
         return 'Not implemented'  # TODO: Prepare some meaningful representation of the dataset
 
+    @classmethod
+    def init_from_meta(cls, data_dir, name, description):
+        pass
 
-class DatasetCC(Dataset):
-    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 DatasetFIPS(Dataset):
+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