Cleanup PdfData attributes.

5 files changed, 46 insertions, 157 deletions

diff --git a/src/sec_certs/sample/cc.py b/src/sec_certs/sample/cc.py
index bb4b0c7c..28099f58 100644
--- a/src/sec_certs/sample/cc.py
+++ b/src/sec_certs/sample/cc.py
@@ -27,15 +27,7 @@ from sec_certs.sample.sar import SAR
 from sec_certs.serialization.json import ComplexSerializableType
 from sec_certs.serialization.pandas import PandasSerializableType
 from sec_certs.utils import helpers
-from sec_certs.utils.extract import normalize_match_string
-
-HEADERS = {
-    "anssi": sec_certs.utils.extract.search_only_headers_anssi,
-    "bsi": sec_certs.utils.extract.search_only_headers_bsi,
-    "nscib": sec_certs.utils.extract.search_only_headers_nscib,
-    "niap": sec_certs.utils.extract.search_only_headers_niap,
-    "canada": sec_certs.utils.extract.search_only_headers_canada,
-}
+from sec_certs.utils.extract import normalize_match_string, scheme_frontpage_functions
 
 
 class CCCertificate(
@@ -181,8 +173,12 @@ class CCCertificate(
         st_metadata: dict[str, Any] | None = field(default=None)
         cert_metadata: dict[str, Any] | None = field(default=None)
         report_frontpage: dict[str, dict[str, Any]] | None = field(default=None)
-        st_frontpage: dict[str, dict[str, Any]] | None = field(default=None)
-        cert_frontpage: dict[str, dict[str, Any]] | None = field(default=None)
+        st_frontpage: dict[str, dict[str, Any]] | None = field(
+            default=None
+        )  # TODO: Unused, we have no frontpage matching for targets
+        cert_frontpage: dict[str, dict[str, Any]] | None = field(
+            default=None
+        )  # TODO: Unused, we have no frontpage matching for certs
         report_keywords: dict[str, Any] | None = field(default=None)
         st_keywords: dict[str, Any] | None = field(default=None)
         cert_keywords: dict[str, Any] | None = field(default=None)
@@ -194,86 +190,33 @@ class CCCertificate(
             return any(x is not None for x in vars(self))
 
         @property
-        def bsi_data(self) -> dict[str, Any] | None:
-            """
-            Returns frontpage data related to BSI-provided information
-            """
-            return self.report_frontpage.get("bsi", None) if self.report_frontpage else None
-
-        @property
-        def niap_data(self) -> dict[str, Any] | None:
-            """
-            Returns frontpage data related to niap-provided information
-            """
-            return self.report_frontpage.get("niap", None) if self.report_frontpage else None
-
-        @property
-        def nscib_data(self) -> dict[str, Any] | None:
-            """
-            Returns frontpage data related to nscib-provided information
-            """
-            return self.report_frontpage.get("nscib", None) if self.report_frontpage else None
-
-        @property
-        def canada_data(self) -> dict[str, Any] | None:
-            """
-            Returns frontpage data related to canada-provided information
-            """
-            return self.report_frontpage.get("canada", None) if self.report_frontpage else None
-
-        @property
-        def anssi_data(self) -> dict[str, Any] | None:
-            """
-            Returns frontpage data related to ANSSI-provided information
-            """
-            return self.report_frontpage.get("anssi", None) if self.report_frontpage else None
-
-        @property
         def cert_lab(self) -> list[str] | None:
             """
             Returns labs for which certificate data was parsed.
             """
+            if not self.report_frontpage:
+                return None
             labs = [
                 data["cert_lab"].split(" ")[0].upper()
-                for data in [self.bsi_data, self.anssi_data, self.niap_data, self.nscib_data, self.canada_data]
+                for scheme, data in self.report_frontpage.items()
                 if data and "cert_lab" in data
             ]
             return labs if labs else None
 
-        @property
-        def bsi_cert_id(self) -> str | None:
-            return self.bsi_data.get("cert_id", None) if self.bsi_data else None
-
-        @property
-        def niap_cert_id(self) -> str | None:
-            return self.niap_data.get("cert_id", None) if self.niap_data else None
-
-        @property
-        def nscib_cert_id(self) -> str | None:
-            return self.nscib_data.get("cert_id", None) if self.nscib_data else None
-
-        @property
-        def canada_cert_id(self) -> str | None:
-            return self.canada_data.get("cert_id", None) if self.canada_data else None
-
-        @property
-        def anssi_cert_id(self) -> str | None:
-            return self.anssi_data.get("cert_id", None) if self.anssi_data else None
-
         def frontpage_cert_id(self, scheme: str) -> dict[str, float]:
             """
             Get cert_id candidate from the frontpage of the report.
             """
-            scheme_map = {
-                "DE": self.bsi_cert_id,
-                "US": self.niap_cert_id,
-                "NL": self.nscib_cert_id,
-                "CA": self.canada_cert_id,
-                "FR": self.anssi_cert_id,
-            }
-            if scheme in scheme_map and (candidate := scheme_map[scheme]):
-                return {candidate: 1.0}
-            return {}
+            if not self.report_frontpage:
+                return {}
+            data = self.report_frontpage.get(scheme)
+            if not data:
+                return {}
+            cert_id = data.get("cert_id")
+            if not cert_id:
+                return {}
+            else:
+                return {cert_id: 1.0}
 
         def filename_cert_id(self, scheme: str) -> dict[str, float]:
             """
@@ -982,9 +925,9 @@ class CCCertificate(
         """
         cert.pdf_data.report_frontpage = {}
 
-        for header_type, associated_header_func in HEADERS.items():
-            response, cert.pdf_data.report_frontpage[header_type] = associated_header_func(cert.state.report.txt_path)
-
+        if cert.scheme in scheme_frontpage_functions:
+            header_func = scheme_frontpage_functions[cert.scheme]
+            response, cert.pdf_data.report_frontpage[cert.scheme] = header_func(cert.state.report.txt_path)
             if response != constants.RETURNCODE_OK:
                 cert.state.report.extract_ok = False
         return cert
diff --git a/src/sec_certs/utils/extract.py b/src/sec_certs/utils/extract.py
index e6312b8d..669cbcfc 100644
--- a/src/sec_certs/utils/extract.py
+++ b/src/sec_certs/utils/extract.py
@@ -818,3 +818,12 @@ def get_sums_for_rules_subset(dct: dict | None, path: str) -> dict[str, float]:
     cc_rules_subset_to_search = rules_get_subset(path)
     paths_to_search = extract_key_paths(cc_rules_subset_to_search, path)
     return {x: get_sum_of_values_from_dict_path(dct, x, np.nan) for x in paths_to_search}
+
+
+scheme_frontpage_functions = {
+    "FR": search_only_headers_anssi,
+    "DE": search_only_headers_bsi,
+    "NL": search_only_headers_nscib,
+    "US": search_only_headers_niap,
+    "CA": search_only_headers_canada,
+}
diff --git a/tests/data/cc/analysis/cc_full_dataset.json b/tests/data/cc/analysis/cc_full_dataset.json
index 4baf4758..64ef38e4 100644
--- a/tests/data/cc/analysis/cc_full_dataset.json
+++ b/tests/data/cc/analysis/cc_full_dataset.json
@@ -130,8 +130,7 @@
                     }
                 },
                 "report_frontpage": {
-                    "anssi": {},
-                    "bsi": {
+                    "DE": {
                         "match_rules": [
                             "(BSI-DSZ-CC-.+?) (?:for|For) (.+?) from (.*)"
                         ],
@@ -142,17 +141,9 @@
                         "ref_protection_profiles": "None",
                         "cc_version": "Product specific Security Target Common Criteria Part 2 conformant",
                         "cc_security_level": "Common Criteria Part 3 conformant EAL 3 augmented by ALC_FLR.1 SOGIS Recognition Agreement"
-                    },
-                    "nscib": {},
-                    "niap": {},
-                    "canada": {}
+                    }
                 },
                 "st_frontpage": {
-                    "anssi": {},
-                    "bsi": {},
-                    "nscib": {},
-                    "niap": {},
-                    "canada": {}
                 },
                 "report_keywords": {
                     "cc_cert_id": {
diff --git a/tests/data/cc/analysis/reference_dataset.json b/tests/data/cc/analysis/reference_dataset.json
index b0ef6683..38fb7fd0 100644
--- a/tests/data/cc/analysis/reference_dataset.json
+++ b/tests/data/cc/analysis/reference_dataset.json
@@ -101,8 +101,7 @@
           "/Title": "Security Target The Océ Digital Access Controller (DAC) R10.1.5, as used in the Océ VarioPrint 1055, 1055 BC, 1055 DP, 1065, 1075, 2062, 2075, 2075 DP printer/copier/scanner products"
         },
         "report_frontpage": {
-          "anssi": {},
-          "bsi": {
+          "DE": {
             "match_rules": [
               "(BSI-DSZ-CC-.+?) (?:for|For) (.+?) from (.*)"
             ],
@@ -110,17 +109,9 @@
             "cert_item": "Océ Digital Access Controller (DAC) R10.1.5 for use in the Océ VarioPrint 1055, 1055 BC, 1055 DP, 1065, 1075, 2062, 2075, 2075 DP printer/copier/scanner products",
             "developer": "Océ Technologies B.V",
             "cert_lab": "BSI"
-          },
-          "nscib": {},
-          "niap": {},
-          "canada": {}
+          }
         },
         "st_frontpage": {
-          "anssi": {},
-          "bsi": {},
-          "nscib": {},
-          "niap": {},
-          "canada": {}
         },
         "report_keywords": {
           "cc_cert_id": {
@@ -678,8 +669,7 @@
           "/Title": "Microsoft Word - Oce Venlo DAC Security Target 2.4.doc"
         },
         "report_frontpage": {
-          "anssi": {},
-          "bsi": {
+          "DE": {
             "match_rules": [
               "(BSI-DSZ-CC-.+?) (?:for|For) (.+?) from (.*)"
             ],
@@ -687,17 +677,9 @@
             "cert_item": "Océ Digital Access Controller (DAC) R9.1.6",
             "developer": "Océ Technologies B.V",
             "cert_lab": "BSI"
-          },
-          "nscib": {},
-          "niap": {},
-          "canada": {}
+          }
         },
         "st_frontpage": {
-          "anssi": {},
-          "bsi": {},
-          "nscib": {},
-          "niap": {},
-          "canada": {}
         },
         "report_keywords": {
           "cc_cert_id": {
@@ -1333,8 +1315,7 @@
           "/Title": "Microsoft Word - Oce Venlo DAC Security Target 1.9.doc"
         },
         "report_frontpage": {
-          "anssi": {},
-          "bsi": {
+          "DE": {
             "match_rules": [
               "(BSI-DSZ-CC-.+?) (?:for|For) (.+?) from (.*)"
             ],
@@ -1342,17 +1323,9 @@
             "cert_item": "Océ Digital Access Controller (DAC) R 8.1.10",
             "developer": "Océ Technologies B.V",
             "cert_lab": "BSI"
-          },
-          "nscib": {},
-          "niap": {},
-          "canada": {}
+          }
         },
         "st_frontpage": {
-          "anssi": {},
-          "bsi": {},
-          "nscib": {},
-          "niap": {},
-          "canada": {}
         },
         "report_keywords": {
           "cc_cert_id": {
diff --git a/tests/data/cc/analysis/transitive_vulnerability_dataset.json b/tests/data/cc/analysis/transitive_vulnerability_dataset.json
index 21391a12..abd4c7a3 100644
--- a/tests/data/cc/analysis/transitive_vulnerability_dataset.json
+++ b/tests/data/cc/analysis/transitive_vulnerability_dataset.json
@@ -108,8 +108,7 @@
           "/CreationDate": "D:20140828154014+02'00'"
         },
         "report_frontpage": {
-          "anssi": {},
-          "bsi": {
+          "DE": {
             "match_rules": [
               "(BSI-DSZ-CC-.+?) (?:for|For) (.+?) from (.*)"
             ],
@@ -117,17 +116,9 @@
             "cert_item": "IBM z/OS Version 2 Release 1",
             "developer": "IBM Corporation",
             "cert_lab": "BSI"
-          },
-          "nscib": {},
-          "niap": {},
-          "canada": {}
+          }
         },
         "st_frontpage": {
-          "anssi": {},
-          "bsi": {},
-          "nscib": {},
-          "niap": {},
-          "canada": {}
         },
         "report_keywords": {
           "cc_cert_id": {
@@ -1432,8 +1423,7 @@
           "/CreationDate": "D:20150508083715+02'00'"
         },
         "report_frontpage": {
-          "anssi": {},
-          "bsi": {
+          "DE": {
             "match_rules": [
               "(BSI-DSZ-CC-.+?) (?:for|For) (.+?) from (.*)"
             ],
@@ -1441,17 +1431,9 @@
             "cert_item": "RACF Element of z/OS Version 2, Release 1",
             "developer": "IBM Corporation",
             "cert_lab": "BSI"
-          },
-          "nscib": {},
-          "niap": {},
-          "canada": {}
+          }
         },
         "st_frontpage": {
-          "anssi": {},
-          "bsi": {},
-          "nscib": {},
-          "niap": {},
-          "canada": {}
         },
         "report_keywords": {
           "cc_cert_id": {
@@ -2399,8 +2381,7 @@
           "/CreationDate": "D:20170503171742+02'00'"
         },
         "report_frontpage": {
-          "anssi": {},
-          "bsi": {
+          "DE": {
             "match_rules": [
               "(BSI-DSZ-CC-.+?) (?:for|For) (.+?) from (.*)"
             ],
@@ -2411,17 +2392,9 @@
             "ref_protection_profiles": "Operating System Protection Profile, Version 2.0, 01 June 2010, BSI-CC-PP-0067-2010, OSPP Extended Packages: Extended Identification and Authentication and Labeled Security, both Version 2.0, 28 May 2010",
             "cc_version": "PP conformant Common Criteria Part 2 extended",
             "cc_security_level": "Common Criteria Part 3 conformant EAL 4 augmented by ALC_FLR.3"
-          },
-          "nscib": {},
-          "niap": {},
-          "canada": {}
+          }
         },
         "st_frontpage": {
-          "anssi": {},
-          "bsi": {},
-          "nscib": {},
-          "niap": {},
-          "canada": {}
         },
         "report_keywords": {
           "cc_cert_id": {