Improve BSI regexes.

3 files changed, 45 insertions, 39 deletions

diff --git a/src/sec_certs/rules.yaml b/src/sec_certs/rules.yaml
index 87acd04b..5604b709 100644
--- a/src/sec_certs/rules.yaml
+++ b/src/sec_certs/rules.yaml
@@ -5,17 +5,17 @@
 #####
 cc_cert_id:
   DE:
-    - "BSI-DSZ-CC-[0-9]+?-[0-9]+"
-    - "BSI-DSZ-CC-[0-9]+?-(?:V|v)[0-9]+-[0-9]+"
-    - "BSI-DSZ-CC-[0-9]+?-(?:V|v)[0-9]+"
-    - "BSI-DSZ-CC-[0-9]+-(?:V|v)[0-9]+(-[0-9]{4})*"  # German BSI (number + version + year or without year)
-    - "BSI-DSZ-CC-[0-9]+-[0-9]{4}"  # German BSI (number + year, no version)
-    - "BSI-DSZ-CC-[0-9]+-(?:V|v)[0-9]+(?!-)"  # German BSI (number + version but no year => no - after version)
-    # - "BSI-DSZ-CC-[0-9]+"  # Maybe?
+    - "BSI-DSZ-CC-(?:(?P<s>S)-)?(?P<counter>[0-9]{3,5})-(?:(?P<version>[vV][0-9])-)?(?P<year>[0-9]{4})?(?:-(?P<doc>(?:RA|MA)(?:-[0-9]+)?))?"
+    # Examples:
+    # BSI-DSZ-CC-1004
+    # BSI-DSZ-CC-0973-2016
+    # BSI-DSZ-CC-0831-V4-2021
+    # BSI-DSZ-CC-0837-V2-2014-MA-01
+    # BSI-DSZ-CC-S-0192-2021
   FR:
     - "DCSS[Ii]-(?P<year>[0-9]{2,4})/(?P<counter>[0-9]+)([vV](?P<version>[0-9]))?"
-    - "Certification Report (?P<year>[0-9]{2,4})/(?P<counter>[0-9]+)([vV](?P<version>[0-9]))?"
     - "Rapport de certification (?P<year>[0-9]{2,4})/(?P<counter>[0-9]+)([vV](?P<version>[0-9]))?"
+    - "Certification Report (?P<year>[0-9]{2,4})/(?P<counter>[0-9]+)([vV](?P<version>[0-9]))?"
     - "ANSS[Ii](?:-CC)?[ -](?P<year>[0-9]{2,4})[/_-](?P<counter>[0-9]+)(?:-(?P<doc>(?:[MSR][0-9]+)))?([vV](?P<version>[0-9]))?"
     # Examples:
     # DCSSI-2009/07
@@ -30,7 +30,9 @@ cc_cert_id:
     # CC-16-31801-CR4  (no NSCIB)
     # NSCIB-CC-98209   (no year, no CR)
   "NO":
-    - "SERTIT-(?P<counter>[0-9]+)"  # Norway
+    - "SERTIT-(?P<counter>[0-9]+)"
+    # Examples:
+    # SERTIT-101
   US:
     - "CCEVS-VR-(?:CC-|VID)?[0-9]+-[0-9]+[a-z]?(?:-[0-9]+)?"  # US NSA (CCEVS-VR-10884-2018 CCEVS-VR-VID10877-2018)
   CA:
diff --git a/src/sec_certs/sample/cc_certificate_id.py b/src/sec_certs/sample/cc_certificate_id.py
index 9bce0764..188a2163 100644
--- a/src/sec_certs/sample/cc_certificate_id.py
+++ b/src/sec_certs/sample/cc_certificate_id.py
@@ -6,6 +6,16 @@ from dataclasses import dataclass
 from sec_certs.cert_rules import rules
 
 
+def _parse_year(year: str | None) -> int | None:
+    if year is None:
+        return None
+    y = int(year)
+    if y < 100:
+        return y + 1900
+    else:
+        return y
+
+
 @dataclass(eq=True, frozen=True)
 class CertificateId:
     """
@@ -20,7 +30,7 @@ class CertificateId:
         for rule in rules["cc_cert_id"]["FR"]:
             if match := re.match(rule, new_cert_id):
                 groups = match.groupdict()
-                year = int(groups["year"]) if len(groups["year"]) == 4 else int(groups["year"]) + 2000
+                year = _parse_year(groups["year"])
                 counter = groups["counter"]
                 doc = groups.get("doc")
                 version = groups.get("version")
@@ -34,35 +44,26 @@ class CertificateId:
         return new_cert_id
 
     def _canonical_de(self) -> str:
-        def extract_parts(bsi_parts: list[str]) -> tuple:
-            cert_num = None
-            cert_version = None
-            cert_year = None
-
-            if len(bsi_parts) > 3:
-                cert_num = bsi_parts[3]
-            if len(bsi_parts) > 4:
-                if bsi_parts[4].startswith("V") or bsi_parts[4].startswith("v"):
-                    cert_version = bsi_parts[4].upper()  # get version in uppercase
-                else:
-                    cert_year = bsi_parts[4]
-            if len(bsi_parts) > 5:
-                cert_year = bsi_parts[5]
-
-            return cert_num, cert_version, cert_year
-
-        bsi_parts = self.clean.split("-")
-
-        cert_num, cert_version, cert_year = extract_parts(bsi_parts)
-
-        # reconstruct BSI number again
-        new_cert_id = "BSI-DSZ-CC"
-        if cert_num is not None:
-            new_cert_id += "-" + cert_num
-        if cert_version is not None:
-            new_cert_id += "-" + cert_version
-        if cert_year is not None:
-            new_cert_id += "-" + cert_year
+        new_cert_id = self.clean
+        for rule in rules["cc_cert_id"]["DE"]:
+            if match := re.match(rule, new_cert_id):
+                groups = match.groupdict()
+                s = groups.get("s")
+                counter = groups["counter"]
+                version = groups.get("version")
+                year = _parse_year(groups.get("year"))
+                doc = groups.get("doc")
+                new_cert_id = "BSI-DSZ-CC"
+                if s:
+                    new_cert_id += f"-{s}"
+                new_cert_id += f"-{counter}"
+                if version:
+                    new_cert_id += f"-{version.upper()}"
+                if year:
+                    new_cert_id += f"-{year}"
+                if doc:
+                    new_cert_id += f"-{doc}"
+                return new_cert_id
 
         return new_cert_id
 
diff --git a/tests/cc/test_cc_misc.py b/tests/cc/test_cc_misc.py
index ae9a83aa..b9654c91 100644
--- a/tests/cc/test_cc_misc.py
+++ b/tests/cc/test_cc_misc.py
@@ -8,6 +8,9 @@ def test_canonicalize_fr():
 
 def test_canonicalize_de():
     assert canonicalize("BSI-DSZ-CC-0420-2007", "DE") == "BSI-DSZ-CC-0420-2007"
+    assert canonicalize("BSI-DSZ-CC-1004", "DE") == "BSI-DSZ-CC-1004"
+    assert canonicalize("BSI-DSZ-CC-0831-V4-2021", "DE") == "BSI-DSZ-CC-0831-V4-2021"
+    assert canonicalize("BSI-DSZ-CC-0837-V2-2014-MA-01", "DE") == "BSI-DSZ-CC-0837-V2-2014-MA-01"
 
 
 def test_canonicalize_es():