Improve US scheme rules.

3 files changed, 38 insertions, 4 deletions

diff --git a/src/sec_certs/rules.yaml b/src/sec_certs/rules.yaml
index 5604b709..fa188052 100644
--- a/src/sec_certs/rules.yaml
+++ b/src/sec_certs/rules.yaml
@@ -34,7 +34,12 @@ cc_cert_id:
     # 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)
+    - "CCEVS-VR-(?:(?P<cc>CC)-)?(?:(?P<VID>VID)-?)?(?P<year>[0-9]{2})-(?P<counter>[0-9]+)"
+    - "CCEVS-VR-(?:(?P<cc>CC)-)?(?:(?P<VID>VID)-?)?(?P<counter>[0-9]{4,5})-(?P<year>[0-9]{4})?"
+    # Examples:
+    # CCEVS-VR-VID10015-2008
+    # CCEVS-VR-10880-2018
+    # CCEVS-VR-04-0082
   CA:
     # '[0-9][0-9\-]+?-CR', # Canada
     - "[0-9][0-9][0-9]-[347]-[0-9][0-9][0-9]?(?:-CR|P)?"  # Canada xxx-{347}-xxx (383-4-438, 383-4-82-CR, 383-4-422P)
diff --git a/src/sec_certs/sample/cc_certificate_id.py b/src/sec_certs/sample/cc_certificate_id.py
index 188a2163..e84f6df2 100644
--- a/src/sec_certs/sample/cc_certificate_id.py
+++ b/src/sec_certs/sample/cc_certificate_id.py
@@ -10,7 +10,9 @@ def _parse_year(year: str | None) -> int | None:
     if year is None:
         return None
     y = int(year)
-    if y < 100:
+    if y < 50:
+        return y + 2000
+    elif y < 100:
         return y + 1900
     else:
         return y
@@ -39,7 +41,7 @@ class CertificateId:
                     new_cert_id += f"-{doc}"
                 if version:
                     new_cert_id += f"v{version}"
-                return new_cert_id
+                break
 
         return new_cert_id
 
@@ -63,7 +65,27 @@ class CertificateId:
                     new_cert_id += f"-{year}"
                 if doc:
                     new_cert_id += f"-{doc}"
-                return new_cert_id
+                break
+
+        return new_cert_id
+
+    def _canonical_us(self) -> str:
+        new_cert_id = self.clean
+        for rule in rules["cc_cert_id"]["US"]:
+            if match := re.match(rule, new_cert_id):
+                groups = match.groupdict()
+                year = _parse_year(groups["year"])
+                counter = groups["counter"]
+                cc = groups.get("cc")
+                vid = groups.get("VID")
+                new_cert_id = "CCEVS-VR"
+                if cc:
+                    new_cert_id += f"-{cc}"
+                if vid:
+                    new_cert_id += f"-{vid}"
+                new_cert_id += f"-{counter}"
+                new_cert_id += f"-{year}"
+                break
 
         return new_cert_id
 
@@ -147,6 +169,7 @@ class CertificateId:
         schemes = {
             "FR": self._canonical_fr,
             "DE": self._canonical_de,
+            "US": self._canonical_us,
             "ES": self._canonical_es,
             "IT": self._canonical_it,
             "IN": self._canonical_in,
diff --git a/tests/cc/test_cc_misc.py b/tests/cc/test_cc_misc.py
index b9654c91..6fde3339 100644
--- a/tests/cc/test_cc_misc.py
+++ b/tests/cc/test_cc_misc.py
@@ -13,6 +13,12 @@ def test_canonicalize_de():
     assert canonicalize("BSI-DSZ-CC-0837-V2-2014-MA-01", "DE") == "BSI-DSZ-CC-0837-V2-2014-MA-01"
 
 
+def test_canonicalize_us():
+    assert canonicalize("CCEVS-VR-VID10015-2008", "US") == "CCEVS-VR-VID-10015-2008"
+    assert canonicalize("CCEVS-VR-10880-2018", "US") == "CCEVS-VR-10880-2018"
+    assert canonicalize("CCEVS-VR-04-0082", "US") == "CCEVS-VR-0082-2004"
+
+
 def test_canonicalize_es():
     assert canonicalize("2011-14-INF-1095-v1", "ES") == "2011-14-INF-1095"