Improve Korean rules.

2 files changed, 18 insertions, 3 deletions

diff --git a/src/sec_certs/rules.yaml b/src/sec_certs/rules.yaml
index c0acd3d4..5e000144 100644
--- a/src/sec_certs/rules.yaml
+++ b/src/sec_certs/rules.yaml
@@ -58,10 +58,12 @@ cc_cert_id:
     # 2020-34-INF-3784- v1
     # 2019-20-INF-3379-v1
   KR:
-    # Korea
+    - "KECS[-‐](?P<word>ISIS|NISS|CISS)[-‐](?P<counter>[0-9]{2,4})[-‐](?P<year>[0-9]{4})"
     # XXX: Do not use KECS-CR as those refer to the certificate report and do not represent the certificate id.
-    # - "KECS[-‐]CR[-‐][0-9]+[-‐][0-9]+"  # Korea KECS-CR-20-61
-    - "KECS[-‐](?:ISIS|NISS|CISS)[-‐][0-9]+[-‐][0-9]{4}"  # Korea KECS-ISIS-1234-2011
+    # Examples:
+    # KECS-ISIS-0579-2015
+    # KECS-NISS-0792-2017
+    # KECS-CISS-1210-2023
   JP:
     - "(?:CRP|ACR)-C(?P<counter>[0-9]+)-(?P<digit>[0-9]+)"
     - "JISEC-CC-CRP-C(?P<counter>[0-9]+)-(?P<digit>[0-9]+)-(?P<year>[0-9]{4})"
diff --git a/src/sec_certs/sample/cc_certificate_id.py b/src/sec_certs/sample/cc_certificate_id.py
index 88e1f4a7..5facf1a5 100644
--- a/src/sec_certs/sample/cc_certificate_id.py
+++ b/src/sec_certs/sample/cc_certificate_id.py
@@ -163,6 +163,18 @@ class CertificateId:
                 break
         return new_cert_id
 
+    def _canonical_kr(self):
+        new_cert_id = self.clean
+        for rule in rules["cc_cert_id"]["KR"]:
+            if match := re.match(rule, new_cert_id):
+                groups = match.groupdict()
+                word = groups["word"]
+                counter = int(groups["counter"])
+                year = _parse_year(groups["year"])
+                new_cert_id = f"KECS-{word}-{counter:04}-{year}"
+                break
+        return new_cert_id
+
     def _canonical_no(self):
         new_cert_id = self.clean
         cert_num = int(new_cert_id.split("-")[1])
@@ -218,6 +230,7 @@ class CertificateId:
             "NO": self._canonical_no,
             "NL": self._canonical_nl,
             "AU": self._canonical_au,
+            "KR": self._canonical_kr,
             # SG is canonical by default
             # IT is canonucal by default
         }