Improve Japanese rules.

2 files changed, 19 insertions, 7 deletions

diff --git a/src/sec_certs/rules.yaml b/src/sec_certs/rules.yaml
index 6b63b991..95a834ba 100644
--- a/src/sec_certs/rules.yaml
+++ b/src/sec_certs/rules.yaml
@@ -56,9 +56,13 @@ cc_cert_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
   JP:
-    - "(?:CRP|ACR)-C[0-9]+-[0-9]+"  # Japan (CRP-C0595-01 ACR-C0417-03)
-    - "JISEC-CC-CRP-C[0-9]+-[0-9]+-[0-9]+"  # Japan (JISEC-CC-CRP-C0689-01-2020)
-    - "Certification No. [cC][0-9]+" # Japan (Certification No. C0090)
+    - "(?: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})"
+    - "Certification No. [cC](?P<counter>[0-9]+)"
+    # Examples:
+    # CRP-C0595-01
+    # JISEC-CC-CRP-C0689-01-2020
+    # Certification No. C0090
   MY:
     - "ISCB-(?P<digit>[0-9])-RPT-C(?P<counter>[0-9]{3})-CR(?:-[0-9])?-(?P<version>[vV][0-9][a-z]?)"
     # Examples:
diff --git a/src/sec_certs/sample/cc_certificate_id.py b/src/sec_certs/sample/cc_certificate_id.py
index 42dc9e6f..fe362b99 100644
--- a/src/sec_certs/sample/cc_certificate_id.py
+++ b/src/sec_certs/sample/cc_certificate_id.py
@@ -147,10 +147,18 @@ class CertificateId:
 
     def _canonical_jp(self):
         new_cert_id = self.clean
-        if match := re.match("Certification No. (C[0-9]+)", new_cert_id):
-            return match.group(1)
-        if match := re.search("CRP-(C[0-9]+)-", new_cert_id):
-            return match.group(1)
+        for rule in rules["cc_cert_id"]["JP"]:
+            if match := re.match(rule, new_cert_id):
+                groups = match.groupdict()
+                counter = groups["counter"]
+                digit = groups.get("digit")
+                year = _parse_year(groups.get("year"))
+                new_cert_id = f"JISEC-CC-CRP-C{counter}"
+                if digit:
+                    new_cert_id += f"-{digit}"
+                if year:
+                    new_cert_id += f"-{year}"
+                break
         return new_cert_id
 
     def _canonical_no(self):