Improve Singaporean and Australian rules.

2 files changed, 26 insertions, 5 deletions

diff --git a/src/sec_certs/rules.yaml b/src/sec_certs/rules.yaml
index b4e54d20..b2b4a979 100644
--- a/src/sec_certs/rules.yaml
+++ b/src/sec_certs/rules.yaml
@@ -92,13 +92,17 @@ cc_cert_id:
     # will miss STQC/CC/14-15/12/ETR/0017
     - "(?:IC3S|STQC/CC)/[^ ]+? ?/CR"
   SG:
-    - "CSA_CC_[0-9]+"  # Singapore (CSA_CC_19001)
+    - "CSA_CC_(?P<year>[0-9]{2})(?P<counter>[0-9]{3})"
+    # Examples:
+    # CSA_CC_19001
   AU:
-    # Australia (EFS-T048 ETR 1.0, EFS-T056-ETR 1.0, DXC-EFC-T092-ETR 1.0)
+    - "(?:Certificate Number:|Certification Report) (?P<year>[0-9]{2,4})/(?P<counter>[0-9]+)"
     # XXX: Do not use Australian ETR numbers, they are not certificate id.
-    # - "(?:EFS|EFT|DXC-EFC)-T[0-9]+(?: |-)ETR [0-9]+.[0-9]+"
-    - "Certificate Number: [0-9]{1,4}/[0-9]{1,4}"
-    - "Certification Report [0-9]+/[0-9]+"
+    # Examples:
+    # Certification Report 2007/06
+    # Certificate Number: 2010/67
+    # Certificate Number: 37/2006    !mistake
+    # Certification Report 97/76     !short year
 
 #####
 # Common Criteria protection profile IDs, grouped by certification body (e.g. BSI)
diff --git a/src/sec_certs/sample/cc_certificate_id.py b/src/sec_certs/sample/cc_certificate_id.py
index be30bcf5..7e0b7f4a 100644
--- a/src/sec_certs/sample/cc_certificate_id.py
+++ b/src/sec_certs/sample/cc_certificate_id.py
@@ -183,6 +183,21 @@ class CertificateId:
             new_cert_id = f"{new_cert_id}-CR"
         return new_cert_id
 
+    def _canonical_au(self):
+        new_cert_id = self.clean
+        for rule in rules["cc_cert_id"]["AU"]:
+            if match := re.match(rule, new_cert_id):
+                groups = match.groupdict()
+                counter = groups["counter"]
+                year_s = groups["year"]
+                if len(year_s) < len(counter):
+                    # Hack for some mistakes in their ordering
+                    year_s, counter = counter, year_s
+                year = _parse_year(year_s)
+                new_cert_id = f"Certificate Number: {year}/{counter}"
+                break
+        return new_cert_id
+
     @property
     def clean(self) -> str:
         """
@@ -210,6 +225,8 @@ class CertificateId:
             "JP": self._canonical_jp,
             "NO": self._canonical_no,
             "NL": self._canonical_nl,
+            "AU": self._canonical_au,
+            # SG is canonical by default
         }
 
         if self.scheme in schemes: