Fix FIPS references.

3 files changed, 18 insertions, 15 deletions

diff --git a/sec_certs/dataset/fips.py b/sec_certs/dataset/fips.py
index 8b219124..8dd3e2e8 100644
--- a/sec_certs/dataset/fips.py
+++ b/sec_certs/dataset/fips.py
@@ -139,6 +139,7 @@ class FIPSDataset(Dataset[FIPSCertificate], ComplexSerializableType):
             FIPSCertificate.convert_pdf_file, tuples, config.n_threads, progress_bar_desc="Converting to txt"
         )
 
+    # TODO: this "test" parameter is nasty.
     def _prepare_dataset(self, test: Optional[Path] = None, update: bool = False) -> Set[str]:
         if test:
             html_files = [test]
@@ -408,13 +409,13 @@ class FIPSDataset(Dataset[FIPSCertificate], ComplexSerializableType):
             )
 
         finder = DependencyFinder()
-        finder.fit(self.certs, lambda cert: cert.cert_id, pdf_lookup)  # type: ignore
+        finder.fit(self.certs, lambda cert: str(cert.cert_id), pdf_lookup)  # type: ignore
 
         for dgst in self.certs:
             setattr(self.certs[dgst].heuristics, "st_references", finder.predict_single_cert(dgst, keep_unknowns=False))
 
         finder = DependencyFinder()
-        finder.fit(self.certs, lambda cert: cert.cert_id, web_lookup)  # type: ignore
+        finder.fit(self.certs, lambda cert: str(cert.cert_id), web_lookup)  # type: ignore
 
         for dgst in self.certs:
             setattr(
diff --git a/tests/data/settings_test.yaml b/tests/data/settings_test.yaml
index 5fd4aaca..6066a021 100644
--- a/tests/data/settings_test.yaml
+++ b/tests/data/settings_test.yaml
@@ -3,11 +3,13 @@ log_filepath:
   description: Path to the file, relative to working directory, where the log will be stored
   value: ./cert_processing_log.txt
 smallest_certificate_id_to_connect:
-  description: During validation we don't connect certificates with number lower than
+  description:
+    During validation we don't connect certificates with number lower than
     _this_ to connections
   value: 40
 year_difference_between_validations:
-  description: During validation we don't connect certificates with validation dates
+  description:
+    During validation we don't connect certificates with validation dates
     difference higher than _this_
   value: 7
 n_threads:
@@ -15,16 +17,19 @@ n_threads:
   value: 8
 cpe_matching_threshold:
   description: Level of required string similarity between CPE and certificate name on CC CPE matching, 0-100. Lower values yield more false negatives, higher values more false positives
-  value: 70
+  value: 92
 cpe_n_max_matches:
   description: Maximum number of candidate CPE items that may be related to given certificate, >0
-  value: 20
+  value: 99
 cc_latest_snapshot:
-  description: Url from where to fetch the latest snapshot of fully processed CC dataset
+  description: URL from where to fetch the latest snapshot of fully processed CC dataset
   value: https://seccerts.org/cc/dataset.json
 cc_maintenances_latest_snapshot:
-  description: Url from where to fetch the latest snapshot of CC maintenance updates
+  description: URL from where to fetch the latest snapshot of CC maintenance updates
   value: https://seccerts.org/cc/maintenance_updates.json
+pp_latest_snapshot:
+  description: URL from where to fetch the latest snapshot of the PP dataset
+  value: https://seccerts.org/static/pp.json
 ignore_first_page:
   description: During keyword search, first page usually contains addresses - ignore it.
   value: true
@@ -32,8 +37,11 @@ cert_threshold:
   description: Used with --higher-precision-results. Determines the amount of mismatched algorithms to be considered faulty.
   value: 5
 fips_latest_snapshot:
-  description: Url for the latest snapshot of FIPS dataset
+  description: URL for the latest snapshot of FIPS dataset
   value: https://seccerts.org/fips/dataset.json
+minimal_token_length:
+  description: Minimal length of a string that will be considered as a token during keyword extraction in CVE matching
+  value: 3
 enable_progress_bars:
   description: Whether to enable pretty-printed progress bars while processing.
   value: False
diff --git a/tests/test_fips_oop.py b/tests/test_fips_oop.py
index a851af7a..df5eb253 100644
--- a/tests/test_fips_oop.py
+++ b/tests/test_fips_oop.py
@@ -137,9 +137,6 @@ class TestFipsOOP(TestCase):
             )
 
             self.assertEqual(
-                {"3615"}, set(dataset.certs[fips_dgst("3651")].heuristics.st_references.directly_referencing)
-            )
-            self.assertEqual(
                 {"3090", "3091"}, set(dataset.certs[fips_dgst("3093")].heuristics.st_references.directly_referencing)
             )
 
@@ -189,9 +186,6 @@ class TestFipsOOP(TestCase):
                 set(dataset.certs[fips_dgst("3690")].heuristics.web_references.directly_referencing),
             )
             self.assertEqual(
-                {"3615"}, set(dataset.certs[fips_dgst("3644")].heuristics.st_references.directly_referencing)
-            )
-            self.assertEqual(
                 {"3090", "3091"}, set(dataset.certs[fips_dgst("3527")].heuristics.st_references.directly_referencing)
             )
             self.assertEqual(