Working CLI

3 files changed, 6 insertions, 3 deletions

diff --git a/fips_cli.py b/fips_cli.py
index 25a57957..adf84392 100755
--- a/fips_cli.py
+++ b/fips_cli.py
@@ -189,6 +189,8 @@ def main(
 
     assert inputpath
     dset: FIPSDataset = FIPSDataset.from_json(inputpath)
+
+    print(f'Have dataset with {len(dset)} certs and {len(dset.algorithms)} algorithms.')
     if output:
         print(
             "Warning: You provided both inputpath and outputpath, dataset will be copied to outputpath (without data)"
diff --git a/sec_certs/dataset/fips.py b/sec_certs/dataset/fips.py
index 0dbfb324..39cb1a92 100644
--- a/sec_certs/dataset/fips.py
+++ b/sec_certs/dataset/fips.py
@@ -89,6 +89,10 @@ class FIPSDataset(Dataset, ComplexSerializableType):
         output = {}
         cert: FIPSCertificate
         for cert in self.certs.values():
+            # if the pdf has not been processed, no matching can be done
+            if not cert.pdf_scan.keywords or not cert.state.txt_state:
+                continue
+            
             output[cert.dgst] = FIPSCertificate.match_web_algs_to_pdf(cert)
             cert.processed.unmatched_algs = output[cert.dgst]
 
@@ -270,8 +274,6 @@ class FIPSDataset(Dataset, ComplexSerializableType):
         for state, cert, algorithms in result:
             self.certs[cert.dgst].state.tables_done = state
             self.certs[cert.dgst].pdf_scan.algorithms += algorithms
-
-        self.match_algs()
         return not_decoded
 
     def remove_algorithms_from_extracted_data(self):
diff --git a/sec_certs/dataset/fips_algorithm.py b/sec_certs/dataset/fips_algorithm.py
index 89b4e4df..0fd39ab0 100644
--- a/sec_certs/dataset/fips_algorithm.py
+++ b/sec_certs/dataset/fips_algorithm.py
@@ -105,7 +105,6 @@ class FIPSAlgorithmDataset(Dataset, ComplexSerializableType):
         return dset
 
     def to_dict(self):
-        print(self.__dict__)
         return self.__dict__ 
 
     def to_json(self, output_path: Union[str, Path]):