Store OCR results into CC cert state.

4 files changed, 32 insertions, 18 deletions

diff --git a/sec_certs/sample/common_criteria.py b/sec_certs/sample/common_criteria.py
index a9298eab..b3ccb8db 100644
--- a/sec_certs/sample/common_criteria.py
+++ b/sec_certs/sample/common_criteria.py
@@ -843,14 +843,18 @@ class CommonCriteriaCert(
         :param CommonCriteriaCert cert: cert to download the pdf report for
         :return CommonCriteriaCert: the modified certificate with updated state
         """
-        exit_code = sec_certs.utils.pdf.convert_pdf_file(cert.state.report_pdf_path, cert.state.report_txt_path)
-        if exit_code != constants.RETURNCODE_OK:
+        ocr_done, ok_result = sec_certs.utils.pdf.convert_pdf_file(
+            cert.state.report_pdf_path, cert.state.report_txt_path
+        )
+        # If OCR was done the result was garbage
+        cert.state.report_convert_garbage = ocr_done
+        # And put the whole result into convert_ok
+        cert.state.report_convert_ok = ok_result
+        if not ok_result:
             error_msg = "failed to convert report pdf->txt"
             logger.error(f"Cert dgst: {cert.dgst} " + error_msg)
-            cert.state.report_convert_ok = False
             cert.state.errors.append(error_msg)
         else:
-            cert.state.report_convert_ok = True
             cert.state.report_txt_hash = helpers.get_sha256_filepath(cert.state.report_txt_path)
         return cert
 
@@ -862,14 +866,16 @@ class CommonCriteriaCert(
         :param CommonCriteriaCert cert: cert to download the pdf security target for
         :return CommonCriteriaCert: the modified certificate with updated state
         """
-        exit_code = sec_certs.utils.pdf.convert_pdf_file(cert.state.st_pdf_path, cert.state.st_txt_path)
-        if exit_code != constants.RETURNCODE_OK:
+        ocr_done, ok_result = sec_certs.utils.pdf.convert_pdf_file(cert.state.st_pdf_path, cert.state.st_txt_path)
+        # If OCR was done the result was garbage
+        cert.state.st_convert_garbage = ocr_done
+        # And put the whole result into convert_ok
+        cert.state.st_convert_ok = ok_result
+        if not ok_result:
             error_msg = "failed to convert security target pdf->txt"
             logger.error(f"Cert dgst: {cert.dgst} " + error_msg)
-            cert.state.st_convert_ok = False
             cert.state.errors.append(error_msg)
         else:
-            cert.state.st_convert_ok = True
             cert.state.st_txt_hash = helpers.get_sha256_filepath(cert.state.st_txt_path)
         return cert
 
diff --git a/sec_certs/sample/fips.py b/sec_certs/sample/fips.py
index cf45aafb..eb8308f5 100644
--- a/sec_certs/sample/fips.py
+++ b/sec_certs/sample/fips.py
@@ -559,8 +559,8 @@ class FIPSCertificate(Certificate["FIPSCertificate", "FIPSCertificate.Heuristics
         """
         cert, pdf_path, txt_path = tup
         if not cert.state.txt_state:
-            exit_code = sec_certs.utils.pdf.convert_pdf_file(pdf_path, txt_path)
-            if exit_code != constants.RETURNCODE_OK:
+            ocr_done, ok_result = sec_certs.utils.pdf.convert_pdf_file(pdf_path, txt_path)
+            if not ok_result:
                 logger.error(f"Cert dgst: {cert.cert_id} failed to convert security policy pdf->txt")
                 cert.state.txt_state = False
             else:
diff --git a/sec_certs/utils/pdf.py b/sec_certs/utils/pdf.py
index 08de2cc7..a40fb18d 100644
--- a/sec_certs/utils/pdf.py
+++ b/sec_certs/utils/pdf.py
@@ -64,15 +64,18 @@ def ocr_pdf_file(pdf_path: Path) -> str:
     return contents
 
 
-def convert_pdf_file(pdf_path: Path, txt_path: Path) -> str:
+def convert_pdf_file(pdf_path: Path, txt_path: Path) -> Tuple[bool, bool]:
     """
     Convert a PDF tile to text and save it on the `txt_path`.
 
     :param pdf_path: Path to the to-be-converted PDF file.
     :param txt_path: Path to the resulting text file.
-    :return: Whether the conversion was successful (see constants).
+    :return: A tuple of two results, whether OCR was done and what the complete result
+             was (OK/NOK).
     """
     txt = None
+    ok = False
+    ocr = False
     try:
         with pdf_path.open("rb") as pdf_handle:
             pdf = pdftotext.PDF(pdf_handle, "", True)  # No password, Raw=True
@@ -82,19 +85,19 @@ def convert_pdf_file(pdf_path: Path, txt_path: Path) -> str:
 
     if txt is None or text_is_garbage(txt):
         logger.warning(f"Detected garbage during conversion of {pdf_path}")
+        ocr = True
         try:
             txt = ocr_pdf_file(pdf_path)
             logger.info(f"OCR OK for {pdf_path}")
         except Exception as e:
             logger.error(f"Error during OCR of {pdf_path}, using garbage: {e}")
 
-    if txt is None:
-        return constants.RETURNCODE_NOK
+    if txt is not None:
+        ok = True
+        with txt_path.open("w", encoding="utf-8") as txt_handle:
+            txt_handle.write(txt)
 
-    with txt_path.open("w", encoding="utf-8") as txt_handle:
-        txt_handle.write(txt)
-
-    return constants.RETURNCODE_OK
+    return ocr, ok
 
 
 def parse_pdf_date(dateval: Optional[bytes]) -> Optional[datetime]:
diff --git a/tests/test_cc_oop.py b/tests/test_cc_oop.py
index 201b3db3..0777a1e1 100644
--- a/tests/test_cc_oop.py
+++ b/tests/test_cc_oop.py
@@ -142,6 +142,11 @@ class TestCommonCriteriaOOP(TestCase):
                     f"Hash of PDF security target for CommonCriteriaCert with digest {cert.dgst} deviates from template.",
                 )
 
+            self.assertFalse(dset["309ac2fd7f2dcf17"].state.report_convert_garbage)
+            self.assertFalse(dset["309ac2fd7f2dcf17"].state.st_convert_garbage)
+            self.assertTrue(dset["309ac2fd7f2dcf17"].state.report_convert_ok)
+            self.assertTrue(dset["309ac2fd7f2dcf17"].state.st_convert_ok)
+
             self.assertTrue(dset["309ac2fd7f2dcf17"].state.report_txt_path.exists())
             self.assertTrue(dset["309ac2fd7f2dcf17"].state.st_txt_path.exists())