Add new CC cert state attributes.

3 files changed, 54 insertions, 31 deletions

diff --git a/sec_certs/sample/common_criteria.py b/sec_certs/sample/common_criteria.py
index 4bdd44ab..a9298eab 100644
--- a/sec_certs/sample/common_criteria.py
+++ b/sec_certs/sample/common_criteria.py
@@ -104,16 +104,18 @@ class CommonCriteriaCert(
     @dataclass(init=False)
     class InternalState(ComplexSerializableType):
         """
-        Holds internal state of the dataset, whether downloads and converts of individual components succeeded. Also
+        Holds internal state of the certificate, whether downloads and converts of individual components succeeded. Also
         holds information about errors and paths to the files.
         """
 
-        st_download_ok: bool
-        report_download_ok: bool
-        st_convert_ok: bool
-        report_convert_ok: bool
-        st_extract_ok: bool
-        report_extract_ok: bool
+        st_download_ok: bool  # Whether target download went OK
+        report_download_ok: bool  # Whether report download went OK
+        st_convert_garbage: bool  # Whether initial target conversion resulted in garbage
+        report_convert_garbage: bool  # Whether initial report conversion resulted in garbage
+        st_convert_ok: bool  # Whether overall target conversion went OK (either pdftotext or via OCR)
+        report_convert_ok: bool  # Whether overall report conversion went OK (either pdftotext or via OCR)
+        st_extract_ok: bool  # Whether target extraction went OK
+        report_extract_ok: bool  # Whether report extraction went OK
 
         errors: List[str]
 
@@ -129,20 +131,25 @@ class CommonCriteriaCert(
 
         def __init__(
             self,
-            st_download_ok: bool = True,
-            report_download_ok: bool = True,
-            st_convert_ok: bool = True,
-            report_convert_ok: bool = True,
-            st_extract_ok: bool = True,
-            report_extract_ok: bool = True,
+            st_download_ok: bool = False,
+            report_download_ok: bool = False,
+            st_convert_garbage: bool = False,
+            report_convert_garbage: bool = False,
+            st_convert_ok: bool = False,
+            report_convert_ok: bool = False,
+            st_extract_ok: bool = False,
+            report_extract_ok: bool = False,
             errors: Optional[List[str]] = None,
             st_pdf_hash: Optional[str] = None,
             report_pdf_hash: Optional[str] = None,
             st_txt_hash: Optional[str] = None,
             report_txt_hash: Optional[str] = None,
         ):
+            super().__init__()
             self.st_download_ok = st_download_ok
             self.report_download_ok = report_download_ok
+            self.st_convert_garbage = st_convert_garbage
+            self.report_convert_garbage = report_convert_garbage
             self.st_convert_ok = st_convert_ok
             self.report_convert_ok = report_convert_ok
             self.st_extract_ok = st_extract_ok
@@ -158,6 +165,8 @@ class CommonCriteriaCert(
             return [
                 "st_download_ok",
                 "report_download_ok",
+                "st_convert_garbage",
+                "report_convert_garbage",
                 "st_convert_ok",
                 "report_convert_ok",
                 "st_extract_ok",
@@ -797,6 +806,7 @@ class CommonCriteriaCert(
             cert.state.report_download_ok = False
             cert.state.errors.append(error_msg)
         else:
+            cert.state.report_download_ok = True
             cert.state.report_pdf_hash = helpers.get_sha256_filepath(cert.state.report_pdf_path)
             cert.pdf_data.report_filename = unquote_plus(str(urlparse(cert.report_link).path).split("/")[-1])
         return cert
@@ -820,6 +830,7 @@ class CommonCriteriaCert(
             cert.state.st_download_ok = False
             cert.state.errors.append(error_msg)
         else:
+            cert.state.st_download_ok = True
             cert.state.st_pdf_hash = helpers.get_sha256_filepath(cert.state.st_pdf_path)
             cert.pdf_data.st_filename = unquote_plus(str(urlparse(cert.st_link).path).split("/")[-1])
         return cert
@@ -839,6 +850,7 @@ class CommonCriteriaCert(
             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
 
@@ -857,6 +869,7 @@ class CommonCriteriaCert(
             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
 
@@ -872,6 +885,8 @@ class CommonCriteriaCert(
         if response != constants.RETURNCODE_OK:
             cert.state.st_extract_ok = False
             cert.state.errors.append(response)
+        else:
+            cert.state.st_extract_ok = True
         return cert
 
     @staticmethod
@@ -886,6 +901,8 @@ class CommonCriteriaCert(
         if response != constants.RETURNCODE_OK:
             cert.state.report_extract_ok = False
             cert.state.errors.append(response)
+        else:
+            cert.state.report_extract_ok = True
         return cert
 
     @staticmethod
diff --git a/tests/data/test_cc_oop/fictional_cert.json b/tests/data/test_cc_oop/fictional_cert.json
index 2b52af75..117f87d2 100644
--- a/tests/data/test_cc_oop/fictional_cert.json
+++ b/tests/data/test_cc_oop/fictional_cert.json
@@ -41,12 +41,14 @@
     },
     "state": {
         "_type": "sec_certs.sample.common_criteria.CommonCriteriaCert.InternalState",
-        "st_download_ok": true,
-        "report_download_ok": true,
-        "st_convert_ok": true,
-        "report_convert_ok": true,
-        "st_extract_ok": true,
-        "report_extract_ok": true,
+        "st_download_ok": false,
+        "report_download_ok": false,
+        "st_convert_garbage": false,
+        "report_convert_garbage": false,
+        "st_convert_ok": false,
+        "report_convert_ok": false,
+        "st_extract_ok": false,
+        "report_extract_ok": false,
         "errors": [],
         "st_pdf_hash": null,
         "report_pdf_hash": null,
diff --git a/tests/data/test_cc_oop/toy_dataset.json b/tests/data/test_cc_oop/toy_dataset.json
index 19f22405..cb733b0c 100644
--- a/tests/data/test_cc_oop/toy_dataset.json
+++ b/tests/data/test_cc_oop/toy_dataset.json
@@ -44,12 +44,14 @@
             },
             "state": {
                 "_type": "sec_certs.sample.common_criteria.CommonCriteriaCert.InternalState",
-                "st_download_ok": true,
-                "report_download_ok": true,
-                "st_convert_ok": true,
-                "report_convert_ok": true,
-                "st_extract_ok": true,
-                "report_extract_ok": true,
+                "st_download_ok": false,
+                "report_download_ok": false,
+                "st_convert_garbage": false,
+                "report_convert_garbage": false,
+                "st_convert_ok": false,
+                "report_convert_ok": false,
+                "st_extract_ok": false,
+                "report_extract_ok": false,
                 "errors": [],
                 "st_pdf_hash": null,
                 "report_pdf_hash": null,
@@ -130,12 +132,14 @@
             },
             "state": {
                 "_type": "sec_certs.sample.common_criteria.CommonCriteriaCert.InternalState",
-                "st_download_ok": true,
-                "report_download_ok": true,
-                "st_convert_ok": true,
-                "report_convert_ok": true,
-                "st_extract_ok": true,
-                "report_extract_ok": true,
+                "st_download_ok": false,
+                "report_download_ok": false,
+                "st_convert_garbage": false,
+                "report_convert_garbage": false,
+                "st_convert_ok": false,
+                "report_convert_ok": false,
+                "st_extract_ok": false,
+                "report_extract_ok": false,
                 "errors": [],
                 "st_pdf_hash": null,
                 "report_pdf_hash": null,