Merge pull request #238 from crocs-muni:fix/document-hashes-constructors

fix CommonCriteriaCert.InternalState constructor

docs/quickstart.md

@@ -61,7 +61,9 @@ It may be handy to create a shared folder between your host machine and the dock
 
 ```bash
 mkdir seccerts-data && \
-docker run -it --mount type=bind,source="$(pwd)"/seccerts-data/,target=/home/user/data seccerts/sec-certs
+docker run -it \
+--mount type=bind,source="$(pwd)"/seccerts-data/,target=/home/user/data \
+seccerts/sec-certs
 ```
 
 The folder should be accessible on your machine on `./seccerts-data` path; from docker on `/home/user/data` path.
@@ -71,7 +73,9 @@ The folder should be accessible on your machine on `./seccerts-data` path; from
 You can also use our docker image to serve `jupyter notebook` instance that you can access from your device. Run
 
 ```bash
-docker run --rm -it -p 8888:8888 seccerts/sec-certs jupyter notebook --no-browser --ip 0.0.0.0 --NotebookApp.token='' --notebook-dir="/home/user/"
+docker run --rm -it -p 8888:8888 \
+seccerts/sec-certs jupyter notebook \
+--no-browser --ip 0.0.0.0 --NotebookApp.token='' --notebook-dir="/home/user/"
 ```
 
 Now, you should be able to access the notebook at `localhost:8888` from your machine. Navigate to `/home/user/sec-certs/notebooks/examples` to see some example notebooks.

sec_certs/sample/common_criteria.py

@@ -95,30 +95,32 @@ class InternalState(ComplexSerializableType):
         report_convert_ok: bool
         st_extract_ok: bool
         report_extract_ok: bool
+
         errors: List[str]
 
-        st_pdf_path: Path
-        report_pdf_path: Path
         st_pdf_hash: Optional[str]
         report_pdf_hash: Optional[str]
-        st_txt_path: Path
-        report_txt_path: Path
         st_txt_hash: Optional[str]
         report_txt_hash: Optional[str]
 
+        st_pdf_path: Path
+        report_pdf_path: Path
+        st_txt_path: Path
+        report_txt_path: Path
+
         def __init__(
             self,
             st_download_ok: bool = True,
             report_download_ok: bool = True,
-            st_pdf_hash: Optional[str] = None,
-            report_pdf_hash: Optional[str] = None,
             st_convert_ok: bool = True,
             report_convert_ok: bool = True,
-            st_txt_hash: Optional[str] = None,
-            report_txt_hash: Optional[str] = None,
             st_extract_ok: bool = True,
             report_extract_ok: bool = True,
             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,
         ):
             self.st_download_ok = st_download_ok
             self.report_download_ok = report_download_ok
@@ -127,21 +129,25 @@ def __init__(
             self.st_extract_ok = st_extract_ok
             self.report_extract_ok = report_extract_ok
             self.errors = errors if errors else []
+            self.st_pdf_hash = st_pdf_hash
+            self.report_pdf_hash = report_pdf_hash
+            self.st_txt_hash = st_txt_hash
+            self.report_txt_hash = report_txt_hash
 
         @property
         def serialized_attributes(self) -> List[str]:
             return [
                 "st_download_ok",
                 "report_download_ok",
-                "st_pdf_hash",
-                "report_pdf_hash",
                 "st_convert_ok",
                 "report_convert_ok",
-                "st_txt_hash",
-                "report_txt_hash",
                 "st_extract_ok",
                 "report_extract_ok",
                 "errors",
+                "st_pdf_hash",
+                "report_pdf_hash",
+                "st_txt_hash",
+                "report_txt_hash",
             ]
 
         def report_is_ok_to_download(self, fresh: bool = True) -> bool:

tests/data/test_cc_oop/fictional_cert.json

@@ -42,7 +42,11 @@
         "report_convert_ok": true,
         "st_extract_ok": true,
         "report_extract_ok": true,
-        "errors": []
+        "errors": [],
+        "st_pdf_hash": null,
+        "report_pdf_hash": null,
+        "st_txt_hash": null,
+        "report_txt_hash": null
     },
     "pdf_data": {
         "_type": "PdfData",

tests/data/test_cc_oop/toy_dataset.json

@@ -49,7 +49,11 @@
                 "report_convert_ok": true,
                 "st_extract_ok": true,
                 "report_extract_ok": true,
-                "errors": []
+                "errors": [],
+                "st_pdf_hash": null,
+                "report_pdf_hash": null,
+                "st_txt_hash": null,
+                "report_txt_hash": null
             },
             "pdf_data": {
                 "_type": "PdfData",
@@ -126,7 +130,11 @@
                 "report_convert_ok": true,
                 "st_extract_ok": true,
                 "report_extract_ok": true,
-                "errors": []
+                "errors": [],
+                "st_pdf_hash": null,
+                "report_pdf_hash": null,
+                "st_txt_hash": null,
+                "report_txt_hash": null
             },
             "pdf_data": {
                 "_type": "PdfData",

tests/test_cc_oop.py

@@ -9,7 +9,6 @@
 from unittest import TestCase
 
 import sec_certs.constants as constants
-import sec_certs.helpers as helpers
 import tests.data.test_cc_oop
 from sec_certs.config.configuration import config
 from sec_certs.dataset import CCDataset
@@ -129,23 +128,18 @@ def test_download_and_convert_pdfs(self):
             dset.download_all_pdfs()
             dset.convert_all_pdfs()
 
-            actual_report_pdf_hashes = {
-                key: helpers.get_sha256_filepath(val.state.report_pdf_path) for key, val in dset.certs.items()
-            }
-            actual_target_pdf_hashes = {
-                key: helpers.get_sha256_filepath(val.state.st_pdf_path) for key, val in dset.certs.items()
-            }
+            for cert in dset:
+                self.assertEqual(
+                    cert.state.report_pdf_hash,
+                    self.template_report_pdf_hashes[cert.dgst],
+                    f"Hash of PDF certification report for CommonCriteriaCert with digest {cert.dgst} deviates from template.",
+                )
 
-            self.assertEqual(
-                actual_report_pdf_hashes,
-                self.template_report_pdf_hashes,
-                "Hashes of downloaded pdfs (sample report) do not the template",
-            )
-            self.assertEqual(
-                actual_target_pdf_hashes,
-                self.template_target_pdf_hashes,
-                "Hashes of downloaded pdfs (security target) do not match the template",
-            )
+                self.assertEqual(
+                    cert.state.st_pdf_hash,
+                    self.template_target_pdf_hashes[cert.dgst],
+                    f"Hash of PDF security target for CommonCriteriaCert with digest {cert.dgst} deviates from template.",
+                )
 
             self.assertTrue(dset["309ac2fd7f2dcf17"].state.report_txt_path.exists())
             self.assertTrue(dset["309ac2fd7f2dcf17"].state.st_txt_path.exists())