From 97f68d8323fdb756cf32795f2cdd1993bfe4ad35 Mon Sep 17 00:00:00 2001
From: Simon K <6615834+simon-20@users.noreply.github.com>
Date: Thu, 1 Aug 2024 16:59:49 +0100
Subject: [PATCH 1/2] fix: falsey tests for var which can be None

* Tests for falsity of file_schema_valid was changed in response to flake8 warnings about use of '==' , but recommended action was not semantically equivalent--this commit restores original behaviour and passes flake8 checks.
---
 src/library/validate.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/src/library/validate.py b/src/library/validate.py
index cca0f28..f8b0be8 100644
--- a/src/library/validate.py
+++ b/src/library/validate.py
@@ -33,7 +33,7 @@ def process_hash_list(document_datasets):
             file_schema_valid = file_data[6]
             publisher_black_flag = file_data[7] is not None
 
-            if not file_schema_valid and downloaded > (
+            if file_schema_valid is False and downloaded > (
                 now - timedelta(hours=config["VALIDATION"]["SAFETY_CHECK_PERIOD"])
             ):
                 logger.info(
@@ -43,7 +43,7 @@ def process_hash_list(document_datasets):
                 )
                 continue
 
-            if not file_schema_valid and publisher_black_flag:
+            if file_schema_valid is False and publisher_black_flag:
                 logger.info(
                     "Skipping Schema Invalid file for Full Validation since publisher: "
                     f"{publisher} is black flagged for hash: {file_hash} and id: {file_id}"
@@ -115,7 +115,7 @@ def process_hash_list(document_datasets):
                     )
                     continue
 
-            if not file_schema_valid and downloaded > (
+            if file_schema_valid is False and downloaded > (
                 now - timedelta(hours=config["VALIDATION"]["SAFETY_CHECK_PERIOD"])
             ):
                 logger.info(
@@ -125,7 +125,7 @@ def process_hash_list(document_datasets):
                 )
                 continue
 
-            if not file_schema_valid and publisher_black_flag:
+            if file_schema_valid is False and publisher_black_flag:
                 logger.info(
                     f"Skipping Schema Invalid file for Full Validation since publisher: {publisher} "
                     f"is flagged for hash: {file_hash} and id: {file_id}"

From c05999a090b65e8cdaa6674ecebda0bed45f6ef9 Mon Sep 17 00:00:00 2001
From: Simon K <6615834+simon-20@users.noreply.github.com>
Date: Thu, 1 Aug 2024 17:37:23 +0100
Subject: [PATCH 2/2] fix: change tests to check explicitly for True, False

* These tests were changed to use 'variable' and 'not variable' in response to flake8 warnings, but should have used 'is True', 'is False', because the variable can be None
---
 src/library/validate.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/src/library/validate.py b/src/library/validate.py
index f8b0be8..71f046d 100644
--- a/src/library/validate.py
+++ b/src/library/validate.py
@@ -43,7 +43,7 @@ def process_hash_list(document_datasets):
                 )
                 continue
 
-            if file_schema_valid is False and publisher_black_flag:
+            if file_schema_valid is False and publisher_black_flag is True:
                 logger.info(
                     "Skipping Schema Invalid file for Full Validation since publisher: "
                     f"{publisher} is black flagged for hash: {file_hash} and id: {file_id}"
@@ -104,7 +104,7 @@ def process_hash_list(document_datasets):
                         )
                 try:
                     body = schema_response.json()
-                    if body["valid"] or not body["valid"]:
+                    if body["valid"] is True or body["valid"] is False:
                         db.updateDocumentSchemaValidationStatus(conn, file_id, body["valid"])
                         file_schema_valid = body["valid"]
                     else:
@@ -125,7 +125,7 @@ def process_hash_list(document_datasets):
                 )
                 continue
 
-            if file_schema_valid is False and publisher_black_flag:
+            if file_schema_valid is False and publisher_black_flag is True:
                 logger.info(
                     f"Skipping Schema Invalid file for Full Validation since publisher: {publisher} "
                     f"is flagged for hash: {file_hash} and id: {file_id}"
@@ -141,7 +141,7 @@ def process_hash_list(document_datasets):
             full_url = config["VALIDATION"]["FULL_VALIDATION_URL"]
 
             # only need meta=true for invalid files to "clean" them later
-            if not file_schema_valid:
+            if file_schema_valid is False:
                 full_url += "?meta=true"
             full_response = requests.post(
                 full_url,