From 5eaa09c10d7d83da48d645b96934d5113bc7ae59 Mon Sep 17 00:00:00 2001
From: Lorenzo Vagliano <lorenzo.ventura.vagliano@cern.ch>
Date: Fri, 25 Oct 2024 14:23:38 +0200
Subject: [PATCH] fix: fetched filetype from article data

Signed-off-by: Lorenzo Vagliano

Also ensured support for different ways to type pdf/a.
---
 scoap3/articles/tasks.py | 10 +++++++++-
 scoap3/tasks.py          | 14 ++++++++++++--
 2 files changed, 21 insertions(+), 3 deletions(-)

diff --git a/scoap3/articles/tasks.py b/scoap3/articles/tasks.py
index 097011acf..7f4cdf0d1 100644
--- a/scoap3/articles/tasks.py
+++ b/scoap3/articles/tasks.py
@@ -57,7 +57,15 @@ def check_required_file_formats(obj):
     for file in obj.related_files.all():
         available_formats.append(file.filetype)
 
-    missing_formats = [f for f in required_formats if f not in available_formats]
+    missing_formats = []
+    for f in required_formats:
+        if f == "pdf/a":
+            if not any(
+                variant in available_formats for variant in ["pdf/a", "pdfa", "pdf_a"]
+            ):
+                missing_formats.append(f)
+        elif f not in available_formats:
+            missing_formats.append(f)
 
     if missing_formats:
         return False, f"Missing required file formats: {', '.join(missing_formats)}."
diff --git a/scoap3/tasks.py b/scoap3/tasks.py
index c9c09ab3c..77a9a2afb 100644
--- a/scoap3/tasks.py
+++ b/scoap3/tasks.py
@@ -144,17 +144,27 @@ def _create_article_file(data, article):
         article_id = article.id
         filename = file.get("key")
         file_path = f"files/{article_id}/{filename}"
+        filetype = file["filetype"]
         article = Article.objects.get(pk=article_id)
-        article_file_data = {"article_id": article, "file": file_path}
+        article_file_data = {
+            "article_id": article,
+            "file": file_path,
+            "filetype": filetype,
+        }
         ArticleFile.objects.get_or_create(**article_file_data)
 
     for file in data.get("files", {}):
         article_id = article.id
         file_path = data["files"][file]
+        filetype = file
         if DEFAULT_STORAGE_PATH:
             file_path = file_path.replace(DEFAULT_STORAGE_PATH, "")
         article = Article.objects.get(pk=article_id)
-        article_file_data = {"article_id": article, "file": file_path}
+        article_file_data = {
+            "article_id": article,
+            "file": file_path,
+            "filetype": filetype,
+        }
         ArticleFile.objects.get_or_create(**article_file_data)