feat: add optional dms integration

This adds the feature to be able to send docx, odt to DMS to be
converted to pdf and saved in alexandria.

README.md

@@ -112,11 +112,17 @@ A list of configuration options which you need
     To use SSE-C in development make sure to generate a certificate for the minio container and set `ALEXANDRIA_S3_VERIFY` to `false`.
 
 - ClamAV
-- `ALEXANDRIA_CLAMD_ENABLED`: Set this to `True` to enable ClamAV (virus scanner).
-- `ALEXANDRIA_CLAMD_SOCKET`: ClamAV socket
-- `ALEXANDRIA_CLAMD_USE_TCP`: Use TCP to connect to ClamAV service
-- `ALEXANDRIA_CLAMD_TCP_SOCKET`: ClamAV service socket
-- `ALEXANDRIA_CLAMD_TCP_ADDR`: ClamAV service address
+  - `ALEXANDRIA_CLAMD_ENABLED`: Set this to `True` to enable ClamAV (virus scanner).
+  - `ALEXANDRIA_CLAMD_SOCKET`: ClamAV socket
+  - `ALEXANDRIA_CLAMD_USE_TCP`: Use TCP to connect to ClamAV service
+  - `ALEXANDRIA_CLAMD_TCP_SOCKET`: ClamAV service socket
+  - `ALEXANDRIA_CLAMD_TCP_ADDR`: ClamAV service address
+
+- [Document Merge Service](https://github.com/adfinis/document-merge-service)
+Enable conversion of docx/odt files to pdf directly in alexandria.
+  - `ALEXANDRIA_ENABLE_PDF_CONVERSION`: Set this to `True` to enable the pdf conversion endpoint.
+  - `ALEXANDRIA_DMS_URL`: URL where the document merge service is running
+
 
 For development, you can also set the following environemnt variables to help you:
 

alexandria/core/tests/test_views.py

@@ -13,6 +13,7 @@
     HTTP_200_OK,
     HTTP_201_CREATED,
     HTTP_400_BAD_REQUEST,
+    HTTP_401_UNAUTHORIZED,
     HTTP_403_FORBIDDEN,
     HTTP_404_NOT_FOUND,
 )
@@ -441,3 +442,58 @@ def test_presigned_url_different_file(admin_client, file, file_factory):
 
     response = admin_client.get(url)
     assert response.status_code == HTTP_400_BAD_REQUEST
+
+
+def test_convert_document(
+    admin_client, document_factory, file_factory, settings, mocker
+):
+    settings.ALEXANDRIA_ENABLE_PDF_CONVERSION = True
+    document = document_factory()
+    file_factory(document=document, name="foo")
+
+    response = mocker.Mock()
+    response.status_code = HTTP_200_OK
+    response.content = b"pdfdata"
+    mocker.patch("requests.post", return_value=response)
+    url = reverse("document-convert", args=[document.pk])
+    response = admin_client.get(url)
+
+    assert response.status_code == HTTP_200_OK
+
+    assert Document.objects.all().count() == 2
+    assert File.objects.all().count() == 3
+    assert File.objects.filter(name="foo.pdf", variant=File.Variant.ORIGINAL).exists()
+
+
+def test_convert_document_not_enabled(
+    admin_client, document_factory, file_factory, settings, mocker
+):
+    settings.ALEXANDRIA_ENABLE_PDF_CONVERSION = False
+    document = document_factory()
+    file_factory(document=document, name="foo")
+
+    response = mocker.Mock()
+    response.status_code = HTTP_200_OK
+    response.content = b"pdfdata"
+    mocker.patch("requests.post", return_value=response)
+    url = reverse("document-convert", args=[document.pk])
+    response = admin_client.get(url)
+
+    assert response.status_code == HTTP_400_BAD_REQUEST
+
+
+def test_convert_document_token_expired(
+    admin_client, document_factory, file_factory, settings, mocker
+):
+    settings.ALEXANDRIA_ENABLE_PDF_CONVERSION = True
+    document = document_factory()
+    file_factory(document=document, name="foo")
+
+    response = mocker.Mock()
+    response.status_code = HTTP_401_UNAUTHORIZED
+    response.content = b"pdfdata"
+    mocker.patch("requests.post", return_value=response)
+    url = reverse("document-convert", args=[document.pk])
+    response = admin_client.get(url)
+
+    assert response.status_code == HTTP_401_UNAUTHORIZED

alexandria/core/views.py

@@ -4,22 +4,30 @@
 import zipfile
 from tempfile import NamedTemporaryFile
 
+import requests
 from django.conf import settings
 from django.core.exceptions import ValidationError as DjangoCoreValidationError
+from django.core.files.base import ContentFile
 from django.http import FileResponse
 from django.utils.translation import gettext as _
 from generic_permissions.permissions import AllowAny, PermissionViewMixin
 from generic_permissions.visibilities import VisibilityViewMixin
+from rest_framework.authentication import get_authorization_header
 from rest_framework.decorators import action, permission_classes
-from rest_framework.exceptions import NotFound, PermissionDenied, ValidationError
+from rest_framework.exceptions import (
+    AuthenticationFailed,
+    NotFound,
+    PermissionDenied,
+    ValidationError,
+)
 from rest_framework.mixins import (
     CreateModelMixin,
     DestroyModelMixin,
     ListModelMixin,
     RetrieveModelMixin,
 )
 from rest_framework.response import Response
-from rest_framework.status import HTTP_201_CREATED
+from rest_framework.status import HTTP_200_OK, HTTP_201_CREATED, HTTP_401_UNAUTHORIZED
 from rest_framework.viewsets import GenericViewSet
 from rest_framework_json_api.views import (
     AutoPrefetchMixin,
@@ -100,6 +108,44 @@ def update(self, request, *args, **kwargs):
 
         return response
 
+    @action(methods=["get"], detail=True)
+    def convert(self, request, pk=None):
+        if not settings.ALEXANDRIA_ENABLE_PDF_CONVERSION:
+            raise ValidationError(_("PDF conversion is not enabled."))
+
+        document = self.get_object()
+        file = document.get_latest_original()
+
+        response = requests.post(
+            settings.ALEXANDRIA_DMS_URL + "/convert",
+            data={"target_format": "pdf"},
+            headers={"authorization": get_authorization_header(request)},
+            files={"file": file.content},
+        )
+
+        if response.status_code == HTTP_401_UNAUTHORIZED:
+            raise AuthenticationFailed(_("Token has expired."))
+
+        response.raise_for_status()
+
+        converted_document = models.Document.objects.create(
+            title={k: v + ".pdf" for k, v in document.title.items()},
+            description=document.description,
+            category=document.category,
+            date=document.date,
+        )
+        file_name = file.name + ".pdf"
+        converted_file = models.File.objects.create(
+            document=converted_document,
+            name=file_name,
+            content=ContentFile(response.content, file_name),
+            mime_type="application/pdf",
+            size=len(response.content),
+        )
+        converted_file.create_thumbnail()
+
+        return Response(status=HTTP_200_OK)
+
 
 class FileViewSet(
     PermissionViewMixin,

alexandria/settings/alexandria.py

@@ -167,3 +167,9 @@ def default(default_dev=env.NOTSET, default_prod=env.NOTSET):
 CLAMD_TCP_SOCKET = env.str("ALEXANDRIA_CLAMD_TCP_SOCKET", default=3310)
 CLAMD_TCP_ADDR = env.str("ALEXANDRIA_CLAMD_TCP_ADDR", default="localhost")
 CLAMD_ENABLED = env.bool("ALEXANDRIA_CLAMD_ENABLED", default=False)
+
+# Document merge service
+ALEXANDRIA_ENABLE_PDF_CONVERSION = env.bool(
+    "ALEXANDRIA_ENABLE_PDF_CONVERSION", default=False
+)
+ALEXANDRIA_DMS_URL = env.str("ALEXANDRIA_DMS_URL", default="http://dms:8000/api/v1")

docker-compose.yml

@@ -26,6 +26,21 @@ services:
       # https://docs.djangoproject.com/en/2.1/ref/settings/#password
       # same as postgres password above
       # - DATABASE_PASSWORD=
+  dms:
+    image: ghcr.io/adfinis/document-merge-service:6.4.4
+    depends_on:
+      - db
+    environment:
+      - DATABASE_ENGINE=django.db.backends.postgresql
+      - DATABASE_HOST=db
+      - DATABASE_PORT=5432
+      - DATABASE_NAME=alexandria
+      - DATABASE_USER=alexandria
+      - DATABASE_PASSWORD=alexandria
+      - ALLOWED_HOSTS=dms
+      - OIDC_BEARER_TOKEN_REVALIDATION_TIME=300
+      - SECRET_KEY=aaa
+      - DOCXTEMPLATE_JINJA_EXTENSIONS=
   minio:
     image: minio/minio:RELEASE.2023-11-06T22-26-08Z
     volumes: