unovonv is deprecated, use unoserver (Fixes #332) (#353)

Author: C4ptainCrunch

documents/tasks.py

@@ -132,38 +132,43 @@ def checksum(self, document_id: int) -> int:
 def convert_office_to_pdf(self, document_id: int) -> int:
     document = Document.objects.get(pk=document_id)
 
-    try:
+    if settings.DEBUG:
         # Check if unoserver is running
         ping_result = subprocess.run(
             ["unoping"], capture_output=True, timeout=5, check=False
         )
 
         if ping_result.returncode != 0:
             # Server not running, start it as a daemon
-            # Note: --daemon causes the process to fork, so we use Popen and don't wait
-            subprocess.Popen(
-                ["unoserver", "--daemon", "--conversion-timeout", "300"],
-                stdout=subprocess.DEVNULL,
-                stderr=subprocess.DEVNULL,
-            )
+            # Here we want to use the system unoserver, as it needs access to LibreOffice
+            try:
+                subprocess.Popen(
+                    [
+                        f"{os.environ['HOME']}/.local/bin/unoserver",
+                        "--daemon",
+                        "--conversion-timeout",
+                        "300",
+                    ],
+                    stdout=subprocess.DEVNULL,
+                    stderr=subprocess.DEVNULL,
+                )
+            except FileNotFoundError as e:
+                raise MissingBinary("unoserver") from e
             # Give the server time to start up and be ready
             time.sleep(2)
 
-        try:
-            result = subprocess.run(
-                ["unoconvert", "-", "-", "--convert-to", "pdf"],
-                input=document.original.read(),
-                capture_output=True,
-                check=True,
-            )
-            sub = result.stdout
-        except subprocess.CalledProcessError as e:
-            raise DocumentProcessingError(
-                document, exc=e, message='"unoconvert" has failed: %s' % e.output[:800]
-            ) from e
-
-    except FileNotFoundError as e:
-        raise MissingBinary("unoserver") from e
+    try:
+        result = subprocess.run(
+            ["unoconvert", "-", "-", "--convert-to", "pdf"],
+            input=document.original.read(),
+            capture_output=True,
+            check=True,
+        )
+        sub = result.stdout
+    except subprocess.CalledProcessError as e:
+        raise DocumentProcessingError(
+            document, exc=e, message='"unoconvert" has failed: %s' % e.stderr[:800]
+        ) from e
 
     document.pdf.save(str(uuid.uuid4()) + ".pdf", ContentFile(sub))
 

documents/templates/documents/viewer.html

@@ -36,7 +36,7 @@ <h1 class="fs-4 mb-0 d-flex align-items-center gap-1">
 
         <div class="d-sm-flex gap-2 align-items-center">
             <div class="mt-2  d-flex gap-2 align-items-center">
-                {% if document.state == "DONE" %}
+                {% if document.state == "DONE" and not document.is_unconvertible %}
                     <a class="btn btn-primary btn-sm d-inline-flex align-items-center gap-1" data-turbo="false"
                        href="{% url 'document_pdf' document.pk %}">
                         <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" fill="currentColor"
@@ -140,7 +140,14 @@ <h1 class="fs-4 mb-0 d-flex align-items-center gap-1">
 {% endblock header %}
 
 {% block content %}
-    {% if document.state == "DONE" %}
+    {% if document.is_unconvertible %}
+        <div class="alert alert-primary" role="alert">
+            DocHub ne sait pas générer d'aperçu pour ce document.
+            Mais tu peux le
+            <a href="{% url 'document_original' document.pk %}">télécharger</a>
+            et l'ouvrir directement chez toi.
+        </div>
+    {% elif document.state == "DONE" %}
         <div class="container-xl" data-controller="viewer"
              data-viewer-src-value="{% url 'document_pdf' document.pk %}?embed">
             <div data-viewer-target="sidebar">
@@ -165,20 +172,12 @@ <h1 class="fs-4 mb-0 d-flex align-items-center gap-1">
 
         </div>
     {% elif document.state == "ERROR" %}
-        {% if document.is_unconvertible %}
-            <div class="alert alert-primary" role="alert">
-                DocHub ne sait pas générer d'aperçu pour ce document.
-                Mais tu peux le
-                <a href="{% url 'document_original' document.pk %}">télécharger</a>
-                et l'ouvrir directement chez toi.
-            </div>
-        {% else %}
-            <div class="alert alert-primary" role="alert">
-                Ce document n'a pas pu être traité par DocHub à cause d'une erreur.
-                Mais tu peux <a href="{% url 'document_original' document.pk %}">télécharger
-                    la version originale </a> si tu le désires.
-            </div>
-        {% endif %}
+
+        <div class="alert alert-primary" role="alert">
+            Ce document n'a pas pu être traité par DocHub à cause d'une erreur.
+            Mais tu peux <a href="{% url 'document_original' document.pk %}">télécharger
+                la version originale </a> si tu le désires.
+        </div>
     {% else %}
         <div class="alert alert-primary" role="alert">
             Ce document est en cours de traitement par DocHub et il n'est donc pas encore

documents/tests/celery_test.py

@@ -1,3 +1,7 @@
+import logging
+import os
+import subprocess
+
 from django.core.files import File
 
 import celery
@@ -8,6 +12,8 @@
 from documents.tasks import mutool_get_pages, process_document
 from users.models import User
 
+logger = logging.getLogger(__name__)
+
 pytestmark = [pytest.mark.django_db, pytest.mark.celery]
 
 
@@ -55,10 +61,53 @@ def test_send_duplicate():
     assert Document.objects.filter(id=doc.id).count() == 0
 
 
+@pytest.fixture
+def unoserver():
+    # Check if unoserver is running
+    ping_result = subprocess.run(
+        ["unoping"], capture_output=True, timeout=5, check=False
+    )
+
+    if ping_result.returncode != 0:
+        logger.debug("Unoserver is not running, starting it ourselves")
+        sub = subprocess.Popen(
+            [
+                f"{os.environ['HOME']}/.local/bin/unoserver",
+                "--daemon",
+                "--conversion-timeout",
+                "300",
+            ],
+            stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE,
+        )
+        logger.debug("Unoserver started")
+
+        yield
+
+        logger.debug("Killing unoserver")
+        sub.kill()
+
+        # Get stdout and stderr
+        stdout, stderr = sub.communicate(timeout=5)
+
+        # Log them
+        if stdout:
+            logger.info(
+                "unoserver stdout:\n%s", stdout.decode("utf-8", errors="replace")
+            )
+        if stderr:
+            logger.error(
+                "unoserver stderr:\n%s", stderr.decode("utf-8", errors="replace")
+            )
+    else:
+        logger.debug("Unoserver is already running")
+        yield
+
+
 # TODO : mock unoserver and provide a fake pdf instead
 @pytest.mark.unoserver
 @pytest.mark.slow
-def test_send_office():
+def test_send_office(unoserver):
     doc = create_doc("My office doc", ".docx")
 
     with open("documents/tests/files/2pages.docx", "rb") as fd:

pyproject.toml

@@ -37,6 +37,7 @@ dependencies = [
     "psycopg[binary,pool]>=3.2.6",
     "python-memcached>=1.62",
     "redis>=5.2.1",
+    "unoserver==3.6",
 ]
 
 
@@ -65,6 +66,9 @@ dev = [
 norecursedirs = ".venv ve ve3 static media .git node_modules"
 DJANGO_SETTINGS_MODULE="www.test_settings"
 addopts = "--reuse-db"
+filterwarnings = [
+    "ignore:builtin type.*has no __module__ attribute:DeprecationWarning"
+]
 markers = """
     slow: marks tests as slow (deselect with '-m "not slow"')
     network: marks tests using the network (deselect with '-m "not network"')