Merge remote-tracking branch 'upstream/dev' into feature-consume-eml

src/documents/templates/index.html

@@ -76,7 +76,7 @@
 					</g>
 				</svg>
 				<h6 class="m-auto">{% translate "Paperless-ngx is loading..." %}</h6>
-				<p class="warning m-auto mt-3 small fade hide">{% translate "Still here?! Hmm, something might be wrong." %} <a href="https://paperless-ngx.readthedocs.io/en/latest/">{% translate "Here's a link to the docs." %}</a></p>
+				<p class="warning m-auto mt-3 small fade hide">{% translate "Still here?! Hmm, something might be wrong." %} <a href="https://docs.paperless-ngx.com">{% translate "Here's a link to the docs." %}</a></p>
 			</div>
 		</div>
 	</app-root>

src/locale/no_NO/LC_MESSAGES/django.po

@@ -3,7 +3,7 @@ msgstr ""
 "Project-Id-Version: paperless-ngx\n"
 "Report-Msgid-Bugs-To: \n"
 "POT-Creation-Date: 2022-11-09 21:50+0000\n"
-"PO-Revision-Date: 2022-11-09 23:11\n"
+"PO-Revision-Date: 2022-11-29 08:29\n"
 "Last-Translator: \n"
 "Language-Team: Norwegian\n"
 "Language: no_NO\n"
@@ -184,11 +184,11 @@ msgstr "Gjeldende arkiv filnavn i lagring"
 
 #: documents/models.py:221
 msgid "original filename"
-msgstr ""
+msgstr "opprinnelig filnavn"
 
 #: documents/models.py:227
 msgid "The original name of the file when it was uploaded"
-msgstr ""
+msgstr "Det opprinnelige filnavnet da den ble lastet opp"
 
 #: documents/models.py:231
 msgid "archive serial number"
@@ -368,15 +368,15 @@ msgstr "har tags i"
 
 #: documents/models.py:410
 msgid "ASN greater than"
-msgstr ""
+msgstr "ASN større enn"
 
 #: documents/models.py:411
 msgid "ASN less than"
-msgstr ""
+msgstr "ASN mindre enn"
 
 #: documents/models.py:412
 msgid "storage path is"
-msgstr ""
+msgstr "lagringssti er"
 
 #: documents/models.py:422
 msgid "rule type"
@@ -396,31 +396,31 @@ msgstr "filtrer regler"
 
 #: documents/models.py:536
 msgid "Task ID"
-msgstr ""
+msgstr "Oppgave ID"
 
 #: documents/models.py:537
 msgid "Celery ID for the Task that was run"
-msgstr ""
+msgstr "Celery ID for oppgaven som ble kjørt"
 
 #: documents/models.py:542
 msgid "Acknowledged"
-msgstr ""
+msgstr "Bekreftet"
 
 #: documents/models.py:543
 msgid "If the task is acknowledged via the frontend or API"
-msgstr ""
+msgstr "Hvis oppgaven bekreftes via frontend eller API"
 
 #: documents/models.py:549 documents/models.py:556
 msgid "Task Name"
-msgstr ""
+msgstr "Oppgavenavn"
 
 #: documents/models.py:550
 msgid "Name of the file which the Task was run for"
-msgstr ""
+msgstr "Navn på filen som oppgaven ble kjørt for"
 
 #: documents/models.py:557
 msgid "Name of the Task which was run"
-msgstr ""
+msgstr "Navn på Oppgaven som ble kjørt"
 
 #: documents/models.py:562
 msgid "Task Positional Arguments"

src/locale/pl_PL/LC_MESSAGES/django.po

@@ -3,7 +3,7 @@ msgstr ""
 "Project-Id-Version: paperless-ngx\n"
 "Report-Msgid-Bugs-To: \n"
 "POT-Creation-Date: 2022-11-09 21:50+0000\n"
-"PO-Revision-Date: 2022-11-09 23:11\n"
+"PO-Revision-Date: 2022-11-28 16:30\n"
 "Last-Translator: \n"
 "Language-Team: Polish\n"
 "Language: pl_PL\n"
@@ -184,11 +184,11 @@ msgstr "Aktualna nazwa pliku archiwum w pamięci"
 
 #: documents/models.py:221
 msgid "original filename"
-msgstr ""
+msgstr "oryginalna nazwa pliku"
 
 #: documents/models.py:227
 msgid "The original name of the file when it was uploaded"
-msgstr ""
+msgstr "Oryginalna nazwa pliku, gdy został przesłany"
 
 #: documents/models.py:231
 msgid "archive serial number"
@@ -368,15 +368,15 @@ msgstr "ma znaczniki w"
 
 #: documents/models.py:410
 msgid "ASN greater than"
-msgstr ""
+msgstr "ASN większy niż"
 
 #: documents/models.py:411
 msgid "ASN less than"
-msgstr ""
+msgstr "ASN mniejszy niż"
 
 #: documents/models.py:412
 msgid "storage path is"
-msgstr ""
+msgstr "ścieżką przechowywania jest"
 
 #: documents/models.py:422
 msgid "rule type"
@@ -396,23 +396,23 @@ msgstr "reguły filtrowania"
 
 #: documents/models.py:536
 msgid "Task ID"
-msgstr ""
+msgstr "ID zadania"
 
 #: documents/models.py:537
 msgid "Celery ID for the Task that was run"
-msgstr ""
+msgstr "ID Celery dla zadania, które zostało uruchomione"
 
 #: documents/models.py:542
 msgid "Acknowledged"
-msgstr ""
+msgstr "Potwierdzono"
 
 #: documents/models.py:543
 msgid "If the task is acknowledged via the frontend or API"
-msgstr ""
+msgstr "Jeśli zadanie jest potwierdzone przez frontend lub API"
 
 #: documents/models.py:549 documents/models.py:556
 msgid "Task Name"
-msgstr ""
+msgstr "Nazwa zadania"
 
 #: documents/models.py:550
 msgid "Name of the file which the Task was run for"
@@ -440,7 +440,7 @@ msgstr ""
 
 #: documents/models.py:578
 msgid "Task State"
-msgstr ""
+msgstr "Stan zadania"
 
 #: documents/models.py:579
 msgid "Current state of the task being run"
@@ -476,19 +476,19 @@ msgstr ""
 
 #: documents/models.py:604
 msgid "The data returned by the task"
-msgstr ""
+msgstr "Dane zwrócone przez zadanie"
 
 #: documents/models.py:613
 msgid "Comment for the document"
-msgstr ""
+msgstr "Komentarz do dokumentu"
 
 #: documents/models.py:642
 msgid "comment"
-msgstr ""
+msgstr "komentarz"
 
 #: documents/models.py:643
 msgid "comments"
-msgstr ""
+msgstr "komentarze"
 
 #: documents/serialisers.py:72
 #, python-format

src/paperless/settings.py

@@ -8,6 +8,7 @@ import tempfile
 from typing import Final
 from typing import Optional
 from typing import Set
+from typing import Tuple
 from urllib.parse import urlparse
 
 from celery.schedules import crontab
@@ -65,6 +66,44 @@ def __get_path(key: str, default: str) -> str:
     return os.path.abspath(os.path.normpath(os.environ.get(key, default)))
 
 
+def _parse_redis_url(env_redis: Optional[str]) -> Tuple[str]:
+    """
+    Gets the Redis information from the environment or a default and handles
+    converting from incompatible django_channels and celery formats.
+
+    Returns a tuple of (celery_url, channels_url)
+    """
+
+    # Not set, return a compatible default
+    if env_redis is None:
+        return ("redis://localhost:6379", "redis://localhost:6379")
+
+    if "unix" in env_redis.lower():
+        # channels_redis socket format, looks like:
+        # "unix:///path/to/redis.sock"
+        _, path = env_redis.split(":")
+        # Optionally setting a db number
+        if "?db=" in env_redis:
+            path, number = path.split("?db=")
+            return (f"redis+socket:{path}?virtual_host={number}", env_redis)
+        else:
+            return (f"redis+socket:{path}", env_redis)
+
+    elif "+socket" in env_redis.lower():
+        # celery socket style, looks like:
+        # "redis+socket:///path/to/redis.sock"
+        _, path = env_redis.split(":")
+        if "?virtual_host=" in env_redis:
+            # Virtual host (aka db number)
+            path, number = path.split("?virtual_host=")
+            return (env_redis, f"unix:{path}?db={number}")
+        else:
+            return (env_redis, f"unix:{path}")
+
+    # Not a socket
+    return (env_redis, env_redis)
+
+
 # NEVER RUN WITH DEBUG IN PRODUCTION.
 DEBUG = __get_boolean("PAPERLESS_DEBUG", "NO")
 
@@ -182,7 +221,9 @@ ASGI_APPLICATION = "paperless.asgi.application"
 STATIC_URL = os.getenv("PAPERLESS_STATIC_URL", BASE_URL + "static/")
 WHITENOISE_STATIC_PREFIX = "/static/"
 
-_REDIS_URL = os.getenv("PAPERLESS_REDIS", "redis://localhost:6379")
+_CELERY_REDIS_URL, _CHANNELS_REDIS_URL = _parse_redis_url(
+    os.getenv("PAPERLESS_REDIS", None),
+)
 
 # TODO: what is this used for?
 TEMPLATES = [
@@ -205,7 +246,7 @@ CHANNEL_LAYERS = {
     "default": {
         "BACKEND": "channels_redis.core.RedisChannelLayer",
         "CONFIG": {
-            "hosts": [_REDIS_URL],
+            "hosts": [_CHANNELS_REDIS_URL],
             "capacity": 2000,  # default 100
             "expiry": 15,  # default 60
         },
@@ -468,7 +509,7 @@ TASK_WORKERS = __get_int("PAPERLESS_TASK_WORKERS", 1)
 
 WORKER_TIMEOUT: Final[int] = __get_int("PAPERLESS_WORKER_TIMEOUT", 1800)
 
-CELERY_BROKER_URL = _REDIS_URL
+CELERY_BROKER_URL = _CELERY_REDIS_URL
 CELERY_TIMEZONE = TIME_ZONE
 
 CELERY_WORKER_HIJACK_ROOT_LOGGER = False
@@ -513,7 +554,7 @@ CELERY_BEAT_SCHEDULE_FILENAME = os.path.join(DATA_DIR, "celerybeat-schedule.db")
 CACHES = {
     "default": {
         "BACKEND": "django.core.cache.backends.redis.RedisCache",
-        "LOCATION": _REDIS_URL,
+        "LOCATION": _CHANNELS_REDIS_URL,
     },
 }
 

src/paperless/tests/test_settings.py

@@ -3,6 +3,7 @@ from unittest import mock
 from unittest import TestCase
 
 from paperless.settings import _parse_ignore_dates
+from paperless.settings import _parse_redis_url
 from paperless.settings import default_threads_per_worker
 
 
@@ -82,3 +83,59 @@ class TestIgnoreDateParsing(TestCase):
                 self.assertGreaterEqual(default_threads, 1)
 
                 self.assertLessEqual(default_workers * default_threads, i)
+
+    def test_redis_socket_parsing(self):
+        """
+        GIVEN:
+            - Various Redis connection URI formats
+        WHEN:
+            - The URI is parsed
+        THEN:
+            - Socket based URIs are translated
+            - Non-socket URIs are unchanged
+            - None provided uses default
+        """
+
+        for input, expected in [
+            # Nothing is set
+            (None, ("redis://localhost:6379", "redis://localhost:6379")),
+            # celery style
+            (
+                "redis+socket:///run/redis/redis.sock",
+                (
+                    "redis+socket:///run/redis/redis.sock",
+                    "unix:///run/redis/redis.sock",
+                ),
+            ),
+            # redis-py / channels-redis style
+            (
+                "unix:///run/redis/redis.sock",
+                (
+                    "redis+socket:///run/redis/redis.sock",
+                    "unix:///run/redis/redis.sock",
+                ),
+            ),
+            # celery style with db
+            (
+                "redis+socket:///run/redis/redis.sock?virtual_host=5",
+                (
+                    "redis+socket:///run/redis/redis.sock?virtual_host=5",
+                    "unix:///run/redis/redis.sock?db=5",
+                ),
+            ),
+            # redis-py / channels-redis style with db
+            (
+                "unix:///run/redis/redis.sock?db=10",
+                (
+                    "redis+socket:///run/redis/redis.sock?virtual_host=10",
+                    "unix:///run/redis/redis.sock?db=10",
+                ),
+            ),
+            # Just a host with a port
+            (
+                "redis://myredishost:6379",
+                ("redis://myredishost:6379", "redis://myredishost:6379"),
+            ),
+        ]:
+            result = _parse_redis_url(input)
+            self.assertTupleEqual(expected, result)

src/paperless/version.py

@@ -1,7 +1,7 @@
 from typing import Final
 from typing import Tuple
 
-__version__: Final[Tuple[int, int, int]] = (1, 10, 0)
+__version__: Final[Tuple[int, int, int]] = (1, 10, 2)
 # Version string like X.Y.Z
 __full_version_str__: Final[str] = ".".join(map(str, __version__))
 # Version string like X.Y

src/paperless_tesseract/checks.py

@@ -16,8 +16,7 @@ def get_tesseract_langs():
     # Decode bytes to string, split on newlines, trim out the header
     proc_lines = proc.stdout.decode("utf8", errors="ignore").strip().split("\n")[1:]
 
-    # Replace _ with - to convert two part languages to the expected code
-    return [x.replace("_", "-") for x in proc_lines]
+    return [x.strip() for x in proc_lines]
 
 
 @register()

src/paperless_tesseract/tests/test_checks.py

@@ -27,3 +27,40 @@ class TestChecks(TestCase):
         msgs = check_default_language_available(None)
         self.assertEqual(len(msgs), 1)
         self.assertEqual(msgs[0].level, ERROR)
+
+    @override_settings(OCR_LANGUAGE="chi_sim")
+    @mock.patch("paperless_tesseract.checks.get_tesseract_langs")
+    def test_multi_part_language(self, m):
+        """
+        GIVEN:
+            - An OCR language which is multi part (ie chi-sim)
+            - The language is correctly formatted
+        WHEN:
+            - Installed packages are checked
+        THEN:
+            - No errors are reported
+        """
+        m.return_value = ["chi_sim", "eng"]
+
+        msgs = check_default_language_available(None)
+
+        self.assertEqual(len(msgs), 0)
+
+    @override_settings(OCR_LANGUAGE="chi-sim")
+    @mock.patch("paperless_tesseract.checks.get_tesseract_langs")
+    def test_multi_part_language_bad_format(self, m):
+        """
+        GIVEN:
+            - An OCR language which is multi part (ie chi-sim)
+            - The language is correctly NOT formatted
+        WHEN:
+            - Installed packages are checked
+        THEN:
+            - No errors are reported
+        """
+        m.return_value = ["chi_sim", "eng"]
+
+        msgs = check_default_language_available(None)
+
+        self.assertEqual(len(msgs), 1)
+        self.assertEqual(msgs[0].level, ERROR)