Refactor after feedback:

- rename PAPERLESS_TIKA to PAPERLESS_TIKA_ENABLED
- all other env params now start with PAPERLESS_TIKA
- convert_to_pdf as class instance method
- smaller details

Signed-off-by: Jo Vandeginste <Jo.Vandeginste@kuleuven.be>

docker/hub/docker-compose.tika.yml

@@ -24,9 +24,9 @@ services:
     env_file: docker-compose.env
     environment:
       PAPERLESS_REDIS: redis://broker:6379
-      PAPERLESS_TIKA: 1
-      GOTENBERG_SERVER_ENDPOINT: http://gotenberg:3000
-      TIKA_SERVER_ENDPOINT: http://tika:9998
+      PAPERLESS_TIKA_ENABLED: 1
+      PAPERLESS_TIKA_GOTENBERG_ENDPOINT: http://gotenberg:3000
+      PAPERLESS_TIKA_ENDPOINT: http://tika:9998
 
   gotenberg:
     image: thecodingmachine/gotenberg

docker/local/docker-compose.tika.yml

@@ -24,9 +24,9 @@ services:
     env_file: docker-compose.env
     environment:
       PAPERLESS_REDIS: redis://broker:6379
-      PAPERLESS_TIKA: 1
-      GOTENBERG_SERVER_ENDPOINT: http://gotenberg:3000
-      TIKA_SERVER_ENDPOINT: http://tika:9998
+      PAPERLESS_TIKA_ENABLED: 1
+      PAPERLESS_TIKA_GOTENBERG_ENDPOINT: http://gotenberg:3000
+      PAPERLESS_TIKA_ENDPOINT: http://tika:9998
 
   gotenberg:
     image: thecodingmachine/gotenberg

docs/configuration.rst

@@ -291,17 +291,17 @@ configure their endpoints, and enable the feature.
 If you run paperless on docker, you can add those services to the docker-compose
 file (see the examples provided).
 
-PAPERLESS_TIKA=<bool>
+PAPERLESS_TIKA_ENABLED=<bool>
     Enable (or disable) the Tika parser.
 
     Defaults to false.
 
-TIKA_SERVER_ENDPOINT=<url>
+PAPERLESS_TIKA_ENDPOINT=<url>
     Set the endpoint URL were Paperless can reach your Tika server.
 
     Defaults to "http://localhost:9998".
 
-GOTENBERG_SERVER_ENDPOINT=<url>
+PAPERLESS_TIKA_GOTENBERG_ENDPOINT=<url>
     Set the endpoint URL were Paperless can reach your Gotenberg server.
 
     Defaults to "http://localhost:3000".

src/paperless/settings.py

@@ -427,5 +427,8 @@ PAPERLESS_FILENAME_FORMAT = os.getenv("PAPERLESS_FILENAME_FORMAT")
 THUMBNAIL_FONT_NAME = os.getenv("PAPERLESS_THUMBNAIL_FONT_NAME", "/usr/share/fonts/liberation/LiberationSerif-Regular.ttf")
 
 # Tika settings
-PAPERLESS_TIKA = __get_boolean("PAPERLESS_TIKA", "NO")
-GOTENBERG_SERVER_ENDPOINT = os.getenv("GOTENBERG_SERVER_ENDPOINT", "http://localhost:3000")
+PAPERLESS_TIKA_ENABLED = __get_boolean("PAPERLESS_TIKA_ENABLED", "NO")
+PAPERLESS_TIKA_ENDPOINT = os.getenv("PAPERLESS_TIKA_ENDPOINT", "http://localhost:9998")
+PAPERLESS_TIKA_GOTENBERG_ENDPOINT = os.getenv(
+    "PAPERLESS_TIKA_GOTENBERG_ENDPOINT", "http://localhost:3000"
+)

src/paperless_tika/apps.py

@@ -9,6 +9,6 @@ class PaperlessTikaConfig(AppConfig):
     def ready(self):
         from documents.signals import document_consumer_declaration
 
-        if settings.PAPERLESS_TIKA:
+        if settings.PAPERLESS_TIKA_ENABLED:
             document_consumer_declaration.connect(tika_consumer_declaration)
         AppConfig.ready(self)

src/paperless_tika/parsers.py

@@ -70,49 +70,46 @@ class TikaDocumentParser(DocumentParser):
 
     def parse(self, document_path, mime_type):
         self.log("info", f"[TIKA_PARSE] Sending {document_path} to Tika server")
+        tika_server = settings.PAPERLESS_TIKA_ENDPOINT
 
         try:
-            parsed = parser.from_file(document_path)
+            parsed = parser.from_file(document_path, tika_server)
         except requests.exceptions.HTTPError as err:
-            raise ParseError(f"Could not parse {document_path} with tika server: {err}")
-
-        try:
-            content = parsed["content"].strip()
-        except:
-            content = ""
-
-        try:
-            creation_date = dateutil.parser.isoparse(
-                parsed["metadata"]["Creation-Date"]
+            raise ParseError(
+                f"Could not parse {document_path} with tika server at {tika_server}: {err}"
             )
+
+        try:
+            self.text = parsed["content"].strip()
         except:
-            creation_date = None
+            pass
+
+        try:
+            self.date = dateutil.parser.isoparse(parsed["metadata"]["Creation-Date"])
+        except:
+            pass
 
         archive_path = os.path.join(self.tempdir, "convert.pdf")
-        convert_to_pdf(self, document_path, archive_path)
-
+        convert_to_pdf(document_path, archive_path)
         self.archive_path = archive_path
-        self.date = creation_date
-        self.text = content
 
+    def convert_to_pdf(document_path, pdf_path):
+        pdf_path = os.path.join(self.tempdir, "convert.pdf")
+        gotenberg_server = settings.PAPERLESS_TIKA_GOTENBERG_ENDPOINT
+        url = gotenberg_server + "/convert/office"
 
-def convert_to_pdf(self, document_path, pdf_path):
-    pdf_path = os.path.join(self.tempdir, "convert.pdf")
-    gotenberg_server = settings.GOTENBERG_SERVER_ENDPOINT
-    url = gotenberg_server + "/convert/office"
+        self.log("info", f"[TIKA] Converting {document_path} to PDF as {pdf_path}")
+        files = {"files": open(document_path, "rb")}
+        headers = {}
 
-    self.log("info", f"[TIKA] Converting {document_path} to PDF as {pdf_path}")
-    files = {"files": open(document_path, "rb")}
-    headers = {}
+        try:
+            response = requests.post(url, files=files, headers=headers)
+            response.raise_for_status()  # ensure we notice bad responses
+        except requests.exceptions.HTTPError as err:
+            raise ParseError(
+                f"Could not contact gotenberg server at {gotenberg_server}: {err}"
+            )
 
-    try:
-        response = requests.post(url, files=files, headers=headers)
-        response.raise_for_status()  # ensure we notice bad responses
-    except requests.exceptions.HTTPError as err:
-        raise ParseError(
-            f"Could not contact gotenberg server at {gotenberg_server}: {err}"
-        )
-
-    file = open(pdf_path, "wb")
-    file.write(response.content)
-    file.close()
+        file = open(pdf_path, "wb")
+        file.write(response.content)
+        file.close()

src/paperless_tika/test.py

@@ -1,3 +0,0 @@
-import magic
-m = magic.from_file("/nfsstorage/jo/syncthing/Documenten/20R-309.153.052.pdf")
-print(m)