hore: update tika ocr and pdf scannning (#248)

* chore: update tika ocr and pdf scannning This Closes #247 * chore: update from 7071 to 8070
saidsef · May 11, 2024 · 5af2887 · 5af2887
1 parent c93f0d7
commit 5af2887
Show file tree

Hide file tree

Showing 7 changed files with 9 additions and 8 deletions.
diff --git a/deployment/k8s/base/server/deployment.yml b/deployment/k8s/base/server/deployment.yml
@@ -65,7 +65,7 @@ spec:
                   fieldPath: status.podIP
           ports:
           - name: server
-            containerPort: 7071
+            containerPort: 8070
             protocol: TCP
           - name: metrics
             containerPort: 7072

diff --git a/deployment/k8s/base/server/service.yml b/deployment/k8s/base/server/service.yml
@@ -17,7 +17,7 @@ spec:
   ports:
   - name: server
     protocol: TCP
-    port: 7071
+    port: 8070
     targetPort: server
   - name: metrics
     protocol: TCP

diff --git a/deployment/k8s/base/ui/deployment.yml b/deployment/k8s/base/ui/deployment.yml
@@ -62,7 +62,7 @@ spec:
           - name: HOST
             value: "tika-server"
           - name: HOST_PORT
-            value: "7071"
+            value: "8070"
           ports:
             - name: http
               containerPort: 8080

diff --git a/function/Dockerfile b/function/Dockerfile
@@ -10,7 +10,7 @@ ENV JAI_IMAGEIO_JPEG2000 1.4.0
 ENV JBIG2_IMAGEIO 3.0.4
 ENV MS 180000
 ENV POD_IP ${POD_IP:-127.0.0.1}
-ENV PORT ${PORT:-7070}
+ENV PORT ${PORT:-8070}
 ENV PYTHONIOENCODING utf8
 ENV SQLITE_JDBC 3.40.0.0
 

diff --git a/function/Dockerfile.server b/function/Dockerfile.server
@@ -6,7 +6,7 @@ LABEL org.opencontainers.image.description "Apache Tika API Server"
 
 ENV TIKA_VERSION 2.9.2
 ENV CLASSPATH "/tika-server-standard-${TIKA_VERSION}.jar:/tika-extras/*:/opt/tika/libs/*"
-ENV PORT ${PORT:-7071}
+ENV PORT ${PORT:-8070}
 ENV PORT_METRICS ${PORT_METRICS:-7072}
 ENV POD_IP ${POD_IP:-127.0.0.1}
 ENV PROMETHEUS_JMX_JAR_VERSION 0.20.0

diff --git a/function/handler.py b/function/handler.py
@@ -6,7 +6,7 @@
 from subprocess import Popen, PIPE
 
 logging.getLogger(__name__)
-logging.basicConfig(level=logging.DEBUG)
+logging.basicConfig(level=logging.DEBUG, handlers=[logging.StreamHandler()])
 
 
 def byte2json(b):

diff --git a/ui/app.js b/ui/app.js
@@ -26,7 +26,7 @@ const uploads = multer({ storage });
 const TIMEOUT = 500000; // Milliseconds
 const PORT = process.env.PORT || 8080;
 const HOST = process.env.HOST || 'server';
-const HOST_PORT = process.env.HOST_PORT || 7071;
+const HOST_PORT = process.env.HOST_PORT || 8070;
 const protocol = process.env.PROTOCOL === 'https' ? https : http;
 
 collectDefaultMetrics({ register: prometheusRegister });
@@ -92,7 +92,8 @@ app.post('/', uploads.single('doc'), async (req, res, next) => {
     if (req.file) {
       payload = req.file.buffer;
       options.headers = {
-        'Content-Type': req.file.mimetype
+        'Content-Type': req.file.mimetype,
+        'X-Tika-PDFocrStrategy': 'auto'
       }
     }
     const post = protocol.request(options, (response) => {