Merge branch 'main' into rag

kubernetes/helm/templates/ollama-namespace.yaml

+apiVersion: v1
+kind: Namespace
+metadata:
+  name: {{ .Values.namespace }}
\ No newline at end of file

kubernetes/helm/templates/ollama-service.yaml

+apiVersion: v1
+kind: Service
+metadata:
+  name: ollama-service
+  namespace: {{ .Values.namespace }}
+spec:
+  type: {{ .Values.ollama.service.type }}
+  selector:
+    app: ollama
+  ports:
+  - protocol: TCP
+    port: {{ .Values.ollama.servicePort }}
+    targetPort: {{ .Values.ollama.servicePort }}
\ No newline at end of file

kubernetes/helm/templates/ollama-statefulset.yaml

+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: ollama
+  namespace: {{ .Values.namespace }}
+spec:
+  serviceName: "ollama"
+  replicas: {{ .Values.ollama.replicaCount }}
+  selector:
+    matchLabels:
+      app: ollama
+  template:
+    metadata:
+      labels:
+        app: ollama
+    spec:
+      containers:
+      - name: ollama
+        image: {{ .Values.ollama.image }}
+        ports:
+        - containerPort: {{ .Values.ollama.servicePort }}
+        env:
+        {{- if .Values.ollama.gpu.enabled }}
+          - name: PATH
+            value: /usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
+          - name: LD_LIBRARY_PATH
+            value: /usr/local/nvidia/lib:/usr/local/nvidia/lib64
+          - name: NVIDIA_DRIVER_CAPABILITIES
+            value: compute,utility
+        {{- end}}
+        {{- if .Values.ollama.resources }}
+        resources: {{- toYaml .Values.ollama.resources | nindent 10 }}
+        {{- end }}
+        volumeMounts:
+        - name: ollama-volume
+          mountPath: /root/.ollama
+        tty: true
+      {{- with .Values.ollama.nodeSelector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      tolerations:
+        {{- if .Values.ollama.gpu.enabled }}
+        - key: nvidia.com/gpu
+          operator: Exists
+          effect: NoSchedule
+        {{- end }}
+  volumeClaimTemplates:
+  - metadata:
+      name: ollama-volume
+    spec:
+      accessModes: [ "ReadWriteOnce" ]
+      resources:
+        requests:
+          storage: {{ .Values.ollama.volumeSize }}
\ No newline at end of file

kubernetes/helm/templates/webui-deployment.yaml

+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: ollama-webui-deployment
+  namespace: {{ .Values.namespace }}
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: ollama-webui
+  template:
+    metadata:
+      labels:
+        app: ollama-webui
+    spec:
+      containers:
+      - name: ollama-webui
+        image: {{ .Values.webui.image }}
+        ports:
+        - containerPort: 8080
+        {{- if .Values.webui.resources }}
+        resources: {{- toYaml .Values.webui.resources | nindent 10 }}
+        {{- end }}
+        volumeMounts:
+        - name: webui-volume
+          mountPath: /app/backend/data
+        env:
+        - name: OLLAMA_API_BASE_URL
+          value: "http://ollama-service.{{ .Values.namespace }}.svc.cluster.local:{{ .Values.ollama.servicePort }}/api"
+        tty: true
+      {{- with .Values.webui.nodeSelector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      volumes:
+      - name: webui-volume
+        persistentVolumeClaim:
+          claimName: ollama-webui-pvc
\ No newline at end of file

kubernetes/helm/templates/webui-ingress.yaml

+{{- if .Values.webui.ingress.enabled }}
+apiVersion: networking.k8s.io/v1
+kind: Ingress
+metadata:
+  name: ollama-webui-ingress
+  namespace: {{ .Values.namespace }}
+{{- if .Values.webui.ingress.annotations }}
+  annotations:
+{{ toYaml .Values.webui.ingress.annotations | trimSuffix "\n" | indent 4 }}
+{{- end }}
+spec:
+  rules:
+  - host: {{ .Values.webui.ingress.host }}
+    http:
+      paths:
+      - path: /
+        pathType: Prefix
+        backend:
+          service:
+            name: ollama-webui-service
+            port:
+              number: {{ .Values.webui.servicePort }}
+{{- end }}

kubernetes/helm/templates/webui-pvc.yaml

+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  labels:
+    app: ollama-webui
+  name: ollama-webui-pvc
+  namespace: {{ .Values.namespace }}  
+spec:
+  accessModes: [ "ReadWriteOnce" ]
+  resources:
+    requests:
+      storage: {{ .Values.webui.volumeSize }}
\ No newline at end of file

kubernetes/helm/templates/webui-service.yaml

+apiVersion: v1
+kind: Service
+metadata:
+  name: ollama-webui-service
+  namespace: {{ .Values.namespace }}
+spec:
+  type: {{ .Values.webui.service.type }} # Default: NodePort  # Use LoadBalancer if you're on a cloud that supports it
+  selector:
+    app: ollama-webui
+  ports:
+    - protocol: TCP
+      port: {{ .Values.webui.servicePort }}
+      targetPort: {{ .Values.webui.servicePort }}
+      # If using NodePort, you can optionally specify the nodePort:
+      # nodePort: 30000
\ No newline at end of file

kubernetes/helm/values.yaml

+namespace: ollama-namespace
+
+ollama:
+  replicaCount: 1
+  image: ollama/ollama:latest
+  servicePort: 11434
+  resources:
+    limits:
+      cpu: "2000m"
+      memory: "2Gi"
+      nvidia.com/gpu: "0"
+  volumeSize: 1Gi
+  nodeSelector: {}
+  tolerations: []
+  service:
+    type: ClusterIP
+  gpu:
+    enabled: false
+
+webui:
+  replicaCount: 1
+  image: ghcr.io/ollama-webui/ollama-webui:main
+  servicePort: 8080
+  resources:
+    limits:
+      cpu: "500m"
+      memory: "500Mi"
+  ingress:
+    enabled: true
+    annotations:
+    # Use appropriate annotations for your Ingress controller, e.g., for NGINX:
+      # nginx.ingress.kubernetes.io/rewrite-target: /
+    host: ollama.minikube.local
+  volumeSize: 1Gi
+  nodeSelector: {}
+  tolerations: []
+  service:
+    type: NodePort
\ No newline at end of file

kubernetes/manifest/base/ollama-namespace.yaml

+apiVersion: v1
+kind: Namespace
+metadata:
+  name: ollama-namespace
\ No newline at end of file

kubernetes/manifest/base/ollama-service.yaml

+apiVersion: v1
+kind: Service
+metadata:
+  name: ollama-service
+  namespace: ollama-namespace
+spec:
+  selector:
+    app: ollama
+  ports:
+  - protocol: TCP
+    port: 11434
+    targetPort: 11434
\ No newline at end of file

kubernetes/manifest/base/ollama-statefulset.yaml

+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: ollama
+  namespace: ollama-namespace
+spec:
+  serviceName: "ollama"
+  replicas: 1
+  selector:
+    matchLabels:
+      app: ollama
+  template:
+    metadata:
+      labels:
+        app: ollama
+    spec:
+      containers:
+      - name: ollama
+        image: ollama/ollama:latest
+        ports:
+        - containerPort: 11434
+        resources:
+          limits:
+            cpu: "2000m"
+            memory: "2Gi"
+        volumeMounts:
+        - name: ollama-volume
+          mountPath: /root/.ollama
+        tty: true
+  volumeClaimTemplates:
+  - metadata:
+      name: ollama-volume
+    spec:
+      accessModes: [ "ReadWriteOnce" ]
+      resources:
+        requests:
+          storage: 1Gi
\ No newline at end of file

kubernetes/manifest/base/webui-deployment.yaml

+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: ollama-webui-deployment
+  namespace: ollama-namespace
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: ollama-webui
+  template:
+    metadata:
+      labels:
+        app: ollama-webui
+    spec:
+      containers:
+      - name: ollama-webui
+        image: ghcr.io/ollama-webui/ollama-webui:main
+        ports:
+        - containerPort: 8080
+        resources:
+          limits:
+            cpu: "500m"
+            memory: "500Mi"
+        env:
+        - name: OLLAMA_API_BASE_URL
+          value: "http://ollama-service.ollama-namespace.svc.cluster.local:11434/api"
+        tty: true
\ No newline at end of file

kubernetes/manifest/base/webui-ingress.yaml

+apiVersion: networking.k8s.io/v1
+kind: Ingress
+metadata:
+  name: ollama-webui-ingress
+  namespace: ollama-namespace
+  #annotations:
+    # Use appropriate annotations for your Ingress controller, e.g., for NGINX:
+    # nginx.ingress.kubernetes.io/rewrite-target: /
+spec:
+  rules:
+  - host: ollama.minikube.local
+    http:
+      paths:
+      - path: /
+        pathType: Prefix
+        backend:
+          service:
+            name: ollama-webui-service
+            port:
+              number: 8080

kubernetes/manifest/base/webui-service.yaml

+apiVersion: v1
+kind: Service
+metadata:
+  name: ollama-webui-service
+  namespace: ollama-namespace
+spec:
+  type: NodePort  # Use LoadBalancer if you're on a cloud that supports it
+  selector:
+    app: ollama-webui
+  ports:
+    - protocol: TCP
+      port: 8080
+      targetPort: 8080
+      # If using NodePort, you can optionally specify the nodePort:
+      # nodePort: 30000
\ No newline at end of file

kubernetes/manifest/kustomization.yaml

+resources:
+- base/ollama-namespace.yaml
+- base/ollama-service.yaml
+- base/ollama-statefulset.yaml
+- base/webui-deployment.yaml
+- base/webui-service.yaml
+- base/webui-ingress.yaml
+
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+patches:
+- path: patches/ollama-statefulset-gpu.yaml

kubernetes/manifest/patches/ollama-statefulset-gpu.yaml

+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: ollama
+  namespace: ollama-namespace
+spec:
+  selector:
+    matchLabels:
+      app: ollama
+  serviceName: "ollama"
+  template:
+    spec:
+      containers:
+      - name: ollama
+        resources:
+          limits:
+            nvidia.com/gpu: "1"

run-compose.sh

+#!/bin/bash
+
+# Define color and formatting codes
+BOLD='\033[1m'
+GREEN='\033[1;32m'
+WHITE='\033[1;37m'
+RED='\033[0;31m'
+NC='\033[0m' # No Color
+# Unicode character for tick mark
+TICK='\u2713'
+
+# Detect GPU driver
+get_gpu_driver() {
+    # Detect NVIDIA GPUs
+    if lspci | grep -i nvidia >/dev/null; then
+        echo "nvidia"
+        return
+    fi
+
+    # Detect AMD GPUs (including GCN architecture check for amdgpu vs radeon)
+    if lspci | grep -i amd >/dev/null; then
+        # List of known GCN and later architecture cards
+        # This is a simplified list, and in a real-world scenario, you'd want a more comprehensive one
+        local gcn_and_later=("Radeon HD 7000" "Radeon HD 8000" "Radeon R5" "Radeon R7" "Radeon R9" "Radeon RX")
+
+        # Get GPU information
+        local gpu_info=$(lspci | grep -i 'vga.*amd')
+
+        for model in "${gcn_and_later[@]}"; do
+            if echo "$gpu_info" | grep -iq "$model"; then
+                echo "amdgpu"
+                return
+            fi
+        done
+
+        # Default to radeon if no GCN or later architecture is detected
+        echo "radeon"
+        return
+    fi
+
+    # Detect Intel GPUs
+    if lspci | grep -i intel >/dev/null; then
+        echo "i915"
+        return
+    fi
+
+    # If no known GPU is detected
+    echo "Unknown or unsupported GPU driver"
+    exit 1
+}
+
+# Function for rolling animation
+show_loading() {
+    local spin='-\|/'
+    local i=0
+
+    printf " "
+
+    while kill -0 $1 2>/dev/null; do
+        i=$(( (i+1) %4 ))
+        printf "\b${spin:$i:1}"
+        sleep .1
+    done
+
+    # Replace the spinner with a tick
+    printf "\b${GREEN}${TICK}${NC}"
+}
+
+# Usage information
+usage() {
+    echo "Usage: $0 [OPTIONS]"
+    echo "Options:"
+    echo "  --enable-gpu[count=COUNT]  Enable GPU support with the specified count."
+    echo "  --enable-api[port=PORT]    Enable API and expose it on the specified port."
+    echo "  --webui[port=PORT]         Set the port for the web user interface."
+    echo "  --data[folder=PATH]        Bind mount for ollama data folder (by default will create the 'ollama' volume)."
+    echo "  --build                    Build the docker image before running the compose project."
+    echo "  --drop                     Drop the compose project."
+    echo "  -q, --quiet                Run script in headless mode."
+    echo "  -h, --help                 Show this help message."
+    echo ""
+    echo "Examples:"
+    echo "  $0 --drop"
+    echo "  $0 --enable-gpu[count=1]"
+    echo "  $0 --enable-api[port=11435]"
+    echo "  $0 --enable-gpu[count=1] --enable-api[port=12345] --webui[port=3000]"
+    echo "  $0 --enable-gpu[count=1] --enable-api[port=12345] --webui[port=3000] --data[folder=./ollama-data]"
+    echo "  $0 --enable-gpu[count=1] --enable-api[port=12345] --webui[port=3000] --data[folder=./ollama-data] --build"
+    echo ""
+    echo "This script configures and runs a docker-compose setup with optional GPU support, API exposure, and web UI configuration."
+    echo "About the gpu to use, the script automatically detects it using the "lspci" command."
+    echo "In this case the gpu detected is: $(get_gpu_driver)"
+}
+
+# Default values
+gpu_count=1
+api_port=11435
+webui_port=3000
+headless=false
+build_image=false
+kill_compose=false
+
+# Function to extract value from the parameter
+extract_value() {
+    echo "$1" | sed -E 's/.*\[.*=(.*)\].*/\1/; t; s/.*//'
+}
+
+# Parse arguments
+while [[ $# -gt 0 ]]; do
+    key="$1"
+
+    case $key in
+        --enable-gpu*)
+            enable_gpu=true
+            value=$(extract_value "$key")
+            gpu_count=${value:-1}
+            ;;
+        --enable-api*)
+            enable_api=true
+            value=$(extract_value "$key")
+            api_port=${value:-11435}
+            ;;
+        --webui*)
+            value=$(extract_value "$key")
+            webui_port=${value:-3000}
+            ;;
+        --data*)
+            value=$(extract_value "$key")
+            data_dir=${value:-"./ollama-data"}
+            ;;
+        --drop)
+            kill_compose=true
+            ;;
+        --build)
+            build_image=true
+            ;;
+        -q|--quiet)
+            headless=true
+            ;;
+        -h|--help)
+            usage
+            exit
+            ;;
+        *)
+            # Unknown option
+            echo "Unknown option: $key"
+            usage
+            exit 1
+            ;;
+    esac
+    shift # past argument or value
+done
+
+if [[ $kill_compose == true ]]; then
+    docker compose down --remove-orphans
+    echo -e "${GREEN}${BOLD}Compose project dropped successfully.${NC}"
+    exit
+else
+    DEFAULT_COMPOSE_COMMAND="docker compose -f docker-compose.yaml"
+    if [[ $enable_gpu == true ]]; then
+        # Validate and process command-line arguments
+        if [[ -n $gpu_count ]]; then
+            if ! [[ $gpu_count =~ ^[0-9]+$ ]]; then
+                echo "Invalid GPU count: $gpu_count"
+                exit 1
+            fi
+            echo "Enabling GPU with $gpu_count GPUs"
+            # Add your GPU allocation logic here
+            export OLLAMA_GPU_DRIVER=$(get_gpu_driver)
+            export OLLAMA_GPU_COUNT=$gpu_count # Set OLLAMA_GPU_COUNT environment variable
+        fi
+        DEFAULT_COMPOSE_COMMAND+=" -f docker-compose.gpu.yaml"
+    fi
+    if [[ $enable_api == true ]]; then
+        DEFAULT_COMPOSE_COMMAND+=" -f docker-compose.api.yaml"
+        if [[ -n $api_port ]]; then
+            export OLLAMA_WEBAPI_PORT=$api_port # Set OLLAMA_WEBAPI_PORT environment variable
+        fi
+    fi
+    if [[ -n $data_dir ]]; then
+        DEFAULT_COMPOSE_COMMAND+=" -f docker-compose.data.yaml"
+        export OLLAMA_DATA_DIR=$data_dir # Set OLLAMA_DATA_DIR environment variable
+    fi
+    DEFAULT_COMPOSE_COMMAND+=" up -d"
+    DEFAULT_COMPOSE_COMMAND+=" --remove-orphans"
+    DEFAULT_COMPOSE_COMMAND+=" --force-recreate"
+    if [[ $build_image == true ]]; then
+        DEFAULT_COMPOSE_COMMAND+=" --build"
+    fi
+fi
+
+# Recap of environment variables
+echo
+echo -e "${WHITE}${BOLD}Current Setup:${NC}"
+echo -e "   ${GREEN}${BOLD}GPU Driver:${NC} ${OLLAMA_GPU_DRIVER:-Not Enabled}"
+echo -e "   ${GREEN}${BOLD}GPU Count:${NC} ${OLLAMA_GPU_COUNT:-Not Enabled}"
+echo -e "   ${GREEN}${BOLD}WebAPI Port:${NC} ${OLLAMA_WEBAPI_PORT:-Not Enabled}"
+echo -e "   ${GREEN}${BOLD}Data Folder:${NC} ${data_dir:-Using ollama volume}"
+echo -e "   ${GREEN}${BOLD}WebUI Port:${NC} $webui_port"
+echo
+
+if [[ $headless == true ]]; then
+    echo -ne "${WHITE}${BOLD}Running in headless mode... ${NC}"
+    choice="y"
+else
+    # Ask for user acceptance
+    echo -ne "${WHITE}${BOLD}Do you want to proceed with current setup? (Y/n): ${NC}"
+    read -n1 -s choice
+fi
+
+echo
+
+if [[ $choice == "" || $choice == "y" ]]; then
+    # Execute the command with the current user
+    eval "$DEFAULT_COMPOSE_COMMAND" &
+
+    # Capture the background process PID
+    PID=$!
+
+    # Display the loading animation
+    #show_loading $PID
+
+    # Wait for the command to finish
+    wait $PID
+
+    echo
+    # Check exit status
+    if [ $? -eq 0 ]; then
+        echo -e "${GREEN}${BOLD}Compose project started successfully.${NC}"
+    else
+        echo -e "${RED}${BOLD}There was an error starting the compose project.${NC}"
+    fi
+else
+    echo "Aborted."
+fi
+
+echo

src/app.css

@@ -16,7 +16,7 @@ html {
 
 code {
 	/* white-space-collapse: preserve !important; */
-	white-space: pre;
+	overflow-x: auto;
 	width: auto;
 }
 

src/lib/components/chat/MessageInput.svelte

@@ -298,6 +298,24 @@
 									submitPrompt(prompt);
 								}
 							}}
+							on:keydown={(e) => {
+								if (prompt === '' && e.key == 'ArrowUp') {
+									e.preventDefault();
+
+									const userMessageElement = [
+										...document.getElementsByClassName('user-message')
+									]?.at(-1);
+
+									const editButton = [
+										...document.getElementsByClassName('edit-user-message-button')
+									]?.at(-1);
+
+									console.log(userMessageElement);
+
+									userMessageElement.scrollIntoView({ block: 'center' });
+									editButton?.click();
+								}
+							}}
 							rows="1"
 							on:input={(e) => {
 								e.target.style.height = '';

src/lib/components/chat/Messages/ResponseMessage.svelte

@@ -88,6 +88,7 @@
 				let code = block.querySelector('code');
 				code.style.borderTopRightRadius = 0;
 				code.style.borderTopLeftRadius = 0;
+				code.style.whiteSpace = 'pre';
 
 				let topBarDiv = document.createElement('div');
 				topBarDiv.style.backgroundColor = '#202123';