HomeAudit/scripts/diagnose_paperless_issues.sh

#!/bin/bash

# Paperless Database Issue Diagnostic Script
# This script helps identify and diagnose issues between Paperless-ngx and Paperless AI

set -e

echo "🔍 Paperless Database Issue Diagnostic"
echo "======================================"

# Colors for output
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
BLUE='\033[0;34m'
NC='\033[0m' # No Color

# Function to print colored output
print_status() {
    echo -e "${BLUE}[INFO]${NC} $1"
}

print_success() {
    echo -e "${GREEN}[SUCCESS]${NC} $1"
}

print_warning() {
    echo -e "${YELLOW}[WARNING]${NC} $1"
}

print_error() {
    echo -e "${RED}[ERROR]${NC} $1"
}

# Check if Docker is running
if ! docker info > /dev/null 2>&1; then
    print_error "Docker is not running. Please start Docker first."
    exit 1
fi

echo ""
print_status "1. Checking container status..."

# Check Paperless-ngx containers
PAPERLESS_CONTAINERS=$(docker ps -a --filter "name=paperless" --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}")
if [[ -n "$PAPERLESS_CONTAINERS" ]]; then
    echo "📄 Paperless-ngx containers:"
    echo "$PAPERLESS_CONTAINERS"
else
    print_warning "No Paperless-ngx containers found"
fi

# Check Paperless AI containers
PAPERLESS_AI_CONTAINERS=$(docker ps -a --filter "name=paperless-ai" --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}")
if [[ -n "$PAPERLESS_AI_CONTAINERS" ]]; then
    echo ""
    echo "🤖 Paperless AI containers:"
    echo "$PAPERLESS_AI_CONTAINERS"
else
    print_warning "No Paperless AI containers found"
fi

echo ""
print_status "2. Checking database configurations..."

# Check if PostgreSQL is running
if docker ps | grep -q "postgres"; then
    print_success "PostgreSQL container is running"
    
    # Get PostgreSQL container details
    PG_CONTAINER=$(docker ps --filter "name=postgres" --format "{{.Names}}" | head -1)
    if [[ -n "$PG_CONTAINER" ]]; then
        echo "   Container: $PG_CONTAINER"
        
        # Check database connectivity
        if docker exec "$PG_CONTAINER" pg_isready -U postgres > /dev/null 2>&1; then
            print_success "   Database is accepting connections"
        else
            print_error "   Database is not accepting connections"
        fi
        
        # Check if paperless database exists
        if docker exec "$PG_CONTAINER" psql -U postgres -lqt | cut -d \| -f 1 | grep -qw paperless; then
            print_success "   Paperless database exists"
        else
            print_error "   Paperless database does not exist"
        fi
    fi
else
    print_warning "PostgreSQL container not found"
fi

echo ""
print_status "3. Checking network connectivity..."

# Check if containers can communicate
if docker ps | grep -q "paperless" && docker ps | grep -q "paperless-ai"; then
    PAPERLESS_CONTAINER=$(docker ps --filter "name=paperless" --format "{{.Names}}" | head -1)
    PAPERLESS_AI_CONTAINER=$(docker ps --filter "name=paperless-ai" --format "{{.Names}}" | head -1)
    
    if [[ -n "$PAPERLESS_CONTAINER" && -n "$PAPERLESS_AI_CONTAINER" ]]; then
        # Check if they're on the same network
        PAPERLESS_NETWORKS=$(docker inspect "$PAPERLESS_CONTAINER" --format '{{range $net, $config := .NetworkSettings.Networks}}{{$net}} {{end}}')
        PAPERLESS_AI_NETWORKS=$(docker inspect "$PAPERLESS_AI_CONTAINER" --format '{{range $net, $config := .NetworkSettings.Networks}}{{$net}} {{end}}')
        
        echo "   Paperless-ngx networks: $PAPERLESS_NETWORKS"
        echo "   Paperless AI networks: $PAPERLESS_AI_NETWORKS"
        
        # Check for common networks
        for net in $PAPERLESS_NETWORKS; do
            if echo "$PAPERLESS_AI_NETWORKS" | grep -q "$net"; then
                print_success "   Both containers are on network: $net"
                break
            fi
        done
    fi
else
    print_warning "Cannot check network connectivity - containers not running"
fi

echo ""
print_status "4. Checking environment variables..."

# Check Paperless-ngx environment
if docker ps | grep -q "paperless"; then
    PAPERLESS_CONTAINER=$(docker ps --filter "name=paperless" --format "{{.Names}}" | head -1)
    if [[ -n "$PAPERLESS_CONTAINER" ]]; then
        echo "📄 Paperless-ngx environment:"
        docker exec "$PAPERLESS_CONTAINER" env | grep -E "(PAPERLESS_DB|PAPERLESS_REDIS|PAPERLESS_URL)" | sort
    fi
fi

# Check Paperless AI environment
if docker ps | grep -q "paperless-ai"; then
    PAPERLESS_AI_CONTAINER=$(docker ps --filter "name=paperless-ai" --format "{{.Names}}" | head -1)
    if [[ -n "$PAPERLESS_AI_CONTAINER" ]]; then
        echo ""
        echo "🤖 Paperless AI environment:"
        docker exec "$PAPERLESS_AI_CONTAINER" env | grep -E "(PAPERLESS_|OPENAI_|OLLAMA_|DEEPSEEK_|AZURE_)" | sort
    fi
fi

echo ""
print_status "5. Checking API connectivity..."

# Check Paperless-ngx API
if curl -f -s "https://paperless.pressmess.duckdns.org/api/" > /dev/null 2>&1; then
    print_success "Paperless-ngx API is accessible"
else
    print_error "Paperless-ngx API is not accessible"
fi

# Check Paperless AI API
if curl -f -s "http://localhost:3000/health" > /dev/null 2>&1; then
    print_success "Paperless AI API is accessible"
else
    print_error "Paperless AI API is not accessible"
fi

echo ""
print_status "6. Checking recent logs for errors..."

# Check Paperless-ngx logs
if docker ps | grep -q "paperless"; then
    PAPERLESS_CONTAINER=$(docker ps --filter "name=paperless" --format "{{.Names}}" | head -1)
    if [[ -n "$PAPERLESS_CONTAINER" ]]; then
        echo "📄 Recent Paperless-ngx errors:"
        docker logs "$PAPERLESS_CONTAINER" --tail 20 2>&1 | grep -i "error\|exception\|failed" || echo "   No recent errors found"
    fi
fi

# Check Paperless AI logs
if docker ps | grep -q "paperless-ai"; then
    PAPERLESS_AI_CONTAINER=$(docker ps --filter "name=paperless-ai" --format "{{.Names}}" | head -1)
    if [[ -n "$PAPERLESS_AI_CONTAINER" ]]; then
        echo ""
        echo "🤖 Recent Paperless AI errors:"
        docker logs "$PAPERLESS_AI_CONTAINER" --tail 20 2>&1 | grep -i "error\|exception\|failed" || echo "   No recent errors found"
    fi
fi

echo ""
print_status "7. Database schema analysis..."

# Check if we can connect to the database and analyze the schema
if docker ps | grep -q "postgres"; then
    PG_CONTAINER=$(docker ps --filter "name=postgres" --format "{{.Names}}" | head -1)
    if [[ -n "$PG_CONTAINER" ]]; then
        echo "📊 Checking Paperless database schema..."
        
        # Check for core tables
        CORE_TABLES=("documents_document" "documents_tag" "documents_correspondent" "documents_documenttype")
        for table in "${CORE_TABLES[@]}"; do
            if docker exec "$PG_CONTAINER" psql -U postgres -d paperless -c "\dt $table" > /dev/null 2>&1; then
                print_success "   Table exists: $table"
            else
                print_error "   Table missing: $table"
            fi
        done
        
        # Check document count
        DOC_COUNT=$(docker exec "$PG_CONTAINER" psql -U postgres -d paperless -t -c "SELECT COUNT(*) FROM documents_document;" 2>/dev/null | tr -d ' ')
        if [[ -n "$DOC_COUNT" && "$DOC_COUNT" != "0" ]]; then
            print_success "   Documents in database: $DOC_COUNT"
        else
            print_warning "   No documents found in database"
        fi
        
        # Check tag count
        TAG_COUNT=$(docker exec "$PG_CONTAINER" psql -U postgres -d paperless -t -c "SELECT COUNT(*) FROM documents_tag;" 2>/dev/null | tr -d ' ')
        if [[ -n "$TAG_COUNT" && "$TAG_COUNT" != "0" ]]; then
            print_success "   Tags in database: $TAG_COUNT"
        else
            print_warning "   No tags found in database"
        fi
    fi
fi

echo ""
print_status "8. Recommendations..."

echo "🔧 Based on the analysis, here are the likely issues and solutions:"
echo ""

# Check for common issues
ISSUES_FOUND=0

# Issue 1: Different databases
if docker ps | grep -q "paperless" && docker ps | grep -q "paperless-ai"; then
    PAPERLESS_CONTAINER=$(docker ps --filter "name=paperless" --format "{{.Names}}" | head -1)
    PAPERLESS_AI_CONTAINER=$(docker ps --filter "name=paperless-ai" --format "{{.Names}}" | head -1)
    
    if [[ -n "$PAPERLESS_CONTAINER" && -n "$PAPERLESS_AI_CONTAINER" ]]; then
        PAPERLESS_DB=$(docker exec "$PAPERLESS_CONTAINER" env | grep PAPERLESS_DBHOST | cut -d= -f2)
        PAPERLESS_AI_DB=$(docker exec "$PAPERLESS_AI_CONTAINER" env | grep PAPERLESS_DBHOST | cut -d= -f2)
        
        if [[ "$PAPERLESS_DB" != "$PAPERLESS_AI_DB" ]]; then
            echo "❌ ISSUE 1: Different database hosts"
            echo "   Paperless-ngx: $PAPERLESS_DB"
            echo "   Paperless AI: $PAPERLESS_AI_DB"
            echo "   SOLUTION: Configure both to use the same database"
            echo ""
            ISSUES_FOUND=$((ISSUES_FOUND + 1))
        fi
    fi
fi

# Issue 2: Missing API configuration
if docker ps | grep -q "paperless-ai"; then
    PAPERLESS_AI_CONTAINER=$(docker ps --filter "name=paperless-ai" --format "{{.Names}}" | head -1)
    if [[ -n "$PAPERLESS_AI_CONTAINER" ]]; then
        if ! docker exec "$PAPERLESS_AI_CONTAINER" env | grep -q "PAPERLESS_URL"; then
            echo "❌ ISSUE 2: Missing Paperless-ngx URL configuration"
            echo "   SOLUTION: Set PAPERLESS_URL environment variable"
            echo ""
            ISSUES_FOUND=$((ISSUES_FOUND + 1))
        fi
        
        if ! docker exec "$PAPERLESS_AI_CONTAINER" env | grep -q "PAPERLESS_API_TOKEN\|PAPERLESS_USERNAME"; then
            echo "❌ ISSUE 3: Missing authentication configuration"
            echo "   SOLUTION: Configure PAPERLESS_API_TOKEN or PAPERLESS_USERNAME/PAPERLESS_PASSWORD"
            echo ""
            ISSUES_FOUND=$((ISSUES_FOUND + 1))
        fi
    fi
fi

# Issue 4: No AI providers configured
if docker ps | grep -q "paperless-ai"; then
    PAPERLESS_AI_CONTAINER=$(docker ps --filter "name=paperless-ai" --format "{{.Names}}" | head -1)
    if [[ -n "$PAPERLESS_AI_CONTAINER" ]]; then
        if ! docker exec "$PAPERLESS_AI_CONTAINER" env | grep -E "(OPENAI_API_KEY|OLLAMA_BASE_URL|DEEPSEEK_API_KEY|AZURE_OPENAI_API_KEY)" | grep -v "=" > /dev/null; then
            echo "❌ ISSUE 4: No AI providers configured"
            echo "   SOLUTION: Configure at least one AI provider (OpenAI, Ollama, DeepSeek, or Azure)"
            echo ""
            ISSUES_FOUND=$((ISSUES_FOUND + 1))
        fi
    fi
fi

if [[ $ISSUES_FOUND -eq 0 ]]; then
    echo "✅ No obvious configuration issues found"
    echo "   The problem might be in the application logic or data processing"
fi

echo ""
echo "🚀 To fix these issues, run:"
echo "   ./scripts/setup_paperless_ai_integration.sh"
echo ""
echo "📚 For more information, see:"
echo "   - stacks/ai/paperless-ai.yml (new configuration)"
echo "   - scripts/setup_paperless_ai_integration.sh (setup script)"
echo "   - https://github.com/clusterzx/paperless-ai (documentation)"