PozoAppOptimandSEO/server/seo-middleware.js

440 lines
18 KiB
JavaScript

// Express middleware to inject dynamic SEO meta tags
import fs from 'fs';
import path from 'path';
import { fileURLToPath } from 'url';
import axios from 'axios';
import { dirname } from 'path';
const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);
const seoMiddleware = async (req, res, next) => {
// Skip non-HTML requests and static assets
// Skip files with extensions (except .html) - CSS, JS, images, etc.
const hasExtension = req.path.match(/\.[a-z]{2,4}$/i);
if (hasExtension && !req.path.endsWith('.html')) {
return next(); // Let static middleware serve CSS, JS, images, etc.
}
// Skip if it's an API route or static asset
if (req.path.startsWith('/api/') || req.path.startsWith('/assets/') || req.path.startsWith('/home/assets/')) {
return next();
}
// CRITICAL: Process ALL routes that could be HTML pages
// This includes: /, /home, /home/blog, /home/pricing, etc.
// Don't call next() - we will send the response ourselves
console.log('🔍 SEO Middleware checking path:', req.path);
console.log('🔍 Request method:', req.method);
console.log('🔍 Full URL:', req.url);
try {
console.log('==== SEO Middleware triggered for:', req.path);
console.log('🔍 Raw req.path:', req.path);
const normalizedPath = normalizePath(mapToHomeNamespace(req.path));
console.log('Normalized path:', normalizedPath);
console.log('🔍 After normalization:', normalizedPath);
// Attempt dynamic SEO fetch by path/slug first, fallback to pageId map
const routeToPageId = {
'/': 1,
'/home': 1,
'/home/': 1,
'/blog': 5,
'/blog/': 5,
'/home/blog': 5,
'/home/blog/': 5,
'/pricing': 3,
'/pricing/': 3,
'/home/pricing': 3,
'/home/pricing/': 3,
'/pricing-pozoapp': 3,
'/pricing-pozoapp/': 3,
'/home/pricing-pozoapp': 3,
'/home/pricing-pozoapp/': 3,
'/contact-us': 4,
'/contact-us/': 4,
'/home/contact-us': 4,
'/home/contact-us/': 4,
'/signin': 2,
'/signin/': 2,
'/home/signin': 2,
'/home/signin/': 2
};
const pageId = routeToPageId[normalizedPath] || 1;
console.log('Page ID matched:', pageId, 'for path:', normalizedPath);
console.log('Available routes:', Object.keys(routeToPageId).filter(k => routeToPageId[k] === 5));
const fullUrl = buildFullUrl(req);
const baseUrl = `${req.protocol || 'http'}://${req.headers.host || 'localhost:3000'}`;
// Production URL for images - always use production domain for OG images
const PRODUCTION_URL = process.env.PRODUCTION_URL || 'https://www.pozo.app';
// Helper function to ensure image URLs are absolute and use production URL
const ensureAbsoluteImageUrl = (imagePath) => {
if (!imagePath) return null;
// If already absolute URL (starts with http/https), use as-is
if (/^https?:\/\//i.test(imagePath)) {
// If it's localhost, replace with production URL
if (/^https?:\/\/localhost/i.test(imagePath)) {
return imagePath.replace(/^https?:\/\/[^/]+/i, PRODUCTION_URL);
}
return imagePath;
}
// If relative path, make it absolute with production URL
const cleanPath = imagePath.startsWith('/') ? imagePath : `/${imagePath}`;
return `${PRODUCTION_URL}${cleanPath}`;
};
const dynamicSeo = await fetchSEOByPathOrSlug(normalizedPath);
let dbSeoData = dynamicSeo || (await fetchSEOFromDB(pageId));
console.log('SEO Data fetched:', dbSeoData ? 'from DB' : 'using fallback');
if (dbSeoData) {
console.log('DB SEO Data:', {
title: dbSeoData.MetaTitle?.substring(0, 50),
image: dbSeoData.ImageUrl || dbSeoData.image,
pageId: pageId
});
}
const fallbackData = {
1: {
title: 'Retail ERP & POS for Indian MSMEs | POZO',
description: 'Fast billing, smart inventory, GST-ready POS. POZO helps kirana, mini-supermarkets & retail chains speed checkout, connect weighing scales, and manage multi-store ops.',
keywords: 'POS software, retail ERP, billing software, inventory management, GST billing, weighing scale POS, multi-store ERP',
image: `${PRODUCTION_URL}/og/home.jpg`,
url: fullUrl
},
2: {
title: 'Sign In to PozoApp',
description: 'Access your business dashboard and manage your retail operations with POZO ERP & POS system.',
keywords: 'PozoApp login, sign in, business dashboard, retail management',
image: `${PRODUCTION_URL}/og/Signin-og.jpg`,
url: fullUrl
},
3: {
title: 'Pricing - Retail ERP & POS Plans | POZO',
description: 'Simple plans for MSMEs. Fast billing, inventory, GST e-invoice, weighing-scale integration, WhatsApp e-bills & multi-store controls. Book a demo.',
keywords: 'pricing, POS software pricing, retail ERP plans, billing software pricing, inventory management pricing, GST billing plans',
image: `${PRODUCTION_URL}/og/pricing-og.jpg`,
url: fullUrl
},
4: {
title: 'Contact PozoApp',
description: 'Get support and sales information. Contact POZO for retail ERP & POS solutions, billing software, and inventory management systems.',
keywords: 'contact PozoApp, support, sales, retail ERP support, POS software support',
image: `${PRODUCTION_URL}/og/contact-og.jpg`,
url: fullUrl
},
5: {
title: 'POZO Blog — Retail ERP, POS & Grocery Billing Guides',
description: 'Practical guides on POS billing, weighing-scale integration, GST e-invoices, multi-store ERP & inventory control for Indian retailers.',
keywords: 'POS billing, weighing scale integration, GST e-invoices, retail ERP, inventory management, grocery billing guides',
image: `${PRODUCTION_URL}/og/blog-og.jpg`,
url: fullUrl
}
};
const chosenFallback = fallbackData[pageId] || fallbackData[1];
// Ensure image URL is always absolute and uses production URL
// If database image URL is missing or points to wrong page (e.g., home.jpg for blog), use fallback
const dbImageUrl = dbSeoData?.ImageUrl || dbSeoData?.image;
let finalImageUrl = ensureAbsoluteImageUrl(dbImageUrl);
// Validate image URL - if it's pointing to wrong page image, use correct fallback
// For blog page (pageId 5), ensure it's not using home.jpg
if (pageId === 5 && finalImageUrl && finalImageUrl.includes('/og/home.jpg')) {
console.log('Warning: Blog page has home.jpg image, using blog-og.jpg instead');
finalImageUrl = chosenFallback.image;
}
// For home page (pageId 1), ensure it's not using blog-og.jpg
if (pageId === 1 && finalImageUrl && finalImageUrl.includes('/og/blog-og.jpg')) {
console.log('Warning: Home page has blog-og.jpg image, using home.jpg instead');
finalImageUrl = chosenFallback.image;
}
// If no valid image URL, use fallback
if (!finalImageUrl) {
finalImageUrl = chosenFallback.image;
}
// Validate URL - ensure it matches the current page path
let finalUrl = dbSeoData?.CanonicalUrl || dbSeoData?.url || fullUrl;
// If database URL doesn't match the current page path, use correct fullUrl
// Blog page (pageId 5) should have /blog in URL
if (pageId === 5 && finalUrl && !finalUrl.includes('/blog')) {
console.log('Warning: Blog page URL is incorrect, using correct path:', fullUrl);
finalUrl = fullUrl;
}
// Home page (pageId 1) should be root or /home, not other paths
if (pageId === 1 && finalUrl && finalUrl.includes('/blog')) {
console.log('Warning: Home page URL is pointing to blog, using correct path:', fullUrl);
finalUrl = fullUrl;
}
// Pricing page (pageId 3) should have /pricing in URL
if (pageId === 3 && finalUrl && !finalUrl.includes('/pricing')) {
console.log('Warning: Pricing page URL is incorrect, using correct path:', fullUrl);
finalUrl = fullUrl;
}
// CRITICAL: Force correct image and URL based on pageId
// Blog page (pageId 5) MUST use blog-og.jpg, not home.jpg
// Home page (pageId 1) MUST use home.jpg, not blog-og.jpg
if (pageId === 5) {
// Blog page - force blog image and URL
if (!finalImageUrl || finalImageUrl.includes('/og/home.jpg')) {
console.log('FORCING blog-og.jpg for blog page (pageId 5)');
finalImageUrl = `${PRODUCTION_URL}/og/blog-og.jpg`;
}
if (!finalUrl || !finalUrl.includes('/blog')) {
console.log('FORCING blog URL for blog page (pageId 5)');
finalUrl = fullUrl;
}
} else if (pageId === 1) {
// Home page - force home image
if (!finalImageUrl || finalImageUrl.includes('/og/blog-og.jpg')) {
console.log('FORCING home.jpg for home page (pageId 1)');
finalImageUrl = `${PRODUCTION_URL}/og/home.jpg`;
}
}
// Force correct data based on pageId - if database returns wrong page data, use fallback
// This ensures blog page always gets blog data, not home data
let seoData;
if (dbSeoData) {
// Use database data but with forced correct image/URL
seoData = {
title: dbSeoData.MetaTitle || dbSeoData.title || chosenFallback.title,
description: dbSeoData.MetaDesc || dbSeoData.description || chosenFallback.description,
keywords: dbSeoData.Keywords || dbSeoData.keywords || chosenFallback.keywords || '',
image: finalImageUrl, // This is now forced to be correct
url: finalUrl // This is now forced to be correct
};
} else {
// No database data, use fallback (includes keywords)
seoData = chosenFallback;
}
console.log('Final SEO Data:', { title: seoData.title.substring(0, 50), image: seoData.image, url: seoData.url });
// Read HTML - try to read from specific path first, fallback to root index.html
let htmlPath;
const distPath = path.join(__dirname, '..');
// Try to find HTML file for the specific route
// For /home/blog, try dist/home/blog/index.html, then dist/index.html
if (normalizedPath !== '/' && normalizedPath !== '/home' && normalizedPath !== '/home/') {
const routeHtmlPath = path.join(distPath, normalizedPath, 'index.html');
if (fs.existsSync(routeHtmlPath)) {
htmlPath = routeHtmlPath;
console.log('Reading HTML from:', routeHtmlPath);
} else {
// Try without leading slash
const routeHtmlPath2 = path.join(distPath, normalizedPath.replace(/^\//, ''), 'index.html');
if (fs.existsSync(routeHtmlPath2)) {
htmlPath = routeHtmlPath2;
console.log('Reading HTML from:', routeHtmlPath2);
} else {
htmlPath = path.join(distPath, 'index.html');
console.log('Using root index.html');
}
}
} else {
htmlPath = path.join(distPath, 'index.html');
console.log('Using root index.html for home page');
}
let html = fs.readFileSync(htmlPath, 'utf8');
// ULTRA AGGRESSIVE: Remove ALL existing SEO tags (handles multiline tags)
// Match from <meta until /> or > (handles multiline with [\s\S]*?)
const metaTagPattern = /<meta\s+[^>]*?\/?>/gs; // 's' flag makes . match newlines
// Remove all OG tags (multiline aware)
html = html.replace(/<meta\s+property=["']og:type["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+property=["']og:url["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+property=["']og:title["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+property=["']og:description["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+property=["']og:image["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+property=["']og:site_name["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+property=["']og:locale["'][\s\S]*?\/?>/gi, '');
// Remove Twitter tags (multiline aware)
html = html.replace(/<meta\s+(name|property)=["']twitter:card["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+(name|property)=["']twitter:url["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+(name|property)=["']twitter:title["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+(name|property)=["']twitter:description["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+(name|property)=["']twitter:image["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+(name|property)=["']twitter:site["'][\s\S]*?\/?>/gi, '');
// Remove description and keywords (multiline aware)
html = html.replace(/<meta\s+name=["']description["'][\s\S]*?\/?>/gi, '');
html = html.replace(/<meta\s+name=["']keywords["'][\s\S]*?\/?>/gi, '');
// Update title
html = html.replace(/<title>[^<]*<\/title>/i, `<title>${escapeHtml(seoData.title)}</title>`);
// Update description
html = html.replace(/<meta\s+name="description"[^>]*\/?>/i, `<meta name="description" content="${escapeHtml(seoData.description)}">`);
// Update canonical
html = html.replace(/<link\s+rel="canonical"[^>]*\/?>/i, `<link rel="canonical" href="${escapeAttribute(seoData.url)}">`);
// Add fresh SEO tags before </head>
// Include keywords if available
const keywordsTag = seoData.keywords ? `<meta name="keywords" content="${escapeHtml(seoData.keywords)}" />` : '';
const seoTags = `
<!-- SEO Meta Tags -->
${keywordsTag}
<!-- Open Graph / Facebook -->
<meta property="og:type" content="website" />
<meta property="og:url" content="${escapeAttribute(seoData.url)}" />
<meta property="og:title" content="${escapeAttribute(seoData.title)}" />
<meta property="og:description" content="${escapeAttribute(seoData.description)}" />
<meta property="og:image" content="${escapeAttribute(seoData.image)}" />
<!-- Twitter -->
<meta name="twitter:card" content="summary_large_image" />
<meta name="twitter:url" content="${escapeAttribute(seoData.url)}" />
<meta name="twitter:title" content="${escapeAttribute(seoData.title)}" />
<meta name="twitter:description" content="${escapeAttribute(seoData.description)}" />
<meta name="twitter:image" content="${escapeAttribute(seoData.image)}" />`;
html = html.replace(/<\/head>/i, `${seoTags}\n</head>`);
res.send(html);
} catch (error) {
console.error('SEO middleware error:', error);
next();
}
};
// Function to fetch SEO data from your database
async function fetchSEOFromDB(pageId) {
try {
const API_URL = process.env.API_URL || 'https://api.pozo.app';
const response = await axios.get(`${API_URL}/Seo?PageId=${pageId}`);
if (response.data?.statusCode === 1 && response.data?.data?.length > 0) {
return response.data.data[0];
}
return null;
} catch (error) {
console.error('Failed to fetch SEO data from database:', error.message);
return null;
}
}
// Try to fetch SEO by exact path or blog slug, with graceful fallbacks
async function fetchSEOByPathOrSlug(normalizedPath) {
try {
const API_URL = process.env.API_URL || 'https://api.pozo.app';
// 1) Try path-based SEO: /Seo?Path=/home/blog/my-post
const byPath = await safeGet(`${API_URL}/Seo`, { Path: normalizedPath });
if (byPath) return byPath;
// 2) If looks like blog detail: /blog/slug or /home/blog/slug
if (/^\/(home\/)?blog\//.test(normalizedPath)) {
const slug = normalizedPath.replace(/^\/(home\/)?blog\//, '').replace(/\/$/, '');
if (slug) {
// Try common blog SEO endpoints
const blogSeo = await safeGet(`${API_URL}/Seo/Blog`, { slug })
|| await safeGet(`${API_URL}/Blog/Seo`, { slug })
|| await safeGet(`${API_URL}/Blog`, { slug });
if (blogSeo) return blogSeo;
}
}
return null;
} catch (e) {
return null;
}
}
async function safeGet(baseUrl, queryObj) {
try {
const qs = new URLSearchParams(queryObj).toString();
const url = `${baseUrl}?${qs}`;
const response = await axios.get(url);
const data = response.data;
if (data?.statusCode === 1 && Array.isArray(data?.data) && data.data.length > 0) {
return data.data[0];
}
// Some APIs return object directly
if (data && typeof data === 'object' && !Array.isArray(data)) return data;
return null;
} catch (e) {
return null;
}
}
function normalizePath(p) {
if (!p) return '/';
try {
// Remove query/hash, ensure leading slash, collapse duplicate slashes
const onlyPath = p.split('?')[0].split('#')[0] || '/';
// Collapse multiple slashes to single slash, but preserve path structure
let normalized = onlyPath.replace(/\/+/g, '/');
if (!normalized.startsWith('/')) normalized = `/${normalized}`;
// Remove trailing slash except for root
if (normalized.length > 1 && normalized.endsWith('/')) {
normalized = normalized.slice(0, -1);
}
console.log('🔍 normalizePath input:', p, '→ output:', normalized);
return normalized;
} catch {
return '/';
}
}
// Map routes - support both /home/* and clean routes
function mapToHomeNamespace(p) {
if (!p) return '/';
const raw = p.split('?')[0].split('#')[0] || '/';
// Just normalize and return - we support both /home/ and clean routes now
return normalizePath(raw);
}
function buildFullUrl(req) {
const proto = (req.headers['x-forwarded-proto'] || req.protocol || 'https').split(',')[0];
const host = req.headers['x-forwarded-host'] || req.headers.host || 'www.pozo.app';
const pathOnly = normalizePath(req.path);
return `${proto}://${host}${pathOnly}`;
}
function replaceOrInsert(html, regex, newTagHtml) {
if (regex.test(html)) {
return html.replace(regex, newTagHtml);
}
// Insert before </head>
if (/<\/head>/i.test(html)) {
return html.replace(/<\/head>/i, `${newTagHtml}\n</head>`);
}
// As a last resort, prepend to document
return `${newTagHtml}\n${html}`;
}
function escapeHtml(str) {
return String(str || '')
.replace(/&/g, '&amp;')
.replace(/</g, '&lt;')
.replace(/>/g, '&gt;');
}
function escapeAttribute(str) {
return String(str || '')
.replace(/&/g, '&amp;')
.replace(/"/g, '&quot;')
.replace(/</g, '&lt;')
.replace(/>/g, '&gt;');
}
export default seoMiddleware;