@paywalls-net/filter 1.1.1 → 1.2.0

package/README.md

@@ -9,9 +9,9 @@ npm install @paywalls-net/filter
 ```
 
 ## Environment Variables
-- `PAYWALLS_PUBLISHER_ID`: The unique identifier for the publisher using Paywalls.net services.
-- `PAYWALLS_CLOUD_API_HOST`: The host for the Paywalls.net API. eg `https://cloud-api.paywalls.net`.
-- `PAYWALLS_CLOUD_API_KEY`: The API key for accessing Paywalls.net services. NOTE: This key should be treated like a password and kept secret and stored in a secure secrets vault or environment variable.
+- `PAYWALLS_PUBLISHER_ID`: The unique identifier for the publisher using paywalls.net services.
+- `PAYWALLS_CLOUD_API_HOST`: The host for the paywalls.net API. eg `https://cloud-api.paywalls.net`.
+- `PAYWALLS_CLOUD_API_KEY`: The API key for accessing paywalls.net services. NOTE: This key should be treated like a password and kept secret and stored in a secure secrets vault or environment variable.
 
 ## Usage
 ```javascript

package/package.json

@@ -3,7 +3,7 @@
   "description": "Client SDK for integrating paywalls.net bot filtering and authorization services into your server or CDN.",
   "author": "paywalls.net",
   "license": "MIT",
-  "version": "1.1.1",
+  "version": "1.2.0",
   "publishConfig": {
     "access": "public"
   },

package/src/index.js

@@ -140,7 +140,7 @@ async function isRecognizedBot(cfg, request) {
 }
 
 
-function sendResponse(authz) {
+function setHeaders(authz) {
     let headers = {
         "Content-Type": "text/html",
     };
@@ -157,6 +157,38 @@ function sendResponse(authz) {
     });
 }
 
+function setCloudFrontHeaders(authz) {
+    const headers = {};
+
+    if (authz.response?.headers) {
+        for (const [key, value] of Object.entries(authz.response.headers)) {
+            headers[key.toLowerCase()] = [
+                {
+                    key: key,
+                    value: value
+                }
+            ];
+        }
+    }
+
+    // Add default Content-Type header if not already set
+    if (!headers["content-type"]) {
+        headers["content-type"] = [
+            {
+                key: "Content-Type",
+                value: "text/html"
+            }
+        ];
+    }
+
+    return {
+        status: authz.response?.code || 402,
+        statusDescription: authz.response?.code === 402 ? "Payment Required" : "Error",
+        headers: headers,
+        body: authz.response?.html || "Payment required."
+    };
+}
+
 /**
  * Detect AI Bot and authorize it using paywalls.net.
  * @param {Request} request 
@@ -174,19 +206,17 @@ async function cloudflare(config = null) {
         };
         await loadAgentPatterns(paywallsConfig);
 
-        if (isRecognizedBot(paywallsConfig, request)) {
+        if (await isRecognizedBot(paywallsConfig, request)) {
             const authz = await checkAgentStatus(paywallsConfig, request);
 
             ctx.waitUntil(logAccess(paywallsConfig, request, authz));
 
             if (authz.access === 'deny') {
-                return sendResponse(authz);
+                return setHeaders(authz);
             } else {
                 console.log("Bot-like request allowed. Proceeding to origin/CDN.");
             }
         }
-
-        return fetch(request); // Proceed to origin/CDN
     };
 }
 
@@ -200,17 +230,37 @@ async function fastly(config) {
     await loadAgentPatterns(paywallsConfig);
 
     return async function handle(request) {
-        if (isRecognizedBot(paywallsConfig,request)) {
+        if (await isRecognizedBot(paywallsConfig, request)) {
+            const authz = await checkAgentStatus(paywallsConfig, request);
+
+            await logAccess(paywallsConfig, request, authz);
+
+            if (authz.access === 'deny') {
+                return setHeaders(authz);
+            }
+        }
+    };
+}
+
+async function cloudfront(config) {
+    const paywallsConfig = {
+        paywallsAPIHost: config.get('PAYWALLS_CLOUD_API_HOST'),
+        paywallsAPIKey: config.get('PAYWALLS_API_KEY'),
+        paywallsPublisherId: config.get('PAYWALLS_PUBLISHER_ID')
+    };
+    await loadAgentPatterns(paywallsConfig);
+
+    return async function handle(request) {
+        if (await isRecognizedBot(paywallsConfig, request)) {
             const authz = await checkAgentStatus(paywallsConfig, request);
 
             await logAccess(paywallsConfig, request, authz);
 
             if (authz.access === 'deny') {
-                return sendResponse(authz);
+                return setCloudFrontHeaders(authz);
             }
         }
 
-        return fetch(request, { backend: 'origin' });
     };
 }
 
@@ -228,6 +278,8 @@ export async function init(cdn, config = {}) {
             return await cloudflare(config);
         case 'fastly':
             return await fastly(config);
+        case 'cloudfront':
+            return await cloudfront(config);
         default:
             throw new Error(`Unsupported CDN: ${cdn}`);
     }

package/src/user-agent-classification.js

@@ -27,15 +27,22 @@ export async function loadAgentPatterns(cfg) {
         });
 
         if (!response.ok) {
-            throw new Error(`Failed to fetch user agent patterns: ${response.status} ${response.statusText}`);
+            throw new Error(`Failed to fetch agent patterns: ${response.status} ${response.statusText}`);
         }
 
-        cachedUserAgentPatterns = await response.json();
+        const serializedPatterns = await response.json();
+
+        // Deserialize RegExp strings back into RegExp objects
+        cachedUserAgentPatterns = serializedPatterns.map((pattern) => ({
+            ...pattern,
+            patterns: pattern.patterns.map((regexString) => new RegExp(regexString.slice(1, -1))) // Remove leading and trailing slashes
+        }));
+
         cacheTimestamp = now;
         return cachedUserAgentPatterns;
     } catch (error) {
-        console.error('Failed to fetch user agent patterns:', error);
-        throw new Error('Could not load user agent patterns');
+        console.error('Error loading agent patterns:', error);
+        throw error;
     }
 }
 
@@ -56,7 +63,7 @@ export async function classifyUserAgent(cfg, userAgent) {
     for (const config of userAgentPatterns) {
         if (!config.patterns) continue;
         for (const pattern of config.patterns) {
-            if (new RegExp(pattern).test(userAgent)) {
+            if (new RegExp(pattern).test(userAgent)) {         
                 return {
                     operator: config.operator,
                     agent: config.agent || browser,