@posthog/ai 6.0.1 → 6.1.1

package/dist/langchain/index.mjs

@@ -535,6 +535,134 @@ class BaseCallbackHandler extends BaseCallbackHandlerMethodsClass {
     }
 }
 
+// Type guards for safer type checking
+
+const isString = value => {
+  return typeof value === 'string';
+};
+const isObject = value => {
+  return value !== null && typeof value === 'object' && !Array.isArray(value);
+};
+
+const REDACTED_IMAGE_PLACEHOLDER = '[base64 image redacted]';
+
+// ============================================
+// Base64 Detection Helpers
+// ============================================
+
+const isBase64DataUrl = str => {
+  return /^data:([^;]+);base64,/.test(str);
+};
+const isValidUrl = str => {
+  try {
+    new URL(str);
+    return true;
+  } catch {
+    // Not an absolute URL, check if it's a relative URL or path
+    return str.startsWith('/') || str.startsWith('./') || str.startsWith('../');
+  }
+};
+const isRawBase64 = str => {
+  // Skip if it's a valid URL or path
+  if (isValidUrl(str)) {
+    return false;
+  }
+
+  // Check if it's a valid base64 string
+  // Base64 images are typically at least a few hundred chars, but we'll be conservative
+  return str.length > 20 && /^[A-Za-z0-9+/]+=*$/.test(str);
+};
+function redactBase64DataUrl(str) {
+  if (!isString(str)) return str;
+
+  // Check for data URL format
+  if (isBase64DataUrl(str)) {
+    return REDACTED_IMAGE_PLACEHOLDER;
+  }
+
+  // Check for raw base64 (Vercel sends raw base64 for inline images)
+  if (isRawBase64(str)) {
+    return REDACTED_IMAGE_PLACEHOLDER;
+  }
+  return str;
+}
+
+// ============================================
+// Common Message Processing
+// ============================================
+
+const processMessages = (messages, transformContent) => {
+  if (!messages) return messages;
+  const processContent = content => {
+    if (typeof content === 'string') return content;
+    if (!content) return content;
+    if (Array.isArray(content)) {
+      return content.map(transformContent);
+    }
+
+    // Handle single object content
+    return transformContent(content);
+  };
+  const processMessage = msg => {
+    if (!isObject(msg) || !('content' in msg)) return msg;
+    return {
+      ...msg,
+      content: processContent(msg.content)
+    };
+  };
+
+  // Handle both arrays and single messages
+  if (Array.isArray(messages)) {
+    return messages.map(processMessage);
+  }
+  return processMessage(messages);
+};
+const sanitizeLangChainImage = item => {
+  if (!isObject(item)) return item;
+
+  // OpenAI style
+  if (item.type === 'image_url' && 'image_url' in item && isObject(item.image_url) && 'url' in item.image_url) {
+    return {
+      ...item,
+      image_url: {
+        ...item.image_url,
+        url: redactBase64DataUrl(item.image_url.url)
+      }
+    };
+  }
+
+  // Direct image with data field
+  if (item.type === 'image' && 'data' in item) {
+    return {
+      ...item,
+      data: redactBase64DataUrl(item.data)
+    };
+  }
+
+  // Anthropic style
+  if (item.type === 'image' && 'source' in item && isObject(item.source) && 'data' in item.source) {
+    return {
+      ...item,
+      source: {
+        ...item.source,
+        data: redactBase64DataUrl(item.source.data)
+      }
+    };
+  }
+
+  // Google style
+  if (item.type === 'media' && 'data' in item) {
+    return {
+      ...item,
+      data: redactBase64DataUrl(item.data)
+    };
+  }
+  return item;
+};
+const sanitizeLangChain = data => {
+  return processMessages(data, sanitizeLangChainImage);
+};
+
 /** A run may either be a Span or a Generation */
 
 /** Storage for run metadata */
@@ -715,7 +843,7 @@ class LangChainCallbackHandler extends BaseCallbackHandler {
     }) || 'generation';
     const generation = {
       name: runNameFound,
-      input: messages,
+      input: sanitizeLangChain(messages),
       startTime: Date.now()
     };
     if (extraParams) {
@@ -980,7 +1108,9 @@ class LangChainCallbackHandler extends BaseCallbackHandler {
         ...message.additional_kwargs
       };
     }
-    return messageDict;
+
+    // Sanitize the message content to redact base64 images
+    return sanitizeLangChain(messageDict);
   }
   _parseUsageModel(usage) {
     const conversionList = [['promptTokens', 'input'], ['completionTokens', 'output'], ['input_tokens', 'input'], ['output_tokens', 'output'], ['prompt_token_count', 'input'], ['candidates_token_count', 'output'], ['inputTokenCount', 'input'], ['outputTokenCount', 'output'], ['input_token_count', 'input'], ['generated_token_count', 'output']];