@posthog/ai 8.6.4 → 8.6.5

package/dist/vercel/index.mjs

@@ -42,14 +42,16 @@ const FILE_FAMILY_TYPES = new Set(['file', 'input_file', 'document', 'media', 'f
 const KNOWN_AUDIO_FORMATS = new Set(['wav', 'mp3', 'ogg', 'flac', 'm4a', 'aac', 'webm']);
 class MediaTypeContext {
   static EMPTY = new MediaTypeContext(undefined, undefined);
-  constructor(parent, key) {
+  constructor(parent, key, explicitMediaType) {
     this.parent = parent;
     this.key = key;
+    this.explicitMediaType = explicitMediaType;
   }
   inferMediaType() {
     return this.inferFromSiblingMime() ?? this.inferFromSiblingFormat() ?? this.inferFromParentType() ?? this.inferFromKey();
   }
   inferFromSiblingMime() {
+    if (this.explicitMediaType) return this.explicitMediaType;
     if (!this.parent) return undefined;
     for (const hint of MIME_HINT_KEYS) {
       const v = this.parent[hint];
@@ -84,7 +86,13 @@ class MediaTypeContext {
     if (key.includes('file') || key.includes('document')) return 'application/octet-stream';
     return undefined;
   }
+  hasExplicitBinaryMediaType() {
+    if (!this.explicitMediaType && (!this.parent || !this.key || !STRONG_CONTEXT_KEYS.has(this.key))) return false;
+    const mediaType = this.inferFromSiblingMime();
+    return mediaType !== undefined && !mediaType.toLowerCase().startsWith('text/');
+  }
   signalsBinary() {
+    if (this.explicitMediaType) return true;
     if (this.parent) {
       for (const hint of MIME_HINT_KEYS) {
         if (typeof this.parent[hint] === 'string') return true;
@@ -106,10 +114,10 @@ class BinaryContentRedactor {
   constructor(recognizer = new Base64Recognizer()) {
     this.recognizer = recognizer;
   }
-  redact(value) {
+  redact(value, mediaType) {
     if (this.isMultimodalEnabled()) return value;
     this.visited = new WeakSet();
-    return this.walk(value, MediaTypeContext.EMPTY);
+    return this.walk(value, mediaType ? new MediaTypeContext(undefined, undefined, mediaType) : MediaTypeContext.EMPTY);
   }
   walk(value, ctx) {
     if (value === null || value === undefined) return value;
@@ -133,8 +141,10 @@ class BinaryContentRedactor {
     return out;
   }
   redactString(value, ctx) {
-    const minLength = ctx.signalsBinary() ? STRONG_CONTEXT_MIN_LENGTH : WEAK_CONTEXT_MIN_LENGTH;
-    const recognition = this.recognizer.recognize(value, minLength);
+    const hasExplicitBinaryMediaType = ctx.hasExplicitBinaryMediaType();
+    const recognitionValue = hasExplicitBinaryMediaType ? value.replace(/[\r\n]/g, '') : value;
+    const minLength = hasExplicitBinaryMediaType ? Math.min(recognitionValue.length, STRONG_CONTEXT_MIN_LENGTH) : ctx.signalsBinary() ? STRONG_CONTEXT_MIN_LENGTH : WEAK_CONTEXT_MIN_LENGTH;
+    const recognition = this.recognizer.recognize(recognitionValue, minLength);
     switch (recognition.kind) {
       case 'data-url':
         return this.placeholderFor(recognition.mediaType);
@@ -156,9 +166,10 @@ class BinaryContentRedactor {
 }
 
 const redactor = new BinaryContentRedactor();
-function redactBase64DataUrl(str) {
-  return redactor.redact(str);
+function redactBase64DataUrl(str, mediaType) {
+  return redactor.redact(str, mediaType);
 }
+const sanitizeVercel = data => redactor.redact(data);
 
 const TOKEN_PROPERTY_KEYS = new Set(['$ai_input_tokens', '$ai_output_tokens', '$ai_cache_read_input_tokens', '$ai_cache_creation_input_tokens', '$ai_total_tokens', '$ai_reasoning_tokens']);
 function getTokensSource(posthogProperties) {
@@ -395,7 +406,7 @@ function sanitizeValues(obj) {
   return jsonSafe;
 }
 
-var version = "8.6.4";
+var version = "8.6.5";
 
 const DEFAULT_MAX_DEPTH = 3;
 const MAX_STACK_LINES = 20;
@@ -645,6 +656,15 @@ function isV3Model(model) {
 
 // Content types for the output array
 
+const redactFileData = (data, mediaType) => {
+  if (data instanceof URL) {
+    return redactBase64DataUrl(data.toString(), data.protocol === 'data:' ? mediaType : undefined);
+  }
+  if (isString(data)) {
+    return redactBase64DataUrl(data, mediaType);
+  }
+  return undefined;
+};
 const mapVercelParams = params => {
   return {
     temperature: params.temperature,
@@ -677,17 +697,8 @@ const mapVercelPrompt = messages => {
               text: truncate(c.text)
             };
           } else if (c.type === 'file') {
-            // For file type, check if it's a data URL and redact if needed
-            let fileData;
-            const contentData = c.data;
-            if (contentData instanceof URL) {
-              fileData = contentData.toString();
-            } else if (isString(contentData)) {
-              // Redact base64 data URLs and raw base64 to prevent oversized events
-              fileData = redactBase64DataUrl(contentData);
-            } else {
-              fileData = 'raw files not supported';
-            }
+            // Redact base64 data URLs and raw base64 to prevent oversized events
+            const fileData = redactFileData(c.data, c.mediaType) ?? 'raw files not supported';
             return {
               type: 'file',
               file: fileData,
@@ -710,7 +721,7 @@ const mapVercelPrompt = messages => {
               type: 'tool-result',
               toolCallId: c.toolCallId,
               toolName: c.toolName,
-              output: c.output,
+              output: sanitizeVercel(c.output),
               isError: c.isError
             };
           }
@@ -796,18 +807,11 @@ const mapVercelOutput = result => {
     }
     if (item.type === 'file') {
       // Handle files similar to input mapping - avoid large base64 data
-      let fileData;
-      if (item.data instanceof URL) {
-        fileData = item.data.toString();
-      } else if (typeof item.data === 'string') {
-        fileData = redactBase64DataUrl(item.data);
+      let fileData = redactFileData(item.data, item.mediaType) ?? `[binary ${item.mediaType} file]`;
 
-        // If not redacted and still large, replace with size indicator
-        if (fileData === item.data && item.data.length > 1000) {
-          fileData = `[${item.mediaType} file - ${item.data.length} bytes]`;
-        }
-      } else {
-        fileData = `[binary ${item.mediaType} file]`;
+      // If not redacted and still large, replace with size indicator
+      if (typeof item.data === 'string' && fileData === item.data && item.data.length > 1000) {
+        fileData = `[${item.mediaType} file - ${item.data.length} bytes]`;
       }
       return {
         type: 'file',