@posthog/ai 8.17.2 → 8.18.1

package/dist/index.mjs

@@ -301,6 +301,12 @@ const getModelParams = (params, responseServiceTier) => {
 	if (responseServiceTier != null) modelParams.service_tier = responseServiceTier;
 	return modelParams;
 };
+const uint8ArrayToBase64 = (data) => {
+	if (typeof Buffer !== "undefined") return Buffer.from(data).toString("base64");
+	let binary = "";
+	for (let i = 0; i < data.length; i++) binary += String.fromCharCode(data[i]);
+	return btoa(binary);
+};
 function toSafeString(input) {
 	if (input === void 0 || input === null) return "";
 	if (typeof input === "string") return input;
@@ -412,7 +418,7 @@ function sanitizeValues(obj) {
 }
 //#endregion
 //#region package.json
-var version = "8.17.2";
+var version = "8.18.1";
 //#endregion
 //#region src/serializeError.ts
 const DEFAULT_MAX_DEPTH = 3;
@@ -718,9 +724,38 @@ function isV3Model(model) {
 function getSpecificationVersion(model) {
 	if (typeof model === "object" && model !== null && "specificationVersion" in model) return model.specificationVersion;
 }
-const redactFileData = (data, mediaType, client) => {
-	if (data instanceof URL) return isFullAiCaptureEnabled(client) ? data.toString() : redactBase64DataUrl(data.toString(), data.protocol === "data:" ? mediaType : void 0);
-	if (isString(data)) return isFullAiCaptureEnabled(client) ? data : redactBase64DataUrl(data, mediaType);
+const dataUrlMediaType = (mediaType) => mediaType && /^[\w.+-]+\/[\w.+-]+$/.test(mediaType) ? mediaType : "application/octet-stream";
+const MAX_EVENT_FILE_BYTES = 7864320;
+const newFileBudget = () => ({ remaining: MAX_EVENT_FILE_BYTES });
+const fitsBudget = (length, budget) => {
+	if (length > budget.remaining) return false;
+	budget.remaining -= length;
+	return true;
+};
+const fullCaptureDataUrl = (dataUrl, budget, mediaType) => fitsBudget(dataUrl.length, budget) ? dataUrl : String(redactBase64DataUrl(dataUrl, mediaType));
+const redactFileData = (data, mediaType, client, budget) => {
+	if (data instanceof Uint8Array) {
+		if (!isFullAiCaptureEnabled(client)) return String(redactBase64DataUrl(data, mediaType));
+		const prefix = `data:${dataUrlMediaType(mediaType)};base64,`;
+		if (!fitsBudget(prefix.length + Math.ceil(data.length / 3) * 4, budget)) return `[base64 ${mediaType} redacted]`;
+		return prefix + uint8ArrayToBase64(data);
+	}
+	if (data instanceof URL) return isFullAiCaptureEnabled(client) ? fullCaptureDataUrl(data.toString(), budget, mediaType) : redactBase64DataUrl(data.toString(), data.protocol === "data:" ? mediaType : void 0);
+	if (isString(data)) {
+		if (!isFullAiCaptureEnabled(client)) return redactBase64DataUrl(data, mediaType);
+		if (/^(data:|https?:\/\/)/i.test(data)) return fullCaptureDataUrl(data, budget, mediaType);
+		return fullCaptureDataUrl(`data:${dataUrlMediaType(mediaType)};base64,${data}`, budget, mediaType);
+	}
+};
+const mapOutputFile = (item, client, budget) => {
+	let fileData = redactFileData(item.data, item.mediaType, client, budget) ?? `[binary ${item.mediaType} file]`;
+	if (!isFullAiCaptureEnabled(client) && typeof item.data === "string" && fileData === item.data && item.data.length > 1e3) fileData = `[${item.mediaType} file - ${item.data.length} bytes]`;
+	return {
+		type: "file",
+		name: "generated_file",
+		mediaType: item.mediaType,
+		data: fileData
+	};
 };
 const mapVercelParams = (params) => {
 	return {
@@ -733,7 +768,7 @@ const mapVercelParams = (params) => {
 		stream: params.stream
 	};
 };
-const mapVercelPrompt = (messages, client) => {
+const mapVercelPrompt = (messages, client, budget = newFileBudget()) => {
 	const inputs = messages.map((message) => {
 		let content;
 		if (message.role === "system") content = [{
@@ -747,7 +782,7 @@ const mapVercelPrompt = (messages, client) => {
 			};
 			else if (c.type === "file") return {
 				type: "file",
-				file: redactFileData(c.data, c.mediaType, client) ?? "raw files not supported",
+				file: redactFileData(c.data, c.mediaType, client, budget) ?? "raw files not supported",
 				mediaType: c.mediaType
 			};
 			else if (c.type === "reasoning") return {
@@ -808,7 +843,7 @@ const mapVercelPrompt = (messages, client) => {
 	}
 	return inputs;
 };
-const mapVercelOutput = (result, client) => {
+const mapVercelOutput = (result, client, budget) => {
 	const content = result.map((item) => {
 		if (item.type === "text") return {
 			type: "text",
@@ -830,16 +865,7 @@ const mapVercelOutput = (result, client) => {
 			type: "reasoning",
 			text: truncate(item.text, client)
 		};
-		if (item.type === "file") {
-			let fileData = redactFileData(item.data, item.mediaType, client) ?? `[binary ${item.mediaType} file]`;
-			if (!isFullAiCaptureEnabled(client) && typeof item.data === "string" && fileData === item.data && item.data.length > 1e3) fileData = `[${item.mediaType} file - ${item.data.length} bytes]`;
-			return {
-				type: "file",
-				name: "generated_file",
-				mediaType: item.mediaType,
-				data: fileData
-			};
-		}
+		if (item.type === "file") return mapOutputFile(item, client, budget);
 		if (item.type === "source") return {
 			type: "source",
 			sourceType: item.sourceType,
@@ -979,7 +1005,8 @@ const wrapVercelLanguageModel = (model, phClient, options) => {
 					const result = await model.doGenerate(params);
 					const modelId = mergedOptions.posthogModelOverride ?? (result.response?.modelId ? result.response.modelId : model.modelId);
 					const provider = mergedOptions.posthogProviderOverride ?? extractProvider(model);
-					const content = mapVercelOutput(result.content ?? [], phClient);
+					const fileBudget = newFileBudget();
+					const content = mapVercelOutput(result.content ?? [], phClient, fileBudget);
 					const latency = (Date.now() - startTime) / 1e3;
 					const providerMetadata = result.providerMetadata;
 					const additionalTokenValues = extractAdditionalTokenValues(providerMetadata, result.usage);
@@ -1009,7 +1036,7 @@ const wrapVercelLanguageModel = (model, phClient, options) => {
 						...baseOptions,
 						model: modelId,
 						provider,
-						input: privacyMode ? "" : mapVercelPrompt(params.prompt, phClient),
+						input: privacyMode ? "" : mapVercelPrompt(params.prompt, phClient, fileBudget),
 						output: content,
 						latency,
 						baseURL,
@@ -1059,6 +1086,8 @@ const wrapVercelLanguageModel = (model, phClient, options) => {
 				const provider = mergedOptions.posthogProviderOverride ?? extractProvider(model);
 				const availableTools = extractAvailableToolCalls("vercel", params);
 				const baseURL = extractBaseURL(model);
+				const fileBudget = newFileBudget();
+				const generatedFiles = [];
 				const toolCallsInProgress = /* @__PURE__ */ new Map();
 				const captureStreamGeneration = async (captureOptions) => {
 					try {
@@ -1102,6 +1131,10 @@ const wrapVercelLanguageModel = (model, phClient, options) => {
 								input: chunk.input
 							});
 						}
+						if (chunk.type === "file") {
+							if (firstTokenTime === void 0) firstTokenTime = Date.now();
+							if (!privacyMode) generatedFiles.push(mapOutputFile(chunk, phClient, fileBudget));
+						}
 						if (chunk.type === "error") {
 							hasInBandError = true;
 							inBandError = chunk.error;
@@ -1144,6 +1177,7 @@ const wrapVercelLanguageModel = (model, phClient, options) => {
 									arguments: toolCall.input
 								}
 							});
+							content.push(...generatedFiles);
 							const output = content.length > 0 ? [{
 								role: "assistant",
 								content: content.length === 1 && content[0].type === "text" ? content[0].text : content
@@ -1163,7 +1197,7 @@ const wrapVercelLanguageModel = (model, phClient, options) => {
 								...baseOptions,
 								model: modelId,
 								provider,
-								input: privacyMode ? "" : mapVercelPrompt(params.prompt, phClient),
+								input: privacyMode ? "" : mapVercelPrompt(params.prompt, phClient, fileBudget),
 								output,
 								latency,
 								timeToFirstToken,