shumengya/sproutclaw
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Revert "fix(ai): own Anthropic SSE parsing to avoid SDK JSON.parse hard-failures"

Browse Source 
This reverts commit 4b926a30a2.
This commit is contained in:
Mario Zechner
2026-04-21 23:19:25 +02:00
parent 81de426f96
commit fc9220d2de
12 changed files with 124 additions and 430 deletions
Download Patch File Download Diff File Expand all files Collapse all files

181
packages/ai/src/providers/anthropic.ts
View File

@@ -4,7 +4,6 @@ import type {
ContentBlockParam, ContentBlockParam,
MessageCreateParamsStreaming, MessageCreateParamsStreaming,
MessageParam, MessageParam,
RawMessageStreamEvent,
} from "@anthropic-ai/sdk/resources/messages.js"; } from "@anthropic-ai/sdk/resources/messages.js";
import { getEnvApiKey } from "../env-api-keys.js"; import { getEnvApiKey } from "../env-api-keys.js";
import { calculateCost } from "../models.js"; import { calculateCost } from "../models.js";
@@ -28,7 +27,7 @@ import type {
} from "../types.js"; } from "../types.js";
import { AssistantMessageEventStream } from "../utils/event-stream.js"; import { AssistantMessageEventStream } from "../utils/event-stream.js";
import { headersToRecord } from "../utils/headers.js"; import { headersToRecord } from "../utils/headers.js";
import { parseJsonWithRepair, parseStreamingJson } from "../utils/json-parse.js"; import { parseStreamingJson } from "../utils/json-parse.js";
import { sanitizeSurrogates } from "../utils/sanitize-unicode.js"; import { sanitizeSurrogates } from "../utils/sanitize-unicode.js";
import { buildCopilotDynamicHeaders, hasCopilotVisionInput } from "./github-copilot-headers.js"; import { buildCopilotDynamicHeaders, hasCopilotVisionInput } from "./github-copilot-headers.js";
@@ -214,176 +213,6 @@ function mergeHeaders(...headerSources: (Record<string, string> | undefined)[]):
return merged; return merged;
} }
interface ServerSentEvent {
event: string | null;
data: string;
raw: string[];
}
interface SseDecoderState {
event: string | null;
data: string[];
raw: string[];
}
function flushSseEvent(state: SseDecoderState): ServerSentEvent | null {
if (!state.event && state.data.length === 0) {
return null;
}
const event: ServerSentEvent = {
event: state.event,
data: state.data.join("\n"),
raw: [...state.raw],
};
state.event = null;
state.data = [];
state.raw = [];
return event;
}
function decodeSseLine(line: string, state: SseDecoderState): ServerSentEvent | null {
if (line === "") {
return flushSseEvent(state);
}
state.raw.push(line);
if (line.startsWith(":")) {
return null;
}
const delimiterIndex = line.indexOf(":");
const fieldName = delimiterIndex === -1 ? line : line.slice(0, delimiterIndex);
let value = delimiterIndex === -1 ? "" : line.slice(delimiterIndex + 1);
if (value.startsWith(" ")) {
value = value.slice(1);
}
if (fieldName === "event") {
state.event = value;
} else if (fieldName === "data") {
state.data.push(value);
}
return null;
}
function nextLineBreakIndex(text: string): number {
const carriageReturnIndex = text.indexOf("\r");
const newlineIndex = text.indexOf("\n");
if (carriageReturnIndex === -1) {
return newlineIndex;
}
if (newlineIndex === -1) {
return carriageReturnIndex;
}
return Math.min(carriageReturnIndex, newlineIndex);
}
function consumeLine(text: string): { line: string; rest: string } | null {
const lineBreakIndex = nextLineBreakIndex(text);
if (lineBreakIndex === -1) {
return null;
}
let nextIndex = lineBreakIndex + 1;
if (text[lineBreakIndex] === "\r" && text[nextIndex] === "\n") {
nextIndex += 1;
}
return {
line: text.slice(0, lineBreakIndex),
rest: text.slice(nextIndex),
};
}
async function* iterateSseMessages(
body: ReadableStream<Uint8Array>,
signal?: AbortSignal,
): AsyncGenerator<ServerSentEvent> {
const reader = body.getReader();
const decoder = new TextDecoder();
const state: SseDecoderState = { event: null, data: [], raw: [] };
let buffer = "";
try {
while (true) {
if (signal?.aborted) {
throw new Error("Request was aborted");
}
const { value, done } = await reader.read();
if (done) {
break;
}
buffer += decoder.decode(value, { stream: true });
let consumed = consumeLine(buffer);
while (consumed) {
buffer = consumed.rest;
const event = decodeSseLine(consumed.line, state);
if (event) {
yield event;
}
consumed = consumeLine(buffer);
}
}
buffer += decoder.decode();
let consumed = consumeLine(buffer);
while (consumed) {
buffer = consumed.rest;
const event = decodeSseLine(consumed.line, state);
if (event) {
yield event;
}
consumed = consumeLine(buffer);
}
if (buffer.length > 0) {
const event = decodeSseLine(buffer, state);
if (event) {
yield event;
}
}
const trailingEvent = flushSseEvent(state);
if (trailingEvent) {
yield trailingEvent;
}
} finally {
reader.releaseLock();
}
}
async function* iterateAnthropicEvents(
response: Response,
signal?: AbortSignal,
): AsyncGenerator<RawMessageStreamEvent> {
if (!response.body) {
throw new Error("Attempted to iterate over an Anthropic response with no body");
}
for await (const sse of iterateSseMessages(response.body, signal)) {
if (!sse.event || sse.event === "ping") {
continue;
}
if (sse.event === "error") {
throw new Error(sse.data);
}
try {
yield parseJsonWithRepair<RawMessageStreamEvent>(sse.data);
} catch (error) {
const message = error instanceof Error ? error.message : String(error);
throw new Error(
`Could not parse Anthropic SSE event ${sse.event}: ${message}; data=${sse.data}; raw=${sse.raw.join("\\n")}`,
);
}
}
}
export const streamAnthropic: StreamFunction<"anthropic-messages", AnthropicOptions> = ( export const streamAnthropic: StreamFunction<"anthropic-messages", AnthropicOptions> = (
model: Model<"anthropic-messages">, model: Model<"anthropic-messages">,
context: Context, context: Context,
@@ -444,16 +273,16 @@ export const streamAnthropic: StreamFunction<"anthropic-messages", AnthropicOpti
if (nextParams !== undefined) { if (nextParams !== undefined) {
params = nextParams as MessageCreateParamsStreaming; params = nextParams as MessageCreateParamsStreaming;
} }
const response = await client.messages const { data: anthropicStream, response } = await client.messages
.create({ ...params, stream: true }, { signal: options?.signal }) .stream({ ...params, stream: true }, { signal: options?.signal })
.asResponse(); .withResponse();
await options?.onResponse?.({ status: response.status, headers: headersToRecord(response.headers) }, model); await options?.onResponse?.({ status: response.status, headers: headersToRecord(response.headers) }, model);
stream.push({ type: "start", partial: output }); stream.push({ type: "start", partial: output });
type Block = (ThinkingContent | TextContent | (ToolCall & { partialJson: string })) & { index: number }; type Block = (ThinkingContent | TextContent | (ToolCall & { partialJson: string })) & { index: number };
const blocks = output.content as Block[]; const blocks = output.content as Block[];
for await (const event of iterateAnthropicEvents(response, options?.signal)) { for await (const event of anthropicStream) {
if (event.type === "message_start") { if (event.type === "message_start") {
output.responseId = event.message.id; output.responseId = event.message.id;
// Capture initial token usage from message_start event // Capture initial token usage from message_start event

108
packages/ai/src/utils/json-parse.ts
View File

@@ -1,99 +1,5 @@
import { parse as partialParse } from "partial-json"; import { parse as partialParse } from "partial-json";
const VALID_JSON_ESCAPES = new Set(['"', "\\", "/", "b", "f", "n", "r", "t", "u"]);
function isControlCharacter(char: string): boolean {
const codePoint = char.codePointAt(0);
return codePoint !== undefined && codePoint >= 0x00 && codePoint <= 0x1f;
}
function escapeControlCharacter(char: string): string {
switch (char) {
case "\b":
return "\\b";
case "\f":
return "\\f";
case "\n":
return "\\n";
case "\r":
return "\\r";
case "\t":
return "\\t";
default:
return `\\u${char.codePointAt(0)?.toString(16).padStart(4, "0") ?? "0000"}`;
}
}
/**
* Repairs malformed JSON string literals by:
* - escaping raw control characters inside strings
* - doubling backslashes before invalid escape characters
*/
export function repairJson(json: string): string {
let repaired = "";
let inString = false;
for (let index = 0; index < json.length; index++) {
const char = json[index];
if (!inString) {
repaired += char;
if (char === '"') {
inString = true;
}
continue;
}
if (char === '"') {
repaired += char;
inString = false;
continue;
}
if (char === "\\") {
const nextChar = json[index + 1];
if (nextChar === undefined) {
repaired += "\\\\";
continue;
}
if (nextChar === "u") {
const unicodeDigits = json.slice(index + 2, index + 6);
if (/^[0-9a-fA-F]{4}$/.test(unicodeDigits)) {
repaired += `\\u${unicodeDigits}`;
index += 5;
continue;
}
}
if (VALID_JSON_ESCAPES.has(nextChar)) {
repaired += `\\${nextChar}`;
index += 1;
continue;
}
repaired += "\\\\";
continue;
}
repaired += isControlCharacter(char) ? escapeControlCharacter(char) : char;
}
return repaired;
}
export function parseJsonWithRepair<T>(json: string): T {
try {
return JSON.parse(json) as T;
} catch (error) {
const repairedJson = repairJson(json);
if (repairedJson !== json) {
return JSON.parse(repairedJson) as T;
}
throw error;
}
}
/** /**
* Attempts to parse potentially incomplete JSON during streaming. * Attempts to parse potentially incomplete JSON during streaming.
* Always returns a valid object, even if the JSON is incomplete. * Always returns a valid object, even if the JSON is incomplete.
@@ -101,24 +7,22 @@ export function parseJsonWithRepair<T>(json: string): T {
* @param partialJson The partial JSON string from streaming * @param partialJson The partial JSON string from streaming
* @returns Parsed object or empty object if parsing fails * @returns Parsed object or empty object if parsing fails
*/ */
export function parseStreamingJson<T = Record<string, unknown>>(partialJson: string | undefined): T { export function parseStreamingJson<T = any>(partialJson: string | undefined): T {
if (!partialJson || partialJson.trim() === "") { if (!partialJson || partialJson.trim() === "") {
return {} as T; return {} as T;
} }
// Try standard parsing first (fastest for complete JSON)
try { try {
return parseJsonWithRepair<T>(partialJson); return JSON.parse(partialJson) as T;
} catch { } catch {
// Try partial-json for incomplete JSON
try { try {
const result = partialParse(partialJson); const result = partialParse(partialJson);
return (result ?? {}) as T; return (result ?? {}) as T;
} catch { } catch {
try { // If all parsing fails, return empty object
const result = partialParse(repairJson(partialJson)); return {} as T;
return (result ?? {}) as T;
} catch {
return {} as T;
}
} }
} }
} }

113
packages/ai/test/anthropic-sse-parsing.test.ts
View File

@@ -1,113 +0,0 @@
import type Anthropic from "@anthropic-ai/sdk";
import { Type } from "@sinclair/typebox";
import { describe, expect, it } from "vitest";
import { getModel } from "../src/models.js";
import { streamAnthropic } from "../src/providers/anthropic.js";
import type { Context, ToolCall } from "../src/types.js";
function createSseResponse(events: Array<{ event: string; data: string }>): Response {
const body = events.map(({ event, data }) => `event: ${event}\ndata: ${data}\n`).join("\n");
return new Response(body, {
status: 200,
headers: { "content-type": "text/event-stream" },
});
}
function createFakeAnthropicClient(response: Response): Anthropic {
return {
messages: {
create: () => ({
asResponse: async () => response,
}),
},
} as unknown as Anthropic;
}
describe("Anthropic raw SSE parsing", () => {
it("repairs malformed SSE JSON and malformed streamed tool JSON", async () => {
const model = getModel("anthropic", "claude-haiku-4-5");
const context: Context = {
messages: [{ role: "user", content: "Use the edit tool.", timestamp: Date.now() }],
tools: [
{
name: "edit",
description: "Edit a file.",
parameters: Type.Object({
path: Type.String(),
text: Type.String(),
}),
},
],
};
const malformedToolJsonDelta = String.raw`{"type":"content_block_delta","index":0,"delta":{"type":"input_json_delta","partial_json":"{\"path\":\"A\H\",\"text\":\"col1 col2\"}"}}`;
const response = createSseResponse([
{
event: "message_start",
data: JSON.stringify({
type: "message_start",
message: {
id: "msg_test",
usage: {
input_tokens: 12,
output_tokens: 0,
cache_read_input_tokens: 0,
cache_creation_input_tokens: 0,
},
},
}),
},
{
event: "content_block_start",
data: JSON.stringify({
type: "content_block_start",
index: 0,
content_block: {
type: "tool_use",
id: "toolu_test",
name: "edit",
input: {},
},
}),
},
{ event: "content_block_delta", data: malformedToolJsonDelta },
{
event: "content_block_stop",
data: JSON.stringify({ type: "content_block_stop", index: 0 }),
},
{
event: "message_delta",
data: JSON.stringify({
type: "message_delta",
delta: { stop_reason: "tool_use" },
usage: {
input_tokens: 12,
output_tokens: 5,
cache_read_input_tokens: 0,
cache_creation_input_tokens: 0,
},
}),
},
{
event: "message_stop",
data: JSON.stringify({ type: "message_stop" }),
},
]);
const stream = streamAnthropic(model, context, {
client: createFakeAnthropicClient(response),
});
const result = await stream.result();
expect(result.stopReason).toBe("toolUse");
expect(result.errorMessage).toBeUndefined();
const toolCall = result.content.find((block): block is ToolCall => block.type === "toolCall");
expect(toolCall).toBeDefined();
expect(toolCall?.arguments).toEqual({
path: "A\\H",
text: "col1\tcol2",
});
});
});

12
packages/ai/test/cache-retention.test.ts
View File

@@ -27,7 +27,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
it.skipIf(!process.env.ANTHROPIC_API_KEY)( it.skipIf(!process.env.ANTHROPIC_API_KEY)(
"should use default cache TTL (no ttl field) when PI_CACHE_RETENTION is not set", "should use default cache TTL (no ttl field) when PI_CACHE_RETENTION is not set",
async () => { async () => {
const model = getModel("anthropic", "claude-haiku-4-5"); const model = getModel("anthropic", "claude-3-5-haiku-20241022");
let capturedPayload: any = null; let capturedPayload: any = null;
const s = stream(model, context, { const s = stream(model, context, {
@@ -50,7 +50,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
it.skipIf(!process.env.ANTHROPIC_API_KEY)("should use 1h cache TTL when PI_CACHE_RETENTION=long", async () => { it.skipIf(!process.env.ANTHROPIC_API_KEY)("should use 1h cache TTL when PI_CACHE_RETENTION=long", async () => {
process.env.PI_CACHE_RETENTION = "long"; process.env.PI_CACHE_RETENTION = "long";
const model = getModel("anthropic", "claude-haiku-4-5"); const model = getModel("anthropic", "claude-3-5-haiku-20241022");
let capturedPayload: any = null; let capturedPayload: any = null;
const s = stream(model, context, { const s = stream(model, context, {
@@ -74,7 +74,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
process.env.PI_CACHE_RETENTION = "long"; process.env.PI_CACHE_RETENTION = "long";
// Create a model with a different baseUrl (simulating a proxy) // Create a model with a different baseUrl (simulating a proxy)
const baseModel = getModel("anthropic", "claude-haiku-4-5"); const baseModel = getModel("anthropic", "claude-3-5-haiku-20241022");
const proxyModel = { const proxyModel = {
...baseModel, ...baseModel,
baseUrl: "https://my-proxy.example.com/v1", baseUrl: "https://my-proxy.example.com/v1",
@@ -114,7 +114,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
}); });
it("should omit cache_control when cacheRetention is none", async () => { it("should omit cache_control when cacheRetention is none", async () => {
const baseModel = getModel("anthropic", "claude-haiku-4-5"); const baseModel = getModel("anthropic", "claude-3-5-haiku-20241022");
let capturedPayload: any = null; let capturedPayload: any = null;
const { streamAnthropic } = await import("../src/providers/anthropic.js"); const { streamAnthropic } = await import("../src/providers/anthropic.js");
@@ -140,7 +140,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
}); });
it("should add cache_control to string user messages", async () => { it("should add cache_control to string user messages", async () => {
const baseModel = getModel("anthropic", "claude-haiku-4-5"); const baseModel = getModel("anthropic", "claude-3-5-haiku-20241022");
let capturedPayload: any = null; let capturedPayload: any = null;
const { streamAnthropic } = await import("../src/providers/anthropic.js"); const { streamAnthropic } = await import("../src/providers/anthropic.js");
@@ -168,7 +168,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
}); });
it("should set 1h cache TTL when cacheRetention is long", async () => { it("should set 1h cache TTL when cacheRetention is long", async () => {
const baseModel = getModel("anthropic", "claude-haiku-4-5"); const baseModel = getModel("anthropic", "claude-3-5-haiku-20241022");
let capturedPayload: any = null; let capturedPayload: any = null;
const { streamAnthropic } = await import("../src/providers/anthropic.js"); const { streamAnthropic } = await import("../src/providers/anthropic.js");

4
packages/ai/test/context-overflow.test.ts
View File

@@ -100,8 +100,8 @@ function logResult(result: OverflowResult) {
describe("Context overflow error handling", () => { describe("Context overflow error handling", () => {
describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic (API Key)", () => { describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic (API Key)", () => {
it("claude-haiku-4-5 - should detect overflow via isContextOverflow", async () => { it("claude-3-5-haiku - should detect overflow via isContextOverflow", async () => {
const model = getModel("anthropic", "claude-haiku-4-5"); const model = getModel("anthropic", "claude-3-5-haiku-20241022");
const result = await testContextOverflow(model, process.env.ANTHROPIC_API_KEY!); const result = await testContextOverflow(model, process.env.ANTHROPIC_API_KEY!);
logResult(result); logResult(result);

4
packages/ai/test/empty.test.ts
View File

@@ -229,7 +229,7 @@ describe("AI Providers Empty Message Tests", () => {
}); });
describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider Empty Messages", () => { describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider Empty Messages", () => {
const llm = getModel("anthropic", "claude-haiku-4-5"); const llm = getModel("anthropic", "claude-3-5-haiku-20241022");
it("should handle empty content array", { retry: 3, timeout: 30000 }, async () => { it("should handle empty content array", { retry: 3, timeout: 30000 }, async () => {
await testEmptyMessage(llm); await testEmptyMessage(llm);
@@ -453,7 +453,7 @@ describe("AI Providers Empty Message Tests", () => {
// ========================================================================= // =========================================================================
describe("Anthropic OAuth Provider Empty Messages", () => { describe("Anthropic OAuth Provider Empty Messages", () => {
const llm = getModel("anthropic", "claude-haiku-4-5"); const llm = getModel("anthropic", "claude-3-5-haiku-20241022");
it.skipIf(!anthropicOAuthToken)("should handle empty content array", { retry: 3, timeout: 30000 }, async () => { it.skipIf(!anthropicOAuthToken)("should handle empty content array", { retry: 3, timeout: 30000 }, async () => {
await testEmptyMessage(llm, { apiKey: anthropicOAuthToken }); await testEmptyMessage(llm, { apiKey: anthropicOAuthToken });

37
packages/ai/test/github-copilot-anthropic.test.ts
View File

@@ -4,42 +4,37 @@ import type { Context } from "../src/types.js";
const mockState = vi.hoisted(() => ({ const mockState = vi.hoisted(() => ({
constructorOpts: undefined as Record<string, unknown> | undefined, constructorOpts: undefined as Record<string, unknown> | undefined,
createParams: undefined as Record<string, unknown> | undefined, streamParams: undefined as Record<string, unknown> | undefined,
})); }));
vi.mock("@anthropic-ai/sdk", () => { vi.mock("@anthropic-ai/sdk", () => {
function createSseResponse(): Response { const fakeStream = {
const body = [ async *[Symbol.asyncIterator]() {
`event: message_start\ndata: ${JSON.stringify({ yield {
type: "message_start", type: "message_start",
message: { message: {
id: "msg_test",
usage: { input_tokens: 10, output_tokens: 0 }, usage: { input_tokens: 10, output_tokens: 0 },
}, },
})}\n`, };
`event: message_delta\ndata: ${JSON.stringify({ yield {
type: "message_delta", type: "message_delta",
delta: { stop_reason: "end_turn" }, delta: { stop_reason: "end_turn" },
usage: { output_tokens: 5 }, usage: { output_tokens: 5 },
})}\n`, };
].join("\n"); },
finalMessage: async () => ({
return new Response(body, { usage: { input_tokens: 10, output_tokens: 5, cache_creation_input_tokens: 0, cache_read_input_tokens: 0 },
status: 200, }),
headers: { "content-type": "text/event-stream" }, };
});
}
class FakeAnthropic { class FakeAnthropic {
constructor(opts: Record<string, unknown>) { constructor(opts: Record<string, unknown>) {
mockState.constructorOpts = opts; mockState.constructorOpts = opts;
} }
messages = { messages = {
create: (params: Record<string, unknown>) => { stream: (params: Record<string, unknown>) => {
mockState.createParams = params; mockState.streamParams = params;
return { return fakeStream;
asResponse: async () => createSseResponse(),
};
}, },
}; };
} }
@@ -84,7 +79,7 @@ describe("Copilot Claude via Anthropic Messages", () => {
expect(beta).not.toContain("fine-grained-tool-streaming"); expect(beta).not.toContain("fine-grained-tool-streaming");
// Payload is valid Anthropic Messages format // Payload is valid Anthropic Messages format
const params = mockState.createParams!; const params = mockState.streamParams!;
expect(params.model).toBe("claude-sonnet-4"); expect(params.model).toBe("claude-sonnet-4");
expect(params.stream).toBe(true); expect(params.stream).toBe(true);
expect(params.max_tokens).toBeGreaterThan(0); expect(params.max_tokens).toBeGreaterThan(0);

4
packages/ai/test/stream.test.ts
View File

@@ -475,8 +475,8 @@ describe("Generate E2E Tests", () => {
}); });
}); });
describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider (claude-haiku-4-5)", () => { describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider (claude-3-5-haiku-20241022)", () => {
const model = getModel("anthropic", "claude-haiku-4-5"); const model = getModel("anthropic", "claude-3-5-haiku-20241022");
it("should complete basic text generation", { retry: 3 }, async () => { it("should complete basic text generation", { retry: 3 }, async () => {
await basicTextGeneration(model, { thinkingEnabled: true }); await basicTextGeneration(model, { thinkingEnabled: true });

4
packages/ai/test/tool-call-without-result.test.ts
View File

@@ -137,7 +137,7 @@ describe("Tool Call Without Result Tests", () => {
}); });
describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider", () => { describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider", () => {
const model = getModel("anthropic", "claude-haiku-4-5"); const model = getModel("anthropic", "claude-3-5-haiku-20241022");
it("should filter out tool calls without corresponding tool results", { retry: 3, timeout: 30000 }, async () => { it("should filter out tool calls without corresponding tool results", { retry: 3, timeout: 30000 }, async () => {
await testToolCallWithoutResult(model); await testToolCallWithoutResult(model);
@@ -229,7 +229,7 @@ describe("Tool Call Without Result Tests", () => {
// ========================================================================= // =========================================================================
describe("Anthropic OAuth Provider", () => { describe("Anthropic OAuth Provider", () => {
const model = getModel("anthropic", "claude-haiku-4-5"); const model = getModel("anthropic", "claude-3-5-haiku-20241022");
it.skipIf(!anthropicOAuthToken)( it.skipIf(!anthropicOAuthToken)(
"should filter out tool calls without corresponding tool results", "should filter out tool calls without corresponding tool results",

4
packages/ai/test/total-tokens.test.ts
View File

@@ -106,10 +106,10 @@ describe("totalTokens field", () => {
describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic (API Key)", () => { describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic (API Key)", () => {
it( it(
"claude-sonnet-4-5 - should return totalTokens equal to sum of components", "claude-3-5-haiku - should return totalTokens equal to sum of components",
{ retry: 3, timeout: 60000 }, { retry: 3, timeout: 60000 },
async () => { async () => {
const llm = getModel("anthropic", "claude-sonnet-4-5"); const llm = getModel("anthropic", "claude-3-5-haiku-20241022");
console.log(`\nAnthropic / ${llm.id}:`); console.log(`\nAnthropic / ${llm.id}:`);
const { first, second } = await testTotalTokensWithCache(llm, { apiKey: process.env.ANTHROPIC_API_KEY }); const { first, second } = await testTotalTokensWithCache(llm, { apiKey: process.env.ANTHROPIC_API_KEY });

4
packages/ai/test/unicode-surrogate.test.ts
View File

@@ -352,7 +352,7 @@ describe("AI Providers Unicode Surrogate Pair Tests", () => {
}); });
describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider Unicode Handling", () => { describe.skipIf(!process.env.ANTHROPIC_API_KEY)("Anthropic Provider Unicode Handling", () => {
const llm = getModel("anthropic", "claude-haiku-4-5"); const llm = getModel("anthropic", "claude-3-5-haiku-20241022");
it("should handle emoji in tool results", { retry: 3, timeout: 30000 }, async () => { it("should handle emoji in tool results", { retry: 3, timeout: 30000 }, async () => {
await testEmojiInToolResults(llm); await testEmojiInToolResults(llm);
@@ -372,7 +372,7 @@ describe("AI Providers Unicode Surrogate Pair Tests", () => {
// ========================================================================= // =========================================================================
describe("Anthropic OAuth Provider Unicode Handling", () => { describe("Anthropic OAuth Provider Unicode Handling", () => {
const llm = getModel("anthropic", "claude-haiku-4-5"); const llm = getModel("anthropic", "claude-3-5-haiku-20241022");
it.skipIf(!anthropicOAuthToken)("should handle emoji in tool results", { retry: 3, timeout: 30000 }, async () => { it.skipIf(!anthropicOAuthToken)("should handle emoji in tool results", { retry: 3, timeout: 30000 }, async () => {
await testEmojiInToolResults(llm, { apiKey: anthropicOAuthToken }); await testEmojiInToolResults(llm, { apiKey: anthropicOAuthToken });

79
packages/coding-agent/test/agent-session-model-switch-thinking.test.ts Normal file
View File

@@ -0,0 +1,79 @@
import { Agent, type ThinkingLevel } from "@mariozechner/pi-agent-core";
import { getModel } from "@mariozechner/pi-ai";
import { describe, expect, it } from "vitest";
import { AgentSession } from "../src/core/agent-session.js";
import { AuthStorage } from "../src/core/auth-storage.js";
import { ModelRegistry } from "../src/core/model-registry.js";
import { SessionManager } from "../src/core/session-manager.js";
import { SettingsManager } from "../src/core/settings-manager.js";
import { createTestResourceLoader } from "./utilities.js";
const reasoningModel = getModel("anthropic", "claude-sonnet-4-5")!;
const nonReasoningModel = getModel("anthropic", "claude-3-5-haiku-latest")!;
function createSession({
thinkingLevel = "high",
defaultThinkingLevel = thinkingLevel,
scopedModels,
}: {
thinkingLevel?: ThinkingLevel;
defaultThinkingLevel?: ThinkingLevel;
scopedModels?: Array<{ model: typeof reasoningModel; thinkingLevel?: ThinkingLevel }>;
} = {}) {
const settingsManager = SettingsManager.inMemory({ defaultThinkingLevel });
const sessionManager = SessionManager.inMemory();
const authStorage = AuthStorage.inMemory();
authStorage.setRuntimeApiKey("anthropic", "test-key");
const session = new AgentSession({
agent: new Agent({
getApiKey: () => "test-key",
initialState: {
model: reasoningModel,
systemPrompt: "You are a helpful assistant.",
tools: [],
thinkingLevel,
},
}),
sessionManager,
settingsManager,
cwd: process.cwd(),
modelRegistry: ModelRegistry.inMemory(authStorage),
resourceLoader: createTestResourceLoader(),
scopedModels,
});
return { session, sessionManager, settingsManager };
}
describe("AgentSession model switching", () => {
it("preserves the saved thinking preference through non-reasoning models", async () => {
const { session, sessionManager, settingsManager } = createSession({
scopedModels: [{ model: reasoningModel }, { model: nonReasoningModel }],
});
try {
await session.setModel(nonReasoningModel);
expect(session.thinkingLevel).toBe("off");
expect(settingsManager.getDefaultThinkingLevel()).toBe("high");
await session.setModel(reasoningModel);
expect(session.thinkingLevel).toBe("high");
await session.cycleModel();
expect(session.thinkingLevel).toBe("off");
expect(settingsManager.getDefaultThinkingLevel()).toBe("high");
await session.cycleModel();
expect(session.thinkingLevel).toBe("high");
expect(settingsManager.getDefaultThinkingLevel()).toBe("high");
expect(
sessionManager
.getEntries()
.filter((entry) => entry.type === "thinking_level_change")
.map((entry) => entry.thinkingLevel),
).toEqual(["off", "high", "off", "high"]);
} finally {
session.dispose();
}
});
});