mirror of
https://github.com/getcompanion-ai/co-mono.git
synced 2026-04-20 09:01:49 +00:00
feat(ai): add partial JSON parsing for streaming tool calls
- Added partial-json package for parsing incomplete JSON during streaming
- Tool call arguments now contain partially parsed JSON during toolcall_delta events
- Enables progressive UI updates (e.g., showing file paths before content is complete)
- Arguments are always valid objects (minimum empty {}), never undefined
- Full validation still occurs at toolcall_end when arguments are complete
- Updated all providers (Anthropic, OpenAI Completions/Responses) to use parseStreamingJson
- Added comprehensive documentation and examples in README
- Added test to verify arguments are always defined during streaming
This commit is contained in:
Mario Zechner
parent
197259c88a
commit
39c626b6c9
10 changed files with 208 additions and 69 deletions
28
packages/ai/src/json-parse.ts
Normal file
28
packages/ai/src/json-parse.ts
Normal file
|
|
@ -0,0 +1,28 @@
|
|||
import { parse as partialParse } from "partial-json";
|
||||
|
||||
/**
|
||||
* Attempts to parse potentially incomplete JSON during streaming.
|
||||
* Always returns a valid object, even if the JSON is incomplete.
|
||||
*
|
||||
* @param partialJson The partial JSON string from streaming
|
||||
* @returns Parsed object or empty object if parsing fails
|
||||
*/
|
||||
export function parseStreamingJson<T = any>(partialJson: string | undefined): T {
|
||||
if (!partialJson || partialJson.trim() === "") {
|
||||
return {} as T;
|
||||
}
|
||||
|
||||
// Try standard parsing first (fastest for complete JSON)
|
||||
try {
|
||||
return JSON.parse(partialJson) as T;
|
||||
} catch {
|
||||
// Try partial-json for incomplete JSON
|
||||
try {
|
||||
const result = partialParse(partialJson);
|
||||
return (result ?? {}) as T;
|
||||
} catch {
|
||||
// If all parsing fails, return empty object
|
||||
return {} as T;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -2714,13 +2714,13 @@ export const MODELS = {
|
|||
reasoning: false,
|
||||
input: ["text"],
|
||||
cost: {
|
||||
input: 0.038000000000000006,
|
||||
output: 0.12,
|
||||
input: 0.012,
|
||||
output: 0.036,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
},
|
||||
contextWindow: 131072,
|
||||
maxTokens: 16384,
|
||||
maxTokens: 8192,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"amazon/nova-lite-v1": {
|
||||
id: "amazon/nova-lite-v1",
|
||||
|
|
@ -2943,23 +2943,6 @@ export const MODELS = {
|
|||
contextWindow: 32768,
|
||||
maxTokens: 4096,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"cohere/command-r-plus-08-2024": {
|
||||
id: "cohere/command-r-plus-08-2024",
|
||||
name: "Cohere: Command R+ (08-2024)",
|
||||
api: "openai-completions",
|
||||
provider: "openrouter",
|
||||
baseUrl: "https://openrouter.ai/api/v1",
|
||||
reasoning: false,
|
||||
input: ["text"],
|
||||
cost: {
|
||||
input: 2.5,
|
||||
output: 10,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
},
|
||||
contextWindow: 128000,
|
||||
maxTokens: 4000,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"cohere/command-r-08-2024": {
|
||||
id: "cohere/command-r-08-2024",
|
||||
name: "Cohere: Command R (08-2024)",
|
||||
|
|
@ -2977,6 +2960,23 @@ export const MODELS = {
|
|||
contextWindow: 128000,
|
||||
maxTokens: 4000,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"cohere/command-r-plus-08-2024": {
|
||||
id: "cohere/command-r-plus-08-2024",
|
||||
name: "Cohere: Command R+ (08-2024)",
|
||||
api: "openai-completions",
|
||||
provider: "openrouter",
|
||||
baseUrl: "https://openrouter.ai/api/v1",
|
||||
reasoning: false,
|
||||
input: ["text"],
|
||||
cost: {
|
||||
input: 2.5,
|
||||
output: 10,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
},
|
||||
contextWindow: 128000,
|
||||
maxTokens: 4000,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"microsoft/phi-3.5-mini-128k-instruct": {
|
||||
id: "microsoft/phi-3.5-mini-128k-instruct",
|
||||
name: "Microsoft: Phi-3.5 Mini 128K Instruct",
|
||||
|
|
@ -3079,23 +3079,6 @@ export const MODELS = {
|
|||
contextWindow: 131072,
|
||||
maxTokens: 128000,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"mistralai/mistral-7b-instruct-v0.3": {
|
||||
id: "mistralai/mistral-7b-instruct-v0.3",
|
||||
name: "Mistral: Mistral 7B Instruct v0.3",
|
||||
api: "openai-completions",
|
||||
provider: "openrouter",
|
||||
baseUrl: "https://openrouter.ai/api/v1",
|
||||
reasoning: false,
|
||||
input: ["text"],
|
||||
cost: {
|
||||
input: 0.028,
|
||||
output: 0.054,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
},
|
||||
contextWindow: 32768,
|
||||
maxTokens: 16384,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"mistralai/mistral-7b-instruct:free": {
|
||||
id: "mistralai/mistral-7b-instruct:free",
|
||||
name: "Mistral: Mistral 7B Instruct (free)",
|
||||
|
|
@ -3130,6 +3113,23 @@ export const MODELS = {
|
|||
contextWindow: 32768,
|
||||
maxTokens: 16384,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"mistralai/mistral-7b-instruct-v0.3": {
|
||||
id: "mistralai/mistral-7b-instruct-v0.3",
|
||||
name: "Mistral: Mistral 7B Instruct v0.3",
|
||||
api: "openai-completions",
|
||||
provider: "openrouter",
|
||||
baseUrl: "https://openrouter.ai/api/v1",
|
||||
reasoning: false,
|
||||
input: ["text"],
|
||||
cost: {
|
||||
input: 0.028,
|
||||
output: 0.054,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
},
|
||||
contextWindow: 32768,
|
||||
maxTokens: 16384,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"microsoft/phi-3-mini-128k-instruct": {
|
||||
id: "microsoft/phi-3-mini-128k-instruct",
|
||||
name: "Microsoft: Phi-3 Mini 128K Instruct",
|
||||
|
|
@ -3300,23 +3300,6 @@ export const MODELS = {
|
|||
contextWindow: 128000,
|
||||
maxTokens: 4096,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"mistralai/mistral-tiny": {
|
||||
id: "mistralai/mistral-tiny",
|
||||
name: "Mistral Tiny",
|
||||
api: "openai-completions",
|
||||
provider: "openrouter",
|
||||
baseUrl: "https://openrouter.ai/api/v1",
|
||||
reasoning: false,
|
||||
input: ["text"],
|
||||
cost: {
|
||||
input: 0.25,
|
||||
output: 0.25,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
},
|
||||
contextWindow: 32768,
|
||||
maxTokens: 4096,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"mistralai/mistral-small": {
|
||||
id: "mistralai/mistral-small",
|
||||
name: "Mistral Small",
|
||||
|
|
@ -3334,6 +3317,23 @@ export const MODELS = {
|
|||
contextWindow: 32768,
|
||||
maxTokens: 4096,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"mistralai/mistral-tiny": {
|
||||
id: "mistralai/mistral-tiny",
|
||||
name: "Mistral Tiny",
|
||||
api: "openai-completions",
|
||||
provider: "openrouter",
|
||||
baseUrl: "https://openrouter.ai/api/v1",
|
||||
reasoning: false,
|
||||
input: ["text"],
|
||||
cost: {
|
||||
input: 0.25,
|
||||
output: 0.25,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
},
|
||||
contextWindow: 32768,
|
||||
maxTokens: 4096,
|
||||
} satisfies Model<"openai-completions">,
|
||||
"mistralai/mixtral-8x7b-instruct": {
|
||||
id: "mistralai/mixtral-8x7b-instruct",
|
||||
name: "Mistral: Mixtral 8x7B Instruct",
|
||||
|
|
|
|||
|
|
@ -5,6 +5,7 @@ import type {
|
|||
MessageParam,
|
||||
} from "@anthropic-ai/sdk/resources/messages.js";
|
||||
import { AssistantMessageEventStream } from "../event-stream.js";
|
||||
import { parseStreamingJson } from "../json-parse.js";
|
||||
import { calculateCost } from "../models.js";
|
||||
import type {
|
||||
Api,
|
||||
|
|
@ -124,6 +125,7 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
|
|||
const block = blocks[index];
|
||||
if (block && block.type === "toolCall") {
|
||||
block.partialJson += event.delta.partial_json;
|
||||
block.arguments = parseStreamingJson(block.partialJson);
|
||||
stream.push({
|
||||
type: "toolcall_delta",
|
||||
contentIndex: index,
|
||||
|
|
|
|||
|
|
@ -8,6 +8,7 @@ import type {
|
|||
ChatCompletionMessageParam,
|
||||
} from "openai/resources/chat/completions.js";
|
||||
import { AssistantMessageEventStream } from "../event-stream.js";
|
||||
import { parseStreamingJson } from "../json-parse.js";
|
||||
import { calculateCost } from "../models.js";
|
||||
import type {
|
||||
AssistantMessage,
|
||||
|
|
@ -210,6 +211,7 @@ export const streamOpenAICompletions: StreamFunction<"openai-completions"> = (
|
|||
if (toolCall.function?.arguments) {
|
||||
delta = toolCall.function.arguments;
|
||||
currentBlock.partialArgs += toolCall.function.arguments;
|
||||
currentBlock.arguments = parseStreamingJson(currentBlock.partialArgs);
|
||||
}
|
||||
stream.push({
|
||||
type: "toolcall_delta",
|
||||
|
|
|
|||
|
|
@ -11,6 +11,7 @@ import type {
|
|||
ResponseReasoningItem,
|
||||
} from "openai/resources/responses/responses.js";
|
||||
import { AssistantMessageEventStream } from "../event-stream.js";
|
||||
import { parseStreamingJson } from "../json-parse.js";
|
||||
import { calculateCost } from "../models.js";
|
||||
import type {
|
||||
Api,
|
||||
|
|
@ -194,12 +195,7 @@ export const streamOpenAIResponses: StreamFunction<"openai-responses"> = (
|
|||
currentBlock.type === "toolCall"
|
||||
) {
|
||||
currentBlock.partialJson += event.delta;
|
||||
try {
|
||||
const args = JSON.parse(currentBlock.partialJson);
|
||||
currentBlock.arguments = args;
|
||||
} catch {
|
||||
// Ignore JSON parse errors - the JSON might be incomplete
|
||||
}
|
||||
currentBlock.arguments = parseStreamingJson(currentBlock.partialJson);
|
||||
stream.push({
|
||||
type: "toolcall_delta",
|
||||
contentIndex: blockIndex(),
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue