Files
claude-code/packages/@ant/model-provider/src/shared/openaiConvertMessages.ts
ymonster 1b10ea391a fix: preserve empty reasoning_content for DeepSeek v4 thinking mode (#399)
DeepSeek v4 in thinking mode sometimes returns reasoning_content: ""
when the model answers directly without internal reasoning. Two places
were filtering the empty string out, which dropped the thinking block
from the assistant turn entirely. The next request then omitted
reasoning_content for that prior turn, and DeepSeek rejected with
400 "reasoning_content ... must be passed back to the API".

Fix:
- openaiStreamAdapter: open a thinking block whenever reasoning_content
  is present (including ""); skip the empty thinking_delta event since
  the empty value is already conveyed by the block's initial state.
- openaiConvertMessages: preserve empty thinking blocks as
  reasoning_content: "" when serializing assistant messages back to
  the OpenAI/DeepSeek format.

Tests:
- New: empty reasoning_content opens a thinking block (adapter).
- Updated: empty thinking blocks now round-trip as reasoning_content: ""
  instead of being dropped.
- New: assistant messages with no thinking block still omit
  reasoning_content (regression guard for non-thinking models).
2026-05-02 14:58:29 +08:00

269 lines
7.9 KiB
TypeScript

import type {
BetaContentBlockParam,
BetaToolResultBlockParam,
BetaToolUseBlock,
} from '@anthropic-ai/sdk/resources/beta/messages/messages.mjs'
import type {
ChatCompletionAssistantMessageParam,
ChatCompletionMessageParam,
ChatCompletionSystemMessageParam,
ChatCompletionToolMessageParam,
ChatCompletionUserMessageParam,
} from 'openai/resources/chat/completions/completions.mjs'
import type { AssistantMessage, UserMessage } from '../types/message.js'
import type { SystemPrompt } from '../types/systemPrompt.js'
export interface ConvertMessagesOptions {
/** When true, preserve thinking blocks as reasoning_content on assistant messages
* (required for DeepSeek thinking mode with tool calls). */
enableThinking?: boolean
}
/**
* Convert internal (UserMessage | AssistantMessage)[] to OpenAI-format messages.
*
* Key conversions:
* - system prompt → role: "system" message prepended
* - tool_use blocks → tool_calls[] on assistant message
* - tool_result blocks → role: "tool" messages
* - thinking blocks → preserved as reasoning_content (DeepSeek requires passing it back)
* - cache_control → stripped
*/
export function anthropicMessagesToOpenAI(
messages: (UserMessage | AssistantMessage)[],
systemPrompt: SystemPrompt,
// options retained for API compatibility; thinking blocks are now always preserved
_options?: ConvertMessagesOptions,
): ChatCompletionMessageParam[] {
const result: ChatCompletionMessageParam[] = []
// Prepend system prompt as system message
const systemText = systemPromptToText(systemPrompt)
if (systemText) {
result.push({
role: 'system',
content: systemText,
} satisfies ChatCompletionSystemMessageParam)
}
for (const msg of messages) {
switch (msg.type) {
case 'user':
result.push(...convertInternalUserMessage(msg))
break
case 'assistant':
result.push(...convertInternalAssistantMessage(msg))
break
default:
break
}
}
return result
}
function systemPromptToText(systemPrompt: SystemPrompt): string {
if (!systemPrompt || systemPrompt.length === 0) return ''
return systemPrompt.filter(Boolean).join('\n\n')
}
function convertInternalUserMessage(
msg: UserMessage,
): ChatCompletionMessageParam[] {
const result: ChatCompletionMessageParam[] = []
const content = msg.message.content
if (typeof content === 'string') {
result.push({
role: 'user',
content,
} satisfies ChatCompletionUserMessageParam)
} else if (Array.isArray(content)) {
const textParts: string[] = []
const toolResults: BetaToolResultBlockParam[] = []
const imageParts: Array<{ type: 'image_url'; image_url: { url: string } }> =
[]
for (const block of content) {
if (typeof block === 'string') {
textParts.push(block)
} else if (block.type === 'text') {
textParts.push(block.text)
} else if (block.type === 'tool_result') {
toolResults.push(block as BetaToolResultBlockParam)
} else if (block.type === 'image') {
const imagePart = convertImageBlockToOpenAI(
block as unknown as Record<string, unknown>,
)
if (imagePart) {
imageParts.push(imagePart)
}
}
}
// CRITICAL: tool messages must come BEFORE any user message in the result.
// OpenAI API requires that a tool message immediately follows the assistant
// message with tool_calls. If we emit a user message first, the API will
// reject the request with "insufficient tool messages following tool_calls".
for (const tr of toolResults) {
result.push(convertToolResult(tr))
}
// 如果有图片,构建多模态 content 数组
if (imageParts.length > 0) {
const multiContent: Array<
| { type: 'text'; text: string }
| { type: 'image_url'; image_url: { url: string } }
> = []
if (textParts.length > 0) {
multiContent.push({ type: 'text', text: textParts.join('\n') })
}
multiContent.push(...imageParts)
result.push({
role: 'user',
content: multiContent,
} satisfies ChatCompletionUserMessageParam)
} else if (textParts.length > 0) {
result.push({
role: 'user',
content: textParts.join('\n'),
} satisfies ChatCompletionUserMessageParam)
}
}
return result
}
function convertToolResult(
block: BetaToolResultBlockParam,
): ChatCompletionToolMessageParam {
let content: string
if (typeof block.content === 'string') {
content = block.content
} else if (Array.isArray(block.content)) {
content = block.content
.map(c => {
if (typeof c === 'string') return c
if ('text' in c) return c.text
return ''
})
.filter(Boolean)
.join('\n')
} else {
content = ''
}
return {
role: 'tool',
tool_call_id: block.tool_use_id,
content,
} satisfies ChatCompletionToolMessageParam
}
function convertInternalAssistantMessage(
msg: AssistantMessage,
): ChatCompletionMessageParam[] {
const content = msg.message.content
if (typeof content === 'string') {
return [
{
role: 'assistant',
content,
} satisfies ChatCompletionAssistantMessageParam,
]
}
if (!Array.isArray(content)) {
return [
{
role: 'assistant',
content: '',
} satisfies ChatCompletionAssistantMessageParam,
]
}
const textParts: string[] = []
const toolCalls: NonNullable<
ChatCompletionAssistantMessageParam['tool_calls']
> = []
const reasoningParts: string[] = []
for (const block of content) {
if (typeof block === 'string') {
textParts.push(block)
} else if (block.type === 'text') {
textParts.push(block.text)
} else if (block.type === 'tool_use') {
const tu = block as BetaToolUseBlock
toolCalls.push({
id: tu.id,
type: 'function',
function: {
name: tu.name,
arguments:
typeof tu.input === 'string' ? tu.input : JSON.stringify(tu.input),
},
})
} else if (block.type === 'thinking') {
// DeepSeek thinking mode: always preserve reasoning_content,
// including the empty-string case. DeepSeek v4 may return
// reasoning_content: "" when the model answers directly, and the
// empty value must be echoed back in the next request — otherwise
// DeepSeek returns 400 ("reasoning_content ... must be passed back").
const thinkingText = (block as unknown as Record<string, unknown>)
.thinking
if (typeof thinkingText === 'string') {
reasoningParts.push(thinkingText)
}
}
// Skip redacted_thinking, server_tool_use, etc.
}
const result: ChatCompletionAssistantMessageParam = {
role: 'assistant',
content: textParts.length > 0 ? textParts.join('\n') : null,
...(toolCalls.length > 0 && { tool_calls: toolCalls }),
...(reasoningParts.length > 0 && {
reasoning_content: reasoningParts.join('\n'),
}),
}
return [result]
}
/**
* 将 Anthropic image 块转换为 OpenAI image_url 格式。
*
* Anthropic 格式: { type: "image", source: { type: "base64", media_type: "image/png", data: "..." } }
* OpenAI 格式: { type: "image_url", image_url: { url: "data:image/png;base64,..." } }
*/
function convertImageBlockToOpenAI(
block: Record<string, unknown>,
): { type: 'image_url'; image_url: { url: string } } | null {
const source = block.source as Record<string, unknown> | undefined
if (!source) return null
if (source.type === 'base64' && typeof source.data === 'string') {
const mediaType = (source.media_type as string) || 'image/png'
return {
type: 'image_url',
image_url: {
url: `data:${mediaType};base64,${source.data}`,
},
}
}
// url 类型的图片直接传递
if (source.type === 'url' && typeof source.url === 'string') {
return {
type: 'image_url',
image_url: {
url: source.url,
},
}
}
return null
}