refactor(*): 平台架构 - 拆分后端领域模块并同步插件协议与模型能力

This commit is contained in:
ddCat committed 2026-09-11 14:02:49 +08:00
1 parent 916043fb5a
commit 6e5cbc8b05
644 files changed
+62731 -2320

No files matched your search

+5
View File
@@ -0,0 +1,5 @@
# OpenAI Audio Speech
该目录是独立官方协议插件源码。后端从生成的 `openai-audio.yingce-plugin` 包加载,不依赖系统内置 `host:` 适配器。
完整接口见 [docs/interface.md](docs/interface.md)。
@@ -0,0 +1,233 @@
# OpenAI Audio Speech 接口字段
## 协议身份
- 插件 ID:`openai-audio`。
- Provider ID:`openai-audio`。
- 能力:`audio`。
- 默认 Base URL:`https://api.openai.com`。
- 鉴权驱动:`bearer`。
- 创建:`POST /v1/audio/speech`。
- 生命周期:同步响应。
## 配置字段
| 字段 | 类型 | 必填 | 含义 |
| --- | --- | --- | --- |
| `apiKey` | secret | 是 | API Key |
## 统一字段映射
| 统一字段 | 类型 | 必填 | 上游映射 | 说明 |
| --- | --- | --- | --- | --- |
| `model` | string | 是 | `model` | 音频模型 ID。 |
| `prompt` | string | 是 | `input` | 待合成文本。 |
| `providerOptions` | object | 否 | `provider-specific fields` | 插件命名空间内的厂商扩展字段。 |
## 上游请求模板逐字段清单
下表由插件请求模板生成,覆盖 body、query、headers 和 multipart 文件声明中的每个字段。
| 上游位置 | 值或转换表达式 |
| --- | --- |
| `create.method` | `"POST"` |
| `create.path` | `"/v1/audio/speech"` |
| `create.contentType` | `"application/json"` |
| `create.body.model` | `{"$ref":"request.model"}` |
| `create.body.input` | `{"$ref":"request.prompt"}` |
| `create.body.voice` | `{"$coalesce":[{"$ref":"request.extra.audioVoice"},{"$ref":"request.providerOptions.openai-audio.voice"},"alloy"]}` |
| `create.body.response_format` | `{"$coalesce":[{"$ref":"request.extra.audioFormat"},{"$ref":"request.providerOptions.openai-audio.response_format"},"mp3"]}` |
| `create.body.speed` | `{"$coalesce":[{"$if":{"condition":{"$ne":[{"$toFloat":{"$ref":"request.extra.audioSpeed"}},0]},"then":{"$toFloat":{"$ref":"request.extra.audioSpeed"}},"else":null}},{"$ref":"request.providerOptions.openai-audio.speed"},1]}` |
| `create.body.instructions` | `{"$omitEmpty":{"$coalesce":[{"$ref":"request.extra.audioInstructions"},{"$ref":"request.providerOptions.openai-audio.instructions"}]}}` |
## Provider 扩展键
- `providerOptions.openai-audio.instructions`
- `providerOptions.openai-audio.response_format`
- `providerOptions.openai-audio.speed`
- `providerOptions.openai-audio.voice`
动态模型或工作流允许使用文档声明的完整 `parameters/input/extra_body` 对象;该对象是协议本身的开放 schema,不会被宿主裁剪。
## 响应映射逐字段清单
| 映射位置 | 上游路径或转换表达式 |
| --- | --- |
| `response.binaryPayload` | `true` |
| `response.resultKind` | `"audio"` |
| `response.status` | `"succeeded"` |
| `response.errorPaths[0]` | `"error.code"` |
| `response.messagePaths[0]` | `"error.message"` |
## 响应与错误
插件把上游 task/status/text/media/usage 映射为统一结果。临时媒体 URL 标记为 ephemeral,由宿主立即下载持久化。HTTP 错误、业务 code 和 error object 保持失败语义,不包装成成功。
## 兼容边界
同步 /v1/audio/speech 返回原始音频流,由 binaryPayload 包装为统一音频结果。
<!-- YINGCE_MANIFEST_CONTRACT_START -->
## Manifest 完整接口定义
以下 JSON 与插件包内实际 `manifest.json` 逐字段一致,覆盖插件身份、权限、配置、鉴权、参数、校验、创建、Agent、查询、取消、结果下载、响应和 Agent 响应映射。`documentation` 字段的值就是当前完整文档;为避免文档在自身内部无限递归,JSON 中仅用等义占位文本表示正文。
```json
{
"apiVersion": "yingce.plugin/v2",
"id": "openai-audio",
"name": "OpenAI Audio Speech",
"version": "2.0.0",
"author": "OpenAI / 影策",
"description": "OpenAI Audio Speech 独立请求协议插件。",
"documentation": "<当前插件的完整 documentation,由 README.md 与 docs/interface.md 拼接而成;为避免 JSON 递归,此处不重复展开正文。>",
"permissions": [
"generation.run",
"media.read"
],
"configuration": {
"fields": [
{
"name": "apiKey",
"type": "secret",
"label": "API Key",
"required": true
}
]
},
"contributes": {
"providers": [
{
"id": "openai-audio",
"label": "OpenAI Audio Speech",
"capabilities": [
"audio"
],
"scopes": [
"admin.system-channel",
"user.custom-channel",
"canvas",
"creation",
"agent"
],
"baseUrl": "https://api.openai.com",
"requiresPublicMediaUrls": false,
"auth": {
"type": "bearer",
"field": "apiKey"
},
"parameters": [
{
"name": "model",
"type": "string",
"required": true,
"mapping": "model",
"description": "音频模型 ID。"
},
{
"name": "prompt",
"type": "string",
"required": true,
"mapping": "input",
"description": "待合成文本。"
},
{
"name": "providerOptions",
"type": "object",
"required": false,
"mapping": "provider-specific fields",
"description": "插件命名空间内的厂商扩展字段。"
}
],
"create": {
"method": "POST",
"path": "/v1/audio/speech",
"contentType": "application/json",
"body": {
"model": {
"$ref": "request.model"
},
"input": {
"$ref": "request.prompt"
},
"voice": {
"$coalesce": [
{
"$ref": "request.extra.audioVoice"
},
{
"$ref": "request.providerOptions.openai-audio.voice"
},
"alloy"
]
},
"response_format": {
"$coalesce": [
{
"$ref": "request.extra.audioFormat"
},
{
"$ref": "request.providerOptions.openai-audio.response_format"
},
"mp3"
]
},
"speed": {
"$coalesce": [
{
"$if": {
"condition": {
"$ne": [
{
"$toFloat": {
"$ref": "request.extra.audioSpeed"
}
},
0
]
},
"then": {
"$toFloat": {
"$ref": "request.extra.audioSpeed"
}
},
"else": null
}
},
{
"$ref": "request.providerOptions.openai-audio.speed"
},
1
]
},
"instructions": {
"$omitEmpty": {
"$coalesce": [
{
"$ref": "request.extra.audioInstructions"
},
{
"$ref": "request.providerOptions.openai-audio.instructions"
}
]
}
}
}
},
"response": {
"binaryPayload": true,
"resultKind": "audio",
"status": "succeeded",
"errorPaths": [
"error.code"
],
"messagePaths": [
"error.message"
]
}
}
]
}
}
```
<!-- YINGCE_MANIFEST_CONTRACT_END -->
+156
View File
@@ -0,0 +1,156 @@
{
"apiVersion": "yingce.plugin/v2",
"id": "openai-audio",
"name": "OpenAI Audio Speech",
"version": "2.0.0",
"author": "OpenAI / 影策",
"description": "OpenAI Audio Speech 独立请求协议插件。",
"documentation": "# OpenAI Audio Speech\n\n该目录是独立官方协议插件源码。后端从生成的 `openai-audio.yingce-plugin` 包加载,不依赖系统内置 `host:` 适配器。\n\n完整接口见 [docs/interface.md](docs/interface.md)。\n\n---\n\n# OpenAI Audio Speech 接口字段\n\n## 协议身份\n\n- 插件 ID:`openai-audio`。\n- Provider ID:`openai-audio`。\n- 能力:`audio`。\n- 默认 Base URL:`https://api.openai.com`。\n- 鉴权驱动:`bearer`。\n- 创建:`POST /v1/audio/speech`。\n- 生命周期:同步响应。\n\n## 配置字段\n\n| 字段 | 类型 | 必填 | 含义 |\n| --- | --- | --- | --- |\n| `apiKey` | secret | 是 | API Key |\n\n## 统一字段映射\n\n| 统一字段 | 类型 | 必填 | 上游映射 | 说明 |\n| --- | --- | --- | --- | --- |\n| `model` | string | 是 | `model` | 音频模型 ID。 |\n| `prompt` | string | 是 | `input` | 待合成文本。 |\n| `providerOptions` | object | 否 | `provider-specific fields` | 插件命名空间内的厂商扩展字段。 |\n\n## 上游请求模板逐字段清单\n\n下表由插件请求模板生成,覆盖 body、query、headers 和 multipart 文件声明中的每个字段。\n\n| 上游位置 | 值或转换表达式 |\n| --- | --- |\n| `create.method` | `\"POST\"` |\n| `create.path` | `\"/v1/audio/speech\"` |\n| `create.contentType` | `\"application/json\"` |\n| `create.body.model` | `{\"$ref\":\"request.model\"}` |\n| `create.body.input` | `{\"$ref\":\"request.prompt\"}` |\n| `create.body.voice` | `{\"$coalesce\":[{\"$ref\":\"request.extra.audioVoice\"},{\"$ref\":\"request.providerOptions.openai-audio.voice\"},\"alloy\"]}` |\n| `create.body.response_format` | `{\"$coalesce\":[{\"$ref\":\"request.extra.audioFormat\"},{\"$ref\":\"request.providerOptions.openai-audio.response_format\"},\"mp3\"]}` |\n| `create.body.speed` | `{\"$coalesce\":[{\"$if\":{\"condition\":{\"$ne\":[{\"$toFloat\":{\"$ref\":\"request.extra.audioSpeed\"}},0]},\"then\":{\"$toFloat\":{\"$ref\":\"request.extra.audioSpeed\"}},\"else\":null}},{\"$ref\":\"request.providerOptions.openai-audio.speed\"},1]}` |\n| `create.body.instructions` | `{\"$omitEmpty\":{\"$coalesce\":[{\"$ref\":\"request.extra.audioInstructions\"},{\"$ref\":\"request.providerOptions.openai-audio.instructions\"}]}}` |\n\n## Provider 扩展键\n\n- `providerOptions.openai-audio.instructions`\n- `providerOptions.openai-audio.response_format`\n- `providerOptions.openai-audio.speed`\n- `providerOptions.openai-audio.voice`\n\n动态模型或工作流允许使用文档声明的完整 `parameters/input/extra_body` 对象;该对象是协议本身的开放 schema,不会被宿主裁剪。\n\n## 响应映射逐字段清单\n\n| 映射位置 | 上游路径或转换表达式 |\n| --- | --- |\n| `response.binaryPayload` | `true` |\n| `response.resultKind` | `\"audio\"` |\n| `response.status` | `\"succeeded\"` |\n| `response.errorPaths[0]` | `\"error.code\"` |\n| `response.messagePaths[0]` | `\"error.message\"` |\n\n## 响应与错误\n\n插件把上游 task/status/text/media/usage 映射为统一结果。临时媒体 URL 标记为 ephemeral,由宿主立即下载持久化。HTTP 错误、业务 code 和 error object 保持失败语义,不包装成成功。\n\n## 兼容边界\n\n同步 /v1/audio/speech 返回原始音频流,由 binaryPayload 包装为统一音频结果。\n\n<!-- YINGCE_MANIFEST_CONTRACT_START -->\n## Manifest 完整接口定义\n\n以下 JSON 与插件包内实际 `manifest.json` 逐字段一致,覆盖插件身份、权限、配置、鉴权、参数、校验、创建、Agent、查询、取消、结果下载、响应和 Agent 响应映射。`documentation` 字段的值就是当前完整文档;为避免文档在自身内部无限递归,JSON 中仅用等义占位文本表示正文。\n\n```json\n{\n \"apiVersion\": \"yingce.plugin/v2\",\n \"id\": \"openai-audio\",\n \"name\": \"OpenAI Audio Speech\",\n \"version\": \"2.0.0\",\n \"author\": \"OpenAI / 影策\",\n \"description\": \"OpenAI Audio Speech 独立请求协议插件。\",\n \"documentation\": \"<当前插件的完整 documentation,由 README.md 与 docs/interface.md 拼接而成;为避免 JSON 递归,此处不重复展开正文。>\",\n \"permissions\": [\n \"generation.run\",\n \"media.read\"\n ],\n \"configuration\": {\n \"fields\": [\n {\n \"name\": \"apiKey\",\n \"type\": \"secret\",\n \"label\": \"API Key\",\n \"required\": true\n }\n ]\n },\n \"contributes\": {\n \"providers\": [\n {\n \"id\": \"openai-audio\",\n \"label\": \"OpenAI Audio Speech\",\n \"capabilities\": [\n \"audio\"\n ],\n \"scopes\": [\n \"admin.system-channel\",\n \"user.custom-channel\",\n \"canvas\",\n \"creation\",\n \"agent\"\n ],\n \"baseUrl\": \"https://api.openai.com\",\n \"requiresPublicMediaUrls\": false,\n \"auth\": {\n Line truncated
"permissions": [
"generation.run",
"media.read"
],
"configuration": {
"fields": [
{
"name": "apiKey",
"type": "secret",
"label": "API Key",
"required": true
}
]
},
"contributes": {
"providers": [
{
"id": "openai-audio",
"label": "OpenAI Audio Speech",
"capabilities": [
"audio"
],
"scopes": [
"admin.system-channel",
"user.custom-channel",
"canvas",
"creation",
"agent"
],
"baseUrl": "https://api.openai.com",
"requiresPublicMediaUrls": false,
"auth": {
"type": "bearer",
"field": "apiKey"
},
"parameters": [
{
"name": "model",
"type": "string",
"required": true,
"mapping": "model",
"description": "音频模型 ID。"
},
{
"name": "prompt",
"type": "string",
"required": true,
"mapping": "input",
"description": "待合成文本。"
},
{
"name": "providerOptions",
"type": "object",
"required": false,
"mapping": "provider-specific fields",
"description": "插件命名空间内的厂商扩展字段。"
}
],
"create": {
"method": "POST",
"path": "/v1/audio/speech",
"contentType": "application/json",
"body": {
"model": {
"$ref": "request.model"
},
"input": {
"$ref": "request.prompt"
},
"voice": {
"$coalesce": [
{
"$ref": "request.extra.audioVoice"
},
{
"$ref": "request.providerOptions.openai-audio.voice"
},
"alloy"
]
},
"response_format": {
"$coalesce": [
{
"$ref": "request.extra.audioFormat"
},
{
"$ref": "request.providerOptions.openai-audio.response_format"
},
"mp3"
]
},
"speed": {
"$coalesce": [
{
"$if": {
"condition": {
"$ne": [
{
"$toFloat": {
"$ref": "request.extra.audioSpeed"
}
},
0
]
},
"then": {
"$toFloat": {
"$ref": "request.extra.audioSpeed"
}
},
"else": null
}
},
{
"$ref": "request.providerOptions.openai-audio.speed"
},
1
]
},
"instructions": {
"$omitEmpty": {
"$coalesce": [
{
"$ref": "request.extra.audioInstructions"
},
{
"$ref": "request.providerOptions.openai-audio.instructions"
}
]
}
}
}
},
"response": {
"binaryPayload": true,
"resultKind": "audio",
"status": "succeeded",
"errorPaths": [
"error.code"
],
"messagePaths": [
"error.message"
]
}
}
]
}
}