@lingjingai/scriptctl 0.64.3 → 0.65.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +3 -1
- package/changes/0.65.0.md +15 -0
- package/changes/0.65.1.md +13 -0
- package/dist/bin.js +1 -1
- package/dist/{chunk-YYKX6SLG.js → chunk-5KDKNZCX.js} +193 -142
- package/dist/cli.d.ts +2 -0
- package/dist/cli.js +1 -1
- package/package.json +17 -18
- package/skills/scriptctl/SKILL.md +1 -1
- package/skills/scriptctl/references/ingest-workflow.md +2 -1
package/README.md
CHANGED
|
@@ -138,11 +138,13 @@ tasks:
|
|
|
138
138
|
direct.episode-titles: default
|
|
139
139
|
```
|
|
140
140
|
|
|
141
|
+
视频拉片默认让模型按逐行记录的 DSL 输出,再解析成与 JSON 相同的结构,输出 token 约为 JSON 的一半;`scriptctl ingest --video-output json` 改用 Schema 约束的 JSON。以下 Schema 字段配置只在 `json` 格式下生效。
|
|
142
|
+
|
|
141
143
|
Gemini provider 可设置 `videoResponseSchemaField: responseSchema`,用于 GRSAI 等仅接受该字段的视频接口;默认仍为 `responseJsonSchema`。
|
|
142
144
|
`responseSchema` 由 Google SDK 转换字段定义,保留类型、必填字段和描述,但不发送 `additionalProperties` 约束。
|
|
143
145
|
|
|
144
146
|
内置 `ingest.video-serial` 与 `ingest.comic-serial` 使用 NNYYAI 的 `gemini-3.8-flash`(`https://api.nnyyai.com`),
|
|
145
|
-
从 `NNYYAI_API_KEY` 读取密钥,使用
|
|
147
|
+
从 `NNYYAI_API_KEY` 读取密钥,使用 base64 视频或 PDF 输入(Schema 字段为 `responseSchema`)。设置密钥后直接运行
|
|
146
148
|
`scriptctl ingest -s <视频或PDF文件或目录> -w <workspace>`。Gemini 型号切换始终使用同一 NNYYAI 渠道;配置了 `ARK_API_KEY` 时保留豆包最终兜底。文本任务与 `review-agent` 保持原有路由。
|
|
147
149
|
显式 LLM 配置仍完全按所选文件执行;GrsAI、Google Vertex 等 provider 可继续使用。
|
|
148
150
|
|
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
# 0.65.0
|
|
2
|
+
|
|
3
|
+
## Added
|
|
4
|
+
- `scriptctl ingest --video-output dsl|json` selects the video model output format. `json` uses schema-constrained JSON output.
|
|
5
|
+
|
|
6
|
+
## Changed
|
|
7
|
+
- Video ingest defaults to a line-based output format. It uses about half the output tokens of JSON and produces the same parsed structure. Generation is about 20% faster.
|
|
8
|
+
- Video ingest saves each episode's raw model output as `ep/ep_NNN.ep` in the default format, and as `ep/ep_NNN.md` with `--video-output json`. `scriptctl view` lists both under 转录.
|
|
9
|
+
|
|
10
|
+
## Fixed
|
|
11
|
+
- Video ingest retries the current model, then the next route, when a provider returns text that is not in the requested output format, such as a Gemini proxy filter notice. Such responses no longer reject the episode as having no scenes.
|
|
12
|
+
- Video and PDF ingest retry a Gemini stream that ends before the model finishes, instead of accepting the truncated response.
|
|
13
|
+
|
|
14
|
+
## Breaking
|
|
15
|
+
- None.
|
package/dist/bin.js
CHANGED
|
@@ -1,2 +1,2 @@
|
|
|
1
1
|
#!/usr/bin/env node
|
|
2
|
-
import{b as s}from"./chunk-
|
|
2
|
+
import{b as s}from"./chunk-5KDKNZCX.js";import{Console as r}from"node:console";var o=new r({stdout:process.stderr,stderr:process.stderr});console.debug=o.debug.bind(o);console.info=o.info.bind(o);console.log=o.log.bind(o);s(process.argv.slice(2)).then(e=>{process.exitCode=e}).catch(e=>{console.error(e),process.exitCode=70});
|