dsh-pf-stream 0.1.0 → 0.1.3

package/README.en.md

@@ -0,0 +1,110 @@
+# dsh-pf-stream
+
+[中文](README.md) | **English**
+
+Video perception plugin for DeepSeek Harness (DSH). One tool lets the model "see" local video files:
+
+- **Continuous events** → adaptive streaming sampling. During motion, the sampler emits low-token process units by displacement integration. On settle or sharpness peaks, it freezes a high-resolution focus unit.
+- **Suspicious time points** → deterministic inspection: metadata probe, frame extraction at given timestamps, uniform slicing, and local RoI zoom-in.
+
+Every canvas aligns with the DeepSeek-V4.1-Flash vision pipeline: 42-px superpixel grid, billing `r×(c+1)+2`. All four forms bill ≤530 tokens, within the 1024-token per-image limit. The server encodes each canvas at its original resolution.
+
+## Capability routing
+
+One tool, `dsh-pf-stream`, routes through `intent.action`:
+
+| action | engine | use |
+| --- | --- | --- |
+| `pf_stream` (default) | adaptive streaming | 336² process units (P) during motion; 672² focus units (F) on settle/peak. Schmitt four-state machine + two-layer adaptation (intent baseline + content feedback R1-R4) |
+| `timestamps` | deterministic inspection | high-resolution frames at given timestamps; `roi_hint` zoom-in |
+| `uniform` | deterministic inspection | uniform slicing by step or target frame count |
+| `probe` | deterministic inspection | metadata only (display size, rotation, last-frame PTS, stream summary); zero vision tokens |
+
+## Install
+
+Requirements:
+
+- DSH host (Node ≥ 20).
+- `ffmpeg` and `ffprobe` executables. Search order: plugin config path → `DSH_FFMPEG_PATH` / `FFMPEG_PATH` → `DSH_FFMPEG_DIR` → `~/.dsh/tools/ffmpeg/bin` → `~/.dsh/bin` → `PATH`.
+
+Install the package `dsh-pf-stream` in the DSH plugin manager. The host mounts the plugin through `dsh.bundle.patch` (`cordis.patch.yml`). Peer dependencies `@deepseek-ai/cordis` and `@deepseek-ai/schemastery` resolve in the host runtime. The only runtime dependency is `pngjs`; canvas PNG encoding uses fixed parameters, so output bytes are deterministic.
+
+## Usage
+
+Streaming perception ("what text is on screen"):
+
+```json
+{
+  "video_path": "D:/videos/demo.mp4",
+  "intent": { "action": "pf_stream", "focus": "detail", "mode": "dense" }
+}
+```
+
+Point inspection (text found at 12.9 s in a streaming result):
+
+```json
+{
+  "video_path": "D:/videos/demo.mp4",
+  "intent": {
+    "action": "timestamps",
+    "timestamps_ms": [12900],
+    "roi_hint": { "x": 0.2, "y": 0.3, "w": 0.4, "h": 0.4 }
+  }
+}
+```
+
+The top level also accepts common convenience fields from the video-inspection domain: `start_time`, `end_time`, `timestamps`, `fps`, `interval`, `num_frames`, `crop`, `probe_only`. The entry adapter normalizes them into the standard `intent` structure.
+
+## Returned content
+
+Each call returns one or more canvas packets:
+
+- **PNG canvas**: one of four forms (table below). Two-tone separator lines block edge-feature crosstalk between units; an optional superpixel badge renders unit IDs.
+- **Structured metadata**: `packet_id`, per-unit PTS, token-grid bbox, source-sampling bbox, dual-carrier SHA-256.
+- **Timeline hint line**: a model-readable P/F timeline, for example `[Packet #42 | Anchor: #105 | #101(P, 12.1s, Left-Top) → …]`.
+
+Session capabilities:
+
+- Intent persists across calls.
+- Streaming cursor resume: the next call on the same video continues from the last position.
+- Two-tier history reference: the hot tier reuses the latest 2 full strips; the cold tier reuses up to 4 independent F-anchor carriers.
+- Epoch compaction: hot-to-cold conversion in batches of 6 packets, keeping the multi-turn Prefix KV-Cache prefix stable.
+
+## Canvas and billing
+
+| form | layout | tokens |
+| --- | --- | --- |
+| `FORM_1_STANDARD_STRIP` | 4P + 1F, 1344×672 | 530 |
+| `FORM_2_COMPACT_4P` | 4P, 672×672 | 274 |
+| `FORM_3_COMPACT_1F` | 1F, 672×672 | 274 |
+| `FORM_4_PARTIAL_STRIP` | 1-3P + 1F, 1344×672 | 530 |
+
+## Configuration
+
+| option | default | description |
+| --- | --- | --- |
+| `renderVisualId` | `true` | superpixel inverse badge at each unit corner |
+| `maxPacketsPerCall` | `64` | packet count limit per call |
+| `maxTokensBudgetPerCall` | `65536` | vision-token budget per call |
+| `maxRequestBytes` | `56 MiB` | request byte watermark, Base64-inclusive |
+| `maxColdRefs` | `4` | cold-tier reference limit |
+| `defaultStreamWindowMs` | `10000` | window length in ms when `pts_window` is absent |
+| `streamWorkMaxEdge` | `2048` | working-resolution long edge for streaming decode; deterministic inspection always uses full resolution |
+| `epochCompactionWindow` | `6` | hot-to-cold batch window, in packets |
+| `meteringProfile.imageCountTiersEnabled` | `false` | gate for the ≥15-image 4096-px rule |
+
+## Testing
+
+```bash
+pnpm test        # boundary (pure logic) + e2e (fixture-driven), 20 boundary cases
+pnpm run build   # TypeScript compile
+pnpm verify:host # host-contract self-check (schema shape, keyword subset, dependency boundary)
+```
+
+## Release
+
+npm releases run automatically through GitHub Actions. Process and version policy: [docs/发布规范.md](docs/发布规范.md) (Chinese).
+
+## License
+
+Apache-2.0.

package/README.md

@@ -1,11 +1,13 @@
 # dsh-pf-stream
 
-DeepSeek Harness（DSH）视频感知插件。模型通过一个工具即可"看"本地视频：
+**中文** | [English](README.en.md)
 
-- **连续事件** → 自适应流式采样：运动时按位移积分抽取低 token 过程单元，停稳与顿挫波峰时定格高清焦点单元；
-- **可疑时间点** → 确定性检视：元数据探测、指定时间戳抽帧、均匀切片、局部 RoI 放大复核。
+DeepSeek Harness（DSH）视频感知插件。模型用一个工具即可「看」本地视频：
 
-全部输出画布对齐 DeepSeek-V4.1-Flash 视觉编码管线：42px 超像素网格、计费 `r×(c+1)+2`、单图 ≤1024 token，不触发服务端二次下采样。
+- **连续事件** → 自适应流式采样。运动时按位移积分抽低 token 过程单元；停稳与顿挫波峰时定格高清焦点单元。
+- **可疑时间点** → 确定性检视。能力：元数据探测、指定时间戳抽帧、均匀切片、局部 RoI 放大复核。
+
+全部画布对齐 DeepSeek-V4.1-Flash 视觉编码管线：42px 超像素网格，计费 `r×(c+1)+2`。四种形态计费均 ≤530 token，满足单图 1024 token 上限。服务端按画布原始分辨率计量。
 
 ## 能力路由
 
@@ -13,7 +15,7 @@ DeepSeek Harness（DSH）视频感知插件。模型通过一个工具即可"看
 
 | action | 引擎 | 用途 |
 | --- | --- | --- |
-| `pf_stream`（默认） | 动态流式采样 | 运动中抽 336² 过程单元（P），停稳/波峰定格 672² 焦点单元（F）；施密特四状态机 + 双层自适应调节（意图基准 + 内容自闭环 R1-R4） |
+| `pf_stream`（默认） | 动态流式采样 | 运动中抽 336² 过程单元（P）；停稳/波峰定格 672² 焦点单元（F）。施密特四状态机 + 双层自适应调节（意图基准 + 内容自闭环 R1-R4） |
 | `timestamps` | 确定性检视 | 指定时间戳单帧/组合高清检视，支持 `roi_hint` 局部放大 |
 | `uniform` | 确定性检视 | 固定步长或目标帧数均匀切片 |
 | `probe` | 确定性检视 | 仅元数据（显示宽高/旋转/末帧 PTS/流摘要），零视觉 token |
@@ -22,14 +24,14 @@ DeepSeek Harness（DSH）视频感知插件。模型通过一个工具即可"看
 
 前置条件：
 
-- DSH 宿主（Node ≥ 20）；
-- `ffmpeg` 与 `ffprobe` 可执行文件。发现顺序：插件配置路径 → `DSH_FFMPEG_PATH` / `FFMPEG_PATH` → `DSH_FFMPEG_DIR` → `~/.dsh/tools/ffmpeg/bin` → `~/.dsh/bin` → `PATH`。
+- DSH 宿主（Node ≥ 20）。
+- `ffmpeg` 与 `ffprobe` 可执行文件。定位顺序：插件配置路径 → `DSH_FFMPEG_PATH` / `FFMPEG_PATH` → `DSH_FFMPEG_DIR` → `~/.dsh/tools/ffmpeg/bin` → `~/.dsh/bin` → `PATH`。
 
-在 DSH 插件管理器中安装包 `dsh-pf-stream`。宿主经 `dsh.bundle.patch`（`cordis.patch.yml`）挂载插件；peer 依赖 `@deepseek-ai/cordis` 与 `@deepseek-ai/schemastery` 由宿主运行时解析。运行时直接依赖仅 `pngjs`（画布 PNG 编码，参数固定保证字节幂等）。
+在 DSH 插件管理器中安装包 `dsh-pf-stream`。宿主经 `dsh.bundle.patch`（`cordis.patch.yml`）挂载插件。peer 依赖 `@deepseek-ai/cordis` 与 `@deepseek-ai/schemastery` 由宿主运行时解析。运行时直接依赖仅 `pngjs`；画布 PNG 编码参数固定，输出字节幂等。
 
 ## 使用
 
-流式感知（问"画面里写了什么"）：
+流式感知（问「画面里写了什么」）：
 
 ```json
 {
@@ -38,7 +40,7 @@ DeepSeek Harness（DSH）视频感知插件。模型通过一个工具即可"看
 }
 ```
 
-定点复核（在流式结果中发现 12.9s 处有文字）：
+定点复核（流式结果中 12.9s 处发现文字）：
 
 ```json
 {
@@ -51,17 +53,22 @@ DeepSeek Harness（DSH）视频感知插件。模型通过一个工具即可"看
 }
 ```
 
-顶层同时接受视频检视领域常见便捷字段（`start_time` / `end_time` / `timestamps` / `fps` / `interval` / `num_frames` / `crop` / `probe_only`），由入口适配器归一化为标准 `intent` 结构。
+顶层同时接受视频检视领域常见便捷字段：`start_time` / `end_time` / `timestamps` / `fps` / `interval` / `num_frames` / `crop` / `probe_only`。入口适配器将其归一化为标准 `intent` 结构。
 
 ## 返回内容
 
 每次调用返回一个或多个画布包：
 
-- **PNG 画布**：四形态之一（见下表），双色隔离线阻断相邻单元边缘串扰，可选超像素角标渲染单元编号；
-- **结构化元数据**：`packet_id`、每单元 PTS、token 网格 bbox、源采样 bbox、双载体 SHA-256；
+- **PNG 画布**：四形态之一（见下表）。单元间双色隔离线阻断相邻边缘特征串扰；可选超像素角标渲染单元编号。
+- **结构化元数据**：`packet_id`、每单元 PTS、token 网格 bbox、源采样 bbox、双载体 SHA-256。
 - **时序提示行**：模型可读的 P/F 单元时间轴（如 `[Packet #42 | Anchor: #105 | #101(P, 12.1s, Left-Top) → …]`）。
 
-会话级能力：意图跨调用持久化；同视频流式游标续窗（下次调用从上次位置继续）；两级历史引用（热区最近 2 包整条带复用、冷区至多 4 个 F 锚点独立载体复用）与 Epoch Compaction（默认 6 包窗口）保持多轮对话 Prefix KV-Cache 前缀稳定。
+会话级能力：
+
+- 意图跨调用持久化。
+- 流式游标续窗：同视频下次调用从上次位置继续。
+- 两级历史引用：热区最近 2 包整条带复用；冷区至多 4 个 F 锚点独立载体复用。
+- Epoch Compaction：默认 6 包窗口批量热转冷，保持多轮对话 Prefix KV-Cache 前缀稳定。
 
 ## 画布与计费
 
@@ -86,8 +93,6 @@ DeepSeek Harness（DSH）视频感知插件。模型通过一个工具即可"看
 | `epochCompactionWindow` | `6` | 热转冷批量收敛窗口（包数） |
 | `meteringProfile.imageCountTiersEnabled` | `false` | ≥15 图单边 4096px 规则门控 |
 
-完整配置、设计依据与 20 项工程边界验证方案见 `docs/dsh-pf-stream 技术设计方案.md`。
-
 ## 测试
 
 ```bash
@@ -96,6 +101,10 @@ pnpm run build   # TypeScript 编译
 pnpm verify:host # 宿主契约自检（Schema 形状/关键字子集/依赖边界）
 ```
 
+## 发布
+
+npm 发布由 GitHub Actions 自动执行。流程与版本策略见 [docs/发布规范.md](docs/发布规范.md)。
+
 ## 许可证
 
-Apache-2.0。实现参考了 dsh-video-see、dsh-video-lens（MIT）与 dsh-see-now（Apache-2.0）的公开实现，无逐字复制；见 `NOTICE`。
+Apache-2.0。

package/cordis.patch.yml

@@ -1,2 +1,4 @@
-insert:
-  - name: dsh-pf-stream
+# dsh bundle patch: inserts this plugin into a profile's layer stack.
+- insert:
+    - id: dsh-pf-stream
+      name: 'dsh-pf-stream'

package/package.json

@@ -1,16 +1,28 @@
 {
   "name": "dsh-pf-stream",
-  "version": "0.1.0",
+  "version": "0.1.3",
   "description": "DeepSeek Harness video perception plugin: adaptive dual-scale streaming sampling plus deterministic inspection, aligned to the DeepSeek-V4.1-Flash vision pipeline (42px superpixel grid, <=1024 tokens per image).",
   "type": "module",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/Weilv-D/dsh-pf-stream"
+  },
   "main": "lib/index.js",
   "types": "lib/index.d.ts",
   "files": [
     "lib",
     "cordis.patch.yml",
     "LICENSE",
-    "NOTICE"
+    "README.md",
+    "README.en.md"
   ],
+  "scripts": {
+    "build": "tsc -p tsconfig.json",
+    "typecheck": "tsc -p tsconfig.json --noEmit",
+    "verify:host": "node scripts/verify-host-contract.mjs",
+    "check:package": "node scripts/check-package.mjs",
+    "test": "vitest run --passWithNoTests"
+  },
   "peerDependencies": {
     "@deepseek-ai/cordis": "^4.0.1",
     "@deepseek-ai/schemastery": "^3.18.1"
@@ -41,11 +53,5 @@
     "vision",
     "ffmpeg"
   ],
-  "license": "Apache-2.0",
-  "scripts": {
-    "build": "tsc -p tsconfig.json",
-    "typecheck": "tsc -p tsconfig.json --noEmit",
-    "verify:host": "node scripts/verify-host-contract.mjs",
-    "test": "vitest run --passWithNoTests"
-  }
-}
+  "license": "Apache-2.0"
+}

package/NOTICE

@@ -1,13 +0,0 @@
-dsh-pf-stream
-Copyright 2026 Weilv-D
-
-This product is licensed under the Apache License, Version 2.0 (the "License");
-you may not use this product except in compliance with the License.
-
-This product includes design references to the following open-source projects
-(no source code is verbatim copied; refer to docs/dsh-pf-stream 技术设计方案.md
-section 2.5 for the reference-reuse inventory):
-
-  - dsh-video-see (MIT) — probing, rotation parsing, frame-size planning patterns
-  - dsh-video-lens (MIT) — bounded-concurrency frame extraction patterns
-  - dsh-see-now (Apache-2.0) — rawvideo streaming pipeline and tool contract patterns