dsh-mic-dictation 0.1.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 dsh-mic-dictation contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,77 @@
+# dsh-mic-dictation
+
+DeepSeek Harness 原生客户端插件：在 Web 提问栏的 **Full access 左侧** 插入一个麦克风按钮，点一下直接说话，识别文字写进提问框。
+
+- 浏览器原生 `SpeechRecognition`，中文识别（`zh-CN`）
+- 无需外部 API Key
+- 纯客户端插件：host 半是空 `apply`，浏览器半通过 `dsh.client` 清单发货
+- 用 `MutationObserver` 盯住 DSH 的 React DOM，把按钮稳定地放在 Full access（访问模式）按钮左边
+
+## 目录
+
+```text
+dsh-mic-dictation/
+├── package.json        # 插件清单 + dsh.client / dsh.bundle 声明
+├── cordis.patch.yml    # bundle patch：把本插件插入 web profile
+├── lib/
+│   ├── index.js        # host half（空 apply）
+│   ├── index.d.ts
+│   ├── client.js       # 浏览器 half（lazy-CJS factory bundle）
+│   └── client.d.ts
+├── README.md
+└── LICENSE
+```
+
+## 安装到自己的 DeepSeek Harness
+
+先发布到 npm 或 GitHub，然后用户在自己的终端里执行：
+
+```bash
+# 从 npm 安装
+dsh plugin --profile web add dsh-mic-dictation
+
+# 或从 GitHub 安装（pnpm 会直接安装该仓库）
+dsh plugin --profile web add github:你的用户名/dsh-mic-dictation
+```
+
+重启 Web：
+
+```bash
+dsh web
+```
+
+刷新页面后，首次点麦克风并允许浏览器使用麦克风即可。
+
+> 说明：当前 DSH 版本的「插件市场」是**只读清单**（设置 → 插件列表）；安装入口是 `dsh plugin --profile web add ...`。等官方市场 UI 上线后，只要这个包发布在 npm/GitHub，就能被市场检索到。
+
+## 发布到 GitHub / npm
+
+### GitHub
+
+1. 把这个目录推到新仓库。
+2. 用户安装：`dsh plugin --profile web add github:OWNER/dsh-mic-dictation`。
+3. 仓库里必须保留 `cordis.patch.yml`、`lib/client.js` 和 `package.json`；它们是发布内容，不是构建产物。
+
+### npm
+
+```bash
+npm login
+npm publish --access public
+```
+
+之后用户直接 `dsh plugin --profile web add dsh-mic-dictation`。
+
+## 工作原理
+
+1. `cordis.patch.yml` 把 `dsh-mic-dictation` 作为一个 Loader entry 插进 profile。
+2. host 扫描该 entry，读到 `package.json` 的 `dsh.client` 声明，把 `exports["./client"]`（`lib/client.js`）作为客户端 bundle 提供给浏览器。
+3. 浏览器用 `window.__ModuleLoader__.load({ id, factory })` 注册这个 lazy-CJS 模块。
+4. `apply(ctx)` 在页面启动后挂载一个 DOM controller，找到 `[data-composer-card]` 里的 Full access / 访问模式按钮，把麦克风按钮插到它前面。
+5. 语音识别结果通过原生 value setter + `input` 事件写入 textarea，React 正常接管草稿。
+
+## 限制
+
+- 需要 Chrome / Edge / Safari（`SpeechRecognition` / `webkitSpeechRecognition`）。
+- 需要页面运行在 localhost / 127.0.0.1（浏览器的安全上下文要求）。
+- Chrome 的语音识别依赖 Google 语音服务，需要联网。
+- 如果 DSH 前端升级改变了 `data-composer-card` 或访问模式按钮的 `aria-label`，需要同步调整 `lib/client.js` 里的选择器。

package/cordis.patch.yml

@@ -0,0 +1,8 @@
+# dsh-mic-dictation bundle patch.
+# Installed as a profile bundle after dsh-web-app, this inserts the client
+# plugin into the web profile's Loader tree. The host half is an empty apply;
+# the browser half ships through exports["./client"] and is discovered from
+# the "dsh.client" manifest in package.json.
+- insert:
+    - id: mic-dictation
+      name: dsh-mic-dictation

package/lib/client.d.ts

@@ -0,0 +1,4 @@
+/** Services required by this client plugin (none). */
+export declare const inject: string[];
+/** Mount the microphone dictation controller in the web composer. */
+export declare function apply(ctx: unknown): void;

package/lib/client.js

@@ -0,0 +1,234 @@
+window.__ModuleLoader__.load({
+  id: "dsh-mic-dictation",
+  factory: function (require) {
+    "use strict";
+    var module = { exports: {} };
+    var exports = module.exports;
+    Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+
+    var inject = [];
+
+    function apply(ctx) {
+      if (typeof window === "undefined" || typeof document === "undefined") return;
+      if (window.__dshMicDictationInstalled) return;
+      window.__dshMicDictationInstalled = true;
+
+      var SR = window.SpeechRecognition || window.webkitSpeechRecognition;
+      if (!SR) {
+        console.warn("[dsh-mic-dictation] SpeechRecognition is not supported by this browser.");
+        return;
+      }
+
+      var style = document.createElement("style");
+      style.dataset.plugin = "dsh-mic-dictation";
+      style.dataset.pluginCss = "dsh-mic-dictation/client.css";
+      style.textContent = [
+        ".dsh-mic-btn{display:inline-flex;align-items:center;justify-content:center;width:28px;height:28px;flex:none;border:none;border-radius:999px;background:transparent;color:var(--dsw-alias-label-secondary,#6b7280);cursor:pointer;padding:0;margin:0;}",
+        ".dsh-mic-btn:hover{background:var(--dsw-alias-interactive-bg-hover,rgba(0,0,0,0.06));}",
+        ".dsh-mic-btn.listening{color:#ef4444;background:rgba(239,68,68,0.12);}",
+        ".dsh-mic-btn.listening svg{animation:dsh-mic-pulse 1.2s ease-in-out infinite;}",
+        "@keyframes dsh-mic-pulse{0%,100%{opacity:1;}50%{opacity:0.45;}}"
+      ].join("");
+      document.head.appendChild(style);
+
+      var button = document.createElement("button");
+      button.type = "button";
+      button.id = "dsh-mic-dictation-btn";
+      button.className = "dsh-mic-btn";
+      button.setAttribute("aria-label", "语音输入");
+      button.title = "语音输入";
+      button.innerHTML =
+        '<svg width="16" height="16" viewBox="0 0 16 16" fill="none" aria-hidden="true">' +
+        '<rect x="6" y="1.5" width="4" height="9" rx="2" stroke="currentColor" stroke-width="1.3"/>' +
+        '<path d="M3.5 7.5a4.5 4.5 0 0 0 9 0M8 12.5V15M5.5 15h5" stroke="currentColor" stroke-width="1.3" stroke-linecap="round"/>' +
+        "</svg>";
+
+      var recognition = null;
+      var listening = false;
+      var scheduled = false;
+      var syncTimer = 0;
+
+      function findAccessButton() {
+        var cards = document.querySelectorAll("[data-composer-card]");
+        for (var i = 0; i < cards.length; i++) {
+          var btn = cards[i].querySelector(
+            'button[aria-label^="Access mode"], button[aria-label^="访问模式"]'
+          );
+          if (btn) return btn;
+        }
+        return null;
+      }
+
+      function scheduleSync() {
+        if (scheduled) return;
+        scheduled = true;
+        requestAnimationFrame(function () {
+          scheduled = false;
+          syncPlacement();
+        });
+      }
+
+      function syncPlacement() {
+        var target = findAccessButton();
+        if (!target) {
+          if (button.isConnected) button.remove();
+          return;
+        }
+        var parent = target.parentElement;
+        if (!parent) return;
+        if (button.parentElement !== parent || button.nextElementSibling !== target) {
+          target.insertAdjacentElement("beforebegin", button);
+        }
+      }
+
+      function setListening(on) {
+        button.classList.toggle("listening", on);
+        button.title = on ? "正在聆听，点击停止" : "语音输入";
+      }
+
+      function setNativeValue(el, value) {
+        var proto = el instanceof HTMLTextAreaElement
+          ? HTMLTextAreaElement.prototype
+          : HTMLInputElement.prototype;
+        var desc = Object.getOwnPropertyDescriptor(proto, "value");
+        if (desc && desc.set) desc.set.call(el, value);
+        else el.value = value;
+        el.dispatchEvent(new Event("input", { bubbles: true }));
+      }
+
+      function appendTranscript(text) {
+        var clean = String(text || "").trim();
+        if (!clean) return;
+        var card = button.closest("[data-composer-card]");
+        var textarea = card ? card.querySelector("textarea") : null;
+        if (!textarea) return;
+
+        var start = textarea.selectionStart != null
+          ? textarea.selectionStart
+          : textarea.value.length;
+        var end = textarea.selectionEnd != null ? textarea.selectionEnd : start;
+        var next = textarea.value.slice(0, start) + clean + textarea.value.slice(end);
+        setNativeValue(textarea, next);
+
+        var caret = start + clean.length;
+        try { textarea.setSelectionRange(caret, caret); } catch (_) {}
+        try { textarea.focus({ preventScroll: true }); } catch (_) { textarea.focus(); }
+      }
+
+      function stop() {
+        listening = false;
+        setListening(false);
+        if (recognition) {
+          try { recognition.stop(); } catch (_) {}
+        }
+      }
+
+      function start() {
+        if (listening) return;
+        var rec = new SR();
+        rec.lang = "zh-CN";
+        rec.continuous = true;
+        rec.interimResults = true;
+        recognition = rec;
+        listening = true;
+        setListening(true);
+
+        rec.onresult = function (event) {
+          var interim = "";
+          for (var i = event.resultIndex; i < event.results.length; i++) {
+            var result = event.results[i];
+            if (result.isFinal) {
+              appendTranscript(result[0].transcript);
+            } else {
+              interim += result[0].transcript;
+            }
+          }
+          if (interim) button.title = interim;
+        };
+
+        rec.onerror = function (event) {
+          console.warn("[dsh-mic-dictation] recognition error:", event.error, event.message || "");
+          if (event.error === "not-allowed" || event.error === "service-not-allowed") {
+            listening = false;
+            setListening(false);
+            button.title = "麦克风权限被拒绝，请在浏览器地址栏允许麦克风";
+          } else if (event.error === "audio-capture") {
+            listening = false;
+            setListening(false);
+            button.title = "未检测到麦克风设备";
+          } else if (event.error === "no-speech") {
+            button.title = "没有听到语音，请再试一次";
+          } else if (event.error === "network") {
+            listening = false;
+            setListening(false);
+            button.title = "网络错误，语音识别不可用";
+          } else if (event.error === "aborted") {
+            // 主动停止或页面切换，交给 onend 处理。
+          } else {
+            listening = false;
+            setListening(false);
+            button.title = "语音识别错误：" + event.error;
+          }
+        };
+
+        rec.onend = function () {
+          if (listening) {
+            try {
+              rec.start();
+            } catch (err) {
+              listening = false;
+              setListening(false);
+              button.title = "语音识别已结束";
+            }
+          } else {
+            setListening(false);
+            var keep = button.title.indexOf("麦克风权限") === 0
+              || button.title.indexOf("网络") === 0
+              || button.title.indexOf("未检测到") === 0
+              || button.title.indexOf("语音识别错误") === 0
+              || button.title.indexOf("没有听到") === 0;
+            if (!keep) button.title = "语音输入";
+          }
+        };
+
+        try {
+          rec.start();
+        } catch (err) {
+          listening = false;
+          setListening(false);
+          button.title = "无法启动语音识别：" + err.message;
+        }
+      }
+
+      button.addEventListener("click", function (event) {
+        event.preventDefault();
+        event.stopPropagation();
+        if (listening) stop();
+        else start();
+      });
+
+      var observer = new MutationObserver(function () { scheduleSync(); });
+      observer.observe(document.documentElement, { childList: true, subtree: true });
+      syncTimer = window.setInterval(syncPlacement, 1000);
+      scheduleSync();
+
+      ctx.effect(function () {
+        return function () {
+          observer.disconnect();
+          window.clearInterval(syncTimer);
+          if (recognition) {
+            try { recognition.abort(); } catch (_) {}
+          }
+          if (button.isConnected) button.remove();
+          if (style.isConnected) style.remove();
+          window.__dshMicDictationInstalled = false;
+        };
+      }, "dsh-mic-dictation: DOM controller");
+    }
+
+    exports.apply = apply;
+    exports.inject = inject;
+    return module.exports;
+  }
+});
+//# sourceMappingURL=client.js.map

package/lib/index.d.ts

@@ -0,0 +1,2 @@
+/** Host plugin body — no host-side behavior for this client-only plugin. */
+export declare function apply(): void;

package/lib/index.js

@@ -0,0 +1,8 @@
+/**
+ * dsh-mic-dictation, host half.
+ *
+ * The empty apply exists so this package is a real Loader entry on the host
+ * and the client half is discovered through the package.json "dsh.client"
+ * declaration. All behavior lives in the browser half (./client).
+ */
+export function apply() {}

package/package.json

@@ -0,0 +1,46 @@
+{
+  "name": "dsh-mic-dictation",
+  "version": "0.1.0",
+  "description": "DeepSeek Harness client plugin: a microphone dictation button to the left of the Full access control in the web composer.",
+  "type": "module",
+  "main": "lib/index.js",
+  "types": "lib/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./lib/index.d.ts",
+      "default": "./lib/index.js"
+    },
+    "./client": {
+      "types": "./lib/client.d.ts",
+      "default": "./lib/client.js"
+    },
+    "./package.json": "./package.json"
+  },
+  "files": [
+    "lib",
+    "cordis.patch.yml",
+    "README.md",
+    "LICENSE"
+  ],
+  "dsh": {
+    "bundle": {
+      "patch": "./cordis.patch.yml"
+    },
+    "client": {
+      "inject": [],
+      "platform": "web"
+    }
+  },
+  "peerDependencies": {
+    "@deepseek-ai/cordis": "^4.0.1"
+  },
+  "license": "MIT",
+  "keywords": [
+    "deepseek-harness",
+    "dsh",
+    "plugin",
+    "voice",
+    "dictation",
+    "speech-recognition"
+  ]
+}