yunti-browser-runtime 0.2.0 → 0.2.1

package/extension/content.js

@@ -338,7 +338,7 @@ function isExtensionContextInvalidated(error) {
 }
 
 function contextInvalidatedResponse() {
-  return { ok: false, error: "插件已重新加载，请刷新 browser page恢复连接。" }
+  return { ok: false, error: "插件已重新加载，正在等待扩展自动恢复连接；必要时再刷新页面。" }
 }
 
 function markExtensionContextInvalidated() {
@@ -382,6 +382,8 @@ async function executeTool(tool, args) {
       return clickElement(args)
     case "yunti_click_at":
       return clickAt(args)
+    case "yunti_hover":
+      return hoverElement(args)
     case "yunti_fill":
       return fillElement(args)
     case "yunti_type_text":
@@ -548,10 +550,13 @@ function rollbackPreviewPatch(patchId) {
 }
 
 function clickElement(args) {
-  const element = mustFind(args.selector)
-  element.scrollIntoView({ block: "center", inline: "center" })
-  element.click()
-  return { clicked: true, element: describeElement(element) }
+  if (args.selector) {
+    const element = mustFind(args.selector)
+    element.scrollIntoView({ block: "center", inline: "center" })
+    element.click()
+    return { clicked: true, element: describeElement(element) }
+  }
+  return clickAt(args)
 }
 
 function clickAt(args) {
@@ -569,12 +574,36 @@ function clickAt(args) {
   }
 }
 
+function hoverElement(args) {
+  const element = resolveTargetElement(args)
+  if (!element) throw new Error("yunti_hover requires selector or x/y coordinates")
+  element.scrollIntoView({ block: "center", inline: "center" })
+  const rect = element.getBoundingClientRect()
+  const x = Number.isFinite(Number(args.x)) ? Number(args.x) : rect.left + rect.width / 2
+  const y = Number.isFinite(Number(args.y)) ? Number(args.y) : rect.top + rect.height / 2
+  focusElement(element)
+  dispatchHoverSequence(element, x, y)
+  return {
+    hovered: true,
+    x: Math.round(x),
+    y: Math.round(y),
+    element: describeElement(element),
+  }
+}
+
 function fillElement(args) {
-  const element = mustFind(args.selector)
+  const element = resolveTargetElement(args)
+  if (!element) throw new Error("yunti_fill requires selector or x/y coordinates")
   const value = String(args.value ?? "")
   element.scrollIntoView({ block: "center", inline: "center" })
   setEditableText(element, value, { replace: true })
-  return { filled: true, element: describeElement(element), valueLength: value.length }
+  return {
+    filled: true,
+    element: describeElement(element),
+    valueLength: value.length,
+    valueApplied: getEditableText(element) === value,
+    method: element.isContentEditable ? "contenteditable" : element instanceof HTMLSelectElement ? "select" : "dom",
+  }
 }
 
 function typeText(args) {
@@ -612,32 +641,55 @@ function pressKey(args) {
 }
 
 function selectElement(args) {
-  const element = mustFind(args.selector)
+  const element = resolveTargetElement(args)
+  if (!element) throw new Error("yunti_select requires selector or x/y coordinates")
   if (!(element instanceof HTMLSelectElement)) throw new Error("Target is not a select element")
   const value = String(args.value ?? "")
+  const text = String(args.text ?? "")
+  const matchMode = text ? "text" : "value"
+  const targetOption = text || value
   const options = Array.from(element.options)
-  const option = options.find((item) => item.value === value)
+  const option = options.find((item) => matchMode === "text" ? item.text.trim() === targetOption : item.value === targetOption)
+  if (!option) {
+    return {
+      selected: false,
+      element: describeElement(element),
+      value: element.value,
+      code: "OPTION_NOT_FOUND",
+      error: matchMode === "text" ? "Option text not found" : "Option value not found",
+      matchMode,
+      targetOption,
+      availableValues: options.map((item) => item.value).slice(0, 50),
+      availableTexts: options.map((item) => item.text.trim()).slice(0, 50),
+      options: summarizeSelectOptions(options),
+    }
+  }
   if (option?.disabled) {
     return {
       selected: false,
       element: describeElement(element),
       value: element.value,
       code: "OPTION_DISABLED",
-      error: "Option value is disabled",
+      error: matchMode === "text" ? "Option text is disabled" : "Option value is disabled",
+      matchMode,
+      targetOption,
       disabledValue: option.value,
       disabledText: option.text.trim(),
-      options: options.slice(0, 50).map((item) => ({
-        value: item.value,
-        text: item.text.trim(),
-        disabled: Boolean(item.disabled),
-        selected: Boolean(item.selected),
-      })),
+      availableValues: options.map((item) => item.value).slice(0, 50),
+      availableTexts: options.map((item) => item.text.trim()).slice(0, 50),
+      options: summarizeSelectOptions(options),
     }
   }
-  element.value = value
+  element.value = option.value
   element.dispatchEvent(new Event("input", { bubbles: true }))
   element.dispatchEvent(new Event("change", { bubbles: true }))
-  return { selected: true, element: describeElement(element), value: element.value }
+  return {
+    selected: true,
+    element: describeElement(element),
+    value: element.value,
+    text: option.text.trim(),
+    selectedIndex: element.selectedIndex,
+  }
 }
 
 function scrollPage(args) {
@@ -693,6 +745,15 @@ function resolveTargetElement(args = {}) {
   return null
 }
 
+function summarizeSelectOptions(options) {
+  return options.slice(0, 50).map((item) => ({
+    value: item.value,
+    text: item.text.trim(),
+    disabled: Boolean(item.disabled),
+    selected: Boolean(item.selected),
+  }))
+}
+
 function focusElement(element) {
   if (element && typeof element.focus === "function") {
     element.focus({ preventScroll: true })
@@ -720,6 +781,27 @@ function dispatchPointerMouseSequence(element, x, y) {
   }
 }
 
+function dispatchHoverSequence(element, x, y) {
+  const common = {
+    bubbles: true,
+    cancelable: true,
+    composed: true,
+    clientX: x,
+    clientY: y,
+    view: window,
+  }
+  for (const type of ["pointerover", "pointerenter", "pointermove"]) {
+    try {
+      element.dispatchEvent(new PointerEvent(type, { ...common, pointerType: "mouse", isPrimary: true }))
+    } catch {
+      break
+    }
+  }
+  for (const type of ["mouseover", "mouseenter", "mousemove"]) {
+    element.dispatchEvent(new MouseEvent(type, common))
+  }
+}
+
 function setEditableText(element, text, options = {}) {
   assertEditableTarget(element)
   if (element.isContentEditable) {

package/extension/manifest.json

@@ -1,11 +1,12 @@
 {
   "manifest_version": 3,
   "name": "Yunti Browser Runtime",
-  "version": "0.2.0",
+  "version": "0.2.1",
   "description": "Local browser runtime for AI agents through MCP.",
   "permissions": [
     "activeTab",
     "debugger",
+    "scripting",
     "storage",
     "tabs",
     "webRequest"

package/extension/popup.js

@@ -33,7 +33,7 @@ async function save() {
     return
   }
   renderSettings(state.settings || {})
-  setStatus("设置已保存。请刷新目标页面完成重新注册。")
+  setStatus("设置已保存。扩展会自动尝试重新注册可访问页面。")
 }
 
 function renderSettings(settings) {
@@ -56,7 +56,7 @@ function statusText(state) {
   if (!state.bridge?.ok) {
     return "Bridge 未连接。请确认 Agent MCP 已启动，或运行 yunti-browser-runtime bridge。"
   }
-  return "默认无需设置。打开或刷新任意 http/https 页面即可连接。"
+  return "默认无需设置。打开任意 http/https 页面后会自动连接。"
 }
 
 function setStatus(text) {

package/extension/session-manager.js

@@ -12,6 +12,8 @@ import {
 export function createSessionManager() {
   const sessionsByTab = new Map()
   const pollers = new Map()
+  const pendingTabRecovery = new Map()
+  const lastInjectionAtByTab = new Map()
   let cachedPlatformMatches = null
   let toolRequestHandler = null
 
@@ -38,6 +40,104 @@ export function createSessionManager() {
     }).catch(() => {})
   }
 
+  async function ensureAllTabsRegistered(options = {}) {
+    const settings = await getSettings()
+    cachedPlatformMatches = settings.platformMatches
+    const tabs = await chrome.tabs.query({})
+    const results = await Promise.allSettled(
+      tabs.map((tab) => ensureTabRegistered(tab.id, { ...options, tab, settings }))
+    )
+    return {
+      ok: true,
+      reason: options.reason || "manual",
+      checked: tabs.length,
+      recovered: results.filter((result) => result.value?.registered || result.value?.injected)
+        .length,
+      skipped: results.filter((result) => result.value?.skipped).length,
+      failed: results.filter((result) => result.status === "rejected" || result.value?.ok === false)
+        .length,
+    }
+  }
+
+  async function ensureTabRegistered(tabId, options = {}) {
+    if (!Number.isFinite(Number(tabId))) {
+      return { ok: false, skipped: true, reason: "missing_tab_id" }
+    }
+    const key = Number(tabId)
+    if (pendingTabRecovery.has(key)) return pendingTabRecovery.get(key)
+    const promise = recoverTabRegistration(key, options).finally(() => {
+      pendingTabRecovery.delete(key)
+    })
+    pendingTabRecovery.set(key, promise)
+    return promise
+  }
+
+  async function recoverTabRegistration(tabId, options = {}) {
+    const settings = options.settings || (await getSettings())
+    cachedPlatformMatches = settings.platformMatches
+    const tab = options.tab || (await chrome.tabs.get(tabId).catch(() => null))
+    if (!isInjectableTab(tab, settings)) {
+      return { ok: true, skipped: true, reason: "unsupported_tab" }
+    }
+    if (sessionsByTab.has(tabId)) {
+      await refreshTabRegistration(tabId).catch(() => null)
+      return { ok: true, registered: true, reason: "already_registered" }
+    }
+    const ping = await refreshTabRegistration(tabId).catch((error) => ({
+      ok: false,
+      error: error?.message || String(error),
+    }))
+    if (ping?.ok) {
+      return { ok: true, registered: true, reason: "content_script_present" }
+    }
+    const lastInjectionAt = lastInjectionAtByTab.get(tabId) || 0
+    if (Date.now() - lastInjectionAt < 5000) {
+      return { ok: true, skipped: true, reason: "recently_injected" }
+    }
+    const injected = await injectContentScripts(tabId).catch((error) => ({
+      ok: false,
+      error: error?.message || String(error),
+    }))
+    if (!injected?.ok) {
+      return { ok: false, reason: "inject_failed", error: injected?.error || "inject failed" }
+    }
+    lastInjectionAtByTab.set(tabId, Date.now())
+    await delay(50)
+    await refreshTabRegistration(tabId).catch(() => null)
+    return { ok: true, injected: true, reason: options.reason || "auto_recovery" }
+  }
+
+  function isInjectableTab(tab, settings) {
+    if (!tab?.id) return false
+    if (tab.discarded) return false
+    return isPlatformUrl(tab.url, settings)
+  }
+
+  async function refreshTabRegistration(tabId) {
+    return chrome.tabs.sendMessage(tabId, { type: "yunti_refresh_registration" })
+  }
+
+  async function injectContentScripts(tabId) {
+    if (!chrome.scripting?.executeScript) {
+      return { ok: false, error: "chrome.scripting permission is unavailable" }
+    }
+    if (chrome.scripting.insertCSS) {
+      await chrome.scripting.insertCSS({
+        target: { tabId },
+        files: ["content.css"],
+      })
+    }
+    await chrome.scripting.executeScript({
+      target: { tabId },
+      files: ["dom-observer.js"],
+    })
+    await chrome.scripting.executeScript({
+      target: { tabId },
+      files: ["content.js"],
+    })
+    return { ok: true }
+  }
+
   async function handleMessage(message, sender) {
     switch (message?.type) {
       case "yunti_content_ready":
@@ -244,9 +344,7 @@ export function createSessionManager() {
   async function refreshActiveTab() {
     const [tab] = await chrome.tabs.query({ active: true, currentWindow: true })
     if (!tab?.id) return { ok: false, error: "no active tab" }
-    await chrome.tabs
-      .sendMessage(tab.id, { type: "yunti_refresh_registration" })
-      .catch(() => null)
+    await ensureTabRegistered(tab.id, { tab, reason: "popup_refresh" }).catch(() => null)
     return getPanelState()
   }
 
@@ -254,6 +352,8 @@ export function createSessionManager() {
     sessionsByTab,
     pollers,
     activateTab,
+    ensureAllTabsRegistered,
+    ensureTabRegistered,
     forgetTab,
     forwardConsoleEvent,
     getPlatformMatches,
@@ -279,7 +379,7 @@ function normalizePageAuth(auth, source = "unknown") {
     return {
       state: "missing_from_content_script",
       loggedIn: false,
-      reason: "Content script did not report page state; reload the extension and refresh the page",
+      reason: "Content script did not report page state; Yunti will try automatic registration recovery before asking for a page refresh",
       checkedAt: new Date().toISOString(),
       checkedBy: source,
     }