dsh-notify-windows 0.6.0 → 0.7.3

package/lib/launcher.ps1

@@ -0,0 +1,310 @@
+# dsh-notify launcher (Windows PowerShell 5.1).
+# Registered as the "dshnotify" protocol handler. Invoked by the OS as:
+#   powershell.exe -NoProfile -ExecutionPolicy Bypass -WindowStyle Hidden -File "launcher.ps1" "%1"
+# It decodes the target URL from dshnotify://open?u=<encoded> and tries, in order:
+#   A) navigate an already-open DSH tab via Chrome DevTools Protocol (WebSocket);
+#   B) focus an already-open DSH window via Win32 P/Invoke;
+#   C) open the target URL in the default browser (Start-Process).
+# Every failure degrades silently. The click must never do nothing.
+param(
+  [Parameter(Mandatory = $true)][string]$ProtocolArgs
+)
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+function Write-Trace {
+  param([string]$Message)
+  try {
+    $dir = Join-Path $env:TEMP "dsh-notify"
+    if (-not (Test-Path $dir)) { New-Item -Path $dir -ItemType Directory -Force | Out-Null }
+    $line = (Get-Date -Format "yyyy-MM-dd HH:mm:ss.fff") + " " + $Message
+    Add-Content -Path (Join-Path $dir "launcher.log") -Value $line -Encoding UTF8
+  } catch {
+    # diagnostics must never break the flow
+  }
+}
+
+function Send-WebSocketJson {
+  # Synchronously sends a JSON string over a Chrome DevTools Protocol WebSocket
+  # and waits briefly for an ack frame. Returns $true on successful send.
+  param(
+    [string]$WsUrl,
+    [string]$Json,
+    [int]$TimeoutMs = 2000
+  )
+  try {
+    $ws = [System.Net.WebSockets.ClientWebSocket]::new()
+    $ct = [System.Threading.CancellationToken]::None
+    $connectTask = $ws.ConnectAsync($WsUrl, $ct)
+    $connectTask.Wait($TimeoutMs)
+    if ($ws.State -ne [System.Net.WebSockets.WebSocketState]::Open) {
+      try { $ws.Dispose() } catch {}
+      return $false
+    }
+    $bytes = [System.Text.Encoding]::UTF8.GetBytes($Json)
+    $seg = New-Object System.ArraySegment[byte] (, $bytes)
+    $sendTask = $ws.SendAsync($seg, [System.Net.WebSockets.WebMessageType]::Text, $true, $ct)
+    $sendTask.Wait($TimeoutMs)
+    if (-not $sendTask.IsCompletedSuccessfully) {
+      try { $ws.Dispose() } catch {}
+      return $false
+    }
+    # Best-effort ack read so the frame is flushed before we close.
+    $buf = New-Object byte[] 4096
+    $recvSeg = New-Object System.ArraySegment[byte] (, $buf)
+    $recvTask = $ws.ReceiveAsync($recvSeg, $ct)
+    if ($recvTask.Wait($TimeoutMs)) {
+      $ackBytes = $recvSeg.Array[0..($recvTask.Result.Count - 1)]
+      Write-Trace ("[cdp] ack=" + [System.Convert]::ToBase64String($ackBytes))
+    }
+    try { $ws.CloseAsync([System.Net.WebSockets.WebSocketCloseStatus]::NormalClosure, "ok", $ct).Wait($TimeoutMs) } catch {}
+    try { $ws.Dispose() } catch {}
+    return $true
+  } catch {
+    Write-Trace ("[fail step] Send-WebSocketJson: " + $_.Exception.Message)
+    try { if ($ws -ne $null) { $ws.Dispose() } } catch {}
+    return $false
+  }
+}
+
+function Invoke-CdpNavigate {
+  # Tries CDP on the standard debug ports plus any port discovered from a
+  # running chrome.exe command line. Returns $true if it navigated a page.
+  param([string]$TargetUrl, [string]$HostPort)
+  $ports = @(9222, 9223, 9229)
+  try {
+    $procs = Get-CimInstance Win32_Process -Filter "Name='chrome.exe'" -ErrorAction SilentlyContinue
+    foreach ($p in $procs) {
+      if ($p.CommandLine -match "--remote-debugging-port=(\d+)") {
+        $ports += [int]$Matches[1]
+      }
+    }
+  } catch {}
+  $ports = $ports | Sort-Object -Unique
+
+  foreach ($port in $ports) {
+    try {
+      $ver = $null
+      try {
+        $ver = Invoke-WebRequest -Uri "http://127.0.0.1:$port/json/version" -TimeoutSec 1 -UseBasicParsing
+      } catch {}
+      if ($ver -eq $null) { continue }
+      $list = Invoke-WebRequest -Uri "http://127.0.0.1:$port/json/list" -TimeoutSec 1 -UseBasicParsing
+      $targets = $list.Content | ConvertFrom-Json
+      $hit = $null
+      foreach ($t in $targets) {
+        if ($t.type -eq "page" -and $t.url -match [regex]::Escape($HostPort)) {
+          $hit = $t
+          break
+        }
+      }
+      if ($hit -eq $null) { continue }
+      $wsUrl = "ws://127.0.0.1:$port/devtools/page/$($hit.id)"
+      $json = '{"id":1,"method":"Page.navigate","params":{"url":"' + $TargetUrl.Replace('\', '\\').Replace('"', '\"') + '"}}'
+      Write-Trace ("[cdp] navigate target id=" + $hit.id + " port=" + $port)
+      if (Send-WebSocketJson -WsUrl $wsUrl -Json $json) {
+        return $true
+      }
+    } catch {
+      Write-Trace ("[fail step] Invoke-CdpNavigate port=$port : " + $_.Exception.Message)
+    }
+  }
+  return $false
+}
+
+# Win32 P/Invoke type compiled once for Step B.
+$win32Source = @'
+using System;
+using System.Collections.Generic;
+using System.Runtime.InteropServices;
+using System.Text;
+
+public class Win32WindowFinder {
+  public delegate bool EnumWindowsProc(IntPtr hWnd, IntPtr lParam);
+  [DllImport("user32.dll")] public static extern bool EnumWindows(EnumWindowsProc lpEnumFunc, IntPtr lParam);
+  [DllImport("user32.dll")] public static extern bool IsWindowVisible(IntPtr hWnd);
+  [DllImport("user32.dll")] public static extern int GetWindowThreadProcessId(IntPtr hWnd, out int lpdwProcessId);
+  [DllImport("user32.dll", CharSet = CharSet.Unicode)] public static extern int GetWindowText(IntPtr hWnd, StringBuilder lpString, int nMaxCount);
+  [DllImport("user32.dll", CharSet = CharSet.Unicode)] public static extern int GetClassName(IntPtr hWnd, StringBuilder lpString, int nMaxCount);
+  [DllImport("user32.dll")] public static extern bool SetForegroundWindow(IntPtr hWnd);
+  [DllImport("user32.dll")] public static extern bool ShowWindow(IntPtr hWnd, int nCmdShow);
+  [DllImport("user32.dll")] public static extern bool IsIconic(IntPtr hWnd);
+
+  public static List<IntPtr> Find(string className, HashSet<int> pids, string titleCue, string[] matchTitles) {
+    var result = new List<IntPtr>();
+    EnumWindows(delegate(IntPtr hWnd, IntPtr lParam) {
+      try {
+        if (!IsWindowVisible(hWnd)) return true;
+        var cls = new StringBuilder(256);
+        GetClassName(hWnd, cls, cls.Capacity);
+        if (cls.ToString() != className) return true;
+        int pid;
+        GetWindowThreadProcessId(hWnd, out pid);
+        if (!pids.Contains(pid)) return true;
+        var title = new StringBuilder(512);
+        GetWindowText(hWnd, title, title.Capacity);
+        var t = title.ToString();
+        bool titleOk = false;
+        foreach (var mt in matchTitles) { if (t.Contains(mt)) { titleOk = true; break; } }
+        if (titleOk || t.Contains(titleCue)) result.Add(hWnd);
+      } catch {}
+      return true;
+    }, IntPtr.Zero);
+    return result;
+  }
+
+  public static string GetTitle(IntPtr hWnd) {
+    var sb = new StringBuilder(512);
+    GetWindowText(hWnd, sb, sb.Capacity);
+    return sb.ToString();
+  }
+}
+'@
+
+function Invoke-Win32Focus {
+  # Enumerates top-level Chrome windows and focuses the best DSH candidate.
+  # Returns $true if a window was focused (navigation is not possible without
+  # CDP, so this is focus-only by design).
+  param([string]$HostPort, [string[]]$MatchTitles)
+  try {
+    if (-not ("Win32WindowFinder" -as [type])) {
+      Add-Type -TypeDefinition $win32Source -Language CSharp
+    }
+    $chromePids = New-Object System.Collections.Generic.HashSet[int]
+    $procs = Get-Process chrome -ErrorAction SilentlyContinue
+    foreach ($p in $procs) { [void]$chromePids.Add($p.Id) }
+
+    $wins = [Win32WindowFinder]::Find("Chrome_WidgetWin_1", $chromePids, $HostPort, $MatchTitles)
+    if ($wins -eq $null -or $wins.Count -eq 0) {
+      Write-Trace ("[win32] no candidate window for " + $HostPort)
+      return $false
+    }
+    # Prefer a window whose title contains the host:port cue.
+    $best = $null
+    foreach ($w in $wins) {
+      $title = [Win32WindowFinder]::GetTitle($w)
+      if ($title.Contains($HostPort)) { $best = $w; break }
+    }
+    if ($best -eq $null) { $best = $wins[0] }
+
+    if ([Win32WindowFinder]::IsIconic($best)) {
+      [void][Win32WindowFinder]::ShowWindow($best, 9)  # SW_RESTORE
+    }
+    $ok = [Win32WindowFinder]::SetForegroundWindow($best)
+    if (-not $ok) {
+      [void][Win32WindowFinder]::ShowWindow($best, 5)  # SW_SHOW
+      $ok = [Win32WindowFinder]::SetForegroundWindow($best)
+    }
+    Write-Trace ("[win32] focused hwnd=" + $best.ToString() + " ok=" + $ok)
+    return $true
+  } catch {
+    Write-Trace ("[fail step] Invoke-Win32Focus: " + $_.Exception.Message)
+    return $false
+  }
+}
+
+# ---------------------------------------------------------------------------
+# Main
+# ---------------------------------------------------------------------------
+function Main {
+  Write-Trace ("[start] args='" + $ProtocolArgs + "'")
+
+  # Space-separated args arrive as one token, but rejoin defensively.
+  $raw = $ProtocolArgs
+  if ($args -and $args.Count -gt 0) {
+    $raw = (@($ProtocolArgs) + $args) -join " "
+  }
+
+  # Strip the scheme prefix and parse the query.
+  $stripped = $raw
+  if ($stripped -match "^dshnotify://") {
+    $stripped = $stripped.Substring("dshnotify://".Length)
+  } elseif ($stripped -match "^dshnotify:") {
+    $stripped = $stripped.Substring("dshnotify:".Length)
+  }
+  $targetUrl = ""
+  if ($stripped -match "\?u=") {
+    $encoded = $stripped -split "\?u=" | Select-Object -Last 1
+    try { $targetUrl = [uri]::UnescapeDataString($encoded) } catch { $targetUrl = $encoded }
+  }
+  if ([string]::IsNullOrWhiteSpace($targetUrl)) {
+    Write-Trace ("[fail] no targetUrl parsed from '" + $raw + "'")
+    return $false
+  }
+
+  # Origin info from the decoded URL (best-effort).
+  $HostPort = ""
+  try {
+    $parsed = [Uri]$targetUrl
+    $HostPort = "$($parsed.Host):$($parsed.Port)"
+  } catch {
+    Write-Trace ("[warn] could not parse targetUrl as Uri: " + $targetUrl)
+  }
+  $MatchTitles = @("DeepSeek Harness")
+
+  $handled = $false
+
+  # Step A: CDP navigate (best-effort, silent on any failure).
+  try {
+    if ($HostPort -ne "") {
+      if (Invoke-CdpNavigate -TargetUrl $targetUrl -HostPort $HostPort) {
+        Write-Trace ("[ok] Step A CDP navigate succeeded for " + $HostPort)
+        $handled = $true
+      } else {
+        Write-Trace ("[step] Step A skipped/failed")
+      }
+    }
+  } catch {
+    Write-Trace ("[fail step] Step A: " + $_.Exception.Message)
+  }
+
+  # Step B: Win32 focus (best-effort, focus-only without CDP navigation).
+  if (-not $handled) {
+    try {
+      if (Invoke-Win32Focus -HostPort $HostPort -MatchTitles $MatchTitles) {
+        Write-Trace ("[ok] Step B focused existing DSH window (focus-only)")
+        $handled = $true
+      } else {
+        Write-Trace ("[step] Step B no candidate")
+      }
+    } catch {
+      Write-Trace ("[fail step] Step B: " + $_.Exception.Message)
+    }
+  }
+
+  # Step C: fallback open in default browser. Always runs if A and B failed.
+  if (-not $handled) {
+    try {
+      Write-Trace ("[fallback] Start-Process " + $targetUrl)
+      Start-Process $targetUrl
+      $handled = $true
+    } catch {
+      Write-Trace ("[fail step] Step C: " + $_.Exception.Message)
+    }
+  }
+
+  return $handled
+}
+
+$success = $false
+try {
+  $success = Main
+} catch {
+  Write-Trace ("[fatal] " + $_.Exception.Message)
+  # Last resort: try to open the URL directly.
+  try {
+    $m = $ProtocolArgs -match "u=([^&\s]+)"
+    if ($m) {
+      $u = [uri]::UnescapeDataString($Matches[1])
+      Start-Process $u
+      $success = $true
+    }
+  } catch {}
+}
+
+if ($success) {
+  exit 0
+} else {
+  exit 1
+}

package/lib/notify.ps1

@@ -1,35 +1,99 @@
-# dsh-notify toast sender (Windows PowerShell 5.1 required: WinRT projection).
-# Parameters are bound by PowerShell from the process argument vector, so this
-# file stays ASCII-only and needs no BOM.
-param(
-  [Parameter(Mandatory = $true)][string]$Title,
-  [Parameter(Mandatory = $true)][string]$Body,
-  [string]$Aumid = "DeepSeekHarness.Notify",
-  [string]$AppName = "DeepSeek Harness"
-)
-$ErrorActionPreference = "Stop"
-
-# Register the AppUserModelId once under HKCU so Win32 toasts are allowed.
-# Best effort: if the write is denied we still try to show the toast below.
-$key = "HKCU:\SOFTWARE\Classes\AppUserModelId\$Aumid"
-try {
-  if (-not (Test-Path $key)) {
-    New-Item -Path $key -Force | Out-Null
-    New-ItemProperty -Path $key -Name DisplayName -Value $AppName -PropertyType String -Force | Out-Null
-  }
-} catch {
-  Write-Warning ("AUMID registration failed: " + $_.Exception.Message)
-}
-
-# Project the WinRT toast API (supported natively by .NET Framework).
-[Windows.UI.Notifications.ToastNotificationManager, Windows.UI.Notifications, ContentType = WindowsRuntime] | Out-Null
-
-$xml = [Windows.UI.Notifications.ToastNotificationManager]::GetTemplateContent([Windows.UI.Notifications.ToastTemplateType]::ToastText02)
-$texts = $xml.GetElementsByTagName("text")
-$texts.Item(0).AppendChild($xml.CreateTextNode($Title)) | Out-Null
-$texts.Item(1).AppendChild($xml.CreateTextNode($Body)) | Out-Null
-
-$toast = New-Object Windows.UI.Notifications.ToastNotification($xml)
-$notifier = [Windows.UI.Notifications.ToastNotificationManager]::CreateToastNotifier($Aumid)
-$notifier.Show($toast)
-Write-Output "toast shown"
+# dsh-notify toast sender (Windows PowerShell 5.1 required: WinRT projection).
+# Parameters are bound by PowerShell from the process argument vector, so this
+# file stays ASCII-only and needs no BOM.
+param(
+  [Parameter(Mandatory = $true)][string]$Title,
+  [Parameter(Mandatory = $true)][string]$Body,
+  [string]$Aumid = "DeepSeekHarness.Notify",
+  [string]$AppName = "DeepSeek Harness",
+  [string]$Url = "",
+  [int]$LaunchProtocol = 1   # 1 = route via dshnotify:// launcher (prefer existing); 0 = open targetUrl directly
+)
+$ErrorActionPreference = "Stop"
+
+# Register the AppUserModelId once under HKCU so Win32 toasts are allowed.
+# Best effort: if the write is denied we still try to show the toast below.
+$key = "HKCU:\SOFTWARE\Classes\AppUserModelId\$Aumid"
+try {
+  if (-not (Test-Path $key)) {
+    New-Item -Path $key -Force | Out-Null
+    New-ItemProperty -Path $key -Name DisplayName -Value $AppName -PropertyType String -Force | Out-Null
+  }
+} catch {
+  Write-Warning ("AUMID registration failed: " + $_.Exception.Message)
+}
+
+# Project the WinRT toast API (supported natively by .NET Framework).
+[Windows.UI.Notifications.ToastNotificationManager, Windows.UI.Notifications, ContentType = WindowsRuntime] | Out-Null
+
+# Optional clickable activation: when $Url is set we make the toast open a URL
+# on click. Best effort - any failure here degrades gracefully (either to a
+# direct protocol activation or to a plain non-clickable toast) and never
+# aborts the Show below.
+$clickable = $false
+$launchValue = ""
+if ($Url -ne "") {
+  try {
+    if ($LaunchProtocol -eq 1) {
+      # Register the dshnotify:// protocol handler idempotently (best effort).
+      try {
+        $progId = "HKCU:\Software\Classes\dshnotify"
+        if (-not (Test-Path $progId)) {
+          New-Item -Path $progId -Force | Out-Null
+        }
+        # Default (unnamed) value names the protocol.
+        Set-Item -Path $progId -Value 'dshnotify' -Force
+        # "URL Protocol" marker value (presence marks the key as a protocol handler).
+        Set-ItemProperty -Path $progId -Name 'URL Protocol' -Value '' -Force
+        $cmdKey = "HKCU:\Software\Classes\dshnotify\shell\open\command"
+        if (-not (Test-Path $cmdKey)) {
+          New-Item -Path $cmdKey -Force | Out-Null
+        }
+        $launcherPath = Join-Path $PSScriptRoot "launcher.ps1"
+        $command = ('"powershell.exe" -NoProfile -ExecutionPolicy Bypass -WindowStyle Hidden -File "' + $launcherPath + '" "%1"')
+        Set-Item -Path $cmdKey -Value $command -Force
+      } catch {
+        Write-Warning ("dshnotify protocol registration failed, falling back to direct URL: " + $_.Exception.Message)
+        # Fall back to direct-URL activation for this toast.
+        $LaunchProtocol = 0
+      }
+    }
+
+    if ($LaunchProtocol -eq 1) {
+      $launchValue = ("dshnotify://open?u=" + [uri]::EscapeDataString($Url))
+    } else {
+      $launchValue = $Url
+    }
+    $clickable = $true
+  } catch {
+    Write-Warning ("Toast activation setup failed, showing plain toast: " + $_.Exception.Message)
+    $clickable = $false
+  }
+}
+
+$xml = [Windows.UI.Notifications.ToastNotificationManager]::GetTemplateContent([Windows.UI.Notifications.ToastTemplateType]::ToastText02)
+$texts = $xml.GetElementsByTagName("text")
+$texts.Item(0).AppendChild($xml.CreateTextNode($Title)) | Out-Null
+$texts.Item(1).AppendChild($xml.CreateTextNode($Body)) | Out-Null
+
+if ($clickable) {
+  try {
+    $root = $xml.DocumentElement
+    if ($root.Name -eq "toast") {
+      $root.SetAttribute("activationType", "protocol") | Out-Null
+      $root.SetAttribute("duration", "long") | Out-Null
+      $root.SetAttribute("launch", $launchValue) | Out-Null
+      # Optional explicit audio cue.
+      $audio = $xml.CreateElement("audio")
+      $audio.SetAttribute("src", "ms-winsoundevent:Notification.Default") | Out-Null
+      $root.AppendChild($audio) | Out-Null
+    }
+  } catch {
+    Write-Warning ("Failed to attach activation attributes: " + $_.Exception.Message)
+  }
+}
+
+$toast = New-Object Windows.UI.Notifications.ToastNotification($xml)
+$notifier = [Windows.UI.Notifications.ToastNotificationManager]::CreateToastNotifier($Aumid)
+$notifier.Show($toast)
+Write-Output "toast shown"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-notify-windows",
-  "version": "0.6.0",
+  "version": "0.7.3",
   "description": "DeepSeek Harness (DSH) 插件：任务完成 / 等待审批 / 等待回答时发送 Windows 系统桌面通知",
   "repository": {
     "type": "git",