Added mailboxes as local data sources (#1022)

app/MindWork AI Studio/Tools/ToolCallingSystem/ToolSelectionRules.cs

@@ -1,4 +1,5 @@
 using AIStudio.Provider;
+using AIStudio.Settings.DataModel;
 
 namespace AIStudio.Tools.ToolCallingSystem;
 
@@ -10,33 +11,10 @@ public static class ToolSelectionRules
     public const string READ_WEB_PAGE_TOOL_ID = "read_web_page";
     public const string SEARCH_CONFLUENCE_TOOL_ID = "search_confluence";
     public const string SEMANTIC_SEARCH_TOOL_ID = "semantic_search";
-
-    /// <summary>
-    /// Turns a set of selected tool IDs into the set which actually runs.
-    /// </summary>
-    /// <remarks>
-    /// Removes duplicates and adds the tools another one depends on: Search Confluence only finds
-    /// pages, so it brings Read Web Page along to open them. An added tool keeps its own rules.
-    /// ToolRegistry still drops it when it is switched off or the provider's confidence is too
-    /// low, and Read Web Page reaches a wiki on a private or VPN address only when its host is
-    /// allowed there.<br/><br/>
-    /// It also removes the tools nobody selects. Semantic Search offers itself whenever the data
-    /// sources of a chat call for it, see ToolActivation.CONTEXT; kept in a selection, it would
-    /// appear on the security card of a plugin and in its audit without the selection having any
-    /// say in whether it runs.<br/><br/>
-    /// Every place which shows or stores a selection normalizes it, the tool selection fields
-    /// included. That way a chat, a template, a policy, or an assistant plugin shows the tools
-    /// which will actually run, and the audit of a plugin judges exactly those.
-    /// </remarks>
-    public static HashSet<string> NormalizeSelection(IEnumerable<string> selectedToolIds)
-    {
-        var normalized = selectedToolIds.ToHashSet(StringComparer.Ordinal);
-        if (normalized.Contains(SEARCH_CONFLUENCE_TOOL_ID))
-            normalized.Add(READ_WEB_PAGE_TOOL_ID);
-
-        normalized.Remove(SEMANTIC_SEARCH_TOOL_ID);
-        return normalized;
-    }
+    public const string SEARCH_MAILS_TOOL_ID = "search_mails";
+    public const string READ_MAIL_TOOL_ID = "read_mail";
+    public const string COUNT_MAILS_TOOL_ID = "count_mails";
+    public const string MAILBOXES_COLLECTION_ID = "mailboxes";
 
     public static string GetMaxToolCallsFinalResponseInstruction() => $"The maximum of {MAX_TOOL_CALLS} tool calls has been reached. No more tools are available. Provide the best possible final answer to the user based on the tool results already available.";
 
@@ -75,4 +53,22 @@ public static class ToolSelectionRules
 
     public static bool IsProviderConfidenceAllowed(ConfidenceLevel providerConfidence, ConfidenceLevel minimumToolConfidence) =>
         minimumToolConfidence is ConfidenceLevel.NONE || providerConfidence >= minimumToolConfidence;
+
+    /// <summary>
+    /// Whether a tool may run in a chat with this outbound data restriction.
+    /// </summary>
+    /// <remarks>
+    /// Services configured in AI Studio stay allowed on every level. Below UNRESTRICTED, both
+    /// queries to third parties and addresses the model chooses are kept back, unless the tool
+    /// keeps to the restriction itself. That is why both stricter levels decide alike here: they
+    /// differ only in what such a tool lets through. A level this version does not know is treated
+    /// like a strict one.
+    /// </remarks>
+    /// <param name="restriction">Where the chat may still send data.</param>
+    /// <param name="implementation">The tool.</param>
+    /// <returns>True when the tool may run.</returns>
+    public static bool IsOutboundDataAllowed(OutboundDataRestriction restriction, IToolImplementation implementation) =>
+        restriction is OutboundDataRestriction.UNRESTRICTED ||
+        implementation.OutboundData is ToolOutboundData.NONE or ToolOutboundData.CONFIGURED_SERVICE ||
+        implementation.EnforcesOutboundDataRestriction;
 }