Improve handling of llama-server start problems

At first launch of codesigned and notarized app llama-server seems to start very slowly. Let's handle this situation more gracefully, try waiting and inform user of the status

Author: abra-code

AIChat.app/Contents/Resources/Scripts/aichat.init.sh

@@ -4,6 +4,8 @@ source "$OMC_APP_BUNDLE_PATH/Contents/Resources/Scripts/aichat.library.sh"
 
 echo "[$(/usr/bin/basename "$0")]"
 
+webui_dir_path="$OMC_APP_BUNDLE_PATH/Contents/Resources/WebUI"
+
 register_started_server()
 {
 	local host_pid="$1"
@@ -90,7 +92,9 @@ stop_orphaned_servers()
 
 wait_for_server_response()
 {
+	local result=0
 	local seconds_count=0
+	
 	while true; do
 		# wait until we get a response from server
 		/usr/bin/curl "http://localhost:$port_num/slots" > /dev/null 2>&1
@@ -100,24 +104,38 @@ wait_for_server_response()
 			break
 		fi
 
-		# or 5 seconds pass
+		# or 20 seconds pass
 		seconds_count=$((seconds_count + 1))
-		if [ "$seconds_count" -ge 5 ]; then	
-			echo "warning: timed out after $seconds_count seconds while waiting for server response"
+		if [ "$seconds_count" -ge 20 ]; then
+			local message=$(echo "Timed out after $seconds_count seconds while waiting for llama-server response.\n\nPlease try again")
+			echo "$message"
+			"$alert" --level "stop" --title "$APPLET_NAME" --ok "OK" "$message"
+			result=13
 			break
+		elif [ "$seconds_count" -eq 5 ]; then
+			echo "$dialog $OMC_NIB_DLG_GUID 2 file://${webui_dir_path}/start_slow.html"
+			"$dialog" "$OMC_NIB_DLG_GUID" 2 "file://${webui_dir_path}/start_slow.html"
 		fi
 
 		sleep 1
 	done
+	
+	return "$result"
 }
 
 report_server_launch_failure()
 {
 	local message=$(echo "llama-server failed to launch! \n\nVerify if the selected large language model is supported by llama.cpp engine.")
 	echo "$message"
 	"$alert" --level "stop" --title "$APPLET_NAME" --ok "OK" "$message"
+	return 11
 }
 
+
+echo "$dialog $OMC_NIB_DLG_GUID 2 file://${webui_dir_path}/start.html"
+"$dialog" "$OMC_NIB_DLG_GUID" 2 "file://${webui_dir_path}/start.html"
+echo ""
+
 echo "OMC_CURRENT_COMMAND_GUID: ${OMC_CURRENT_COMMAND_GUID}"
 echo "OMC_NIB_DLG_GUID: ${OMC_NIB_DLG_GUID}"
 echo "OMC_FRONT_PROCESS_ID: ${OMC_FRONT_PROCESS_ID}"
@@ -148,13 +166,14 @@ stop_orphaned_servers
 
 # /usr/bin/curl "http://localhost:$port_num/slots" > /dev/null 2>&1
 
+server_result=0
+
 echo "Check if the required llama-server with selected model is already running"
 running_process=$(/bin/ps -U $USER | /usr/bin/grep -E "$OMC_APP_BUNDLE_PATH/Contents/Support/Llama.cpp/llama-server" | /usr/bin/grep -E "$port_num" | /usr/bin/grep -E "$AICHAT_MODEL_PATH")
 
 if [ $? != 0 ]; then
 	echo "Starting llama-server..."
 	# start the server
-	webui_dir_path="$OMC_APP_BUNDLE_PATH/Contents/Resources/WebUI"
 	"$OMC_APP_BUNDLE_PATH/Contents/Support/Llama.cpp/llama-server" --host 127.0.0.1 --port $port_num --path "$webui_dir_path" --model "$AICHAT_MODEL_PATH" &
 	llama_server_pid=$!
 	if [ "$llama_server_pid" != "" ]; then
@@ -164,23 +183,33 @@ if [ $? != 0 ]; then
 		if [ "$server_process_exists" != 0 ]; then
 			# server exited. most likely something wrong with selected gguf model
 			report_server_launch_failure
+			server_result=$?
 		else
 			# server process running, check if it is responsive
 			wait_for_server_response
+			server_result=$?
 
 			echo "Register server with pid $llama_server_pid"
 			register_started_server "${OMC_FRONT_PROCESS_ID}" "${llama_server_pid}" "$AICHAT_MODEL_PATH"	
 		fi
 	else
 		report_server_launch_failure
+		server_result=$?
 	fi
 
 else
 	llama_server_pid=$(echo "$running_process" | /usr/bin/grep -E --only-matching '^ *[[:digit:]]+ ' | /usr/bin/tr -d ' ')
 	echo "llama-server already running with pid: $running_process"
+	server_result=0
 fi
 
-echo ""
-echo "$dialog $OMC_NIB_DLG_GUID 2 http://localhost:$port_num/"
-"$dialog" "$OMC_NIB_DLG_GUID" 2 "http://localhost:$port_num/"
+if [ "$server_result" = 0 ]; then
+	echo ""
+	echo "$dialog $OMC_NIB_DLG_GUID 2 http://localhost:$port_num/"
+	"$dialog" "$OMC_NIB_DLG_GUID" 2 "http://localhost:$port_num/"
+else
+	echo ""
+	echo "$dialog $OMC_NIB_DLG_GUID 2 file://${webui_dir_path}/start_error.html?port=$port_num"
+	"$dialog" "$OMC_NIB_DLG_GUID" 2 "file://${webui_dir_path}/start_error.html?port=$port_num"
+fi
 

AIChat.app/Contents/Resources/WebUI/start.html

@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title>Starting Local Llama Server</title>    
+<body>
+	<div style="display: block; margin-top: 200px; text-align: center; height: 300vh; align-items: center; justify-content: center;">
+        <p>Starting local Llama server...</p>
+    </div>
+</body>
+</html>

AIChat.app/Contents/Resources/WebUI/start_error.html

@@ -0,0 +1,28 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title>Local Llama Server Error</title>    
+    <script>
+        window.onload = function () {
+            const currentURL = window.location.href;
+            const url = new URL(currentURL);
+            const params = new URLSearchParams(url.search);
+            port = params.get('port');
+
+            // Check if port exists and construct the URL
+            if (!port) {
+            	port = 8080
+            }
+			const constructedURL = `http://localhost:${port}/`;
+			document.getElementById('server_link').innerHTML = `<a href="${constructedURL}">Try loading anyway</a>`;
+        };
+    </script>
+<body>
+	<div style="display: block; margin-top: 200px; text-align: center; height: 300vh; align-items: center; justify-content: center;">
+        <h3>Local Llama server error</h3>
+        <p>There was a problem starting local Llama server</p>
+    </div>
+    <div id="server_link" style="display: block; margin-top: 200px; text-align: center; height: 300vh; align-items: center; justify-content: center;"></div>
+</body>
+</html>

AIChat.app/Contents/Resources/WebUI/start_slow.html

@@ -0,0 +1,13 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title>Starting Local Llama Server</title>    
+<body>
+	<div style="display: block; margin-top: 200px; text-align: center; height: 300vh; align-items: center; justify-content: center;">
+        <p>It is taking longer than expected to start the local Llama server.</br></p>
+        <p>This may happen on the first launch.</br></p>
+        <p>Hang on...</br></p>
+    </div>
+</body>
+</html>