Use eventlet instead of gevent-websocket

aiserver.py

@@ -6,7 +6,12 @@
 #==================================================================#
 
 # External packages
+import eventlet
+eventlet.monkey_patch()
 import os
+os.environ['EVENTLET_THREADPOOL_SIZE'] = '1'
+from eventlet import tpool
+
 from os import path, getcwd
 import re
 import tkinter as tk
@@ -559,7 +564,7 @@ from flask import Flask, render_template, Response, request
 from flask_socketio import SocketIO, emit
 app = Flask(__name__)
 app.config['SECRET KEY'] = 'secret!'
-socketio = SocketIO(app)
+socketio = SocketIO(app, async_method="eventlet")
 print("{0}OK!{1}".format(colors.GREEN, colors.END))
 
 # Start transformers and create pipeline
@@ -990,8 +995,8 @@ def load_lua_scripts():
 
     try:
         vars.lua_koboldbridge.obliterate_multiverse()
-        vars.lua_koboldbridge.load_corescript("default.lua")
+        tpool.execute(vars.lua_koboldbridge.load_corescript, "default.lua")
-        vars.lua_koboldbridge.load_userscripts(filenames, modulenames, descriptions)
+        tpool.execute(vars.lua_koboldbridge.load_userscripts, filenames, modulenames, descriptions)
     except lupa.LuaError as e:
         vars.lua_koboldbridge.obliterate_multiverse()
         if(vars.serverstarted):
@@ -1293,7 +1298,7 @@ def lua_is_custommodel():
 def execute_inmod():
     vars.lua_logname = ...
     try:
-        vars.lua_koboldbridge.execute_inmod()
+        tpool.execute(vars.lua_koboldbridge.execute_inmod)
     except lupa.LuaError as e:
         vars.lua_koboldbridge.obliterate_multiverse()
         emit('from_server', {'cmd': 'errmsg', 'data': 'Lua script error, please check console.'}, broadcast=True)
@@ -1307,7 +1312,7 @@ def execute_genmod():
 
 def execute_outmod():
     try:
-        vars.lua_koboldbridge.execute_outmod()
+        tpool.execute(vars.lua_koboldbridge.execute_outmod)
     except lupa.LuaError as e:
         vars.lua_koboldbridge.obliterate_multiverse()
         emit('from_server', {'cmd': 'errmsg', 'data': 'Lua script error, please check console.'}, broadcast=True)
@@ -1315,6 +1320,9 @@ def execute_outmod():
         print("{0}{1}{2}".format(colors.RED, str(e).replace("\033", ""), colors.END), file=sys.stderr)
         print("{0}{1}{2}".format(colors.YELLOW, "Lua engine stopped; please open 'Userscripts' and press Load to reinitialize scripts.", colors.END), file=sys.stderr)
         set_aibusy(0)
+    if(vars.lua_koboldbridge.resend_settings_required):
+        vars.lua_koboldbridge.resend_settings_required = False
+        lua_resend_settings()
 
 #==================================================================#
 #  Lua runtime startup
@@ -1348,7 +1356,6 @@ bridged = {
     "has_setting": lua_has_setting,
     "get_setting": lua_get_setting,
     "set_setting": lua_set_setting,
-    "resend_settings": lua_resend_settings,
     "set_chunk": lua_set_chunk,
     "get_modeltype": lua_get_modeltype,
     "get_modelbackend": lua_get_modelbackend,
@@ -2157,23 +2164,8 @@ def calcsubmit(txt):
 #==================================================================#
 # Send text to generator and deal with output
 #==================================================================#
-def generate(txt, minimum, maximum, found_entries=None):    
-    if(found_entries is None):
-        found_entries = set()
-    found_entries = tuple(found_entries.copy() for _ in range(vars.numseqs))
 
-    print("{0}Min:{1}, Max:{2}, Txt:{3}{4}".format(colors.YELLOW, minimum, maximum, txt, colors.END))
+def _generate(txt, minimum, maximum, found_entries):
-    
-    # Store context in memory to use it for comparison with generated content
-    vars.lastctx = txt
-    
-    # Clear CUDA cache if using GPU
-    if(vars.hascuda and (vars.usegpu or vars.breakmodel)):
-        gc.collect()
-        torch.cuda.empty_cache()
-    
-    # Submit input text to generator
-    try:
     gen_in = tokenizer.encode(txt, return_tensors="pt", truncation=True).long()
     if(vars.sp is not None):
         soft_tokens = torch.arange(
@@ -2251,6 +2243,27 @@ def generate(txt, minimum, maximum, found_entries=None):
             model.kai_scanner_head_length = encoded.shape[-1]
             numseqs = 1
     
+    return genout, already_generated
+    
+
+def generate(txt, minimum, maximum, found_entries=None):    
+    if(found_entries is None):
+        found_entries = set()
+    found_entries = tuple(found_entries.copy() for _ in range(vars.numseqs))
+
+    print("{0}Min:{1}, Max:{2}, Txt:{3}{4}".format(colors.YELLOW, minimum, maximum, txt, colors.END))
+
+    # Store context in memory to use it for comparison with generated content
+    vars.lastctx = txt
+
+    # Clear CUDA cache if using GPU
+    if(vars.hascuda and (vars.usegpu or vars.breakmodel)):
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    # Submit input text to generator
+    try:
+        genout, already_generated = tpool.execute(_generate, txt, minimum, maximum, found_entries)
     except Exception as e:
         if(issubclass(type(e), lupa.LuaError)):
             vars.lua_koboldbridge.obliterate_multiverse()
@@ -2448,7 +2461,8 @@ def tpumtjgenerate(txt, minimum, maximum, found_entries=None):
             dtype=np.uint32
         )
 
-        genout = tpu_mtj_backend.infer(
+        genout = tpool.execute(
+            tpu_mtj_backend.infer,
             txt,
             gen_len = maximum-minimum+1,
             temp=vars.temp,
@@ -3813,4 +3827,4 @@ if __name__ == "__main__":
         webbrowser.open_new('http://localhost:5000')
         print("{0}\nServer started!\nYou may now connect with a browser at http://127.0.0.1:5000/{1}".format(colors.GREEN, colors.END))
         vars.serverstarted = True
-        socketio.run(app)
+        socketio.run(app, port=5000)

bridge.lua

@@ -1679,9 +1679,6 @@ return function(_python, _bridged)
                 end
             end
         end
-        if koboldbridge.resend_settings_required then
-            bridged.resend_settings()
-        end
         koboldbridge.userstate = nil
         return r
     end

environments/finetuneanon.yml

@@ -10,11 +10,11 @@ dependencies:
   - cudatoolkit=11.1
   - tensorflow-gpu
   - python=3.8.*
-  - gevent-websocket
+  - eventlet
   - pip
   - git
   - pip:
     - git+https://github.com/finetuneanon/transformers@gpt-neo-localattention3-rp-b
     - flask-cloudflared
     - flask-ngrok
-    - lupa
+    - lupa==1.10

environments/huggingface.yml

@@ -11,10 +11,10 @@ dependencies:
   - python=3.8.*
   - cudatoolkit=11.1
   - transformers
-  - gevent-websocket
+  - eventlet
   - pip
   - git
   - pip:
     - flask-cloudflared
     - flask-ngrok
-    - lupa
+    - lupa==1.10

environments/rocm-finetune.yml

@@ -6,7 +6,7 @@ dependencies:
   - colorama
   - flask-socketio
   - python=3.8.*
-  - gevent-websocket
+  - eventlet
   - pip
   - git
   - pip:
@@ -16,4 +16,4 @@ dependencies:
     - flask-cloudflared
     - git+https://github.com/finetuneanon/transformers@gpt-neo-localattention3-rp-b
     - flask-ngrok
-    - lupa
+    - lupa==1.10

environments/rocm.yml

@@ -8,7 +8,7 @@ dependencies:
   - colorama
   - flask-socketio
   - python=3.8.*
-  - gevent-websocket
+  - eventlet
   - pip
   - git
   - pip:
@@ -17,4 +17,4 @@ dependencies:
     - torchvision==0.11.1
     - flask-cloudflared
     - flask-ngrok
-    - lupa
+    - lupa==1.10

static/application.js

@@ -1619,7 +1619,7 @@ $(document).ready(function(){
 	seqselcontents    = $("#seqselcontents");
 
 	// Connect to SocketIO server
-	socket = io.connect(window.document.origin);
+	socket = io.connect(window.document.origin, {transports: ['websocket', 'polling']});
 	
 	socket.on('from_server', function(msg) {
 		if(msg.cmd == "connected") {

templates/index.html

@@ -7,7 +7,7 @@
 	<script src="static/jquery-3.6.0.min.js"></script>
 	<script src="static/jquery-ui.sortable.min.js"></script>
 	<script src="static/socket.io.min.js"></script>
-	<script src="static/application.js?ver=1.16.4e"></script>
+	<script src="static/application.js?ver=1.16.4f"></script>
 	<script src="static/bootstrap.min.js"></script>
 	<script src="static/bootstrap-toggle.min.js"></script>
 	<script src="static/rangy-core.min.js"></script>