Merge pull request #581 from paul-gauthier/switch-model

New /model command to switch models, /models to search them

aider/coders/base_coder.py

@@ -66,6 +66,7 @@ class Coder:
         main_model=None,
         edit_format=None,
         io=None,
+        from_coder=None,
         **kwargs,
     ):
         from . import EditBlockCoder, UnifiedDiffCoder, WholeFileCoder
@@ -76,15 +77,42 @@ class Coder:
         if edit_format is None:
             edit_format = main_model.edit_format
 
+        if from_coder:
+            use_kwargs = dict(from_coder.original_kwargs)  # copy orig kwargs
+
+            # If the edit format changes, we can't leave old ASSISTANT
+            # messages in the chat history. The old edit format will
+            # confused the new LLM. It may try and imitate it, disobeying
+            # the system prompt.
+            done_messages = from_coder.done_messages
+            if edit_format != from_coder.edit_format and done_messages:
+                done_messages = from_coder.summarizer.summarize_all(done_messages)
+
+            # Bring along context from the old Coder
+            update = dict(
+                fnames=from_coder.get_inchat_relative_files(),
+                done_messages=done_messages,
+                cur_messages=from_coder.cur_messages,
+            )
+
+            use_kwargs.update(update)  # override to complete the switch
+            use_kwargs.update(kwargs)  # override passed kwargs
+
+            kwargs = use_kwargs
+
         if edit_format == "diff":
-            return EditBlockCoder(main_model, io, **kwargs)
+            res = EditBlockCoder(main_model, io, **kwargs)
         elif edit_format == "whole":
-            return WholeFileCoder(main_model, io, **kwargs)
+            res = WholeFileCoder(main_model, io, **kwargs)
         elif edit_format == "udiff":
-            return UnifiedDiffCoder(main_model, io, **kwargs)
+            res = UnifiedDiffCoder(main_model, io, **kwargs)
         else:
             raise ValueError(f"Unknown edit format {edit_format}")
 
+        res.original_kwargs = dict(kwargs)
+
+        return res
+
     def get_announcements(self):
         lines = []
         lines.append(f"Aider v{__version__}")
@@ -153,6 +181,8 @@ class Coder:
         use_git=True,
         voice_language=None,
         aider_ignore_file=None,
+        cur_messages=None,
+        done_messages=None,
     ):
         if not fnames:
             fnames = []
@@ -166,8 +196,16 @@ class Coder:
 
         self.verbose = verbose
         self.abs_fnames = set()
-        self.cur_messages = []
+
-        self.done_messages = []
+        if cur_messages:
+            self.cur_messages = cur_messages
+        else:
+            self.cur_messages = []
+
+        if done_messages:
+            self.done_messages = done_messages
+        else:
+            self.done_messages = []
 
         self.io = io
         self.stream = stream

aider/coders/wholefile_prompts.py

@@ -18,14 +18,14 @@ Once you understand the request you MUST:
 You MUST use this *file listing* format:
 
 path/to/filename.js
-{fence[0]}javascript
+{fence[0]}
 // entire file content ...
 // ... goes in between
 {fence[1]}
 
 Every *file listing* MUST use this format:
 - First line: the filename with any originally provided path
-- Second line: opening {fence[0]} including the code language
+- Second line: opening {fence[0]}
 - ... entire content of the file ...
 - Final line: closing {fence[1]}
 

aider/commands.py

@@ -5,15 +5,23 @@ import sys
 from pathlib import Path
 
 import git
+import litellm
 import openai
 from prompt_toolkit.completion import Completion
 
-from aider import prompts, voice
+from aider import models, prompts, voice
 from aider.scrape import Scraper
 from aider.utils import is_image_file
 
 from .dump import dump  # noqa: F401
 
+litellm.suppress_debug_info = True
+
+
+class SwitchModel(Exception):
+    def __init__(self, model):
+        self.model = model
+
 
 class Commands:
     voice = None
@@ -28,6 +36,30 @@ class Commands:
 
         self.voice_language = voice_language
 
+    def cmd_model(self, args):
+        "Switch to a new LLM"
+
+        model_name = args.strip()
+        model = models.Model(model_name)
+        models.sanity_check_models(self.io, model)
+        raise SwitchModel(model)
+
+    def completions_model(self, partial):
+        models = litellm.model_cost.keys()
+        for model in models:
+            if partial.lower() in model.lower():
+                yield Completion(model, start_position=-len(partial))
+
+    def cmd_models(self, args):
+        "Search the list of available models"
+
+        args = args.strip()
+
+        if args:
+            models.print_matching_models(self.io, args)
+        else:
+            self.io.tool_output("Please provide a partial model name to search for.")
+
     def cmd_web(self, args):
         "Use headless selenium to scrape a webpage and add the content to the chat"
         url = args.strip()
@@ -99,6 +131,8 @@ class Commands:
         matching_commands, first_word, rest_inp = res
         if len(matching_commands) == 1:
             return self.do_run(matching_commands[0][1:], rest_inp)
+        elif first_word in matching_commands:
+            return self.do_run(first_word[1:], rest_inp)
         elif len(matching_commands) > 1:
             self.io.tool_error(f"Ambiguous command: {', '.join(matching_commands)}")
         else:

aider/main.py

@@ -10,6 +10,7 @@ from streamlit.web import cli
 from aider import __version__, models
 from aider.args import get_parser
 from aider.coders import Coder
+from aider.commands import SwitchModel
 from aider.io import InputOutput
 from aider.repo import GitRepo
 from aider.versioncheck import check_version
@@ -270,17 +271,7 @@ def main(argv=None, input=None, output=None, force_git_root=None, return_coder=F
         return 0 if not update_available else 1
 
     if args.models:
-        matches = models.fuzzy_match_models(args.models)
+        models.print_matching_models(io, args.models)
-        if matches:
-            io.tool_output(f'Models which match "{args.models}":')
-            for model in matches:
-                fq, m = model
-                if fq == m:
-                    io.tool_output(f"- {m}")
-                else:
-                    io.tool_output(f"- {m} ({fq})")
-        else:
-            io.tool_output(f'No models match "{args.models}".')
         return 0
 
     if args.git:
@@ -337,6 +328,7 @@ def main(argv=None, input=None, output=None, force_git_root=None, return_coder=F
             voice_language=args.voice_language,
             aider_ignore_file=args.aiderignore,
         )
+
     except ValueError as err:
         io.tool_error(str(err))
         return 1
@@ -398,7 +390,13 @@ def main(argv=None, input=None, output=None, force_git_root=None, return_coder=F
             return 1
         return
 
-    coder.run()
+    while True:
+        try:
+            coder.run()
+            return
+        except SwitchModel as switch:
+            coder = Coder.create(main_model=switch.model, io=io, from_coder=coder)
+            coder.show_announcements()
 
 
 if __name__ == "__main__":

aider/models.py

@@ -409,6 +409,20 @@ def fuzzy_match_models(name):
     return list(zip(matching_models, matching_models))
 
 
+def print_matching_models(io, search):
+    matches = fuzzy_match_models(search)
+    if matches:
+        io.tool_output(f'Models which match "{search}":')
+        for model in matches:
+            fq, m = model
+            if fq == m:
+                io.tool_output(f"- {m}")
+            else:
+                io.tool_output(f"- {m} ({fq})")
+    else:
+        io.tool_output(f'No models match "{search}".')
+
+
 def main():
     if len(sys.argv) != 2:
         print("Usage: python models.py <model_name>")