llama-stack/llama_toolchain/tools/ipython_tool/code_env_prefix.py
Ashwin Bharambe 7bc7785b0d
API Updates: fleshing out RAG APIs, introduce "llama stack" CLI command (#51)
* add tools to chat completion request

* use templates for generating system prompts

* Moved ToolPromptFormat and jinja templates to llama_models.llama3.api

* <WIP> memory changes

- inlined AgenticSystemInstanceConfig so API feels more ergonomic
- renamed it to AgentConfig, AgentInstance -> Agent
- added a MemoryConfig and `memory` parameter
- added `attachments` to input and `output_attachments` to the response

- some naming changes

* InterleavedTextAttachment -> InterleavedTextMedia, introduce memory tool

* flesh out memory banks API

* agentic loop has a RAG implementation

* faiss provider implementation

* memory client works

* re-work tool definitions, fix FastAPI issues, fix tool regressions

* fix agentic_system utils

* basic RAG seems to work

* small bug fixes for inline attachments

* Refactor custom tool execution utilities

* Bug fix, show memory retrieval steps in EventLogger

* No need for api_key for Remote providers

* add special unicode character ↵ to showcase newlines in model prompt templates

* remove api.endpoints imports

* combine datatypes.py and endpoints.py into api.py

* Attachment / add TTL api

* split batch_inference from inference

* minor import fixes

* use a single impl for ChatFormat.decode_assistant_mesage

* use interleaved_text_media_as_str() utilityt

* Fix api.datatypes imports

* Add blobfile for tiktoken

* Add ToolPromptFormat to ChatFormat.encode_message so that tools are encoded properly

* templates take optional --format={json,function_tag}

* Rag Updates

* Add `api build` subcommand -- WIP

* fix

* build + run image seems to work

* <WIP> adapters

* bunch more work to make adapters work

* api build works for conda now

* ollama remote adapter works

* Several smaller fixes to make adapters work

Also, reorganized the pattern of __init__ inside providers so
configuration can stay lightweight

* llama distribution -> llama stack + containers (WIP)

* All the new CLI for api + stack work

* Make Fireworks and Together into the Adapter format

* Some quick fixes to the CLI behavior to make it consistent

* Updated README phew

* Update cli_reference.md

* llama_toolchain/distribution -> llama_toolchain/core

* Add termcolor

* update paths

* Add a log just for consistency

* chmod +x scripts

* Fix api dependencies not getting added to configuration

* missing import lol

* Delete utils.py; move to agentic system

* Support downloading of URLs for attachments for code interpreter

* Simplify and generalize `llama api build` yay

* Update `llama stack configure` to be very simple also

* Fix stack start

* Allow building an "adhoc" distribution

* Remote `llama api []` subcommands

* Fixes to llama stack commands and update docs

* Update documentation again and add error messages to llama stack start

* llama stack start -> llama stack run

* Change name of build for less confusion

* Add pyopenapi fork to the repository, update RFC assets

* Remove conflicting annotation

* Added a "--raw" option for model template printing

---------

Co-authored-by: Hardik Shah <hjshah@fb.com>
Co-authored-by: Ashwin Bharambe <ashwin@meta.com>
Co-authored-by: Dalton Flanagan <6599399+dltn@users.noreply.github.com>
2024-09-03 22:39:39 -07:00

133 lines
3.4 KiB
Python

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
import errno
# Disabling potentially dangerous functions
import os as _os
from functools import partial
os_funcs_to_disable = [
"kill",
"system",
"putenv",
"remove",
"removedirs",
"rmdir",
"fchdir",
"setuid",
"fork",
"forkpty",
"killpg",
"rename",
"renames",
"truncate",
"replace",
# "unlink", # Commenting as this was blocking matpltlib from rendering plots correctly
"fchmod",
"fchown",
"chmod",
"chown",
"chroot",
"fchdir",
"lchflags",
"lchmod",
"lchown",
"chdir",
]
def call_not_allowed(*args, **kwargs):
raise OSError(errno.EPERM, "Call are not permitted in this environment")
for func_name in os_funcs_to_disable:
if hasattr(_os, func_name):
setattr(_os, func_name, partial(call_not_allowed, _func_name=f"os.{func_name}"))
import shutil as _shutil
for func_name in ["rmtree", "move", "chown"]:
if hasattr(_shutil, func_name):
setattr(
_shutil,
func_name,
partial(call_not_allowed, _func_name=f"shutil.{func_name}"),
)
import subprocess as _subprocess
def popen_not_allowed(*args, **kwargs):
raise _subprocess.CalledProcessError(
-1,
args[0] if args else "unknown",
stderr="subprocess.Popen is not allowed in this environment",
)
_subprocess.Popen = popen_not_allowed
import atexit as _atexit
import builtins as _builtins
import io as _io
import json as _json
import sys as _sys
# NB! The following "unused" imports crucial, make sure not not to remove
# them with linters - they're used in code_execution.py
from contextlib import ( # noqa
contextmanager as _contextmanager,
redirect_stderr as _redirect_stderr,
redirect_stdout as _redirect_stdout,
)
from multiprocessing.connection import Connection as _Connection
# Mangle imports to avoid polluting model execution namespace.
_IO_SINK = _io.StringIO()
_NETWORK_TIMEOUT = 5
_NETWORK_CONNECTIONS = None
def _open_connections():
global _NETWORK_CONNECTIONS
if _NETWORK_CONNECTIONS is not None:
# Ensure connections only opened once.
return _NETWORK_CONNECTIONS
req_w_fd, resp_r_fd = _sys.argv[1], _sys.argv[2]
req_con = _Connection(int(req_w_fd), readable=False)
resp_con = _Connection(int(resp_r_fd), writable=False)
_NETWORK_CONNECTIONS = (req_con, resp_con)
return _NETWORK_CONNECTIONS
_builtins._open_connections = _open_connections
@_atexit.register
def _close_connections():
global _NETWORK_CONNECTIONS
if _NETWORK_CONNECTIONS is None:
return
for con in _NETWORK_CONNECTIONS:
con.close()
del _NETWORK_CONNECTIONS
def _network_call(request):
# NOTE: We communicate with the parent process in json, encoded
# in raw bytes. We do this because native send/recv methods use
# pickle which involves execution of arbitrary code.
_open_connections()
req_con, resp_con = _NETWORK_CONNECTIONS
req_con.send_bytes(_json.dumps(request).encode("utf-8"))
if resp_con.poll(timeout=_NETWORK_TIMEOUT) is None:
raise Exception(f"Network request timed out: {_json.dumps(request)}")
else:
return _json.loads(resp_con.recv_bytes().decode("utf-8"))