From 6132c7c9fe49e0023e67cd2fe60e4fb4e32af253 Mon Sep 17 00:00:00 2001
From: Faiz Surani <faiz.surani@gmail.com>
Date: Sat, 9 Nov 2024 12:05:37 -0800
Subject: [PATCH] Split off more deps into extras

---
 pyproject.toml   | 12 +++++++-----
 rl/cli/main.py   |  6 +++---
 rl/utils/core.py |  5 ++---
 uv.lock          | 22 ++++++++++------------
 4 files changed, 22 insertions(+), 23 deletions(-)

diff --git a/pyproject.toml b/pyproject.toml
index bd2e820..81f62d6 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -21,18 +21,20 @@ dependencies = [
     "tqdm",
     "ipdb",
     "requests",
-    "regex",
     "python-dotenv",
-    "unidecode",
     "StrEnum",
-    # TODO: Maybe move into an extra?
-    "watchdog",
     "pydantic>=2",
     "typing-extensions",
-    "legal-segmenter @ git+https://github.com/lexeme-dev/legal-segmenter@main",
 ]
 
 [project.optional-dependencies]
+segmenter = [
+    "unidecode",
+    "legal-segmenter @ git+https://github.com/lexeme-dev/legal-segmenter@main",
+]
+devsync = [
+    "watchdog",
+]
 sherlock = [
     "pexpect",
     "pyotp",
diff --git a/rl/cli/main.py b/rl/cli/main.py
index ac0eee0..203b190 100644
--- a/rl/cli/main.py
+++ b/rl/cli/main.py
@@ -3,6 +3,7 @@
 import json
 import os
 import random
+import re
 import subprocess
 import threading
 import time
@@ -13,7 +14,6 @@
 import paramiko
 import pexpect  # type: ignore
 import questionary
-import regex
 import rich
 import rich.progress
 import rich.table
@@ -75,8 +75,8 @@
 _DEFAULT_SSH_TUNNEL_PORT = 5549
 
 
-_MFA_LINE_REGEX = regex.compile(
-    r"\s*(?P<number>\d+)\. Duo Push to XXX-XXX-0199\s*", regex.IGNORECASE
+_MFA_LINE_REGEX = re.compile(
+    r"\s*(?P<number>\d+)\. Duo Push to XXX-XXX-0199\s*", re.IGNORECASE
 )
 
 # endregion
diff --git a/rl/utils/core.py b/rl/utils/core.py
index 3beb952..c165f23 100644
--- a/rl/utils/core.py
+++ b/rl/utils/core.py
@@ -1,9 +1,8 @@
 import collections
+import re
 from collections.abc import Callable, Iterable
 from typing import Any, TypeVar
 
-import regex
-
 import rl.utils.io
 
 K = TypeVar("K")
@@ -31,7 +30,7 @@ def dig(d: dict, *keys: str) -> Any:
     return d
 
 
-def safe_extract(pattern: regex.Pattern, text: str, key: str) -> str | None:
+def safe_extract(pattern: re.Pattern, text: str, key: str) -> str | None:
     match = pattern.search(text)
     if match:
         return match.group(key)
diff --git a/uv.lock b/uv.lock
index 368c39d..bfcc3c5 100644
--- a/uv.lock
+++ b/uv.lock
@@ -2952,23 +2952,20 @@ source = { editable = "." }
 dependencies = [
     { name = "click" },
     { name = "ipdb" },
-    { name = "legal-segmenter" },
-    { name = "paramiko" },
     { name = "pydantic" },
     { name = "python-dotenv" },
-    { name = "questionary" },
-    { name = "regex" },
     { name = "requests" },
     { name = "rich" },
     { name = "rich-click" },
     { name = "strenum" },
     { name = "tqdm" },
     { name = "typing-extensions" },
-    { name = "unidecode" },
-    { name = "watchdog" },
 ]
 
 [package.optional-dependencies]
+devsync = [
+    { name = "watchdog" },
+]
 llm = [
     { name = "anthropic" },
     { name = "google-generativeai" },
@@ -2994,6 +2991,10 @@ plot = [
     { name = "matplotlib" },
     { name = "pandas" },
 ]
+segmenter = [
+    { name = "legal-segmenter" },
+    { name = "unidecode" },
+]
 sherlock = [
     { name = "appdirs" },
     { name = "paramiko" },
@@ -3026,13 +3027,12 @@ requires-dist = [
     { name = "google-generativeai", marker = "extra == 'llm'" },
     { name = "huggingface-hub", marker = "extra == 'local-llm'" },
     { name = "ipdb" },
-    { name = "legal-segmenter", git = "https://github.com/lexeme-dev/legal-segmenter?rev=main" },
+    { name = "legal-segmenter", marker = "extra == 'segmenter'", git = "https://github.com/lexeme-dev/legal-segmenter?rev=main" },
     { name = "matplotlib", marker = "extra == 'plot'" },
     { name = "modal", marker = "extra == 'llm'" },
     { name = "openai", marker = "extra == 'llm'" },
     { name = "pandas", marker = "extra == 'local-llm'" },
     { name = "pandas", marker = "extra == 'plot'" },
-    { name = "paramiko", specifier = ">=3.4.1" },
     { name = "paramiko", marker = "extra == 'sherlock'" },
     { name = "peft", marker = "extra == 'local-llm'" },
     { name = "pexpect", marker = "extra == 'sherlock'" },
@@ -3041,9 +3041,7 @@ requires-dist = [
     { name = "pyotp", marker = "extra == 'sherlock'" },
     { name = "pyspark", marker = "extra == 'spark'" },
     { name = "python-dotenv" },
-    { name = "questionary", specifier = ">=1.10.0" },
     { name = "questionary", marker = "extra == 'sherlock'" },
-    { name = "regex" },
     { name = "requests" },
     { name = "rich" },
     { name = "rich-click" },
@@ -3054,10 +3052,10 @@ requires-dist = [
     { name = "transformers", marker = "extra == 'local-llm'", specifier = ">=4.43" },
     { name = "trl", marker = "extra == 'local-llm'" },
     { name = "typing-extensions" },
-    { name = "unidecode" },
+    { name = "unidecode", marker = "extra == 'segmenter'" },
     { name = "vllm", marker = "platform_system == 'Linux' and extra == 'local-llm'", specifier = "~=0.5.3.post1" },
     { name = "wandb", marker = "extra == 'local-llm'" },
-    { name = "watchdog" },
+    { name = "watchdog", marker = "extra == 'devsync'" },
 ]
 
 [package.metadata.requires-dev]