From 162f83de6024bcdd1378a8424072daee4766f178 Mon Sep 17 00:00:00 2001
From: Paul Stretenowich <paul.stretenowich@mcgill.ca>
Date: Fri, 14 Feb 2025 16:43:29 -0500
Subject: [PATCH] Removing lxml and using html5lib to avoid BS4 failing with
 lxml on HPCs

---
 pt_cli/connect.py | 3 +--
 pyproject.toml    | 2 +-
 2 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/pt_cli/connect.py b/pt_cli/connect.py
index f1406e6..3776cbd 100755
--- a/pt_cli/connect.py
+++ b/pt_cli/connect.py
@@ -10,7 +10,6 @@
 
 import requests
 import bs4
-import lxml
 
 logger = logging.getLogger(__name__)
 
@@ -128,7 +127,7 @@ def maybe_json(self, data):
             return loads
         except json.decoder.JSONDecodeError:
             if isinstance(data, str):
-                soup = bs4.BeautifulSoup(data, features="lxml")
+                soup = bs4.BeautifulSoup(data, features="html5lib")
                 if soup.get_text().startswith("----------"):
                     sys.stdout.write(soup.get_text())
                 elif soup.get_text().startswith("Welcome"):
diff --git a/pyproject.toml b/pyproject.toml
index 3f7f7fd..9c270ee 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -21,7 +21,7 @@ dependencies = [
     "pyyaml>=6.0",
     "requests>=2.28",
     "beautifulsoup4>=4.12",
-    "lxml",
+    "html5lib",
     "shtab",
 ]