From 756bb2d77538d052ef14cecbf59917672629187a Mon Sep 17 00:00:00 2001
From: Ghraven <115199279+Ghraven@users.noreply.github.com>
Date: Wed, 20 May 2026 18:54:42 +0000
Subject: [PATCH] fix: add explicit encoding="utf-8" to text-mode file I/O

Several text-mode open() calls relied on the platform default encoding
(cp1252 on Windows) when reading/writing JSON test-case files, the
JSONL streamlit log, and the bundled index.html. JSON and JSONL are
UTF-8 per RFC 8259, so non-ASCII content could raise UnicodeDecodeError
or be corrupted on non-UTF-8 platforms. Pinning encoding="utf-8" makes
behavior consistent everywhere.
---
 burr/cli/__main__.py           | 4 ++--
 burr/integrations/streamlit.py | 2 +-
 burr/testing/__init__.py       | 2 +-
 burr/tracking/server/run.py    | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/burr/cli/__main__.py b/burr/cli/__main__.py
index 5bbc99dd5..8967602d9 100644
--- a/burr/cli/__main__.py
+++ b/burr/cli/__main__.py
@@ -477,14 +477,14 @@ def create_test_case(
     if target_file_name:
         # if it already exists, load it up and append to it
         if os.path.exists(target_file_name):
-            with open(target_file_name, "r") as f:
+            with open(target_file_name, "r", encoding="utf-8") as f:
                 # assumes it's a list of test cases
                 current_testcases = json.load(f)
             current_testcases.append(tc_json)
         else:
             current_testcases = [tc_json]
         print(f"\nWriting data to file {target_file_name}")
-        with open(target_file_name, "w") as f:
+        with open(target_file_name, "w", encoding="utf-8") as f:
             json.dump(current_testcases, f, indent=2)
     else:
         logger.info(json.dumps(tc_json, indent=2))
diff --git a/burr/integrations/streamlit.py b/burr/integrations/streamlit.py
index 71b63bce0..c3b1df159 100644
--- a/burr/integrations/streamlit.py
+++ b/burr/integrations/streamlit.py
@@ -95,7 +95,7 @@ def load_state_from_log_file(jsonl_log_file: str, app: Application) -> AppState:
     :return: AppState
     """
     out = []
-    for i, line in enumerate(open(jsonl_log_file)):
+    for i, line in enumerate(open(jsonl_log_file, encoding="utf-8")):
         json_line = json.loads(line)
         record = Record(
             state=json_line["state"],
diff --git a/burr/testing/__init__.py b/burr/testing/__init__.py
index f1caa19bf..bd89af77f 100644
--- a/burr/testing/__init__.py
+++ b/burr/testing/__init__.py
@@ -26,7 +26,7 @@
 # and then load it for the test.
 def load_test_cases(file_name: str) -> tuple:
     """Load test cases from a json file."""
-    with open(file_name, "r") as f:
+    with open(file_name, "r", encoding="utf-8") as f:
         json_test_cases = json.load(f)
         test_cases = [(tc.get("input_state"), tc.get("expected_state")) for tc in json_test_cases]
         test_ids = [
diff --git a/burr/tracking/server/run.py b/burr/tracking/server/run.py
index ad253cc6b..4c6b941f3 100644
--- a/burr/tracking/server/run.py
+++ b/burr/tracking/server/run.py
@@ -375,7 +375,7 @@ async def version() -> dict:
         ui_app.mount("/public", StaticFiles(directory=base_asset_directory, html=True), "/public")
 
         # Read index.html once at startup
-        with open(os.path.join(base_asset_directory, "index.html")) as f:
+        with open(os.path.join(base_asset_directory, "index.html"), encoding="utf-8") as f:
             _index_html_template = f.read()
 
         @ui_app.get("/manifest.json")