Fix: strip UTF-8 BOM from .env files

[h1whelan]

Summary

Fixes #637

When a .env file is saved with a UTF-8 BOM (Byte Order Mark), the BOM character (\ufeff) is prepended to the first variable name, making it silently inaccessible via its intended key. This is a common issue with JetBrains IDEs on Windows, which often save files with BOM encoding by default.

Before:

dotenv_values("file_with_bom.env")
# {'\ufeffFIRST': 'value1', 'SECOND': 'value2'}  ← first key is corrupted

After:

dotenv_values("file_with_bom.env")
# {'FIRST': 'value1', 'SECOND': 'value2'}  ← all keys are correct

Changes

• src/dotenv/parser.py: Strip the UTF-8 BOM (\ufeff) from the beginning of the stream content in Reader.__init__ using str.removeprefix().
• tests/test_parser.py: Added two test cases for BOM handling — single variable and multiple variables with BOM prefix.

Test plan

• All 45 parser tests pass (43 existing + 2 new)
• All 114 main tests pass
• Verified fix with actual BOM-encoded .env file
• Verified no regression for files without BOM

🤖 Generated with Claude Code