Blob leaks a UTF-8 BOM into as_string()

Submission checklist

• This is a bug, not a usage question.
• I added a clear and descriptive title that summarizes this issue.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).
• This is not related to the langchain-community package.
• I posted a self-contained, minimal, reproducible example. A maintainer can copy it and run it AS IS.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Related Issues / PRs

No response

Reproduction Steps / Example Code (Python)

import tempfile
from pathlib import Path
from langchain_core.documents.base import Blob

p = Path(tempfile.mkdtemp()) / "doc.md"
p.write_text("# Heading\n\nbody\n", encoding="utf-8-sig")  # a BOM, as Notepad writes it

print(repr(Blob.from_path(p).as_string()))
# actual:   '\ufeff# Heading\n\nbody\n'
# expected: '# Heading\n\nbody\n'

Error Message and Stack Trace (if applicable)

Description

• I'm loading local files through Blob — a Markdown file that was saved on Windows, so it carries a UTF-8 byte order mark.
• I expect as_string() to return the file's text.
• Instead it returns the text with a zero-width U+FEFF prepended, because Blob decodes with encoding="utf-8", which preserves the BOM.

Notepad, Excel's "CSV UTF-8" export and PowerShell > / Out-File all write a BOM by default, so any file opened and re-saved on Windows arrives with one. The BOM then flows into splitters, embeddings and prompts. For Markdown it lands before the leading #, exactly where a heading parser expects line-start. Blob.from_data() behaves the same way for BOM-prefixed bytes.

This isn't platform-specific. The BOM is in the file's bytes, so it reproduces identically on Linux and macOS. Windows is just where BOM files come from.

Suggested fix: default to utf-8-sig. It decodes plain UTF-8 byte-for-byte identically and strips a BOM when one is present, so files without a BOM are unaffected and passing encoding explicitly keeps its exact meaning.

System Info

System Information

OS: Windows OS Version: 10.0.26200 Python Version: 3.12.7 | packaged by Anaconda, Inc. | (main, Oct 4 2024, 13:17:27) [MSC v.1929 64 bit (AMD64)]

Package Information

langchain_core: 1.6.3 langchain_community: 0.4.2 langsmith: 0.12.4 langchain_classic: 1.0.8 langchain_protocol: 0.0.19 langchain_text_splitters: 1.1.2

Social handles (optional)

www.linkedin.com/in/rehan-ur-rahman-8b1a6722a