Skip to main content

🌐 lncrawl-translator

Translate entire web novels through one small, stateless API — tuned for Chinese · Korean · Japanese → English.

PyPI Python Docker CI License

A translation service for web-novel metadata (titles, author, synopsis, tags) and HTML chapter content — markup preserved, only human-readable text translated. Run it as a standalone container or embed it as a Python library inside your own app.

Built as the stateless translation engine behind lightnovel-crawler, with switchable engines (free-tier hosted APIs and CPU-friendly local models). Every request can carry a per-novel glossary that gets injected into translations and returned with any new terms — so the caller keeps names and terms consistent across thousands of chapters.


✨ Highlights

🈶 CJK → English first Purpose-tuned for Chinese/Korean/Japanese source text; other directions work best-effort.
🧾 Text and HTML Batch short strings, or translate a whole chapter's HTML with tags kept intact.
📖 Glossary-aware Pass a glossary in, get new terms out — consistent character/place names at book scale.
🔀 Switchable engines OpenAI-compatible LLMs, DeepL, Baidu, and a keyless Bing lane behind one router.
♻️ Resilient routing Client-side rate limiting, retries, cooldowns, and quota-aware failover across lanes.
🎛️ Live config Browser dashboard + CRUD API; edits apply atomically and persist to a sparse YAML overlay.
🧭 Offline detection Unicode-script heuristics + langdetect — no network, no engine quota.
🧩 Standalone or embedded Docker/uvicorn service, or a thread-safe synchronous Python facade.

🚀 Quick start

No config file needed — a curated set of free providers is pre-wired, and the keyless Bing lane is the default, so translation works even with zero keys:

docker compose up -d
curl http://localhost:8184/health   # shows which engines came up

Then open http://localhost:8184/ and paste your provider API keys — the matching engines enable instantly, no restart needed.

📚 See the deployment guide for engine keys, the keyless Bing lane, and API examples · design doc for API & architecture · engine research.


🔌 API

Endpoint What it does
GET / Browser demo & config UI — try translations, watch engine status, manage config in place.
GET /health Liveness / readiness.
GET /engines Configured engines with live status (quota, cooldowns).
POST /detect Local language detection (no engine quota).
POST /translate/text Batched short strings (titles, tags, synopsis).
POST /translate/html One chapter per call — glossary in, new terms out.
GET /config + CRUD on /providers, /engines, /routing Runtime config; changes apply atomically and persist to config.yml.

⚠️ The API is unauthenticated by design — run it on localhost or a private network only (see the deployment guide).


📦 Use as a library

pip install lncrawl-translator   # Python 3.9+

The embedded service is a thread-safe, synchronous facade that runs the same engine router on its own event loop:

from translator import TranslatorService

service = TranslatorService(config_path="translator.yml")

service.detect(["どこから来ましたか"])                 # local, no quota
response = service.translate_text({
    "texts": ["少年は勇者になった"],
    "target_lang": "en",
    "glossary": {"勇者": "Hero"},
})
print(response.translations, response.engine, response.new_terms)

service.close()  # on shutdown

translate_text / translate_html accept an optional signal (threading.Event) for cooperative cancellation and a timeout in seconds.

Error handling. Failures share a common TranslatorError base, re-exported from the top-level namespace — map them without importing pydantic or reaching into internals:

from translator import TranslatorError, ApiError, AbortedError, InvalidRequestError

Invalid dict payloads raise InvalidRequestError (not pydantic's ValidationError). Need only a language code? from translator import detect_code returns the bare ISO 639-1 code (or None) without ever constructing a service; detect_language is also available.

Mount the dashboard into a host app

app.mount("/translator", service.create_app())

The mounted app shares the service's live config — edits in the dashboard apply to the embedded service immediately. It carries no authentication of its own; the host gates access.

When the host authenticates the mount, pass create_app(auth=True): the OpenAPI then declares HTTPBasic and HTTPBearer schemes (so the docs' Authorize button works), and the dashboard reads an admin token from the page's URL fragment (#token=…) and sends it as a Bearer header. The schemes are only declared here — the host still verifies the credential.


🛠️ Development

uv sync
uv run poe check      # ruff + mypy + pytest
uv run poe dev        # dev server with auto-reload
uv run poe start      # production-style server (uvicorn on :8184)
uv run poe live-test  # smoke-test real engines from config.yml (opt-in)

See AGENTS.md for project decisions and conventions, and CHANGELOG.md for release notes.


📜 License

Licensed under the Apache License 2.0.

Release files for lncrawl-translator 0.2.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for lncrawl-translator 0.2.2
File Size Uploaded
lncrawl_translator-0.2.2.tar.gz 157.9 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for lncrawl-translator 0.2.2
File Interpreter ABI Platform
lncrawl_translator-0.2.2-py3-none-any.whl Python 3 none any Details

Total release size: 348.5 kB

Release files / lncrawl_translator-0.2.2.tar.gz

Download URL lncrawl_translator-0.2.2.tar.gz
Size 157.9 kB
Tags Source
SHA-256 checksum
How to use checksums
f5298a6b311baa00bd49029a5eb803731445f34f40aa70b7d55174a91b89684b
BLAKE2b-256 checksum
How to use checksums
59f2440f5afe89eaa4307ece94067d12074a6965b816830586929c48516e9b67
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jul 31, 2026.

Transparency log

Release files / lncrawl_translator-0.2.2-py3-none-any.whl

Download URL lncrawl_translator-0.2.2-py3-none-any.whl
Size 190.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
bc69c31a29485a0d6d7beca144da50136bffa73218935cb9b40c3b6fb65d9897
BLAKE2b-256 checksum
How to use checksums
a1c4d2594fc7bf99e4eb05b444fffc4a9cf680cac36be43b20f3be6f6d04f33b
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jul 31, 2026.

Transparency log

Release history Release notifications | RSS feed

This release

0.2.2 This release

2 release files

0.2.1

2 release files

0.2.0

2 release files

0.1.3

2 release files

0.1.2

2 release files

0.1.1

2 release files

0.1.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page