A Shakespearean nickname generator with source-linked provenance and a browser composing room.
简体中文 · Live demo · Published v1.0.0
Stage: Experimental. Python and browser workflows have automated checks; independent installation, upgrade and sustained user validation remain outstanding. Current source declares 1.1.0; the latest published release is v1.0.0. Source examples below describe the current source, not a promise about the older wheel.
- Generate invented names from Shakespearean ingredients and inspect speaker ties, source locations, formation choices and historical date explanations.
- The current evidence atlas indexes 195 words from Henry IV, Part 1, Henry IV, Part 2 and Henry V, with source hashes and XML identifiers.
- Compare ranked blends and character Markov candidates. Seeded records make runs replayable within one engine/version.
- The browser runs locally without a generation API or key. Inspect the four-stage formation, select alternatives, share replay settings and export JSON.
Evidence attests the ingredients, not the invented name. The scoring and editorial choices are explained in Methodology.
Open the live demo and choose FORGE NAME. Inspect the paper trail and formation steps before copying a name or exporting its record. To serve the same static UI locally, clone the repository and run from its root:
git clone https://github.com/styayur/Nickspeare.git
cd Nickspeare
python -m http.server 8765 --directory docsOpen http://localhost:8765. Python serves the files; nickname generation runs in the browser. The packaged atlas is included, so a corpus download is not needed for ordinary generation.
The browser keeps the current results in memory, without a saved-history store. Export JSON to keep a record; replay links expose the rule, style, suffix, seed, affinity and selected impression in their URL. They are not private backups. Exported JSON is a record, not an importable project file; replay is tied to the browser engine/version.
Use Python 3.9+; there are no third-party runtime dependencies. A virtual environment is recommended.
git clone https://github.com/styayur/Nickspeare.git
cd Nickspeare
python -m pip install -e .
python -m nickspeare generate -n 10 --seed 42
python -m nickspeare generate -n 5 --rule rogue_to_king --year-theme agincourt --royal-affinity 0.85 --json
python -m nickspeare generate -n 5 --rule markov_coin --suffix none --seed 42Python offers rogue_to_king, quote_splice, archaic_coinage, markov_coin and markov_epithet; the browser offers the first four. --suffix none adds no year, number adds a nonhistorical ornament, and both chooses a historical or numeric suffix. Run python -m nickspeare generate --help for styles and other options.
from nickspeare import NicknameGenerator
engine = NicknameGenerator(seed=42, royal_affinity=0.85)
name = engine.generate_one(rule="rogue_to_king", year_theme="agincourt")
print(name.variant)
print(name.as_dict()["provenance"])For a fixed published package, download the wheel attached to v1.0.0 and install that local file. No PyPI availability or 1.1.0 release is implied. Corpus rebuilding and optional R/RStudio analysis are covered in Development.
From the download directory, in a separate virtual environment from the editable source install:
python -m pip install ./nickspeare-1.0.0-py3-none-any.whl
From the repository root, use Python 3.9+ and Node.js 22 (CI; the manifest allows Node 20+):
python -m pip install -e '.[dev]'
python -m unittest discover -s tests -v
npm test
python -m buildFor browser integration, install Chromium with python -m playwright install chromium, serve docs as above, then run python scripts/check_browser.py in a second terminal from the repository root.
The Verify workflow tests Python 3.9/3.12/3.14 and headless Chromium on Ubuntu, including replay, export, clipboard, formation steps, responsive layout and load failure. CodeForge regression is a separate development gate. These checks do not certify every browser or operating system; GitHub Pages serves main:/docs.
- Generated names are neologisms, not Shakespeare quotations or historical pronunciation reconstructions. Speaker groups are editorial interpretations, and rare-word associations remain uncertain.
- Python and browser use different RNGs and transformation rules. Seeds replay within an engine/version, not across engines.
- The provenance atlas covers three plays and 195 words. Supplemental training corpora do not automatically extend the committed evidence index.
- R is optional analysis tooling; its exploratory output is not the published evidence atlas. Independent installation, upgrade and sustained user acceptance have not been audited.
Current source already provides generation and per-name provenance. Next work is a reviewed corpus extension and clearer evidence views; broader computational-humanities comparisons remain future experiments. See the corpus acceptance workflow. Ungrounded generation and commercial name services are not planned.
Read CONTRIBUTING.md and Development. Bug reports should include the engine/version, seed, rule, suffix and exported record. Use Issues for reproducible bugs and SECURITY.md for vulnerabilities. New corpus evidence needs reviewed hashes and attribution.
Code is AGPL-3.0-or-later. The cached Folger edition and its derived atlas retain CC BY-NC 3.0 attribution and noncommercial restrictions; the code license does not relicense those data. See Sources and rights for edition credits, hashes and optional corpora. Thanks to the Folger Shakespeare Library and its editors/encoders; Markov influences are recorded there.