#!/usr/bin/env python3 # SPDX-License-Identifier: MIT # Copyright (c) 2026 Pablo Nogueira Grossi / G6 LLC """ key_register.py -- the series' own dictionary of its operator keys, generated from the pages. Book XIV ch 4 found that the letters C, K, F, U work as translation keys: one head sense per key, and a domain-specific filler on each page ("K: portão de Heaviside em pH = 4.5"). This tool writes that finding down as a register and keeps it current (R8: never hand-edit). python3 tools/key_register.py check: exit 1 if the register is stale python3 tools/key_register.py --write regenerate book14/key-register.html and docs/key-register.tsv For each letter: senses the gloss words ("K: threshold", "U — Union"), Portuguese folded into English sense Book XIV ch 3's rule: glosses joined by a page that uses both are one sense (connected components) head the most-used gloss of the largest component key set a component whose pages mostly also name Genesis and Logos (the first two keys of the Omega chain G–L–R–U) belongs to that chain, not the dm³ chain fillers the text that follows the gloss on the page, grouped by the page's subject FIELD (docs/subjects.tsv): what the key means in that domain Run order (R25/R26): key_register → subject_tags → run_yourself → crossref → gold_standard. Standard library only. """ import csv, html, os, re, subprocess, sys from collections import defaultdict ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__))) OUT_HTML = os.path.join(ROOT, 'book14', 'key-register.html') OUT_TSV = os.path.join(ROOT, 'docs', 'key-register.tsv') RETIRED = ('docs/ml-evidence/', '_archive/', '_to_delete/') # pages that QUOTE glosses as examples (use/mention): not counted as usage MENTION = {'book14/ch04-translation-keys.html', 'book14/key-register.html', 'book14/ch06-words-that-went-wrong.html', 'book14/ch02-how-mathematical-prose-parses.html', 'book14/ch09-what-types-decide.html'} SENSES = { 'C': {'compression': ['compression', 'compress', 'compressão'], 'contact': ['contact']}, 'K': {'threshold': ['threshold', 'limiar'], 'curvature': ['curvature', 'curvatura'], 'gate': ['gate', 'portão']}, 'F': {'fold': ['fold', 'folding', 'dobramento', 'dobra']}, 'U': {'unfolding': ['unfolding', 'unfold', 'desdobramento'], 'union': ['union', 'união'], 'universal': ['universal'], 'unification': ['unification', 'unificação']}, } GEN = re.compile(r'.*?', re.S) STY = re.compile(r'<(style|script)\b.*?', re.S | re.I) TAG = re.compile(r'<[^>]+>') # the Omega chain's first two keys; a page naming both is working in that key set OMEGA_WORDS = (re.compile(r'\bGenesis\b'), re.compile(r'\bLogos\b')) # a filler ends at a sentence mark, a separator, or the next key ("F — ...") STOP = re.compile(r'\s[A-Z]\s*(?:—|–|=|:)\s|[.;|·•]\s|\s—\s|\s\(\s*[A-Z]\s*\)|→|\s\d+(?:\.\d+)+\s') def pages(): out = subprocess.run(['git', '--no-optional-locks', '-C', ROOT, 'ls-files', '*.html'], capture_output=True, text=True).stdout.split() return sorted(f for f in out if not f.startswith(RETIRED) and f not in MENTION) def text(f): t = open(os.path.join(ROOT, f), encoding='utf-8', errors='ignore').read() return re.sub(r'\s+', ' ', html.unescape(TAG.sub(' ', STY.sub(' ', GEN.sub(' ', t))))) def field_of(): d = {} with open(os.path.join(ROOT, 'docs', 'subjects.tsv'), encoding='utf-8') as fh: for row in csv.reader(fh, delimiter='\t'): if len(row) >= 2 and row[0] and not row[0].startswith('#'): d[row[0]] = row[1].split(' · ')[0].strip() return d def build(): T = {f: text(f) for f in pages()} field = field_of() use = {L: defaultdict(set) for L in SENSES} fill = {L: defaultdict(lambda: defaultdict(list)) for L in SENSES} for L, senses in SENSES.items(): word = {w: s for s, ws in senses.items() for w in ws} rx = re.compile(r'(? 0.5 else 'dm³ chain (C→K→F→U)' head = max(sorted(c), key=lambda s: len(use[L][s])) for s in sorted(c, key=lambda s: (-len(use[L][s]), s)): reg.append(dict(letter=L, sense=s, component=ci + 1, pages=len(use[L][s]), head=(s == head and ci == 0), comp_head=head, keyset=keyset, fillers=fill[L][s], n_comp=len(comps))) return reg, len(T) def render(reg, npages): src = open(os.path.join(ROOT, 'book20', 'ch01-the-planes-that-came-back.html'), encoding='utf-8').read() css = src[src.find('') + 8] e = html.escape parts = [] for L in SENSES: rows = [r for r in reg if r['letter'] == L] n_comp = rows[0]['n_comp'] if rows else 0 head = next((r['sense'] for r in rows if r['head']), '—') parts.append(f'
key {L}

{L} — head sense: {e(head)}

') parts.append(f'

{n_comp} sense(s) by Book XIV ch 3’s rule.

') parts.append('
') for r in rows: if r['head']: tag = 'head' elif r['comp_head'] != r['sense']: tag = 'same sense as ' + e(r['comp_head']) else: tag = 'separate sense' parts.append(f'' f'') parts.append('
glosspagessensekey set
{e(r["sense"])}{r["pages"]}{tag}{e(r["keyset"])}
') for r in rows: if not r['fillers']: continue parts.append(f'

{L} = {e(r["sense"])}: fillers by domain

') body = '\n'.join(parts) return f''' Key Register · Book XIV — Principia Orthogona {css}
⚜ PRINCIPIA ORTHOGONA · Book XIV · Key Register ← Ch 4 · Translation Keys · Book XIV contents
Book XIV · Reference · generated

Key Register

The operator letters as dictionary entries: one head sense per key, the key set it belongs to, and what it is filled with in each domain.
Source{npages} published pages, regenerated by tools/key_register.py
MethodBook XIV ch 3 (senses as chains) and ch 4 (keys and fillers)
Idealetters as translation keys: Pablo Nogueira Grossi
Each entry follows Chapter 3's template. The head sense comes first. Glosses that pages join count as one sense. A gloss that belongs to another chain is marked as such. Under each entry, the fillers show what the key becomes in each domain, each one linked to the page that says it.
{body}
How to read the fillersA filler is the text that follows the gloss on a page, cut at the next sentence mark or key, so some are fragments. The domain is the page's subject tag; untagged pages are listed last. A filler shows usage and does not verify it.
''' def tsv(reg): lines = ['letter\tsense\tpages\tcomponent\thead\tkeyset\tdomains'] for r in reg: lines.append(f"{r['letter']}\t{r['sense']}\t{r['pages']}\t{r['component']}\t" f"{'yes' if r['head'] else ''}\t{r['keyset']}\t{len(r['fillers'])}") return '\n'.join(lines) + '\n' def strip_gen(s): # later generators stamp their own blocks onto the page; staleness ignores them s = re.sub(r'.*?\n?', '', s, flags=re.S) return re.sub(r']*>\n?', '', s) def main(): reg, n = build() h, t = render(reg, n), tsv(reg) old_h = open(OUT_HTML, encoding='utf-8').read() if os.path.exists(OUT_HTML) else '' old_t = open(OUT_TSV, encoding='utf-8').read() if os.path.exists(OUT_TSV) else '' stale = strip_gen(old_h) != strip_gen(h) or old_t != t heads = {r['letter']: r['sense'] for r in reg if r['head']} if '--write' in sys.argv: if stale: open(OUT_HTML, 'w', encoding='utf-8').write(h) open(OUT_TSV, 'w', encoding='utf-8').write(t) print(f"{n} pages; {len(reg)} glosses; heads {heads}; {'written' if stale else '0 stale'}") return 0 print(f"{n} pages; {len(reg)} glosses; {'STALE' if stale else '0 stale'}") return 1 if stale else 0 if __name__ == '__main__': sys.exit(main())