#!/usr/bin/env python3 """ words-drill -- generate a personalized typing drill HTML page from weak bigrams/trigrams. Usage: words-drill [--name NAME] [--corpus PATH] [--count N] [--out PATH] BIGRAM [BIGRAM ...] words-drill --stdin [other flags] # read bigrams one per line from stdin Bigrams are 2- or 3-character sequences (e.g. th, qu, tio). For symbol bigrams (=>, ;;) pass --corpus pointing at a code wordlist. """ import argparse import os import random import re import sys from datetime import date from pathlib import Path # --------------------------------------------------------------------------- # Paths # --------------------------------------------------------------------------- WORDS_ROOT = Path("/home/krisyotam/dev/words") DEFAULT_CORPUS = WORDS_ROOT / "sources/google-10000-english/google-10000-english-usa-no-swears.txt" DRILLS_DIR = WORDS_ROOT / "drills" # --------------------------------------------------------------------------- # HTML helpers (match build.py style exactly) # --------------------------------------------------------------------------- HEAD_TEMPLATE = """\ {title} -- Kris Yotam's Word Lists """ FOOT = """\ """ def render_drill_page(name: str, words: list, bigrams: list) -> str: bigram_list = ", ".join(bigrams) word_count = len(words) body = "\n".join(words) return ( HEAD_TEMPLATE.format(title=name) + '

\n' + '← back to drills\n' + f"

{name}

\n" + f'

{word_count:,} words — bigrams: {bigram_list}

\n' + f'

{body}

\n' + "

\n" + FOOT ) def render_drills_index(drill_files: list) -> str: """Render drills/index.html from a list of (stem, mtime) pairs, sorted by mtime desc.""" listing_items = "" for stem, _mtime in drill_files: title = stem.replace("-", " ").title() listing_items += ( '

{title}

\n" ) return ( HEAD_TEMPLATE.format(title="Drills") + '

\n' + '← back to word lists\n' + "

Drills

\n" + "

\n" + listing_items + FOOT ) def render_drills_placeholder() -> str: return ( HEAD_TEMPLATE.format(title="Drills") + '

\n' + '← back to word lists\n' + "

Drills

\n" + '

Coming soon -- LLM-generated symbol drills.

\n' + "

\n" + FOOT ) # --------------------------------------------------------------------------- # Core logic # --------------------------------------------------------------------------- def load_corpus(path: Path) -> list: if not path.exists(): print(f"error: corpus file not found: {path}", file=sys.stderr) sys.exit(1) lines = path.read_text(encoding="utf-8").splitlines() return [l.strip() for l in lines if l.strip()] def find_words_for_bigram(corpus: list, bigram: str) -> list: """Return all corpus words that contain the given bigram.""" return [w for w in corpus if bigram in w] def build_drill_wordlist(corpus: list, bigrams: list, count: int, seed: int) -> tuple: """ Distribute `count` words across bigrams proportionally. Returns (final_word_list, counts_dict). counts_dict maps each bigram to how many words were found/used. """ per_bigram = count // len(bigrams) if bigrams else 0 remainder = count - per_bigram * len(bigrams) rng = random.Random(seed) all_words = [] counts = {} for i, bigram in enumerate(bigrams): matches = find_words_for_bigram(corpus, bigram) quota = per_bigram + (1 if i < remainder else 0) if not matches: counts[bigram] = 0 continue rng.shuffle(matches) selected = matches[:quota] counts[bigram] = len(selected) all_words.extend(selected) # Shuffle final combined list with same rng for reproducibility rng.shuffle(all_words) return all_words, counts def rebuild_drills_index() -> None: """Scan drills/*.html (skip index.html), sort by mtime desc, write index.""" html_files = [ p for p in DRILLS_DIR.glob("*.html") if p.name != "index.html" ] if not html_files: DRILLS_DIR.joinpath("index.html").write_text(render_drills_placeholder(), encoding="utf-8") return drill_files = sorted( [(p.stem, p.stat().st_mtime) for p in html_files], key=lambda x: x[1], reverse=True, ) DRILLS_DIR.joinpath("index.html").write_text( render_drills_index(drill_files), encoding="utf-8" ) # --------------------------------------------------------------------------- # CLI # --------------------------------------------------------------------------- def main(): parser = argparse.ArgumentParser( description="Generate a personalized typing drill HTML page from weak bigrams.", formatter_class=argparse.RawDescriptionHelpFormatter, ) parser.add_argument( "bigrams", nargs="*", metavar="BIGRAM", help="2- or 3-character sequences to drill (e.g. th qu tio)", ) parser.add_argument( "--stdin", action="store_true", help="Read bigrams one per line from stdin instead of positional args", ) parser.add_argument( "--name", default=None, help="Name for the drill (default: weakness-YYYY-MM-DD)", ) parser.add_argument( "--corpus", default=str(DEFAULT_CORPUS), help="Path to corpus wordlist file (default: google-10000-english-usa-no-swears.txt)", ) parser.add_argument( "--count", type=int, default=200, help="Total words in output drill (default: 200)", ) parser.add_argument( "--out", default=None, help="Output HTML file path (default: drills/.html)", ) args = parser.parse_args() # Collect bigrams if args.stdin: bigrams = [line.strip() for line in sys.stdin if line.strip()] else: bigrams = args.bigrams if not bigrams: print( "usage: words-drill [--name NAME] [--corpus PATH] [--count N] [--out PATH] BIGRAM [BIGRAM ...]\n" " words-drill --stdin [other flags]\n" "error: no bigrams specified", file=sys.stderr, ) sys.exit(2) # Resolve name today = date.today().isoformat() name = args.name if args.name else f"weakness-{today}" # Resolve output path DRILLS_DIR.mkdir(parents=True, exist_ok=True) out_path = Path(args.out) if args.out else DRILLS_DIR / f"{name}.html" # Load corpus corpus_path = Path(args.corpus) corpus = load_corpus(corpus_path) # Derive seed from bigrams + name for reproducibility seed_str = name + "".join(sorted(bigrams)) seed = int.from_bytes(seed_str.encode(), "little") % (2**32) # Build drill words, counts = build_drill_wordlist(corpus, bigrams, args.count, seed) # Write HTML html = render_drill_page(name, words, bigrams) out_path.write_text(html, encoding="utf-8") # Rebuild drills index rebuild_drills_index() # Report total_written = len(words) detail = ", ".join(f"{b}: {counts.get(b, 0)}" for b in bigrams) print(out_path) print( f"wrote {total_written}/{args.count} words across {len(bigrams)} bigrams ({detail})" ) if __name__ == "__main__": main()