split-master.py

"""One-shot: split the single-file master into index.html + css/app.css + js/*.js.Content-preserving: every line moves verbatim into exactly one output file.Run from D:/workspace/guessword. Writes build/guessword.presplit.bak.html first."""import os, shutil, reROOT = "D:/workspace/guessword"BAK = os.path.join(ROOT, "build", "guessword.presplit.bak.html")SRC = os.path.join(ROOT, "guessword.html")if not os.path.exists(SRC):          # already split once: re-run from the pre-split backup    SRC = BAKsrc = open(SRC, encoding="utf-8", newline="").read()assert "\r\n" not in srclines = src.split("\n")          # last element is "" (file ends with \n)print("lines:", len(lines))def find(pred, start=0):    for i in range(start, len(lines)):        if pred(lines[i]):            return i    raise SystemExit("marker not found")i_style_open = find(lambda l: l.strip() == "<style>")i_style_close = find(lambda l: l.strip() == "</style>", i_style_open)i_head_close = find(lambda l: l.strip() == "</head>")i_script_open = find(lambda l: l.strip() == "<script>")i_script_close = find(lambda l: l.strip() == "</script>", i_script_open)i_body_end = find(lambda l: l.strip() == "</body>", i_script_close)print(i_style_open, i_style_close, i_head_close, i_script_open, i_script_close, i_body_end)head = lines[:i_style_open]                       # doctype .. favicon linkcss = lines[i_style_open + 1:i_style_close]        # style bodymid = lines[i_head_close + 1:i_script_open]        # </head> is re-emitted, so start AFTER itscript = lines[i_script_open + 1:i_script_close]   # js bodytail = lines[i_body_end:]                          # </body>, </html>, ""# --- split the script body into data files + app ---data_rx = [    ("en", re.compile(r"^const GWD=")),    ("en", re.compile(r"^const GWMAP=")),    ("en", re.compile(r"^const GWDICT_B64=")),    ("ja", re.compile(r"^const GW_JA=")),    ("ja", re.compile(r"^const GWDICT_JA_B64=")),    ("ko", re.compile(r"^const GW_KO=")),    ("ko", re.compile(r"^const GWDICT_KO_B64=")),    ("extra", re.compile(r"^let GW_FR=")),]buckets = {"en": [], "ja": [], "ko": [], "extra": []}app = []used = set()for idx, ln in enumerate(script):    for name, rx in data_rx:        if rx.match(ln):            buckets[name].append(ln)            used.add(idx)            breakfor idx, ln in enumerate(script):    if idx not in used:        app.append(ln)print("data lines:", {k: len(v) for k, v in buckets.items()}, "app lines:", len(app))assert all(len(v) == (3 if k == "en" else 2 if k in ("ja", "ko") else 1) for k, v in buckets.items())HEADERS = {    "en": "/* GuessWord data — English pack (GWD: vocab + Numberbatch 300d int8 vectors; GWMAP: lemma map)\n   and the English dictionary (ECDICT/Wiktionary, gzip+base64). Generated file: edit via build/ scripts. */",    "ja": "/* GuessWord data — Japanese pack + dictionary (gzip+base64). Generated file. */",    "ko": "/* GuessWord data — Korean pack + dictionary (gzip+base64). Generated file. */",    "extra": "/* GuessWord data — fr/de/es placeholders. The real fr/de/es packs live outside the repo\n   (D:/gw-dl/lang/pack-<L>.json, dict-<L>.json) and are injected by build-variant.mjs at deploy time. */",}os.makedirs(os.path.join(ROOT, "css"), exist_ok=True)os.makedirs(os.path.join(ROOT, "js"), exist_ok=True)for k, body in buckets.items():    with open(os.path.join(ROOT, "js", f"data-{k}.js"), "w", encoding="utf-8", newline="\n") as f:        f.write(HEADERS[k] + "\n" + "\n".join(body) + "\n")with open(os.path.join(ROOT, "js", "app.js"), "w", encoding="utf-8", newline="\n") as f:    f.write("\n".join(app) + "\n")with open(os.path.join(ROOT, "css", "app.css"), "w", encoding="utf-8", newline="\n") as f:    f.write("\n".join(css) + "\n")index = "\n".join(head + [    '<link rel="stylesheet" href="css/app.css">',    "</head>",] + mid + [    '<script src="js/data-en.js"></script>',    '<script src="js/data-extra.js"></script>',    '<script src="js/data-ja.js"></script>',    '<script src="js/data-ko.js"></script>',    '<script src="js/app.js"></script>',] + tail)with open(os.path.join(ROOT, "index.html"), "w", encoding="utf-8", newline="\n") as f:    f.write(index)if os.path.abspath(SRC) != os.path.abspath(BAK):    shutil.copy2(SRC, BAK)    os.remove(SRC)print("wrote index.html, css/app.css, js/{data-en,data-extra,data-ja,data-ko,app}.js")for p in ["index.html", "css/app.css", "js/app.js", "js/data-en.js", "js/data-extra.js", "js/data-ja.js", "js/data-ko.js"]:    print("  ", p, f"{os.path.getsize(os.path.join(ROOT, p))/1048576:.2f}MB")print("backup:", BAK)