e1de66e75a
Checking the live instance a day later showed 2,023 Chinese characters back in the served bundles, from two causes. The comment merge rebuilt its dictionary from the scratch files in ~/.cache only, so the committed translations were never an input. Once a later pass added comments, a rebuild discarded them: the count stayed at 274 while eight new translations vanished. src/dictionary.comments.json is now loaded first and the scratch files overlay it, so a rebuild can only grow the dictionary. Two plugins had also shipped new Chinese since the last sweep: dsh-builtin-browser (a new plugin, 834 characters of settings copy) and dsh-our-free-model 2.5.4 (1,082 characters, mostly sponsor copy and comments). Their strings were harvested from the live-served bytes and translated: 70 UI strings into dictionary.residual.json and 8 comments into the comment dictionary. Verified on a fresh instance, escapes decoded: total Chinese across the 79 served bundles 2,023 -> 107, and in the browser both visible text and full page HTML report 0 CJK. The 107 remaining characters are the functional regular expressions that match Chinese input, as before.
76 lines
3.4 KiB
JavaScript
76 lines
3.4 KiB
JavaScript
// Builds src/dictionary.comments.json from the translated comments.
|
|
//
|
|
// The committed dictionary is the source of truth: it is loaded first, then any
|
|
// newer translation passes overlay it. Earlier this read only the scratch files
|
|
// in ~/.cache, so once those were cleared a rebuild silently produced a smaller
|
|
// dictionary and translated comments reverted to Chinese.
|
|
//
|
|
// Overlays, when present:
|
|
// ~/.cache/c2e/cmt-authored.json {wholeComment: translatedComment}
|
|
// ~/.cache/c2e/css-authored.json {body: translatedBody} (delimiters added)
|
|
// ~/.cache/c2e/c2-authored.json {wholeComment: translatedComment}
|
|
//
|
|
// Output: src/dictionary.comments.json {wholeComment: translatedComment}
|
|
import { readFileSync, writeFileSync, existsSync } from "node:fs";
|
|
|
|
const C2E = "/home/oiupoyt/.cache/c2e";
|
|
const OUT = "/home/oiupoyt/my stuff yea/projects/chinese2english/src/dictionary.comments.json";
|
|
|
|
const HAN = /[\u4e00-\u9fff\u3400-\u4dbf\uf900-\ufaff]/;
|
|
const BOX = /[\u2500-\u257f\u2580-\u259f]/g;
|
|
const boxOf = (t) => (t.match(BOX) ?? []).sort().join("");
|
|
const newlines = (t) => (t.match(/\n/g) ?? []).length;
|
|
|
|
function load(name, fallback) {
|
|
const path = `${C2E}/${name}`;
|
|
if (!existsSync(path)) return fallback ?? {};
|
|
return JSON.parse(readFileSync(path, "utf8"));
|
|
}
|
|
|
|
const pairs = new Map();
|
|
|
|
// Existing committed translations survive a rebuild.
|
|
if (existsSync(OUT)) {
|
|
for (const [zh, en] of Object.entries(JSON.parse(readFileSync(OUT, "utf8")))) pairs.set(zh, en);
|
|
}
|
|
|
|
for (const [zh, en] of Object.entries(load("cmt-authored.json"))) pairs.set(zh, en);
|
|
for (const [zh, en] of Object.entries(load("c2-authored.json"))) pairs.set(zh, en);
|
|
for (const [zh, en] of Object.entries(load("css-authored.json"))) {
|
|
// A stylesheet body is stored without delimiters; the emitted key carries them.
|
|
if (zh.includes("/*")) pairs.set(zh, en);
|
|
else pairs.set(`/*${zh}*/`, `/*${en}*/`);
|
|
}
|
|
|
|
const clean = {};
|
|
const reasons = { cjk: 0, latin: 0, identity: 0, box: 0, lines: 0, syntax: 0 };
|
|
|
|
for (const [zh, en] of pairs) {
|
|
if (typeof en !== "string" || !en.trim()) continue;
|
|
if (HAN.test(en)) { reasons.cjk++; continue; }
|
|
if (!/[A-Za-z]/.test(en)) { reasons.latin++; continue; }
|
|
if (en === zh) { reasons.identity++; continue; }
|
|
// Replacing a comment must keep it a well-formed comment, or the emitted
|
|
// bundle stops parsing.
|
|
if (zh.includes("/*") !== en.includes("/*")) { reasons.syntax++; continue; }
|
|
if (zh.includes("*/") !== en.includes("*/")) { reasons.syntax++; continue; }
|
|
if (zh.startsWith("//") !== en.startsWith("//")) { reasons.syntax++; continue; }
|
|
// Box-drawing runs are the structure of block comments, where they align the
|
|
// `*` column. In a single-line `//` banner the run is pure decoration and its
|
|
// length shifts freely with the translated text.
|
|
const isLineComment = zh.startsWith("//");
|
|
if (!isLineComment && boxOf(zh) !== boxOf(en)) { reasons.box++; continue; }
|
|
if (newlines(zh) !== newlines(en)) { reasons.lines++; continue; }
|
|
clean[zh] = en;
|
|
}
|
|
|
|
const sorted = Object.fromEntries(
|
|
Object.entries(clean).sort((a, b) => b[0].length - a[0].length || a[0].localeCompare(b[0])),
|
|
);
|
|
writeFileSync(OUT, JSON.stringify(sorted, null, 1) + "\n");
|
|
const dropped = Object.values(reasons).reduce((a, b) => a + b, 0);
|
|
console.log(
|
|
`comments: ${Object.keys(sorted).length} entries (dropped ${dropped}` +
|
|
Object.entries(reasons).filter(([, v]) => v).map(([k, v]) => `, ${k} ${v}`).join("") +
|
|
")",
|
|
);
|