Seed from NAS prod snapshot 20260130-190531
This commit is contained in:
archicratia
285
scripts/import-docx.mjs
Normal file
285
scripts/import-docx.mjs
Normal file
@@ -0,0 +1,285 @@
|
||||
// scripts/import-docx.mjs
|
||||
// Import DOCX -> MDX (manifest-driven), preferring pandoc if available.
|
||||
// Usage examples:
|
||||
// node scripts/import-docx.mjs --manifest sources/manifest.yml --only archicrat-ia/prologue,archicrat-ia/chapitre-1
|
||||
// node scripts/import-docx.mjs --manifest sources/manifest.yml --all
|
||||
//
|
||||
// Output: src/content/<collection>/<slug>.mdx
|
||||
// Assets (images): public/imported/<collection>/<slug>/...
|
||||
|
||||
import fs from "node:fs/promises";
|
||||
import path from "node:path";
|
||||
import { spawnSync } from "node:child_process";
|
||||
import process from "node:process";
|
||||
|
||||
import { parse as parseYAML } from "yaml";
|
||||
import mammoth from "mammoth";
|
||||
|
||||
function parseArgs(argv) {
|
||||
const out = { manifest: "sources/manifest.yml", only: null, all: false, dryRun: false, force: false };
|
||||
for (let i = 2; i < argv.length; i++) {
|
||||
const a = argv[i];
|
||||
if (a === "--manifest") out.manifest = argv[++i];
|
||||
else if (a === "--only") out.only = (argv[++i] || "").split(",").map(s => s.trim()).filter(Boolean);
|
||||
else if (a === "--all") out.all = true;
|
||||
else if (a === "--dry-run") out.dryRun = true;
|
||||
else if (a === "--force") out.force = true;
|
||||
else if (a === "--help" || a === "-h") {
|
||||
console.log(`DOCX importer
|
||||
--manifest <path> (default: sources/manifest.yml)
|
||||
--only <slug,slug> import only those slugs
|
||||
--all import all entries in manifest
|
||||
--dry-run show what would be done, write nothing
|
||||
--force overwrite existing mdx files
|
||||
`);
|
||||
process.exit(0);
|
||||
}
|
||||
}
|
||||
if (!out.all && (!out.only || out.only.length === 0)) {
|
||||
throw new Error("Missing --only or --all. Example: --only archicrat-ia/prologue,archicrat-ia/chapitre-1");
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
async function readManifest(filePath) {
|
||||
const raw = await fs.readFile(filePath, "utf8");
|
||||
const data = parseYAML(raw);
|
||||
|
||||
// Accept either:
|
||||
// - root array
|
||||
// - { items: [...] }
|
||||
const items = Array.isArray(data) ? data : (data?.items || data?.docs || []);
|
||||
if (!Array.isArray(items) || items.length === 0) {
|
||||
throw new Error(`Manifest has no items: ${filePath}`);
|
||||
}
|
||||
|
||||
// Normalize keys
|
||||
return items.map((it, idx) => {
|
||||
const source = it.source || it.src || it.path;
|
||||
const collection = it.collection;
|
||||
const slug = it.slug;
|
||||
const title = it.title || it.titre || slug;
|
||||
const order = it.order ?? it.ordre ?? idx;
|
||||
|
||||
if (!source || !collection || !slug) {
|
||||
throw new Error(`Manifest item missing fields (need source+collection+slug). Item: ${JSON.stringify(it)}`);
|
||||
}
|
||||
|
||||
return { source, collection, slug, title, order };
|
||||
});
|
||||
}
|
||||
|
||||
function havePandoc() {
|
||||
const r = spawnSync("pandoc", ["--version"], { stdio: "ignore" });
|
||||
return r.status === 0;
|
||||
}
|
||||
|
||||
function runPandoc(docxPath, assetsOutDir) {
|
||||
// Extract media into assetsOutDir (pandoc will create subfolders).
|
||||
// We output GitHub-flavored markdown (good enough for MDX).
|
||||
const args = [
|
||||
docxPath,
|
||||
"-f", "docx",
|
||||
"-t", "gfm",
|
||||
"--wrap=none",
|
||||
"--extract-media", assetsOutDir,
|
||||
];
|
||||
const r = spawnSync("pandoc", args, { encoding: "utf8" });
|
||||
if (r.status !== 0) {
|
||||
throw new Error(`pandoc failed for ${docxPath}\n${r.stderr || ""}`);
|
||||
}
|
||||
return r.stdout || "";
|
||||
}
|
||||
|
||||
async function runMammoth(docxPath, assetsOutDirWebRoot) {
|
||||
// Mammoth -> HTML; we keep HTML inside MDX.
|
||||
// Images are saved into public/... so they can be referenced.
|
||||
const assetsDiskDir = path.resolve(assetsOutDirWebRoot);
|
||||
await fs.mkdir(assetsDiskDir, { recursive: true });
|
||||
|
||||
let imgCount = 0;
|
||||
const result = await mammoth.convertToHtml(
|
||||
{ path: docxPath },
|
||||
{
|
||||
convertImage: mammoth.images.imgElement(async (image) => {
|
||||
imgCount++;
|
||||
const ext = image.contentType?.split("/")?.[1] || "png";
|
||||
const fileName = `image-${String(imgCount).padStart(2, "0")}.${ext}`;
|
||||
const buf = await image.read();
|
||||
await fs.writeFile(path.join(assetsDiskDir, fileName), buf);
|
||||
// Return public URL path (we'll set correct prefix outside)
|
||||
return { src: fileName };
|
||||
}),
|
||||
}
|
||||
);
|
||||
|
||||
let html = result.value || "";
|
||||
|
||||
// Mammoth gives relative src="image-xx.png" ; we will prefix later
|
||||
return html;
|
||||
}
|
||||
|
||||
function escapeFrontmatterString(s) {
|
||||
return String(s).replace(/"/g, '\\"');
|
||||
}
|
||||
|
||||
function stripDuplicateTitle(markdownOrHtml, title) {
|
||||
const t = String(title || "").trim();
|
||||
if (!t) return markdownOrHtml;
|
||||
|
||||
// Remove leading "# Title"
|
||||
const md = markdownOrHtml.replace(/^\s*#\s+([^\n]+)\n+/, (m, h1) => {
|
||||
return h1.trim() === t ? "" : m;
|
||||
});
|
||||
|
||||
// Remove leading <h1>Title</h1>
|
||||
return md.replace(/^\s*<h1[^>]*>(.*?)<\/h1>\s*/i, (m, h1) => {
|
||||
const plain = String(h1).replace(/<[^>]+>/g, "").trim();
|
||||
return plain === t ? "" : m;
|
||||
});
|
||||
}
|
||||
|
||||
function stripWordToc(markdownOrHtml) {
|
||||
// Remove a Word-generated TOC block near the start of the document.
|
||||
// Heuristic: in the first ~4000 chars, detect a "Sommaire" / "Table des matières" heading
|
||||
// followed by a run of links/list items, and cut until the next real heading.
|
||||
const text = String(markdownOrHtml || "");
|
||||
const head = text.slice(0, 4000);
|
||||
|
||||
const m = head.match(/(^|\n)\s{0,3}(#{1,6}\s*)?(Sommaire|Table des matières)\s*\n/i);
|
||||
if (!m) return markdownOrHtml;
|
||||
|
||||
const startIdx = m.index ?? 0;
|
||||
const rest = text.slice(startIdx);
|
||||
|
||||
// Cut until next heading (markdown-style). If none, drop everything from startIdx.
|
||||
const nextHeading = rest.slice(1).match(/\n#{1,6}\s+[^\n]+\n/);
|
||||
if (!nextHeading) return (text.slice(0, startIdx)).trim() + "\n";
|
||||
|
||||
const endIdx = startIdx + 1 + (nextHeading.index ?? 0);
|
||||
return (text.slice(0, startIdx) + text.slice(endIdx)).trim() + "\n";
|
||||
}
|
||||
|
||||
function rewriteLocalImageLinks(text, publicPrefix) {
|
||||
// pandoc tends to output  or 
|
||||
// mammoth fallback gives <img src="image-01.png">
|
||||
let out = text;
|
||||
|
||||
// Markdown image links: ](media/...
|
||||
out = out.replace(/\]\(\s*media\//g, `](${publicPrefix}/media/`);
|
||||
|
||||
// HTML img src="image-xx.ext"
|
||||
out = out.replace(/src="(image-\d+\.[a-z0-9]+)"/gi, `src="${publicPrefix}/$1"`);
|
||||
|
||||
return out;
|
||||
}
|
||||
|
||||
function stripHtmlComments(text) {
|
||||
return String(text || "").replace(/<!--[\s\S]*?-->/g, "");
|
||||
}
|
||||
|
||||
async function exists(p) {
|
||||
try { await fs.access(p); return true; } catch { return false; }
|
||||
}
|
||||
|
||||
async function main() {
|
||||
const args = parseArgs(process.argv);
|
||||
const manifestPath = path.resolve(args.manifest);
|
||||
|
||||
const items = await readManifest(manifestPath);
|
||||
const selected = args.all ? items : items.filter(it => args.only.includes(it.slug));
|
||||
|
||||
if (!args.all && selected.length !== args.only.length) {
|
||||
const found = new Set(selected.map(s => s.slug));
|
||||
const missing = args.only.filter(s => !found.has(s));
|
||||
throw new Error(`Some --only slugs not found in manifest: ${missing.join(", ")}`);
|
||||
}
|
||||
|
||||
const pandocOk = havePandoc();
|
||||
|
||||
console.log(`Manifest: ${manifestPath}`);
|
||||
console.log(`Pandoc: ${pandocOk ? "YES (preferred)" : "NO (fallback to mammoth/html)"}`);
|
||||
console.log(`Items: ${selected.length}`);
|
||||
|
||||
for (const it of selected) {
|
||||
const docxPath = path.resolve(it.source);
|
||||
const outFile = path.resolve("src/content", it.collection, `${it.slug}.mdx`);
|
||||
const outDir = path.dirname(outFile);
|
||||
|
||||
const assetsPublicDir = path.posix.join("/imported", it.collection, it.slug);
|
||||
const assetsDiskDir = path.resolve("public", "imported", it.collection, it.slug);
|
||||
|
||||
if (!(await exists(docxPath))) {
|
||||
throw new Error(`Missing source docx: ${docxPath}`);
|
||||
}
|
||||
if ((await exists(outFile)) && !args.force) {
|
||||
throw new Error(`Refusing to overwrite existing: ${outFile} (use --force)`);
|
||||
}
|
||||
|
||||
console.log(`\n→ ${it.slug}`);
|
||||
console.log(` source: ${it.source}`);
|
||||
console.log(` out: ${path.relative(process.cwd(), outFile)}`);
|
||||
|
||||
if (args.dryRun) continue;
|
||||
|
||||
await fs.mkdir(outDir, { recursive: true });
|
||||
await fs.mkdir(assetsDiskDir, { recursive: true });
|
||||
|
||||
let body = "";
|
||||
if (pandocOk) {
|
||||
// pandoc extract-media wants a directory; it will create media/ inside.
|
||||
body = runPandoc(docxPath, assetsDiskDir);
|
||||
body = stripDuplicateTitle(body, it.title);
|
||||
body = stripWordToc(body);
|
||||
body = stripHtmlComments(body);
|
||||
body = rewriteLocalImageLinks(body, assetsPublicDir);
|
||||
} else {
|
||||
let html = await runMammoth(docxPath, assetsDiskDir);
|
||||
html = stripDuplicateTitle(html, it.title);
|
||||
html = stripWordToc(html);
|
||||
html = stripHtmlComments(html);
|
||||
html = rewriteLocalImageLinks(html, assetsPublicDir);
|
||||
body = html.trim() ? html : "<p>(Import vide)</p>";
|
||||
}
|
||||
|
||||
const defaultVersion = process.env.PUBLIC_RELEASE || "0.1.0";
|
||||
|
||||
const schemaDefaultsByCollection = {
|
||||
archicratie: { edition: "archicratie", status: "modele_sociopolitique", level: 1 },
|
||||
ia: { edition: "ia", status: "cas_pratique", level: 1 },
|
||||
traite: { edition: "traite", status: "ontodynamique", level: 1 },
|
||||
glossaire: { edition: "glossaire", status: "lexique", level: 1 },
|
||||
atlas: { edition: "atlas", status: "atlas", level: 1 },
|
||||
};
|
||||
|
||||
const defaults = schemaDefaultsByCollection[it.collection] || { edition: it.collection, status: "draft", level: 1 };
|
||||
|
||||
const fm = [
|
||||
"---",
|
||||
`title: "${escapeFrontmatterString(it.title)}"`,
|
||||
`edition: "${escapeFrontmatterString(defaults.edition)}"`,
|
||||
`status: "${escapeFrontmatterString(defaults.status)}"`,
|
||||
`level: ${Number(defaults.level)}`,
|
||||
`version: "${escapeFrontmatterString(defaultVersion)}"`,
|
||||
`concepts: []`,
|
||||
`links: []`,
|
||||
`order: ${Number(it.order)}`,
|
||||
`summary: ""`,
|
||||
`source:`,
|
||||
` kind: docx`,
|
||||
` path: "${escapeFrontmatterString(it.source)}"`,
|
||||
"---",
|
||||
"",
|
||||
].join("\n");
|
||||
|
||||
const mdx = fm + body.trim() + "\n";
|
||||
await fs.writeFile(outFile, mdx, "utf8");
|
||||
}
|
||||
|
||||
console.log("\nDone.");
|
||||
}
|
||||
|
||||
main().catch((e) => {
|
||||
console.error("\nERROR:", e?.message || e);
|
||||
process.exit(1);
|
||||
});
|
||||
Reference in New Issue
Block a user