@contentrain/wp-import 0.16.4 → 0.17.0

package/README.md

@@ -77,10 +77,16 @@ models get `i18n: true`, content and meta are written per locale
 (`content/{domain}/{model}/{locale}.json`), and every translation in a group
 shares one entry id — the canonical member is the default-locale post (else
 the lowest id), so `entry_source_map[wp_id]` addresses each post by that id
-and its own locale. Locales are primary subtags (`tr_TR` → `tr`). Taxonomies
-stay per-store; the `language` / `post_translations` bookkeeping taxonomies
-never become models. `report.locales` and `report.translation_groups` say
-what happened. A monolingual site is unchanged (`i18n: false`, `data.json`).
+and its own locale. Locales are primary subtags (`tr_TR` → `tr`). A taxonomy
+stays one store-wide model unless its terms carry a language (`RawTerm.lang`,
+from the Bridge): then it is i18n too, and the Bridge's `term_pairs` fold each
+term translation group into one entry id the same way (default-locale term,
+else the lowest id). Every reference to a term — a post's taxonomy field, its
+primary category, an ACF term field, a menu target, a parent — resolves to its
+group's id. The `language` / `post_translations` bookkeeping taxonomies never
+become models. `report.locales`, `report.translation_groups` and
+`report.term_translation_groups` say what happened. A monolingual site is
+unchanged (`i18n: false`, `data.json`).
 
 Few sites are translated evenly. A post type whose content exists in some of the
 site's languages gets `model.locales` — the locales it actually has posts in,

package/dist/index.d.mts

@@ -126,10 +126,12 @@ interface ImportReport {
     fields: number;
     entries: number;
   }>;
-  /** Default first, then every other language seen on a post. One entry on a monolingual site. */
+  /** Default first, then every other language seen on a post or a translated term. One entry on a monolingual site. */
   locales: string[];
   /** Translation groups that were folded into one entry id across locales. */
   translation_groups: number;
+  /** Term translation groups (`term_pairs`) folded into one entry id across locales. */
+  term_translation_groups: number;
   /**
    * Password-protected posts imported as `draft`, whatever their WordPress
    * status. Their body is the protected text, and nothing downstream (query

package/dist/index.d.mts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.mts","names":[],"sources":["../src/wxr.ts","../src/rest.ts","../src/contentrain.ts","../src/comments.ts","../src/php-unserialize.ts","../src/core.ts","../src/delta.ts"],"mappings":";;;;UA2BiB,QAAA;EACf,UAAA;IAAc,QAAA;IAAkB,OAAA;IAAiB,MAAA;IAAgB,IAAA;EAAA;EACjE,UAAA;EACA,gBAAA;EACA,MAAA,EAAQ,MAAA;AAAA;AAAA,UAGO,cAAA;EACf,GAAA,EAAK,KAAA;EACL,KAAA,EAAO,QAAA;AAAA;AAAA,iBAiBa,QAAA,CAAS,KAAA,WAAgB,QAAA,EAAU,IAAA;EAAS,IAAA;AAAA,IAAkB,OAAA,CAAQ,cAAA;;;UCnC3E,iBAAA;EACf,MAAA;;EAEA,SAAA,UAAmB,KAAA;EDMI;;;;;;;;ECGvB,IAAA;IAAS,IAAA;IAAc,WAAA;EAAA;EACvB,OAAA;EDAc;AAGhB;;;;ECGE,WAAA;EDFK;;;;ECOL,QAAA;EACA,IAAA;AAAA;AAAA,UAGe,gBAAA;EACf,GAAA,EAAK,KAAA;EACL,QAAA;EDKkF;;;;;;;;ECIlF,UAAA;IAAc,MAAA;IAA0C,SAAA;EAAA;;;AAvC1D;;;;;;;;;;EAoDE,IAAA;AAAA;AAAA,iBA4EoB,cAAA,CAAe,OAAA,EAAS,iBAAA,GAAoB,OAAA,CAAQ,gBAAA;;;;UClHzD,QAAA;EACf,KAAA;EACA,IAAA;EFPuB;EESvB,IAAA;EFLc;EEOd,IAAA;EFVc;;;;EEed,MAAA;AAAA;AAAA,UAkBe,YAAA;EACf,cAAA;EACA,aAAA;EFhCc;EEkCd,UAAA,EAAY,QAAA;EACZ,cAAA;EACA,WAAA,EAAa,MAAA;EACb,UAAA,EAAY,MAAA;EFjCZ;EEmCA,WAAA,EAAa,MAAA;EFlCb;EEoCA,mBAAA,EAAqB,MAAA;EFpCN;EEsCf,mBAAA,EAAqB,MAAA;EFrBD;EEuBpB,oBAAA;;EAEA,iBAAA,EAAmB,MAAA;EFzBuE;EE2B1F,sBAAA,EAAwB,MAAA;EACxB,aAAA;EACA,iBAAA;EACA,MAAA,EAAQ,MAAA;IAAiB,IAAA;IAAc,MAAA;IAAgB,MAAA;IAAgB,OAAA;EAAA;EF9BiC;EEgCxG,OAAA;;EAEA,kBAAA;;ADrEF;;;;;EC4EE,yBAAA;EDzEmB;;;;EC8EnB,kBAAA;AAAA;AAAA,UAGe,iBAAA;EACf,KAAA,EAAO,MAAA;EACP,gBAAA,EAAkB,cAAA;EAClB,MAAA,EAAQ,YAAA;AAAA;AAAA,UAqDO,uBAAA;;EAEf,SAAA;EDjHA;;;;;;;ECyHA,aAAA;AAAA;AAAA,iBAqBc,gBAAA,CAAiB,KAAA,EAAO,KAAA,EAAO,IAAA,GAAO,uBAAA,GAA0B,iBAAA;;;;;;AFlKhF;;;iBGbgB,cAAA,CAAe,GAAA,EAAK,KAAA;EAAU,QAAA,EAAU,UAAA;EAAc,QAAA,EAAU,WAAA,CAAY,cAAA;AAAA;;;;;;;;;cAwB/E,oBAAA,EAAoB,MAAA;AAAA,UAuBhB,qBAAA;EACf,YAAA;EH1Be;;;;;;EGiCf,aAAA;AAAA;AAAA,iBAGc,mBAAA,CAAoB,GAAA,EAAK,KAAA,EAAO,OAAA,EAAS,cAAA,EAAgB,IAAA,GAAO,qBAAA,GAAwB,cAAA;;iBAoBxF,iBAAA,CAAkB,GAAA,EAAK,cAAA,GAAiB,eAAA;;;UCxFvC,iBAAA;EACf,EAAA;EACA,UAAA;EACA,KAAA;EACA,KAAA;AAAA;AAAA,iBAGc,eAAA,CAAgB,CAAA,YAAa,CAAA;AAAA,iBAI7B,cAAA,CAAe,GAAA;;iBAyFf,cAAA,CAAe,CAAA,WAAY,iBAAA;;;;;;;;cCrE9B,KAAA,GAAS,CAAA;;cAGT,KAAA,GAAS,CAAA;;cAcT,KAAA,GAAS,CAAA;;;;;;;;ALCtB;;cK8Ba,UAAA,EAAU,MAAA;;;;;;;cASV,iBAAA;AAAA,cACA,UAAA,GAAc,IAAA;;cAGd,WAAA,EAAW,MAAA;;cAOX,UAAA,GAAc,CAAA;AAAA,cAGd,OAAA,GAAW,CAAA;;AJxFxB;;;;;cIsGa,WAAA,GAAe,GAAA;;;;UCrGX,UAAA;EACf,KAAA,EAAO,MAAA;EACP,gBAAA,EAAkB,cAAA;AAAA;AAAA,UAGH,oBAAA;ENOD;EMLd,KAAA,EAAO,eAAA;ENEO;EMAd,KAAA,EAAO,UAAA;ENA0C;EMEjD,QAAA,GAAW,UAAA;ENDX;;;;;EMOA,UAAA;ENFe;;;;EMOf,SAAA;AAAA;AAAA,cAGW,kBAAA;AAAA,cACA,iBAAA;;;ANQb;;;;;;;;;;iBM8NgB,eAAA,CAAgB,KAAA,EAAO,oBAAA,GAAuB,eAAA;;iBAkC9C,uBAAA,CAAwB,IAAA,EAAM,eAAA"}
+{"version":3,"file":"index.d.mts","names":[],"sources":["../src/wxr.ts","../src/rest.ts","../src/contentrain.ts","../src/comments.ts","../src/php-unserialize.ts","../src/core.ts","../src/delta.ts"],"mappings":";;;;UA2BiB,QAAA;EACf,UAAA;IAAc,QAAA;IAAkB,OAAA;IAAiB,MAAA;IAAgB,IAAA;EAAA;EACjE,UAAA;EACA,gBAAA;EACA,MAAA,EAAQ,MAAA;AAAA;AAAA,UAGO,cAAA;EACf,GAAA,EAAK,KAAA;EACL,KAAA,EAAO,QAAA;AAAA;AAAA,iBAiBa,QAAA,CAAS,KAAA,WAAgB,QAAA,EAAU,IAAA;EAAS,IAAA;AAAA,IAAkB,OAAA,CAAQ,cAAA;;;UCnC3E,iBAAA;EACf,MAAA;;EAEA,SAAA,UAAmB,KAAA;EDMI;;;;;;;;ECGvB,IAAA;IAAS,IAAA;IAAc,WAAA;EAAA;EACvB,OAAA;EDAc;AAGhB;;;;ECGE,WAAA;EDFK;;;;ECOL,QAAA;EACA,IAAA;AAAA;AAAA,UAGe,gBAAA;EACf,GAAA,EAAK,KAAA;EACL,QAAA;EDKkF;;;;;;;;ECIlF,UAAA;IAAc,MAAA;IAA0C,SAAA;EAAA;;;AAvC1D;;;;;;;;;;EAoDE,IAAA;AAAA;AAAA,iBA4EoB,cAAA,CAAe,OAAA,EAAS,iBAAA,GAAoB,OAAA,CAAQ,gBAAA;;;;UClHzD,QAAA;EACf,KAAA;EACA,IAAA;EFPuB;EESvB,IAAA;EFLc;EEOd,IAAA;EFVc;;;;EEed,MAAA;AAAA;AAAA,UAkBe,YAAA;EACf,cAAA;EACA,aAAA;EFhCc;EEkCd,UAAA,EAAY,QAAA;EACZ,cAAA;EACA,WAAA,EAAa,MAAA;EACb,UAAA,EAAY,MAAA;EFjCZ;EEmCA,WAAA,EAAa,MAAA;EFlCb;EEoCA,mBAAA,EAAqB,MAAA;EFpCN;EEsCf,mBAAA,EAAqB,MAAA;EFrBD;EEuBpB,oBAAA;;EAEA,iBAAA,EAAmB,MAAA;EFzBuE;EE2B1F,sBAAA,EAAwB,MAAA;EACxB,aAAA;EACA,iBAAA;EACA,MAAA,EAAQ,MAAA;IAAiB,IAAA;IAAc,MAAA;IAAgB,MAAA;IAAgB,OAAA;EAAA;EF9BiC;EEgCxG,OAAA;;EAEA,kBAAA;;EAEA,uBAAA;EDvEgC;;;;;;EC8EhC,yBAAA;EDlES;;;;ECuET,kBAAA;AAAA;AAAA,UAGe,iBAAA;EACf,KAAA,EAAO,MAAA;EACP,gBAAA,EAAkB,cAAA;EAClB,MAAA,EAAQ,YAAA;AAAA;AAAA,UAqDO,uBAAA;EDjHf;ECmHA,SAAA;EDlHA;;;;;;;EC0HA,aAAA;AAAA;AAAA,iBAqBc,gBAAA,CAAiB,KAAA,EAAO,KAAA,EAAO,IAAA,GAAO,uBAAA,GAA0B,iBAAA;;;;;;AFpKhF;;;iBGbgB,cAAA,CAAe,GAAA,EAAK,KAAA;EAAU,QAAA,EAAU,UAAA;EAAc,QAAA,EAAU,WAAA,CAAY,cAAA;AAAA;;;;;;;;;cAwB/E,oBAAA,EAAoB,MAAA;AAAA,UAuBhB,qBAAA;EACf,YAAA;EH1Be;;;;;;EGiCf,aAAA;AAAA;AAAA,iBAGc,mBAAA,CAAoB,GAAA,EAAK,KAAA,EAAO,OAAA,EAAS,cAAA,EAAgB,IAAA,GAAO,qBAAA,GAAwB,cAAA;;iBAoBxF,iBAAA,CAAkB,GAAA,EAAK,cAAA,GAAiB,eAAA;;;UCxFvC,iBAAA;EACf,EAAA;EACA,UAAA;EACA,KAAA;EACA,KAAA;AAAA;AAAA,iBAGc,eAAA,CAAgB,CAAA,YAAa,CAAA;AAAA,iBAI7B,cAAA,CAAe,GAAA;;iBAyFf,cAAA,CAAe,CAAA,WAAY,iBAAA;;;;;;;;cCrE9B,KAAA,GAAS,CAAA;;cAGT,KAAA,GAAS,CAAA;;cAcT,KAAA,GAAS,CAAA;;;;;;;;ALCtB;;cK8Ba,UAAA,EAAU,MAAA;;;;;;;cASV,iBAAA;AAAA,cACA,UAAA,GAAc,IAAA;;cAGd,WAAA,EAAW,MAAA;;cAOX,UAAA,GAAc,CAAA;AAAA,cAGd,OAAA,GAAW,CAAA;;AJxFxB;;;;;cIsGa,WAAA,GAAe,GAAA;;;;UCrGX,UAAA;EACf,KAAA,EAAO,MAAA;EACP,gBAAA,EAAkB,cAAA;AAAA;AAAA,UAGH,oBAAA;ENOD;EMLd,KAAA,EAAO,eAAA;ENEO;EMAd,KAAA,EAAO,UAAA;ENA0C;EMEjD,QAAA,GAAW,UAAA;ENDX;;;;;EMOA,UAAA;ENFe;;;;EMOf,SAAA;AAAA;AAAA,cAGW,kBAAA;AAAA,cACA,iBAAA;;;ANQb;;;;;;;;;;iBM8NgB,eAAA,CAAgB,KAAA,EAAO,oBAAA,GAAuB,eAAA;;iBAkC9C,uBAAA,CAAwB,IAAA,EAAM,eAAA"}

package/dist/index.mjs

@@ -2233,7 +2233,9 @@ function rawToContentrain(input, opts) {
 	const localeContents = {};
 	const localeMetas = {};
 	const postLocale = (p) => normLocale(p.lang ?? p.terms.find((t) => t.taxonomy === "language")?.slug) || locale;
-	const locales = [locale, ...[...new Set(raw.posts.filter((p) => !SKIP_TYPES.test(p.type)).map(postLocale))].filter((l) => l !== locale).toSorted()];
+	const termLocale = (t) => normLocale(t.lang) || locale;
+	const termLocales = raw.terms.filter((t) => t.lang && !LANGUAGE_TAXONOMIES.has(t.taxonomy)).map(termLocale);
+	const locales = [locale, ...[...new Set([...raw.posts.filter((p) => !SKIP_TYPES.test(p.type)).map(postLocale), ...termLocales])].filter((l) => l !== locale).toSorted()];
 	const multilingual = locales.length > 1;
 	const postById = new Map(raw.posts.map((p) => [p.id, p]));
 	const canonicalOf = /* @__PURE__ */ new Map();
@@ -2267,6 +2269,7 @@ function rawToContentrain(input, opts) {
 		models: {},
 		locales,
 		translation_groups: translationGroups,
+		term_translation_groups: 0,
 		password_protected_drafts: 0,
 		site_title_missing: false
 	};
@@ -2291,6 +2294,19 @@ function rawToContentrain(input, opts) {
 	};
 	const authorId = (login) => hexId(`authors:${login}`);
 	const termId = (tax, slug) => hexId(`${tax}:${slug}`);
+	const termByWpId = new Map(raw.terms.filter((t) => t.id !== null).map((t) => [t.id, t]));
+	const canonicalTermSlug = /* @__PURE__ */ new Map();
+	for (const pair of raw.term_pairs ?? []) {
+		const members = [...new Set(Object.values(pair.translations))].map((id) => termByWpId.get(id)).filter((t) => t !== void 0 && t.taxonomy === pair.taxonomy);
+		if (members.length < 2 || new Set(members.map(termLocale)).size !== members.length) continue;
+		const canonical = members.find((t) => termLocale(t) === locale) ?? members.reduce((a, b) => a.id <= b.id ? a : b);
+		for (const t of members) canonicalTermSlug.set(`${t.taxonomy}:${t.slug}`, canonical.slug);
+		report.term_translation_groups++;
+	}
+	/** The entry id a reference to a term resolves to: its group's canonical member's. */
+	const termRef = (tax, slug) => termId(tax, canonicalTermSlug.get(`${tax}:${slug}`) ?? slug);
+	const i18nTaxonomies = new Set(multilingual ? raw.terms.filter((t) => t.lang).map((t) => t.taxonomy) : []);
+	const termExists = (mid, id) => contents[mid]?.[id] !== void 0 || Object.values(localeContents[mid] ?? {}).some((bucket) => bucket[id] !== void 0);
 	const usedSlugs = /* @__PURE__ */ new Map();
 	const postSlug = (p) => {
 		let s = addressSlug(p.slug);
@@ -2378,7 +2394,7 @@ function rawToContentrain(input, opts) {
 			const t = termById.get(id);
 			return t && models[taxModelId(t.taxonomy)] ? {
 				model: taxModelId(t.taxonomy),
-				ref: termId(t.taxonomy, t.slug)
+				ref: termRef(t.taxonomy, t.slug)
 			} : null;
 		}
 		return null;
@@ -2548,12 +2564,15 @@ function rawToContentrain(input, opts) {
 	const taxonomies = [...new Set(raw.terms.map((t) => t.taxonomy))].filter((t) => t !== "nav_menu" && t !== "post_format" && !LANGUAGE_TAXONOMIES.has(t));
 	for (const tax of taxonomies) {
 		const mid = taxModelId(tax);
+		const i18n = i18nTaxonomies.has(tax);
+		const covered = locales.filter((l) => l === locale || raw.terms.some((t) => t.taxonomy === tax && termLocale(t) === l));
 		addModel({
 			id: mid,
 			name: mid,
 			kind: "collection",
 			domain: "blog",
-			i18n: false,
+			i18n,
+			...i18n && covered.length < locales.length ? { locales: covered } : {},
 			title_field: "name",
 			fields: {
 				name: {
@@ -2588,12 +2607,15 @@ function rawToContentrain(input, opts) {
 			}
 		});
 		for (const t of raw.terms.filter((x) => x.taxonomy === tax)) {
-			const id = termId(tax, t.slug);
-			contents[mid][id] = pick({
+			const id = termRef(tax, t.slug);
+			const loc = termLocale(t);
+			const content = i18n ? (localeContents[mid] ??= {})[loc] ??= {} : contents[mid];
+			const meta = i18n ? (localeMetas[mid] ??= {})[loc] ??= {} : metas[mid];
+			content[id] = pick({
 				name: strip(t.name),
 				slug: addressSlug(t.slug) || `term-${t.id}`,
 				description: strip(t.description),
-				parent: t.parent && t.parent_resolved ? termId(tax, t.parent) : null,
+				parent: t.parent && t.parent_resolved ? termRef(tax, t.parent) : null,
 				wp_id: t.id
 			}, new Set([
 				"name",
@@ -2602,19 +2624,22 @@ function rawToContentrain(input, opts) {
 				"parent",
 				"wp_id"
 			]));
-			metas[mid][id] = importMeta("published");
+			meta[id] = importMeta("published");
 		}
 	}
 	for (const p of raw.posts) for (const t of p.terms) {
 		if (t.taxonomy === "nav_menu" || t.taxonomy === "post_format" || t.resolved) continue;
 		const mid = taxModelId(t.taxonomy);
 		if (!models[mid]) continue;
-		const id = termId(t.taxonomy, t.slug);
-		contents[mid][id] ??= {
+		const id = termRef(t.taxonomy, t.slug);
+		if (termExists(mid, id)) continue;
+		const content = models[mid].i18n ? (localeContents[mid] ??= {})[locale] ??= {} : contents[mid];
+		const meta = models[mid].i18n ? (localeMetas[mid] ??= {})[locale] ??= {} : metas[mid];
+		content[id] = {
 			name: strip(t.name),
 			slug: addressSlug(t.slug) || `term-${hexId(`${t.taxonomy}:${t.slug}`).slice(0, 8)}`
 		};
-		metas[mid][id] ??= importMeta("published");
+		meta[id] = importMeta("published");
 	}
 	const postTypes = [...new Set(raw.posts.map((p) => p.type))].filter((t) => {
 		if (SKIP_TYPES.test(t)) {
@@ -2886,7 +2911,7 @@ function rawToContentrain(input, opts) {
 			if (p.modified) e.modified_at = p.modified;
 			if (p.link) e.link = p.link;
 			if (fields.author && p.author) e.author = authorId(slugify(p.author));
-			for (const t of taxes) e[taxModelId(t)] = p.terms.filter((x) => x.taxonomy === t && contents[taxModelId(t)][termId(t, x.slug)]).map((x) => termId(t, x.slug));
+			for (const t of taxes) e[taxModelId(t)] = p.terms.filter((x) => x.taxonomy === t && termExists(taxModelId(t), termRef(t, x.slug))).map((x) => termRef(t, x.slug));
 			if (fields.cover && p.meta._thumbnail_id) {
 				const m = mediaRef(Number(p.meta._thumbnail_id));
 				if (m) e.cover = m;
@@ -2902,9 +2927,9 @@ function rawToContentrain(input, opts) {
 			if (fields.primary_category) {
 				const termWpId = primaryCategoryId(p);
 				const term = termWpId === void 0 ? void 0 : termById.get(termWpId);
-				const ref = term && term.taxonomy === "category" ? termId("category", term.slug) : void 0;
+				const ref = term && term.taxonomy === "category" ? termRef("category", term.slug) : void 0;
 				const assigned = ref !== void 0 && Array.isArray(e[taxModelId("category")]) && e[taxModelId("category")].includes(ref);
-				if (ref && assigned && contents[taxModelId("category")]?.[ref]) e.primary_category = ref;
+				if (ref && assigned && termExists(taxModelId("category"), ref)) e.primary_category = ref;
 				else if (termWpId !== void 0) report.dropped_relations++;
 			}
 			if (fields.template && typeof p.meta._wp_page_template === "string" && p.meta._wp_page_template !== "default") e.template = p.meta._wp_page_template;
@@ -3087,7 +3112,7 @@ function rawToContentrain(input, opts) {
 				}
 				if (i.target.kind === "term" && i.target.resolved && i.target.slug && models[taxModelId(i.target.taxonomy)]) e.target = toTarget.ref({
 					model: taxModelId(i.target.taxonomy),
-					ref: termId(i.target.taxonomy, i.target.slug)
+					ref: termRef(i.target.taxonomy, i.target.slug)
 				});
 				if (i.classes?.length) e.classes = i.classes;
 				if (i.description) e.description = strip(i.description);