@contentrain/wp-import 0.10.4 → 0.11.0

package/dist/index.d.mts

@@ -84,9 +84,25 @@ interface RestImportResult {
 declare function fetchRestRawIR(options: RestImportOptions): Promise<RestImportResult>;
 //#endregion
 //#region src/contentrain.d.ts
+/** A post whose imported slug is not the address it was served at: it gets a 301 from `from`. */
+interface SlugMove {
+  wp_id: number;
+  type: string;
+  /** The path the source site served it at (its `link`), when the source named one. */
+  from: string | null;
+  /** The slug the entry was written under. */
+  slug: string;
+  /**
+   * `empty`: the source slug left no word, `collision`: another entry of the same type already holds it,
+   * `rewritten`: characters outside the slug alphabet (ZWNJ/ZWJ, `・`, an NFD spelling...) were changed.
+   */
+  reason: 'empty' | 'collision' | 'rewritten';
+}
 interface ImportReport {
   slug_rewritten: number;
   slug_fallback: number;
+  /** Posts whose address changed (empty or colliding slug): each needs a 301 from its old address. */
+  slug_moves: SlugMove[];
   title_fallback: number;
   meta_fields: Record<string, string>;
   acf_fields: Record<string, string>;
@@ -183,6 +199,13 @@ declare const PLUGIN_META: RegExp;
 /** Taxonomy → model id. */
 declare const taxModelId: (t: string) => string;
 declare const slugify: (s: unknown) => string;
+/**
+ * A WordPress `post_name` as the address it is: percent-encoded UTF-8 decoded, NFC, lowercase, anything outside letters,
+ * digits and marks (the contract's slug alphabet) a hyphen. Nothing is transliterated: `日本語`, `ışık` and `straße`
+ * stay what the source site served, because the slug is the page's address. ASCII comes out exactly as `slugify`
+ * gives it. Empty when nothing of a word is left, or when the slug would not fit a file name (200 bytes).
+ */
+declare const addressSlug: (raw: string) => string;
 //#endregion
 //#region src/delta.d.ts
 /** A `.contentrain` store as a file map (repo-relative path → file text), with the source map it was built from. */
@@ -227,5 +250,5 @@ declare function planSourceDelta(input: PlanSourceDeltaInput): SourceDeltaPlan;
 /** The plan as text, for a dry run: one line per record, then conflicts, unmapped records, redirects and warnings. */
 declare function formatSourceDeltaReport(plan: SourceDeltaPlan): string;
 //#endregion
-export { type ContentrainResult, DEFAULT_IMPORTERS, DEFAULT_TAXONOMIES, type DeltaStore, type ImportReport, PLUGIN_META, type PlanSourceDeltaInput, type RestImportOptions, type RestImportResult, SKIP_TYPES, type UnserializeResult, type WxrParseResult, type WxrStats, buildCommentsExport, canon, fetchRestRawIR, formatSourceDeltaReport, hexId, looksSerialized, parseWxr, phpUnserialize, planSourceDelta, rawToContentrain, selectComments, slugify, strip, summarizeComments, taxModelId, tryUnserialize };
+export { type ContentrainResult, DEFAULT_IMPORTERS, DEFAULT_TAXONOMIES, type DeltaStore, type ImportReport, PLUGIN_META, type PlanSourceDeltaInput, type RestImportOptions, type RestImportResult, SKIP_TYPES, type UnserializeResult, type WxrParseResult, type WxrStats, addressSlug, buildCommentsExport, canon, fetchRestRawIR, formatSourceDeltaReport, hexId, looksSerialized, parseWxr, phpUnserialize, planSourceDelta, rawToContentrain, selectComments, slugify, strip, summarizeComments, taxModelId, tryUnserialize };
 //# sourceMappingURL=index.d.mts.map

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,iBA2EoB,cAAA,CAAe,OAAA,EAAS,iBAAA,GAAoB,OAAA,CAAQ,gBAAA;;;UCrHzD,YAAA;EACf,cAAA;EACA,aAAA;EACA,cAAA;EACA,WAAA,EAAa,MAAA;EACb,UAAA,EAAY,MAAA;EFFE;EEId,WAAA,EAAa,MAAA;EFPC;EESd,mBAAA,EAAqB,MAAA;EFT4B;EEWjD,mBAAA,EAAqB,MAAA;EFVrB;EEYA,oBAAA;EACA,aAAA;EACA,iBAAA;EACA,MAAA,EAAQ,MAAA;IAAiB,IAAA;IAAc,MAAA;IAAgB,MAAA;IAAgB,OAAA;EAAA;EFTvE;EEWA,OAAA;EFVA;EEYA,kBAAA;EFZe;;AAiBjB;;;;EEEE,yBAAA;EFFkF;;;;EEOlF,kBAAA;AAAA;AAAA,UAGe,iBAAA;EACf,KAAA,EAAO,MAAA;EACP,gBAAA,EAAkB,cAAA;EAClB,MAAA,EAAQ,YAAA;AAAA;AAAA,iBA0CM,gBAAA,CAAiB,GAAA,EAAK,KAAA,EAAO,IAAA;EAAS,SAAA;AAAA,IAAuB,iBAAA;;;;;;AFjF7E;;;iBGbgB,cAAA,CAAe,GAAA,EAAK,KAAA;EAAU,QAAA,EAAU,UAAA;EAAc,QAAA,EAAU,WAAA,CAAY,cAAA;AAAA;AAAA,iBAgB5E,mBAAA,CAAoB,GAAA,EAAK,KAAA,EAAO,OAAA,EAAS,cAAA,EAAgB,IAAA;EAAS,YAAA;AAAA,IAA0B,cAAA;;iBAiB5F,iBAAA,CAAkB,GAAA,EAAK,cAAA,GAAiB,eAAA;;;UC3CvC,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;;;;cCzE9B,KAAA,GAAS,CAAA;;cAGT,KAAA,GAAS,CAAA;;cAcT,KAAA,GAAS,CAAA;;ALdtB;;;;;;;;cK6Ca,UAAA,EAAU,MAAA;;cAIV,WAAA,EAAW,MAAA;;cAOX,UAAA,GAAc,CAAA;AAAA,cAGd,OAAA,GAAW,CAAA;;;;UC1EP,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,iBA2EoB,cAAA,CAAe,OAAA,EAAS,iBAAA,GAAoB,OAAA,CAAQ,gBAAA;;;;UCnHzD,QAAA;EACf,KAAA;EACA,IAAA;EFLuB;EEOvB,IAAA;EFHc;EEKd,IAAA;EFRc;;;;EEad,MAAA;AAAA;AAAA,UAkBe,YAAA;EACf,cAAA;EACA,aAAA;EF9Bc;EEgCd,UAAA,EAAY,QAAA;EACZ,cAAA;EACA,WAAA,EAAa,MAAA;EACb,UAAA,EAAY,MAAA;EF/BZ;EEiCA,WAAA,EAAa,MAAA;EFhCb;EEkCA,mBAAA,EAAqB,MAAA;EFlCN;EEoCf,mBAAA,EAAqB,MAAA;EFnBD;EEqBpB,oBAAA;EACA,aAAA;EACA,iBAAA;EACA,MAAA,EAAQ,MAAA;IAAiB,IAAA;IAAc,MAAA;IAAgB,MAAA;IAAgB,OAAA;EAAA;EFxBP;EE0BhE,OAAA;EF1BkF;EE4BlF,kBAAA;EF5BwG;;;;;ACnC1G;ECsEE,yBAAA;;;;;EAKA,kBAAA;AAAA;AAAA,UAGe,iBAAA;EACf,KAAA,EAAO,MAAA;EACP,gBAAA,EAAkB,cAAA;EAClB,MAAA,EAAQ,YAAA;AAAA;AAAA,iBA0CM,gBAAA,CAAiB,GAAA,EAAK,KAAA,EAAO,IAAA;EAAS,SAAA;AAAA,IAAuB,iBAAA;;;;;;AFlH7E;;;iBGbgB,cAAA,CAAe,GAAA,EAAK,KAAA;EAAU,QAAA,EAAU,UAAA;EAAc,QAAA,EAAU,WAAA,CAAY,cAAA;AAAA;AAAA,iBAgB5E,mBAAA,CAAoB,GAAA,EAAK,KAAA,EAAO,OAAA,EAAS,cAAA,EAAgB,IAAA;EAAS,YAAA;AAAA,IAA0B,cAAA;;iBAiB5F,iBAAA,CAAkB,GAAA,EAAK,cAAA,GAAiB,eAAA;;;UC3CvC,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;;;;cCzE9B,KAAA,GAAS,CAAA;;cAGT,KAAA,GAAS,CAAA;;cAcT,KAAA,GAAS,CAAA;;ALdtB;;;;;;;;cK6Ca,UAAA,EAAU,MAAA;;cAIV,WAAA,EAAW,MAAA;;cAOX,UAAA,GAAc,CAAA;AAAA,cAGd,OAAA,GAAW,CAAA;;;;;;;cAcX,WAAA,GAAe,GAAA;;;;UCxFX,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

@@ -181,6 +181,16 @@ const PLUGIN_META = /^(jetpack|wpdc|discourse|footnotes|inline_featured|spay_|ad
 /** Taxonomy → model id. */
 const taxModelId = (t) => t === "post_tag" ? "tags" : t === "category" ? "categories" : t.replace(/_/g, "-");
 const slugify = (s) => String(s ?? "").normalize("NFKD").replace(/[̀-ͯ]/g, "").toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "");
+/**
+* A WordPress `post_name` as the address it is: percent-encoded UTF-8 decoded, NFC, lowercase, anything outside letters,
+* digits and marks (the contract's slug alphabet) a hyphen. Nothing is transliterated: `日本語`, `ışık` and `straße`
+* stay what the source site served, because the slug is the page's address. ASCII comes out exactly as `slugify`
+* gives it. Empty when nothing of a word is left, or when the slug would not fit a file name (200 bytes).
+*/
+const addressSlug = (raw) => {
+	const slug = decodeSlug(raw).normalize("NFC").toLowerCase().normalize("NFC").replace(/[^\p{Ll}\p{Lo}\p{Lm}\p{N}\p{M}]+/gu, "-").replace(/^-+|-+$/g, "");
+	return new TextEncoder().encode(slug).length > 200 ? "" : slug;
+};
 const decodeSlug = (s) => {
 	try {
 		return decodeURIComponent(s);
@@ -2102,6 +2112,19 @@ async function fetchRestRawIR(options) {
 }
 //#endregion
 //#region src/contentrain.ts
+const utf8Length = (s) => new TextEncoder().encode(s).length;
+/** `s` cut on a code-point boundary to at most `max` UTF-8 bytes, without a trailing hyphen. */
+const fitBytes = (s, max) => {
+	let out = "";
+	let bytes = 0;
+	for (const ch of s) {
+		const n = utf8Length(ch);
+		if (bytes + n > max) break;
+		out += ch;
+		bytes += n;
+	}
+	return out.replace(/-+$/, "");
+};
 const pick = (o, keys) => Object.fromEntries(Object.entries(o).filter(([k, v]) => keys.has(k) && v !== void 0 && v !== null && v !== ""));
 const typeModelId = (t) => t === "post" ? "posts" : t === "page" ? "pages" : t.replace(/_/g, "-");
 const domainOf = (t) => t === "post" ? "blog" : t === "page" ? "site" : "custom";
@@ -2162,6 +2185,7 @@ function rawToContentrain(raw, opts) {
 	const report = {
 		slug_rewritten: 0,
 		slug_fallback: 0,
+		slug_moves: [],
 		title_fallback: 0,
 		meta_fields: {},
 		acf_fields: {},
@@ -2200,15 +2224,44 @@ function rawToContentrain(raw, opts) {
 	const termId = (tax, slug) => hexId(`${tax}:${slug}`);
 	const usedSlugs = /* @__PURE__ */ new Map();
 	const postSlug = (p) => {
-		let s = slugify(decodeSlug(p.slug));
+		let s = addressSlug(p.slug);
+		const moved = (reason, slug) => {
+			let from = null;
+			try {
+				from = p.link ? new URL(p.link).pathname : null;
+			} catch {
+				from = null;
+			}
+			report.slug_moves.push({
+				wp_id: p.id,
+				type: p.type,
+				from,
+				slug,
+				reason
+			});
+		};
+		let empty = false;
+		let rewritten = false;
 		if (!s) {
 			s = `${p.type}-${p.id}`;
+			empty = true;
 			report.slug_fallback++;
-		} else if (s !== p.slug) report.slug_rewritten++;
+		} else if (s !== decodeSlug(p.slug)) {
+			report.slug_rewritten++;
+			rewritten = s !== decodeSlug(p.slug).toLowerCase();
+		}
 		const key = multilingual ? `${p.type}:${postLocale(p)}:${s}` : `${p.type}:${s}`;
 		const n = usedSlugs.get(key) ?? 0;
 		usedSlugs.set(key, n + 1);
-		return n ? `${s}-${p.id}` : s;
+		if (n) {
+			const suffix = `-${p.id}`;
+			const unique = `${fitBytes(s, 200 - utf8Length(suffix))}${suffix}`;
+			moved("collision", unique);
+			return unique;
+		}
+		if (empty) moved("empty", s);
+		else if (rewritten) moved("rewritten", s);
+		return s;
 	};
 	const slugOf = /* @__PURE__ */ new Map();
 	const postEntry = /* @__PURE__ */ new Map();
@@ -2448,7 +2501,7 @@ function rawToContentrain(raw, opts) {
 			const id = termId(tax, t.slug);
 			contents[mid][id] = pick({
 				name: strip(t.name),
-				slug: slugify(decodeSlug(t.slug)) || `term-${t.id}`,
+				slug: addressSlug(t.slug) || `term-${t.id}`,
 				description: strip(t.description),
 				parent: t.parent && t.parent_resolved ? termId(tax, t.parent) : null,
 				wp_id: t.id
@@ -2469,7 +2522,7 @@ function rawToContentrain(raw, opts) {
 		const id = termId(t.taxonomy, t.slug);
 		contents[mid][id] ??= {
 			name: strip(t.name),
-			slug: slugify(t.slug)
+			slug: addressSlug(t.slug) || `term-${hexId(`${t.taxonomy}:${t.slug}`).slice(0, 8)}`
 		};
 		metas[mid][id] ??= importMeta("published");
 	}
@@ -2565,7 +2618,7 @@ function rawToContentrain(raw, opts) {
 	const mediaSlugs = /* @__PURE__ */ new Set();
 	for (const a of raw.attachments) {
 		const id = hexId(`media:${a.id}`);
-		let slug = slugify(decodeSlug(a.slug)) || `media-${a.id}`;
+		let slug = addressSlug(a.slug) || `media-${a.id}`;
 		if (mediaSlugs.has(slug)) slug = `${slug}-${a.id}`;
 		mediaSlugs.add(slug);
 		const target = a.parent ? postEntry.get(a.parent) : void 0;
@@ -2889,7 +2942,7 @@ function rawToContentrain(raw, opts) {
 			const mid = hexId(`menus:${m.slug}`);
 			contents.menus[mid] = {
 				name: strip(m.name),
-				slug: slugify(m.slug) || `menu-${m.id}`,
+				slug: addressSlug(m.slug) || `menu-${m.id}`,
 				items: m.items.map((i) => itemRef(i.id)),
 				...m.id !== null && m.id > 0 ? { wp_id: m.id } : {},
 				...m.locations?.length ? { locations: m.locations } : {}
@@ -2905,7 +2958,7 @@ function rawToContentrain(raw, opts) {
 					open_in_new_tab: i.target_attr === "_blank"
 				};
 				if (i.parent && !i.parent_unresolved) e.parent = itemRef(i.parent);
-				if (i.target.kind === "url" && i.url) e.url = i.url;
+				if ((i.target.kind === "url" || i.target.kind === "archive") && i.url) e.url = i.url;
 				if (i.target.kind === "post" && i.target.resolved && i.target.id != null) {
 					const pe = postEntry.get(i.target.id);
 					if (pe) e.target = toTarget.ref(pe);
@@ -3492,6 +3545,6 @@ function formatSourceDeltaReport(plan) {
 	return `${lines.join("\n")}\n`;
 }
 //#endregion
-export { DEFAULT_IMPORTERS, DEFAULT_TAXONOMIES, PLUGIN_META, SKIP_TYPES, buildCommentsExport, canon, fetchRestRawIR, formatSourceDeltaReport, hexId, looksSerialized, parseWxr, phpUnserialize, planSourceDelta, rawToContentrain, selectComments, slugify, strip, summarizeComments, taxModelId, tryUnserialize };
+export { DEFAULT_IMPORTERS, DEFAULT_TAXONOMIES, PLUGIN_META, SKIP_TYPES, addressSlug, buildCommentsExport, canon, fetchRestRawIR, formatSourceDeltaReport, hexId, looksSerialized, parseWxr, phpUnserialize, planSourceDelta, rawToContentrain, selectComments, slugify, strip, summarizeComments, taxModelId, tryUnserialize };
 
 //# sourceMappingURL=index.mjs.map