@cogenta/import 0.1.0

package/dist/wordpress/import.js

@@ -0,0 +1,241 @@
+import { createUserStore, ensureAuthTables } from '@cogenta/auth';
+import { createDatabaseMediaStore } from '@cogenta/core';
+import { buildPath, createContentStore, createRedirectStore, createSchemaTables, } from '@cogenta/schema';
+import { WORDPRESS_IMPORT_COLLECTIONS, wpCategory, wpComment, wpPage, wpPost, wpTag, } from './collections.js';
+import { convertContent, mediaUrlsOf, resolveMediaReferences, } from './content-convert.js';
+import { downloadAndStoreMedia } from './media.js';
+import { parseWxr } from './parse.js';
+import { emptyReport } from './report.js';
+const APPROVED = '1';
+const IMPORTABLE_STATUSES = new Set(['publish', 'draft', 'pending', 'private', 'future']);
+function wpDateToIso(wpDate) {
+    if (wpDate.length === 0 || wpDate === '0000-00-00 00:00:00')
+        return null;
+    const iso = wpDate.includes('T') ? wpDate : `${wpDate.replace(' ', 'T')}Z`;
+    const date = new Date(iso);
+    return Number.isNaN(date.getTime()) ? null : date.toISOString();
+}
+function toContentBlocks(blocks) {
+    return blocks.map((block) => {
+        const { _key, _type, ...data } = block;
+        return { key: _key, type: _type, data: data };
+    });
+}
+function relativePathOf(link, baseUrl) {
+    try {
+        const url = new URL(link, baseUrl || link);
+        return url.pathname;
+    }
+    catch {
+        return null;
+    }
+}
+/**
+ * Imports a WordPress "Export All Content" WXR file into Cogenta.
+ *
+ * Every step degrades to a report entry rather than aborting the import: a
+ * dead media URL, an unmappable Gutenberg block, an author with no email, a
+ * redirect that would loop — none of them stop the run, because the lot's own
+ * framing is explicit that a reported partial loss beats a silent one. The
+ * only thing that stops the run is a document `parseWxr` cannot read at all.
+ */
+export async function importWordPress(xml, options) {
+    const { db, storage } = options;
+    const parsed = parseWxr(xml);
+    const acc = emptyReport();
+    await createSchemaTables(db, WORDPRESS_IMPORT_COLLECTIONS);
+    await ensureAuthTables(db);
+    const users = createUserStore(db);
+    const redirects = createRedirectStore({ db });
+    await redirects.ensureTable();
+    const mediaStore = createDatabaseMediaStore({ db });
+    const categoryStore = createContentStore({ db, collection: wpCategory });
+    const tagStore = createContentStore({ db, collection: wpTag });
+    const postStore = createContentStore({ db, collection: wpPost });
+    const pageStore = createContentStore({ db, collection: wpPage });
+    const commentStore = createContentStore({ db, collection: wpComment });
+    // ---- Authors -------------------------------------------------------
+    const loginToUserId = new Map();
+    for (const author of parsed.authors) {
+        const email = author.email.trim().length > 0
+            ? author.email.trim()
+            : `${author.login || 'author'}@imported.invalid`;
+        if (author.email.trim().length === 0) {
+            acc.warnings.push(`Author "${author.login}" had no email in the export; used a placeholder (${email}).`);
+        }
+        try {
+            const user = await users.create({ email, roles: ['author'] });
+            loginToUserId.set(author.login, user.id);
+            acc.imported.authors += 1;
+        }
+        catch {
+            // AUTH_USER_EXISTS: a previous import (or the site itself) already has
+            // this email — reuse that account rather than failing the whole run.
+            const existing = await users.byEmail(email);
+            if (existing !== null)
+                loginToUserId.set(author.login, existing.id);
+        }
+    }
+    // ---- Categories & tags ----------------------------------------------
+    const categoryByNiceName = new Map();
+    for (const category of parsed.categories) {
+        const entry = await categoryStore.create({
+            values: { name: category.name || category.niceName, slug: category.niceName },
+            status: 'published',
+        });
+        categoryByNiceName.set(category.niceName, entry.id);
+        acc.imported.categories += 1;
+    }
+    const tagByNiceName = new Map();
+    for (const tag of parsed.tags) {
+        const entry = await tagStore.create({
+            values: { name: tag.name || tag.slug, slug: tag.slug },
+            status: 'published',
+        });
+        tagByNiceName.set(tag.slug, entry.id);
+        acc.imported.tags += 1;
+    }
+    // ---- Media --------------------------------------------------------
+    const posts = parsed.items.filter((item) => item.postType === 'post');
+    const pages = parsed.items.filter((item) => item.postType === 'page');
+    const attachments = parsed.items.filter((item) => item.postType === 'attachment');
+    const attachmentUrlById = new Map(attachments
+        .filter((item) => item.attachmentUrl !== null)
+        .map((item) => [item.postId, item.attachmentUrl]));
+    const converted = new Map();
+    const allMediaUrls = new Set();
+    for (const item of [...posts, ...pages]) {
+        const { blocks, notes } = convertContent(item.contentEncoded);
+        converted.set(item.postId, { blocks, notes: [...notes] });
+        for (const url of mediaUrlsOf(blocks))
+            allMediaUrls.add(url);
+        const thumbnail = item.thumbnailId === null ? undefined : attachmentUrlById.get(item.thumbnailId);
+        if (thumbnail !== undefined)
+            allMediaUrls.add(thumbnail);
+    }
+    for (const url of attachmentUrlById.values())
+        allMediaUrls.add(url);
+    const mediaResult = await downloadAndStoreMedia([...allMediaUrls], {
+        mediaStore,
+        storage,
+        createdBy: null,
+        ...(options.fetchImpl === undefined ? {} : { fetchImpl: options.fetchImpl }),
+    });
+    acc.imported.media = mediaResult.imported.length;
+    for (const failure of mediaResult.failed) {
+        acc.warnings.push(`Media "${failure.url}" could not be downloaded: ${failure.reason}`);
+    }
+    for (const asset of mediaResult.imported) {
+        acc.warnings.push(`Media "${asset.filename}" was imported with a synthesised alt text; review it.`);
+    }
+    // ---- Posts & pages --------------------------------------------------
+    async function writeEntry(item, store, kind) {
+        if (item.status === 'trash') {
+            acc.skipped.push({
+                type: kind,
+                wpId: item.postId,
+                title: item.title,
+                reason: 'Trashed in WordPress; not imported.',
+            });
+            return null;
+        }
+        if (!IMPORTABLE_STATUSES.has(item.status)) {
+            acc.skipped.push({
+                type: kind,
+                wpId: item.postId,
+                title: item.title,
+                reason: `Unrecognised WordPress status "${item.status}".`,
+            });
+            return null;
+        }
+        const draft = converted.get(item.postId);
+        const notes = draft?.notes ?? [];
+        const resolvedBlocks = resolveMediaReferences(draft?.blocks ?? [], mediaResult.urlToMediaId, notes);
+        for (const note of notes) {
+            acc.unconvertedBlocks.push({
+                ...note,
+                postTitle: item.title || `(untitled ${kind} ${item.postId})`,
+            });
+        }
+        const status = item.status === 'publish' ? 'published' : 'draft';
+        const createdBy = loginToUserId.get(item.creator) ?? null;
+        if (createdBy === null && item.creator.length > 0) {
+            acc.warnings.push(`Author "${item.creator}" of "${item.title}" was not found among the export's authors.`);
+        }
+        const customFields = Object.fromEntries(item.postMeta
+            .filter((meta) => !meta.key.startsWith('_'))
+            .map((meta) => [meta.key, meta.value]));
+        const values = {
+            title: item.title || `(untitled ${item.postId})`,
+            slug: item.postName || item.postId,
+            publishedAt: wpDateToIso(item.postDate),
+            customFields,
+        };
+        if (kind === 'post') {
+            values['excerpt'] = item.excerptEncoded;
+            const categoryRef = item.categories.find((ref) => ref.domain === 'category');
+            const category = categoryRef === undefined ? null : (categoryByNiceName.get(categoryRef.niceName) ?? null);
+            if (category !== null)
+                values['category'] = category;
+            values['tags'] = item.categories
+                .filter((ref) => ref.domain === 'post_tag')
+                .map((ref) => tagByNiceName.get(ref.niceName))
+                .filter((id) => id !== undefined);
+        }
+        const entry = await store.create({
+            id: item.postId,
+            status,
+            createdBy,
+            values,
+            blocks: { body: toContentBlocks(resolvedBlocks) },
+        });
+        const collection = kind === 'post' ? wpPost : wpPage;
+        const to = buildPath(collection, { slug: values['slug'] });
+        const from = relativePathOf(item.link, parsed.baseUrl);
+        if (from !== null && from !== to) {
+            try {
+                await redirects.add({ from, to, reason: 'import', collection: kind, entryId: entry.id });
+                acc.redirectsCreated += 1;
+            }
+            catch (error) {
+                acc.warnings.push(`Redirect ${from} → ${to} was not created: ${error instanceof Error ? error.message : String(error)}`);
+            }
+        }
+        return entry.id;
+    }
+    for (const item of posts) {
+        const id = await writeEntry(item, postStore, 'post');
+        if (id !== null) {
+            acc.imported.posts += 1;
+            for (const comment of item.comments) {
+                if (comment.approved !== APPROVED)
+                    continue;
+                await commentStore.create({
+                    status: 'published',
+                    values: {
+                        post: id,
+                        author: comment.author || 'Anonymous',
+                        authorEmail: comment.authorEmail,
+                        body: comment.content,
+                        publishedAt: wpDateToIso(comment.date),
+                    },
+                });
+                acc.imported.comments += 1;
+            }
+        }
+    }
+    for (const item of pages) {
+        const id = await writeEntry(item, pageStore, 'page');
+        if (id !== null)
+            acc.imported.pages += 1;
+    }
+    const skipped = acc.skipped;
+    return {
+        imported: acc.imported,
+        redirectsCreated: acc.redirectsCreated,
+        skipped,
+        unconvertedBlocks: acc.unconvertedBlocks,
+        warnings: acc.warnings,
+    };
+}
+//# sourceMappingURL=import.js.map

package/dist/wordpress/import.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"import.js","sourceRoot":"","sources":["../../src/wordpress/import.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,eAAe,EAAE,gBAAgB,EAAE,MAAM,eAAe,CAAA;AACjE,OAAO,EAAE,wBAAwB,EAA2C,MAAM,eAAe,CAAA;AACjG,OAAO,EACL,SAAS,EACT,kBAAkB,EAClB,mBAAmB,EACnB,kBAAkB,GACnB,MAAM,iBAAiB,CAAA;AACxB,OAAO,EACL,4BAA4B,EAC5B,UAAU,EACV,SAAS,EACT,MAAM,EACN,MAAM,EACN,KAAK,GACN,MAAM,kBAAkB,CAAA;AACzB,OAAO,EAEL,cAAc,EAEd,WAAW,EACX,sBAAsB,GACvB,MAAM,sBAAsB,CAAA;AAC7B,OAAO,EAAE,qBAAqB,EAAE,MAAM,YAAY,CAAA;AAClD,OAAO,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAA;AACrC,OAAO,EAAyB,WAAW,EAAwB,MAAM,aAAa,CAAA;AAUtF,MAAM,QAAQ,GAAG,GAAG,CAAA;AACpB,MAAM,mBAAmB,GAAG,IAAI,GAAG,CAAC,CAAC,SAAS,EAAE,OAAO,EAAE,SAAS,EAAE,SAAS,EAAE,QAAQ,CAAC,CAAC,CAAA;AAEzF,SAAS,WAAW,CAAC,MAAc;IACjC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,KAAK,qBAAqB;QAAE,OAAO,IAAI,CAAA;IACxE,MAAM,GAAG,GAAG,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,MAAM,CAAC,OAAO,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,CAAA;IAC1E,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,GAAG,CAAC,CAAA;IAC1B,OAAO,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,WAAW,EAAE,CAAA;AACjE,CAAC;AAeD,SAAS,eAAe,CAAC,MAA6B;IACpD,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;QAC1B,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,IAAI,EAAE,GAAG,KAAK,CAAA;QACtC,OAAO,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAA+B,EAAE,CAAA;IAC1E,CAAC,CAAC,CAAA;AACJ,CAAC;AAED,SAAS,cAAc,CAAC,IAAY,EAAE,OAAe;IACnD,IAAI,CAAC;QACH,MAAM,GAAG,GAAG,IAAI,GAAG,CAAC,IAAI,EAAE,OAAO,IAAI,IAAI,CAAC,CAAA;QAC1C,OAAO,GAAG,CAAC,QAAQ,CAAA;IACrB,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,CAAA;IACb,CAAC;AACH,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,CAAC,KAAK,UAAU,eAAe,CACnC,GAAW,EACX,OAA+B;IAE/B,MAAM,EAAE,EAAE,EAAE,OAAO,EAAE,GAAG,OAAO,CAAA;IAC/B,MAAM,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAA;IAC5B,MAAM,GAAG,GAAG,WAAW,EAAE,CAAA;IAEzB,MAAM,kBAAkB,CAAC,EAAE,EAAE,4BAA4B,CAAC,CAAA;IAC1D,MAAM,gBAAgB,CAAC,EAAE,CAAC,CAAA;IAE1B,MAAM,KAAK,GAAG,eAAe,CAAC,EAAE,CAAC,CAAA;IACjC,MAAM,SAAS,GAAG,mBAAmB,CAAC,EAAE,EAAE,EAAE,CAAC,CAAA;IAC7C,MAAM,SAAS,CAAC,WAAW,EAAE,CAAA;IAC7B,MAAM,UAAU,GAAG,wBAAwB,CAAC,EAAE,EAAE,EAAE,CAAC,CAAA;IAEnD,MAAM,aAAa,GAAG,kBAAkB,CAAC,EAAE,EAAE,EAAE,UAAU,EAAE,UAAU,EAAE,CAAC,CAAA;IACxE,MAAM,QAAQ,GAAG,kBAAkB,CAAC,EAAE,EAAE,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC,CAAA;IAC9D,MAAM,SAAS,GAAG,kBAAkB,CAAC,EAAE,EAAE,EAAE,UAAU,EAAE,MAAM,EAAE,CAAC,CAAA;IAChE,MAAM,SAAS,GAAG,kBAAkB,CAAC,EAAE,EAAE,EAAE,UAAU,EAAE,MAAM,EAAE,CAAC,CAAA;IAChE,MAAM,YAAY,GAAG,kBAAkB,CAAC,EAAE,EAAE,EAAE,UAAU,EAAE,SAAS,EAAE,CAAC,CAAA;IAEtE,uEAAuE;IACvE,MAAM,aAAa,GAAG,IAAI,GAAG,EAAkB,CAAA;IAC/C,KAAK,MAAM,MAAM,IAAI,MAAM,CAAC,OAAO,EAAE,CAAC;QACpC,MAAM,KAAK,GACT,MAAM,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;YAC5B,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,EAAE;YACrB,CAAC,CAAC,GAAG,MAAM,CAAC,KAAK,IAAI,QAAQ,mBAAmB,CAAA;QACpD,IAAI,MAAM,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACrC,GAAG,CAAC,QAAQ,CAAC,IAAI,CACf,WAAW,MAAM,CAAC,KAAK,qDAAqD,KAAK,IAAI,CACtF,CAAA;QACH,CAAC;QACD,IAAI,CAAC;YACH,MAAM,IAAI,GAAG,MAAM,KAAK,CAAC,MAAM,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAA;YAC7D,aAAa,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,EAAE,CAAC,CAAA;YACxC,GAAG,CAAC,QAAQ,CAAC,OAAO,IAAI,CAAC,CAAA;QAC3B,CAAC;QAAC,MAAM,CAAC;YACP,uEAAuE;YACvE,qEAAqE;YACrE,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;YAC3C,IAAI,QAAQ,KAAK,IAAI;gBAAE,aAAa,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,EAAE,CAAC,CAAA;QACrE,CAAC;IACH,CAAC;IAED,wEAAwE;IACxE,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAkB,CAAA;IACpD,KAAK,MAAM,QAAQ,IAAI,MAAM,CAAC,UAAU,EAAE,CAAC;QACzC,MAAM,KAAK,GAAG,MAAM,aAAa,CAAC,MAAM,CAAC;YACvC,MAAM,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,QAAQ,EAAE,IAAI,EAAE,QAAQ,CAAC,QAAQ,EAAE;YAC7E,MAAM,EAAE,WAAW;SACpB,CAAC,CAAA;QACF,kBAAkB,CAAC,GAAG,CAAC,QAAQ,CAAC,QAAQ,EAAE,KAAK,CAAC,EAAE,CAAC,CAAA;QACnD,GAAG,CAAC,QAAQ,CAAC,UAAU,IAAI,CAAC,CAAA;IAC9B,CAAC;IAED,MAAM,aAAa,GAAG,IAAI,GAAG,EAAkB,CAAA;IAC/C,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,EAAE,CAAC;QAC9B,MAAM,KAAK,GAAG,MAAM,QAAQ,CAAC,MAAM,CAAC;YAClC,MAAM,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE;YACtD,MAAM,EAAE,WAAW;SACpB,CAAC,CAAA;QACF,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,CAAC,CAAA;QACrC,GAAG,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,CAAA;IACxB,CAAC;IAED,sEAAsE;IACtE,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,KAAK,MAAM,CAAC,CAAA;IACrE,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,KAAK,MAAM,CAAC,CAAA;IACrE,MAAM,WAAW,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,KAAK,YAAY,CAAC,CAAA;IAEjF,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAC/B,WAAW;SACR,MAAM,CAAC,CAAC,IAAI,EAA+C,EAAE,CAAC,IAAI,CAAC,aAAa,KAAK,IAAI,CAAC;SAC1F,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,aAAa,CAAU,CAAC,CAC7D,CAAA;IAED,MAAM,SAAS,GAAG,IAAI,GAAG,EAGtB,CAAA;IACH,MAAM,YAAY,GAAG,IAAI,GAAG,EAAU,CAAA;IACtC,KAAK,MAAM,IAAI,IAAI,CAAC,GAAG,KAAK,EAAE,GAAG,KAAK,CAAC,EAAE,CAAC;QACxC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG,cAAc,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;QAC7D,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,GAAG,KAAK,CAAC,EAAE,CAAC,CAAA;QACzD,KAAK,MAAM,GAAG,IAAI,WAAW,CAAC,MAAM,CAAC;YAAE,YAAY,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAC5D,MAAM,SAAS,GACb,IAAI,CAAC,WAAW,KAAK,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,iBAAiB,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;QACjF,IAAI,SAAS,KAAK,SAAS;YAAE,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;IAC1D,CAAC;IACD,KAAK,MAAM,GAAG,IAAI,iBAAiB,CAAC,MAAM,EAAE;QAAE,YAAY,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;IAEnE,MAAM,WAAW,GAAG,MAAM,qBAAqB,CAAC,CAAC,GAAG,YAAY,CAAC,EAAE;QACjE,UAAU;QACV,OAAO;QACP,SAAS,EAAE,IAAI;QACf,GAAG,CAAC,OAAO,CAAC,SAAS,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,SAAS,EAAE,OAAO,CAAC,SAAS,EAAE,CAAC;KAC7E,CAAC,CAAA;IACF,GAAG,CAAC,QAAQ,CAAC,KAAK,GAAG,WAAW,CAAC,QAAQ,CAAC,MAAM,CAAA;IAChD,KAAK,MAAM,OAAO,IAAI,WAAW,CAAC,MAAM,EAAE,CAAC;QACzC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,UAAU,OAAO,CAAC,GAAG,8BAA8B,OAAO,CAAC,MAAM,EAAE,CAAC,CAAA;IACxF,CAAC;IACD,KAAK,MAAM,KAAK,IAAI,WAAW,CAAC,QAAQ,EAAE,CAAC;QACzC,GAAG,CAAC,QAAQ,CAAC,IAAI,CACf,UAAU,KAAK,CAAC,QAAQ,wDAAwD,CACjF,CAAA;IACH,CAAC;IAED,wEAAwE;IACxE,KAAK,UAAU,UAAU,CACvB,IAAa,EACb,KAA0C,EAC1C,IAAqB;QAErB,IAAI,IAAI,CAAC,MAAM,KAAK,OAAO,EAAE,CAAC;YAC5B,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC;gBACf,IAAI,EAAE,IAAI;gBACV,IAAI,EAAE,IAAI,CAAC,MAAM;gBACjB,KAAK,EAAE,IAAI,CAAC,KAAK;gBACjB,MAAM,EAAE,qCAAqC;aAC9C,CAAC,CAAA;YACF,OAAO,IAAI,CAAA;QACb,CAAC;QACD,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;YAC1C,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC;gBACf,IAAI,EAAE,IAAI;gBACV,IAAI,EAAE,IAAI,CAAC,MAAM;gBACjB,KAAK,EAAE,IAAI,CAAC,KAAK;gBACjB,MAAM,EAAE,kCAAkC,IAAI,CAAC,MAAM,IAAI;aAC1D,CAAC,CAAA;YACF,OAAO,IAAI,CAAA;QACb,CAAC;QAED,MAAM,KAAK,GAAG,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QACxC,MAAM,KAAK,GAAG,KAAK,EAAE,KAAK,IAAI,EAAE,CAAA;QAChC,MAAM,cAAc,GAAG,sBAAsB,CAC3C,KAAK,EAAE,MAAM,IAAI,EAAE,EACnB,WAAW,CAAC,YAAY,EACxB,KAAK,CACN,CAAA;QACD,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YACzB,GAAG,CAAC,iBAAiB,CAAC,IAAI,CAAC;gBACzB,GAAG,IAAI;gBACP,SAAS,EAAE,IAAI,CAAC,KAAK,IAAI,aAAa,IAAI,IAAI,IAAI,CAAC,MAAM,GAAG;aAC7D,CAAC,CAAA;QACJ,CAAC;QAED,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,OAAO,CAAA;QAChE,MAAM,SAAS,GAAG,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,IAAI,CAAA;QACzD,IAAI,SAAS,KAAK,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAClD,GAAG,CAAC,QAAQ,CAAC,IAAI,CACf,WAAW,IAAI,CAAC,OAAO,SAAS,IAAI,CAAC,KAAK,6CAA6C,CACxF,CAAA;QACH,CAAC;QAED,MAAM,YAAY,GAAG,MAAM,CAAC,WAAW,CACrC,IAAI,CAAC,QAAQ;aACV,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC;aAC3C,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,CACzC,CAAA;QAED,MAAM,MAAM,GAA4B;YACtC,KAAK,EAAE,IAAI,CAAC,KAAK,IAAI,aAAa,IAAI,CAAC,MAAM,GAAG;YAChD,IAAI,EAAE,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,MAAM;YAClC,WAAW,EAAE,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC;YACvC,YAAY;SACb,CAAA;QACD,IAAI,IAAI,KAAK,MAAM,EAAE,CAAC;YACpB,MAAM,CAAC,SAAS,CAAC,GAAG,IAAI,CAAC,cAAc,CAAA;YACvC,MAAM,WAAW,GAAG,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,KAAK,UAAU,CAAC,CAAA;YAC5E,MAAM,QAAQ,GACZ,WAAW,KAAK,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,kBAAkB,CAAC,GAAG,CAAC,WAAW,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,CAAA;YAC3F,IAAI,QAAQ,KAAK,IAAI;gBAAE,MAAM,CAAC,UAAU,CAAC,GAAG,QAAQ,CAAA;YACpD,MAAM,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,UAAU;iBAC7B,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,KAAK,UAAU,CAAC;iBAC1C,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;iBAC7C,MAAM,CAAC,CAAC,EAAE,EAAgB,EAAE,CAAC,EAAE,KAAK,SAAS,CAAC,CAAA;QACnD,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,KAAK,CAAC,MAAM,CAAC;YAC/B,EAAE,EAAE,IAAI,CAAC,MAAM;YACf,MAAM;YACN,SAAS;YACT,MAAM;YACN,MAAM,EAAE,EAAE,IAAI,EAAE,eAAe,CAAC,cAAc,CAAC,EAAE;SAClD,CAAC,CAAA;QAEF,MAAM,UAAU,GAAG,IAAI,KAAK,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAA;QACpD,MAAM,EAAE,GAAG,SAAS,CAAC,UAAU,EAAE,EAAE,IAAI,EAAE,MAAM,CAAC,MAAM,CAAW,EAAE,CAAC,CAAA;QACpE,MAAM,IAAI,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,EAAE,MAAM,CAAC,OAAO,CAAC,CAAA;QACtD,IAAI,IAAI,KAAK,IAAI,IAAI,IAAI,KAAK,EAAE,EAAE,CAAC;YACjC,IAAI,CAAC;gBACH,MAAM,SAAS,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,CAAA;gBACxF,GAAG,CAAC,gBAAgB,IAAI,CAAC,CAAA;YAC3B,CAAC;YAAC,OAAO,KAAK,EAAE,CAAC;gBACf,GAAG,CAAC,QAAQ,CAAC,IAAI,CACf,YAAY,IAAI,MAAM,EAAE,qBAAqB,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CACtG,CAAA;YACH,CAAC;QACH,CAAC;QAED,OAAO,KAAK,CAAC,EAAE,CAAA;IACjB,CAAC;IAED,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,EAAE,GAAG,MAAM,UAAU,CAAC,IAAI,EAAE,SAAS,EAAE,MAAM,CAAC,CAAA;QACpD,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;YAChB,GAAG,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,CAAA;YAEvB,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;gBACpC,IAAI,OAAO,CAAC,QAAQ,KAAK,QAAQ;oBAAE,SAAQ;gBAC3C,MAAM,YAAY,CAAC,MAAM,CAAC;oBACxB,MAAM,EAAE,WAAW;oBACnB,MAAM,EAAE;wBACN,IAAI,EAAE,EAAE;wBACR,MAAM,EAAE,OAAO,CAAC,MAAM,IAAI,WAAW;wBACrC,WAAW,EAAE,OAAO,CAAC,WAAW;wBAChC,IAAI,EAAE,OAAO,CAAC,OAAO;wBACrB,WAAW,EAAE,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC;qBACvC;iBACF,CAAC,CAAA;gBACF,GAAG,CAAC,QAAQ,CAAC,QAAQ,IAAI,CAAC,CAAA;YAC5B,CAAC;QACH,CAAC;IACH,CAAC;IAED,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,EAAE,GAAG,MAAM,UAAU,CAAC,IAAI,EAAE,SAAS,EAAE,MAAM,CAAC,CAAA;QACpD,IAAI,EAAE,KAAK,IAAI;YAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,CAAA;IAC1C,CAAC;IAED,MAAM,OAAO,GAAsB,GAAG,CAAC,OAAO,CAAA;IAC9C,OAAO;QACL,QAAQ,EAAE,GAAG,CAAC,QAAQ;QACtB,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,OAAO;QACP,iBAAiB,EAAE,GAAG,CAAC,iBAAiB;QACxC,QAAQ,EAAE,GAAG,CAAC,QAAQ;KACvB,CAAA;AACH,CAAC"}

package/dist/wordpress/media.d.ts

@@ -0,0 +1,27 @@
+import type { MediaAsset, MediaStore, StorageDriver } from '@cogenta/core';
+export interface MediaDownloadFailure {
+    readonly url: string;
+    readonly reason: string;
+}
+export interface MediaImportResult {
+    readonly urlToMediaId: ReadonlyMap<string, string>;
+    readonly imported: readonly MediaAsset[];
+    readonly failed: readonly MediaDownloadFailure[];
+}
+export interface DownloadAndStoreMediaOptions {
+    readonly mediaStore: MediaStore;
+    readonly storage: StorageDriver;
+    readonly createdBy: string | null;
+    /** Injected for tests — real `fetch` by default. */
+    readonly fetchImpl?: typeof fetch;
+}
+/**
+ * Downloads every distinct media URL a WXR import references and stores it
+ * for real (rule: no mocking the storage layer — the fetch boundary is the
+ * only thing tests stand in for). A URL that fails to fetch is named in
+ * `failed`, never thrown: one dead upstream image, common in a real-world
+ * export, must not abort the whole import (the lot's own framing: a reported
+ * partial loss is the acceptable outcome, a silent one is not).
+ */
+export declare function downloadAndStoreMedia(urls: readonly string[], options: DownloadAndStoreMediaOptions): Promise<MediaImportResult>;
+//# sourceMappingURL=media.d.ts.map

package/dist/wordpress/media.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"media.d.ts","sourceRoot":"","sources":["../../src/wordpress/media.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAoB,UAAU,EAAE,UAAU,EAAE,aAAa,EAAE,MAAM,eAAe,CAAA;AAE5F,MAAM,WAAW,oBAAoB;IACnC,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAA;IACpB,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAA;CACxB;AAED,MAAM,WAAW,iBAAiB;IAChC,QAAQ,CAAC,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAClD,QAAQ,CAAC,QAAQ,EAAE,SAAS,UAAU,EAAE,CAAA;IACxC,QAAQ,CAAC,MAAM,EAAE,SAAS,oBAAoB,EAAE,CAAA;CACjD;AAED,MAAM,WAAW,4BAA4B;IAC3C,QAAQ,CAAC,UAAU,EAAE,UAAU,CAAA;IAC/B,QAAQ,CAAC,OAAO,EAAE,aAAa,CAAA;IAC/B,QAAQ,CAAC,SAAS,EAAE,MAAM,GAAG,IAAI,CAAA;IACjC,oDAAoD;IACpD,QAAQ,CAAC,SAAS,CAAC,EAAE,OAAO,KAAK,CAAA;CAClC;AAmBD;;;;;;;GAOG;AACH,wBAAsB,qBAAqB,CACzC,IAAI,EAAE,SAAS,MAAM,EAAE,EACvB,OAAO,EAAE,4BAA4B,GACpC,OAAO,CAAC,iBAAiB,CAAC,CA2C5B"}

package/dist/wordpress/media.js

@@ -0,0 +1,68 @@
+import { basename } from 'node:path';
+function filenameFrom(url) {
+    try {
+        const path = new URL(url).pathname;
+        const name = basename(path);
+        return name.length > 0 ? name : 'file';
+    }
+    catch {
+        return 'file';
+    }
+}
+function kindFor(mimeType) {
+    if (mimeType.startsWith('image/'))
+        return 'image';
+    if (mimeType.startsWith('video/'))
+        return 'video';
+    if (mimeType.startsWith('audio/'))
+        return 'audio';
+    return 'file';
+}
+/**
+ * Downloads every distinct media URL a WXR import references and stores it
+ * for real (rule: no mocking the storage layer — the fetch boundary is the
+ * only thing tests stand in for). A URL that fails to fetch is named in
+ * `failed`, never thrown: one dead upstream image, common in a real-world
+ * export, must not abort the whole import (the lot's own framing: a reported
+ * partial loss is the acceptable outcome, a silent one is not).
+ */
+export async function downloadAndStoreMedia(urls, options) {
+    const doFetch = options.fetchImpl ?? fetch;
+    const urlToMediaId = new Map();
+    const imported = [];
+    const failed = [];
+    for (const url of new Set(urls)) {
+        try {
+            const response = await doFetch(url);
+            if (!response.ok) {
+                failed.push({ url, reason: `HTTP ${response.status}` });
+                continue;
+            }
+            const mimeType = response.headers.get('content-type')?.split(';')[0]?.trim() ?? 'application/octet-stream';
+            const buffer = Buffer.from(await response.arrayBuffer());
+            const filename = filenameFrom(url);
+            const storageKey = `imports/wordpress/${Date.now()}-${Math.random().toString(36).slice(2)}-${filename}`;
+            await options.storage.put(storageKey, buffer, { contentType: mimeType });
+            const asset = await options.mediaStore.create({
+                kind: kindFor(mimeType),
+                filename,
+                mimeType,
+                size: buffer.byteLength,
+                // WXR carries no alt text for attachments; a synthesised value keeps
+                // the media library's own alt-text requirement honest rather than
+                // inventing a description of an image nobody described, and the gap
+                // is named in the conversion report so an editor knows to revisit it.
+                alt: `Imported from WordPress: ${filename}`,
+                storageKey,
+                createdBy: options.createdBy,
+            });
+            urlToMediaId.set(url, asset.id);
+            imported.push(asset);
+        }
+        catch (error) {
+            failed.push({ url, reason: error instanceof Error ? error.message : String(error) });
+        }
+    }
+    return { urlToMediaId, imported, failed };
+}
+//# sourceMappingURL=media.js.map

package/dist/wordpress/media.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"media.js","sourceRoot":"","sources":["../../src/wordpress/media.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAA;AAsBpC,SAAS,YAAY,CAAC,GAAW;IAC/B,IAAI,CAAC;QACH,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAA;QAClC,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAA;QAC3B,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,CAAA;IACxC,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,MAAM,CAAA;IACf,CAAC;AACH,CAAC;AAED,SAAS,OAAO,CAAC,QAAgB;IAC/B,IAAI,QAAQ,CAAC,UAAU,CAAC,QAAQ,CAAC;QAAE,OAAO,OAAO,CAAA;IACjD,IAAI,QAAQ,CAAC,UAAU,CAAC,QAAQ,CAAC;QAAE,OAAO,OAAO,CAAA;IACjD,IAAI,QAAQ,CAAC,UAAU,CAAC,QAAQ,CAAC;QAAE,OAAO,OAAO,CAAA;IACjD,OAAO,MAAM,CAAA;AACf,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,CAAC,KAAK,UAAU,qBAAqB,CACzC,IAAuB,EACvB,OAAqC;IAErC,MAAM,OAAO,GAAG,OAAO,CAAC,SAAS,IAAI,KAAK,CAAA;IAC1C,MAAM,YAAY,GAAG,IAAI,GAAG,EAAkB,CAAA;IAC9C,MAAM,QAAQ,GAAiB,EAAE,CAAA;IACjC,MAAM,MAAM,GAA2B,EAAE,CAAA;IAEzC,KAAK,MAAM,GAAG,IAAI,IAAI,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;QAChC,IAAI,CAAC;YACH,MAAM,QAAQ,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,CAAA;YACnC,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;gBACjB,MAAM,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,QAAQ,QAAQ,CAAC,MAAM,EAAE,EAAE,CAAC,CAAA;gBACvD,SAAQ;YACV,CAAC;YACD,MAAM,QAAQ,GACZ,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,0BAA0B,CAAA;YAC3F,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAA;YACxD,MAAM,QAAQ,GAAG,YAAY,CAAC,GAAG,CAAC,CAAA;YAClC,MAAM,UAAU,GAAG,qBAAqB,IAAI,CAAC,GAAG,EAAE,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,QAAQ,EAAE,CAAA;YAEvG,MAAM,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,UAAU,EAAE,MAAM,EAAE,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,CAAA;YAExE,MAAM,KAAK,GAAG,MAAM,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC;gBAC5C,IAAI,EAAE,OAAO,CAAC,QAAQ,CAAC;gBACvB,QAAQ;gBACR,QAAQ;gBACR,IAAI,EAAE,MAAM,CAAC,UAAU;gBACvB,qEAAqE;gBACrE,kEAAkE;gBAClE,oEAAoE;gBACpE,sEAAsE;gBACtE,GAAG,EAAE,4BAA4B,QAAQ,EAAE;gBAC3C,UAAU;gBACV,SAAS,EAAE,OAAO,CAAC,SAAS;aAC7B,CAAC,CAAA;YAEF,YAAY,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,EAAE,CAAC,CAAA;YAC/B,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACtB,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,MAAM,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAA;QACtF,CAAC;IACH,CAAC;IAED,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAA;AAC3C,CAAC"}

package/dist/wordpress/parse.d.ts

@@ -0,0 +1,12 @@
+import type { ParsedWxr } from './types.js';
+/**
+ * Reads a WordPress "Export All Content" WXR file into plain data.
+ *
+ * No conversion decisions happen here — this module only decodes the XML into
+ * the shape `content-convert.ts` and `import.ts` work from. `WXR_POST_STATUSES`
+ * is not enforced here either: an export from a fork or a very old WordPress
+ * version may carry a status this list does not name, and refusing to parse it
+ * would lose everything else in the file over one field.
+ */
+export declare function parseWxr(source: string): ParsedWxr;
+//# sourceMappingURL=parse.d.ts.map

package/dist/wordpress/parse.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"parse.d.ts","sourceRoot":"","sources":["../../src/wordpress/parse.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EACV,SAAS,EAQV,MAAM,YAAY,CAAA;AAiFnB;;;;;;;;GAQG;AACH,wBAAgB,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,CAqBlD"}

package/dist/wordpress/parse.js

@@ -0,0 +1,95 @@
+import { CogentaError } from '@cogenta/core';
+import { children, firstChild, parseXmlDocument, textOf, textOfChild, } from './xml.js';
+function parseAuthor(el) {
+    return {
+        login: textOfChild(el, 'wp:author_login'),
+        email: textOfChild(el, 'wp:author_email'),
+        displayName: textOfChild(el, 'wp:author_display_name'),
+    };
+}
+function parseCategoryTerm(el) {
+    return {
+        termId: textOfChild(el, 'wp:term_id'),
+        niceName: textOfChild(el, 'wp:category_nicename'),
+        name: textOfChild(el, 'wp:cat_name'),
+    };
+}
+function parseTagTerm(el) {
+    return {
+        termId: textOfChild(el, 'wp:term_id'),
+        slug: textOfChild(el, 'wp:tag_slug'),
+        name: textOfChild(el, 'wp:tag_name'),
+    };
+}
+function parseTermRef(el) {
+    return {
+        domain: el.attrs['domain'] ?? '',
+        niceName: el.attrs['nicename'] ?? '',
+        name: textOf(el),
+    };
+}
+function parsePostMeta(el) {
+    return { key: textOfChild(el, 'wp:meta_key'), value: textOfChild(el, 'wp:meta_value') };
+}
+function parseComment(el) {
+    return {
+        id: textOfChild(el, 'wp:comment_id'),
+        author: textOfChild(el, 'wp:comment_author'),
+        authorEmail: textOfChild(el, 'wp:comment_author_email'),
+        date: textOfChild(el, 'wp:comment_date_gmt') || textOfChild(el, 'wp:comment_date'),
+        content: textOfChild(el, 'wp:comment_content'),
+        approved: textOfChild(el, 'wp:comment_approved'),
+        parentId: textOfChild(el, 'wp:comment_parent'),
+    };
+}
+function parseItem(el) {
+    const postMeta = children(el, 'wp:postmeta').map(parsePostMeta);
+    const thumbnailId = postMeta.find((meta) => meta.key === '_thumbnail_id')?.value ?? null;
+    return {
+        postId: textOfChild(el, 'wp:post_id'),
+        postType: textOfChild(el, 'wp:post_type'),
+        title: textOfChild(el, 'title'),
+        link: textOfChild(el, 'link'),
+        postName: textOfChild(el, 'wp:post_name'),
+        status: textOfChild(el, 'wp:status'),
+        postDate: textOfChild(el, 'wp:post_date_gmt') || textOfChild(el, 'wp:post_date'),
+        creator: textOfChild(el, 'dc:creator'),
+        contentEncoded: textOfChild(el, 'content:encoded'),
+        excerptEncoded: textOfChild(el, 'excerpt:encoded'),
+        categories: children(el, 'category').map(parseTermRef),
+        postMeta,
+        comments: children(el, 'wp:comment').map(parseComment),
+        attachmentUrl: textOfChild(el, 'wp:attachment_url') || null,
+        thumbnailId,
+    };
+}
+/**
+ * Reads a WordPress "Export All Content" WXR file into plain data.
+ *
+ * No conversion decisions happen here — this module only decodes the XML into
+ * the shape `content-convert.ts` and `import.ts` work from. `WXR_POST_STATUSES`
+ * is not enforced here either: an export from a fork or a very old WordPress
+ * version may carry a status this list does not name, and refusing to parse it
+ * would lose everything else in the file over one field.
+ */
+export function parseWxr(source) {
+    const root = parseXmlDocument(source);
+    const channel = firstChild(root, 'channel');
+    if (channel === null) {
+        throw new CogentaError({
+            code: 'IMPORT_WXR_PARSE_FAILED',
+            message: 'The document has no <channel> element.',
+            hint: 'This does not look like a WordPress WXR export — check the file was produced by Tools → Export.',
+        });
+    }
+    const items = children(channel, 'item').map(parseItem);
+    return {
+        siteTitle: textOfChild(channel, 'title'),
+        baseUrl: textOfChild(channel, 'wp:base_site_url') || textOfChild(channel, 'wp:base_blog_url'),
+        authors: children(channel, 'wp:author').map(parseAuthor),
+        categories: children(channel, 'wp:category').map(parseCategoryTerm),
+        tags: children(channel, 'wp:tag').map(parseTagTerm),
+        items,
+    };
+}
+//# sourceMappingURL=parse.js.map

package/dist/wordpress/parse.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"parse.js","sourceRoot":"","sources":["../../src/wordpress/parse.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,eAAe,CAAA;AAW5C,OAAO,EACL,QAAQ,EACR,UAAU,EACV,gBAAgB,EAChB,MAAM,EACN,WAAW,GAEZ,MAAM,UAAU,CAAA;AAEjB,SAAS,WAAW,CAAC,EAAc;IACjC,OAAO;QACL,KAAK,EAAE,WAAW,CAAC,EAAE,EAAE,iBAAiB,CAAC;QACzC,KAAK,EAAE,WAAW,CAAC,EAAE,EAAE,iBAAiB,CAAC;QACzC,WAAW,EAAE,WAAW,CAAC,EAAE,EAAE,wBAAwB,CAAC;KACvD,CAAA;AACH,CAAC;AAED,SAAS,iBAAiB,CAAC,EAAc;IACvC,OAAO;QACL,MAAM,EAAE,WAAW,CAAC,EAAE,EAAE,YAAY,CAAC;QACrC,QAAQ,EAAE,WAAW,CAAC,EAAE,EAAE,sBAAsB,CAAC;QACjD,IAAI,EAAE,WAAW,CAAC,EAAE,EAAE,aAAa,CAAC;KACrC,CAAA;AACH,CAAC;AAED,SAAS,YAAY,CAAC,EAAc;IAClC,OAAO;QACL,MAAM,EAAE,WAAW,CAAC,EAAE,EAAE,YAAY,CAAC;QACrC,IAAI,EAAE,WAAW,CAAC,EAAE,EAAE,aAAa,CAAC;QACpC,IAAI,EAAE,WAAW,CAAC,EAAE,EAAE,aAAa,CAAC;KACrC,CAAA;AACH,CAAC;AAED,SAAS,YAAY,CAAC,EAAc;IAClC,OAAO;QACL,MAAM,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,EAAE;QAChC,QAAQ,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI,EAAE;QACpC,IAAI,EAAE,MAAM,CAAC,EAAE,CAAC;KACjB,CAAA;AACH,CAAC;AAED,SAAS,aAAa,CAAC,EAAc;IACnC,OAAO,EAAE,GAAG,EAAE,WAAW,CAAC,EAAE,EAAE,aAAa,CAAC,EAAE,KAAK,EAAE,WAAW,CAAC,EAAE,EAAE,eAAe,CAAC,EAAE,CAAA;AACzF,CAAC;AAED,SAAS,YAAY,CAAC,EAAc;IAClC,OAAO;QACL,EAAE,EAAE,WAAW,CAAC,EAAE,EAAE,eAAe,CAAC;QACpC,MAAM,EAAE,WAAW,CAAC,EAAE,EAAE,mBAAmB,CAAC;QAC5C,WAAW,EAAE,WAAW,CAAC,EAAE,EAAE,yBAAyB,CAAC;QACvD,IAAI,EAAE,WAAW,CAAC,EAAE,EAAE,qBAAqB,CAAC,IAAI,WAAW,CAAC,EAAE,EAAE,iBAAiB,CAAC;QAClF,OAAO,EAAE,WAAW,CAAC,EAAE,EAAE,oBAAoB,CAAC;QAC9C,QAAQ,EAAE,WAAW,CAAC,EAAE,EAAE,qBAAqB,CAAC;QAChD,QAAQ,EAAE,WAAW,CAAC,EAAE,EAAE,mBAAmB,CAAC;KAC/C,CAAA;AACH,CAAC;AAED,SAAS,SAAS,CAAC,EAAc;IAC/B,MAAM,QAAQ,GAAG,QAAQ,CAAC,EAAE,EAAE,aAAa,CAAC,CAAC,GAAG,CAAC,aAAa,CAAC,CAAA;IAC/D,MAAM,WAAW,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,KAAK,eAAe,CAAC,EAAE,KAAK,IAAI,IAAI,CAAA;IAExF,OAAO;QACL,MAAM,EAAE,WAAW,CAAC,EAAE,EAAE,YAAY,CAAC;QACrC,QAAQ,EAAE,WAAW,CAAC,EAAE,EAAE,cAAc,CAAC;QACzC,KAAK,EAAE,WAAW,CAAC,EAAE,EAAE,OAAO,CAAC;QAC/B,IAAI,EAAE,WAAW,CAAC,EAAE,EAAE,MAAM,CAAC;QAC7B,QAAQ,EAAE,WAAW,CAAC,EAAE,EAAE,cAAc,CAAC;QACzC,MAAM,EAAE,WAAW,CAAC,EAAE,EAAE,WAAW,CAAC;QACpC,QAAQ,EAAE,WAAW,CAAC,EAAE,EAAE,kBAAkB,CAAC,IAAI,WAAW,CAAC,EAAE,EAAE,cAAc,CAAC;QAChF,OAAO,EAAE,WAAW,CAAC,EAAE,EAAE,YAAY,CAAC;QACtC,cAAc,EAAE,WAAW,CAAC,EAAE,EAAE,iBAAiB,CAAC;QAClD,cAAc,EAAE,WAAW,CAAC,EAAE,EAAE,iBAAiB,CAAC;QAClD,UAAU,EAAE,QAAQ,CAAC,EAAE,EAAE,UAAU,CAAC,CAAC,GAAG,CAAC,YAAY,CAAC;QACtD,QAAQ;QACR,QAAQ,EAAE,QAAQ,CAAC,EAAE,EAAE,YAAY,CAAC,CAAC,GAAG,CAAC,YAAY,CAAC;QACtD,aAAa,EAAE,WAAW,CAAC,EAAE,EAAE,mBAAmB,CAAC,IAAI,IAAI;QAC3D,WAAW;KACZ,CAAA;AACH,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,QAAQ,CAAC,MAAc;IACrC,MAAM,IAAI,GAAG,gBAAgB,CAAC,MAAM,CAAC,CAAA;IACrC,MAAM,OAAO,GAAG,UAAU,CAAC,IAAI,EAAE,SAAS,CAAC,CAAA;IAC3C,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;QACrB,MAAM,IAAI,YAAY,CAAC;YACrB,IAAI,EAAE,yBAAyB;YAC/B,OAAO,EAAE,wCAAwC;YACjD,IAAI,EAAE,iGAAiG;SACxG,CAAC,CAAA;IACJ,CAAC;IAED,MAAM,KAAK,GAAG,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;IAEtD,OAAO;QACL,SAAS,EAAE,WAAW,CAAC,OAAO,EAAE,OAAO,CAAC;QACxC,OAAO,EAAE,WAAW,CAAC,OAAO,EAAE,kBAAkB,CAAC,IAAI,WAAW,CAAC,OAAO,EAAE,kBAAkB,CAAC;QAC7F,OAAO,EAAE,QAAQ,CAAC,OAAO,EAAE,WAAW,CAAC,CAAC,GAAG,CAAC,WAAW,CAAC;QACxD,UAAU,EAAE,QAAQ,CAAC,OAAO,EAAE,aAAa,CAAC,CAAC,GAAG,CAAC,iBAAiB,CAAC;QACnE,IAAI,EAAE,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC,GAAG,CAAC,YAAY,CAAC;QACnD,KAAK;KACN,CAAA;AACH,CAAC"}

package/dist/wordpress/report.d.ts

@@ -0,0 +1,45 @@
+import type { ContentConversionNote } from './content-convert.js';
+export interface UnconvertedItem {
+    readonly type: string;
+    readonly wpId: string;
+    readonly title: string;
+    readonly reason: string;
+}
+export interface ConversionReport {
+    readonly imported: {
+        readonly posts: number;
+        readonly pages: number;
+        readonly categories: number;
+        readonly tags: number;
+        readonly media: number;
+        readonly authors: number;
+        readonly comments: number;
+    };
+    readonly redirectsCreated: number;
+    readonly skipped: readonly UnconvertedItem[];
+    readonly unconvertedBlocks: readonly (ContentConversionNote & {
+        readonly postTitle: string;
+    })[];
+    readonly warnings: readonly string[];
+}
+export interface MutableConversionAccumulator {
+    imported: {
+        posts: number;
+        pages: number;
+        categories: number;
+        tags: number;
+        media: number;
+        authors: number;
+        comments: number;
+    };
+    redirectsCreated: number;
+    skipped: UnconvertedItem[];
+    unconvertedBlocks: (ContentConversionNote & {
+        postTitle: string;
+    })[];
+    warnings: string[];
+}
+export declare function emptyReport(): MutableConversionAccumulator;
+/** A human-readable summary — what `cogenta import wordpress` prints. */
+export declare function formatConversionReport(report: ConversionReport): string;
+//# sourceMappingURL=report.d.ts.map

package/dist/wordpress/report.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"report.d.ts","sourceRoot":"","sources":["../../src/wordpress/report.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,sBAAsB,CAAA;AAEjE,MAAM,WAAW,eAAe;IAC9B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;IACrB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;IACrB,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAA;IACtB,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAA;CACxB;AAED,MAAM,WAAW,gBAAgB;IAC/B,QAAQ,CAAC,QAAQ,EAAE;QACjB,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAA;QACtB,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAA;QACtB,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAA;QAC3B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;QACrB,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAA;QACtB,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAA;QACxB,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAA;KAC1B,CAAA;IACD,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAA;IACjC,QAAQ,CAAC,OAAO,EAAE,SAAS,eAAe,EAAE,CAAA;IAC5C,QAAQ,CAAC,iBAAiB,EAAE,SAAS,CAAC,qBAAqB,GAAG;QAAE,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAA;KAAE,CAAC,EAAE,CAAA;IAC/F,QAAQ,CAAC,QAAQ,EAAE,SAAS,MAAM,EAAE,CAAA;CACrC;AAED,MAAM,WAAW,4BAA4B;IAC3C,QAAQ,EAAE;QACR,KAAK,EAAE,MAAM,CAAA;QACb,KAAK,EAAE,MAAM,CAAA;QACb,UAAU,EAAE,MAAM,CAAA;QAClB,IAAI,EAAE,MAAM,CAAA;QACZ,KAAK,EAAE,MAAM,CAAA;QACb,OAAO,EAAE,MAAM,CAAA;QACf,QAAQ,EAAE,MAAM,CAAA;KACjB,CAAA;IACD,gBAAgB,EAAE,MAAM,CAAA;IACxB,OAAO,EAAE,eAAe,EAAE,CAAA;IAC1B,iBAAiB,EAAE,CAAC,qBAAqB,GAAG;QAAE,SAAS,EAAE,MAAM,CAAA;KAAE,CAAC,EAAE,CAAA;IACpE,QAAQ,EAAE,MAAM,EAAE,CAAA;CACnB;AAED,wBAAgB,WAAW,IAAI,4BAA4B,CAQ1D;AAED,yEAAyE;AACzE,wBAAgB,sBAAsB,CAAC,MAAM,EAAE,gBAAgB,GAAG,MAAM,CAqCvE"}

package/dist/wordpress/report.js

@@ -0,0 +1,41 @@
+export function emptyReport() {
+    return {
+        imported: { posts: 0, pages: 0, categories: 0, tags: 0, media: 0, authors: 0, comments: 0 },
+        redirectsCreated: 0,
+        skipped: [],
+        unconvertedBlocks: [],
+        warnings: [],
+    };
+}
+/** A human-readable summary — what `cogenta import wordpress` prints. */
+export function formatConversionReport(report) {
+    const lines = [];
+    const { imported } = report;
+    lines.push('Imported:');
+    lines.push(`  ${imported.posts} posts, ${imported.pages} pages`);
+    lines.push(`  ${imported.categories} categories, ${imported.tags} tags`);
+    lines.push(`  ${imported.media} media files, ${imported.authors} authors, ${imported.comments} comments`);
+    lines.push(`  ${report.redirectsCreated} redirects`);
+    if (report.skipped.length > 0) {
+        lines.push('');
+        lines.push(`Not imported (${report.skipped.length}):`);
+        for (const item of report.skipped) {
+            lines.push(`  [${item.type} ${item.wpId}] "${item.title}" — ${item.reason}`);
+        }
+    }
+    if (report.unconvertedBlocks.length > 0) {
+        lines.push('');
+        lines.push(`Content that could not be converted to a block or a rich-text node (${report.unconvertedBlocks.length}):`);
+        for (const note of report.unconvertedBlocks) {
+            lines.push(`  "${note.postTitle}" — ${note.source}: ${note.reason}`);
+        }
+    }
+    if (report.warnings.length > 0) {
+        lines.push('');
+        lines.push('Warnings:');
+        for (const warning of report.warnings)
+            lines.push(`  ${warning}`);
+    }
+    return lines.join('\n');
+}
+//# sourceMappingURL=report.js.map

package/dist/wordpress/report.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"report.js","sourceRoot":"","sources":["../../src/wordpress/report.ts"],"names":[],"mappings":"AAyCA,MAAM,UAAU,WAAW;IACzB,OAAO;QACL,QAAQ,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,EAAE;QAC3F,gBAAgB,EAAE,CAAC;QACnB,OAAO,EAAE,EAAE;QACX,iBAAiB,EAAE,EAAE;QACrB,QAAQ,EAAE,EAAE;KACb,CAAA;AACH,CAAC;AAED,yEAAyE;AACzE,MAAM,UAAU,sBAAsB,CAAC,MAAwB;IAC7D,MAAM,KAAK,GAAa,EAAE,CAAA;IAC1B,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,CAAA;IAE3B,KAAK,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;IACvB,KAAK,CAAC,IAAI,CAAC,KAAK,QAAQ,CAAC,KAAK,WAAW,QAAQ,CAAC,KAAK,QAAQ,CAAC,CAAA;IAChE,KAAK,CAAC,IAAI,CAAC,KAAK,QAAQ,CAAC,UAAU,gBAAgB,QAAQ,CAAC,IAAI,OAAO,CAAC,CAAA;IACxE,KAAK,CAAC,IAAI,CACR,KAAK,QAAQ,CAAC,KAAK,iBAAiB,QAAQ,CAAC,OAAO,aAAa,QAAQ,CAAC,QAAQ,WAAW,CAC9F,CAAA;IACD,KAAK,CAAC,IAAI,CAAC,KAAK,MAAM,CAAC,gBAAgB,YAAY,CAAC,CAAA;IAEpD,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC9B,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;QACd,KAAK,CAAC,IAAI,CAAC,iBAAiB,MAAM,CAAC,OAAO,CAAC,MAAM,IAAI,CAAC,CAAA;QACtD,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,OAAO,EAAE,CAAC;YAClC,KAAK,CAAC,IAAI,CAAC,MAAM,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,MAAM,IAAI,CAAC,KAAK,OAAO,IAAI,CAAC,MAAM,EAAE,CAAC,CAAA;QAC9E,CAAC;IACH,CAAC;IAED,IAAI,MAAM,CAAC,iBAAiB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACxC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;QACd,KAAK,CAAC,IAAI,CACR,uEAAuE,MAAM,CAAC,iBAAiB,CAAC,MAAM,IAAI,CAC3G,CAAA;QACD,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,iBAAiB,EAAE,CAAC;YAC5C,KAAK,CAAC,IAAI,CAAC,MAAM,IAAI,CAAC,SAAS,OAAO,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,MAAM,EAAE,CAAC,CAAA;QACtE,CAAC;IACH,CAAC;IAED,IAAI,MAAM,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC/B,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;QACd,KAAK,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;QACvB,KAAK,MAAM,OAAO,IAAI,MAAM,CAAC,QAAQ;YAAE,KAAK,CAAC,IAAI,CAAC,KAAK,OAAO,EAAE,CAAC,CAAA;IACnE,CAAC;IAED,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AACzB,CAAC"}

package/dist/wordpress/types.d.ts

@@ -0,0 +1,66 @@
+/** The WXR document, decoded into plain data — no XML concepts leak past this module. */
+export interface WxrAuthor {
+    readonly login: string;
+    readonly email: string;
+    readonly displayName: string;
+}
+export interface WxrCategory {
+    readonly termId: string;
+    readonly niceName: string;
+    readonly name: string;
+}
+export interface WxrTag {
+    readonly termId: string;
+    readonly slug: string;
+    readonly name: string;
+}
+/** A `<category>` element on an `<item>` — the term it actually applies. */
+export interface WxrTermRef {
+    readonly domain: 'category' | 'post_tag' | string;
+    readonly niceName: string;
+    readonly name: string;
+}
+export interface WxrPostMeta {
+    readonly key: string;
+    readonly value: string;
+}
+export interface WxrComment {
+    readonly id: string;
+    readonly author: string;
+    readonly authorEmail: string;
+    readonly date: string;
+    readonly content: string;
+    /** `'1'` approved, `'0'` pending, `'spam'` — only `'1'` is imported. */
+    readonly approved: string;
+    readonly parentId: string;
+}
+export declare const WXR_POST_STATUSES: readonly ['publish', 'draft', 'pending', 'private', 'future', 'trash'];
+export type WxrPostStatus = (typeof WXR_POST_STATUSES)[number];
+export interface WxrItem {
+    readonly postId: string;
+    readonly postType: string;
+    readonly title: string;
+    readonly link: string;
+    readonly postName: string;
+    readonly status: string;
+    readonly postDate: string;
+    readonly creator: string;
+    readonly contentEncoded: string;
+    readonly excerptEncoded: string;
+    readonly categories: readonly WxrTermRef[];
+    readonly postMeta: readonly WxrPostMeta[];
+    readonly comments: readonly WxrComment[];
+    /** Set only when `postType === 'attachment'`. */
+    readonly attachmentUrl: string | null;
+    /** The thumbnail's `wp:post_id`, from `_thumbnail_id` postmeta, resolved later against attachments. */
+    readonly thumbnailId: string | null;
+}
+export interface ParsedWxr {
+    readonly siteTitle: string;
+    readonly baseUrl: string;
+    readonly authors: readonly WxrAuthor[];
+    readonly categories: readonly WxrCategory[];
+    readonly tags: readonly WxrTag[];
+    readonly items: readonly WxrItem[];
+}
+//# sourceMappingURL=types.d.ts.map