@jant/core 0.3.42 → 0.3.43

package/bin/commands/import-site.js

@@ -20,7 +20,7 @@ import {
 } from "../lib/site-media-parser.js";
 import { openNodeDatabase } from "../lib/node-database.js";
 import { loadNodeRuntime } from "../lib/load-node-runtime.js";
-import { parseFrontMatter as parseFrontMatterShared } from "../../src/lib/hugo-markdown.ts";
+import { parseFrontMatter as parseFrontMatterShared } from "../lib/hugo-markdown.js";
 
 /**
  * Parse front matter from a Markdown file.

package/bin/commands/search-reindex.js

@@ -0,0 +1,175 @@
+import { parseArgs } from "node:util";
+import { resolveSiteUrl } from "../lib/site-url.js";
+
+const INTERNAL_ADMIN_TOKEN_ENV_VAR = "INTERNAL_ADMIN_TOKEN";
+const DEFAULT_BATCH_SIZE = 50;
+const MAX_BATCH_SIZE = 500;
+
+function normalizeBaseUrl(value) {
+  const trimmed = value.trim();
+  return trimmed.endsWith("/") ? trimmed : `${trimmed}/`;
+}
+
+function buildReindexUrl(siteUrl) {
+  return new URL(
+    "api/internal/search/reindex",
+    normalizeBaseUrl(siteUrl),
+  ).toString();
+}
+
+function parseBatchSize(rawLimit) {
+  if (rawLimit === undefined) {
+    return DEFAULT_BATCH_SIZE;
+  }
+
+  const limit = Number.parseInt(rawLimit, 10);
+  if (!Number.isInteger(limit) || limit < 1 || limit > MAX_BATCH_SIZE) {
+    throw new Error(
+      `Batch size must be an integer between 1 and ${MAX_BATCH_SIZE}.`,
+    );
+  }
+  return limit;
+}
+
+async function requestBatch(url, token, { limit, cursor }) {
+  const response = await fetch(url, {
+    method: "POST",
+    headers: {
+      Authorization: `Bearer ${token}`,
+      "Content-Type": "application/json",
+    },
+    body: JSON.stringify(cursor ? { limit, cursor } : { limit }),
+  });
+
+  if (!response.ok) {
+    const text = await response.text();
+    throw new Error(`HTTP ${response.status}: ${text || response.statusText}`);
+  }
+
+  return response.json();
+}
+
+function logBatchResult(batchIndex, result) {
+  console.log(
+    `  batch ${batchIndex}: processed=${result.processed} updated=${result.updated} skipped=${result.skipped}`,
+  );
+}
+
+export async function run(argv) {
+  const { values } = parseArgs({
+    args: argv,
+    options: {
+      config: { type: "string" },
+      env: { type: "string" },
+      help: { type: "boolean", short: "h" },
+      limit: { type: "string" },
+      once: { type: "boolean", default: false },
+      token: { type: "string" },
+      url: { type: "string" },
+    },
+  });
+
+  if (values.help) {
+    console.log("Usage: jant search-reindex [--url <url>] [options]");
+    console.log("");
+    console.log("Rebuild the search index for every non-deleted post by");
+    console.log("recomputing `post.body_text` from the stored TipTap body.");
+    console.log("Useful after changes to the text extraction logic — for");
+    console.log("example, indexing link URLs from inline markdown links.");
+    console.log("");
+    console.log(
+      "Idempotent: rows whose body_text is already up to date are skipped.",
+    );
+    console.log("");
+    console.log("Options:");
+    console.log("  --url           Target site URL");
+    console.log(
+      `  --limit         Batch size per request (default: ${DEFAULT_BATCH_SIZE}, max: ${MAX_BATCH_SIZE})`,
+    );
+    console.log(
+      "  --once          Run a single batch and exit (default: loop until drained)",
+    );
+    console.log("  --token         Internal admin token");
+    console.log(
+      "  --config        Wrangler config file (default: wrangler.toml)",
+    );
+    console.log("  --env           Wrangler environment name");
+    console.log("");
+    console.log("Authentication:");
+    console.log(
+      `  export ${INTERNAL_ADMIN_TOKEN_ENV_VAR}=your-internal-admin-token`,
+    );
+    console.log("  jant search-reindex --url https://your-site.example");
+    console.log("");
+    console.log(
+      "If --url is omitted, uses SITE_ORIGIN + SITE_PATH_PREFIX from env or wrangler.toml.",
+    );
+    process.exit(0);
+  }
+
+  const siteUrl = resolveSiteUrl({
+    url: values.url,
+    config: values.config,
+    env: values.env,
+  });
+  if (!siteUrl) {
+    console.error(
+      "Error: search-reindex requires --url or SITE_ORIGIN in the environment or wrangler.toml.",
+    );
+    process.exit(1);
+  }
+
+  const token =
+    values.token?.trim() || process.env[INTERNAL_ADMIN_TOKEN_ENV_VAR]?.trim();
+  if (!token) {
+    console.error(
+      `Error: search-reindex requires --token or ${INTERNAL_ADMIN_TOKEN_ENV_VAR}.`,
+    );
+    process.exit(1);
+  }
+
+  const batchSize = parseBatchSize(values.limit);
+  const reindexUrl = buildReindexUrl(siteUrl);
+
+  console.log(`Rebuilding search index for ${siteUrl}`);
+  console.log(
+    `  batch size: ${batchSize}${values.once ? " (single batch)" : ""}`,
+  );
+
+  let totalProcessed = 0;
+  let totalUpdated = 0;
+  let totalSkipped = 0;
+  let batchIndex = 0;
+  let cursor;
+
+  for (;;) {
+    batchIndex += 1;
+    const result = await requestBatch(reindexUrl, token, {
+      limit: batchSize,
+      cursor,
+    });
+    logBatchResult(batchIndex, result);
+
+    totalProcessed += result.processed;
+    totalUpdated += result.updated;
+    totalSkipped += result.skipped;
+
+    if (values.once) break;
+    if (result.done) break;
+
+    // Stop if the server didn't advance the cursor — shouldn't happen with
+    // the service-level contract, but avoids an infinite loop just in case.
+    if (!result.nextCursor || result.nextCursor === cursor) {
+      console.error(
+        "Server returned no next cursor but reported more work; aborting to avoid an infinite loop.",
+      );
+      break;
+    }
+    cursor = result.nextCursor;
+  }
+
+  console.log("");
+  console.log(
+    `Done. processed=${totalProcessed} updated=${totalUpdated} skipped=${totalSkipped} batches=${batchIndex}`,
+  );
+}

package/bin/lib/hugo-markdown.js

@@ -0,0 +1,102 @@
+/**
+ * Plain-JS runtime mirror of `src/lib/hugo-markdown.ts` for the published
+ * CLI. The `bin/` scripts ship as `.js` and are executed by Node directly
+ * from `node_modules`, where Node 24 refuses to strip TypeScript types.
+ * Importing the `.ts` source from these scripts fails with
+ * ERR_UNSUPPORTED_NODE_MODULES_TYPE_STRIPPING, so we keep the runtime here
+ * as real JavaScript.
+ *
+ * Keep this file in lock-step with `src/lib/hugo-markdown.ts` — same front
+ * matter shapes, same key order, same YAML formatting options.
+ */
+
+const FRONT_MATTER_KEY_ORDER = [
+  // Identity & routing
+  "id",
+  "title",
+  "date",
+  "updated",
+  "slug",
+  "type",
+  "draft",
+  "aliases",
+  "build",
+
+  // Post payload
+  "format",
+  "status",
+  "visibility",
+  "summary_text",
+  "link_url",
+  "source_name",
+  "source_url",
+  "quote_text",
+  "rating",
+  "featured_at",
+  "pinned_at",
+
+  // Bookkeeping / attachments
+  "root_aliases",
+  "collections",
+  "media",
+];
+
+/**
+ * Parse front matter from a Markdown file. Accepts YAML (`---...---`) or
+ * TOML (`+++...+++`) delimiters.
+ *
+ * @param {string} content
+ * @returns {Promise<{ frontMatter: Record<string, unknown>, body: string }>}
+ */
+export async function parseFrontMatter(content) {
+  const yamlMatch = content.match(
+    /^---\r?\n([\s\S]*?)\r?\n---\r?\n?([\s\S]*)$/,
+  );
+  if (yamlMatch) {
+    const { parse } = await import("yaml");
+    const frontMatter = parse(yamlMatch[1] ?? "") ?? {};
+    return { frontMatter, body: yamlMatch[2] ?? "" };
+  }
+
+  const tomlMatch = content.match(
+    /^\+\+\+\r?\n([\s\S]*?)\r?\n\+\+\+\r?\n?([\s\S]*)$/,
+  );
+  if (tomlMatch) {
+    const { parse } = await import("smol-toml");
+    const frontMatter = parse(tomlMatch[1] ?? "");
+    return { frontMatter, body: tomlMatch[2] ?? "" };
+  }
+
+  return { frontMatter: {}, body: content };
+}
+
+/**
+ * Serialize a front-matter object as a YAML `---...---` block with the
+ * canonical key order.
+ *
+ * @param {Record<string, unknown>} frontMatter
+ * @returns {Promise<string>}
+ */
+export async function formatFrontMatter(frontMatter) {
+  const { stringify } = await import("yaml");
+
+  const ordered = {};
+  for (const key of FRONT_MATTER_KEY_ORDER) {
+    if (key in frontMatter) {
+      const value = frontMatter[key];
+      if (value !== undefined) ordered[key] = value;
+    }
+  }
+  for (const key of Object.keys(frontMatter)) {
+    if (key in ordered) continue;
+    const value = frontMatter[key];
+    if (value !== undefined) ordered[key] = value;
+  }
+
+  const yaml = stringify(ordered, {
+    defaultKeyType: "PLAIN",
+    defaultStringType: "QUOTE_DOUBLE",
+    lineWidth: 0,
+  });
+  return `---\n${yaml.trimEnd()}\n---\n`;
+}

package/bin/lib/site-pull-media.js

@@ -19,10 +19,7 @@ import {
   isSkippableUrl,
   rewriteMediaReferences,
 } from "./site-media-parser.js";
-import {
-  formatFrontMatter,
-  parseFrontMatter,
-} from "../../src/lib/hugo-markdown.ts";
+import { formatFrontMatter, parseFrontMatter } from "./hugo-markdown.js";
 
 export function getSitePathPrefix(baseUrl) {
   if (typeof baseUrl !== "string" || baseUrl.trim() === "") {

package/dist/app-Ctl0T0zO.js

@@ -0,0 +1,5 @@
+import "./url-umUptr5z.js";
+import { t as createApp } from "./app-GbfwoeDJ.js";
+import "./github-sync-7y_nTXx1.js";
+import "./env-CgaH9Mut.js";
+export { createApp };