feat(crawler): fix async crawling

.vscode/settings.json

@@ -9,8 +9,8 @@
     "editor.defaultFormatter": "biomejs.biome"
   },
   "editor.codeActionsOnSave": {
-    "source.organizeImports.biome": "explicit",
-    "source.fixAll.biome": "explicit"
+    "source.fixAll.biome": "explicit",
+    "source.organizeImports.biome": "explicit"
   },
   "editor.defaultFormatter": "biomejs.biome",
   "editor.formatOnSave": true,

apps/crawler/src/config.ts

@@ -2,6 +2,7 @@ import path from "node:path";
 
 import { loadConfig as defineConfig } from "@devscast/config";
 import { z } from "zod";
+
 import {
   DateRangeSchema,
   HtmlSourceConfigSchema,

apps/crawler/src/constants.ts

@@ -1,6 +1,29 @@
+/**
+ * Default date format used for parsing and formatting dates.
+ * Follows the "yyyy-LL-dd" pattern (e.g., "2024-06-15").
+ */
 export const DEFAULT_DATE_FORMAT = "yyyy-LL-dd";
+
+/**
+ * Default User-Agent string for HTTP requests made by the crawler.
+ * Some websites may block requests with missing or generic User-Agent headers.
+ */
 export const DEFAULT_USER_AGENT = "Basango/0.1 (+https://github.com/bernard-ng/basango)";
+
+/**
+ * User-Agent string used for Open Graph requests.
+ * Some services require a specific User-Agent to return Open Graph data.
+ */
 export const OPEN_GRAPH_USER_AGENT = "facebookexternalhit/1.1";
 
+/**
+ * HTTP status codes considered transient errors.
+ * Used for retry logic in HTTP clients.
+ */
 export const TRANSIENT_HTTP_STATUSES = [429, 500, 502, 503, 504];
+
+/**
+ * Default header name for Retry-After responses.
+ * Used when handling rate limiting.
+ */
 export const DEFAULT_RETRY_AFTER_HEADER = "retry-after";

apps/crawler/src/errors.ts

@@ -0,0 +1,39 @@
+/**
+ * Error thrown when an article is invalid or cannot be processed.
+ */
+export class InvalidArticleError extends Error {
+  constructor(message: string) {
+    super(message);
+    this.name = "InvalidArticleError";
+  }
+}
+
+/**
+ * Error thrown when a source kind is not supported by the crawler.
+ */
+export class UnsupportedSourceKindError extends Error {
+  constructor(message: string) {
+    super(message);
+    this.name = "UnsupportedSourceKindError";
+  }
+}
+
+/**
+ * Error thrown when a source's selectors are invalid or missing.
+ */
+export class InvalidSourceSelectorsError extends Error {
+  constructor(message: string) {
+    super(message);
+    this.name = "InvalidSourceSelectorsError";
+  }
+}
+
+/**
+ * Error thrown when an article's publication date is outside the specified date range.
+ */
+export class ArticleOutOfDateRangeError extends Error {
+  constructor(message: string, _meta: Record<string, unknown>) {
+    super(message);
+    this.name = "ArticleOutOfDateRangeError";
+  }
+}

apps/crawler/src/http/http-client.ts

@@ -1,4 +1,5 @@
 import { setTimeout as delay } from "node:timers/promises";
+
 import { FetchClientConfig } from "@/config";
 import {
   DEFAULT_RETRY_AFTER_HEADER,

apps/crawler/src/http/open-graph.ts

@@ -1,4 +1,5 @@
 import { parse } from "node-html-parser";
+
 import { config } from "@/config";
 import { OPEN_GRAPH_USER_AGENT } from "@/constants";
 import { SyncHttpClient } from "@/http/http-client";

apps/crawler/src/process/async/handlers.ts

@@ -1,17 +1,17 @@
 import { logger } from "@basango/logger";
 
 import { config, env } from "@/config";
+import { UnsupportedSourceKindError } from "@/errors";
 import { SyncHttpClient } from "@/http/http-client";
-import { createQueueManager, QueueManager } from "@/process/async/queue";
+import { QueueManager, createQueueManager } from "@/process/async/queue";
 import {
   DetailsTaskPayload,
   ListingTaskPayload,
   ProcessingTaskPayload,
 } from "@/process/async/schemas";
-import { resolveCrawlerConfig } from "@/process/crawler";
+import { createPersistors, resolveCrawlerConfig } from "@/process/crawler";
 import { HtmlCrawler } from "@/process/parsers/html";
 import { WordPressCrawler } from "@/process/parsers/wordpress";
-import { JsonlPersistor } from "@/process/persistence";
 import { Article, HtmlSourceConfig, SourceKindSchema, WordPressSourceConfig } from "@/schema";
 import { createDateRange, formatDateRange, formatPageRange, resolveSourceConfig } from "@/utils";
 
@@ -30,7 +30,7 @@ export const collectHtmlListing = async (
 
   let queued = 0;
   for (let page = pageRange.start; page <= pageRange.end; page += 1) {
-    const target = crawler.buildPageUrl(page) ?? `${source.sourceUrl}`;
+    const target = crawler.buildEndpointUrl(page) ?? `${source.sourceUrl}`;
 
     try {
       const items = await crawler.fetchLinks(target, source.sourceSelectors.articles);
@@ -69,7 +69,7 @@ export const collectWordPressListing = async (
 
   let queued = 0;
   for (let page = pageRange.start; page <= pageRange.end; page += 1) {
-    const url = crawler.postsEndpoint(page);
+    const url = crawler.buildEndpointUrl(page);
 
     try {
       const entries = await crawler.fetchLinks(url);
@@ -94,7 +94,10 @@ export const collectWordPressListing = async (
   return queued;
 };
 
-export const collectArticle = async (payload: DetailsTaskPayload): Promise<unknown> => {
+export const collectArticle = async (
+  payload: DetailsTaskPayload,
+  manager: QueueManager = createQueueManager(),
+): Promise<unknown> => {
   const source = resolveSourceConfig(payload.sourceId);
   const settings = resolveCrawlerConfig(source, {
     category: payload.category,
@@ -102,26 +105,30 @@ export const collectArticle = async (payload: DetailsTaskPayload): Promise<unkno
     pageRange: payload.pageRange ? formatPageRange(payload.pageRange) : undefined,
     sourceId: payload.sourceId,
   });
-  const persistors = [
-    new JsonlPersistor({
-      directory: config.paths.data,
-      sourceId: String(source.sourceId),
-    }),
-  ];
+  const persistors = createPersistors(source);
 
   if (source.sourceKind === SourceKindSchema.enum.html) {
-    if (!payload.url) throw new Error("Missing article url");
     const crawler = new HtmlCrawler(settings, { persistors });
     const html = await crawler.crawl(payload.url);
-    return await crawler.fetchOne(html, settings.dateRange);
+
+    const article = await crawler.fetchOne(html, settings.dateRange);
+    await manager.enqueueProcessed({
+      article,
+      sourceId: payload.sourceId,
+    } as ProcessingTaskPayload);
   }
 
   if (source.sourceKind === SourceKindSchema.enum.wordpress) {
     const crawler = new WordPressCrawler(settings, { persistors });
-    return await crawler.fetchOne(payload.data ?? {}, settings.dateRange);
+
+    const article = await crawler.fetchOne(payload.data ?? {}, settings.dateRange);
+    await manager.enqueueProcessed({
+      article,
+      sourceId: payload.sourceId,
+    } as ProcessingTaskPayload);
   }
 
-  throw new Error(`Unsupported source kind`);
+  throw new UnsupportedSourceKindError(`Unsupported source kind`);
 };
 
 export const forwardForProcessing = async (payload: ProcessingTaskPayload): Promise<Article> => {

apps/crawler/src/process/async/queue.ts

@@ -1,7 +1,9 @@
 import { randomUUID } from "node:crypto";
+
 import { JobsOptions, Queue, QueueOptions } from "bullmq";
 import IORedis from "ioredis";
-import { config, FetchAsyncConfig } from "@/config";
+
+import { FetchAsyncConfig, config } from "@/config";
 import {
   DetailsTaskPayload,
   DetailsTaskPayloadSchema,
@@ -97,9 +99,9 @@ export const createQueueManager = (options: CreateQueueManagerOptions = {}): Que
       return queue.add("forward_for_processing", data);
     },
     iterQueueNames: () => [
-      `${settings.prefix}:${settings.queues.listing}`,
-      `${settings.prefix}:${settings.queues.details}`,
-      `${settings.prefix}:${settings.queues.processing}`,
+      settings.queues.listing,
+      settings.queues.details,
+      settings.queues.processing,
     ],
     queueName: (suffix: string) => `${settings.prefix}:${suffix}`,
     settings,

apps/crawler/src/process/async/schemas.ts

@@ -1,4 +1,5 @@
 import { z } from "zod";
+
 import { ArticleSchema, DateRangeSchema, PageRangeSchema } from "@/schema";
 
 export const ListingTaskPayloadSchema = z.object({

apps/crawler/src/process/async/tasks.ts

@@ -1,4 +1,5 @@
 import { logger } from "@basango/logger";
+
 import * as handlers from "@/process/async/handlers";
 import { createQueueManager } from "@/process/async/queue";
 import {

apps/crawler/src/process/async/worker.ts

@@ -45,6 +45,7 @@ export const startWorker = (options: WorkerOptions): WorkerHandle => {
       {
         concurrency: options.concurrency ?? 5,
         connection,
+        prefix: manager.settings.prefix,
       },
     );
 
@@ -53,7 +54,10 @@ export const startWorker = (options: WorkerOptions): WorkerHandle => {
       worker.on("error", (err) => options.onError?.(err as Error));
     }
 
-    const queueEvents = new QueueEvents(queueName, { connection });
+    const queueEvents = new QueueEvents(queueName, {
+      connection,
+      prefix: manager.settings.prefix,
+    });
 
     workers.push(worker);
     events.push(queueEvents);

apps/crawler/src/process/crawler.ts

@@ -1,5 +1,6 @@
 import logger from "@basango/logger";
-import { config, FetchCrawlerConfig } from "@/config";
+
+import { FetchCrawlerConfig, config } from "@/config";
 import { JsonlPersistor, Persistor } from "@/process/persistence";
 import { AnySourceConfig } from "@/schema";
 import { createDateRange, createPageRange } from "@/utils";

apps/crawler/src/process/parsers/base.ts

@@ -1,5 +1,6 @@
 import { HTMLElement, parse as parseHtml } from "node-html-parser";
-import { config, FetchCrawlerConfig } from "@/config";
+
+import { FetchCrawlerConfig, config } from "@/config";
 import { SyncHttpClient } from "@/http/http-client";
 import { OpenGraph } from "@/http/open-graph";
 import type { Persistor } from "@/process/persistence";

apps/crawler/src/process/parsers/html.ts

@@ -2,10 +2,17 @@ import { logger } from "@basango/logger";
 import { getUnixTime, isMatch as isDateMatch, parse as parseDateFns } from "date-fns";
 import { HTMLElement } from "node-html-parser";
 import TurndownService from "turndown";
+
 import { FetchCrawlerConfig } from "@/config";
+import {
+  ArticleOutOfDateRangeError,
+  InvalidArticleError,
+  InvalidSourceSelectorsError,
+  UnsupportedSourceKindError,
+} from "@/errors";
 import { BaseCrawler } from "@/process/parsers/base";
 import { Persistor, persist } from "@/process/persistence";
-import { DateRange, HtmlSourceConfig } from "@/schema";
+import { Article, DateRange, HtmlSourceConfig } from "@/schema";
 import { createAbsoluteUrl, isTimestampInRange } from "@/utils";
 
 const md = new TurndownService({
@@ -32,13 +39,13 @@ const safeRegExp = (pattern?: string | null): RegExp | null => {
  */
 export class HtmlCrawler extends BaseCrawler {
   readonly source: HtmlSourceConfig;
-  private currentArticleUrl: string | null = null;
+  private currentNode: string | null = null;
 
   constructor(settings: FetchCrawlerConfig, options: { persistors?: Persistor[] } = {}) {
     super(settings, options);
 
     if (!settings.source || settings.source.sourceKind !== "html") {
-      throw new Error("HtmlCrawler requires a source of kind 'html'");
+      throw new UnsupportedSourceKindError("HtmlCrawler requires a source of kind 'html'");
     }
     this.source = this.settings.source as HtmlSourceConfig;
   }
@@ -46,69 +53,64 @@ export class HtmlCrawler extends BaseCrawler {
   async fetch(): Promise<void> {
     const pageRange = this.settings.pageRange ?? (await this.getPagination());
     const dateRange = this.settings.dateRange;
+    const selectors = this.source.sourceSelectors;
 
-    const articleSelector = this.source.sourceSelectors.articles;
-    if (!articleSelector) {
-      logger.error(
-        { source: this.source.sourceId },
-        "No article selector configured for HTML source",
-      );
-      return;
+    if (!selectors.articles) {
+      throw new InvalidSourceSelectorsError("No article selector configured for HTML source");
     }
 
-    let stop = false;
     for (let page = pageRange.start; page <= pageRange.end; page += 1) {
-      const pageUrl = this.buildPageUrl(page);
+      const endpoint = this.buildEndpointUrl(page);
       let html: string;
+
       try {
-        html = await this.crawl(pageUrl);
+        html = await this.crawl(endpoint);
       } catch (error) {
-        logger.error({ error, page, pageUrl }, "> page %s => [failed]", page);
+        logger.error({ endpoint, error, page }, `Failed to crawl page ${page}`);
         continue;
       }
 
       const root = this.parseHtml(html);
-      const articles = this.extractAll(root, articleSelector);
+      const articles = this.extractAll(root, selectors.articles);
       if (!articles.length) {
-        logger.info({ page }, "No articles found on page");
+        logger.error({ page }, "No articles found on page");
         continue;
       }
 
       for (const node of articles) {
         try {
-          this.currentArticleUrl = this.extractLink(node);
-          let targetHtml = node.toString();
+          this.currentNode = this.extractLink(node);
+          let nodeHtml = node.toString();
 
           if (this.source.requiresDetails) {
-            if (!this.currentArticleUrl) {
-              logger.debug({ page }, "Skipping article without link for details");
+            if (!this.currentNode) {
+              logger.error({ page }, "Skipping article without link for details");
               continue;
             }
+
             try {
-              targetHtml = await this.crawl(this.currentArticleUrl);
-            } catch (err) {
-              logger.error(
-                { error: err, url: this.currentArticleUrl },
-                "Failed to fetch detail page",
-              );
+              nodeHtml = await this.crawl(this.currentNode);
+            } catch (error) {
+              logger.error({ error, url: this.currentNode }, "Failed to fetch detail page");
               continue;
             }
           }
 
-          const saved = await this.fetchOne(targetHtml, dateRange);
-          // stop early on first out-of-range if pages are sorted by date desc
-          if (saved === null) {
-            stop = true;
+          await this.fetchOne(nodeHtml, dateRange);
+        } catch (error: unknown) {
+          if (error instanceof ArticleOutOfDateRangeError) {
+            logger.info(
+              { url: this.currentNode },
+              "Article out of date range, stopping further processing",
+            );
             break;
           }
-        } catch (error) {
-          logger.error({ error, pageUrl }, "Failed to process article on page");
+
+          logger.error({ error, url: this.currentNode }, "Failed to process HTML article");
         } finally {
-          this.currentArticleUrl = null;
+          this.currentNode = null;
         }
       }
-
-      if (stop) break;
     }
   }
 
@@ -117,43 +119,43 @@ export class HtmlCrawler extends BaseCrawler {
    * @param html - The HTML content of the article
    * @param dateRange - Optional date range for filtering
    */
-  async fetchOne(html: string, dateRange?: DateRange | null) {
+  async fetchOne(html: string, dateRange?: DateRange | null): Promise<Article> {
     const root = this.parseHtml(html);
-    const sel = this.source.sourceSelectors;
+    const selectors = this.source.sourceSelectors;
 
-    const titleText = this.extractText(root, sel.articleTitle) ?? "Untitled";
-    const link = this.currentArticleUrl ?? this.extractLink(root);
+    const title = this.extractText(root, selectors.articleTitle) ?? "Untitled";
+    const link = this.currentNode ?? this.extractLink(root);
     if (!link) {
-      logger.warn({ title: titleText }, "Skipping article without link");
-      return null;
+      throw new InvalidArticleError("Missing article link");
     }
 
-    const body = this.extractBody(root, sel.articleBody);
-    const categories = this.extractCategories(root, sel.articleCategories);
-    const rawDate = this.extractText(root, sel.articleDate);
-    const timestamp = this.computeTimestamp(rawDate);
+    const body = this.extractBody(root, selectors.articleBody);
+    const categories = this.extractCategories(root, selectors.articleCategories);
+    const date = this.extractText(root, selectors.articleDate);
+    const timestamp = this.computeTimestamp(date);
 
     if (dateRange && !isTimestampInRange(dateRange, timestamp)) {
-      logger.info(
-        { date: rawDate, link, timestamp, title: titleText },
-        "Skipping article outside date range",
-      );
-      return null;
+      throw new ArticleOutOfDateRangeError("Article outside date range", {
+        date,
+        link,
+        timestamp,
+        title,
+      });
     }
 
-    const enriched = await this.enrichWithOpenGraph(
+    const data = await this.enrichWithOpenGraph(
       {
         body,
         categories,
         link,
         source: this.source.sourceId,
         timestamp,
-        title: titleText,
+        title,
       },
       link,
     );
 
-    return await persist(enriched, this.persistors);
+    return await persist(data, this.persistors);
   }
 
   /**
@@ -211,7 +213,7 @@ export class HtmlCrawler extends BaseCrawler {
    * Build the URL for a given page number.
    * @param page - The page number
    */
-  buildPageUrl(page: number): string {
+  buildEndpointUrl(page: number): string {
     let template = this.applyCategory(this.source.paginationTemplate);
     if (template.includes("{page}")) {
       template = template.replace("{page}", String(page));

apps/crawler/src/process/parsers/wordpress.ts

@@ -1,9 +1,16 @@
 import { logger } from "@basango/logger";
 import TurndownService from "turndown";
+
 import { FetchCrawlerConfig } from "@/config";
+import {
+  ArticleOutOfDateRangeError,
+  InvalidArticleError,
+  UnsupportedSourceKindError,
+} from "@/errors";
 import { BaseCrawler } from "@/process/parsers/base";
 import { Persistor, persist } from "@/process/persistence";
-import { DateRange, PageRange, WordPressSourceConfig } from "@/schema";
+import { Article, DateRange, PageRange, WordPressSourceConfig } from "@/schema";
+import { isTimestampInRange } from "@/utils";
 
 const md = new TurndownService({
   bulletListMarker: "-",
@@ -38,7 +45,9 @@ export class WordPressCrawler extends BaseCrawler {
     super(settings, options);
 
     if (!settings.source || settings.source.sourceKind !== "wordpress") {
-      throw new Error("HtmlCrawler requires a source of kind 'wordpress'");
+      throw new UnsupportedSourceKindError(
+        "WordPressCrawler requires a source of kind 'wordpress'",
+      );
     }
     this.source = this.settings.source as WordPressSourceConfig;
   }
@@ -50,29 +59,31 @@ export class WordPressCrawler extends BaseCrawler {
     const pageRange = this.settings.pageRange ?? (await this.getPagination());
     const dateRange = this.settings.dateRange;
 
-    let stop = false;
     for (let page = pageRange.start; page <= pageRange.end; page += 1) {
-      const endpoint = this.postsEndpoint(page);
+      const endpoint = this.buildEndpointUrl(page);
+
       try {
         const response = await this.http.get(endpoint);
-        const data = (await response.json()) as unknown;
-        const articles = Array.isArray(data) ? (data as WordPressPost[]) : [];
-        if (!Array.isArray(data)) {
-          logger.warn({ page, type: typeof data }, "Unexpected WordPress payload type");
-        }
+        const articles = (await response.json()) as WordPressPost[];
 
-        for (const entry of articles) {
-          const saved = await this.fetchOne(entry, dateRange);
-          if (saved === null) {
-            stop = true;
-            break;
+        for (const node of articles) {
+          try {
+            await this.fetchOne(node, dateRange);
+          } catch (error: unknown) {
+            if (error instanceof ArticleOutOfDateRangeError) {
+              logger.info(
+                { url: node.link },
+                "Article out of date range, stopping further processing",
+              );
+              break;
+            }
+
+            logger.error({ error, url: node.link }, "Failed to process WordPress article");
           }
         }
       } catch (error) {
-        logger.error({ error, page }, "> page %s => [failed]", page);
-        continue;
+        logger.error({ error, page }, `Failed to fetch WordPress page ${page}`);
       }
-      if (stop) break;
     }
   }
 
@@ -95,7 +106,7 @@ export class WordPressCrawler extends BaseCrawler {
    * @param input - Decoded JSON object or raw JSON string
    * @param dateRange - Optional date range for filtering
    */
-  async fetchOne(input: unknown, dateRange?: DateRange | null) {
+  async fetchOne(input: unknown, dateRange?: DateRange | null): Promise<Article> {
     // input can be the decoded JSON object or a raw JSON string
     let data: WordPressPost | null = null;
     try {
@@ -110,35 +121,29 @@ export class WordPressCrawler extends BaseCrawler {
     }
 
     if (!data || typeof data !== "object") {
-      throw new Error("Unexpected WordPress payload type");
+      throw new InvalidArticleError("Unexpected WordPress payload type");
     }
 
     const link = data.link;
     if (!link) {
-      logger.error("Skipping WordPress article without link");
-      return null;
+      throw new InvalidArticleError("Missing article link");
     }
 
-    const titleHtml = data.title?.rendered ?? "";
-    const bodyHtml = data.content?.rendered ?? "";
-    const title = this.textContent(this.parseHtml(titleHtml)) ?? data.slug ?? "Untitled";
-    const body = md.turndown(bodyHtml);
+    const title =
+      this.textContent(this.parseHtml(data.title?.rendered ?? "")) ?? data.slug ?? "Untitled";
+    const body = md.turndown(data.content?.rendered ?? "");
     const timestamp = this.computeTimestamp(data.date);
     const categories = await this.mapCategories(data.categories ?? []);
 
-    // date range skip as in HTML crawler
-    if (dateRange) {
-      const { isTimestampInRange } = await import("@/utils");
-      if (!isTimestampInRange(dateRange, timestamp)) {
-        logger.info(
-          { date: data.date, link, timestamp, title },
-          "Skipping article outside date range",
-        );
-        return null;
-      }
+    if (dateRange && !isTimestampInRange(dateRange, timestamp)) {
+      throw new ArticleOutOfDateRangeError("Article outside date range", {
+        link,
+        timestamp,
+        title,
+      });
     }
 
-    const enriched = await this.enrichWithOpenGraph(
+    const article = await this.enrichWithOpenGraph(
       {
         body,
         categories,
@@ -150,7 +155,7 @@ export class WordPressCrawler extends BaseCrawler {
       link,
     );
 
-    return await persist(enriched, this.persistors);
+    return await persist(article, this.persistors);
   }
 
   /**
@@ -188,7 +193,7 @@ export class WordPressCrawler extends BaseCrawler {
    * Construct posts endpoint URL for a given page.
    * @param page - Page number
    */
-  postsEndpoint(page: number): string {
+  buildEndpointUrl(page: number): string {
     return `${this.baseUrl()}wp-json/wp/v2/posts?${WordPressCrawler.POST_QUERY}&page=${page}&per_page=100`;
   }
 

apps/crawler/src/process/persistence.ts

@@ -1,6 +1,8 @@
 import fs from "node:fs";
 import path from "node:path";
+
 import logger from "@basango/logger";
+
 import { Article } from "@/schema";
 import { countTokens } from "@/utils";
 

apps/crawler/src/process/sync/tasks.ts

@@ -1,4 +1,5 @@
 import logger from "@basango/logger";
+
 import {
   CrawlingOptions,
   closePersistors,

apps/crawler/src/scripts/crawl.ts

@@ -1,4 +1,5 @@
 import { logger } from "@basango/logger";
+
 import { runSyncCrawl } from "@/process/sync/tasks";
 import { CRAWLING_USAGE, parseCrawlingCliArgs } from "@/scripts/utils";
 

apps/crawler/src/scripts/queue.ts

@@ -1,4 +1,5 @@
 import { logger } from "@basango/logger";
+
 import { scheduleAsyncCrawl } from "@/process/async/tasks";
 import { CRAWLING_USAGE, parseCrawlingCliArgs } from "@/scripts/utils";
 

apps/crawler/src/scripts/utils.ts

@@ -1,4 +1,5 @@
 import { parseArgs } from "node:util";
+
 import { CrawlingOptions } from "@/process/crawler";
 
 interface WorkerCliOptions {
@@ -6,13 +7,12 @@ interface WorkerCliOptions {
 }
 
 export const CRAWLING_USAGE = `
-    Usage: bun run crawl:[async|sync] -- --sourceId <id> [options]
+    Usage: bun run crawler:[async|sync] -- --sourceId <id> [options]
     
     Options:
       --pageRange <range>   Optional page range filter (e.g. 1:5)
       --dateRange <range>   Optional date range filter (e.g. 2024-01-01:2024-01-31)
       --category <slug>     Optional category to crawl
-      -h, --help            Show this message
 `;
 
 export const parseWorkerCliArgs = (): WorkerCliOptions => {

apps/crawler/src/scripts/worker.ts

@@ -8,9 +8,7 @@ const main = async (): Promise<void> => {
   const options = parseWorkerCliArgs();
 
   const manager = createQueueManager();
-  const queues = options.queue?.length
-    ? options.queue.map((name) => manager.queueName(name))
-    : undefined;
+  const queues = options.queue?.length ? options.queue : undefined;
 
   const handle = startWorker({
     queueManager: manager,

apps/crawler/src/utils.ts

@@ -1,6 +1,7 @@
 import { format, getUnixTime, isMatch, parse } from "date-fns";
 import type { RedisOptions } from "ioredis";
-import { get_encoding, TiktokenEncoding } from "tiktoken";
+import { TiktokenEncoding, get_encoding } from "tiktoken";
+
 import { config } from "@/config";
 import { DEFAULT_DATE_FORMAT } from "@/constants";
 import {

apps/mobile-legacy/_package.json

@@ -1,8 +1,6 @@
 {
   "commitlint": {
-    "extends": [
-      "@commitlint/config-conventional"
-    ]
+    "extends": ["@commitlint/config-conventional"]
   },
   "config": {
     "commitizen": {
@@ -85,14 +83,8 @@
     "preset": "jest-expo"
   },
   "lint-staged": {
-    "*.ts": [
-      "prettier --write",
-      "eslint --fix"
-    ],
-    "*.tsx": [
-      "prettier --write",
-      "eslint --fix"
-    ]
+    "*.ts": ["prettier --write", "eslint --fix"],
+    "*.tsx": ["prettier --write", "eslint --fix"]
   },
   "main": "expo-router/entry",
   "name": "drc-news",

apps/mobile-legacy/src/api/request/feed-management/bookmark.ts

@@ -1,8 +1,8 @@
 import { endpoint } from "@/api/endpoint";
 import {
   Bookmark,
-  BookmarkedArticle,
   BookmarkPayload,
+  BookmarkedArticle,
 } from "@/api/schema/feed-management/bookmark";
 import {
   ArticleFilters,

apps/mobile-legacy/src/hooks/use-relative-time.ts

@@ -1,4 +1,4 @@
-import { formatDistanceToNowStrict, Locale } from "date-fns";
+import { Locale, formatDistanceToNowStrict } from "date-fns";
 import { fr } from "date-fns/locale";
 import { useEffect, useState } from "react";
 

apps/mobile-legacy/src/providers/root-providers.tsx

@@ -1,5 +1,4 @@
 import type React from "react";
-
 import { GestureHandlerRootView } from "react-native-gesture-handler";
 import { SafeAreaProvider } from "react-native-safe-area-context";
 

apps/mobile-legacy/src/providers/tamagui-config-provider.tsx

@@ -1,5 +1,4 @@
 import type React from "react";
-
 import { TamaguiProvider } from "tamagui";
 
 import { config } from "~/tamagui.config";

apps/mobile-legacy/src/ui/components/content/article/ArticleList.tsx

@@ -1,5 +1,4 @@
 import React, { useCallback } from "react";
-
 import { ActivityIndicator, Dimensions, FlatList, FlatListProps } from "react-native";
 import { View, XStack, YStack } from "tamagui";
 

apps/mobile-legacy/src/ui/components/content/article/ArticleSkeleton.tsx

@@ -1,5 +1,4 @@
 import { useCallback } from "react";
-
 import ContentLoader, { Circle, Rect } from "react-content-loader/native";
 import { Dimensions, FlatList } from "react-native";
 import { View } from "tamagui";

apps/mobile-legacy/src/ui/components/content/bookmark/BookmarkList.tsx

@@ -1,5 +1,4 @@
 import React, { useCallback } from "react";
-
 import { ActivityIndicator, FlatList, FlatListProps } from "react-native";
 import { YStack } from "tamagui";
 

apps/mobile-legacy/src/ui/components/content/source/SourceFollowButton.tsx

@@ -1,5 +1,4 @@
 import { useCallback, useState } from "react";
-
 import { ActivityIndicator, Alert } from "react-native";
 import { Button, GetProps } from "tamagui";
 

apps/mobile-legacy/src/ui/components/content/source/SourceList.tsx

@@ -1,5 +1,4 @@
 import React, { useCallback } from "react";
-
 import { FlatList, FlatListProps } from "react-native";
 import { Paragraph, XStack, YStack } from "tamagui";
 

apps/mobile-legacy/src/ui/components/content/source/SourceOverviewCard.tsx

@@ -1,5 +1,5 @@
 import { Link } from "expo-router";
-import { GetProps, styled, XStack, YStack } from "tamagui";
+import { GetProps, XStack, YStack, styled } from "tamagui";
 
 import { SourceOverview } from "@/api/schema/feed-management/source";
 import { SourceFollowButton } from "@/ui/components/content/source/SourceFollowButton";

apps/mobile-legacy/src/ui/components/content/source/SourceSkeleton.tsx

@@ -1,5 +1,4 @@
 import { useCallback } from "react";
-
 import ContentLoader, { Circle, Rect } from "react-content-loader/native";
 import { FlatList } from "react-native";
 import { YStack } from "tamagui";

apps/mobile-legacy/src/ui/components/controls/forms/Input.tsx

@@ -5,10 +5,10 @@ import {
   GetProps,
   Label,
   SizeTokens,
-  styled,
   Input as TamaguiInput,
   XStack,
   YStack,
+  styled,
 } from "tamagui";
 
 import { Caption } from "@/ui/components/typography";

apps/mobile-legacy/src/ui/components/controls/forms/TextArea.tsx

@@ -1,4 +1,4 @@
-import { GetProps, Label, styled, TextArea as TamaguiTextArea, XStack, YStack } from "tamagui";
+import { GetProps, Label, TextArea as TamaguiTextArea, XStack, YStack, styled } from "tamagui";
 
 import { withController } from "@/ui/components/controls/forms/withController";
 import { Caption } from "@/ui/components/typography";

apps/mobile-legacy/src/ui/components/controls/forms/withController.tsx

@@ -1,5 +1,4 @@
 import React from "react";
-
 import { Controller, ControllerProps } from "react-hook-form";
 
 type WithControllerProps = {

apps/mobile-legacy/src/ui/components/layout/ScreenHeading.tsx

@@ -1,6 +1,5 @@
 import React from "react";
-
-import { styled, View, XStack } from "tamagui";
+import { View, XStack, styled } from "tamagui";
 
 import { Text } from "@/ui/components/typography";
 

apps/mobile-legacy/src/ui/components/layout/ScreenSection.tsx

@@ -1,6 +1,6 @@
 import { ArrowRight } from "@tamagui/lucide-icons";
 import { Href, Link } from "expo-router";
-import { GetProps, Paragraph, styled, XStack } from "tamagui";
+import { GetProps, Paragraph, XStack, styled } from "tamagui";
 
 import { Text } from "@/ui/components/typography";
 

apps/mobile-legacy/src/ui/components/layout/ScreenView.tsx

@@ -1,7 +1,7 @@
 import { StatusBar } from "expo-status-bar";
 import React from "react";
 import { useSafeAreaInsets } from "react-native-safe-area-context";
-import { styled, YStack } from "tamagui";
+import { YStack, styled } from "tamagui";
 
 import { ScreenHeading } from "@/ui/components/layout/ScreenHeading";
 import { ScreenSection } from "@/ui/components/layout/ScreenSection";

apps/mobile-legacy/src/ui/components/typography/Caption.tsx

@@ -1,5 +1,4 @@
 import type React from "react";
-
 import { Paragraph, ParagraphProps } from "tamagui";
 
 export const Caption = (props: React.PropsWithChildren<ParagraphProps>) => {

apps/mobile-legacy/src/ui/components/typography/Display.tsx

@@ -1,5 +1,4 @@
 import type React from "react";
-
 import { H2, ParagraphProps } from "tamagui";
 
 export const Display = (props: React.PropsWithChildren<ParagraphProps>) => {

apps/mobile-legacy/src/ui/components/typography/Heading.tsx

@@ -1,5 +1,4 @@
 import type React from "react";
-
 import { H4, ParagraphProps } from "tamagui";
 
 export const Heading = (props: React.PropsWithChildren<ParagraphProps>) => {

apps/mobile-legacy/src/ui/components/typography/Text.tsx

@@ -1,5 +1,4 @@
 import type React from "react";
-
 import { Paragraph, ParagraphProps } from "tamagui";
 
 export const Text = (props: React.PropsWithChildren<ParagraphProps>) => {

biome.json

@@ -39,7 +39,7 @@
   },
   "files": {
     "ignoreUnknown": true,
-    "includes": ["**/apps", "**/packages", "!/apps/api-legacy", "!/apps/mobile-legacy"]
+    "includes": ["**", "!apps/mobile-legacy", "!apps/api-legacy"]
   },
   "formatter": {
     "enabled": true,

packages/db/src/client.ts

@@ -1,5 +1,6 @@
 import { drizzle } from "drizzle-orm/node-postgres";
 import { Pool } from "pg";
+
 import * as schema from "@/schema";
 
 const isDevelopment = process.env.NODE_ENV === "development";
@@ -21,7 +22,7 @@ const pool = new Pool({
 export const getConnectionPoolStats = () => {
   const stats = {
     active: Math.max(0, (pool.totalCount ?? 0) - (pool.idleCount ?? 0)),
-    ended: (pool as any).ended ?? false,
+    ended: pool.ended ?? false,
     idle: pool.idleCount ?? 0,
     name: "primary",
     total: pool.options.max ?? 0,

packages/db/src/queries/articles.ts

@@ -4,13 +4,13 @@ import { and, asc, desc, eq, gt, lt, or, sql } from "drizzle-orm";
 import type { Database } from "@/client";
 import { articles, bookmarkArticles, bookmarks, comments, sources, users } from "@/schema";
 import {
-  buildPaginationResult,
-  createPageState,
-  decodeCursor,
   type PageRequest,
   type PageState,
   type PaginationMeta,
   type SortDirection,
+  buildPaginationResult,
+  createPageState,
+  decodeCursor,
 } from "@/utils/pagination";
 
 export interface ArticleFilters {

packages/db/src/queries/bookmarks.ts

@@ -4,11 +4,11 @@ import { and, desc, eq, lt, sql } from "drizzle-orm";
 import type { Database } from "@/client";
 import { bookmarkArticles, bookmarks } from "@/schema";
 import {
+  type PageRequest,
+  type PaginationMeta,
   buildPaginationResult,
   createPageState,
   decodeCursor,
-  type PageRequest,
-  type PaginationMeta,
 } from "@/utils/pagination";
 
 export interface BookmarkRow {

packages/db/src/queries/sources.ts

@@ -5,11 +5,11 @@ import type { Database } from "@/client";
 import { PUBLICATION_GRAPH_DAYS, SOURCE_IMAGE_BASE } from "@/constant";
 import { articles, followedSources, sources } from "@/schema";
 import {
+  type PageRequest,
+  type PaginationMeta,
   buildPaginationResult,
   createPageState,
   decodeCursor,
-  type PageRequest,
-  type PaginationMeta,
 } from "@/utils/pagination";
 
 export interface SourceOverviewRow {

packages/db/src/utils/health.ts

@@ -1,4 +1,5 @@
 import { sql } from "drizzle-orm";
+
 import { db } from "@/client";
 
 export async function checkHealth() {

packages/tsconfig/base.json

@@ -7,14 +7,15 @@
     "incremental": false,
     "isolatedModules": true,
     "lib": ["es2022", "DOM", "DOM.Iterable"],
-    "module": "NodeNext",
+    "module": "ESNext",
     "moduleDetection": "force",
-    "moduleResolution": "NodeNext",
+    "moduleResolution": "Bundler",
     "noUncheckedIndexedAccess": true,
     "resolveJsonModule": true,
     "skipLibCheck": true,
     "strict": true,
-    "target": "ES2022"
+    "target": "ES2022",
+    "verbatimModuleSyntax": false
   },
   "display": "Default"
 }

packages/ui/src/components/button.tsx

@@ -1,6 +1,6 @@
 import { cn } from "@basango/ui/lib/utils";
 import { Slot } from "@radix-ui/react-slot";
-import { cva, type VariantProps } from "class-variance-authority";
+import { type VariantProps, cva } from "class-variance-authority";
 import * as React from "react";
 
 const buttonVariants = cva(