Merge pull request #516 from kevinswiber/fix/use-db-auth-in-single-url-scraper

Ensuring USE_DB_AUTHENTICATION is true in single URL scraper.

apps/api/src/lib/logger.ts

@@ -25,7 +25,8 @@ export class Logger {
       const color = Logger.colors[level];
       console[level.toLowerCase()](color, `[${new Date().toISOString()}]${level} - ${message}`);
 
-      // if (process.env.USE_DB_AUTH) {
+      // const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+      // if (useDbAuthentication) {
       // save to supabase? another place?
       // supabase.from('logs').insert({ level: level, message: message, timestamp: new Date().toISOString(), success: boolean });
       // }

apps/api/src/lib/scrape-events.ts

@@ -36,7 +36,8 @@ export class ScrapeEvents {
   static async insert(jobId: string, content: ScrapeEvent) {
     if (jobId === "TEST") return null;
     
-    if (process.env.USE_DB_AUTHENTICATION) {
+    const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+    if (useDbAuthentication) {
       try {
         const result = await supabase.from("scrape_events").insert({
           job_id: jobId,

apps/api/src/lib/withAuth.ts

@@ -7,7 +7,8 @@ export function withAuth<T extends AuthResponse, U extends any[]>(
   originalFunction: (...args: U) => Promise<T>
 ) {
   return async function (...args: U): Promise<T> {
-    if (process.env.USE_DB_AUTHENTICATION === "false") {
+    const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+    if (!useDbAuthentication) {
       if (warningCount < 5) {
         Logger.warn("You're bypassing authentication");
         warningCount++;

apps/api/src/main/runWebScraper.ts

@@ -144,7 +144,8 @@ export async function runWebScraper({
 
 const saveJob = async (job: Job, result: any, token: string, mode: string) => {
   try {
-    if (process.env.USE_DB_AUTHENTICATION === "true") {
+    const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+    if (useDbAuthentication) {
       const { data, error } = await supabase_service
         .from("firecrawl_jobs")
         .update({ docs: result })

apps/api/src/scraper/WebScraper/single_url.ts

@@ -23,12 +23,15 @@ import { clientSideError } from "../../strings";
 
 dotenv.config();
 
+const useScrapingBee = process.env.SCRAPING_BEE_API_KEY !== '' && process.env.SCRAPING_BEE_API_KEY !== undefined;
+const useFireEngine = process.env.FIRE_ENGINE_BETA_URL !== '' && process.env.FIRE_ENGINE_BETA_URL !== undefined;
+
 export const baseScrapers = [
-  "fire-engine;chrome-cdp",
-  "fire-engine",
-  "scrapingBee",
-  process.env.USE_DB_AUTHENTICATION ? undefined : "playwright",
-  "scrapingBeeLoad",
+  useFireEngine ? "fire-engine;chrome-cdp" : undefined,
+  useFireEngine ? "fire-engine" : undefined,
+  useScrapingBee ? "scrapingBee" : undefined,
+  useFireEngine ? undefined : "playwright",
+  useScrapingBee ? "scrapingBeeLoad" : undefined,
   "fetch",
 ].filter(Boolean);
 
@@ -85,18 +88,18 @@ function getScrapingFallbackOrder(
   });
 
   let defaultOrder = [
-    !process.env.USE_DB_AUTHENTICATION ? undefined : "fire-engine;chrome-cdp",
-    !process.env.USE_DB_AUTHENTICATION ? undefined : "fire-engine",
-    "scrapingBee",
-    process.env.USE_DB_AUTHENTICATION ? undefined : "playwright",
-    "scrapingBeeLoad",
+    useFireEngine ? "fire-engine;chrome-cdp" : undefined,
+    useFireEngine ? "fire-engine" : undefined,
+    useScrapingBee ? "scrapingBee" : undefined,
+    useScrapingBee ? "scrapingBeeLoad" : undefined,
+    useFireEngine ? undefined : "playwright",
     "fetch",
   ].filter(Boolean);
 
   if (isWaitPresent || isScreenshotPresent || isHeadersPresent) {
     defaultOrder = [
       "fire-engine",
-      process.env.USE_DB_AUTHENTICATION ? undefined : "playwright",
+      useFireEngine ? undefined : "playwright",
       ...defaultOrder.filter(
         (scraper) => scraper !== "fire-engine" && scraper !== "playwright"
       ),

apps/api/src/services/logging/crawl_log.ts

@@ -3,7 +3,8 @@ import { Logger } from "../../../src/lib/logger";
 import "dotenv/config";
 
 export async function logCrawl(job_id: string, team_id: string) {
-  if (process.env.USE_DB_AUTHENTICATION === 'true') {
+  const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+  if (useDbAuthentication) {
     try {
       const { data, error } = await supabase_service
       .from("bulljobs_teams")

apps/api/src/services/logging/log_job.ts

@@ -7,7 +7,8 @@ import { Logger } from "../../lib/logger";
 
 export async function logJob(job: FirecrawlJob) {
   try {
-    if (process.env.USE_DB_AUTHENTICATION === "false") {
+    const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+    if (!useDbAuthentication) {
       return;
     }
 

apps/api/src/services/logging/scrape_log.ts

@@ -8,7 +8,8 @@ export async function logScrape(
   scrapeLog: ScrapeLog,
   pageOptions?: PageOptions
 ) {
-  if (process.env.USE_DB_AUTHENTICATION === "false") {
+  const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+  if (!useDbAuthentication) {
     Logger.debug("Skipping logging scrape to Supabase");
     return;
   }

apps/api/src/services/supabase.ts

@@ -8,8 +8,9 @@ class SupabaseService {
   constructor() {
     const supabaseUrl = process.env.SUPABASE_URL;
     const supabaseServiceToken = process.env.SUPABASE_SERVICE_TOKEN;
+    const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
     // Only initialize the Supabase client if both URL and Service Token are provided.
-    if (process.env.USE_DB_AUTHENTICATION === "false") {
+    if (!useDbAuthentication) {
       // Warn the user that Authentication is disabled by setting the client to null
       Logger.warn(
         "Authentication is disabled. Supabase client will not be initialized."

apps/test-suite/utils/supabase.ts

@@ -9,7 +9,8 @@ class SupabaseService {
     const supabaseUrl = process.env.SUPABASE_URL;
     const supabaseServiceToken = process.env.SUPABASE_SERVICE_TOKEN;
     // Only initialize the Supabase client if both URL and Service Token are provided.
-    if (process.env.USE_DB_AUTHENTICATION === "false") {
+    const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+    if (!useDbAuthentication) {
       // Warn the user that Authentication is disabled by setting the client to null
       console.warn(
         "Authentication is disabled. Supabase client will not be initialized."
@@ -36,7 +37,8 @@ export const supabase_service: SupabaseClient = new Proxy(
   new SupabaseService(),
   {
     get: function (target, prop, receiver) {
-      if (process.env.USE_DB_AUTHENTICATION === "false") {
+      const useDbAuthentication = process.env.USE_DB_AUTHENTICATION === 'true';
+      if (!useDbAuthentication) {
         console.debug(
           "Attempted to access Supabase client when it's not configured."
         );