Merge branch 'main' of https://github.com/mendableai/firecrawl

mendableai · Jan 25, 2025 · cf17479 · cf17479
2 parents d8d159b + eb22848
commit cf17479
Show file tree

Hide file tree

Showing 4 changed files with 34 additions and 6 deletions.
diff --git a/apps/api/src/lib/__tests__/html-transformer.test.ts b/apps/api/src/lib/__tests__/html-transformer.test.ts
@@ -315,5 +315,33 @@ describe("HTML Transformer", () => {
       expect(result).toContain("á é í ó ú ñ");
       expect(result).toContain("🎉 👍 🚀");
     });
+
+    it("should make all URLs absolute", async () => {
+      const options = {
+        html: `
+          <div>
+            <a href="https://example.com/fullurl">hi</a>
+            <a href="http://example.net/fullurl">hi</a>
+            <a href="/pathurl">hi</a>
+            <a href="//example.net/proturl">hi</a>
+            <a href="?queryurl">hi</a>
+            <a href="#hashurl">hi</a>
+          </div>
+        `,
+        url: "https://example.com",
+        include_tags: [],
+        exclude_tags: [],
+        only_main_content: true,
+      };
+
+      const result = await transformHtml(options);
+      console.log(result)
+      expect(result).toContain("https://example.com/fullurl");
+      expect(result).toContain("http://example.net/fullurl");
+      expect(result).toContain("https://example.com/pathurl");
+      expect(result).toContain("https://example.net/proturl");
+      expect(result).toContain("https://example.com/?queryurl");
+      expect(result).toContain("https://example.com/#hashurl");
+    });
   });
 });
diff --git a/apps/api/src/services/billing/credit_billing.ts b/apps/api/src/services/billing/credit_billing.ts
@@ -48,7 +48,7 @@ export async function supaBillTeam(
     credits,
   });
 
-  if (team_id === "preview") {
+  if (team_id === "preview" || team_id.startsWith("preview_")) {
     return { success: true, message: "Preview team, no credits used" };
   }
   _logger.info(`Billing team ${team_id} for ${credits} credits`);
@@ -109,7 +109,7 @@ export async function supaCheckTeamCredits(
   credits: number,
 ): Promise<CheckTeamCreditsResponse> {
   // WARNING: chunk will be null if team_id is preview -- do not perform operations on it under ANY circumstances - mogery
-  if (team_id === "preview") {
+  if (team_id === "preview" || team_id.startsWith("preview_")) {
     return {
       success: true,
       message: "Preview team, no credits used",

diff --git a/apps/api/src/services/logging/log_job.ts b/apps/api/src/services/logging/log_job.ts
@@ -50,7 +50,7 @@ export async function logJob(job: FirecrawlJob, force: boolean = false) {
       num_docs: job.num_docs,
       docs: cleanOfNull(job.docs),
       time_taken: job.time_taken,
-      team_id: job.team_id === "preview" ? null : job.team_id,
+      team_id: (job.team_id === "preview" || job.team_id?.startsWith("preview_"))? null : job.team_id,
       mode: job.mode,
       url: job.url,
       crawler_options: job.crawlerOptions,
@@ -112,7 +112,7 @@ export async function logJob(job: FirecrawlJob, force: boolean = false) {
     if (process.env.POSTHOG_API_KEY && !job.crawl_id) {
       let phLog = {
         distinctId: "from-api", //* To identify this on the group level, setting distinctid to a static string per posthog docs: https://posthog.com/docs/product-analytics/group-analytics#advanced-server-side-only-capturing-group-events-without-a-user
-        ...(job.team_id !== "preview" && {
+        ...((job.team_id !== "preview" && !job.team_id?.startsWith("preview_")) && {
           groups: { team: job.team_id },
         }), //* Identifying event on this team
         event: "job-logged",
@@ -121,7 +121,7 @@ export async function logJob(job: FirecrawlJob, force: boolean = false) {
           message: job.message,
           num_docs: job.num_docs,
           time_taken: job.time_taken,
-          team_id: job.team_id === "preview" ? null : job.team_id,
+          team_id: (job.team_id === "preview" || job.team_id?.startsWith("preview_"))? null : job.team_id,
           mode: job.mode,
           url: job.url,
           crawler_options: job.crawlerOptions,

diff --git a/apps/api/src/services/notification/email_notification.ts b/apps/api/src/services/notification/email_notification.ts
@@ -86,7 +86,7 @@ export async function sendNotificationInternal(
   chunk: AuthCreditUsageChunk,
   bypassRecentChecks: boolean = false,
 ): Promise<{ success: boolean }> {
-  if (team_id === "preview") {
+  if (team_id === "preview" || team_id.startsWith("preview_")) {
     return { success: true };
   }
   return await redlock.using(