From 6369d5f4627065f5285feecb54adbdde16bfd998 Mon Sep 17 00:00:00 2001
From: Predrag Stojadinovic <predrag@stojadinovic.net>
Date: Sat, 11 May 2024 18:43:55 +0200
Subject: [PATCH] chore: confluence data connector can now handle custom urls,
 in addition to default {subdomain}.atlassian.net ones

---
 .../utils/extensions/Confluence/index.js      | 44 ++++++++++++-------
 1 file changed, 29 insertions(+), 15 deletions(-)

diff --git a/collector/utils/extensions/Confluence/index.js b/collector/utils/extensions/Confluence/index.js
index 5a473f65..63094a1f 100644
--- a/collector/utils/extensions/Confluence/index.js
+++ b/collector/utils/extensions/Confluence/index.js
@@ -2,20 +2,30 @@ const fs = require("fs");
 const path = require("path");
 const { default: slugify } = require("slugify");
 const { v4 } = require("uuid");
+const UrlPattern = require("url-pattern");
 const { writeToServerDocuments } = require("../../files");
 const { tokenizeString } = require("../../tokenizer");
-const {
-  ConfluencePagesLoader,
-} = require("langchain/document_loaders/web/confluence");
+const { ConfluencePagesLoader } = require("langchain/document_loaders/web/confluence");
 
 function validSpaceUrl(spaceUrl = "") {
-  const UrlPattern = require("url-pattern");
-  const pattern = new UrlPattern(
-    "https\\://(:subdomain).atlassian.net/wiki/spaces/(:spaceKey)*"
-  );
-  const match = pattern.match(spaceUrl);
-  if (!match) return { valid: false, result: null };
-  return { valid: true, result: match };
+  // Atlassian default URL match
+  const atlassianPattern = new UrlPattern("https\\://(:subdomain).atlassian.net/wiki/spaces/(:spaceKey)/*");
+  const atlassianMatch = atlassianPattern.match(spaceUrl);
+  if (atlassianMatch) {
+    return { valid: true, result: atlassianMatch };
+  }
+
+  // Custom Confluence URL match
+  const customPattern = new UrlPattern("https\\://(:subdomain.):domain.:tld/wiki/spaces/(:spaceKey)/*");
+  const customMatch = customPattern.match(spaceUrl);
+  if (customMatch) {
+    customMatch.customDomain = (customMatch.subdomain ? `${customMatch.subdomain}.` : "") + //
+      (`${customMatch.domain}.${customMatch.tld}`);
+    return { valid: true, result: customMatch, custom: true };
+  }
+
+  // No match
+  return { valid: false, result: null };
 }
 
 async function loadConfluence({ pageUrl, username, accessToken }) {
@@ -31,15 +41,19 @@ async function loadConfluence({ pageUrl, username, accessToken }) {
   if (!validSpace.result) {
     return {
       success: false,
-      reason:
-        "Confluence space URL is not in the expected format of https://domain.atlassian.net/wiki/space/~SPACEID/*",
+      reason: "Confluence space URL is not in the expected format of https://domain.atlassian.net/wiki/space/~SPACEID/* or https://customDomain/wiki/space/~SPACEID/*",
     };
   }
 
-  const { subdomain, spaceKey } = validSpace.result;
-  console.log(`-- Working Confluence ${subdomain}.atlassian.net --`);
+  const { subdomain, customDomain, spaceKey } = validSpace.result;
+  let baseUrl = `https://${subdomain}.atlassian.net/wiki`;
+  if (customDomain) {
+    baseUrl = `https://${customDomain}/wiki`;
+  }
+
+  console.log(`-- Working Confluence ${baseUrl} --`);
   const loader = new ConfluencePagesLoader({
-    baseUrl: `https://${subdomain}.atlassian.net/wiki`,
+    baseUrl,
     spaceKey,
     username,
     accessToken,