Update Jira endpoint to v3

Update Jira query to return all issue fields, otherwise, no fields will
be returned.
Optimise fetch logic to fix duplicates and avoid looping through all
issues

Author: phong-phuong

.changeset/large-cycles-thank.md

@@ -0,0 +1,5 @@
+---
+"@langchain/community": major
+---
+
+Update Jira document loader with updated API endpoint, handle edge case where jira fields are missing.

libs/langchain-community/src/document_loaders/web/jira.ts

@@ -324,7 +324,7 @@ export interface JiraProjectLoaderParams {
 }
 
 const API_ENDPOINTS = {
-  SEARCH: "/rest/api/2/search",
+  SEARCH: "/rest/api/3/search/jql",
 };
 
 /**
@@ -411,39 +411,59 @@ export class JiraProjectLoader extends BaseDocumentLoader {
     const url = `${this.host}${API_ENDPOINTS.SEARCH}`;
     const createdAfterAsString = this.toJiraDateString(this.createdAfter);
     let startAt = 0;
+    const seenIds = new Set<string>();
+    let hitDuplicate = false;
 
-    while (true) {
+    while (!hitDuplicate) {
       try {
         const jqlProps = [
           `project=${this.projectKey}`,
-          ...(createdAfterAsString ? [`created>=${createdAfterAsString}`] : []),
-        ];
-        const params = new URLSearchParams({
-          jql: jqlProps.join(" AND "),
-          startAt: `${startAt}`,
-          maxResults: `${this.limitPerRequest}`,
-        });
-        const pageUrl = `${url}?${params}`;
-
-        const options = {
-          method: "GET",
-          headers: {
-            Authorization: authorizationHeader,
-            Accept: "application/json",
-          },
-        };
-
-        const response = await fetch(pageUrl, options);
-        const data: JiraAPIResponse = await response.json();
-
-        if (!data.issues || data.issues.length === 0) break;
-
-        yield data.issues;
-        startAt += this.limitPerRequest;
+          ...(createdAfterAsString ? [`created>= "${createdAfterAsString}"`] : []),
+      ];
+      const jql = `${jqlProps.join(" AND ")} ORDER BY created ASC, key ASC`;
+
+      const params = new URLSearchParams({
+        jql,
+        startAt: `${startAt}`,
+        maxResults: `${this.limitPerRequest}`,
+        fields: '*all',
+      });
+      const pageUrl = `${url}?${params}`;
+
+      const options = {
+        method: "GET",
+        headers: {
+          Authorization: authorizationHeader,
+          Accept: "application/json",
+        },
+      };
+
+      const response = await fetch(pageUrl, options);
+      const data: JiraAPIResponse = await response.json();
+
+      if (!data.issues || data.issues.length === 0) break;
+
+      const uniqueIssues = [];
+      for (const issue of data.issues) {
+        if (seenIds.has(issue.id)) {
+          hitDuplicate = true;
+          break;
+        }
+        seenIds.add(issue.id);
+        uniqueIssues.push(issue);
+      }
+
+      if (uniqueIssues.length > 0) yield uniqueIssues;
+
+      startAt += this.limitPerRequest;
+
+      // Exit if we have fewer issues than limit (end of result set)
+      if (data.issues.length < this.limitPerRequest) break;
+
       } catch (error) {
-        console.error(error);
-        yield [];
+        console.error("Error fetching Jira issues:", error);
+        break;
       }
     }
   }
-}
+}