Pull in more fields for Jira

backend/onyx/connectors/onyx_jira/connector.py

@@ -44,6 +44,18 @@
 _JIRA_SLIM_PAGE_SIZE = 500
 _JIRA_FULL_PAGE_SIZE = 50
 
+# Constants for Jira field names
+_FIELD_REPORTER = "reporter"
+_FIELD_ASSIGNEE = "assignee"
+_FIELD_PRIORITY = "priority"
+_FIELD_STATUS = "status"
+_FIELD_RESOLUTION = "resolution"
+_FIELD_LABELS = "labels"
+_FIELD_KEY = "key"
+_FIELD_CREATED = "created"
+_FIELD_DUEDATE = "duedate"
+_FIELD_ISSUETYPE = "issuetype"
+
 
 def _perform_jql_search(
     jira_client: JIRA,
@@ -107,32 +119,40 @@ def process_jira_issue(
 
     page_url = build_jira_url(jira_client, issue.key)
 
+    metadata_dict: dict[str, str | list[str]] = {}
     people = set()
     try:
-        creator = best_effort_get_field_from_issue(issue, "creator")
+        creator = best_effort_get_field_from_issue(issue, _FIELD_REPORTER)
         if basic_expert_info := best_effort_basic_expert_info(creator):
             people.add(basic_expert_info)
+            metadata_dict[_FIELD_REPORTER] = basic_expert_info.get_semantic_name()
     except Exception:
         # Author should exist but if not, doesn't matter
         pass
 
     try:
-        assignee = best_effort_get_field_from_issue(issue, "assignee")
+        assignee = best_effort_get_field_from_issue(issue, _FIELD_ASSIGNEE)
         if basic_expert_info := best_effort_basic_expert_info(assignee):
             people.add(basic_expert_info)
+            metadata_dict[_FIELD_ASSIGNEE] = basic_expert_info.get_semantic_name()
     except Exception:
         # Author should exist but if not, doesn't matter
         pass
 
-    metadata_dict = {}
-    if priority := best_effort_get_field_from_issue(issue, "priority"):
-        metadata_dict["priority"] = priority.name
-    if status := best_effort_get_field_from_issue(issue, "status"):
-        metadata_dict["status"] = status.name
-    if resolution := best_effort_get_field_from_issue(issue, "resolution"):
-        metadata_dict["resolution"] = resolution.name
-    if labels := best_effort_get_field_from_issue(issue, "labels"):
-        metadata_dict["labels"] = labels
+    if priority := best_effort_get_field_from_issue(issue, _FIELD_PRIORITY):
+        metadata_dict[_FIELD_PRIORITY] = priority.name
+    if status := best_effort_get_field_from_issue(issue, _FIELD_STATUS):
+        metadata_dict[_FIELD_STATUS] = status.name
+    if resolution := best_effort_get_field_from_issue(issue, _FIELD_RESOLUTION):
+        metadata_dict[_FIELD_RESOLUTION] = resolution.name
+    if labels := best_effort_get_field_from_issue(issue, _FIELD_LABELS):
+        metadata_dict[_FIELD_LABELS] = labels
+    if created := best_effort_get_field_from_issue(issue, _FIELD_CREATED):
+        metadata_dict[_FIELD_CREATED] = created
+    if duedate := best_effort_get_field_from_issue(issue, _FIELD_DUEDATE):
+        metadata_dict[_FIELD_DUEDATE] = duedate
+    if issuetype := best_effort_get_field_from_issue(issue, _FIELD_ISSUETYPE):
+        metadata_dict[_FIELD_ISSUETYPE] = issuetype.name
 
     return Document(
         id=page_url,
@@ -277,7 +297,7 @@ def retrieve_all_slim_documents(
             max_results=_JIRA_SLIM_PAGE_SIZE,
             fields="key",
         ):
-            issue_key = best_effort_get_field_from_issue(issue, "key")
+            issue_key = best_effort_get_field_from_issue(issue, _FIELD_KEY)
             id = build_jira_url(self.jira_client, issue_key)
             slim_doc_batch.append(
                 SlimDocument(

backend/onyx/connectors/onyx_jira/utils.py

@@ -23,8 +23,8 @@
 def best_effort_basic_expert_info(obj: Any) -> BasicExpertInfo | None:
     display_name = None
     email = None
-    if hasattr(obj, "display_name"):
-        display_name = obj.display_name
+    if hasattr(obj, "displayName"):
+        display_name = obj.displayName
     else:
         display_name = obj.get("displayName")
 

backend/tests/daily/connectors/jira/test_jira_basic.py

@@ -4,6 +4,7 @@
 import pytest
 
 from onyx.configs.constants import DocumentSource
+from onyx.connectors.models import Document
 from onyx.connectors.onyx_jira.connector import JiraConnector
 from tests.daily.connectors.utils import load_all_docs_from_checkpoint_connector
 
@@ -30,19 +31,60 @@ def test_jira_connector_basic(jira_connector: JiraConnector) -> None:
         start=0,
         end=time.time(),
     )
-    assert len(docs) == 1
-    doc = docs[0]
-
-    assert doc.id == "https://danswerai.atlassian.net/browse/AS-2"
-    assert doc.semantic_identifier == "AS-2: test123small"
-    assert doc.source == DocumentSource.JIRA
-    assert doc.metadata == {"priority": "Medium", "status": "Backlog"}
-    assert doc.secondary_owners is None
-    assert doc.title == "AS-2 test123small"
-    assert doc.from_ingestion_api is False
-    assert doc.additional_info is None
-
-    assert len(doc.sections) == 1
-    section = doc.sections[0]
+    assert len(docs) == 2
+
+    # Find story and epic
+    story: Document | None = None
+    epic: Document | None = None
+    for doc in docs:
+        if doc.metadata["issuetype"] == "Story":
+            story = doc
+        elif doc.metadata["issuetype"] == "Epic":
+            epic = doc
+
+    assert story is not None
+    assert epic is not None
+
+    # Check task
+    assert story.id == "https://danswerai.atlassian.net/browse/AS-3"
+    assert story.semantic_identifier == "AS-3: test123small"
+    assert story.source == DocumentSource.JIRA
+    assert story.metadata == {
+        "priority": "Medium",
+        "status": "Backlog",
+        "reporter": "Chris Weaver",
+        "assignee": "Chris Weaver",
+        "issuetype": "Story",
+        "created": "2025-04-16T16:44:06.716-0700",
+    }
+    assert story.secondary_owners is None
+    assert story.title == "AS-3 test123small"
+    assert story.from_ingestion_api is False
+    assert story.additional_info is None
+
+    assert len(story.sections) == 1
+    section = story.sections[0]
+    assert section.text == "example_text\n"
+    assert section.link == "https://danswerai.atlassian.net/browse/AS-3"
+
+    # Check epic
+    assert epic.id == "https://danswerai.atlassian.net/browse/AS-4"
+    assert epic.semantic_identifier == "AS-4: EPIC"
+    assert epic.source == DocumentSource.JIRA
+    assert epic.metadata == {
+        "priority": "Medium",
+        "status": "Backlog",
+        "reporter": "Founder Onyx",
+        "assignee": "Chris Weaver",
+        "issuetype": "Epic",
+        "created": "2025-04-16T16:55:53.068-0700",
+    }
+    assert epic.secondary_owners is None
+    assert epic.title == "AS-4 EPIC"
+    assert epic.from_ingestion_api is False
+    assert epic.additional_info is None
+
+    assert len(epic.sections) == 1
+    section = epic.sections[0]
     assert section.text == "example_text\n"
-    assert section.link == "https://danswerai.atlassian.net/browse/AS-2"
+    assert section.link == "https://danswerai.atlassian.net/browse/AS-4"