Fix Asignee and Reporter metadata for Epics in JiraReader (#16842)

llama-index-integrations/readers/llama-index-readers-jira/CHANGELOG.md

@@ -1,5 +1,11 @@
 # CHANGELOG
 
+## [0.3.2] - 2024-11-05
+
+- Add assignee and reporter emails as metadata
+- Fix getting assignee and reporter for epics
+- Add include_epics parameter
+
 ## [0.1.2] - 2024-02-13
 
 - Add maintainers and keywords from library.json (llamahub)

llama-index-integrations/readers/llama-index-readers-jira/llama_index/readers/jira/base.py

@@ -39,6 +39,8 @@ class JiraReader(BaseReader):
         }
     """
 
+    include_epics: bool = True
+
     def __init__(
         self,
         email: Optional[str] = None,
@@ -47,6 +49,7 @@ def __init__(
         BasicAuth: Optional[BasicAuth] = None,
         Oauth2: Optional[Oauth2] = None,
         PATauth: Optional[PATauth] = None,
+        include_epics: bool = True,
     ) -> None:
         from jira import JIRA
 
@@ -75,6 +78,8 @@ def __init__(
                 server=f"https://{BasicAuth['server_url']}",
             )
 
+        self.include_epics = include_epics
+
     def load_data(
         self, query: str, start_at: int = 0, max_results: int = 50
     ) -> List[Document]:
@@ -91,14 +96,26 @@ def load_data(
         epic_descripton = ""
 
         for issue in relevant_issues:
-            # Iterates through only issues and not epics
-            if "parent" in (issue.raw["fields"]):
-                if issue.fields.assignee:
-                    assignee = issue.fields.assignee.displayName
-
-                if issue.fields.reporter:
-                    reporter = issue.fields.reporter.displayName
-
+            issue_type = issue.fields.issuetype.name
+            if issue_type == "Epic" and not self.include_epics:
+                continue
+
+            assignee = ""
+            assignee_email = ""
+            reporter = ""
+            reporter_email = ""
+            epic_key = ""
+            epic_summary = ""
+            epic_descripton = ""
+
+            if issue.fields.assignee:
+                assignee = issue.fields.assignee.displayName
+                assignee_email = issue.fields.assignee.emailAddress
+            if issue.fields.reporter:
+                reporter = issue.fields.reporter.displayName
+                reporter_email = issue.fields.reporter.emailAddress
+
+            if "parent" in issue.raw["fields"]:
                 if issue.raw["fields"]["parent"]["key"]:
                     epic_key = issue.raw["fields"]["parent"]["key"]
 
@@ -123,9 +140,11 @@ def load_data(
                         "labels": issue.fields.labels,
                         "status": issue.fields.status.name,
                         "assignee": assignee,
+                        "assignee_email": assignee_email,
                         "reporter": reporter,
+                        "reporter_email": reporter_email,
                         "project": issue.fields.project.name,
-                        "issue_type": issue.fields.issuetype.name,
+                        "issue_type": issue_type,
                         "priority": issue.fields.priority.name,
                         "epic_key": epic_key,
                         "epic_summary": epic_summary,

llama-index-integrations/readers/llama-index-readers-jira/pyproject.toml

@@ -29,7 +29,7 @@ license = "MIT"
 maintainers = ["bearguy"]
 name = "llama-index-readers-jira"
 readme = "README.md"
-version = "0.3.1"
+version = "0.3.2"
 
 [tool.poetry.dependencies]
 python = ">=3.8.1,<4.0"

llama-index-integrations/readers/llama-index-readers-jira/tests/BUILD

@@ -1 +1,3 @@
-python_tests()
+python_tests(
+    dependencies=["llama-index-integrations/readers/llama-index-readers-jira:poetry"],
+)

llama-index-integrations/readers/llama-index-readers-jira/tests/test_readers_jira.py

@@ -1,7 +1,173 @@
 from llama_index.core.readers.base import BaseReader
 from llama_index.readers.jira import JiraReader
+import pytest
+from unittest.mock import patch, MagicMock
 
 
 def test_class():
     names_of_base_classes = [b.__name__ for b in JiraReader.__mro__]
     assert BaseReader.__name__ in names_of_base_classes
+
+
+@pytest.fixture(autouse=True)
+def mock_jira():
+    with patch("jira.JIRA") as mock_jira:
+        mock_jira.return_value = MagicMock()
+        yield mock_jira
+
+
+@pytest.fixture()
+def mock_issue():
+    issue = MagicMock()
+    # Setup basic issue fields
+    issue.id = "TEST-123"
+    issue.fields.summary = "Test Summary"
+    issue.fields.description = "Test Description"
+    issue.fields.issuetype.name = "Story"
+    issue.fields.created = "2024-01-01"
+    issue.fields.updated = "2024-01-02"
+    issue.fields.labels = ["test-label"]
+    issue.fields.status.name = "In Progress"
+    issue.fields.project.name = "Test Project"
+    issue.fields.priority.name = "High"
+
+    # Setup assignee and reporter
+    issue.fields.assignee = MagicMock()
+    issue.fields.assignee.displayName = "Test Assignee"
+    issue.fields.assignee.emailAddress = "[email protected]"
+    issue.fields.reporter = MagicMock()
+    issue.fields.reporter.displayName = "Test Reporter"
+    issue.fields.reporter.emailAddress = "[email protected]"
+
+    # Setup raw fields for parent/epic info
+    issue.raw = {
+        "fields": {
+            "parent": {
+                "key": "EPIC-1",
+                "fields": {
+                    "summary": "Epic Summary",
+                    "status": {"description": "Epic Description"},
+                },
+            }
+        }
+    }
+
+    issue.permalink.return_value = "https://test.atlassian.net/browse/TEST-123"
+    return issue
+
+
+def test_basic_auth(mock_jira):
+    reader = JiraReader(
+        email="[email protected]",
+        api_token="test-token",
+        server_url="example.atlassian.net",
+    )
+
+    mock_jira.assert_called_once_with(
+        basic_auth=("[email protected]", "test-token"),
+        server="https://example.atlassian.net",
+    )
+
+
+def test_oauth2(mock_jira):
+    reader = JiraReader(Oauth2={"cloud_id": "test-cloud", "api_token": "test-token"})
+
+    mock_jira.assert_called_once_with(
+        options={
+            "server": "https://api.atlassian.com/ex/jira/test-cloud",
+            "headers": {"Authorization": "Bearer test-token"},
+        }
+    )
+
+
+def test_pat_auth(mock_jira):
+    reader = JiraReader(
+        PATauth={
+            "server_url": "https://example.atlassian.net",
+            "api_token": "test-token",
+        }
+    )
+
+    mock_jira.assert_called_once_with(
+        options={
+            "server": "https://example.atlassian.net",
+            "headers": {"Authorization": "Bearer test-token"},
+        }
+    )
+
+
+def test_load_data_basic(mock_jira, mock_issue):
+    # Setup mock JIRA instance
+    jira_instance = mock_jira.return_value
+    jira_instance.search_issues.return_value = [mock_issue]
+
+    reader = JiraReader(
+        email="[email protected]",
+        api_token="test-token",
+        server_url="example.atlassian.net",
+    )
+
+    documents = reader.load_data("project = TEST")
+
+    # Verify search_issues was called correctly
+    jira_instance.search_issues.assert_called_once_with(
+        "project = TEST", startAt=0, maxResults=50
+    )
+
+    # Verify returned document
+    assert len(documents) == 1
+    doc = documents[0]
+    assert doc.doc_id == "TEST-123"
+    assert doc.text == "Test Summary \n Test Description"
+
+    # Verify extra_info
+    assert doc.extra_info["assignee"] == "Test Assignee"
+    assert doc.extra_info["assignee_email"] == "[email protected]"
+    assert doc.extra_info["reporter"] == "Test Reporter"
+    assert doc.extra_info["reporter_email"] == "[email protected]"
+    assert doc.extra_info["epic_key"] == "EPIC-1"
+    assert doc.extra_info["epic_summary"] == "Epic Summary"
+    assert doc.extra_info["epic_description"] == "Epic Description"
+
+
+def test_load_data_exclude_epics(mock_jira, mock_issue):
+    # Modify mock issue to be an epic
+    mock_issue.fields.issuetype.name = "Epic"
+    jira_instance = mock_jira.return_value
+    jira_instance.search_issues.return_value = [mock_issue]
+
+    reader = JiraReader(
+        email="[email protected]",
+        api_token="test-token",
+        server_url="example.atlassian.net",
+        include_epics=False,
+    )
+
+    documents = reader.load_data("project = TEST")
+
+    # Verify no documents returned since epics are excluded
+    assert len(documents) == 0
+
+
+def test_load_data_no_assignee_reporter(mock_jira, mock_issue):
+    # Remove assignee and reporter
+    mock_issue.fields.assignee = None
+    mock_issue.fields.reporter = None
+    jira_instance = mock_jira.return_value
+    jira_instance.search_issues.return_value = [mock_issue]
+
+    reader = JiraReader(
+        email="[email protected]",
+        api_token="test-token",
+        server_url="example.atlassian.net",
+    )
+
+    documents = reader.load_data("project = TEST")
+
+    # Verify document has empty assignee/reporter fields
+    assert len(documents) == 1
+    doc = documents[0]
+    assert doc.extra_info["assignee"] == ""
+    assert doc.extra_info["assignee_email"] == ""
+    assert doc.extra_info["reporter"] == ""
+    assert doc.extra_info["reporter_email"] == ""