Add size limit to jira tickets

backend/danswer/configs/app_configs.py

@@ -247,6 +247,10 @@
     for ignored_tag in os.environ.get("JIRA_CONNECTOR_LABELS_TO_SKIP", "").split(",")
     if ignored_tag
 ]
+# Maximum size for Jira tickets in bytes (default: 100KB)
+JIRA_CONNECTOR_MAX_TICKET_SIZE = int(
+    os.environ.get("JIRA_CONNECTOR_MAX_TICKET_SIZE", 100 * 1024)
+)
 
 GONG_CONNECTOR_START_TIME = os.environ.get("GONG_CONNECTOR_START_TIME")
 

backend/danswer/connectors/danswer_jira/connector.py

@@ -9,6 +9,7 @@
 
 from danswer.configs.app_configs import INDEX_BATCH_SIZE
 from danswer.configs.app_configs import JIRA_CONNECTOR_LABELS_TO_SKIP
+from danswer.configs.app_configs import JIRA_CONNECTOR_MAX_TICKET_SIZE
 from danswer.configs.constants import DocumentSource
 from danswer.connectors.cross_connector_utils.miscellaneous_utils import time_str_to_utc
 from danswer.connectors.interfaces import GenerateDocumentsOutput
@@ -134,10 +135,20 @@ def fetch_jira_issues_batch(
             else extract_text_from_adf(jira.raw["fields"]["description"])
         )
         comments = _get_comment_strs(jira, comment_email_blacklist)
-        semantic_rep = f"{description}\n" + "\n".join(
+        ticket_content = f"{description}\n" + "\n".join(
             [f"Comment: {comment}" for comment in comments if comment]
         )
 
+        # Check ticket size
+        if len(ticket_content.encode("utf-8")) > JIRA_CONNECTOR_MAX_TICKET_SIZE:
+            logger.info(
+                f"Skipping {jira.key} because it exceeds the maximum size of "
+                f"{JIRA_CONNECTOR_MAX_TICKET_SIZE} bytes."
+            )
+            continue
+
+        semantic_rep = ticket_content
+
         page_url = f"{jira_client.client_info()}/browse/{jira.key}"
 
         people = set()

backend/tests/unit/danswer/connectors/jira/test_large_ticket_handling.py

@@ -0,0 +1,119 @@
+from unittest.mock import MagicMock
+from unittest.mock import patch
+
+import pytest
+from jira.resources import Issue
+
+from danswer.connectors.danswer_jira.connector import fetch_jira_issues_batch
+
+
+@pytest.fixture
+def mock_jira_client():
+    return MagicMock()
+
+
+@pytest.fixture
+def mock_issue_small():
+    issue = MagicMock()
+    issue.key = "SMALL-1"
+    issue.fields.description = "Small description"
+    issue.fields.comment.comments = [
+        MagicMock(body="Small comment 1"),
+        MagicMock(body="Small comment 2"),
+    ]
+    issue.fields.creator.displayName = "John Doe"
+    issue.fields.creator.emailAddress = "[email protected]"
+    issue.fields.summary = "Small Issue"
+    issue.fields.updated = "2023-01-01T00:00:00+0000"
+    issue.fields.labels = []
+    return issue
+
+
+@pytest.fixture
+def mock_issue_large():
+    # This will be larger than 100KB
+    issue = MagicMock()
+    issue.key = "LARGE-1"
+    issue.fields.description = "a" * 99_000
+    issue.fields.comment.comments = [
+        MagicMock(body="Large comment " * 1000),
+        MagicMock(body="Another large comment " * 1000),
+    ]
+    issue.fields.creator.displayName = "Jane Doe"
+    issue.fields.creator.emailAddress = "[email protected]"
+    issue.fields.summary = "Large Issue"
+    issue.fields.updated = "2023-01-02T00:00:00+0000"
+    issue.fields.labels = []
+    return issue
+
+
+@pytest.fixture
+def patched_type():
+    def _patched_type(obj):
+        if isinstance(obj, MagicMock):
+            return Issue
+        return type(obj)
+
+    return _patched_type
+
+
+@pytest.fixture
+def mock_jira_api_version():
+    with patch("danswer.connectors.danswer_jira.connector.JIRA_API_VERSION", "2"):
+        yield
+
+
+@pytest.fixture
+def patched_environment(patched_type, mock_jira_api_version):
+    with patch("danswer.connectors.danswer_jira.connector.type", patched_type):
+        yield
+
+
+def test_fetch_jira_issues_batch_small_ticket(
+    mock_jira_client, mock_issue_small, patched_environment
+):
+    mock_jira_client.search_issues.return_value = [mock_issue_small]
+
+    docs, count = fetch_jira_issues_batch("project = TEST", 0, mock_jira_client)
+
+    assert count == 1
+    assert len(docs) == 1
+    assert docs[0].id.endswith("/SMALL-1")
+    assert "Small description" in docs[0].sections[0].text
+    assert "Small comment 1" in docs[0].sections[0].text
+    assert "Small comment 2" in docs[0].sections[0].text
+
+
+def test_fetch_jira_issues_batch_large_ticket(
+    mock_jira_client, mock_issue_large, patched_environment
+):
+    mock_jira_client.search_issues.return_value = [mock_issue_large]
+
+    docs, count = fetch_jira_issues_batch("project = TEST", 0, mock_jira_client)
+
+    assert count == 1
+    assert len(docs) == 0  # The large ticket should be skipped
+
+
+def test_fetch_jira_issues_batch_mixed_tickets(
+    mock_jira_client, mock_issue_small, mock_issue_large, patched_environment
+):
+    mock_jira_client.search_issues.return_value = [mock_issue_small, mock_issue_large]
+
+    docs, count = fetch_jira_issues_batch("project = TEST", 0, mock_jira_client)
+
+    assert count == 2
+    assert len(docs) == 1  # Only the small ticket should be included
+    assert docs[0].id.endswith("/SMALL-1")
+
+
+@patch("danswer.connectors.danswer_jira.connector.JIRA_CONNECTOR_MAX_TICKET_SIZE", 50)
+def test_fetch_jira_issues_batch_custom_size_limit(
+    mock_jira_client, mock_issue_small, mock_issue_large, patched_environment
+):
+    mock_jira_client.search_issues.return_value = [mock_issue_small, mock_issue_large]
+
+    docs, count = fetch_jira_issues_batch("project = TEST", 0, mock_jira_client)
+
+    assert count == 2
+    assert len(docs) == 0  # Both tickets should be skipped due to the low size limit