Bug 1311840 - Bail if we try to run crossreference_error_lines on a job with an existing TextLogSummary (#1942)

tests/log_parser/test_crossreference_error_lines.py

@@ -92,3 +92,26 @@ def test_crossreference_error_lines_missing(test_job):
         assert summary_line.failure_line == failure_line
         assert summary_line.verified is False
         assert summary_line.bug_number is None
+
+
+def test_crossreference_error_lines_repeat(test_job):
+
+    lines = [(test_line, {})]
+
+    create_failure_lines(test_job, lines)
+    create_text_log_errors(test_job, lines)
+
+    call_command('crossreference_error_lines', str(test_job.id))
+
+    failure_lines = FailureLine.objects.all()
+    summary_lines = TextLogSummaryLine.objects.all()
+    summary = TextLogSummary.objects.all()[0]
+    assert len(summary_lines) == 1
+
+    summary_line = summary_lines[0]
+    assert summary_line.summary == summary
+    assert summary_line.failure_line == failure_lines[0]
+    assert summary_line.verified is False
+    assert summary_line.bug_number is None
+
+    call_command('crossreference_error_lines', str(test_job.id))

treeherder/log_parser/crossreference.py

@@ -1,6 +1,8 @@
 import logging
 import re
 
+from django.db import (IntegrityError,
+                       transaction)
 from mozlog.formatters.tbplformatter import TbplFormatter
 
 from treeherder.model.models import (FailureLine,
@@ -21,6 +23,19 @@ def crossreference_job(job):
 
     :job: - Job for which to perform the crossreferencing
     """
+
+    try:
+        return _crossreference(job)
+    except IntegrityError:
+        logger.warning("IntegrityError crossreferencing error lines for job %s" % job.id)
+
+
+@transaction.atomic
+def _crossreference(job):
+    if TextLogSummary.objects.filter(job_guid=job.guid).exists():
+        logger.info("crossreference_error_lines already ran for job %s" % job.id)
+        return
+
     failure_lines = FailureLine.objects.filter(job_guid=job.guid)
 
     text_log_errors = TextLogError.objects.filter(
@@ -31,7 +46,7 @@ def crossreference_job(job):
     if not (failure_lines.exists() and text_log_errors.exists()):
         return []
 
-    summary, _ = TextLogSummary.objects.get_or_create(job_guid=job.guid,
+    summary = TextLogSummary.objects.create(job_guid=job.guid,
                                             repository=job.repository)
 
     match_iter = structured_iterator(failure_lines)