PR fetch should be more lenient

This change tries to account for issue like the one described in
mc1arke/sonarqube-community-branch-plugin#759 (comment)

src/main/java/org/jenkinsci/plugins/sonargerrit/sonar/pull_request_analysis/PullRequestAnalysisTask.java

@@ -134,17 +134,25 @@ public List<Issue> fetchIssues(TaskListener listener) throws InterruptedExceptio
  String.format("No pull request found for SonarQube task '%s'", taskId));
  }
 
- ProjectPullRequests.PullRequest pullRequest =
- sonarClient.projectPullRequests().list(new ListRequest().setProject(componentKey))
- .getPullRequestsList().stream()
- .filter(pr -> pullRequestKey.equals(pr.getKey()))
- .findFirst()
- .orElseThrow(
- () ->
- new IllegalStateException(
- String.format(
- "No pull request found for project '%s' and key '%s'.",
- componentKey, pullRequestKey)));
+ ProjectPullRequests.PullRequest pullRequest;
+ while (true) {
+ pullRequest =
+ sonarClient.projectPullRequests().list(new ListRequest().setProject(componentKey))
+ .getPullRequestsList().stream()
+ .filter(pr -> pullRequestKey.equals(pr.getKey()))
+ .findFirst()
+ .orElse(null);
+ if (pullRequest != null) {
+ break;
+ }
+ TaskListenerLogger.log(
+ listener,
+ "Waiting %s before re-performing lookup of pull request '%s' on project '%s' ...",
+ CE_TASK_COMPLETION_CHECK_INTERVAL,
+ pullRequestKey,
+ componentKey);
+ Thread.sleep(CE_TASK_COMPLETION_CHECK_INTERVAL.toMillis());
+ }
 
  SearchRequest issueSearchRequest =
  new SearchRequest()
@@ -159,8 +167,10 @@ public List<Issue> fetchIssues(TaskListener listener) throws InterruptedExceptio
  .map(PullRequestComponent::new)
  .collect(Collectors.toList()));
 
+ ProjectPullRequests.PullRequest finalPullRequest = pullRequest;
+
  return issueSearchResponse.getIssuesList().stream()
- .map(issue -> new PullRequestIssue(pullRequest, components, issue, serverUrl))
+ .map(issue -> new PullRequestIssue(finalPullRequest, components, issue, serverUrl))
  .collect(Collectors.toList());
  }