|
| 1 | +<?php |
| 2 | + |
| 3 | +// Find projects that were posted to PG a while ago and prune page data. |
| 4 | +class PrunePageData extends BackgroundJob |
| 5 | +{ |
| 6 | + public function work() |
| 7 | + { |
| 8 | + $database = SiteConfig::get()->archive_db_name ? SiteConfig::get()->archive_db_name : SiteConfig::get()->db_name; |
| 9 | + |
| 10 | + // We need to iteratively delete data in batches by projectid because: |
| 11 | + // 1. we base the pruning on the project's posted time |
| 12 | + // 2. there are indexes for these tables based on projectid |
| 13 | + // 3. to possibly limit our runtime |
| 14 | + $sql = sprintf( |
| 15 | + " |
| 16 | + SELECT projectid |
| 17 | + FROM projects |
| 18 | + WHERE |
| 19 | + modifieddate <= UNIX_TIMESTAMP() - (24 * 60 * 60) * %d |
| 20 | + AND state = '%s' |
| 21 | + ORDER BY modifieddate |
| 22 | + ", |
| 23 | + SiteConfig::get()->days_to_retain_page_data_after_posting, |
| 24 | + DPDatabase::escape(PROJ_SUBMIT_PG_POSTED) |
| 25 | + ); |
| 26 | + $result = DPDatabase::query($sql); |
| 27 | + $num_projects = mysqli_num_rows($result); |
| 28 | + |
| 29 | + echo "Pruning data for $num_projects projects...\n"; |
| 30 | + |
| 31 | + $num_projects_pruned = 0; |
| 32 | + while ([$project_id] = mysqli_fetch_row($result)) { |
| 33 | + if ($this->watch->read() >= $this->web_context_max_runtime_s) { |
| 34 | + break; |
| 35 | + } |
| 36 | + |
| 37 | + // first wordcheck_events |
| 38 | + $sql = sprintf( |
| 39 | + " |
| 40 | + DELETE FROM $database.wordcheck_events |
| 41 | + WHERE projectid = '%s' |
| 42 | + ", |
| 43 | + DPDatabase::escape($project_id) |
| 44 | + ); |
| 45 | + DPDatabase::query($sql); |
| 46 | + |
| 47 | + // then page_events |
| 48 | + $sql = sprintf( |
| 49 | + " |
| 50 | + DELETE FROM $database.page_events |
| 51 | + WHERE projectid = '%s' |
| 52 | + ", |
| 53 | + DPDatabase::escape($project_id) |
| 54 | + ); |
| 55 | + DPDatabase::query($sql); |
| 56 | + |
| 57 | + $num_projects_pruned += 1; |
| 58 | + } |
| 59 | + |
| 60 | + $leftover_projects = $num_projects - $num_projects_pruned; |
| 61 | + if ($leftover_projects) { |
| 62 | + echo "Reached runtime limit, skipping pruning of page data for remaining $leftover_projects projects.\n"; |
| 63 | + $this->stop_message = "Pruned page data for $num_projects_pruned projects, ran out of time to prune page data for remaining $leftover_projects"; |
| 64 | + } else { |
| 65 | + $this->stop_message = "Pruned page data for $num_projects_pruned projects"; |
| 66 | + } |
| 67 | + } |
| 68 | +} |
0 commit comments