|
| 1 | +#!/usr/bin/env tsx |
| 2 | +/** |
| 3 | + * Aggregates search index failures from multiple language runs into a single |
| 4 | + * consolidated report. Groups failures by page path to show which versions |
| 5 | + * and languages failed for each page. |
| 6 | + * |
| 7 | + * Usage: tsx aggregate-search-index-failures.ts <artifacts-dir> [--workflow-url <url>] |
| 8 | + * |
| 9 | + * Reads failures-summary.json files from subdirectories and outputs a formatted |
| 10 | + * message suitable for Slack notifications. |
| 11 | + */ |
| 12 | + |
| 13 | +import fs from 'fs' |
| 14 | +import path from 'path' |
| 15 | + |
| 16 | +interface Failure { |
| 17 | + url?: string |
| 18 | + relativePath?: string |
| 19 | + error: string |
| 20 | + errorType: string |
| 21 | +} |
| 22 | + |
| 23 | +interface LanguageFailures { |
| 24 | + indexName: string |
| 25 | + languageCode: string |
| 26 | + indexVersion: string |
| 27 | + failures: Failure[] |
| 28 | +} |
| 29 | + |
| 30 | +export interface FailuresSummary { |
| 31 | + totalFailedPages: number |
| 32 | + failures: LanguageFailures[] |
| 33 | +} |
| 34 | + |
| 35 | +interface PageFailure { |
| 36 | + versions: Set<string> |
| 37 | + languages: Set<string> |
| 38 | +} |
| 39 | + |
| 40 | +export interface AggregationResult { |
| 41 | + hasFailures: boolean |
| 42 | + message: string |
| 43 | + totalCount?: number |
| 44 | +} |
| 45 | + |
| 46 | +/** |
| 47 | + * Aggregates failures from multiple summaries into a single report. |
| 48 | + * Groups failures by page path to show which versions and languages failed for each. |
| 49 | + */ |
| 50 | +export function aggregateFailures( |
| 51 | + allFailures: FailuresSummary[], |
| 52 | + workflowUrl?: string, |
| 53 | +): AggregationResult { |
| 54 | + if (allFailures.length === 0) { |
| 55 | + return { hasFailures: false, message: '' } |
| 56 | + } |
| 57 | + |
| 58 | + // Group failures by page path |
| 59 | + const pageFailures = new Map<string, PageFailure>() |
| 60 | + |
| 61 | + for (const summary of allFailures) { |
| 62 | + for (const langFailures of summary.failures) { |
| 63 | + for (const failure of langFailures.failures) { |
| 64 | + const pagePath = failure.relativePath || failure.url || 'unknown' |
| 65 | + |
| 66 | + if (!pageFailures.has(pagePath)) { |
| 67 | + pageFailures.set(pagePath, { |
| 68 | + versions: new Set(), |
| 69 | + languages: new Set(), |
| 70 | + }) |
| 71 | + } |
| 72 | + |
| 73 | + const pageData = pageFailures.get(pagePath)! |
| 74 | + pageData.versions.add(langFailures.indexVersion) |
| 75 | + pageData.languages.add(langFailures.languageCode) |
| 76 | + } |
| 77 | + } |
| 78 | + } |
| 79 | + |
| 80 | + // Use unique page count, not total failure instances |
| 81 | + const uniquePageCount = pageFailures.size |
| 82 | + |
| 83 | + // Format the message |
| 84 | + const lines: string[] = [ |
| 85 | + `:warning: ${uniquePageCount} page(s) failed to scrape for general search indexing`, |
| 86 | + '', |
| 87 | + 'The indexing completed but some pages could not be scraped. This may affect search results for those pages.', |
| 88 | + '', |
| 89 | + ] |
| 90 | + |
| 91 | + // Sort pages alphabetically and format each |
| 92 | + const sortedPages = Array.from(pageFailures.entries()).sort((a, b) => a[0].localeCompare(b[0])) |
| 93 | + |
| 94 | + for (const [pagePath, data] of sortedPages) { |
| 95 | + const versions = Array.from(data.versions).sort().join(', ') |
| 96 | + const languages = Array.from(data.languages).sort().join(', ') |
| 97 | + lines.push(`• \`${pagePath}\` (versions: ${versions}, languages: ${languages})`) |
| 98 | + } |
| 99 | + |
| 100 | + if (workflowUrl) { |
| 101 | + lines.push('') |
| 102 | + lines.push(`Workflow: ${workflowUrl}`) |
| 103 | + } |
| 104 | + |
| 105 | + const message = lines.join('\n') |
| 106 | + |
| 107 | + return { hasFailures: true, message, totalCount: uniquePageCount } |
| 108 | +} |
| 109 | + |
| 110 | +/** |
| 111 | + * Reads failure summaries from artifact directories. |
| 112 | + */ |
| 113 | +export function readFailureSummaries(artifactsDir: string): FailuresSummary[] { |
| 114 | + const allFailures: FailuresSummary[] = [] |
| 115 | + const subdirs = fs.readdirSync(artifactsDir, { withFileTypes: true }) |
| 116 | + |
| 117 | + for (const subdir of subdirs) { |
| 118 | + if (!subdir.isDirectory()) continue |
| 119 | + |
| 120 | + const summaryPath = path.join(artifactsDir, subdir.name, 'failures-summary.json') |
| 121 | + if (fs.existsSync(summaryPath)) { |
| 122 | + const content = fs.readFileSync(summaryPath, 'utf-8') |
| 123 | + try { |
| 124 | + allFailures.push(JSON.parse(content) as FailuresSummary) |
| 125 | + } catch (error) { |
| 126 | + const message = error instanceof Error ? error.message : String(error) |
| 127 | + console.warn(`Warning: Failed to parse JSON in ${summaryPath}: ${message}`) |
| 128 | + } |
| 129 | + } |
| 130 | + } |
| 131 | + |
| 132 | + return allFailures |
| 133 | +} |
| 134 | + |
| 135 | +function main() { |
| 136 | + const args = process.argv.slice(2) |
| 137 | + const artifactsDir = args[0] |
| 138 | + const workflowUrlIndex = args.indexOf('--workflow-url') |
| 139 | + const workflowUrl = workflowUrlIndex !== -1 ? args[workflowUrlIndex + 1] : undefined |
| 140 | + |
| 141 | + if (!artifactsDir) { |
| 142 | + console.error( |
| 143 | + 'Usage: tsx aggregate-search-index-failures.ts <artifacts-dir> [--workflow-url <url>]', |
| 144 | + ) |
| 145 | + process.exit(1) |
| 146 | + } |
| 147 | + |
| 148 | + const allFailures = readFailureSummaries(artifactsDir) |
| 149 | + const result = aggregateFailures(allFailures, workflowUrl) |
| 150 | + console.log(JSON.stringify(result)) |
| 151 | +} |
| 152 | + |
| 153 | +// Only run main when executed directly (not imported) |
| 154 | +if (import.meta.url === `file://${process.argv[1]}`) { |
| 155 | + main() |
| 156 | +} |
0 commit comments