|
| 1 | +import { createReadStream } from "node:fs"; |
| 2 | +import * as fs from "node:fs/promises"; |
| 3 | +import * as path from "node:path"; |
| 4 | +import * as readline from "node:readline"; |
| 5 | +import { z } from "zod"; |
| 6 | + |
| 7 | +import { |
| 8 | + isTimestampInRange, |
| 9 | + rangeOverlapsUtcDay, |
| 10 | + type TelemetryDateRange, |
| 11 | +} from "./range"; |
| 12 | + |
| 13 | +import type { ExportTelemetryEvent } from "./types"; |
| 14 | + |
| 15 | +const TELEMETRY_FILE_PATTERN = |
| 16 | + /^telemetry-(\d{4}-\d{2}-\d{2})-([a-zA-Z0-9]+)(?:\.(\d+))?\.jsonl$/; |
| 17 | + |
| 18 | +const StoredTelemetryEventSchema = z.object({ |
| 19 | + event_id: z.string(), |
| 20 | + event_name: z.string(), |
| 21 | + timestamp: z.string(), |
| 22 | + event_sequence: z.number().finite(), |
| 23 | + context: z.object({ |
| 24 | + extension_version: z.string(), |
| 25 | + machine_id: z.string(), |
| 26 | + session_id: z.string(), |
| 27 | + os_type: z.string(), |
| 28 | + os_version: z.string(), |
| 29 | + host_arch: z.string(), |
| 30 | + platform_name: z.string(), |
| 31 | + platform_version: z.string(), |
| 32 | + deployment_url: z.string(), |
| 33 | + }), |
| 34 | + properties: z.record(z.string(), z.string()), |
| 35 | + measurements: z.record(z.string(), z.number().finite()), |
| 36 | + trace_id: z.string().optional(), |
| 37 | + parent_event_id: z.string().optional(), |
| 38 | + error: z |
| 39 | + .object({ |
| 40 | + message: z.string(), |
| 41 | + type: z.string().optional(), |
| 42 | + code: z.string().optional(), |
| 43 | + }) |
| 44 | + .optional(), |
| 45 | +}); |
| 46 | + |
| 47 | +const ExportTelemetryEventSchema = StoredTelemetryEventSchema.transform( |
| 48 | + (event): ExportTelemetryEvent => ({ |
| 49 | + eventId: event.event_id, |
| 50 | + eventName: event.event_name, |
| 51 | + timestamp: event.timestamp, |
| 52 | + eventSequence: event.event_sequence, |
| 53 | + context: { |
| 54 | + extensionVersion: event.context.extension_version, |
| 55 | + machineId: event.context.machine_id, |
| 56 | + sessionId: event.context.session_id, |
| 57 | + osType: event.context.os_type, |
| 58 | + osVersion: event.context.os_version, |
| 59 | + hostArch: event.context.host_arch, |
| 60 | + platformName: event.context.platform_name, |
| 61 | + platformVersion: event.context.platform_version, |
| 62 | + deploymentUrl: event.context.deployment_url, |
| 63 | + }, |
| 64 | + properties: event.properties, |
| 65 | + measurements: event.measurements, |
| 66 | + ...(event.trace_id !== undefined && { traceId: event.trace_id }), |
| 67 | + ...(event.parent_event_id !== undefined && { |
| 68 | + parentEventId: event.parent_event_id, |
| 69 | + }), |
| 70 | + ...(event.error !== undefined && { error: event.error }), |
| 71 | + }), |
| 72 | +); |
| 73 | + |
| 74 | +type StoredTelemetryEvent = z.infer<typeof StoredTelemetryEventSchema>; |
| 75 | + |
| 76 | +interface TelemetryFileCandidate { |
| 77 | + readonly name: string; |
| 78 | + readonly date: string; |
| 79 | + readonly sessionSlug: string; |
| 80 | + readonly segment: number; |
| 81 | +} |
| 82 | + |
| 83 | +export async function listTelemetryFilesForRange( |
| 84 | + telemetryDir: string, |
| 85 | + range: TelemetryDateRange, |
| 86 | +): Promise<string[]> { |
| 87 | + let names: string[]; |
| 88 | + try { |
| 89 | + names = await fs.readdir(telemetryDir); |
| 90 | + } catch (err) { |
| 91 | + if ((err as NodeJS.ErrnoException).code === "ENOENT") { |
| 92 | + return []; |
| 93 | + } |
| 94 | + throw err; |
| 95 | + } |
| 96 | + |
| 97 | + return names |
| 98 | + .map((name) => telemetryFileCandidate(name)) |
| 99 | + .filter( |
| 100 | + (candidate): candidate is TelemetryFileCandidate => |
| 101 | + candidate !== undefined && rangeOverlapsUtcDay(candidate.date, range), |
| 102 | + ) |
| 103 | + .sort(compareTelemetryFiles) |
| 104 | + .map(({ name }) => path.join(telemetryDir, name)); |
| 105 | +} |
| 106 | + |
| 107 | +function telemetryFileCandidate( |
| 108 | + name: string, |
| 109 | +): TelemetryFileCandidate | undefined { |
| 110 | + const match = TELEMETRY_FILE_PATTERN.exec(name); |
| 111 | + if (!match) { |
| 112 | + return undefined; |
| 113 | + } |
| 114 | + return { |
| 115 | + name, |
| 116 | + date: match[1], |
| 117 | + sessionSlug: match[2], |
| 118 | + segment: match[3] === undefined ? 0 : Number(match[3]), |
| 119 | + }; |
| 120 | +} |
| 121 | + |
| 122 | +function compareTelemetryFiles( |
| 123 | + a: TelemetryFileCandidate, |
| 124 | + b: TelemetryFileCandidate, |
| 125 | +): number { |
| 126 | + return ( |
| 127 | + a.date.localeCompare(b.date) || |
| 128 | + a.sessionSlug.localeCompare(b.sessionSlug) || |
| 129 | + a.segment - b.segment || |
| 130 | + a.name.localeCompare(b.name) |
| 131 | + ); |
| 132 | +} |
| 133 | + |
| 134 | +export async function* readTelemetryEvents( |
| 135 | + filePaths: readonly string[], |
| 136 | + range: TelemetryDateRange, |
| 137 | +): AsyncGenerator<ExportTelemetryEvent> { |
| 138 | + for (const filePath of filePaths) { |
| 139 | + let lineNumber = 0; |
| 140 | + const lines = readline.createInterface({ |
| 141 | + input: createReadStream(filePath, { encoding: "utf8" }), |
| 142 | + crlfDelay: Infinity, |
| 143 | + }); |
| 144 | + try { |
| 145 | + for await (const line of lines) { |
| 146 | + lineNumber += 1; |
| 147 | + if (line.trim() === "") { |
| 148 | + continue; |
| 149 | + } |
| 150 | + const event = parseStoredTelemetryEvent(line, filePath, lineNumber); |
| 151 | + if (isTimestampInRange(event.timestamp, range)) { |
| 152 | + yield event; |
| 153 | + } |
| 154 | + } |
| 155 | + } catch (err) { |
| 156 | + throw wrapReadError(err, filePath, lineNumber); |
| 157 | + } |
| 158 | + } |
| 159 | +} |
| 160 | + |
| 161 | +export function parseStoredTelemetryEvent( |
| 162 | + line: string, |
| 163 | + filePath = "<telemetry>", |
| 164 | + lineNumber = 1, |
| 165 | +): ExportTelemetryEvent { |
| 166 | + try { |
| 167 | + return ExportTelemetryEventSchema.parse(JSON.parse(line)); |
| 168 | + } catch (err) { |
| 169 | + throw new Error( |
| 170 | + `Failed to parse telemetry file ${path.basename(filePath)}:${lineNumber}: ${errorMessage(err)}`, |
| 171 | + { cause: err }, |
| 172 | + ); |
| 173 | + } |
| 174 | +} |
| 175 | + |
| 176 | +export function toStoredTelemetryEvent( |
| 177 | + event: ExportTelemetryEvent, |
| 178 | +): StoredTelemetryEvent { |
| 179 | + return { |
| 180 | + event_id: event.eventId, |
| 181 | + event_name: event.eventName, |
| 182 | + timestamp: event.timestamp, |
| 183 | + event_sequence: event.eventSequence, |
| 184 | + context: { |
| 185 | + extension_version: event.context.extensionVersion, |
| 186 | + machine_id: event.context.machineId, |
| 187 | + session_id: event.context.sessionId, |
| 188 | + os_type: event.context.osType, |
| 189 | + os_version: event.context.osVersion, |
| 190 | + host_arch: event.context.hostArch, |
| 191 | + platform_name: event.context.platformName, |
| 192 | + platform_version: event.context.platformVersion, |
| 193 | + deployment_url: event.context.deploymentUrl, |
| 194 | + }, |
| 195 | + properties: event.properties, |
| 196 | + measurements: event.measurements, |
| 197 | + ...(event.traceId !== undefined && { trace_id: event.traceId }), |
| 198 | + ...(event.parentEventId !== undefined && { |
| 199 | + parent_event_id: event.parentEventId, |
| 200 | + }), |
| 201 | + ...(event.error !== undefined && { error: event.error }), |
| 202 | + }; |
| 203 | +} |
| 204 | + |
| 205 | +function wrapReadError( |
| 206 | + err: unknown, |
| 207 | + filePath: string, |
| 208 | + lineNumber: number, |
| 209 | +): Error { |
| 210 | + if (err instanceof Error && err.message.includes(path.basename(filePath))) { |
| 211 | + return err; |
| 212 | + } |
| 213 | + const location = lineNumber > 0 ? `:${lineNumber}` : ""; |
| 214 | + return new Error( |
| 215 | + `Failed to read telemetry file ${path.basename(filePath)}${location}: ${errorMessage(err)}`, |
| 216 | + { cause: err }, |
| 217 | + ); |
| 218 | +} |
| 219 | + |
| 220 | +function errorMessage(err: unknown): string { |
| 221 | + return err instanceof z.ZodError |
| 222 | + ? z.prettifyError(err) |
| 223 | + : err instanceof Error |
| 224 | + ? err.message |
| 225 | + : String(err); |
| 226 | +} |
0 commit comments