peekinside 0.0.0-stage → 1.2.0

package/src/timeline/timeline.js

@@ -0,0 +1,642 @@
+/**
+ * The timeline of one request: everything that happened while it was handled, as spans (database
+ * queries, outgoing HTTP calls, console lines, exceptions, custom measures, event-loop blocks),
+ * plus counters for lists and filters (see the timeline data contract).
+ *
+ * Instrumentations find the timeline of the code running now with Timeline.current() and record
+ * through it. The timeline closes when the request's record is built: what starts later (work
+ * the app keeps doing after the response) is ignored, and spans still open at that moment are kept
+ * without a duration.
+ *
+ * Queries are grouped into statements by fingerprint (the statement without its values), which is
+ * what N+1 detection, duplicate detection and the per-statement numbers are built on. Every
+ * collection is capped, and the serialized timeline stays within 512 KB, so a request that runs
+ * ten thousand queries costs a bounded amount of memory and bandwidth; the counters still count
+ * everything.
+ */
+
+import { createHash } from 'crypto';
+import { performance } from 'perf_hooks';
+import { RequestContext } from '../exceptions/request-context.js';
+import { Operation } from '../instrumentation/operation.js';
+import { LoopMonitor } from '../runtime/loop-monitor.js';
+import { Caller } from './caller.js';
+import { ExecutionHash } from './execution-hash.js';
+import { LruCache } from './lru.js';
+import { Masker } from './masking.js';
+import { SqlText } from './sql.js';
+
+export const TIMELINE_DEFAULTS = Object.freeze({
+  captureQueries: true,
+  captureOutgoing: true,
+  captureLogs: true,
+  queryParams: true,
+  slowQueryMs: 100,
+  nPlusOneThreshold: 5,
+  loopBlockMs: 100
+});
+
+// Spans listed per request, by type.
+const SPAN_CAPS = Object.freeze({ query: 500, http: 100, log: 200, measure: 100, exception: 10, block: 20 });
+const MAX_STATEMENTS = 200;
+// Executions per statement whose parameters are kept.
+const PARAMS_PER_STATEMENT = 20;
+const MAX_PARAMS = 50;
+const BUDGET_BYTES = 512 * 1024;
+const MAX_TEXT = 4096;
+const MAX_ERROR = 500;
+const MAX_LOG = 2000;
+const MAX_NAME = 200;
+const MAX_EXCEPTION_MESSAGE = 500;
+// Distinct executions remembered per statement for duplicate detection.
+const MAX_SEEN = 1000;
+
+// Shared by every timeline: statements repeat across requests.
+const sqlText = new SqlText();
+const shapeFingerprints = new LruCache(500);
+
+/**
+ * What one PeekInside instance records on its timelines.
+ */
+export class TimelineSettings {
+  /**
+   * @param {Object} options - PeekInside options (see TIMELINE_DEFAULTS)
+   * @param {string} root - The project folder, where callers are looked for
+   */
+  constructor(options, root) {
+    this.captureQueries = options.captureQueries !== false;
+    this.captureOutgoing = options.captureOutgoing !== false;
+    this.captureLogs = options.captureLogs !== false;
+    this.queryParams = options.queryParams !== false;
+    this.slowQueryMs = Number.isFinite(options.slowQueryMs) && options.slowQueryMs >= 0 ? options.slowQueryMs : TIMELINE_DEFAULTS.slowQueryMs;
+    this.nPlusOneThreshold = Number.isInteger(options.nPlusOneThreshold) && options.nPlusOneThreshold >= 2
+      ? options.nPlusOneThreshold
+      : TIMELINE_DEFAULTS.nPlusOneThreshold;
+    // 0 turns block detection off (loopBlockMs: false).
+    this.loopBlockMs = options.loopBlockMs === false || options.loopBlockMs === 0
+      ? 0
+      : Number.isFinite(options.loopBlockMs) && options.loopBlockMs >= 50 ? options.loopBlockMs : TIMELINE_DEFAULTS.loopBlockMs;
+    this.caller = new Caller(root);
+  }
+}
+
+/**
+ * One request's timeline.
+ */
+export class Timeline {
+  /**
+   * The open timeline of the request the calling code runs for.
+   *
+   * @returns {Timeline|null}
+   */
+  static current() {
+    const context = RequestContext.current();
+    const timeline = context && context.timeline;
+    return timeline && !timeline.closed ? timeline : null;
+  }
+
+  /**
+   * The fingerprint a SQL statement gets, for code that needs it before running the statement
+   * (caller hints).
+   *
+   * @param {string} text
+   * @param {string} system
+   * @returns {string|null} null when the text can't be analyzed
+   */
+  static sqlFingerprint(text, system) {
+    try {
+      return sqlText.analyze(text, system).fingerprint;
+    } catch (error) {
+      return null;
+    }
+  }
+
+  /**
+   * @param {TimelineSettings} settings
+   * @param {number} [origin=performance.now()] - performance.now() at the instant the request's
+   *   occurredAt was taken: span starts are offsets from it
+   */
+  constructor(settings, origin = performance.now()) {
+    this.settings = settings;
+    this.origin = origin;
+    this.closed = false;
+    this.spans = [];
+    this.listed = { query: 0, http: 0, log: 0, measure: 0, exception: 0, block: 0 };
+    this.dropped = 0;
+    this.statements = [];
+    this.byFingerprint = new Map();
+    // Statements past the cap: counted, not listed.
+    this.overflow = new Map();
+    this.stats = { queries: 0, queryTime: 0, slowQueries: 0, duplicates: 0, http: 0, httpTime: 0, logs: 0, loopBlock: 0 };
+    this._result = null;
+    if (settings.loopBlockMs > 0) LoopMonitor.track(this);
+  }
+
+  /**
+   * Whether statements need a caller hint: true when `fingerprint` hasn't run in this request yet.
+   *
+   * @param {string} fingerprint
+   * @returns {boolean}
+   */
+  needsCaller(fingerprint) {
+    return typeof fingerprint === 'string' && !this.byFingerprint.has(fingerprint) && this.statements.length < MAX_STATEMENTS;
+  }
+
+  /**
+   * Starts a query span.
+   *
+   * @param {Object} query
+   * @param {'postgresql'|'mysql'|'mongodb'|'sqlite'|'sqlserver'|'other'} query.system
+   * @param {'pg'|'mysql2'|'mongodb'|'prisma'} query.client
+   * @param {string} query.text - SQL as sent; for MongoDB and Prisma, the value-free shape
+   * @param {boolean} [query.sql=true] - Whether `text` is SQL (analyzed and masked) or a shape
+   * @param {Array|Object} [query.params] - The values sent, unmasked (they are masked here)
+   * @param {string} [query.operation] - The ORM call that ran it, such as "User.findMany"
+   * @param {() => (import('./caller.js').CallerInfo|null)} [query.caller] - Finds the caller when
+   *   the stack can't (Prisma)
+   * @param {number} [query.startedAt] - performance.now() when the query started, for a query
+   *   recorded once it is over; now by default
+   * @returns {SpanHandle|null} null when queries aren't recorded, or this one couldn't be (the
+   *   app's query runs all the same)
+   */
+  query(query) {
+    try {
+      return this._query(query);
+    } catch (error) {
+      return null;
+    }
+  }
+
+  _query({ system, client, text, sql = true, params, operation, caller, startedAt = performance.now() }) {
+    if (this.closed || !this.settings.captureQueries) return null;
+    const analysis = sql ? sqlText.analyze(text, system) : shapeAnalysis(text);
+    this.stats.queries++;
+
+    let statement = this.byFingerprint.get(analysis.fingerprint);
+    if (!statement && this.statements.length < MAX_STATEMENTS) {
+      statement = new Statement({
+        index: this.statements.length,
+        system,
+        client,
+        text: cut(analysis.maskedText, MAX_TEXT),
+        fingerprint: analysis.fingerprint,
+        caller: this._caller(analysis.fingerprint, caller)
+      });
+      this.statements.push(statement);
+      this.byFingerprint.set(analysis.fingerprint, statement);
+    }
+    const tracked = statement || this._overflow(analysis.fingerprint);
+    tracked.count++;
+    const execution = ExecutionHash.of(text, params);
+    if (tracked.seen.has(execution)) {
+      tracked.duplicates++;
+      this.stats.duplicates++;
+    } else if (tracked.seen.size < MAX_SEEN) {
+      tracked.seen.add(execution);
+    }
+
+    let span = null;
+    if (statement && this.listed.query < SPAN_CAPS.query) {
+      this.listed.query++;
+      span = { type: 'query', start: this._offset(startedAt), duration: null, statement: statement.index };
+      const shown = cut(analysis.maskedText, MAX_TEXT);
+      if (shown !== statement.text) span.text = shown;
+      if (this.settings.queryParams && params != null && statement.paramsKept < PARAMS_PER_STATEMENT) {
+        const masked = maskParams(params, analysis.sensitive);
+        if (masked !== undefined) {
+          statement.paramsKept++;
+          span.params = masked;
+        }
+      }
+      if (operation) span.operation = String(operation).slice(0, MAX_NAME);
+      this.spans.push(span);
+    } else {
+      this.dropped++;
+    }
+    return new SpanHandle(this, 'query', span, startedAt, tracked);
+  }
+
+  /**
+   * Starts an outgoing HTTP call span.
+   *
+   * @param {Object} call
+   * @param {'http'|'fetch'} call.client
+   * @param {string} call.method
+   * @param {string} call.url - Masked here (credentials, sensitive query values)
+   * @param {string} call.host
+   * @returns {SpanHandle|null} null when outgoing calls aren't recorded, or this one couldn't be
+   */
+  http(call) {
+    try {
+      return this._http(call);
+    } catch (error) {
+      return null;
+    }
+  }
+
+  _http({ client, method, url, host }) {
+    if (this.closed || !this.settings.captureOutgoing) return null;
+    const startedAt = performance.now();
+    let span = null;
+    if (this.listed.http < SPAN_CAPS.http) {
+      span = {
+        type: 'http',
+        start: this._offset(startedAt),
+        duration: null,
+        client,
+        method: String(method || 'GET').toUpperCase().slice(0, 16),
+        url: Masker.url(url),
+        host: String(host || '').toLowerCase().slice(0, 255),
+        status: null
+      };
+      this.listed.http++;
+      this.spans.push(span);
+    } else {
+      this.dropped++;
+    }
+    this.stats.http++;
+    return new SpanHandle(this, 'http', span, startedAt, null);
+  }
+
+  /**
+   * Records a log line.
+   *
+   * @param {'debug'|'info'|'warn'|'error'} level
+   * @param {string} message - Formatted as it was written
+   * @param {'console'|'pino'|'winston'} [source='console']
+   * @returns {void}
+   */
+  log(level, message, source = 'console') {
+    if (this.closed || !this.settings.captureLogs) return;
+    try {
+      const span = { type: 'log', start: this._offset(performance.now()), level, message: cut(String(message), MAX_LOG), source };
+      this.stats.logs++;
+      if (this.listed.log < SPAN_CAPS.log) {
+        this.listed.log++;
+        this.spans.push(span);
+      } else {
+        this.dropped++;
+      }
+    } catch (error) {
+      // A message that can't be read as text isn't recorded.
+    }
+  }
+
+  /**
+   * Records an exception recorded during the request.
+   *
+   * @param {{ id: string, summary: { name: string, message: string } }} record - An ExceptionRecord
+   * @returns {void}
+   */
+  exception(record) {
+    if (this.closed) return;
+    try {
+      if (this.listed.exception < SPAN_CAPS.exception) {
+        this.spans.push({
+          type: 'exception',
+          start: this._offset(performance.now()),
+          id: record.id,
+          name: record.summary.name,
+          message: cut(record.summary.message, MAX_EXCEPTION_MESSAGE)
+        });
+        this.listed.exception++;
+      } else {
+        this.dropped++;
+      }
+    } catch (error) {
+      // Not recorded on the timeline; the exception itself is.
+    }
+  }
+
+  /**
+   * Records that the event loop was blocked while the request was in flight (see LoopMonitor):
+   * the request waited, whichever code blocked it. Blocks shorter than the settings' loopBlockMs
+   * are left out.
+   *
+   * @param {number} start - performance.now() when the block began
+   * @param {number} duration - ms
+   * @returns {void}
+   */
+  block(start, duration) {
+    if (this.closed || !this.settings.loopBlockMs) return;
+    // A block that began before the request counts from the request's start.
+    const from = Math.max(start, this.origin);
+    const length = start + duration - from;
+    if (length < this.settings.loopBlockMs) return;
+    if (length > this.stats.loopBlock) this.stats.loopBlock = length;
+    if (this.listed.block < SPAN_CAPS.block) {
+      this.listed.block++;
+      this.spans.push({ type: 'block', start: this._offset(from), duration: round(length) });
+    } else {
+      this.dropped++;
+    }
+  }
+
+  /**
+   * Starts a measure span (PeekInside#measure()).
+   *
+   * @param {string} name
+   * @returns {SpanHandle|null} null once the timeline is closed, or for a name that can't be read
+   *   as text
+   */
+  measure(name) {
+    if (this.closed) return null;
+    const startedAt = performance.now();
+    let span = null;
+    try {
+      if (this.listed.measure < SPAN_CAPS.measure) {
+        span = { type: 'measure', start: this._offset(startedAt), duration: null, name: cut(String(name), MAX_NAME) };
+        this.listed.measure++;
+        this.spans.push(span);
+      } else {
+        this.dropped++;
+      }
+    } catch (error) {
+      return null;
+    }
+    return new SpanHandle(this, 'measure', span, startedAt, null);
+  }
+
+  /**
+   * Closes the timeline, for the request's record. Later spans are ignored; spans still open keep
+   * a null duration. Calling it again returns the same result.
+   *
+   * @returns {{ stats: Object, timeline: Object|undefined }} `timeline` only when something was
+   *   recorded
+   */
+  close() {
+    if (this._result) return this._result;
+    // The request may have blocked the loop itself and finished before the monitor could tell.
+    const pending = this.settings.loopBlockMs ? LoopMonitor.pending() : null;
+    if (pending) this.block(pending.start, pending.duration);
+    LoopMonitor.untrack(this);
+    this.closed = true;
+    const threshold = this.settings.nPlusOneThreshold;
+    let nPlusOne = 0;
+    for (const statement of this.statements) if (statement.count >= threshold) nPlusOne++;
+    for (const statement of this.overflow.values()) if (statement.count >= threshold) nPlusOne++;
+
+    const stats = {
+      queries: this.stats.queries,
+      queryTime: round(this.stats.queryTime),
+      slowQueries: this.stats.slowQueries,
+      nPlusOne,
+      duplicates: this.stats.duplicates,
+      http: this.stats.http,
+      httpTime: round(this.stats.httpTime),
+      logs: this.stats.logs,
+      loopBlock: round(this.stats.loopBlock)
+    };
+
+    // By start, then in the order they were created (a stable sort): spans recorded once they were
+    // over (a Prisma operation) start before spans created in the meantime.
+    const spans = this.spans.filter((span) => !span.cancelled).sort((a, b) => a.start - b.start);
+    let timeline;
+    if (spans.length > 0 || this.dropped > 0) {
+      timeline = fitBudget({
+        version: 1,
+        spans,
+        statements: this.statements.map((statement) => statement.toJSON(threshold)),
+        dropped: this.dropped,
+        slowQueryMs: this.settings.slowQueryMs,
+        nPlusOneThreshold: threshold
+      });
+    }
+    this._result = { stats, timeline };
+    return this._result;
+  }
+
+  // Milliseconds since the request started, 2 decimals.
+  _offset(at) {
+    return round(at - this.origin);
+  }
+
+  _caller(fingerprint, find) {
+    try {
+      if (typeof find === 'function') return find();
+      // A query an ORM call runs: the line that made the call, when the ORM captured it.
+      const found = Operation.caller();
+      if (found) return found;
+      const hint = Caller.hinted(fingerprint);
+      return hint ? hint.caller : this.settings.caller.capture();
+    } catch (error) {
+      return null;
+    }
+  }
+
+  _overflow(fingerprint) {
+    let entry = this.overflow.get(fingerprint);
+    if (!entry) {
+      entry = { count: 0, duplicates: 0, time: 0, maxTime: 0, slow: 0, seen: new Set() };
+      this.overflow.set(fingerprint, entry);
+    }
+    return entry;
+  }
+}
+
+/**
+ * A span in progress. Ending it twice, or after the timeline closed, does nothing.
+ */
+export class SpanHandle {
+  /**
+   * @param {Timeline} timeline
+   * @param {'query'|'http'|'measure'} type
+   * @param {Object|null} span - The listed span, or null when a cap left it out
+   * @param {number} startedAt - performance.now() at the start
+   * @param {Object|null} statement - For queries: the statement it counts towards
+   */
+  constructor(timeline, type, span, startedAt, statement) {
+    this.timeline = timeline;
+    this.type = type;
+    this.span = span;
+    this.startedAt = startedAt;
+    this.statement = statement;
+    this.done = false;
+  }
+
+  /**
+   * Ends the span.
+   *
+   * @param {Object} [result]
+   * @param {number|null} [result.rows] - Queries: rows returned or affected
+   * @param {number|null} [result.status] - HTTP: the response status
+   * @param {*} [result.error] - What made it fail (an Error, or a message)
+   * @returns {void}
+   */
+  end(result) {
+    if (this.done) return;
+    this.done = true;
+    try {
+      this._end(result || {});
+    } catch (error) {
+      // Called from the driver's own completion path: whatever happens here stays here.
+    }
+  }
+
+  _end({ rows, status, error }) {
+    const timeline = this.timeline;
+    if (timeline.closed) return;
+    const duration = performance.now() - this.startedAt;
+    const span = this.span;
+    const failure = error === undefined || error === null ? null : errorText(error);
+
+    if (this.type === 'query') {
+      const statement = this.statement;
+      timeline.stats.queryTime += duration;
+      statement.time += duration;
+      statement.maxTime = Math.max(statement.maxTime, duration);
+      if (duration >= timeline.settings.slowQueryMs) {
+        statement.slow++;
+        timeline.stats.slowQueries++;
+      }
+      if (span && Number.isInteger(rows) && rows >= 0) span.rows = rows;
+    } else if (this.type === 'http') {
+      timeline.stats.httpTime += duration;
+      if (span && Number.isInteger(status)) span.status = status;
+    }
+    if (span) {
+      span.duration = round(duration);
+      if (failure) span.error = failure;
+    }
+  }
+
+  /**
+   * Takes the span back: the operation never ran (its call threw synchronously).
+   *
+   * @returns {void}
+   */
+  cancel() {
+    if (this.done) return;
+    this.done = true;
+    const timeline = this.timeline;
+    if (timeline.closed) return;
+    if (this.type === 'query') {
+      timeline.stats.queries--;
+      this.statement.count--;
+    } else if (this.type === 'http') {
+      timeline.stats.http--;
+    }
+    if (this.span) this.span.cancelled = true;
+  }
+}
+
+/**
+ * One distinct statement of a request.
+ */
+class Statement {
+  /**
+   * @param {Object} fields - index, system, client, text, fingerprint, caller
+   */
+  constructor({ index, system, client, text, fingerprint, caller }) {
+    this.index = index;
+    this.system = system;
+    this.client = client;
+    this.text = text;
+    this.fingerprint = fingerprint;
+    this.caller = caller || null;
+    this.count = 0;
+    this.time = 0;
+    this.maxTime = 0;
+    this.slow = 0;
+    this.duplicates = 0;
+    this.paramsKept = 0;
+    this.seen = new Set();
+  }
+
+  /**
+   * @param {number} threshold - nPlusOneThreshold
+   * @returns {Object} The statement as the contract lists it
+   */
+  toJSON(threshold) {
+    return {
+      system: this.system,
+      client: this.client,
+      text: this.text,
+      fingerprint: this.fingerprint,
+      count: this.count,
+      time: round(this.time),
+      maxTime: round(this.maxTime),
+      slow: this.slow,
+      nPlusOne: this.count >= threshold,
+      duplicates: this.duplicates,
+      caller: this.caller
+    };
+  }
+}
+
+// Shapes are value-free already: the fingerprint is the hash of the text itself.
+function shapeAnalysis(text) {
+  const shape = String(text);
+  let fingerprint = shapeFingerprints.get(shape);
+  if (!fingerprint) {
+    fingerprint = createHash('sha256').update(shape).digest('hex').slice(0, 16);
+    shapeFingerprints.set(shape, fingerprint);
+  }
+  return { fingerprint, maskedText: shape, sensitive: new Set() };
+}
+
+// A query span's parameters, masked; undefined when they can't be read (the span goes without).
+function maskParams(params, sensitive) {
+  try {
+    if (Array.isArray(params)) return Masker.sqlParams(params, sensitive, MAX_PARAMS);
+    return typeof params === 'object' ? Masker.fields(params, MAX_PARAMS) : null;
+  } catch (error) {
+    return undefined;
+  }
+}
+
+// Keeps the serialized timeline within the budget: first the parameters of the newest query spans
+// go, then the newest spans themselves (counted in `dropped`), and as a last resort the statements'
+// texts are shortened.
+function fitBudget(timeline) {
+  let size = Buffer.byteLength(JSON.stringify(timeline));
+  if (size <= BUDGET_BYTES) return timeline;
+
+  const spans = timeline.spans.map((span) => ({ ...span }));
+  const lengths = spans.map((span) => Buffer.byteLength(JSON.stringify(span)) + 1);
+  for (let i = spans.length - 1; i >= 0 && size > BUDGET_BYTES; i--) {
+    if (spans[i].type !== 'query' || spans[i].params === undefined) continue;
+    delete spans[i].params;
+    const length = Buffer.byteLength(JSON.stringify(spans[i])) + 1;
+    size -= lengths[i] - length;
+    lengths[i] = length;
+  }
+  let dropped = timeline.dropped;
+  while (size > BUDGET_BYTES && spans.length > 0) {
+    size -= lengths.pop();
+    spans.pop();
+    dropped++;
+  }
+  let result = { ...timeline, spans, dropped };
+  for (const max of [1024, 256, 64]) {
+    if (Buffer.byteLength(JSON.stringify(result)) <= BUDGET_BYTES) break;
+    result = { ...result, statements: result.statements.map((statement) => ({ ...statement, text: cut(statement.text, max) })) };
+  }
+  return result;
+}
+
+// "<code>: <message>" or "<message>", at most 500 characters.
+function errorText(error) {
+  if (typeof error === 'string') return cut(error, MAX_ERROR);
+  let message;
+  let code;
+  try {
+    message = typeof error.message === 'string' ? error.message : String(error);
+    code = error.code;
+  } catch (failure) {
+    message = 'Error';
+  }
+  // Some libraries (Prisma) format messages over several lines with blank lines between.
+  const tidy = String(message).trim().replace(/\n\s*\n+/g, '\n');
+  const text = (typeof code === 'string' || typeof code === 'number') && code !== '' ? `${code}: ${tidy}` : tidy;
+  return cut(text, MAX_ERROR);
+}
+
+function cut(text, max) {
+  if (typeof text !== 'string') return text;
+  return text.length > max ? `${text.slice(0, max - 1)}…` : text;
+}
+
+function round(value) {
+  return Math.round(value * 100) / 100;
+}