@ogcio/o11y-sdk-node 0.3.1 → 0.4.1

package/test/internals/redactors/ip.test.ts

@@ -0,0 +1,93 @@
+import {
+  describe,
+  expect,
+  it,
+  vi,
+  beforeEach,
+  beforeAll,
+  afterAll,
+} from "vitest";
+
+import * as sharedMetrics from "../../../lib/internals/shared-metrics.js";
+import { ipRedactor } from "../../../lib/internals/redaction/redactors/ip";
+
+describe("IP Redaction utils", () => {
+  describe("tracks metrics", () => {
+    const mockMetricAdd = vi.fn();
+
+    beforeEach(() => {
+      vi.restoreAllMocks();
+      vi.spyOn(sharedMetrics, "_getPIICounterRedactionMetric").mockReturnValue({
+        add: mockMetricAdd,
+      });
+    });
+
+    it("redacts plain PII and tracks redaction with metric", () => {
+      const input = "255.255.255.255";
+      const output = ipRedactor(input, "log", "string");
+
+      expect(output).toBe("[REDACTED IPV4]");
+      expect(mockMetricAdd).toHaveBeenCalledWith(
+        1,
+        expect.objectContaining({
+          pii_format: "string",
+          pii_type: "IPv4",
+          redaction_source: "log",
+        }),
+      );
+    });
+
+    it("handles strings without PII unchanged", () => {
+      const input = "hello world";
+      const output = ipRedactor(input, "log", "string");
+
+      expect(output).toBe("hello world");
+      expect(mockMetricAdd).not.toHaveBeenCalled();
+    });
+  });
+
+  describe("Redacts IPv4 and IPv6", () => {
+    beforeAll(() => {
+      vi.spyOn(sharedMetrics, "_getPIICounterRedactionMetric").mockReturnValue({
+        add: vi.fn(),
+      });
+    });
+
+    afterAll(() => {
+      vi.restoreAllMocks();
+    });
+
+    it.each`
+      value                                           | expectedRedactedValue
+      ${"hello world"}                                | ${"hello world"}
+      ${"hello 127.0.0.1"}                            | ${"hello [REDACTED IPV4]"}
+      ${"127.0.0.1, hello!"}                          | ${"[REDACTED IPV4], hello!"}
+      ${"127.0.0.1,127.0.0.1"}                        | ${"[REDACTED IPV4],[REDACTED IPV4]"}
+      ${"127.0.0.1127.0.0.1"}                         | ${"127.0.0.1127.0.0.1"}
+      ${"256.1.1.1"}                                  | ${"256.1.1.1"}
+      ${"0.0.0.0!"}                                   | ${"[REDACTED IPV4]!"}
+      ${"text0.0.0.0"}                                | ${"text[REDACTED IPV4]"}
+      ${"%A00.0.0.0"}                                 | ${"[REDACTED IPV4]"}
+      ${"0.0.0.0%20"}                                 | ${"[REDACTED IPV4]"}
+      ${"0.0.text0.0"}                                | ${"0.0.text0.0"}
+      ${"2001:0db8::1"}                               | ${"[REDACTED IPV6]"}
+      ${"::1"}                                        | ${"[REDACTED IPV6]"}
+      ${"%20::1"}                                     | ${"[REDACTED IPV6]"}
+      ${"::1%A0"}                                     | ${"[REDACTED IPV6]"}
+      ${"text::1"}                                    | ${"text[REDACTED IPV6]"}
+      ${"::1text"}                                    | ${"[REDACTED IPV6]text"}
+      ${"sentence ending with f::1"}                  | ${"sentence ending with [REDACTED IPV6]"}
+      ${"sentence ending with :::1"}                  | ${"sentence ending with :::1"}
+      ${"2001:0DB8:85A3:0000:0000:8A2E:0370:7334::1"} | ${"2001:0DB8:85A3:0000:0000:8A2E:0370:7334::1"}
+      ${"2001:0DB8:85A3:0000:text:8A2E:0370:7334"}    | ${"2001:0DB8:85A3:0000:text:8A2E:0370:7334"}
+      ${"2001:0db8::12001:0db8::1"}                   | ${"2001:0db8::12001:0db8::1"}
+      ${"2001:0db8::1,2001:0db8::1"}                  | ${"[REDACTED IPV6],[REDACTED IPV6]"}
+    `(
+      "returns $expectedRedactedValue for value '$value'",
+      async ({ value, expectedRedactedValue }: Record<string, string>) => {
+        const result = ipRedactor(value, "log", "string");
+        expect(result).toBe(expectedRedactedValue);
+      },
+    );
+  });
+});

package/test/traces/active-span.test.ts

@@ -1,5 +1,5 @@
 import { beforeAll, beforeEach, describe, expect, it, vi } from "vitest";
-import * as piiDetection from "../../lib/internals/pii-detection.js";
+import * as piiDetection from "../../lib/internals/redaction/pii-detection.js";
 import { getActiveSpan } from "../../lib/traces.js";
 import { MockSpan } from "../utils/mock-signals.js";
 import { setNodeSdkConfig } from "../../lib/config-manager.js";

package/vitest.config.ts

@@ -27,10 +27,10 @@ export default defineConfig({
         test: {
           include: [
             "**/test/*.test.ts",
-            "**/test/processor/*.test.ts",
-            "**/test/traces/*.test.ts",
-            "**/test/internals/*.test.ts",
-            "**/test/exporter/*.test.ts",
+            "**/test/processor/**/*.test.ts",
+            "**/test/traces/**/*.test.ts",
+            "**/test/internals/**/*.test.ts",
+            "**/test/exporter/**/*.test.ts",
           ],
           name: "unit",
         },

package/dist/lib/internals/pii-detection.d.ts

@@ -1,17 +0,0 @@
-import type { AnyValue } from "@opentelemetry/api-logs";
-export type PIISource = "trace" | "log" | "metric";
-/**
- * Cleans a string by redacting email addresses and emitting metrics for PII.
- *
- * If the string is URL-encoded, it will be decoded before redaction.
- * Metrics are emitted for each domain found in redacted email addresses.
- *
- * @param {string} value - The input string to sanitize.
- * @param {"trace" | "log"} source - The source context of the input, used in metrics.
- * @returns {string} The cleaned string with any email addresses replaced by `[REDACTED EMAIL]`.
- */
-export declare function _cleanStringPII(value: AnyValue, source: PIISource): AnyValue;
-export declare function _cleanObjectPII(entry: object, source: PIISource): {
-    [k: string]: AnyValue;
-};
-export declare function _cleanLogBodyPII(value: AnyValue): AnyValue;

package/dist/lib/internals/pii-detection.js

@@ -1,116 +0,0 @@
-import { _getPIICounterRedactionMetric } from "./shared-metrics.js";
-const EMAIL_REGEX = /[a-zA-Z0-9._%+-]+@([a-zA-Z0-9.-]+\.[a-z]{2,})/gi;
-const decoder = new TextDecoder();
-const encoder = new TextEncoder();
-/**
- * Redacts all email addresses in the input string and collects metadata.
- *
- * @param {string} value The input string potentially containing email addresses.
- * @returns {{
- *   redacted: string,
- *   count: number,
- *   domains: Record<string, number>
- * }}
- *
- * An object containing:
- *   - `redacted`: the string with email addresses replaced by `[REDACTED EMAIL]`
- *   - `count`: total number of email addresses redacted
- *   - `domains`: a map of domain names to the number of times they were redacted
- */
-function _redactEmails(value) {
-    let count = 0;
-    const domains = {};
-    const redacted = value.replace(EMAIL_REGEX, (_, domain) => {
-        count++;
-        domains[domain] = (domains[domain] || 0) + 1;
-        return "[REDACTED EMAIL]";
-    });
-    return { redacted, count, domains };
-}
-/**
- * Checks whether a string contains URI-encoded components.
- *
- * @param {string} value - The string to inspect.
- * @returns {boolean} `true` if the string is encoded, `false` otherwise.
- */
-function _containsEncodedComponents(value) {
-    try {
-        return decodeURI(value) !== decodeURIComponent(value);
-    }
-    catch {
-        return false;
-    }
-}
-/**
- * Cleans a string by redacting email addresses and emitting metrics for PII.
- *
- * If the string is URL-encoded, it will be decoded before redaction.
- * Metrics are emitted for each domain found in redacted email addresses.
- *
- * @param {string} value - The input string to sanitize.
- * @param {"trace" | "log"} source - The source context of the input, used in metrics.
- * @returns {string} The cleaned string with any email addresses replaced by `[REDACTED EMAIL]`.
- */
-export function _cleanStringPII(value, source) {
-    if (Array.isArray(value)) {
-        return value.map((v) => _cleanStringPII(v, source));
-    }
-    if (typeof value !== "string") {
-        return value;
-    }
-    let kind = "string";
-    let decodedValue = value;
-    if (_containsEncodedComponents(value)) {
-        decodedValue = decodeURIComponent(value);
-        kind = "url";
-    }
-    const { redacted, count, domains } = _redactEmails(decodedValue);
-    if (count > 0) {
-        for (const [domain, domainCount] of Object.entries(domains)) {
-            _getPIICounterRedactionMetric().add(domainCount, {
-                pii_type: "email",
-                redaction_source: source,
-                pii_email_domain: domain,
-                pii_format: kind,
-            });
-        }
-    }
-    return redacted;
-}
-export function _cleanObjectPII(entry, source) {
-    if (!entry) {
-        return entry;
-    }
-    return Object.fromEntries(Object.entries(entry).map(([k, v]) => [k, _cleanStringPII(v, source)]));
-}
-export function _cleanLogBodyPII(value) {
-    if (typeof value === "string") {
-        return _cleanStringPII(value, "log");
-    }
-    if (typeof value === "number" ||
-        typeof value === "boolean" ||
-        value == null) {
-        return value;
-    }
-    if (value instanceof Uint8Array) {
-        try {
-            const decoded = decoder.decode(value);
-            const sanitized = _cleanStringPII(decoded, "log");
-            return encoder.encode(sanitized);
-        }
-        catch {
-            return value;
-        }
-    }
-    if (Array.isArray(value)) {
-        return value.map(_cleanLogBodyPII);
-    }
-    if (typeof value === "object") {
-        const sanitized = {};
-        for (const [key, val] of Object.entries(value)) {
-            sanitized[key] = _cleanLogBodyPII(val);
-        }
-        return sanitized;
-    }
-    return value;
-}

package/lib/internals/pii-detection.ts

@@ -1,145 +0,0 @@
-import type { AnyValue, AnyValueMap } from "@opentelemetry/api-logs";
-import { _getPIICounterRedactionMetric } from "./shared-metrics.js";
-
-const EMAIL_REGEX = /[a-zA-Z0-9._%+-]+@([a-zA-Z0-9.-]+\.[a-z]{2,})/gi;
-
-const decoder = new TextDecoder();
-const encoder = new TextEncoder();
-
-export type PIISource = "trace" | "log" | "metric";
-
-/**
- * Redacts all email addresses in the input string and collects metadata.
- *
- * @param {string} value The input string potentially containing email addresses.
- * @returns {{
- *   redacted: string,
- *   count: number,
- *   domains: Record<string, number>
- * }}
- *
- * An object containing:
- *   - `redacted`: the string with email addresses replaced by `[REDACTED EMAIL]`
- *   - `count`: total number of email addresses redacted
- *   - `domains`: a map of domain names to the number of times they were redacted
- */
-function _redactEmails(value: string): {
-  redacted: string;
-  count: number;
-  domains: Record<string, number>;
-} {
-  let count = 0;
-  const domains: Record<string, number> = {};
-
-  const redacted = value.replace(EMAIL_REGEX, (_, domain) => {
-    count++;
-    domains[domain] = (domains[domain] || 0) + 1;
-    return "[REDACTED EMAIL]";
-  });
-
-  return { redacted, count, domains };
-}
-
-/**
- * Checks whether a string contains URI-encoded components.
- *
- * @param {string} value - The string to inspect.
- * @returns {boolean} `true` if the string is encoded, `false` otherwise.
- */
-function _containsEncodedComponents(value: string) {
-  try {
-    return decodeURI(value) !== decodeURIComponent(value);
-  } catch {
-    return false;
-  }
-}
-
-/**
- * Cleans a string by redacting email addresses and emitting metrics for PII.
- *
- * If the string is URL-encoded, it will be decoded before redaction.
- * Metrics are emitted for each domain found in redacted email addresses.
- *
- * @param {string} value - The input string to sanitize.
- * @param {"trace" | "log"} source - The source context of the input, used in metrics.
- * @returns {string} The cleaned string with any email addresses replaced by `[REDACTED EMAIL]`.
- */
-export function _cleanStringPII(value: AnyValue, source: PIISource): AnyValue {
-  if (Array.isArray(value)) {
-    return value.map((v) => _cleanStringPII(v, source));
-  }
-
-  if (typeof value !== "string") {
-    return value;
-  }
-
-  let kind: "string" | "url" = "string";
-  let decodedValue = value;
-
-  if (_containsEncodedComponents(value)) {
-    decodedValue = decodeURIComponent(value);
-    kind = "url";
-  }
-
-  const { redacted, count, domains } = _redactEmails(decodedValue);
-
-  if (count > 0) {
-    for (const [domain, domainCount] of Object.entries(domains)) {
-      _getPIICounterRedactionMetric().add(domainCount, {
-        pii_type: "email",
-        redaction_source: source,
-        pii_email_domain: domain,
-        pii_format: kind,
-      });
-    }
-  }
-  return redacted;
-}
-
-export function _cleanObjectPII(entry: object, source: PIISource) {
-  if (!entry) {
-    return entry;
-  }
-
-  return Object.fromEntries(
-    Object.entries(entry).map(([k, v]) => [k, _cleanStringPII(v, source)]),
-  );
-}
-
-export function _cleanLogBodyPII(value: AnyValue): AnyValue {
-  if (typeof value === "string") {
-    return _cleanStringPII(value, "log");
-  }
-
-  if (
-    typeof value === "number" ||
-    typeof value === "boolean" ||
-    value == null
-  ) {
-    return value;
-  }
-
-  if (value instanceof Uint8Array) {
-    try {
-      const decoded = decoder.decode(value);
-      const sanitized = _cleanStringPII(decoded, "log") as string;
-      return encoder.encode(sanitized);
-    } catch {
-      return value;
-    }
-  }
-
-  if (Array.isArray(value)) {
-    return value.map(_cleanLogBodyPII);
-  }
-
-  if (typeof value === "object") {
-    const sanitized: AnyValueMap = {};
-    for (const [key, val] of Object.entries(value)) {
-      sanitized[key] = _cleanLogBodyPII(val);
-    }
-    return sanitized;
-  }
-
-  return value;
-}