feat: regex sampler

This fixes Redocly#152 except in an edge case where both `pattern` and
`maxLength` are used, and the sampler skips over the valid length range.

It introduces a tiny dependency (<10 KB uncompressed) which doesn't
really have any prominent competitors.

package.json

@@ -88,6 +88,7 @@
   },
   "dependencies": {
     "@types/json-schema": "^7.0.7",
-    "json-pointer": "0.6.2"
+    "json-pointer": "0.6.2",
+    "randexp": "^0.5.3"
   }
 }

src/samplers/string.js

@@ -1,8 +1,51 @@
 'use strict';
 
+import RandExp from 'randexp';
+
 import { ensureMinLength, toRFCDateTime, uuid } from '../utils';
 
 const passwordSymbols = 'qwerty!@#$%^123456';
+const MAX_REGEX_SAMPLES = 100;
+
+function sampleRegex(pattern, placeholder, min, max) {
+  if (!pattern) return placeholder;
+
+  let res;
+  let i = 0;
+
+  do {
+    RandExp.prototype.randInt = (from, to) => Math.min(from + i, to);
+    res = new RandExp(pattern).gen();
+    i++;
+  } while (res.length < min && i < MAX_REGEX_SAMPLES);
+
+  // Handle case where we skipped over the range.
+  // Example: /\d*\d*foo/, will sample foo, 11foo, 2222foo, etc.
+  // HACK: RandExp doesn't expose an API to set the value of a specific sample,
+  //       so we'll just fuzz it. If the number of * and + is more than 100 times
+  //       the size of the range, it may fail.
+  if (max && max >= min && res.length > max) {
+    for (let j = 0; j < MAX_REGEX_SAMPLES; j++) {
+      RandExp.prototype.randInt = (from, to) => Math.max(
+        from,
+        Math.min(from + i - 2 + Math.floor(Math.random() * 2), to),
+      );
+      const candidate = new RandExp(pattern).gen();
+      console.log(candidate);
+      if (candidate.length <= max && candidate.length >= min) return candidate;
+    }
+  }
+
+  return res;
+}
+
+function truncateString(str, min, max) {
+  let res = ensureMinLength(str, min);
+  if (max && res.length > max) {
+    res = res.substring(0, max);
+  }
+  return res;
+}
 
 function emailSample() {
   return '[email protected]';
@@ -42,12 +85,9 @@ function timeSample(min, max) {
   return commonDateTimeSample({ min, max, omitTime: false, omitDate: true }).slice(1);
 }
 
-function defaultSample(min, max) {
-  let res = ensureMinLength('string', min);
-  if (max && res.length > max) {
-    res = res.substring(0, max);
-  }
-  return res;
+function defaultSample(min, max, _propertyName, pattern) {
+  const res = sampleRegex(pattern, 'string', min, max);
+  return truncateString(res, min, max)
 }
 
 function ipv4Sample() {
@@ -96,8 +136,9 @@ function relativeJsonPointerSample() {
   return '1/relative/json/pointer';
 }
 
-function regexSample() {
-  return '/regex/';
+function regexSample(min, max, _propertyName, pattern) {
+  const res = sampleRegex(pattern, '/regex/', min, max);
+  return truncateString(res, min, max);
 }
 
 const stringFormats = {
@@ -127,5 +168,10 @@ export function sampleString(schema, options, spec, context) {
   let format = schema.format || 'default';
   let sampler = stringFormats[format] || defaultSample;
   let propertyName = context && context.propertyName;
-  return sampler(schema.minLength | 0, schema.maxLength, propertyName);
+  return sampler(
+    schema.minLength || 0,
+    schema.maxLength,
+    propertyName,
+    schema.pattern,
+  );
 }

test/unit/string.spec.js

@@ -130,6 +130,17 @@ describe('sampleString', () => {
     expect(res).to.equal('fb4274c7-4fcd-4035-8958-a680548957ff');
   });
 
+  it('should return valid string for regex with min and max', () => {
+    const schema = {
+      format: 'regex',
+      pattern: 'foo-\\d+\\d+\\d+\\d+\\d+-bar',
+      minLength: 19,
+      maxLength: 19,
+    };
+    res = sampleString(schema, null, null, {propertyName: 'fooId'});
+    expect(res).to.match(/foo-\d{11}-bar/);
+  });
+
   it.each([
     'email',
     // 'idn-email', // unsupported by ajv-formats