enum inline fixes (#190)

Author: eviltester

packages/core/js/data_generation/inline-schema-rule.js

@@ -0,0 +1,33 @@
+function startsConstraint(trimmedLine) {
+  return /^IF\s+(?:\[|\(|NOT\b)/i.test(trimmedLine);
+}
+
+function looksLikeInlineRuleSpec(ruleText) {
+  const trimmed = String(ruleText ?? '').trim();
+  if (trimmed.length === 0 || startsConstraint(trimmed)) {
+    return false;
+  }
+
+  if (
+    /^(?:enum|literal|regex|datatype\.(?:enum|literal|regex)|awd\.datatype\.(?:enum|literal|regex))\s*\(/i.test(trimmed)
+  ) {
+    return true;
+  }
+
+  if (/^(?:faker\.)?[A-Za-z][A-Za-z0-9_]*(?:\.[A-Za-z][A-Za-z0-9_]*)+(?:\s*\(.*\)\s*|\s*)$/i.test(trimmed)) {
+    return true;
+  }
+
+  if (!trimmed.includes(',')) {
+    return false;
+  }
+
+  const values = trimmed.split(',').map((value) => value.trim());
+  if (values.length < 2 || values.some((value) => value.length === 0 || value.length > 50)) {
+    return false;
+  }
+
+  return !values.some((value) => /[[\]{}()^$*+?|\\]/.test(value) || (value.includes('.') && /[A-Z]/.test(value)));
+}
+
+export { startsConstraint, looksLikeInlineRuleSpec };

packages/core/js/data_generation/rulesParser.js

@@ -1,38 +1,7 @@
 import { TestDataRules } from './testDataRules.js';
 import { SchemaParsingErrors } from './schema-parsing-errors.js';
 import { parseConstraintText } from './schema-constraint-parser.js';
-
-function startsConstraint(trimmedLine) {
-  return /^IF\s+(?:\[|\(|NOT\b)/i.test(trimmedLine);
-}
-
-function looksLikeInlineRuleSpec(ruleText) {
-  const trimmed = String(ruleText ?? '').trim();
-  if (trimmed.length === 0 || startsConstraint(trimmed)) {
-    return false;
-  }
-
-  if (
-    /^(?:enum|literal|regex|datatype\.(?:enum|literal|regex)|awd\.datatype\.(?:enum|literal|regex))\s*\(/i.test(trimmed)
-  ) {
-    return true;
-  }
-
-  if (/^(?:faker\.)?[A-Za-z][A-Za-z0-9_]*(?:\.[A-Za-z][A-Za-z0-9_]*)+(?:\s*\(.*\)\s*|\s*)$/i.test(trimmed)) {
-    return true;
-  }
-
-  if (!trimmed.includes(',')) {
-    return false;
-  }
-
-  const values = trimmed.split(',').map((value) => value.trim());
-  if (values.length < 2 || values.some((value) => value.length === 0 || value.length > 50)) {
-    return false;
-  }
-
-  return !values.some((value) => /[[\]{}()^$*+?|\\]/.test(value) || (value.includes('.') && /[A-Z]/.test(value)));
-}
+import { looksLikeInlineRuleSpec, startsConstraint } from './inline-schema-rule.js';
 
 function parseInlineRuleDefinition(line) {
   const source = String(line ?? '');
@@ -42,15 +11,18 @@ function parseInlineRuleDefinition(line) {
     }
 
     const name = source.slice(0, index).trim();
-    const rule = source.slice(index + 1).trim();
+    const rawRulePortion = source.slice(index + 1);
+    const rule = rawRulePortion.trim();
     if (name.length === 0 || !looksLikeInlineRuleSpec(rule)) {
       continue;
     }
 
+    const leadingWhitespaceLength = rawRulePortion.length - rawRulePortion.trimStart().length;
+
     return {
       name,
       rule,
-      separator: ': ',
+      separator: `:${rawRulePortion.slice(0, leadingWhitespaceLength)}`,
     };
   }
 

packages/core/src/index.js

@@ -11,6 +11,7 @@ import {
   CombinationAlgorithm,
   CombinationsTestDataGenerator,
 } from '../js/data_generation/n-wise/combinationsTestDataGenerator.js';
+import { looksLikeInlineRuleSpec, startsConstraint } from '../js/data_generation/inline-schema-rule.js';
 import { parseSchemaText } from '../js/data_generation/schema-conversion.js';
 import { hasSafeFakerLiteralArguments } from '../js/data_generation/faker/safeLiteralArgumentParser.js';
 import {
@@ -78,34 +79,6 @@ const SUPPORTED_FORMATS = [
   'asciitable',
 ];
 
-function looksLikeInlineSchemaRule(ruleText) {
-  const trimmed = String(ruleText ?? '').trim();
-  if (trimmed.length === 0 || /^IF\s+(?:\[|\(|NOT\b)/i.test(trimmed)) {
-    return false;
-  }
-
-  if (
-    /^(?:enum|literal|regex|datatype\.(?:enum|literal|regex)|awd\.datatype\.(?:enum|literal|regex))\s*\(/i.test(trimmed)
-  ) {
-    return true;
-  }
-
-  if (/^(?:faker\.)?[A-Za-z][A-Za-z0-9_]*(?:\.[A-Za-z][A-Za-z0-9_]*)+(?:\s*\(.*\)\s*|\s*)$/i.test(trimmed)) {
-    return true;
-  }
-
-  if (!trimmed.includes(',')) {
-    return false;
-  }
-
-  const values = trimmed.split(',').map((value) => value.trim());
-  if (values.length < 2 || values.some((value) => value.length === 0 || value.length > 50)) {
-    return false;
-  }
-
-  return !values.some((value) => /[[\]{}()^$*+?|\\]/.test(value) || (value.includes('.') && /[A-Z]/.test(value)));
-}
-
 function extractRuleLines(textSpec) {
   if (typeof textSpec !== 'string') {
     return [];
@@ -115,7 +88,7 @@ function extractRuleLines(textSpec) {
   let pendingName = null;
   for (const line of lines) {
     const trimmed = line.trim();
-    if (trimmed.length === 0 || /^\s*#/.test(line) || /^IF\s+(?:\[|\(|NOT\b)/i.test(trimmed)) {
+    if (trimmed.length === 0 || /^\s*#/.test(line) || startsConstraint(trimmed)) {
       pendingName = null;
       continue;
     }
@@ -126,7 +99,7 @@ function extractRuleLines(textSpec) {
         continue;
       }
       const rule = line.slice(separatorIndex + 1).trim();
-      if (looksLikeInlineSchemaRule(rule)) {
+      if (looksLikeInlineRuleSpec(rule)) {
         ruleLines.push(rule);
         pendingName = null;
         matchedInlineRule = true;

packages/core/src/tests/data_generation/unit/rulesParser.test.js

@@ -140,6 +140,20 @@ Status: person.jobTitle`;
     expect(parser.renderSpecFromRulesWithTokens(parser.testDataRules.rules)).toBe(inputText);
   });
 
+  test('preserves authored inline separator spacing when rebuilding from parsed tokens', () => {
+    const inputText = `Name:person.fullName
+Role:   enum(admin,user)`;
+
+    const parser = new RulesParser(faker, RandExp);
+    parser.parseText(inputText);
+
+    expect(parser.getSchemaTokens()).toEqual([
+      expect.objectContaining({ kind: 'rule', separator: ':' }),
+      expect.objectContaining({ kind: 'rule', separator: ':   ' }),
+    ]);
+    expect(parser.renderSpecFromRulesWithTokens(parser.testDataRules.rules)).toBe(inputText);
+  });
+
   test('preserves comments and blank lines when rebuilding from rule comments', () => {
     const inputText = `# one