From bd8b8e2772a39744c8ced370107642904153e9b2 Mon Sep 17 00:00:00 2001
From: Divyansh Singh <40380293+brc-dd@users.noreply.github.com>
Date: Thu, 27 Feb 2025 22:30:55 +0530
Subject: [PATCH] handle nested regions

---
 .../node/markdown/plugins/snippet.test.ts     | 297 +++++++++++-------
 src/node/markdown/plugins/snippet.ts          |  76 +++--
 2 files changed, 238 insertions(+), 135 deletions(-)

diff --git a/__tests__/unit/node/markdown/plugins/snippet.test.ts b/__tests__/unit/node/markdown/plugins/snippet.test.ts
index 0781b373..aa940784 100644
--- a/__tests__/unit/node/markdown/plugins/snippet.test.ts
+++ b/__tests__/unit/node/markdown/plugins/snippet.test.ts
@@ -99,145 +99,228 @@ describe('node/markdown/plugins/snippet', () => {
     })
   })
 
-  test('rawPathToToken', () => {
-    rawPathTokenMap.forEach(([rawPath, token]) => {
+  describe('rawPathToToken', () => {
+    test.each(rawPathTokenMap)('%s', (rawPath, token) => {
       expect(removeEmptyKeys(rawPathToToken(rawPath))).toEqual(token)
     })
   })
 
   describe('findRegion', () => {
-    test('when c# region with matching tag', () => {
-      const lines = `Console.WriteLine("Before region");
-#region hello
-Console.WriteLine("Hello, World!");
-#endregion hello
-Console.WriteLine("After region");`.split('\n')
-      const result = findRegion(lines, 'hello')
+    it('returns null when no region markers are present', () => {
+      const lines = ['function foo() {', '  console.log("hello");', '}']
+      expect(findRegion(lines, 'foo')).toBeNull()
+    })
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        'Console.WriteLine("Hello, World!");'
-      )
+    it('ignores non-matching region names', () => {
+      const lines = [
+        '// #region regionA',
+        'some code here',
+        '// #endregion regionA'
+      ]
+      expect(findRegion(lines, 'regionC')).toBeNull()
     })
-    test('when c# region is not indented with spaces and no matching tag', () => {
-      const lines = `Console.WriteLine("Before region");
-#region hello
-Console.WriteLine("Hello, World!");
-#endregion
-Console.WriteLine("After region");`.split('\n')
-      const result = findRegion(lines, 'hello')
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        'Console.WriteLine("Hello, World!");'
-      )
+    it('returns null if a region start marker exists without a matching end marker', () => {
+      const lines = [
+        '// #region missingEnd',
+        'console.log("inside region");',
+        'console.log("still inside");'
+      ]
+      expect(findRegion(lines, 'missingEnd')).toBeNull()
     })
-    test('when c# region is indented with spaces and no matching tag', () => {
-      const lines = `  Console.WriteLine("Before region");
-  #region hello
-  Console.WriteLine("Hello, World!");
-  #endregion hello
-  Console.WriteLine("After region");`.split('\n')
-      const result = findRegion(lines, 'hello')
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        '  Console.WriteLine("Hello, World!");'
-      )
+    it('returns null if an end marker exists without a preceding start marker', () => {
+      const lines = [
+        '// #endregion ghostRegion',
+        'console.log("stray end marker");'
+      ]
+      expect(findRegion(lines, 'ghostRegion')).toBeNull()
     })
-    test('when c# region with matching tag', () => {
-      const lines = `Console.WriteLine("Before region");
-#region hello
-Console.WriteLine("Hello, World!");
-#endregion hello
-Console.WriteLine("After region");`.split('\n')
-      const result = findRegion(lines, 'hello')
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        'Console.WriteLine("Hello, World!");'
-      )
+    it('detects C#/JavaScript style region markers with matching tags', () => {
+      const lines = [
+        'Console.WriteLine("Before region");',
+        '#region hello',
+        'Console.WriteLine("Hello, World!");',
+        '#endregion hello',
+        'Console.WriteLine("After region");'
+      ]
+      const result = findRegion(lines, 'hello')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          'Console.WriteLine("Hello, World!");'
+        )
+      }
     })
-    test('when c# region is not indented with spaces and no matching tag', () => {
-      const lines = `Console.WriteLine("Before region");
-#region hello
-Console.WriteLine("Hello, World!");
-#endregion
-Console.WriteLine("After region");`.split('\n')
+
+    it('detects region markers even when the end marker omits the region name', () => {
+      const lines = [
+        'Console.WriteLine("Before region");',
+        '#region hello',
+        'Console.WriteLine("Hello, World!");',
+        '#endregion',
+        'Console.WriteLine("After region");'
+      ]
       const result = findRegion(lines, 'hello')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          'Console.WriteLine("Hello, World!");'
+        )
+      }
+    })
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        'Console.WriteLine("Hello, World!");'
-      )
+    it('handles indented region markers correctly', () => {
+      const lines = [
+        '  Console.WriteLine("Before region");',
+        '  #region hello',
+        '  Console.WriteLine("Hello, World!");',
+        '  #endregion hello',
+        '  Console.WriteLine("After region");'
+      ]
+      const result = findRegion(lines, 'hello')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          '  Console.WriteLine("Hello, World!");'
+        )
+      }
     })
 
-    test('when typescript region has matching tag', () => {
-      const lines = `let regexp: RegExp[] = []
-// #region foo
-let start = -1
-// #endregion foo`.split('\n')
+    it('detects TypeScript style region markers', () => {
+      const lines = [
+        'let regexp: RegExp[] = [];',
+        '// #region foo',
+        'let start = -1;',
+        '// #endregion foo'
+      ]
       const result = findRegion(lines, 'foo')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          'let start = -1;'
+        )
+      }
+    })
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        'let start = -1'
-      )
+    it('detects CSS style region markers', () => {
+      const lines = [
+        '.body-content {',
+        '/* #region foo */',
+        '  padding-left: 15px;',
+        '/* #endregion foo */',
+        '  padding-right: 15px;',
+        '}'
+      ]
+      const result = findRegion(lines, 'foo')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          '  padding-left: 15px;'
+        )
+      }
     })
-    test('when typescript region is indented with spaces and no matching tag', () => {
-      const lines = `  let regexp: RegExp[] = []
-  // #region foo
-  let start = -1
-  // #endregion`.split('\n')
+
+    it('detects HTML style region markers', () => {
+      const lines = [
+        '<div>Some content</div>',
+        '<!-- #region foo -->',
+        '  <h1>Hello world</h1>',
+        '<!-- #endregion foo -->',
+        '<div>Other content</div>'
+      ]
       const result = findRegion(lines, 'foo')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          '  <h1>Hello world</h1>'
+        )
+      }
+    })
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        '  let start = -1'
-      )
+    it('detects Visual Basic style region markers (with case-insensitive "End")', () => {
+      const lines = [
+        'Console.WriteLine("VB")',
+        '#Region VBRegion',
+        '  Console.WriteLine("Inside region")',
+        '#End Region VBRegion',
+        'Console.WriteLine("Done")'
+      ]
+      const result = findRegion(lines, 'VBRegion')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          '  Console.WriteLine("Inside region")'
+        )
+      }
     })
 
-    test('when css region has matching tag', () => {
-      const lines = `.body-content {
-/* #region foo */
-  padding-left: 15px;
-/* #endregion foo */
-  padding-right: 15px;
-}`.split('\n')
+    it('detects Bat style region markers', () => {
+      const lines = ['::#region foo', 'echo off', '::#endregion foo']
       const result = findRegion(lines, 'foo')
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          'echo off'
+        )
+      }
+    })
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        '  padding-left: 15px;'
-      )
-    })
-    test('when css region is indented with spaces and no matching tag', () => {
-      const lines = `.body-content {
-  /* #region foo */
-  padding-left: 15px;
-  /* #endregion */
-  padding-right: 15px;
-}`.split('\n')
+    it('detects C/C++ style region markers using #pragma', () => {
+      const lines = [
+        '#pragma region foo',
+        'int a = 1;',
+        '#pragma endregion foo'
+      ]
       const result = findRegion(lines, 'foo')
-
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        '  padding-left: 15px;'
-      )
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          'int a = 1;'
+        )
+      }
     })
 
-    test('when html region has matching tag', () => {
-      const lines = `<!-- #region foo -->
-  <h1>Hello world</h1>
-<!-- #endregion foo -->
-  <p>more text</p>`.split('\n')
+    it('returns the first complete region when multiple regions exist', () => {
+      const lines = [
+        '// #region foo',
+        'first region content',
+        '// #endregion foo',
+        '// #region foo',
+        'second region content',
+        '// #endregion foo'
+      ]
       const result = findRegion(lines, 'foo')
-
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        '  <h1>Hello world</h1>'
-      )
+      expect(result).not.toBeNull()
+      if (result) {
+        expect(lines.slice(result.start, result.end).join('\n')).toBe(
+          'first region content'
+        )
+      }
     })
-    test('when html region is indented with spaces and no matching tag', () => {
-      const lines = `  <!-- #region foo -->
-  <h1>Hello world</h1>
-  <!-- #endregion foo -->
-  <p>more text</p>`.split('\n')
-      const result = findRegion(lines, 'foo')
 
-      expect(lines.slice(result?.start, result?.end).join('\n')).toBe(
-        '  <h1>Hello world</h1>'
-      )
+    it('handles nested regions with different names properly', () => {
+      const lines = [
+        '// #region foo',
+        "console.log('line before nested');",
+        '// #region bar',
+        "console.log('nested content');",
+        '// #endregion bar',
+        '// #endregion foo'
+      ]
+      const result = findRegion(lines, 'foo')
+      expect(result).not.toBeNull()
+      if (result) {
+        const extracted = lines.slice(result.start, result.end).join('\n')
+        const expected = [
+          "console.log('line before nested');",
+          '// #region bar',
+          "console.log('nested content');",
+          '// #endregion bar'
+        ].join('\n')
+        expect(extracted).toBe(expected)
+      }
     })
   })
 })
diff --git a/src/node/markdown/plugins/snippet.ts b/src/node/markdown/plugins/snippet.ts
index c03a8081..05356f72 100644
--- a/src/node/markdown/plugins/snippet.ts
+++ b/src/node/markdown/plugins/snippet.ts
@@ -51,22 +51,8 @@ export function dedent(text: string): string {
   return text
 }
 
-function testLine(
-  line: string,
-  regexp: RegExp,
-  regionName: string,
-  end: boolean = false
-) {
-  const [full, tag, name] = regexp.exec(line.trim()) || []
-
-  return full && tag && end
-    ? true
-    : name === regionName &&
-        tag.match(end ? /^[Ee]nd ?[rR]egion$/ : /^[rR]egion$/)
-}
-
 export function findRegion(lines: Array<string>, regionName: string) {
-  const regionRegexps = [
+  const regionRegexps: [RegExp, RegExp][] = [
     [
       /^[ \t]*\/\/ ?#?(region) ([\w*-]+)$/,
       /^[ \t]*\/\/ ?#?(endregion) ?([\w*-]*)$/
@@ -82,20 +68,54 @@ export function findRegion(lines: Array<string>, regionName: string) {
     [/^[ \t]*# ?(region) ([\w*-]+)$/, /^[ \t]*# ?(endregion) ?([\w*-]*)$/] // C#, PHP, Powershell, Python, perl & misc
   ]
 
-  let regexp: RegExp[] = []
-  let start = -1
-
-  for (const [lineId, line] of lines.entries()) {
-    if (regexp.length === 0) {
-      for (const reg of regionRegexps) {
-        if (testLine(line, reg[0], regionName)) {
-          start = lineId + 1
-          regexp = reg
-          break
-        }
+  let chosenRegex: [RegExp, RegExp] | null = null
+  let startLine = -1
+  // find the regex pair for a start marker that matches the given region name
+  for (let i = 0; i < lines.length; i++) {
+    const line = lines[i].trim()
+    for (const [startRegex, endRegex] of regionRegexps) {
+      const startMatch = startRegex.exec(line)
+      if (
+        startMatch &&
+        startMatch[2] === regionName &&
+        /^[rR]egion$/.test(startMatch[1])
+      ) {
+        chosenRegex = [startRegex, endRegex]
+        startLine = i + 1
+        break
+      }
+    }
+    if (chosenRegex) break
+  }
+  if (!chosenRegex) return null
+
+  const [startRegex, endRegex] = chosenRegex
+  let counter = 1
+  // scan the rest of the lines to find the matching end marker, handling nested markers
+  for (let i = startLine; i < lines.length; i++) {
+    const trimmed = lines[i].trim()
+    // check for an inner start marker for the same region
+    const startMatch = startRegex.exec(trimmed)
+    if (
+      startMatch &&
+      startMatch[2] === regionName &&
+      /^[rR]egion$/.test(startMatch[1])
+    ) {
+      counter++
+      continue
+    }
+    // check for an end marker for the same region
+    const endMatch = endRegex.exec(trimmed)
+    if (
+      endMatch &&
+      // allow empty region name on the end marker as a fallback
+      (endMatch[2] === regionName || endMatch[2] === '') &&
+      /^[Ee]nd ?[rR]egion$/.test(endMatch[1])
+    ) {
+      counter--
+      if (counter === 0) {
+        return { start: startLine, end: i, regexp: chosenRegex }
       }
-    } else if (testLine(line, regexp[1], regionName, true)) {
-      return { start, end: lineId, regexp }
     }
   }