Fix line folding for quoted strings + check for sufficient indentatio…

…n (QB6E)

__tests__/YAML-1.2.spec.js

@@ -298,10 +298,10 @@ tie-fighter: '|\\-*-/|'`,
   spans many lines.
 
 quoted: "So does this
-  quoted scalar.\n"`,
+  quoted scalar.\\n"`,
       tgt: [ {
         plain: 'This unquoted scalar spans many lines.',
-        quoted: 'So does this quoted scalar. ' } ]
+        quoted: 'So does this quoted scalar.\n' } ]
     },
   },
 

__tests__/ast/YAML-1.2.spec.js

@@ -346,10 +346,10 @@ tie-fighter: '|\\-*-/|'`,
   spans many lines.
 
 quoted: "So does this
-  quoted scalar.\n"`,
+  quoted scalar.\\n"`,
       tgt: [ { contents: [ { items: [
         'plain', { type: Type.MAP_VALUE, node: 'This unquoted scalar spans many lines.' },
-        'quoted', { type: Type.MAP_VALUE, node: 'So does this quoted scalar. ' }
+        'quoted', { type: Type.MAP_VALUE, node: 'So does this quoted scalar.\n' }
       ] } ] } ]
     },
   },

__tests__/ast/corner-cases.js

@@ -16,10 +16,18 @@ describe('folded block with chomp: keep', () => {
   })
 })
 
-test('multiple linebreaks in plain scalar', () => {
-  const src = `trimmed\n\n\n\nlines\n`
-  const doc = parse(src)[0]
-  expect(doc.contents[0].strValue).toBe('trimmed\n\n\nlines')
+describe('multiple linebreaks in scalars', () => {
+  test('plain', () => {
+    const src = `trimmed\n\n\n\nlines\n`
+    const doc = parse(src)[0]
+    expect(doc.contents[0].strValue).toBe('trimmed\n\n\nlines')
+  })
+
+  test('single-quoted', () => {
+    const src = `'trimmed\n\n\n\nlines'\n`
+    const doc = parse(src)[0]
+    expect(doc.contents[0].strValue).toBe('trimmed\n\n\nlines')
+  })
 })
 
 test('no null document for document-end marker', () => {
@@ -44,3 +52,14 @@ test('seq with anchor as explicit key', () => {
   expect(doc.contents).toHaveLength(1)
   expect(doc.contents[0].items[0].node.rawValue).toBe('- a')
 })
+
+test('unindented single-quoted string', () => {
+  const src = `key: 'two\nlines'\n`
+  const doc = parse(src)[0]
+  const { node } = doc.contents[0].items[1]
+  expect(node.error).toBeNull()
+  expect(node.strValue).toMatchObject({
+    str: 'two lines',
+    errors: [new SyntaxError('Multi-line single-quoted string needs to be sufficiently indented')]
+  })
+})

__tests__/yaml-test-suite.js

@@ -24,7 +24,6 @@ const skipErrors = [
   '9C9N',
   '9KBC',
   'B63P',
-  'QB6E',
   'SY6V',
   'ZCZ6',
   'ZL4Z',

src/ast/Node.js

@@ -127,6 +127,36 @@ export default class Node {
       : Node.endOfWhiteSpace(src, offset)
   }
 
+  // fold single newline into space, multiple newlines to N - 1 newlines
+  // presumes src[offset] === '\n'
+  static foldNewline (src, offset, indent) {
+    let inCount = 0
+    let error = false
+    let fold = ''
+    let ch = src[offset + 1]
+    while (ch === ' ' || ch === '\t' || ch === '\n') {
+      switch (ch) {
+        case '\n':
+          inCount = 0
+          offset += 1
+          fold += '\n'
+          break
+        case '\t':
+          if (inCount <= indent) error = true
+          offset = Node.endOfWhiteSpace(src, offset + 2) - 1
+          break
+        case ' ':
+          inCount += 1
+          offset += 1
+          break
+      }
+      ch = src[offset + 1]
+    }
+    if (!fold) fold = ' '
+    if (ch && inCount <= indent) error = true
+    return { fold, offset, error }
+  }
+
   constructor (type, props, context) {
     this.context = context || null
     this.error = null

src/ast/PlainValue.js

@@ -28,18 +28,9 @@ export default class PlainValue extends Node {
     for (let i = start; i < end; ++i) {
       let ch = src[i]
       if (ch === '\n') {
-        // fold single newline into space, multiple newlines to N - 1 newlines
-        let nlCount = 1
-        ch = src[i + 1]
-        while (ch === ' ' || ch === '\t' || ch === '\n') {
-          if (ch === '\n') {
-            ++nlCount
-            str += '\n'
-          }
-          i += 1
-          ch = src[i + 1]
-        }
-        if (nlCount === 1) str += ' '
+        const { fold, offset } = Node.foldNewline(src, i, -1)
+        str += fold
+        i = offset
       } else if (ch === ' ' || ch === '\t') {
         // trim trailing whitespace
         const wsStart = i

src/ast/QuoteDouble.js

@@ -30,7 +30,7 @@ export default class QuoteDouble extends Node {
     if (!this.valueRange || !this.context) return null
     const errors = []
     const { start, end } = this.valueRange
-    const { src } = this.context
+    const { indent, src } = this.context
     if (src[end - 1] !== '"') errors.push(new SyntaxError('Missing closing "quote'))
     // Using String#replace is too painful with escaped newlines preceded by
     // escaped backslashes; also, this should be faster.
@@ -40,15 +40,11 @@ export default class QuoteDouble extends Node {
       if (ch === '\n') {
         if (Node.atDocumentBoundary(src, i + 1)) errors.push(new SyntaxError(
           'Document boundary indicators are not allowed within string values'))
-        // fold single newline into space, multiple newlines to just one
-        let nlCount = 1
-        ch = src[i + 1]
-        while (ch === ' ' || ch === '\t' || ch === '\n') {
-          if (ch === '\n') ++nlCount
-          i += 1
-          ch = src[i + 1]
-        }
-        str += nlCount > 1 ? '\n' : ' '
+        const { fold, offset, error } = Node.foldNewline(src, i, indent)
+        str += fold
+        i = offset
+        if (error) errors.push(new SyntaxError(
+          'Multi-line double-quoted string needs to be sufficiently indented'))
       } else if (ch === '\\') {
         i += 1
         switch (src[i]) {

src/ast/QuoteSingle.js

@@ -23,15 +23,36 @@ export default class QuoteSingle extends Node {
     if (!this.valueRange || !this.context) return null
     const errors = []
     const { start, end } = this.valueRange
-    const { src } = this.context
+    const { indent, src } = this.context
     if (src[end - 1] !== "'") errors.push(new SyntaxError('Missing closing \'quote'))
-    const raw = src.slice(start + 1, end - 1)
-    if (/\n(?:---|\.\.\.)(?:[\n\t ]|$)/.test(raw)) errors.push(new SyntaxError(
-      'Document boundary indicators are not allowed within string values'))
-    const str = raw
-      .replace(/''/g, "'")
-      .replace(/[ \t]*\n[ \t]*/g, '\n')
-      .replace(/\n+/g, nl => nl.length === 1 ? ' ' : '\n')
+    let str = ''
+    for (let i = start + 1; i < end - 1; ++i) {
+      let ch = src[i]
+      if (ch === '\n') {
+        if (Node.atDocumentBoundary(src, i + 1)) errors.push(new SyntaxError(
+          'Document boundary indicators are not allowed within string values'))
+        const { fold, offset, error } = Node.foldNewline(src, i, indent)
+        str += fold
+        i = offset
+        if (error) errors.push(new SyntaxError(
+          'Multi-line single-quoted string needs to be sufficiently indented'))
+      } else if (ch === "'") {
+        str += ch
+        i += 1
+        if (src[i] !== "'") errors.push(new SyntaxError('Unescaped single quote? This should not happen.'))
+      } else if (ch === ' ' || ch === '\t') {
+        // trim trailing whitespace
+        const wsStart = i
+        let next = src[i + 1]
+        while (next === ' ' || next === '\t') {
+          i += 1
+          next = src[i + 1]
+        }
+        if (next !== '\n') str += i > wsStart ? src.slice(wsStart, i + 1) : ch
+      } else {
+        str += ch
+      }
+    }
     return errors.length > 0 ? { errors, str } : str
   }