Add setOrigRanges() to YAML.parseCST() output

__tests__/cst/corner-cases.js

@@ -122,3 +122,17 @@ test('eemeli/yaml#l19', () => {
   expect(items).toHaveLength(2)
   expect(items[1].comment).toBe(' 123')
 })
+
+test('eemeli/yaml#20', () => {
+  const src = 'a:\r\n  123\r\nb:\r\n  456\r\n'
+  const docStream = parse(src)
+  const a = docStream[0].contents[0].items[1].node
+  expect(a.strValue).toBe('123')
+  expect(docStream.setOrigRanges()).toBe(true)
+  const { origStart: a0, origEnd: a1 } = a.valueRange
+  expect(src.slice(a0, a1)).toBe('123')
+  const b = docStream[0].contents[0].items[3].node
+  expect(b.strValue).toBe('456')
+  const { origStart: b0, origEnd: b1 } = b.valueRange
+  expect(src.slice(b0, b1)).toBe('456')
+})

__tests__/cst/parse.js

@@ -0,0 +1,203 @@
+import parse from '../../src/cst/parse'
+
+test('return value', () => {
+  const src = '- foo\n- bar\n'
+  const cst = parse(src)
+  expect(cst).toHaveLength(1)
+  expect(cst[0]).toMatchObject({
+    contents: [
+      {
+        error: null,
+        items: [
+          {
+            error: null,
+            node: {
+              error: null,
+              props: [],
+              range: { end: 6, start: 2 },
+              type: 'PLAIN',
+              value: null,
+              valueRange: { end: 5, start: 2 }
+            },
+            props: [],
+            range: { end: 6, start: 0 },
+            type: 'SEQ_ITEM',
+            value: null,
+            valueRange: { end: 5, start: 0 }
+          },
+          {
+            error: null,
+            node: {
+              error: null,
+              props: [],
+              range: { end: 12, start: 8 },
+              type: 'PLAIN',
+              value: null,
+              valueRange: { end: 11, start: 8 }
+            },
+            props: [],
+            range: { end: 12, start: 6 },
+            type: 'SEQ_ITEM',
+            value: null,
+            valueRange: { end: 11, start: 6 }
+          }
+        ],
+        props: [],
+        range: { end: 12, start: 0 },
+        type: 'SEQ',
+        value: null,
+        valueRange: { end: 11, start: 0 }
+      }
+    ],
+    directives: [],
+    error: null,
+    props: [],
+    range: null,
+    type: 'DOCUMENT',
+    value: null,
+    valueRange: { end: 12, start: 0 }
+  })
+})
+
+describe('toString()', () => {
+  test('plain document', () => {
+    const src = '- foo\n- bar\n'
+    const cst = parse(src)
+    expect(String(cst)).toBe(src)
+  })
+
+  test('stream of two documents', () => {
+    const src = 'foo\n...\nbar\n'
+    const cst = parse(src)
+    expect(cst).toHaveLength(2)
+    expect(String(cst)).toBe(src)
+  })
+
+  test('document with CRLF line separators', () => {
+    const src = '- foo\r\n- bar\r\n'
+    const cst = parse(src)
+    expect(cst.toString()).toBe('- foo\n- bar\n')
+  })
+})
+
+describe('setOrigRanges()', () => {
+  test('return false for no CRLF', () => {
+    const src = '- foo\n- bar\n'
+    const cst = parse(src)
+    expect(cst.setOrigRanges()).toBe(false)
+    expect(cst[0].valueRange).toMatchObject({ start: 0, end: 12 })
+    expect(cst[0].valueRange.origStart).toBeUndefined()
+    expect(cst[0].valueRange.origEnd).toBeUndefined()
+  })
+
+  test('single document', () => {
+    const src = '- foo\r\n- bar\r\n'
+    const cst = parse(src)
+    expect(cst.setOrigRanges()).toBe(true)
+    expect(cst).toHaveLength(1)
+    const { range, valueRange } = cst[0].contents[0].items[1].node
+    expect(src.slice(range.origStart, range.origEnd)).toBe('bar\r\n')
+    expect(src.slice(valueRange.origStart, valueRange.origEnd)).toBe('bar')
+    expect(cst[0]).toMatchObject({
+      contents: [
+        {
+          error: null,
+          items: [
+            {
+              error: null,
+              node: {
+                error: null,
+                props: [],
+                range: { end: 6, origEnd: 7, origStart: 2, start: 2 },
+                type: 'PLAIN',
+                value: null,
+                valueRange: { end: 5, origEnd: 5, origStart: 2, start: 2 }
+              },
+              props: [],
+              range: { end: 6, origEnd: 7, origStart: 0, start: 0 },
+              type: 'SEQ_ITEM',
+              value: null,
+              valueRange: { end: 5, origEnd: 5, origStart: 0, start: 0 }
+            },
+            {
+              error: null,
+              node: {
+                error: null,
+                props: [],
+                range: { end: 12, origEnd: 14, origStart: 9, start: 8 },
+                type: 'PLAIN',
+                value: null,
+                valueRange: { end: 11, origEnd: 12, origStart: 9, start: 8 }
+              },
+              props: [],
+              range: { end: 12, origEnd: 14, origStart: 7, start: 6 },
+              type: 'SEQ_ITEM',
+              value: null,
+              valueRange: { end: 11, origEnd: 12, origStart: 7, start: 6 }
+            }
+          ],
+          props: [],
+          range: { end: 12, origEnd: 14, origStart: 0, start: 0 },
+          type: 'SEQ',
+          value: null,
+          valueRange: { end: 11, origEnd: 12, origStart: 0, start: 0 }
+        }
+      ],
+      directives: [],
+      error: null,
+      props: [],
+      range: null,
+      type: 'DOCUMENT',
+      value: null,
+      valueRange: { end: 12, origEnd: 14, origStart: 0, start: 0 }
+    })
+  })
+
+  test('stream of two documents', () => {
+    const src = 'foo\r\n...\r\nbar\r\n'
+    const cst = parse(src)
+    expect(cst.setOrigRanges()).toBe(true)
+    expect(cst).toHaveLength(2)
+    const { range, valueRange } = cst[1].contents[0]
+    expect(src.slice(range.origStart, range.origEnd)).toBe('bar\r\n')
+    expect(src.slice(valueRange.origStart, valueRange.origEnd)).toBe('bar')
+    expect(cst[0]).toMatchObject({
+      contents: [
+        {
+          error: null,
+          props: [],
+          range: { end: 4, origEnd: 5, origStart: 0, start: 0 },
+          type: 'PLAIN',
+          value: null,
+          valueRange: { end: 3, origEnd: 3, origStart: 0, start: 0 }
+        }
+      ],
+      directives: [],
+      error: null,
+      props: [],
+      range: null,
+      type: 'DOCUMENT',
+      value: null,
+      valueRange: { end: 4, origEnd: 5, origStart: 0, start: 0 }
+    })
+    expect(cst[1]).toMatchObject({
+      contents: [
+        {
+          error: null,
+          props: [],
+          range: { end: 12, origEnd: 15, origStart: 10, start: 8 },
+          type: 'PLAIN',
+          value: null,
+          valueRange: { end: 11, origEnd: 13, origStart: 10, start: 8 }
+        }
+      ],
+      directives: [],
+      error: null,
+      props: [],
+      range: null,
+      type: 'DOCUMENT',
+      value: null,
+      valueRange: { end: 12, origEnd: 15, origStart: 10, start: 8 }
+    })
+  })
+})

src/Document.js

@@ -107,7 +107,7 @@ export default class Document {
     const comments = { before: [], after: [] }
     const contentNodes = []
     contents.forEach(node => {
-      if (node.valueRange && !node.valueRange.isEmpty) {
+      if (node.valueRange && !node.valueRange.isEmpty()) {
         if (contentNodes.length === 1) {
           const msg = 'Document is not valid YAML (bad indentation?)'
           this.errors.push(new YAMLSyntaxError(node, msg))

src/cst/BlockValue.js

@@ -19,7 +19,7 @@ export default class BlockValue extends Node {
     if (!this.valueRange || !this.context) return null
     let { start, end } = this.valueRange
     const { indent, src } = this.context
-    if (this.valueRange.isEmpty) return ''
+    if (this.valueRange.isEmpty()) return ''
     let lastNewLine = null
     let ch = src[end - 1]
     while (ch === '\n' || ch === '\t' || ch === ' ') {
@@ -179,4 +179,9 @@ export default class BlockValue extends Node {
       JSON.stringify(this.rawValue)
     return offset
   }
+
+  setOrigRanges(cr, offset) {
+    offset = super.setOrigRanges(cr, offset)
+    return this.header ? this.header.setOrigRange(cr, offset) : offset
+  }
 }

src/cst/Collection.js

@@ -115,6 +115,14 @@ export default class Collection extends Node {
     return offset
   }
 
+  setOrigRanges(cr, offset) {
+    offset = super.setOrigRanges(cr, offset)
+    this.items.forEach(node => {
+      offset = node.setOrigRanges(cr, offset)
+    })
+    return offset
+  }
+
   toString() {
     const {
       context: { src },

src/cst/CollectionItem.js

@@ -65,6 +65,11 @@ export default class CollectionItem extends Node {
     return offset
   }
 
+  setOrigRanges(cr, offset) {
+    offset = super.setOrigRanges(cr, offset)
+    return this.node ? this.node.setOrigRanges(cr, offset) : offset
+  }
+
   toString() {
     const {
       context: { src },

src/cst/Document.js

@@ -162,6 +162,17 @@ export default class Document extends Node {
     return offset
   }
 
+  setOrigRanges(cr, offset) {
+    offset = super.setOrigRanges(cr, offset)
+    this.directives.forEach(node => {
+      offset = node.setOrigRanges(cr, offset)
+    })
+    this.contents.forEach(node => {
+      offset = node.setOrigRanges(cr, offset)
+    })
+    return offset
+  }
+
   toString() {
     const {
       contents,

src/cst/FlowCollection.js

@@ -110,6 +110,14 @@ export default class FlowCollection extends Node {
     return offset
   }
 
+  setOrigRanges(cr, offset) {
+    offset = super.setOrigRanges(cr, offset)
+    this.items.forEach(node => {
+      offset = node.setOrigRanges(cr, offset)
+    })
+    return offset
+  }
+
   toString() {
     const {
       context: { src },

src/cst/Node.js

@@ -288,6 +288,21 @@ export default class Node {
     return start
   }
 
+  /**
+   * Populates the `origStart` and `origEnd` values of all ranges for this
+   * node. Extended by child classes to handle descendant nodes.
+   *
+   * @param {number[]} cr - Positions of dropped CR characters
+   * @param {number} offset - Starting index of `cr` from the last call
+   * @returns {number} - The next offset, matching the one found for `origStart`
+   */
+  setOrigRanges(cr, offset) {
+    if (this.range) offset = this.range.setOrigRange(cr, offset)
+    this.valueRange.setOrigRange(cr, offset)
+    this.props.forEach(prop => prop.setOrigRange(cr, offset))
+    return offset
+  }
+
   toString() {
     const {
       context: { src },

src/cst/PlainValue.js

@@ -66,7 +66,7 @@ export default class PlainValue extends Node {
       if (end === null || src[end] === '#') break
       offset = PlainValue.endOfLine(src, end, inFlow)
     }
-    if (this.valueRange.isEmpty) this.valueRange.start = start
+    if (this.valueRange.isEmpty()) this.valueRange.start = start
     this.valueRange.end = offset
     trace: this.valueRange, JSON.stringify(this.rawValue)
     return offset
@@ -112,7 +112,7 @@ export default class PlainValue extends Node {
     trace: 'first line',
       { valueRange: this.valueRange, comment: this.comment },
       JSON.stringify(this.rawValue)
-    if (!this.hasComment || this.valueRange.isEmpty) {
+    if (!this.hasComment || this.valueRange.isEmpty()) {
       offset = this.parseBlockValue(offset)
     }
     trace: this.type,

src/cst/Range.js

@@ -8,15 +8,38 @@ export default class Range {
     this.end = end || start
   }
 
-  get isEmpty() {
+  isEmpty() {
     return typeof this.start !== 'number' || !this.end || this.end <= this.start
   }
 
-  get length() {
-    return this.isEmpty ? 0 : this.end - this.start
-  }
-
-  apply(src) {
-    return this.isEmpty ? '' : src.slice(this.start, this.end)
+  /**
+   * Set `origStart` and `origEnd` to point to the original source range for
+   * this node, which may differ due to dropped CR characters.
+   *
+   * @param {number[]} cr - Positions of dropped CR characters
+   * @param {number} offset - Starting index of `cr` from the last call
+   * @returns {number} - The next offset, matching the one found for `origStart`
+   */
+  setOrigRange(cr, offset) {
+    const { start, end } = this
+    if (cr.length === 0 || end <= cr[0]) {
+      this.origStart = start
+      this.origEnd = end
+      return offset
+    }
+    let i = offset
+    while (i < cr.length) {
+      if (cr[i] > start) break
+      else ++i
+    }
+    this.origStart = start + i
+    const nextOffset = i
+    while (i < cr.length) {
+      // if end was at \n, it should now be at \r
+      if (cr[i] >= end) break
+      else ++i
+    }
+    this.origEnd = end + i
+    return nextOffset
   }
 }