Merge pull request #68 from MikeRalphson/lineOffsets

Optionally compute lineOffsets, fixes #67

src/cst/Document.js

@@ -179,6 +179,7 @@ export default class Document extends Node {
    * @returns {number} - Index of the character after this
    */
   parse(context, start) {
+    context.root = this
     this.context = context
     const { src } = context
     trace: 'DOC START', JSON.stringify(src.slice(start))

src/cst/Node.js

@@ -1,3 +1,4 @@
+import getLinePos from './getLinePos'
 import Range from './Range'
 
 export const Type = {
@@ -248,6 +249,14 @@ export default class Node {
     return jsonLikeTypes.indexOf(this.type) !== -1
   }
 
+  get rangeAsLinePos() {
+    if (!this.range || !this.context) return undefined
+    const start = getLinePos(this.range.start, this.context.root)
+    if (!start) return undefined
+    const end = getLinePos(this.range.end, this.context.root)
+    return { start, end }
+  }
+
   get rawValue() {
     if (!this.valueRange || !this.context) return null
     const { start, end } = this.valueRange

src/cst/ParseContext.js

@@ -65,6 +65,7 @@ export default class ParseContext {
     this.indent = indent != null ? indent : orig.indent
     this.lineStart = lineStart != null ? lineStart : orig.lineStart
     this.parent = parent != null ? parent : orig.parent || {}
+    this.root = orig.root
     this.src = orig.src
   }
 

src/cst/getLinePos.js

@@ -0,0 +1,54 @@
+function findLineStarts(src) {
+  const ls = [0]
+  let offset = src.indexOf('\n')
+  while (offset !== -1) {
+    offset += 1
+    ls.push(offset)
+    offset = src.indexOf('\n', offset)
+  }
+  return ls
+}
+
+/**
+ * Determine the line/col position matching a character offset.
+ *
+ * Accepts a source string or a CST document as the second parameter. With
+ * the latter, starting indices for lines are cached in the document as
+ * `lineStarts: number[]`.
+ *
+ * Returns a zero-indexed `{ line, col }` location if found, or
+ * `undefined` otherwise.
+ *
+ * @param {number} offset
+ * @param {string|Document|Document[]} cst
+ * @returns {{ line: number, col: number }|undefined}
+ */
+export default function getLinePos(offset, cst) {
+  if (typeof offset === 'number' && offset >= 0) {
+    let lineStarts, srcLength
+    if (typeof cst === 'string') {
+      lineStarts = findLineStarts(cst)
+      srcLength = cst.length
+    } else {
+      if (Array.isArray(cst)) cst = cst[0]
+      if (cst) {
+        if (!cst.lineStarts) cst.lineStarts = findLineStarts(cst.context.src)
+        lineStarts = cst.lineStarts
+        srcLength = cst.context.src.length
+      }
+    }
+    if (lineStarts && offset <= srcLength) {
+      for (let i = 0; i < lineStarts.length; ++i) {
+        const start = lineStarts[i]
+        if (offset < start) {
+          const line = i - 1
+          return { line, col: offset - lineStarts[line] }
+        }
+        if (offset === start) return { line: i, col: 0 }
+      }
+      const line = lineStarts.length - 1
+      return { line, col: offset - lineStarts[line] }
+    }
+  }
+  return undefined
+}

src/cst/parse.js

@@ -11,11 +11,11 @@ export default function parse(src) {
       return '\n'
     })
   }
-  const context = new ParseContext({ src })
   const documents = []
   let offset = 0
   do {
     const doc = new Document()
+    const context = new ParseContext({ src })
     offset = doc.parse(context, offset)
     documents.push(doc)
   } while (offset < src.length)

tests/cst/getLinePos.js

@@ -0,0 +1,49 @@
+import getLinePos from '../../src/cst/getLinePos'
+import parse from '../../src/cst/parse'
+
+test('lineStarts for empty document', () => {
+  const src = ''
+  const cst = parse(src)
+  expect(() => getLinePos(0, cst)).not.toThrow()
+  expect(cst[0].lineStarts).toMatchObject([0])
+})
+
+test('lineStarts for multiple documents', () => {
+  const src = 'foo\n...\nbar\n'
+  const cst = parse(src)
+  expect(() => getLinePos(0, cst)).not.toThrow()
+  expect(cst[0].lineStarts).toMatchObject([0, 4, 8, 12])
+})
+
+test('lineStarts for malformed document', () => {
+  const src = '- foo\n\t- bar\n'
+  const cst = parse(src)
+  expect(() => getLinePos(0, cst)).not.toThrow()
+  expect(cst[0].lineStarts).toMatchObject([0, 6, 13])
+})
+
+test('getLinePos()', () => {
+  const src = '- foo\n- bar\n'
+  const cst = parse(src)
+  expect(cst[0].lineStarts).toBeUndefined()
+  expect(getLinePos(0, cst)).toMatchObject({ line: 0, col: 0 })
+  expect(getLinePos(1, cst)).toMatchObject({ line: 0, col: 1 })
+  expect(getLinePos(2, cst)).toMatchObject({ line: 0, col: 2 })
+  expect(getLinePos(5, cst)).toMatchObject({ line: 0, col: 5 })
+  expect(getLinePos(6, cst)).toMatchObject({ line: 1, col: 0 })
+  expect(getLinePos(7, cst)).toMatchObject({ line: 1, col: 1 })
+  expect(getLinePos(11, cst)).toMatchObject({ line: 1, col: 5 })
+  expect(getLinePos(12, cst)).toMatchObject({ line: 2, col: 0 })
+  expect(cst[0].lineStarts).toMatchObject([0, 6, 12])
+})
+
+test('invalid args for getLinePos()', () => {
+  const src = '- foo\n- bar\n'
+  const cst = parse(src)
+  expect(getLinePos()).toBeUndefined()
+  expect(getLinePos(0)).toBeUndefined()
+  expect(getLinePos(1)).toBeUndefined()
+  expect(getLinePos(-1, cst)).toBeUndefined()
+  expect(getLinePos(13, cst)).toBeUndefined()
+  expect(getLinePos(Math.MAXINT, cst)).toBeUndefined()
+})

tests/cst/parse.js

@@ -156,6 +156,8 @@ describe('setOrigRanges()', () => {
       value: null,
       valueRange: { end: 12, origEnd: 14, origStart: 0, start: 0 }
     })
+    expect(cst[0].context.root).toBe(cst[0])
+    expect(cst[0].contents[0].items[1].node.context.root).toBe(cst[0])
   })
 
   test('stream of two documents', () => {
@@ -204,6 +206,8 @@ describe('setOrigRanges()', () => {
       value: null,
       valueRange: { end: 12, origEnd: 15, origStart: 10, start: 8 }
     })
+    expect(cst[0].context.root).toBe(cst[0])
+    expect(cst[1].context.root).toBe(cst[1])
   })
 
   test('flow collections', () => {
@@ -242,5 +246,7 @@ describe('setOrigRanges()', () => {
       value: null,
       valueRange: { end: 7, origEnd: 9, origStart: 2, start: 1 }
     })
+    expect(cst[0].context.root).toBe(cst[0])
+    expect(cst[0].contents[1].context.root).toBe(cst[0])
   })
 })

tests/doc/corner-cases.js

@@ -1,6 +1,5 @@
 import fs from 'fs'
 import path from 'path'
-import Node from '../../src/cst/Node'
 import YAML from '../../src/index'
 
 test('eemeli/yaml#2', () => {
@@ -25,36 +24,6 @@ test('eemeli/yaml#3', () => {
   expect(doc.contents.items[0].value.value).toBe(123)
 })
 
-test('eemeli/yaml#6', () => {
-  const src = 'abc: 123\ndef'
-  const doc = YAML.parseDocument(src)
-  expect(doc.errors).toHaveLength(1)
-  expect(doc.errors[0].name).toBe('YAMLSemanticError')
-  expect(doc.errors[0].source).toBeInstanceOf(Node)
-})
-
-describe('eemeli/yaml#7', () => {
-  test('map', () => {
-    const src = '{ , }\n---\n{ 123,,, }\n'
-    const docs = YAML.parseAllDocuments(src)
-    expect(docs[0].errors).toHaveLength(1)
-    expect(docs[1].errors).toHaveLength(2)
-  })
-  test('seq', () => {
-    const src = '[ , ]\n---\n[ 123,,, ]\n'
-    const docs = YAML.parseAllDocuments(src)
-    expect(docs[0].errors).toHaveLength(1)
-    expect(docs[1].errors).toHaveLength(2)
-  })
-})
-
-test('eemeli/yaml#8', () => {
-  const src = '{'
-  const doc = YAML.parseDocument(src)
-  expect(doc.errors).toHaveLength(1)
-  expect(doc.errors[0].name).toBe('YAMLSemanticError')
-})
-
 describe('eemeli/yaml#10', () => {
   test('reported', () => {
     const src = `

tests/doc/errors.js

@@ -0,0 +1,59 @@
+import Node from '../../src/cst/Node'
+import YAML from '../../src/index'
+
+test('eemeli/yaml#6', () => {
+  const src = 'abc: 123\ndef'
+  const doc = YAML.parseDocument(src)
+  expect(doc.errors).toMatchObject([{ name: 'YAMLSemanticError' }])
+  const node = doc.errors[0].source
+  expect(node).toBeInstanceOf(Node)
+  expect(node.rangeAsLinePos).toMatchObject({
+    start: { line: 1, col: 0 },
+    end: { line: 1, col: 3 }
+  })
+})
+
+describe('eemeli/yaml#7', () => {
+  test('map', () => {
+    const src = '{ , }\n---\n{ 123,,, }\n'
+    const docs = YAML.parseAllDocuments(src)
+    expect(docs[0].errors).toMatchObject([{ name: 'YAMLSyntaxError' }])
+    expect(docs[1].errors).toMatchObject([
+      { name: 'YAMLSyntaxError' },
+      { name: 'YAMLSyntaxError' }
+    ])
+    const node = docs[0].errors[0].source
+    expect(node).toBeInstanceOf(Node)
+    expect(node.rangeAsLinePos).toMatchObject({
+      start: { line: 0, col: 0 },
+      end: { line: 0, col: 5 }
+    })
+  })
+  test('seq', () => {
+    const src = '[ , ]\n---\n[ 123,,, ]\n'
+    const docs = YAML.parseAllDocuments(src)
+    expect(docs[0].errors).toMatchObject([{ name: 'YAMLSyntaxError' }])
+    expect(docs[1].errors).toMatchObject([
+      { name: 'YAMLSyntaxError' },
+      { name: 'YAMLSyntaxError' }
+    ])
+    const node = docs[1].errors[0].source
+    expect(node).toBeInstanceOf(Node)
+    expect(node.rangeAsLinePos).toMatchObject({
+      start: { line: 2, col: 0 },
+      end: { line: 2, col: 10 }
+    })
+  })
+})
+
+test('eemeli/yaml#8', () => {
+  const src = '{'
+  const doc = YAML.parseDocument(src)
+  expect(doc.errors).toMatchObject([{ name: 'YAMLSemanticError' }])
+  const node = doc.errors[0].source
+  expect(node).toBeInstanceOf(Node)
+  expect(node.rangeAsLinePos).toMatchObject({
+    start: { line: 0, col: 0 },
+    end: { line: 0, col: 1 }
+  })
+})