fix: expression and string literal parser, #186

package-lock.json

@@ -1,6 +1,6 @@
 {
   "name": "liquidjs",
-  "version": "9.1.1",
+  "version": "9.6.2",
   "lockfileVersion": 1,
   "requires": true,
   "dependencies": {
@@ -10814,8 +10814,7 @@
         },
         "commander": {
           "version": "2.17.1",
-          "bundled": true,
-          "optional": true
+          "bundled": true
         },
         "commondir": {
           "version": "1.0.1",
@@ -11579,8 +11578,7 @@
           "dependencies": {
             "source-map": {
               "version": "0.6.1",
-              "bundled": true,
-              "optional": true
+              "bundled": true
             }
           }
         },

src/parser/expression-tokenizer.ts

@@ -0,0 +1,50 @@
+const rBlank = /\s/
+const rPunctuation = /[<>=!]/
+
+enum ParseState {
+  INIT = 1,
+  SINGLE_QUOTE = 2,
+  DOUBLE_QUOTE = 4,
+  QUOTE = 6,
+  BRACKET = 8
+}
+
+export function * tokenize (expr: string): IterableIterator<string> {
+  const N = expr.length
+  const stack = [ParseState.INIT]
+  let str = ''
+  let lastIsPunc = false
+
+  for (let i = 0; i < N; i++) {
+    const c = expr[i]
+    const top = stack[stack.length - 1]
+    const isPunc = rPunctuation.test(c)
+    if (c === '\\') {
+      str += expr.substr(i++, 2)
+    } else if (top === ParseState.SINGLE_QUOTE && c === "'") {
+      str += c
+      stack.pop()
+    } else if (top === ParseState.DOUBLE_QUOTE && c === '"') {
+      str += c
+      stack.pop()
+    } else if (ParseState.QUOTE & top) {
+      str += c
+    } else if (top === ParseState.BRACKET && c === ']') {
+      str += c
+      stack.pop()
+    } else if (top === ParseState.INIT && rBlank.exec(c)) {
+      if (str) yield str
+      str = ''
+    } else if (top === ParseState.INIT && isPunc !== lastIsPunc) {
+      if (str) yield str
+      str = c
+    } else {
+      if (c === '"') stack.push(ParseState.DOUBLE_QUOTE)
+      else if (c === "'") stack.push(ParseState.SINGLE_QUOTE)
+      else if (c === '[') stack.push(ParseState.BRACKET)
+      str += c
+    }
+    lastIsPunc = isPunc
+  }
+  if (str) yield str
+}

src/parser/literal.ts

@@ -5,6 +5,24 @@ import { BlankDrop } from '../drop/blank-drop'
 
 type literal = true | false | NullDrop | EmptyDrop | BlankDrop | number | string
 
+const rHex = /[\da-fA-F]/
+const rOct = /[0-7]/
+const escapeChar = {
+  b: '\b',
+  f: '\f',
+  n: '\n',
+  r: '\r',
+  t: '\t',
+  v: '\x0B'
+}
+
+function hexVal (c: string) {
+  const code = c.charCodeAt(0)
+  if (code >= 97) return code - 87
+  if (code >= 65) return code - 55
+  return code - 48
+}
+
 export function parseLiteral (str: string): literal | undefined {
   str = str.trim()
 
@@ -14,5 +32,37 @@ export function parseLiteral (str: string): literal | undefined {
   if (str === 'empty') return new EmptyDrop()
   if (str === 'blank') return new BlankDrop()
   if (!isNaN(Number(str))) return Number(str)
-  if ((str[0] === '"' || str[0] === "'") && str[0] === last(str)) return str.slice(1, -1)
+  if ((str[0] === '"' || str[0] === "'") && str[0] === last(str)) return parseStringLiteral(str)
+}
+
+export function parseStringLiteral (str: string): string {
+  let ret = ''
+  for (let i = 1; i < str.length - 1; i++) {
+    if (str[i] !== '\\') {
+      ret += str[i]
+      continue
+    }
+    if (escapeChar[str[i + 1]] !== undefined) {
+      ret += escapeChar[str[++i]]
+    } else if (str[i + 1] === 'u') {
+      let val = 0
+      let j = i + 2
+      while (j <= i + 5 && rHex.test(str[j])) {
+        val = val * 16 + hexVal(str[j++])
+      }
+      i = j - 1
+      ret += String.fromCharCode(val)
+    } else if (!rOct.test(str[i + 1])) {
+      ret += str[++i]
+    } else {
+      let j = i + 1
+      let val = 0
+      while (j <= i + 3 && rOct.test(str[j])) {
+        val = val * 8 + hexVal(str[j++])
+      }
+      i = j - 1
+      ret += String.fromCharCode(val)
+    }
+  }
+  return ret
 }

src/render/expression.ts

@@ -4,20 +4,24 @@ import { Value } from './value'
 import { Context } from '../context/context'
 import { toValue } from '../util/underscore'
 import { isOperator, precedence, operatorImpls } from './operator'
+import { tokenize } from '../parser/expression-tokenizer'
 
 export class Expression {
   private operands: any[] = []
   private postfix: string[]
 
   public constructor (str = '') {
-    this.postfix = [...toPostfix(str)]
+    this.postfix = [...toPostfix(tokenize(str))]
   }
   public * evaluate (ctx: Context) {
     assert(ctx, 'unable to evaluate: context not defined')
 
     for (const token of this.postfix) {
       if (isOperator(token)) {
-        this.evaluateOnce(token)
+        const r = this.operands.pop()
+        const l = this.operands.pop()
+        const result = operatorImpls[token](l, r)
+        this.operands.push(result)
       } else if (isRange(token)) {
         this.operands.push(yield rangeValue(token, ctx))
       } else this.operands.push(yield new Value(token).evaluate(ctx))
@@ -27,47 +31,11 @@ export class Expression {
   public * value (ctx: Context) {
     return toValue(yield this.evaluate(ctx))
   }
-  private evaluateOnce (token: string) {
-    const r = this.operands.pop()
-    const l = this.operands.pop()
-    const result = operatorImpls[token](l, r)
-    this.operands.push(result)
-  }
 }
 
-function * tokenize (expr: string): IterableIterator<string> {
-  const N = expr.length
-  let str = ''
-  const pairs = { '"': '"', "'": "'", '[': ']', '(': ')' }
-
-  for (let i = 0; i < N; i++) {
-    const c = expr[i]
-    switch (c) {
-      case '[':
-      case '"':
-      case "'":
-        str += c
-        while (i + 1 < N) {
-          str += expr[++i]
-          if (expr[i] === pairs[c]) break
-        }
-        break
-      case ' ':
-      case '\t':
-      case '\n':
-        if (str) yield str
-        str = ''
-        break
-      default:
-        str += c
-    }
-  }
-  if (str) yield str
-}
-
-function * toPostfix (expr: string): IterableIterator<string> {
+function * toPostfix (tokens: IterableIterator<string>): IterableIterator<string> {
   const ops = []
-  for (const token of tokenize(expr)) {
+  for (const token of tokens) {
     if (isOperator(token)) {
       while (ops.length && precedence[ops[ops.length - 1]] > precedence[token]) {
         yield ops.pop()!

test/integration/builtin/tags/if.ts

@@ -65,8 +65,18 @@ describe('tags/if', function () {
       const html = await liquid.parseAndRender(src)
       return expect(html).to.equal('')
     })
+    it('should allow no spaces around operator for literal', async function () {
+      const src = `{% if true==true %}success{%else%}fail{% endif %}`
+      const html = await liquid.parseAndRender(src)
+      return expect(html).to.equal('success')
+    })
+    it('should allow no spaces around operator for variables', async function () {
+      const src = `{%assign var = 1%}{%if var ==1%}success{%else%}fail{%endif%}`
+      const html = await liquid.parseAndRender(src)
+      return expect(html).to.equal('success')
+    })
   })
-  describe('comparasion to null', function () {
+  describe('compare to null', function () {
     it('should evaluate false for null < 10', async function () {
       const src = '{% if null < 10 %}yes{% else %}no{% endif %}'
       const html = await liquid.parseAndRender(src, scope)

test/unit/parser/expression-tokenizer.ts

@@ -0,0 +1,43 @@
+import { tokenize } from '../../../src/parser/expression-tokenizer'
+import { expect } from 'chai'
+
+describe('expression tokenizer', () => {
+  describe('spaces', () => {
+    it('should tokenize a + b', () => {
+      expect([...tokenize('a + b')]).to.deep.equal(['a', '+', 'b'])
+    })
+    it('should tokenize a==1', () => {
+      expect([...tokenize('a==1')]).to.deep.equal(['a', '==', '1'])
+    })
+  })
+
+  describe('range', () => {
+    it('should tokenize (1..3) contains 3', () => {
+      expect([...tokenize('(1..3)')]).to.deep.equal(['(1..3)'])
+    })
+  })
+
+  describe('bracket', () => {
+    it('should tokenize a[b] = c', () => {
+      expect([...tokenize('a[b] = c')]).to.deep.equal(['a[b]', '=', 'c'])
+    })
+    it('should tokenize c[a["b"]] < c', () => {
+      expect([...tokenize('c[a["b"]] < c')]).to.deep.equal(['c[a["b"]]', '<', 'c'])
+    })
+    it('should tokenize "][" == var', () => {
+      expect([...tokenize('"][" == var')]).to.deep.equal(['"]["', '==', 'var'])
+    })
+  })
+
+  describe('quotes', () => {
+    it('should tokenize " " == var', () => {
+      expect([...tokenize('" " == var')]).to.deep.equal(['" "', '==', 'var'])
+    })
+    it('should tokenize "\\\'" == var', () => {
+      expect([...tokenize('"\\\'" == var')]).to.deep.equal(['"\\\'"', '==', 'var'])
+    })
+    it('should tokenize "\\"" == var', () => {
+      expect([...tokenize('"\\"" == var')]).to.deep.equal(['"\\""', '==', 'var'])
+    })
+  })
+})

test/unit/parser/literal.ts

@@ -1,8 +1,8 @@
 import { expect } from 'chai'
-import { parseLiteral } from '../../../src/parser/literal'
+import { parseLiteral, parseStringLiteral } from '../../../src/parser/literal'
 import { NullDrop } from '../../../src/drop/null-drop'
 
-describe('parseLiteral', function () {
+describe('parseLiteral()', function () {
   it('should eval boolean literal', async function () {
     expect(parseLiteral('true')).to.equal(true)
     expect(parseLiteral('TrUE')).to.equal(undefined)
@@ -25,3 +25,36 @@ describe('parseLiteral', function () {
     expect(parseLiteral('null')).to.be.instanceOf(NullDrop)
   })
 })
+
+describe('parseStringLiteral()', function () {
+  it('should parse octal escape', () => {
+    expect(parseStringLiteral(String.raw`"\1010"`)).to.equal('A0')
+    expect(parseStringLiteral(String.raw`"\12"`)).to.equal('\n')
+    expect(parseStringLiteral(String.raw`"\01"`)).to.equal('\u0001')
+    expect(parseStringLiteral(String.raw`"\0"`)).to.equal('\0')
+  })
+  it('should skip invalid octal escape', () => {
+    expect(parseStringLiteral(String.raw`"\9"`)).to.equal('9')
+  })
+  it('should parse \n, \t, \r', () => {
+    expect(parseStringLiteral(String.raw`"fo\no"`)).to.equal('fo\no')
+    expect(parseStringLiteral(String.raw`'fo\to'`)).to.equal('fo\to')
+    expect(parseStringLiteral(String.raw`'fo\ro'`)).to.equal('fo\ro')
+  })
+  it('should parse unicode(hex) escape', () => {
+    expect(parseStringLiteral('"\\u003C"')).to.equal('<')
+    expect(parseStringLiteral('"\\u003cZ"')).to.equal('<Z')
+    expect(parseStringLiteral('"\\u41"')).to.equal('A')
+  })
+  it('should skip invalid unicode(hex) escape', () => {
+    expect(parseStringLiteral('"\\u41Z"')).to.equal('AZ')
+    expect(parseStringLiteral('"\\uZ"')).to.equal('\0Z')
+  })
+  it('should parse quote escape', () => {
+    expect(parseStringLiteral(String.raw`"fo\'o"`)).to.equal("fo'o")
+    expect(parseStringLiteral(String.raw`'fo\"o'`)).to.equal('fo"o')
+  })
+  it('should parse slash escape', () => {
+    expect(parseStringLiteral(String.raw`'fo\\o'`)).to.equal('fo\\o')
+  })
+})

test/unit/render/expression.ts

@@ -11,9 +11,14 @@ describe('Expression', function () {
       one: 1,
       two: 2,
       empty: '',
+      quote: '"',
+      space: ' ',
       x: 'XXX',
       y: undefined,
-      z: null
+      z: null,
+      obj: {
+        ']': 'right bracket'
+      }
     })
   })
 
@@ -26,6 +31,8 @@ describe('Expression', function () {
   })
 
   it('should eval simple expression', async function () {
+    expect(await toThenable(new Expression('1==2').value(ctx))).to.equal(false)
+    expect(await toThenable(new Expression('1<2').value(ctx))).to.equal(true)
     expect(await toThenable(new Expression('1 < 2').value(ctx))).to.equal(true)
     expect(await toThenable(new Expression('1   <   2').value(ctx))).to.equal(true)
     expect(await toThenable(new Expression('2 <= 2').value(ctx))).to.equal(true)
@@ -40,6 +47,19 @@ describe('Expression', function () {
     expect(await toThenable(new Expression('"<=" == "<="').value(ctx))).to.equal(true)
   })
 
+  it('should allow space in quoted value', async function () {
+    expect(await toThenable(new Expression('" " == space').value(ctx))).to.equal(true)
+  })
+
+  describe('escape', () => {
+    it('should escape quote', async function () {
+      expect(await toThenable(new Expression('"\\"" == quote').value(ctx))).to.equal(true)
+    })
+    it('should escape square bracket', async function () {
+      expect(await toThenable(new Expression('obj["]"] == "right bracket"').value(ctx))).to.equal(true)
+    })
+  })
+
   describe('complex expression', function () {
     it('should support value or value', async function () {
       expect(await toThenable(new Expression('false or true').value(ctx))).to.equal(true)