fix: expression and string literal parser, #186

This commit is contained in:
harttle
2020-02-08 03:15:55 +08:00
parent 46b69db87a
commit fc0cf6fe7b
8 changed files with 222 additions and 50 deletions
+50
View File
@@ -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
}
+51 -1
View File
@@ -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
}