feat: export unescape functions

feat: unescape unicode chars

Author: mdvorak

src/properties.spec.ts

@@ -1,4 +1,4 @@
-import * as properties from '.'
+import properties from '.'
 
 describe('parse', () => {
   it('should parse all lines', () => {
@@ -129,10 +129,10 @@ describe('data access', () => {
     ['foo9=', 'bar9', 'foo9\\==bar9'],
     ['foo10=', 'bar10', 'foo10\\==bar10'],
     ['foo11 ', 'bar11', 'foo11\\ =bar11'],
-    [' foo12', 'bar12 ', '\\ foo12=bar12\\ '],
+    [' foo12', 'bar12 ', '\\ foo12=bar12 '],
     ['#foo13', 'bar13', '\\#foo13=bar13'],
     ['!foo14#', 'bar14', '\\!foo14\\#=bar14'],
-    ['foo15', '#bar15', 'foo15=#bar15'],
+    ['foo15', '#bar15', 'foo15=\\#bar15'],
     ['f o  o18', ' bar18', 'f\\ o\\ \\ o18=\\ bar18'],
     ['foo19\n', 'bar\t\f\r19\n', 'foo19\\n=bar\\t\\f\\r19\\n'],
     ['foo20', '', 'foo20='],
@@ -252,3 +252,50 @@ describe('data access', () => {
     ])
   })
 })
+
+
+describe('The property key escaping', () => {
+  it.each([
+    ['foo1', 'foo1'],
+    ['foo2:', 'foo2\\:'],
+    ['foo3=', 'foo3\\='],
+    ['foo4\t', 'foo4\\t'],
+    ['foo5 ', 'foo5\\ '],
+    [' foo6', '\\ foo6'],
+    ['#foo7', '\\#foo7'],
+    ['!foo8#', '\\!foo8\\#'],
+    ['fo  o9', 'fo\\ \\ o9'],
+    ['foo10\n', 'foo10\\n'],
+    ['f\r\f\n\too11', 'f\\r\\f\\n\\too11'],
+    ['\\foo12\\', '\\\\foo12\\\\'],
+    ['\0\u0001', '\\u0000\\u0001'],
+    ['\u3053\u3093\u306B\u3061\u306F', '\\u3053\\u3093\\u306b\\u3061\\u306f'],
+    ['こんにちは', '\\u3053\\u3093\\u306b\\u3061\\u306f'],
+  ])('should escape key "%s" as "%s"', (key: string, expected: string) => {
+    const result = properties.escapeKey(key)
+    expect(result).toEqual(expected)
+  })
+})
+
+describe('The property value escaping', () => {
+  it.each([
+    ['foo1', 'foo1'],
+    ['foo2:', 'foo2\\:'],
+    ['foo3=', 'foo3\\='],
+    ['foo4\t', 'foo4\\t'],
+    ['foo5 ', 'foo5 '],
+    [' foo6', '\\ foo6'],
+    ['#foo7', '\\#foo7'],
+    ['!foo8#', '\\!foo8\\#'],
+    ['fo  o9', 'fo  o9'],
+    ['foo10\n', 'foo10\\n'],
+    ['f\r\f\n\too11', 'f\\r\\f\\n\\too11'],
+    ['\\foo12\\', '\\\\foo12\\\\'],
+    ['\0\u0001', '\\u0000\\u0001'],
+    ['\u3053\u3093\u306B\u3061\u306F', '\\u3053\\u3093\\u306b\\u3061\\u306f'],
+    ['こんにちは', '\\u3053\\u3093\\u306b\\u3061\\u306f'],
+  ])('should escape value "%s" as "%s"', (key: string, expected: string) => {
+    const result = properties.escapeValue(key)
+    expect(result).toEqual(expected)
+  })
+})

src/properties.ts

@@ -77,7 +77,7 @@ export const stringify = (config: Properties): string => {
  */
 export function* list(config: Properties): Generator<KeyValuePair> {
   for (const {key, rawValue} of listPairs(config.lines)) {
-    yield {key, value: unescapeValue(rawValue)}
+    yield {key, value: unescape(rawValue)}
   }
 }
 
@@ -91,7 +91,7 @@ export function* list(config: Properties): Generator<KeyValuePair> {
 export const get = (config: Properties, key: string): string | undefined => {
   // Find existing
   const {rawValue} = findValue(config.lines, key)
-  return typeof rawValue === 'string' ? unescapeValue(rawValue) : undefined
+  return typeof rawValue === 'string' ? unescape(rawValue) : undefined
 }
 
 /**
@@ -105,7 +105,7 @@ export const toMap = (config: Properties): Map<string, string> => {
   const result = new Map<string, string>()
 
   for (const {key, rawValue} of listPairs(config.lines)) {
-    result.set(key, unescapeValue(rawValue))
+    result.set(key, unescape(rawValue))
   }
 
   return result
@@ -308,52 +308,105 @@ const unescapeChar = (c: string): string => {
   }
 }
 
-const unescapeValue = (str: string): string =>
+/**
+ * Unescape key or value.
+ *
+ * @param str Escaped string.
+ * @return Actual string.
+ */
+export const unescape = (str: string): string =>
   str.replace(/\\(.)/g, s => unescapeChar(s[1]))
 
-// Very simple implementation, does not handle unicode etc
-const escapeValue = (str: string, escapeChars = ''): string => {
+/**
+ * Escape property key.
+ *
+ * @param unescapedKey Property key to be escaped.
+ * @param escapeUnicode Escape unicode chars (below 0x0020 and above 0x007e). Default is true.
+ * @return Escaped string.
+ */
+export const escapeKey = (unescapedKey: string, escapeUnicode = true): string => {
+  return escape(unescapedKey, true, escapeUnicode)
+}
+
+/**
+ * Escape property value.
+ *
+ * @param unescapedValue Property value to be escaped.
+ * @param escapeUnicode Escape unicode chars (below 0x0020 and above 0x007e). Default is true.
+ * @return Escaped string.
+ */
+export const escapeValue = (unescapedValue: string, escapeUnicode = true): string => {
+  return escape(unescapedValue, false, escapeUnicode)
+}
+
+/**
+ * Internal escape method.
+ *
+ * @param unescapedContent Text to be escaped.
+ * @param escapeSpace Whether all spaces should be escaped
+ * @param escapeUnicode Whether unicode chars should be escaped
+ * @return Escaped string.
+ */
+const escape = (
+  unescapedContent: string,
+  escapeSpace: boolean,
+  escapeUnicode: boolean
+): string => {
   const result: string[] = []
-  let escapeNext = str.startsWith(' ') // always escape space at beginning
 
-  for (let index = 0; index < str.length; index++) {
-    const char = str[index]
+  // eslint-disable-next-line unicorn/no-for-loop
+  for (let index = 0; index < unescapedContent.length; index++) {
+    const char = unescapedContent[index]
     switch (char) {
+      case ' ': {
+        // Escape space if required, or if it is first character
+        if (escapeSpace || index === 0) {
+          result.push('\\ ')
+        } else {
+          result.push(' ')
+        }
+        break
+      }
       case '\\': {
         result.push('\\\\')
         break
       }
       case '\f': {
-        // Formfeed/
+        // Form-feed
         result.push('\\f')
         break
       }
       case '\n': {
-        // Newline.
+        // Newline
         result.push('\\n')
         break
       }
       case '\r': {
-        // Carriage return.
+        // Carriage return
         result.push('\\r')
         break
       }
       case '\t': {
-        // Tab.
+        // Tab
         result.push('\\t')
         break
       }
+      case '=': // Fall through
+      case ':': // Fall through
+      case '#': // Fall through
+      case '!': {
+        result.push('\\', char)
+        break
+      }
       default: {
-        // Escape trailing space
-        if (index === str.length - 1 && char === ' ') {
-          escapeNext = true
+        if (escapeUnicode) {
+          const codePoint: number = char.codePointAt(0) as number // can never be undefined
+          if (codePoint < 0x0020 || codePoint > 0x007e) {
+            result.push('\\u', codePoint.toString(16).padStart(4, '0'))
+            break
+          }
         }
-        // Escape if required
-        if (escapeNext || escapeChars.includes(char)) {
-          result.push('\\')
-          escapeNext = false
-        }
-
+        // Normal char
         result.push(char)
         break
       }
@@ -362,5 +415,3 @@ const escapeValue = (str: string, escapeChars = ''): string => {
 
   return result.join('')
 }
-
-const escapeKey = (str: string): string => escapeValue(str, ' #!:=')