suport bash's ANSI-C quoted strings

yargs · Mar 15, 2021 · c1420f7 · c1420f7
1 parent 6c60415
commit c1420f7
Show file tree

Hide file tree

Showing 2 changed files with 88 additions and 5 deletions.
diff --git a/lib/yargs-parser.ts b/lib/yargs-parser.ts
@@ -600,11 +600,14 @@ export class YargsParser {
 
     function processValue (key: string, val: any) {
       // strings may be quoted, clean this up as we assign values.
-      if (typeof val === 'string' &&
-        (val[0] === "'" || val[0] === '"') &&
-        val[val.length - 1] === val[0]
-      ) {
-        val = val.substring(1, val.length - 1)
+      if (typeof val === 'string') {
+        if ((val[0] === "'" || val[0] === '"') &&
+          val[val.length - 1] === val[0]
+        ) {
+          val = val.substring(1, val.length - 1)
+        } else if (val.slice(0, 2) === "$'" && val[val.length - 1] === "'") {
+          val = parseAnsiCQuote(val)
+        }
       }
 
       // handle parsing boolean arguments --foo=true --bar false.
@@ -629,6 +632,59 @@ export class YargsParser {
       return value
     }
 
+    // ANSI-C quoted string are a bash-only feature and have the form $'some text'
+    // https://www.gnu.org/software/bash/manual/html_node/ANSI_002dC-Quoting.html
+    function parseAnsiCQuote (str: string): string {
+      function unescapeChar (x: string): string {
+        switch (x.slice(0, 2)) {
+          case '\\\\':
+            return '\\'
+          case '\\a':
+            return '\a' // eslint-disable-line
+          case '\\b':
+            return '\b'
+          case '\\e':
+            return '\u001b'
+          case '\\E':
+            return '\u001b'
+          case '\\f':
+            return '\f'
+          case '\\n':
+            return '\n'
+          case '\\r':
+            return '\r'
+          case '\\t':
+            return '\t'
+          case '\\v':
+            return '\v'
+          case "\\'":
+            return "'"
+          case '\\"':
+            return '"'
+          case '\\?':
+            return '?'
+          case '\\c':
+            // Control codes
+            // "\c1" -> 11, "\c2" -> 12 and so on
+            if (x.match(/\\c[0-9]/)) {
+              return String.fromCharCode(parseInt(x.slice(2), 10) + 16)
+            }
+            // "\ca" -> 01, "\cb" -> 02 and so on
+            return String.fromCharCode(x.toLowerCase().charCodeAt(2) - 'a'.charCodeAt(0) + 1)
+          case '\\x':
+          case '\\u':
+          case '\\U':
+            // Hexadecimal character literal
+            return String.fromCharCode(parseInt(x.slice(2), 16))
+        }
+        // Octal character literal
+        return String.fromCharCode(parseInt(x.slice(1), 8) % 256)
+      }
+
+      const ANSI_BACKSLASHES = /\\(\\|a|b|e|E|f|n|r|t|v|'|"|\?|[0-7]{1,3}|x[0-9A-Fa-f]{1,2}|u[0-9A-Fa-f]{1,4}|U[0-9A-Fa-f]{1,8}|c[0-9a-zA-Z])/g
+      return str.slice(2, -1).replace(ANSI_BACKSLASHES, unescapeChar)
+    }
+
     function maybeCoerceNumber (key: string, value: string | number | null | undefined) {
       if (!configuration['parse-positional-numbers'] && key === '_') return value
       if (!checkAllAliases(key, flags.strings) && !checkAllAliases(key, flags.bools) && !Array.isArray(value)) {

diff --git a/test/yargs-parser.cjs b/test/yargs-parser.cjs
@@ -3587,6 +3587,33 @@ describe('yargs-parser', function () {
       args.foo.should.equal('-hello world')
       args.bar.should.equal('--goodnight moon')
     })
+
+    it('handles bash ANSI-C quoted strings', () => {
+      const args = parser("--foo $'text with \\n newline'")
+      args.foo.should.equal('text with \n newline')
+
+      // Double quotes shouldn't work
+      const args2 = parser('--foo $"text without \\n newline"')
+      args2.foo.should.equal('$"text without \\n newline"')
+
+      const characters = '\\\\' + '\\a' + '\\b' + '\\e' + '\\E' + '\\f' + '\\n' + '\\r' + '\\t' + '\\v' + "\\'" + '\\"' + '\\?'
+      const args3 = parser("--foo $'" + characters + "'")
+      args3.foo.should.equal('\\\a\b\u001b\u001b\f\n\r\t\v\'"?') // eslint-disable-line
+
+      const args4 = parser("--foo $'text \\xFFFF with \\xFF hex'")
+      args4.foo.should.equal('text \u00FFFF with \u00FF hex')
+      const args5 = parser("--foo $'text \\uFFFFFF\\uFFFF with \\uFF hex'")
+      args5.foo.should.equal('text \uFFFFFF\uFFFF with \u00FF hex')
+      const args6 = parser("--foo $'text \\UFFFFFF\\UFFFF with \\U00FF hex'")
+      const longCodePoint = String.fromCharCode(parseInt('FFFFFF', 16))
+      args6.foo.should.equal(`text ${longCodePoint}\uFFFF with \u00FF hex`)
+
+      const args7 = parser("--foo $'text \\cAB \\cz with \\c12 control \\c011 chars'")
+      args7.foo.should.equal('text \u0001B \u001A with \u00112 control \u001011 chars')
+
+      const args8 = parser("--foo $'text \\0 \\001 with \\12 \\123 \\129 octal'")
+      args8.foo.should.equal('text \u0000 \u0001 with \u000A \u0053 \u000A9 octal')
+    })
   })
 
   // see: https://github.com/yargs/yargs-parser/issues/144