builder/lib/processors/nonAsciiEscaper.js

  1. import escapeUnicode from "escape-unicode";
  3. /**
  4.  * @public
  5.  * @module @ui5/builder/processors/nonAsciiEscaper
  6.  */
  8. /**
  9.  * @see https://en.wikipedia.org/wiki/ASCII
  10.  * ascii contains 128 characters.
  11.  * its char codes reach from 0 to 127.
  12.  * @type {number}
  13.  */
  14. const CHAR_CODE_OF_LAST_ASCII_CHARACTER = 127;
  16. // use memoization for escapeUnicode function for performance
  17. const memoizeEscapeUnicodeMap = Object.create(null);
  18. const memoizeEscapeUnicode = function(sChar) {
  19. 	if (memoizeEscapeUnicodeMap[sChar]) {
  20. 		return memoizeEscapeUnicodeMap[sChar];
  21. 	}
  22. 	memoizeEscapeUnicodeMap[sChar] = escapeUnicode(sChar);
  23. 	return memoizeEscapeUnicodeMap[sChar];
  24. };
  26. /**
  27.  * Escapes non ASCII characters with unicode escape sequences.
  28.  *
  29.  * @see https://en.wikipedia.org/wiki/ASCII
  30.  * @see https://tools.ietf.org/html/rfc5137#section-6.1
  31.  *
  32.  *
  33.  * @param {string} string input string with non ascii characters, e.g. L♥VE
  34.  * @returns {{string: (string), modified: boolean}} output string with all non ascii
  35.  * characters being escaped by unicode sequence, e.g. L\u2665VE
  36.  */
  37. const escapeNonAscii = function(string) {
  38. 	let result = "";
  39. 	let modified = false;
  40. 	for (let i = 0; i < string.length; i++) {
  41. 		const char = string[i];
  42. 		// check for non ascii characters (characters which have a char code
  43. 		// greater than the ascii character code range)
  44. 		if (string.charCodeAt(i) > CHAR_CODE_OF_LAST_ASCII_CHARACTER) {
  45. 			result += memoizeEscapeUnicode(char);
  46. 			modified = true;
  47. 		} else {
  48. 			result += char;
  49. 		}
  50. 	}
  51. 	return {
  52. 		modified,
  53. 		string: result
  54. 	};
  55. };
  57. /**
  58.  * Escapes non ASCII characters with unicode escape sequences.
  59.  *
  60.  * @example
  61.  * const encoding = nonAsciiEscaper.getEncodingFromAlias("ISO-8859-1");
  62.  * nonAsciiEscaper({resources, options: {encoding}});
  63.  *
  64.  *
  65.  * @public
  66.  * @function default
  67.  * @static
  68.  *
  69.  * @param {object} parameters Parameters
  70.  * @param {@ui5/fs/Resource[]} parameters.resources List of resources to be processed
  71.  * @param {object} [parameters.options] Options
  72.  * @param {string} [parameters.options.encoding="utf8"] resource file encoding
  73.  *   ({@link https://nodejs.org/api/buffer.html#buffer_buffers_and_character_encodings Node.js character encodings}).
  74.  *   Use #getEncodingFromAlias to get the encoding string
  75.  * @returns {Promise<@ui5/fs/Resource[]>} Promise resolving with the processed resources
  76.  */
  77. async function nonAsciiEscaper({resources, options: {encoding}}) {
  78. 	encoding = encoding || "utf8";
  80. 	async function processResource(resource) {
  81. 		const resourceString = (await resource.getBuffer()).toString(encoding);
  82. 		const escaped = escapeNonAscii(resourceString);
  83. 		// only modify the resource's string if it was changed
  84. 		if (escaped.modified) {
  85. 			resource.setString(escaped.string);
  86. 		}
  87. 		return resource;
  88. 	}
  90. 	return Promise.all(resources.map(processResource));
  91. }
  93. const encodingMap = {
  94. 	"UTF-8": "utf8",
  95. 	"ISO-8859-1": "latin1",
  96. };
  98. /**
  99.  * Provides a mapping from user-friendly encoding name (alias) such as "UTF-8" and "ISO-8859-1" to node
  100.  * specific encoding name such as "utf8" or "latin1". Simplifies usage of nonAsciiEscaper encoding option
  101.  * such that it can be used standalone without the respective task (e.g. in Splitter, Bundler and related projects).
  102.  *
  103.  * @public
  104.  * @function getEncodingFromAlias
  105.  * @alias @ui5/builder/processors/nonAsciiEscaper․getEncodingFromAlias
  106.  * @static
  107.  *
  108.  * @param {string} encoding encoding labels: "UTF-8" and "ISO-8859-1"
  109.  * @returns {string} node.js character encoding string, e.g. utf8 and latin1
  110.  */
  111. nonAsciiEscaper.getEncodingFromAlias = function(encoding) {
  112. 	if (!encodingMap[encoding]) {
  113. 		throw new Error(
  114. 			`Encoding "${encoding}" is not supported. Only ${Object.keys(encodingMap).join(", ")} are allowed values` );
  115. 	}
  116. 	return encodingMap[encoding];
  117. };
  119. export default nonAsciiEscaper;