builder/lib/processors/nonAsciiEscaper.js

  1. const escapeUnicode = require("escape-unicode");
  3. /**
  4.  * @see https://en.wikipedia.org/wiki/ASCII
  5.  * ascii contains 128 characters.
  6.  * its char codes reach from 0 to 127.
  7.  * @type {number}
  8.  */
  9. const CHAR_CODE_OF_LAST_ASCII_CHARACTER = 127;
  11. // use memoization for escapeUnicode function for performance
  12. const memoizeEscapeUnicodeMap = {};
  13. const memoizeEscapeUnicode = function(sChar) {
  14. 	if (memoizeEscapeUnicodeMap[sChar]) {
  15. 		return memoizeEscapeUnicodeMap[sChar];
  16. 	}
  17. 	memoizeEscapeUnicodeMap[sChar] = escapeUnicode(sChar);
  18. 	return memoizeEscapeUnicodeMap[sChar];
  19. };
  21. /**
  22.  * Escapes non ASCII characters with unicode escape sequences.
  23.  *
  24.  * @see https://en.wikipedia.org/wiki/ASCII
  25.  * @see https://tools.ietf.org/html/rfc5137#section-6.1
  26.  *
  27.  *
  28.  * @param {string} string input string with non ascii characters, e.g. L♥VE
  29.  * @returns {{string: (string), modified: boolean}} output string with all non ascii
  30.  * characters being escaped by unicode sequence, e.g. L\u2665VE
  31.  */
  32. const escapeNonAscii = function(string) {
  33. 	let result = "";
  34. 	let modified = false;
  35. 	for (let i = 0; i < string.length; i++) {
  36. 		const char = string[i];
  37. 		// check for non ascii characters (characters which have a char code
  38. 		// greater than the ascii character code range)
  39. 		if (string.charCodeAt(i) > CHAR_CODE_OF_LAST_ASCII_CHARACTER) {
  40. 			result += memoizeEscapeUnicode(char);
  41. 			modified = true;
  42. 		} else {
  43. 			result += char;
  44. 		}
  45. 	}
  46. 	return {
  47. 		modified,
  48. 		string: result
  49. 	};
  50. };
  52. /**
  53.  * Escapes non ASCII characters with unicode escape sequences.
  54.  *
  55.  * @example
  56.  * const encoding = nonAsciiEscaper.getEncodingFromAlias("ISO-8859-1");
  57.  * nonAsciiEscaper({resources, options: {encoding}});
  58.  *
  59.  *
  60.  * @public
  61.  * @alias module:@ui5/builder.processors.nonAsciiEscaper
  62.  * @param {object} parameters Parameters
  63.  * @param {module:@ui5/fs.Resource[]} parameters.resources List of resources to be processed
  64.  * @param {object} [parameters.options] Options
  65.  * @param {string} [parameters.options.encoding="utf8"] resource file encoding
  66.  *   ({@link https://nodejs.org/api/buffer.html#buffer_buffers_and_character_encodings Node.js character encodings}).
  67.  *   Use #getEncodingFromAlias to get the encoding string
  68.  * @returns {Promise<module:@ui5/fs.Resource[]>} Promise resolving with the processed resources
  69.  */
  70. async function nonAsciiEscaper({resources, options: {encoding}}) {
  71. 	encoding = encoding || "utf8";
  73. 	async function processResource(resource) {
  74. 		const resourceString = (await resource.getBuffer()).toString(encoding);
  75. 		const escaped = escapeNonAscii(resourceString);
  76. 		// only modify the resource's string if it was changed
  77. 		if (escaped.modified) {
  78. 			resource.setString(escaped.string);
  79. 		}
  80. 		return resource;
  81. 	}
  83. 	return Promise.all(resources.map(processResource));
  84. }
  86. const encodingMap = {
  87. 	"UTF-8": "utf8",
  88. 	"ISO-8859-1": "latin1",
  89. };
  91. /**
  92.  * Provides a mapping from user-friendly encoding name (alias) such as "UTF-8" and "ISO-8859-1" to node
  93.  * specific encoding name such as "utf8" or "latin1". Simplifies usage of nonAsciiEscaper encoding option
  94.  * such that it can be used standalone without the respective task (e.g. in Splitter, Bundler and related projects).
  95.  *
  96.  * @public
  97.  * @alias module:@ui5/builder.processors.nonAsciiEscaper․getEncodingFromAlias
  98.  * @param {string} encoding encoding labels: "UTF-8" and "ISO-8859-1"
  99.  * @returns {string} node.js character encoding string, e.g. utf8 and latin1
  100.  */
  101. nonAsciiEscaper.getEncodingFromAlias = function(encoding) {
  102. 	if (!encodingMap[encoding]) {
  103. 		throw new Error(
  104. 			`Encoding "${encoding}" is not supported. Only ${Object.keys(encodingMap).join(", ")} are allowed values` );
  105. 	}
  106. 	return encodingMap[encoding];
  107. };
  109. module.exports = nonAsciiEscaper;