blockly/core/utils/string.js

/**
 * @license
 * Copyright 2019 Google LLC
 * SPDX-License-Identifier: Apache-2.0
 */

/**
 * @fileoverview Utility methods for string manipulation.
 * These methods are not specific to Blockly, and could be factored out into
 * a JavaScript framework such as Closure.
 * @author fraser@google.com (Neil Fraser)
 */
'use strict';

/**
 * @name Blockly.utils.string
 * @namespace
 */
goog.provide('Blockly.utils.string');


/**
 * Fast prefix-checker.
 * Copied from Closure's goog.string.startsWith.
 * @param {string} str The string to check.
 * @param {string} prefix A string to look for at the start of `str`.
 * @return {boolean} True if `str` begins with `prefix`.
 */
Blockly.utils.string.startsWith = function(str, prefix) {
  return str.lastIndexOf(prefix, 0) == 0;
};

/**
 * Given an array of strings, return the length of the shortest one.
 * @param {!Array.<string>} array Array of strings.
 * @return {number} Length of shortest string.
 */
Blockly.utils.string.shortestStringLength = function(array) {
  if (!array.length) {
    return 0;
  }
  return array.reduce(function(a, b) {
    return a.length < b.length ? a : b;
  }).length;
};

/**
 * Given an array of strings, return the length of the common prefix.
 * Words may not be split.  Any space after a word is included in the length.
 * @param {!Array.<string>} array Array of strings.
 * @param {number=} opt_shortest Length of shortest string.
 * @return {number} Length of common prefix.
 */
Blockly.utils.string.commonWordPrefix = function(array, opt_shortest) {
  if (!array.length) {
    return 0;
  } else if (array.length == 1) {
    return array[0].length;
  }
  var wordPrefix = 0;
  var max = opt_shortest || Blockly.utils.string.shortestStringLength(array);
  for (var len = 0; len < max; len++) {
    var letter = array[0][len];
    for (var i = 1; i < array.length; i++) {
      if (letter != array[i][len]) {
        return wordPrefix;
      }
    }
    if (letter == ' ') {
      wordPrefix = len + 1;
    }
  }
  for (var i = 1; i < array.length; i++) {
    var letter = array[i][len];
    if (letter && letter != ' ') {
      return wordPrefix;
    }
  }
  return max;
};

/**
 * Given an array of strings, return the length of the common suffix.
 * Words may not be split.  Any space after a word is included in the length.
 * @param {!Array.<string>} array Array of strings.
 * @param {number=} opt_shortest Length of shortest string.
 * @return {number} Length of common suffix.
 */
Blockly.utils.string.commonWordSuffix = function(array, opt_shortest) {
  if (!array.length) {
    return 0;
  } else if (array.length == 1) {
    return array[0].length;
  }
  var wordPrefix = 0;
  var max = opt_shortest || Blockly.utils.string.shortestStringLength(array);
  for (var len = 0; len < max; len++) {
    var letter = array[0].substr(-len - 1, 1);
    for (var i = 1; i < array.length; i++) {
      if (letter != array[i].substr(-len - 1, 1)) {
        return wordPrefix;
      }
    }
    if (letter == ' ') {
      wordPrefix = len + 1;
    }
  }
  for (var i = 1; i < array.length; i++) {
    var letter = array[i].charAt(array[i].length - len - 1);
    if (letter && letter != ' ') {
      return wordPrefix;
    }
  }
  return max;
};

/**
 * Wrap text to the specified width.
 * @param {string} text Text to wrap.
 * @param {number} limit Width to wrap each line.
 * @return {string} Wrapped text.
 */
Blockly.utils.string.wrap = function(text, limit) {
  var lines = text.split('\n');
  for (var i = 0; i < lines.length; i++) {
    lines[i] = Blockly.utils.string.wrapLine_(lines[i], limit);
  }
  return lines.join('\n');
};

/**
 * Wrap single line of text to the specified width.
 * @param {string} text Text to wrap.
 * @param {number} limit Width to wrap each line.
 * @return {string} Wrapped text.
 * @private
 */
Blockly.utils.string.wrapLine_ = function(text, limit) {
  if (text.length <= limit) {
    // Short text, no need to wrap.
    return text;
  }
  // Split the text into words.
  var words = text.trim().split(/\s+/);
  // Set limit to be the length of the largest word.
  for (var i = 0; i < words.length; i++) {
    if (words[i].length > limit) {
      limit = words[i].length;
    }
  }

  var lastScore;
  var score = -Infinity;
  var lastText;
  var lineCount = 1;
  do {
    lastScore = score;
    lastText = text;
    // Create a list of booleans representing if a space (false) or
    // a break (true) appears after each word.
    var wordBreaks = [];
    // Seed the list with evenly spaced linebreaks.
    var steps = words.length / lineCount;
    var insertedBreaks = 1;
    for (var i = 0; i < words.length - 1; i++) {
      if (insertedBreaks < (i + 1.5) / steps) {
        insertedBreaks++;
        wordBreaks[i] = true;
      } else {
        wordBreaks[i] = false;
      }
    }
    wordBreaks = Blockly.utils.string.wrapMutate_(words, wordBreaks, limit);
    score = Blockly.utils.string.wrapScore_(words, wordBreaks, limit);
    text = Blockly.utils.string.wrapToText_(words, wordBreaks);
    lineCount++;
  } while (score > lastScore);
  return lastText;
};

/**
 * Compute a score for how good the wrapping is.
 * @param {!Array.<string>} words Array of each word.
 * @param {!Array.<boolean>} wordBreaks Array of line breaks.
 * @param {number} limit Width to wrap each line.
 * @return {number} Larger the better.
 * @private
 */
Blockly.utils.string.wrapScore_ = function(words, wordBreaks, limit) {
  // If this function becomes a performance liability, add caching.
  // Compute the length of each line.
  var lineLengths = [0];
  var linePunctuation = [];
  for (var i = 0; i < words.length; i++) {
    lineLengths[lineLengths.length - 1] += words[i].length;
    if (wordBreaks[i] === true) {
      lineLengths.push(0);
      linePunctuation.push(words[i].charAt(words[i].length - 1));
    } else if (wordBreaks[i] === false) {
      lineLengths[lineLengths.length - 1]++;
    }
  }
  var maxLength = Math.max.apply(Math, lineLengths);

  var score = 0;
  for (var i = 0; i < lineLengths.length; i++) {
    // Optimize for width.
    // -2 points per char over limit (scaled to the power of 1.5).
    score -= Math.pow(Math.abs(limit - lineLengths[i]), 1.5) * 2;
    // Optimize for even lines.
    // -1 point per char smaller than max (scaled to the power of 1.5).
    score -= Math.pow(maxLength - lineLengths[i], 1.5);
    // Optimize for structure.
    // Add score to line endings after punctuation.
    if ('.?!'.indexOf(linePunctuation[i]) != -1) {
      score += limit / 3;
    } else if (',;)]}'.indexOf(linePunctuation[i]) != -1) {
      score += limit / 4;
    }
  }
  // All else being equal, the last line should not be longer than the
  // previous line.  For example, this looks wrong:
  // aaa bbb
  // ccc ddd eee
  if (lineLengths.length > 1 && lineLengths[lineLengths.length - 1] <=
      lineLengths[lineLengths.length - 2]) {
    score += 0.5;
  }
  return score;
};

/**
 * Mutate the array of line break locations until an optimal solution is found.
 * No line breaks are added or deleted, they are simply moved around.
 * @param {!Array.<string>} words Array of each word.
 * @param {!Array.<boolean>} wordBreaks Array of line breaks.
 * @param {number} limit Width to wrap each line.
 * @return {!Array.<boolean>} New array of optimal line breaks.
 * @private
 */
Blockly.utils.string.wrapMutate_ = function(words, wordBreaks, limit) {
  var bestScore = Blockly.utils.string.wrapScore_(words, wordBreaks, limit);
  var bestBreaks;
  // Try shifting every line break forward or backward.
  for (var i = 0; i < wordBreaks.length - 1; i++) {
    if (wordBreaks[i] == wordBreaks[i + 1]) {
      continue;
    }
    var mutatedWordBreaks = [].concat(wordBreaks);
    mutatedWordBreaks[i] = !mutatedWordBreaks[i];
    mutatedWordBreaks[i + 1] = !mutatedWordBreaks[i + 1];
    var mutatedScore =
        Blockly.utils.string.wrapScore_(words, mutatedWordBreaks, limit);
    if (mutatedScore > bestScore) {
      bestScore = mutatedScore;
      bestBreaks = mutatedWordBreaks;
    }
  }
  if (bestBreaks) {
    // Found an improvement.  See if it may be improved further.
    return Blockly.utils.string.wrapMutate_(words, bestBreaks, limit);
  }
  // No improvements found.  Done.
  return wordBreaks;
};

/**
 * Reassemble the array of words into text, with the specified line breaks.
 * @param {!Array.<string>} words Array of each word.
 * @param {!Array.<boolean>} wordBreaks Array of line breaks.
 * @return {string} Plain text.
 * @private
 */
Blockly.utils.string.wrapToText_ = function(words, wordBreaks) {
  var text = [];
  for (var i = 0; i < words.length; i++) {
    text.push(words[i]);
    if (wordBreaks[i] !== undefined) {
      text.push(wordBreaks[i] ? '\n' : ' ');
    }
  }
  return text.join('');
};