From 10f7bbf007c8cf79248d3dfd4f49b5a305bbec34 Mon Sep 17 00:00:00 2001 From: Timo Tijhof Date: Mon, 13 Nov 2023 00:27:06 +0000 Subject: [PATCH] [WIP] Core: Update diff library, integrate with TapReporter Ref https://github.com/qunitjs/qunit/issues/1412. --- src/core/core.js | 4 +- src/core/diff.js | 3098 +++++++++++++++++++--------- src/core/reporters/HtmlReporter.js | 16 +- src/core/reporters/TapReporter.js | 29 +- test/main/TapReporter.js | 35 +- test/main/diff.js | 15 +- 6 files changed, 2186 insertions(+), 1011 deletions(-) diff --git a/src/core/core.js b/src/core/core.js index e7a6a9ee2..5833f4b08 100644 --- a/src/core/core.js +++ b/src/core/core.js @@ -14,7 +14,7 @@ import ProcessingQueue from './processing-queue.js'; import { urlParams } from './urlparams.js'; import { on } from './events.js'; import onUncaughtException from './on-uncaught-exception.js'; -import diff from './diff.js'; +import { diffHtml } from './diff.js'; import version from './version.js'; import { createStartFunction } from './start.js'; @@ -40,7 +40,7 @@ const QUnit = { stack, urlParams, - diff, + diff: diffHtml, dump, equiv, reporters, diff --git a/src/core/diff.js b/src/core/diff.js index a9e3bee04..351430c97 100644 --- a/src/core/diff.js +++ b/src/core/diff.js @@ -1,1122 +1,2250 @@ -/* eslint-disable indent */ -import { escapeText } from './utilities.js'; - -/* - * This file is a modified version of google-diff-match-patch's JavaScript implementation - * (https://code.google.com/p/google-diff-match-patch/source/browse/trunk/javascript/diff_match_patch_uncompressed.js), - * modifications are licensed as more fully set forth in LICENSE.txt. +/* eslint-disable camelcase, eqeqeq, new-cap, no-var, no-redeclare */ +/** + * Diff Match and Patch. https://github.com/google/diff-match-patch/tree/62f2e689f4/ * - * The original source of google-diff-match-patch is attributable and licensed as follows: + * Modifications for QUnit: + * - diff_prettyHtml: Remove "para" entity and "
". + * - diff_prettyHtml: Remove inline styles. + * - diff_prettyHtml: Remove redundant escapes. + * - diff_linesToCharsMunge_: Avoid unsafe hasOwnProperty call. + * - Error: Change "Null input" to "Cannot diff null input". * - * Copyright 2006 Google Inc. - * https://code.google.com/p/google-diff-match-patch/ + * Copyright 2018 Neil Fraser + * Copyright 2018 The diff-match-patch Authors, https://github.com/google/diff-match-patch * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * - * https://www.apache.org/licenses/LICENSE-2.0 + * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. - * - * More Info: - * https://code.google.com/p/google-diff-match-patch/ - * - * Usage: QUnit.diff(expected, actual) - * */ - function DiffMatchPatch () { - } - // DIFF FUNCTIONS +/** + * @fileoverview Computes the difference between two texts to create a patch. + * Applies the patch onto another text, allowing for errors. + * @author fraser@google.com (Neil Fraser) + */ - /** - * The data structure representing a diff is an array of tuples: - * [[DIFF_DELETE, 'Hello'], [DIFF_INSERT, 'Goodbye'], [DIFF_EQUAL, ' world.']] - * which means: delete 'Hello', add 'Goodbye' and keep ' world.' - */ - const DIFF_DELETE = -1; - const DIFF_INSERT = 1; - const DIFF_EQUAL = 0; - const hasOwn = Object.prototype.hasOwnProperty; +/** + * Class containing the diff, match and patch methods. + * @constructor + */ +var diff_match_patch = function () { + // Defaults. + // Redefine these in your program to override the defaults. + + // Number of seconds to map a diff before giving up (0 for infinity). + this.Diff_Timeout = 1.0; + // Cost of an empty edit operation in terms of edit characters. + this.Diff_EditCost = 4; + // At what point is no match declared (0.0 = perfection, 1.0 = very loose). + this.Match_Threshold = 0.5; + // How far to search for a match (0 = exact location, 1000+ = broad match). + // A match this many characters away from the expected location will add + // 1.0 to the score (0.0 is a perfect match). + this.Match_Distance = 1000; + // When deleting a large block of text (over ~64 characters), how close do + // the contents have to be to match the expected contents. (0.0 = perfection, + // 1.0 = very loose). Note that Match_Threshold controls how closely the + // end points of a delete need to match. + this.Patch_DeleteThreshold = 0.5; + // Chunk size for context length. + this.Patch_Margin = 4; + + // The number of bits in an int. + this.Match_MaxBits = 32; +}; + +// DIFF FUNCTIONS + +/** + * The data structure representing a diff is an array of tuples: + * [[DIFF_DELETE, 'Hello'], [DIFF_INSERT, 'Goodbye'], [DIFF_EQUAL, ' world.']] + * which means: delete 'Hello', add 'Goodbye' and keep ' world.' + */ +var DIFF_DELETE = -1; +var DIFF_INSERT = 1; +var DIFF_EQUAL = 0; + +/** + * Class representing one diff tuple. + * Attempts to look like a two-element array (which is what this used to be). + * @param {number} op Operation, one of: DIFF_DELETE, DIFF_INSERT, DIFF_EQUAL. + * @param {string} text Text to be deleted, inserted, or retained. + * @constructor + */ +diff_match_patch.Diff = function (op, text) { + this[0] = op; + this[1] = text; +}; - /** - * Find the differences between two texts. Simplifies the problem by stripping - * any common prefix or suffix off the texts before diffing. - * @param {string} text1 Old string to be diffed. - * @param {string} text2 New string to be diffed. - * @param {boolean=} optChecklines Optional speedup flag. If present and false, - * then don't run a line-level diff first to identify the changed areas. - * Defaults to true, which does a faster, slightly less optimal diff. - * @return {!Array.} Array of diff tuples. - */ - DiffMatchPatch.prototype.DiffMain = function (text1, text2, optChecklines) { - // The diff must be complete in up to 1 second. - let deadline = Date.now() + 1000; +diff_match_patch.Diff.prototype.length = 2; - // Check for null inputs. - if (text1 === null || text2 === null) { - throw new Error('Cannot diff null input.'); +/** + * Emulate the output of a two-element array. + * @return {string} Diff operation as a string. + */ +diff_match_patch.Diff.prototype.toString = function () { + return this[0] + ',' + this[1]; +}; + +/** + * Find the differences between two texts. Simplifies the problem by stripping + * any common prefix or suffix off the texts before diffing. + * @param {string} text1 Old string to be diffed. + * @param {string} text2 New string to be diffed. + * @param {boolean=} opt_checklines Optional speedup flag. If present and false, + * then don't run a line-level diff first to identify the changed areas. + * Defaults to true, which does a faster, slightly less optimal diff. + * @param {number=} opt_deadline Optional time when the diff should be complete + * by. Used internally for recursive calls. Users should set DiffTimeout + * instead. + * @return {!Array.} Array of diff tuples. + */ +diff_match_patch.prototype.diff_main = function (text1, text2, opt_checklines, + opt_deadline) { + // Set a deadline by which time the diff must be complete. + if (typeof opt_deadline == 'undefined') { + if (this.Diff_Timeout <= 0) { + opt_deadline = Number.MAX_VALUE; + } else { + opt_deadline = (new Date()).getTime() + this.Diff_Timeout * 1000; } + } + var deadline = opt_deadline; - // Check for equality (speedup). - if (text1 === text2) { - if (text1) { - return [ - [DIFF_EQUAL, text1] - ]; - } - return []; - } + // Check for null inputs. + if (text1 == null || text2 == null) { + throw new Error('Cannot diff null input.'); + } - if (typeof optChecklines === 'undefined') { - optChecklines = true; + // Check for equality (speedup). + if (text1 == text2) { + if (text1) { + return [new diff_match_patch.Diff(DIFF_EQUAL, text1)]; } + return []; + } - // Trim off common prefix (speedup). - let commonlength = this.diffCommonPrefix(text1, text2); - let commonprefix = text1.substring(0, commonlength); - text1 = text1.substring(commonlength); - text2 = text2.substring(commonlength); + if (typeof opt_checklines == 'undefined') { + opt_checklines = true; + } + var checklines = opt_checklines; + + // Trim off common prefix (speedup). + var commonlength = this.diff_commonPrefix(text1, text2); + var commonprefix = text1.substring(0, commonlength); + text1 = text1.substring(commonlength); + text2 = text2.substring(commonlength); + + // Trim off common suffix (speedup). + commonlength = this.diff_commonSuffix(text1, text2); + var commonsuffix = text1.substring(text1.length - commonlength); + text1 = text1.substring(0, text1.length - commonlength); + text2 = text2.substring(0, text2.length - commonlength); + + // Compute the diff on the middle block. + var diffs = this.diff_compute_(text1, text2, checklines, deadline); + + // Restore the prefix and suffix. + if (commonprefix) { + diffs.unshift(new diff_match_patch.Diff(DIFF_EQUAL, commonprefix)); + } + if (commonsuffix) { + diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, commonsuffix)); + } + this.diff_cleanupMerge(diffs); + return diffs; +}; + +/** + * Find the differences between two texts. Assumes that the texts do not + * have any common prefix or suffix. + * @param {string} text1 Old string to be diffed. + * @param {string} text2 New string to be diffed. + * @param {boolean} checklines Speedup flag. If false, then don't run a + * line-level diff first to identify the changed areas. + * If true, then run a faster, slightly less optimal diff. + * @param {number} deadline Time when the diff should be complete by. + * @return {!Array.} Array of diff tuples. + * @private + */ +diff_match_patch.prototype.diff_compute_ = function (text1, text2, checklines, + deadline) { + var diffs; - // Trim off common suffix (speedup). - commonlength = this.diffCommonSuffix(text1, text2); - let commonsuffix = text1.substring(text1.length - commonlength); - text1 = text1.substring(0, text1.length - commonlength); - text2 = text2.substring(0, text2.length - commonlength); + if (!text1) { + // Just add some text (speedup). + return [new diff_match_patch.Diff(DIFF_INSERT, text2)]; + } - // Compute the diff on the middle block. - let diffs = this.diffCompute(text1, text2, optChecklines, deadline); + if (!text2) { + // Just delete some text (speedup). + return [new diff_match_patch.Diff(DIFF_DELETE, text1)]; + } - // Restore the prefix and suffix. - if (commonprefix) { - diffs.unshift([DIFF_EQUAL, commonprefix]); - } - if (commonsuffix) { - diffs.push([DIFF_EQUAL, commonsuffix]); + var longtext = text1.length > text2.length ? text1 : text2; + var shorttext = text1.length > text2.length ? text2 : text1; + var i = longtext.indexOf(shorttext); + if (i != -1) { + // Shorter text is inside the longer text (speedup). + diffs = [new diff_match_patch.Diff(DIFF_INSERT, longtext.substring(0, i)), + new diff_match_patch.Diff(DIFF_EQUAL, shorttext), + new diff_match_patch.Diff(DIFF_INSERT, + longtext.substring(i + shorttext.length))]; + // Swap insertions for deletions if diff is reversed. + if (text1.length > text2.length) { + diffs[0][0] = diffs[2][0] = DIFF_DELETE; } - this.diffCleanupMerge(diffs); return diffs; - }; + } - /** - * Reduce the number of edits by eliminating operationally trivial equalities. - * @param {!Array.} diffs Array of diff tuples. - */ - DiffMatchPatch.prototype.diffCleanupEfficiency = function (diffs) { - let changes, equalities, equalitiesLength, lastequality, - pointer, preIns, preDel, postIns, postDel; - changes = false; - equalities = []; // Stack of indices where equalities are found. - equalitiesLength = 0; // Keeping our own length var is faster in JS. - /** @type {?string} */ - lastequality = null; - - // Always equal to diffs[equalities[equalitiesLength - 1]][1] - pointer = 0; // Index of current position. - - // Is there an insertion operation before the last equality. - preIns = false; - - // Is there a deletion operation before the last equality. - preDel = false; - - // Is there an insertion operation after the last equality. - postIns = false; - - // Is there a deletion operation after the last equality. - postDel = false; - while (pointer < diffs.length) { - // Equality found. - if (diffs[pointer][0] === DIFF_EQUAL) { - if (diffs[pointer][1].length < 4 && (postIns || postDel)) { - // Candidate found. - equalities[equalitiesLength++] = pointer; - preIns = postIns; - preDel = postDel; - lastequality = diffs[pointer][1]; - } else { - // Not a candidate, and can never become one. - equalitiesLength = 0; - lastequality = null; - } - postIns = postDel = false; + if (shorttext.length == 1) { + // Single character string. + // After the previous speedup, the character can't be an equality. + return [new diff_match_patch.Diff(DIFF_DELETE, text1), + new diff_match_patch.Diff(DIFF_INSERT, text2)]; + } - // An insertion or deletion. - } else { - if (diffs[pointer][0] === DIFF_DELETE) { - postDel = true; - } else { - postIns = true; - } + // Check to see if the problem can be split in two. + var hm = this.diff_halfMatch_(text1, text2); + if (hm) { + // A half-match was found, sort out the return data. + var text1_a = hm[0]; + var text1_b = hm[1]; + var text2_a = hm[2]; + var text2_b = hm[3]; + var mid_common = hm[4]; + // Send both pairs off for separate processing. + var diffs_a = this.diff_main(text1_a, text2_a, checklines, deadline); + var diffs_b = this.diff_main(text1_b, text2_b, checklines, deadline); + // Merge the results. + return diffs_a.concat([new diff_match_patch.Diff(DIFF_EQUAL, mid_common)], + diffs_b); + } - /* - * Five types to be split: - * ABXYCD - * AXCD - * ABXC - * AXCD - * ABXC - */ - if (lastequality && ((preIns && preDel && postIns && postDel) || - ((lastequality.length < 2) && - (preIns + preDel + postIns + postDel) === 3))) { - // Duplicate record. - diffs.splice( - equalities[equalitiesLength - 1], - 0, - [DIFF_DELETE, lastequality] - ); - - // Change second copy to insert. - diffs[equalities[equalitiesLength - 1] + 1][0] = DIFF_INSERT; - equalitiesLength--; // Throw away the equality we just deleted; - lastequality = null; - if (preIns && preDel) { - // No changes made which could affect previous entry, keep going. - postIns = postDel = true; - equalitiesLength = 0; - } else { - equalitiesLength--; // Throw away the previous equality. - pointer = equalitiesLength > 0 ? equalities[equalitiesLength - 1] : -1; - postIns = postDel = false; + if (checklines && text1.length > 100 && text2.length > 100) { + return this.diff_lineMode_(text1, text2, deadline); + } + + return this.diff_bisect_(text1, text2, deadline); +}; + +/** + * Do a quick line-level diff on both strings, then rediff the parts for + * greater accuracy. + * This speedup can produce non-minimal diffs. + * @param {string} text1 Old string to be diffed. + * @param {string} text2 New string to be diffed. + * @param {number} deadline Time when the diff should be complete by. + * @return {!Array.} Array of diff tuples. + * @private + */ +diff_match_patch.prototype.diff_lineMode_ = function (text1, text2, deadline) { + // Scan the text on a line-by-line basis first. + var a = this.diff_linesToChars_(text1, text2); + text1 = a.chars1; + text2 = a.chars2; + var linearray = a.lineArray; + + var diffs = this.diff_main(text1, text2, false, deadline); + + // Convert the diff back to original text. + this.diff_charsToLines_(diffs, linearray); + // Eliminate freak matches (e.g. blank lines) + this.diff_cleanupSemantic(diffs); + + // Rediff any replacement blocks, this time character-by-character. + // Add a dummy entry at the end. + diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, '')); + var pointer = 0; + var count_delete = 0; + var count_insert = 0; + var text_delete = ''; + var text_insert = ''; + while (pointer < diffs.length) { + switch (diffs[pointer][0]) { + case DIFF_INSERT: + count_insert++; + text_insert += diffs[pointer][1]; + break; + case DIFF_DELETE: + count_delete++; + text_delete += diffs[pointer][1]; + break; + case DIFF_EQUAL: + // Upon reaching an equality, check for prior redundancies. + if (count_delete >= 1 && count_insert >= 1) { + // Delete the offending records and add the merged ones. + diffs.splice(pointer - count_delete - count_insert, + count_delete + count_insert); + pointer = pointer - count_delete - count_insert; + var subDiff = + this.diff_main(text_delete, text_insert, false, deadline); + for (var j = subDiff.length - 1; j >= 0; j--) { + diffs.splice(pointer, 0, subDiff[j]); } - changes = true; + pointer = pointer + subDiff.length; } - } - pointer++; + count_insert = 0; + count_delete = 0; + text_delete = ''; + text_insert = ''; + break; } - - if (changes) { - this.diffCleanupMerge(diffs); + pointer++; + } + diffs.pop(); // Remove the dummy entry at the end. + + return diffs; +}; + +/** + * Find the 'middle snake' of a diff, split the problem in two + * and return the recursively constructed diff. + * See Myers 1986 paper: An O(ND) Difference Algorithm and Its Variations. + * @param {string} text1 Old string to be diffed. + * @param {string} text2 New string to be diffed. + * @param {number} deadline Time at which to bail if not yet complete. + * @return {!Array.} Array of diff tuples. + * @private + */ +diff_match_patch.prototype.diff_bisect_ = function (text1, text2, deadline) { + // Cache the text lengths to prevent multiple calls. + var text1_length = text1.length; + var text2_length = text2.length; + var max_d = Math.ceil((text1_length + text2_length) / 2); + var v_offset = max_d; + var v_length = 2 * max_d; + var v1 = new Array(v_length); + var v2 = new Array(v_length); + // Setting all elements to -1 is faster in Chrome & Firefox than mixing + // integers and undefined. + for (var x = 0; x < v_length; x++) { + v1[x] = -1; + v2[x] = -1; + } + v1[v_offset + 1] = 0; + v2[v_offset + 1] = 0; + var delta = text1_length - text2_length; + // If the total number of characters is odd, then the front path will collide + // with the reverse path. + var front = (delta % 2 != 0); + // Offsets for start and end of k loop. + // Prevents mapping of space beyond the grid. + var k1start = 0; + var k1end = 0; + var k2start = 0; + var k2end = 0; + for (var d = 0; d < max_d; d++) { + // Bail out if deadline is reached. + if ((new Date()).getTime() > deadline) { + break; } - }; - /** - * Convert a diff array into a pretty HTML report. - * @param {!Array.} diffs Array of diff tuples. - * @param {integer} string to be beautified. - * @return {string} HTML representation. - */ - DiffMatchPatch.prototype.diffPrettyHtml = function (diffs) { - const html = []; - for (let x = 0; x < diffs.length; x++) { - const op = diffs[x][0]; // Operation (insert, delete, equal) - const data = diffs[x][1]; // Text of change. - switch (op) { - case DIFF_INSERT: - html[x] = '' + escapeText(data) + ''; - break; - case DIFF_DELETE: - html[x] = '' + escapeText(data) + ''; - break; - case DIFF_EQUAL: - html[x] = '' + escapeText(data) + ''; - break; + // Walk the front path one step. + for (var k1 = -d + k1start; k1 <= d - k1end; k1 += 2) { + var k1_offset = v_offset + k1; + var x1; + if (k1 == -d || (k1 != d && v1[k1_offset - 1] < v1[k1_offset + 1])) { + x1 = v1[k1_offset + 1]; + } else { + x1 = v1[k1_offset - 1] + 1; + } + var y1 = x1 - k1; + while (x1 < text1_length && y1 < text2_length && + text1.charAt(x1) == text2.charAt(y1)) { + x1++; + y1++; + } + v1[k1_offset] = x1; + if (x1 > text1_length) { + // Ran off the right of the graph. + k1end += 2; + } else if (y1 > text2_length) { + // Ran off the bottom of the graph. + k1start += 2; + } else if (front) { + var k2_offset = v_offset + delta - k1; + if (k2_offset >= 0 && k2_offset < v_length && v2[k2_offset] != -1) { + // Mirror x2 onto top-left coordinate system. + var x2 = text1_length - v2[k2_offset]; + if (x1 >= x2) { + // Overlap detected. + return this.diff_bisectSplit_(text1, text2, x1, y1, deadline); + } + } } - } - return html.join(''); - }; - - /** - * Determine the common prefix of two strings. - * @param {string} text1 First string. - * @param {string} text2 Second string. - * @return {number} The number of characters common to the start of each - * string. - */ - DiffMatchPatch.prototype.diffCommonPrefix = function (text1, text2) { - let pointermid, pointermax, pointermin, pointerstart; - - // Quick check for common null cases. - if (!text1 || !text2 || text1.charAt(0) !== text2.charAt(0)) { - return 0; } - // Binary search. - // Performance analysis: https://neil.fraser.name/news/2007/10/09/ - pointermin = 0; - pointermax = Math.min(text1.length, text2.length); - pointermid = pointermax; - pointerstart = 0; - while (pointermin < pointermid) { - if (text1.substring(pointerstart, pointermid) === - text2.substring(pointerstart, pointermid)) { - pointermin = pointermid; - pointerstart = pointermin; + // Walk the reverse path one step. + for (var k2 = -d + k2start; k2 <= d - k2end; k2 += 2) { + var k2_offset = v_offset + k2; + var x2; + if (k2 == -d || (k2 != d && v2[k2_offset - 1] < v2[k2_offset + 1])) { + x2 = v2[k2_offset + 1]; } else { - pointermax = pointermid; + x2 = v2[k2_offset - 1] + 1; + } + var y2 = x2 - k2; + while (x2 < text1_length && y2 < text2_length && + text1.charAt(text1_length - x2 - 1) == + text2.charAt(text2_length - y2 - 1)) { + x2++; + y2++; + } + v2[k2_offset] = x2; + if (x2 > text1_length) { + // Ran off the left of the graph. + k2end += 2; + } else if (y2 > text2_length) { + // Ran off the top of the graph. + k2start += 2; + } else if (!front) { + var k1_offset = v_offset + delta - k2; + if (k1_offset >= 0 && k1_offset < v_length && v1[k1_offset] != -1) { + var x1 = v1[k1_offset]; + var y1 = v_offset + x1 - k1_offset; + // Mirror x2 onto top-left coordinate system. + x2 = text1_length - x2; + if (x1 >= x2) { + // Overlap detected. + return this.diff_bisectSplit_(text1, text2, x1, y1, deadline); + } + } } - pointermid = Math.floor((pointermax - pointermin) / 2 + pointermin); } - return pointermid; - }; + } + // Diff took too long and hit the deadline or + // number of diffs equals number of characters, no commonality at all. + return [new diff_match_patch.Diff(DIFF_DELETE, text1), + new diff_match_patch.Diff(DIFF_INSERT, text2)]; +}; + +/** + * Given the location of the 'middle snake', split the diff in two parts + * and recurse. + * @param {string} text1 Old string to be diffed. + * @param {string} text2 New string to be diffed. + * @param {number} x Index of split point in text1. + * @param {number} y Index of split point in text2. + * @param {number} deadline Time at which to bail if not yet complete. + * @return {!Array.} Array of diff tuples. + * @private + */ +diff_match_patch.prototype.diff_bisectSplit_ = function (text1, text2, x, y, + deadline) { + var text1a = text1.substring(0, x); + var text2a = text2.substring(0, y); + var text1b = text1.substring(x); + var text2b = text2.substring(y); + + // Compute both diffs serially. + var diffs = this.diff_main(text1a, text2a, false, deadline); + var diffsb = this.diff_main(text1b, text2b, false, deadline); + + return diffs.concat(diffsb); +}; + +/** + * Split two texts into an array of strings. Reduce the texts to a string of + * hashes where each Unicode character represents one line. + * @param {string} text1 First string. + * @param {string} text2 Second string. + * @return {{chars1: string, chars2: string, lineArray: !Array.}} + * An object containing the encoded text1, the encoded text2 and + * the array of unique strings. + * The zeroth element of the array of unique strings is intentionally blank. + * @private + */ +diff_match_patch.prototype.diff_linesToChars_ = function (text1, text2) { + var lineArray = []; // e.g. lineArray[4] == 'Hello\n' + var lineHash = {}; // e.g. lineHash['Hello\n'] == 4 + var hasOwn = Object.prototype.hasOwnProperty; + + // '\x00' is a valid character, but various debuggers don't like it. + // So we'll insert a junk entry to avoid generating a null character. + lineArray[0] = ''; /** - * Determine the common suffix of two strings. - * @param {string} text1 First string. - * @param {string} text2 Second string. - * @return {number} The number of characters common to the end of each string. + * Split a text into an array of strings. Reduce the texts to a string of + * hashes where each Unicode character represents one line. + * Modifies linearray and linehash through being a closure. + * @param {string} text String to encode. + * @return {string} Encoded string. + * @private */ - DiffMatchPatch.prototype.diffCommonSuffix = function (text1, text2) { - let pointermid, pointermax, pointermin, pointerend; - - // Quick check for common null cases. - if (!text1 || - !text2 || - text1.charAt(text1.length - 1) !== text2.charAt(text2.length - 1)) { - return 0; - } + function diff_linesToCharsMunge_ (text) { + var chars = ''; + // Walk the text, pulling out a substring for each line. + // text.split('\n') would would temporarily double our memory footprint. + // Modifying text would create many large strings to garbage collect. + var lineStart = 0; + var lineEnd = -1; + // Keeping our own length variable is faster than looking it up. + var lineArrayLength = lineArray.length; + while (lineEnd < text.length - 1) { + lineEnd = text.indexOf('\n', lineStart); + if (lineEnd == -1) { + lineEnd = text.length - 1; + } + var line = text.substring(lineStart, lineEnd + 1); - // Binary search. - // Performance analysis: https://neil.fraser.name/news/2007/10/09/ - pointermin = 0; - pointermax = Math.min(text1.length, text2.length); - pointermid = pointermax; - pointerend = 0; - while (pointermin < pointermid) { - if (text1.substring(text1.length - pointermid, text1.length - pointerend) === - text2.substring(text2.length - pointermid, text2.length - pointerend)) { - pointermin = pointermid; - pointerend = pointermin; + if (hasOwn.call(lineHash, line)) { + chars += String.fromCharCode(lineHash[line]); } else { - pointermax = pointermid; + if (lineArrayLength == maxLines) { + // Bail out at 65535 because + // String.fromCharCode(65536) == String.fromCharCode(0) + line = text.substring(lineStart); + lineEnd = text.length; + } + chars += String.fromCharCode(lineArrayLength); + lineHash[line] = lineArrayLength; + lineArray[lineArrayLength++] = line; } - pointermid = Math.floor((pointermax - pointermin) / 2 + pointermin); + lineStart = lineEnd + 1; } - return pointermid; - }; - - /** - * Find the differences between two texts. Assumes that the texts do not - * have any common prefix or suffix. - * @param {string} text1 Old string to be diffed. - * @param {string} text2 New string to be diffed. - * @param {boolean} checklines Speedup flag. If false, then don't run a - * line-level diff first to identify the changed areas. - * If true, then run a faster, slightly less optimal diff. - * @param {number} deadline Time when the diff should be complete by. - * @return {!Array.} Array of diff tuples. - * @private - */ - DiffMatchPatch.prototype.diffCompute = function (text1, text2, checklines, deadline) { - let diffs, longtext, shorttext, i, hm, - text1A, text2A, text1B, text2B, - midCommon, diffsA, diffsB; - - if (!text1) { - // Just add some text (speedup). - return [ - [DIFF_INSERT, text2] - ]; - } - - if (!text2) { - // Just delete some text (speedup). - return [ - [DIFF_DELETE, text1] - ]; + return chars; + } + // Allocate 2/3rds of the space for text1, the rest for text2. + var maxLines = 40000; + var chars1 = diff_linesToCharsMunge_(text1); + maxLines = 65535; + var chars2 = diff_linesToCharsMunge_(text2); + return { chars1: chars1, chars2: chars2, lineArray: lineArray }; +}; + +/** + * Rehydrate the text in a diff from a string of line hashes to real lines of + * text. + * @param {!Array.} diffs Array of diff tuples. + * @param {!Array.} lineArray Array of unique strings. + * @private + */ +diff_match_patch.prototype.diff_charsToLines_ = function (diffs, lineArray) { + for (var i = 0; i < diffs.length; i++) { + var chars = diffs[i][1]; + var text = []; + for (var j = 0; j < chars.length; j++) { + text[j] = lineArray[chars.charCodeAt(j)]; } - - longtext = text1.length > text2.length ? text1 : text2; - shorttext = text1.length > text2.length ? text2 : text1; - i = longtext.indexOf(shorttext); - if (i !== -1) { - // Shorter text is inside the longer text (speedup). - diffs = [ - [DIFF_INSERT, longtext.substring(0, i)], - [DIFF_EQUAL, shorttext], - [DIFF_INSERT, longtext.substring(i + shorttext.length)] - ]; - - // Swap insertions for deletions if diff is reversed. - if (text1.length > text2.length) { - diffs[0][0] = diffs[2][0] = DIFF_DELETE; - } - return diffs; + diffs[i][1] = text.join(''); + } +}; + +/** + * Determine the common prefix of two strings. + * @param {string} text1 First string. + * @param {string} text2 Second string. + * @return {number} The number of characters common to the start of each + * string. + */ +diff_match_patch.prototype.diff_commonPrefix = function (text1, text2) { + // Quick check for common null cases. + if (!text1 || !text2 || text1.charAt(0) != text2.charAt(0)) { + return 0; + } + // Binary search. + // Performance analysis: https://neil.fraser.name/news/2007/10/09/ + var pointermin = 0; + var pointermax = Math.min(text1.length, text2.length); + var pointermid = pointermax; + var pointerstart = 0; + while (pointermin < pointermid) { + if (text1.substring(pointerstart, pointermid) == + text2.substring(pointerstart, pointermid)) { + pointermin = pointermid; + pointerstart = pointermin; + } else { + pointermax = pointermid; } - - if (shorttext.length === 1) { - // Single character string. - // After the previous speedup, the character can't be an equality. - return [ - [DIFF_DELETE, text1], - [DIFF_INSERT, text2] - ]; + pointermid = Math.floor((pointermax - pointermin) / 2 + pointermin); + } + return pointermid; +}; + +/** + * Determine the common suffix of two strings. + * @param {string} text1 First string. + * @param {string} text2 Second string. + * @return {number} The number of characters common to the end of each string. + */ +diff_match_patch.prototype.diff_commonSuffix = function (text1, text2) { + // Quick check for common null cases. + if (!text1 || !text2 || + text1.charAt(text1.length - 1) != text2.charAt(text2.length - 1)) { + return 0; + } + // Binary search. + // Performance analysis: https://neil.fraser.name/news/2007/10/09/ + var pointermin = 0; + var pointermax = Math.min(text1.length, text2.length); + var pointermid = pointermax; + var pointerend = 0; + while (pointermin < pointermid) { + if (text1.substring(text1.length - pointermid, text1.length - pointerend) == + text2.substring(text2.length - pointermid, text2.length - pointerend)) { + pointermin = pointermid; + pointerend = pointermin; + } else { + pointermax = pointermid; } + pointermid = Math.floor((pointermax - pointermin) / 2 + pointermin); + } + return pointermid; +}; + +/** + * Determine if the suffix of one string is the prefix of another. + * @param {string} text1 First string. + * @param {string} text2 Second string. + * @return {number} The number of characters common to the end of the first + * string and the start of the second string. + * @private + */ +diff_match_patch.prototype.diff_commonOverlap_ = function (text1, text2) { + // Cache the text lengths to prevent multiple calls. + var text1_length = text1.length; + var text2_length = text2.length; + // Eliminate the null case. + if (text1_length == 0 || text2_length == 0) { + return 0; + } + // Truncate the longer string. + if (text1_length > text2_length) { + text1 = text1.substring(text1_length - text2_length); + } else if (text1_length < text2_length) { + text2 = text2.substring(0, text1_length); + } + var text_length = Math.min(text1_length, text2_length); + // Quick check for the worst case. + if (text1 == text2) { + return text_length; + } - // Check to see if the problem can be split in two. - hm = this.diffHalfMatch(text1, text2); - if (hm) { - // A half-match was found, sort out the return data. - text1A = hm[0]; - text1B = hm[1]; - text2A = hm[2]; - text2B = hm[3]; - midCommon = hm[4]; - - // Send both pairs off for separate processing. - diffsA = this.DiffMain(text1A, text2A, checklines, deadline); - diffsB = this.DiffMain(text1B, text2B, checklines, deadline); - - // Merge the results. - return diffsA.concat([ - [DIFF_EQUAL, midCommon] - ], diffsB); + // Start by looking for a single character match + // and increase length until no match is found. + // Performance analysis: https://neil.fraser.name/news/2010/11/04/ + var best = 0; + var length = 1; + while (true) { + var pattern = text1.substring(text_length - length); + var found = text2.indexOf(pattern); + if (found == -1) { + return best; } - - if (checklines && text1.length > 100 && text2.length > 100) { - return this.diffLineMode(text1, text2, deadline); + length += found; + if (found == 0 || text1.substring(text_length - length) == + text2.substring(0, length)) { + best = length; + length++; } - - return this.diffBisect(text1, text2, deadline); - }; + } +}; + +/** + * Do the two texts share a substring which is at least half the length of the + * longer text? + * This speedup can produce non-minimal diffs. + * @param {string} text1 First string. + * @param {string} text2 Second string. + * @return {Array.} Five element Array, containing the prefix of + * text1, the suffix of text1, the prefix of text2, the suffix of + * text2 and the common middle. Or null if there was no match. + * @private + */ +diff_match_patch.prototype.diff_halfMatch_ = function (text1, text2) { + if (this.Diff_Timeout <= 0) { + // Don't risk returning a non-optimal diff if we have unlimited time. + return null; + } + var longtext = text1.length > text2.length ? text1 : text2; + var shorttext = text1.length > text2.length ? text2 : text1; + if (longtext.length < 4 || shorttext.length * 2 < longtext.length) { + return null; // Pointless. + } + var dmp = this; // 'this' becomes 'window' in a closure. /** - * Do the two texts share a substring which is at least half the length of the - * longer text? - * This speedup can produce non-minimal diffs. - * @param {string} text1 First string. - * @param {string} text2 Second string. + * Does a substring of shorttext exist within longtext such that the substring + * is at least half the length of longtext? + * Closure, but does not reference any external variables. + * @param {string} longtext Longer string. + * @param {string} shorttext Shorter string. + * @param {number} i Start index of quarter length substring within longtext. * @return {Array.} Five element Array, containing the prefix of - * text1, the suffix of text1, the prefix of text2, the suffix of - * text2 and the common middle. Or null if there was no match. + * longtext, the suffix of longtext, the prefix of shorttext, the suffix + * of shorttext and the common middle. Or null if there was no match. * @private */ - DiffMatchPatch.prototype.diffHalfMatch = function (text1, text2) { - let longtext, shorttext, dmp, - text1A, text2B, text2A, text1B, midCommon, - hm1, hm2, hm; - - longtext = text1.length > text2.length ? text1 : text2; - shorttext = text1.length > text2.length ? text2 : text1; - if (longtext.length < 4 || shorttext.length * 2 < longtext.length) { - return null; // Pointless. - } - dmp = this; // 'this' becomes 'window' in a closure. - - /** - * Does a substring of shorttext exist within longtext such that the substring - * is at least half the length of longtext? - * Closure, but does not reference any external variables. - * @param {string} longtext Longer string. - * @param {string} shorttext Shorter string. - * @param {number} i Start index of quarter length substring within longtext. - * @return {Array.} Five element Array, containing the prefix of - * longtext, the suffix of longtext, the prefix of shorttext, the suffix - * of shorttext and the common middle. Or null if there was no match. - * @private - */ - function diffHalfMatchI (longtext, shorttext, i) { - let seed, j, bestCommon, prefixLength, suffixLength, - bestLongtextA, bestLongtextB, bestShorttextA, bestShorttextB; - - // Start with a 1/4 length substring at position i as a seed. - seed = longtext.substring(i, i + Math.floor(longtext.length / 4)); - j = -1; - bestCommon = ''; - while ((j = shorttext.indexOf(seed, j + 1)) !== -1) { - prefixLength = dmp.diffCommonPrefix(longtext.substring(i), - shorttext.substring(j)); - suffixLength = dmp.diffCommonSuffix(longtext.substring(0, i), - shorttext.substring(0, j)); - if (bestCommon.length < suffixLength + prefixLength) { - bestCommon = shorttext.substring(j - suffixLength, j) + + function diff_halfMatchI_ (longtext, shorttext, i) { + // Start with a 1/4 length substring at position i as a seed. + var seed = longtext.substring(i, i + Math.floor(longtext.length / 4)); + var j = -1; + var best_common = ''; + var best_longtext_a, best_longtext_b, best_shorttext_a, best_shorttext_b; + while ((j = shorttext.indexOf(seed, j + 1)) != -1) { + var prefixLength = dmp.diff_commonPrefix(longtext.substring(i), + shorttext.substring(j)); + var suffixLength = dmp.diff_commonSuffix(longtext.substring(0, i), + shorttext.substring(0, j)); + if (best_common.length < suffixLength + prefixLength) { + best_common = shorttext.substring(j - suffixLength, j) + shorttext.substring(j, j + prefixLength); - bestLongtextA = longtext.substring(0, i - suffixLength); - bestLongtextB = longtext.substring(i + prefixLength); - bestShorttextA = shorttext.substring(0, j - suffixLength); - bestShorttextB = shorttext.substring(j + prefixLength); - } - } - if (bestCommon.length * 2 >= longtext.length) { - return [bestLongtextA, bestLongtextB, - bestShorttextA, bestShorttextB, bestCommon - ]; - } else { - return null; + best_longtext_a = longtext.substring(0, i - suffixLength); + best_longtext_b = longtext.substring(i + prefixLength); + best_shorttext_a = shorttext.substring(0, j - suffixLength); + best_shorttext_b = shorttext.substring(j + prefixLength); } } - - // First check if the second quarter is the seed for a half-match. - hm1 = diffHalfMatchI(longtext, shorttext, - Math.ceil(longtext.length / 4)); - - // Check again based on the third quarter. - hm2 = diffHalfMatchI(longtext, shorttext, - Math.ceil(longtext.length / 2)); - if (!hm1 && !hm2) { - return null; - } else if (!hm2) { - hm = hm1; - } else if (!hm1) { - hm = hm2; + if (best_common.length * 2 >= longtext.length) { + return [best_longtext_a, best_longtext_b, + best_shorttext_a, best_shorttext_b, best_common]; } else { - // Both matched. Select the longest. - hm = hm1[4].length > hm2[4].length ? hm1 : hm2; + return null; } + } - // A half-match was found, sort out the return data. - if (text1.length > text2.length) { - text1A = hm[0]; - text1B = hm[1]; - text2A = hm[2]; - text2B = hm[3]; - } else { - text2A = hm[0]; - text2B = hm[1]; - text1A = hm[2]; - text1B = hm[3]; + // First check if the second quarter is the seed for a half-match. + var hm1 = diff_halfMatchI_(longtext, shorttext, + Math.ceil(longtext.length / 4)); + // Check again based on the third quarter. + var hm2 = diff_halfMatchI_(longtext, shorttext, + Math.ceil(longtext.length / 2)); + var hm; + if (!hm1 && !hm2) { + return null; + } else if (!hm2) { + hm = hm1; + } else if (!hm1) { + hm = hm2; + } else { + // Both matched. Select the longest. + hm = hm1[4].length > hm2[4].length ? hm1 : hm2; + } + + // A half-match was found, sort out the return data. + var text1_a, text1_b, text2_a, text2_b; + if (text1.length > text2.length) { + text1_a = hm[0]; + text1_b = hm[1]; + text2_a = hm[2]; + text2_b = hm[3]; + } else { + text2_a = hm[0]; + text2_b = hm[1]; + text1_a = hm[2]; + text1_b = hm[3]; + } + var mid_common = hm[4]; + return [text1_a, text1_b, text2_a, text2_b, mid_common]; +}; + +/** + * Reduce the number of edits by eliminating semantically trivial equalities. + * @param {!Array.} diffs Array of diff tuples. + */ +diff_match_patch.prototype.diff_cleanupSemantic = function (diffs) { + var changes = false; + var equalities = []; // Stack of indices where equalities are found. + var equalitiesLength = 0; // Keeping our own length var is faster in JS. + /** @type {?string} */ + var lastEquality = null; + // Always equal to diffs[equalities[equalitiesLength - 1]][1] + var pointer = 0; // Index of current position. + // Number of characters that changed prior to the equality. + var length_insertions1 = 0; + var length_deletions1 = 0; + // Number of characters that changed after the equality. + var length_insertions2 = 0; + var length_deletions2 = 0; + while (pointer < diffs.length) { + if (diffs[pointer][0] == DIFF_EQUAL) { // Equality found. + equalities[equalitiesLength++] = pointer; + length_insertions1 = length_insertions2; + length_deletions1 = length_deletions2; + length_insertions2 = 0; + length_deletions2 = 0; + lastEquality = diffs[pointer][1]; + } else { // An insertion or deletion. + if (diffs[pointer][0] == DIFF_INSERT) { + length_insertions2 += diffs[pointer][1].length; + } else { + length_deletions2 += diffs[pointer][1].length; + } + // Eliminate an equality that is smaller or equal to the edits on both + // sides of it. + if (lastEquality && (lastEquality.length <= + Math.max(length_insertions1, length_deletions1)) && + (lastEquality.length <= Math.max(length_insertions2, + length_deletions2))) { + // Duplicate record. + diffs.splice(equalities[equalitiesLength - 1], 0, + new diff_match_patch.Diff(DIFF_DELETE, lastEquality)); + // Change second copy to insert. + diffs[equalities[equalitiesLength - 1] + 1][0] = DIFF_INSERT; + // Throw away the equality we just deleted. + equalitiesLength--; + // Throw away the previous equality (it needs to be reevaluated). + equalitiesLength--; + pointer = equalitiesLength > 0 ? equalities[equalitiesLength - 1] : -1; + length_insertions1 = 0; // Reset the counters. + length_deletions1 = 0; + length_insertions2 = 0; + length_deletions2 = 0; + lastEquality = null; + changes = true; + } } - midCommon = hm[4]; - return [text1A, text1B, text2A, text2B, midCommon]; - }; + pointer++; + } - /** - * Do a quick line-level diff on both strings, then rediff the parts for - * greater accuracy. - * This speedup can produce non-minimal diffs. - * @param {string} text1 Old string to be diffed. - * @param {string} text2 New string to be diffed. - * @param {number} deadline Time when the diff should be complete by. - * @return {!Array.} Array of diff tuples. - * @private - */ - DiffMatchPatch.prototype.diffLineMode = function (text1, text2, deadline) { - let a, diffs, linearray, pointer, countInsert, - countDelete, textInsert, textDelete, j; - - // Scan the text on a line-by-line basis first. - a = this.diffLinesToChars(text1, text2); - text1 = a.chars1; - text2 = a.chars2; - linearray = a.lineArray; - - diffs = this.DiffMain(text1, text2, false, deadline); - - // Convert the diff back to original text. - this.diffCharsToLines(diffs, linearray); - - // Eliminate freak matches (e.g. blank lines) - this.diffCleanupSemantic(diffs); - - // Rediff any replacement blocks, this time character-by-character. - // Add a dummy entry at the end. - diffs.push([DIFF_EQUAL, '']); - pointer = 0; - countDelete = 0; - countInsert = 0; - textDelete = ''; - textInsert = ''; - while (pointer < diffs.length) { - switch (diffs[pointer][0]) { - case DIFF_INSERT: - countInsert++; - textInsert += diffs[pointer][1]; - break; - case DIFF_DELETE: - countDelete++; - textDelete += diffs[pointer][1]; - break; - case DIFF_EQUAL: - - // Upon reaching an equality, check for prior redundancies. - if (countDelete >= 1 && countInsert >= 1) { - // Delete the offending records and add the merged ones. - diffs.splice(pointer - countDelete - countInsert, - countDelete + countInsert); - pointer = pointer - countDelete - countInsert; - a = this.DiffMain(textDelete, textInsert, false, deadline); - for (j = a.length - 1; j >= 0; j--) { - diffs.splice(pointer, 0, a[j]); - } - pointer = pointer + a.length; - } - countInsert = 0; - countDelete = 0; - textDelete = ''; - textInsert = ''; - break; + // Normalize the diff. + if (changes) { + this.diff_cleanupMerge(diffs); + } + this.diff_cleanupSemanticLossless(diffs); + + // Find any overlaps between deletions and insertions. + // e.g: abcxxxxxxdef + // -> abcxxxdef + // e.g: xxxabcdefxxx + // -> defxxxabc + // Only extract an overlap if it is as big as the edit ahead or behind it. + pointer = 1; + while (pointer < diffs.length) { + if (diffs[pointer - 1][0] == DIFF_DELETE && + diffs[pointer][0] == DIFF_INSERT) { + var deletion = diffs[pointer - 1][1]; + var insertion = diffs[pointer][1]; + var overlap_length1 = this.diff_commonOverlap_(deletion, insertion); + var overlap_length2 = this.diff_commonOverlap_(insertion, deletion); + if (overlap_length1 >= overlap_length2) { + if (overlap_length1 >= deletion.length / 2 || + overlap_length1 >= insertion.length / 2) { + // Overlap found. Insert an equality and trim the surrounding edits. + diffs.splice(pointer, 0, new diff_match_patch.Diff(DIFF_EQUAL, + insertion.substring(0, overlap_length1))); + diffs[pointer - 1][1] = + deletion.substring(0, deletion.length - overlap_length1); + diffs[pointer + 1][1] = insertion.substring(overlap_length1); + pointer++; + } + } else { + if (overlap_length2 >= deletion.length / 2 || + overlap_length2 >= insertion.length / 2) { + // Reverse overlap found. + // Insert an equality and swap and trim the surrounding edits. + diffs.splice(pointer, 0, new diff_match_patch.Diff(DIFF_EQUAL, + deletion.substring(0, overlap_length2))); + diffs[pointer - 1][0] = DIFF_INSERT; + diffs[pointer - 1][1] = + insertion.substring(0, insertion.length - overlap_length2); + diffs[pointer + 1][0] = DIFF_DELETE; + diffs[pointer + 1][1] = + deletion.substring(overlap_length2); + pointer++; + } } pointer++; } - diffs.pop(); // Remove the dummy entry at the end. - - return diffs; - }; + pointer++; + } +}; +/** + * Look for single edits surrounded on both sides by equalities + * which can be shifted sideways to align the edit to a word boundary. + * e.g: The cat came. -> The cat came. + * @param {!Array.} diffs Array of diff tuples. + */ +diff_match_patch.prototype.diff_cleanupSemanticLossless = function (diffs) { /** - * Find the 'middle snake' of a diff, split the problem in two - * and return the recursively constructed diff. - * See Myers 1986 paper: An O(ND) Difference Algorithm and Its Variations. - * @param {string} text1 Old string to be diffed. - * @param {string} text2 New string to be diffed. - * @param {number} deadline Time at which to bail if not yet complete. - * @return {!Array.} Array of diff tuples. + * Given two strings, compute a score representing whether the internal + * boundary falls on logical boundaries. + * Scores range from 6 (best) to 0 (worst). + * Closure, but does not reference any external variables. + * @param {string} one First string. + * @param {string} two Second string. + * @return {number} The score. * @private */ - DiffMatchPatch.prototype.diffBisect = function (text1, text2, deadline) { - let text1Length, text2Length, maxD, vOffset, vLength, - v1, v2, x, delta, front, k1start, k1end, k2start, - k2end, k2Offset, k1Offset, x1, x2, y1, y2, d, k1, k2; - - // Cache the text lengths to prevent multiple calls. - text1Length = text1.length; - text2Length = text2.length; - maxD = Math.ceil((text1Length + text2Length) / 2); - vOffset = maxD; - vLength = 2 * maxD; - v1 = new Array(vLength); - v2 = new Array(vLength); - - // Setting all elements to -1 is faster in Chrome & Firefox than mixing - // integers and undefined. - for (x = 0; x < vLength; x++) { - v1[x] = -1; - v2[x] = -1; + function diff_cleanupSemanticScore_ (one, two) { + if (!one || !two) { + // Edges are the best. + return 6; } - v1[vOffset + 1] = 0; - v2[vOffset + 1] = 0; - delta = text1Length - text2Length; - - // If the total number of characters is odd, then the front path will collide - // with the reverse path. - front = (delta % 2 !== 0); - - // Offsets for start and end of k loop. - // Prevents mapping of space beyond the grid. - k1start = 0; - k1end = 0; - k2start = 0; - k2end = 0; - for (d = 0; d < maxD; d++) { - // Bail out if deadline is reached. - if (Date.now() > deadline) { - break; + + // Each port of this function behaves slightly differently due to + // subtle differences in each language's definition of things like + // 'whitespace'. Since this function's purpose is largely cosmetic, + // the choice has been made to use each language's native features + // rather than force total conformity. + var char1 = one.charAt(one.length - 1); + var char2 = two.charAt(0); + var nonAlphaNumeric1 = char1.match(diff_match_patch.nonAlphaNumericRegex_); + var nonAlphaNumeric2 = char2.match(diff_match_patch.nonAlphaNumericRegex_); + var whitespace1 = nonAlphaNumeric1 && + char1.match(diff_match_patch.whitespaceRegex_); + var whitespace2 = nonAlphaNumeric2 && + char2.match(diff_match_patch.whitespaceRegex_); + var lineBreak1 = whitespace1 && + char1.match(diff_match_patch.linebreakRegex_); + var lineBreak2 = whitespace2 && + char2.match(diff_match_patch.linebreakRegex_); + var blankLine1 = lineBreak1 && + one.match(diff_match_patch.blanklineEndRegex_); + var blankLine2 = lineBreak2 && + two.match(diff_match_patch.blanklineStartRegex_); + + if (blankLine1 || blankLine2) { + // Five points for blank lines. + return 5; + } else if (lineBreak1 || lineBreak2) { + // Four points for line breaks. + return 4; + } else if (nonAlphaNumeric1 && !whitespace1 && whitespace2) { + // Three points for end of sentences. + return 3; + } else if (whitespace1 || whitespace2) { + // Two points for whitespace. + return 2; + } else if (nonAlphaNumeric1 || nonAlphaNumeric2) { + // One point for non-alphanumeric. + return 1; + } + return 0; + } + + var pointer = 1; + // Intentionally ignore the first and last element (don't need checking). + while (pointer < diffs.length - 1) { + if (diffs[pointer - 1][0] == DIFF_EQUAL && + diffs[pointer + 1][0] == DIFF_EQUAL) { + // This is a single edit surrounded by equalities. + var equality1 = diffs[pointer - 1][1]; + var edit = diffs[pointer][1]; + var equality2 = diffs[pointer + 1][1]; + + // First, shift the edit as far left as possible. + var commonOffset = this.diff_commonSuffix(equality1, edit); + if (commonOffset) { + var commonString = edit.substring(edit.length - commonOffset); + equality1 = equality1.substring(0, equality1.length - commonOffset); + edit = commonString + edit.substring(0, edit.length - commonOffset); + equality2 = commonString + equality2; } - // Walk the front path one step. - for (k1 = -d + k1start; k1 <= d - k1end; k1 += 2) { - k1Offset = vOffset + k1; - if (k1 === -d || (k1 !== d && v1[k1Offset - 1] < v1[k1Offset + 1])) { - x1 = v1[k1Offset + 1]; - } else { - x1 = v1[k1Offset - 1] + 1; - } - y1 = x1 - k1; - while (x1 < text1Length && y1 < text2Length && - text1.charAt(x1) === text2.charAt(y1)) { - x1++; - y1++; - } - v1[k1Offset] = x1; - if (x1 > text1Length) { - // Ran off the right of the graph. - k1end += 2; - } else if (y1 > text2Length) { - // Ran off the bottom of the graph. - k1start += 2; - } else if (front) { - k2Offset = vOffset + delta - k1; - if (k2Offset >= 0 && k2Offset < vLength && v2[k2Offset] !== -1) { - // Mirror x2 onto top-left coordinate system. - x2 = text1Length - v2[k2Offset]; - if (x1 >= x2) { - // Overlap detected. - return this.diffBisectSplit(text1, text2, x1, y1, deadline); - } - } + // Second, step character by character right, looking for the best fit. + var bestEquality1 = equality1; + var bestEdit = edit; + var bestEquality2 = equality2; + var bestScore = diff_cleanupSemanticScore_(equality1, edit) + + diff_cleanupSemanticScore_(edit, equality2); + while (edit.charAt(0) === equality2.charAt(0)) { + equality1 += edit.charAt(0); + edit = edit.substring(1) + equality2.charAt(0); + equality2 = equality2.substring(1); + var score = diff_cleanupSemanticScore_(equality1, edit) + + diff_cleanupSemanticScore_(edit, equality2); + // The >= encourages trailing rather than leading whitespace on edits. + if (score >= bestScore) { + bestScore = score; + bestEquality1 = equality1; + bestEdit = edit; + bestEquality2 = equality2; } } - // Walk the reverse path one step. - for (k2 = -d + k2start; k2 <= d - k2end; k2 += 2) { - k2Offset = vOffset + k2; - if (k2 === -d || (k2 !== d && v2[k2Offset - 1] < v2[k2Offset + 1])) { - x2 = v2[k2Offset + 1]; + if (diffs[pointer - 1][1] != bestEquality1) { + // We have an improvement, save it back to the diff. + if (bestEquality1) { + diffs[pointer - 1][1] = bestEquality1; } else { - x2 = v2[k2Offset - 1] + 1; - } - y2 = x2 - k2; - while (x2 < text1Length && y2 < text2Length && - text1.charAt(text1Length - x2 - 1) === - text2.charAt(text2Length - y2 - 1)) { - x2++; - y2++; + diffs.splice(pointer - 1, 1); + pointer--; } - v2[k2Offset] = x2; - if (x2 > text1Length) { - // Ran off the left of the graph. - k2end += 2; - } else if (y2 > text2Length) { - // Ran off the top of the graph. - k2start += 2; - } else if (!front) { - k1Offset = vOffset + delta - k2; - if (k1Offset >= 0 && k1Offset < vLength && v1[k1Offset] !== -1) { - x1 = v1[k1Offset]; - y1 = vOffset + x1 - k1Offset; - - // Mirror x2 onto top-left coordinate system. - x2 = text1Length - x2; - if (x1 >= x2) { - // Overlap detected. - return this.diffBisectSplit(text1, text2, x1, y1, deadline); - } - } + diffs[pointer][1] = bestEdit; + if (bestEquality2) { + diffs[pointer + 1][1] = bestEquality2; + } else { + diffs.splice(pointer + 1, 1); + pointer--; } } } - - // Diff took too long and hit the deadline or - // number of diffs equals number of characters, no commonality at all. - return [ - [DIFF_DELETE, text1], - [DIFF_INSERT, text2] - ]; - }; - - /** - * Given the location of the 'middle snake', split the diff in two parts - * and recurse. - * @param {string} text1 Old string to be diffed. - * @param {string} text2 New string to be diffed. - * @param {number} x Index of split point in text1. - * @param {number} y Index of split point in text2. - * @param {number} deadline Time at which to bail if not yet complete. - * @return {!Array.} Array of diff tuples. - * @private - */ - DiffMatchPatch.prototype.diffBisectSplit = function (text1, text2, x, y, deadline) { - let text1a, text1b, text2a, text2b, diffs, diffsb; - text1a = text1.substring(0, x); - text2a = text2.substring(0, y); - text1b = text1.substring(x); - text2b = text2.substring(y); - - // Compute both diffs serially. - diffs = this.DiffMain(text1a, text2a, false, deadline); - diffsb = this.DiffMain(text1b, text2b, false, deadline); - - return diffs.concat(diffsb); - }; - - /** - * Reduce the number of edits by eliminating semantically trivial equalities. - * @param {!Array.} diffs Array of diff tuples. - */ - DiffMatchPatch.prototype.diffCleanupSemantic = function (diffs) { - let changes = false; - let equalities = []; // Stack of indices where equalities are found. - let equalitiesLength = 0; // Keeping our own length var is faster in JS. - /** @type {?string} */ - let lastequality = null; - - // Always equal to diffs[equalities[equalitiesLength - 1]][1] - let pointer = 0; // Index of current position. - - // Number of characters that changed prior to the equality. - let lengthInsertions1 = 0; - let lengthDeletions1 = 0; - - // Number of characters that changed after the equality. - let lengthInsertions2 = 0; - let lengthDeletions2 = 0; - while (pointer < diffs.length) { - if (diffs[pointer][0] === DIFF_EQUAL) { // Equality found. + pointer++; + } +}; + +// Define some regex patterns for matching boundaries. +diff_match_patch.nonAlphaNumericRegex_ = /[^a-zA-Z0-9]/; +diff_match_patch.whitespaceRegex_ = /\s/; +diff_match_patch.linebreakRegex_ = /[\r\n]/; +diff_match_patch.blanklineEndRegex_ = /\n\r?\n$/; +diff_match_patch.blanklineStartRegex_ = /^\r?\n\r?\n/; + +/** + * Reduce the number of edits by eliminating operationally trivial equalities. + * @param {!Array.} diffs Array of diff tuples. + */ +diff_match_patch.prototype.diff_cleanupEfficiency = function (diffs) { + var changes = false; + var equalities = []; // Stack of indices where equalities are found. + var equalitiesLength = 0; // Keeping our own length var is faster in JS. + /** @type {?string} */ + var lastEquality = null; + // Always equal to diffs[equalities[equalitiesLength - 1]][1] + var pointer = 0; // Index of current position. + // Is there an insertion operation before the last equality. + var pre_ins = false; + // Is there a deletion operation before the last equality. + var pre_del = false; + // Is there an insertion operation after the last equality. + var post_ins = false; + // Is there a deletion operation after the last equality. + var post_del = false; + while (pointer < diffs.length) { + if (diffs[pointer][0] == DIFF_EQUAL) { // Equality found. + if (diffs[pointer][1].length < this.Diff_EditCost && + (post_ins || post_del)) { + // Candidate found. equalities[equalitiesLength++] = pointer; - lengthInsertions1 = lengthInsertions2; - lengthDeletions1 = lengthDeletions2; - lengthInsertions2 = 0; - lengthDeletions2 = 0; - lastequality = diffs[pointer][1]; - } else { // An insertion or deletion. - if (diffs[pointer][0] === DIFF_INSERT) { - lengthInsertions2 += diffs[pointer][1].length; + pre_ins = post_ins; + pre_del = post_del; + lastEquality = diffs[pointer][1]; + } else { + // Not a candidate, and can never become one. + equalitiesLength = 0; + lastEquality = null; + } + post_ins = post_del = false; + } else { // An insertion or deletion. + if (diffs[pointer][0] == DIFF_DELETE) { + post_del = true; + } else { + post_ins = true; + } + /* + * Five types to be split: + * ABXYCD + * AXCD + * ABXC + * AXCD + * ABXC + */ + if (lastEquality && ((pre_ins && pre_del && post_ins && post_del) || + ((lastEquality.length < this.Diff_EditCost / 2) && + (pre_ins + pre_del + post_ins + post_del) == 3))) { + // Duplicate record. + diffs.splice(equalities[equalitiesLength - 1], 0, + new diff_match_patch.Diff(DIFF_DELETE, lastEquality)); + // Change second copy to insert. + diffs[equalities[equalitiesLength - 1] + 1][0] = DIFF_INSERT; + equalitiesLength--; // Throw away the equality we just deleted; + lastEquality = null; + if (pre_ins && pre_del) { + // No changes made which could affect previous entry, keep going. + post_ins = post_del = true; + equalitiesLength = 0; } else { - lengthDeletions2 += diffs[pointer][1].length; - } - - // Eliminate an equality that is smaller or equal to the edits on both - // sides of it. - if (lastequality && (lastequality.length <= - Math.max(lengthInsertions1, lengthDeletions1)) && - (lastequality.length <= Math.max(lengthInsertions2, - lengthDeletions2))) { - // Duplicate record. - diffs.splice( - equalities[equalitiesLength - 1], - 0, - [DIFF_DELETE, lastequality] - ); - - // Change second copy to insert. - diffs[equalities[equalitiesLength - 1] + 1][0] = DIFF_INSERT; - - // Throw away the equality we just deleted. - equalitiesLength--; - - // Throw away the previous equality (it needs to be reevaluated). - equalitiesLength--; - pointer = equalitiesLength > 0 ? equalities[equalitiesLength - 1] : -1; - - // Reset the counters. - lengthInsertions1 = 0; - lengthDeletions1 = 0; - lengthInsertions2 = 0; - lengthDeletions2 = 0; - lastequality = null; - changes = true; + equalitiesLength--; // Throw away the previous equality. + pointer = equalitiesLength > 0 + ? equalities[equalitiesLength - 1] + : -1; + post_ins = post_del = false; } + changes = true; } - pointer++; } + pointer++; + } - // Normalize the diff. - if (changes) { - this.diffCleanupMerge(diffs); - } + if (changes) { + this.diff_cleanupMerge(diffs); + } +}; - let deletion, insertion, overlapLength1, overlapLength2; - - // Find any overlaps between deletions and insertions. - // e.g: abcxxxxxxdef - // -> abcxxxdef - // e.g: xxxabcdefxxx - // -> defxxxabc - // Only extract an overlap if it is as big as the edit ahead or behind it. - pointer = 1; - while (pointer < diffs.length) { - if (diffs[pointer - 1][0] === DIFF_DELETE && - diffs[pointer][0] === DIFF_INSERT) { - deletion = diffs[pointer - 1][1]; - insertion = diffs[pointer][1]; - overlapLength1 = this.diffCommonOverlap(deletion, insertion); - overlapLength2 = this.diffCommonOverlap(insertion, deletion); - if (overlapLength1 >= overlapLength2) { - if (overlapLength1 >= deletion.length / 2 || - overlapLength1 >= insertion.length / 2) { - // Overlap found. Insert an equality and trim the surrounding edits. - diffs.splice( - pointer, - 0, - [DIFF_EQUAL, insertion.substring(0, overlapLength1)] - ); - diffs[pointer - 1][1] = - deletion.substring(0, deletion.length - overlapLength1); - diffs[pointer + 1][1] = insertion.substring(overlapLength1); +/** + * Reorder and merge like edit sections. Merge equalities. + * Any edit section can move as long as it doesn't cross an equality. + * @param {!Array.} diffs Array of diff tuples. + */ +diff_match_patch.prototype.diff_cleanupMerge = function (diffs) { + // Add a dummy entry at the end. + diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, '')); + var pointer = 0; + var count_delete = 0; + var count_insert = 0; + var text_delete = ''; + var text_insert = ''; + var commonlength; + while (pointer < diffs.length) { + switch (diffs[pointer][0]) { + case DIFF_INSERT: + count_insert++; + text_insert += diffs[pointer][1]; + pointer++; + break; + case DIFF_DELETE: + count_delete++; + text_delete += diffs[pointer][1]; + pointer++; + break; + case DIFF_EQUAL: + // Upon reaching an equality, check for prior redundancies. + if (count_delete + count_insert > 1) { + if (count_delete !== 0 && count_insert !== 0) { + // Factor out any common prefixies. + commonlength = this.diff_commonPrefix(text_insert, text_delete); + if (commonlength !== 0) { + if ((pointer - count_delete - count_insert) > 0 && + diffs[pointer - count_delete - count_insert - 1][0] == + DIFF_EQUAL) { + diffs[pointer - count_delete - count_insert - 1][1] += + text_insert.substring(0, commonlength); + } else { + diffs.splice(0, 0, new diff_match_patch.Diff(DIFF_EQUAL, + text_insert.substring(0, commonlength))); + pointer++; + } + text_insert = text_insert.substring(commonlength); + text_delete = text_delete.substring(commonlength); + } + // Factor out any common suffixies. + commonlength = this.diff_commonSuffix(text_insert, text_delete); + if (commonlength !== 0) { + diffs[pointer][1] = text_insert.substring(text_insert.length - + commonlength) + diffs[pointer][1]; + text_insert = text_insert.substring(0, text_insert.length - + commonlength); + text_delete = text_delete.substring(0, text_delete.length - + commonlength); + } + } + // Delete the offending records and add the merged ones. + pointer -= count_delete + count_insert; + diffs.splice(pointer, count_delete + count_insert); + if (text_delete.length) { + diffs.splice(pointer, 0, + new diff_match_patch.Diff(DIFF_DELETE, text_delete)); pointer++; } - } else { - if (overlapLength2 >= deletion.length / 2 || - overlapLength2 >= insertion.length / 2) { - // Reverse overlap found. - // Insert an equality and swap and trim the surrounding edits. - diffs.splice( - pointer, - 0, - [DIFF_EQUAL, deletion.substring(0, overlapLength2)] - ); - - diffs[pointer - 1][0] = DIFF_INSERT; - diffs[pointer - 1][1] = - insertion.substring(0, insertion.length - overlapLength2); - diffs[pointer + 1][0] = DIFF_DELETE; - diffs[pointer + 1][1] = - deletion.substring(overlapLength2); + if (text_insert.length) { + diffs.splice(pointer, 0, + new diff_match_patch.Diff(DIFF_INSERT, text_insert)); pointer++; } + pointer++; + } else if (pointer !== 0 && diffs[pointer - 1][0] == DIFF_EQUAL) { + // Merge this equality with the previous one. + diffs[pointer - 1][1] += diffs[pointer][1]; + diffs.splice(pointer, 1); + } else { + pointer++; } - pointer++; + count_insert = 0; + count_delete = 0; + text_delete = ''; + text_insert = ''; + break; + } + } + if (diffs[diffs.length - 1][1] === '') { + diffs.pop(); // Remove the dummy entry at the end. + } + + // Second pass: look for single edits surrounded on both sides by equalities + // which can be shifted sideways to eliminate an equality. + // e.g: ABAC -> ABAC + var changes = false; + pointer = 1; + // Intentionally ignore the first and last element (don't need checking). + while (pointer < diffs.length - 1) { + if (diffs[pointer - 1][0] == DIFF_EQUAL && + diffs[pointer + 1][0] == DIFF_EQUAL) { + // This is a single edit surrounded by equalities. + if (diffs[pointer][1].substring(diffs[pointer][1].length - + diffs[pointer - 1][1].length) == diffs[pointer - 1][1]) { + // Shift the edit over the previous equality. + diffs[pointer][1] = diffs[pointer - 1][1] + + diffs[pointer][1].substring(0, diffs[pointer][1].length - + diffs[pointer - 1][1].length); + diffs[pointer + 1][1] = diffs[pointer - 1][1] + diffs[pointer + 1][1]; + diffs.splice(pointer - 1, 1); + changes = true; + } else if (diffs[pointer][1].substring(0, diffs[pointer + 1][1].length) == + diffs[pointer + 1][1]) { + // Shift the edit over the next equality. + diffs[pointer - 1][1] += diffs[pointer + 1][1]; + diffs[pointer][1] = + diffs[pointer][1].substring(diffs[pointer + 1][1].length) + + diffs[pointer + 1][1]; + diffs.splice(pointer + 1, 1); + changes = true; } - pointer++; } - }; + pointer++; + } + // If shifts were made, the diff needs reordering and another shift sweep. + if (changes) { + this.diff_cleanupMerge(diffs); + } +}; + +/** + * loc is a location in text1, compute and return the equivalent location in + * text2. + * e.g. 'The cat' vs 'The big cat', 1->1, 5->8 + * @param {!Array.} diffs Array of diff tuples. + * @param {number} loc Location within text1. + * @return {number} Location within text2. + */ +diff_match_patch.prototype.diff_xIndex = function (diffs, loc) { + var chars1 = 0; + var chars2 = 0; + var last_chars1 = 0; + var last_chars2 = 0; + var x; + for (x = 0; x < diffs.length; x++) { + if (diffs[x][0] !== DIFF_INSERT) { // Equality or deletion. + chars1 += diffs[x][1].length; + } + if (diffs[x][0] !== DIFF_DELETE) { // Equality or insertion. + chars2 += diffs[x][1].length; + } + if (chars1 > loc) { // Overshot the location. + break; + } + last_chars1 = chars1; + last_chars2 = chars2; + } + // Was the location was deleted? + if (diffs.length != x && diffs[x][0] === DIFF_DELETE) { + return last_chars2; + } + // Add the remaining character length. + return last_chars2 + (loc - last_chars1); +}; + +/** + * Convert a diff array into a pretty HTML report. + * @param {!Array.} diffs Array of diff tuples. + * @return {string} HTML representation. + */ +diff_match_patch.prototype.diff_prettyHtml = function (diffs) { + var html = []; + var pattern_amp = /&/g; + var pattern_lt = /'; + break; + case DIFF_DELETE: + html[x] = '' + text + ''; + break; + case DIFF_EQUAL: + html[x] = '' + text + ''; + break; + } + } + return html.join(''); +}; + +/** + * Compute and return the source text (all equalities and deletions). + * @param {!Array.} diffs Array of diff tuples. + * @return {string} Source text. + */ +diff_match_patch.prototype.diff_text1 = function (diffs) { + var text = []; + for (var x = 0; x < diffs.length; x++) { + if (diffs[x][0] !== DIFF_INSERT) { + text[x] = diffs[x][1]; + } + } + return text.join(''); +}; + +/** + * Compute and return the destination text (all equalities and insertions). + * @param {!Array.} diffs Array of diff tuples. + * @return {string} Destination text. + */ +diff_match_patch.prototype.diff_text2 = function (diffs) { + var text = []; + for (var x = 0; x < diffs.length; x++) { + if (diffs[x][0] !== DIFF_DELETE) { + text[x] = diffs[x][1]; + } + } + return text.join(''); +}; + +/** + * Compute the Levenshtein distance; the number of inserted, deleted or + * substituted characters. + * @param {!Array.} diffs Array of diff tuples. + * @return {number} Number of changes. + */ +diff_match_patch.prototype.diff_levenshtein = function (diffs) { + var levenshtein = 0; + var insertions = 0; + var deletions = 0; + for (var x = 0; x < diffs.length; x++) { + var op = diffs[x][0]; + var data = diffs[x][1]; + switch (op) { + case DIFF_INSERT: + insertions += data.length; + break; + case DIFF_DELETE: + deletions += data.length; + break; + case DIFF_EQUAL: + // A deletion and an insertion is one substitution. + levenshtein += Math.max(insertions, deletions); + insertions = 0; + deletions = 0; + break; + } + } + levenshtein += Math.max(insertions, deletions); + return levenshtein; +}; + +/** + * Crush the diff into an encoded string which describes the operations + * required to transform text1 into text2. + * E.g. =3\t-2\t+ing -> Keep 3 chars, delete 2 chars, insert 'ing'. + * Operations are tab-separated. Inserted text is escaped using %xx notation. + * @param {!Array.} diffs Array of diff tuples. + * @return {string} Delta text. + */ +diff_match_patch.prototype.diff_toDelta = function (diffs) { + var text = []; + for (var x = 0; x < diffs.length; x++) { + switch (diffs[x][0]) { + case DIFF_INSERT: + text[x] = '+' + encodeURI(diffs[x][1]); + break; + case DIFF_DELETE: + text[x] = '-' + diffs[x][1].length; + break; + case DIFF_EQUAL: + text[x] = '=' + diffs[x][1].length; + break; + } + } + return text.join('\t').replace(/%20/g, ' '); +}; + +/** + * Given the original text1, and an encoded string which describes the + * operations required to transform text1 into text2, compute the full diff. + * @param {string} text1 Source string for the diff. + * @param {string} delta Delta text. + * @return {!Array.} Array of diff tuples. + * @throws {!Error} If invalid input. + */ +diff_match_patch.prototype.diff_fromDelta = function (text1, delta) { + var diffs = []; + var diffsLength = 0; // Keeping our own length var is faster in JS. + var pointer = 0; // Cursor in text1 + var tokens = delta.split(/\t/g); + for (var x = 0; x < tokens.length; x++) { + // Each token begins with a one character parameter which specifies the + // operation of this token (delete, insert, equality). + var param = tokens[x].substring(1); + switch (tokens[x].charAt(0)) { + case '+': + try { + diffs[diffsLength++] = + new diff_match_patch.Diff(DIFF_INSERT, decodeURI(param)); + } catch (ex) { + // Malformed URI sequence. + throw new Error('Illegal escape in diff_fromDelta: ' + param); + } + break; + case '-': + // Fall through. + case '=': + var n = parseInt(param, 10); + if (isNaN(n) || n < 0) { + throw new Error('Invalid number in diff_fromDelta: ' + param); + } + var text = text1.substring(pointer, pointer += n); + if (tokens[x].charAt(0) == '=') { + diffs[diffsLength++] = new diff_match_patch.Diff(DIFF_EQUAL, text); + } else { + diffs[diffsLength++] = new diff_match_patch.Diff(DIFF_DELETE, text); + } + break; + default: + // Blank tokens are ok (from a trailing \t). + // Anything else is an error. + if (tokens[x]) { + throw new Error('Invalid diff operation in diff_fromDelta: ' + + tokens[x]); + } + } + } + if (pointer != text1.length) { + throw new Error('Delta length (' + pointer + + ') does not equal source text length (' + text1.length + ').'); + } + return diffs; +}; + +// MATCH FUNCTIONS + +/** + * Locate the best instance of 'pattern' in 'text' near 'loc'. + * @param {string} text The text to search. + * @param {string} pattern The pattern to search for. + * @param {number} loc The location to search around. + * @return {number} Best match index or -1. + */ +diff_match_patch.prototype.match_main = function (text, pattern, loc) { + // Check for null inputs. + if (text == null || pattern == null || loc == null) { + throw new Error('Cannot diff null input.'); + } + + loc = Math.max(0, Math.min(loc, text.length)); + if (text == pattern) { + // Shortcut (potentially not guaranteed by the algorithm) + return 0; + } else if (!text.length) { + // Nothing to match. + return -1; + } else if (text.substring(loc, loc + pattern.length) == pattern) { + // Perfect match at the perfect spot! (Includes case of null pattern) + return loc; + } else { + // Do a fuzzy compare. + return this.match_bitap_(text, pattern, loc); + } +}; + +/** + * Locate the best instance of 'pattern' in 'text' near 'loc' using the + * Bitap algorithm. + * @param {string} text The text to search. + * @param {string} pattern The pattern to search for. + * @param {number} loc The location to search around. + * @return {number} Best match index or -1. + * @private + */ +diff_match_patch.prototype.match_bitap_ = function (text, pattern, loc) { + if (pattern.length > this.Match_MaxBits) { + throw new Error('Pattern too long for this browser.'); + } + + // Initialise the alphabet. + var s = this.match_alphabet_(pattern); + + var dmp = this; // 'this' becomes 'window' in a closure. /** - * Determine if the suffix of one string is the prefix of another. - * @param {string} text1 First string. - * @param {string} text2 Second string. - * @return {number} The number of characters common to the end of the first - * string and the start of the second string. + * Compute and return the score for a match with e errors and x location. + * Accesses loc and pattern through being a closure. + * @param {number} e Number of errors in match. + * @param {number} x Location of match. + * @return {number} Overall score for match (0.0 = good, 1.0 = bad). * @private */ - DiffMatchPatch.prototype.diffCommonOverlap = function (text1, text2) { - // Cache the text lengths to prevent multiple calls. - const text1Length = text1.length; - const text2Length = text2.length; - - // Eliminate the null case. - if (text1Length === 0 || text2Length === 0) { - return 0; + function match_bitapScore_ (e, x) { + var accuracy = e / pattern.length; + var proximity = Math.abs(loc - x); + if (!dmp.Match_Distance) { + // Dodge divide by zero error. + return proximity ? 1.0 : accuracy; } + return accuracy + (proximity / dmp.Match_Distance); + } - // Truncate the longer string. - if (text1Length > text2Length) { - text1 = text1.substring(text1Length - text2Length); - } else if (text1Length < text2Length) { - text2 = text2.substring(0, text1Length); + // Highest score beyond which we give up. + var score_threshold = this.Match_Threshold; + // Is there a nearby exact match? (speedup) + var best_loc = text.indexOf(pattern, loc); + if (best_loc != -1) { + score_threshold = Math.min(match_bitapScore_(0, best_loc), score_threshold); + // What about in the other direction? (speedup) + best_loc = text.lastIndexOf(pattern, loc + pattern.length); + if (best_loc != -1) { + score_threshold = + Math.min(match_bitapScore_(0, best_loc), score_threshold); } - const textLength = Math.min(text1Length, text2Length); + } - // Quick check for the worst case. - if (text1 === text2) { - return textLength; + // Initialise the bit arrays. + var matchmask = 1 << (pattern.length - 1); + best_loc = -1; + + var bin_min, bin_mid; + var bin_max = pattern.length + text.length; + var last_rd; + for (var d = 0; d < pattern.length; d++) { + // Scan for the best match; each iteration allows for one more error. + // Run a binary search to determine how far from 'loc' we can stray at this + // error level. + bin_min = 0; + bin_mid = bin_max; + while (bin_min < bin_mid) { + if (match_bitapScore_(d, loc + bin_mid) <= score_threshold) { + bin_min = bin_mid; + } else { + bin_max = bin_mid; + } + bin_mid = Math.floor((bin_max - bin_min) / 2 + bin_min); } - - // Start by looking for a single character match - // and increase length until no match is found. - // Performance analysis: https://neil.fraser.name/news/2010/11/04/ - let best = 0; - let length = 1; - while (true) { - const pattern = text1.substring(textLength - length); - const found = text2.indexOf(pattern); - if (found === -1) { - return best; + // Use the result from this iteration as the maximum for the next. + bin_max = bin_mid; + var start = Math.max(1, loc - bin_mid + 1); + var finish = Math.min(loc + bin_mid, text.length) + pattern.length; + + var rd = Array(finish + 2); + rd[finish + 1] = (1 << d) - 1; + for (var j = finish; j >= start; j--) { + // The alphabet (s) is a sparse hash, so the following line generates + // warnings. + var charMatch = s[text.charAt(j - 1)]; + if (d === 0) { // First pass: exact match. + rd[j] = ((rd[j + 1] << 1) | 1) & charMatch; + } else { // Subsequent passes: fuzzy match. + rd[j] = (((rd[j + 1] << 1) | 1) & charMatch) | + (((last_rd[j + 1] | last_rd[j]) << 1) | 1) | + last_rd[j + 1]; } - length += found; - if (found === 0 || text1.substring(textLength - length) === - text2.substring(0, length)) { - best = length; - length++; + if (rd[j] & matchmask) { + var score = match_bitapScore_(d, j - 1); + // This match will almost certainly be better than any existing match. + // But check anyway. + if (score <= score_threshold) { + // Told you so. + score_threshold = score; + best_loc = j - 1; + if (best_loc > loc) { + // When passing loc, don't exceed our current distance from loc. + start = Math.max(1, 2 * loc - best_loc); + } else { + // Already passed loc, downhill from here on in. + break; + } + } } } - }; + // No hope for a (better) match at greater error levels. + if (match_bitapScore_(d + 1, loc) > score_threshold) { + break; + } + last_rd = rd; + } + return best_loc; +}; + +/** + * Initialise the alphabet for the Bitap algorithm. + * @param {string} pattern The text to encode. + * @return {!Object} Hash of character locations. + * @private + */ +diff_match_patch.prototype.match_alphabet_ = function (pattern) { + var s = {}; + for (var i = 0; i < pattern.length; i++) { + s[pattern.charAt(i)] = 0; + } + for (var i = 0; i < pattern.length; i++) { + s[pattern.charAt(i)] |= 1 << (pattern.length - i - 1); + } + return s; +}; + +// PATCH FUNCTIONS + +/** + * Increase the context until it is unique, + * but don't let the pattern expand beyond Match_MaxBits. + * @param {!diff_match_patch.patch_obj} patch The patch to grow. + * @param {string} text Source text. + * @private + */ +diff_match_patch.prototype.patch_addContext_ = function (patch, text) { + if (text.length == 0) { + return; + } + if (patch.start2 === null) { + throw Error('patch not initialized'); + } + var pattern = text.substring(patch.start2, patch.start2 + patch.length1); + var padding = 0; + + // Look for the first and last matches of pattern in text. If two different + // matches are found, increase the pattern length. + while (text.indexOf(pattern) != text.lastIndexOf(pattern) && + pattern.length < this.Match_MaxBits - this.Patch_Margin - + this.Patch_Margin) { + padding += this.Patch_Margin; + pattern = text.substring(patch.start2 - padding, + patch.start2 + patch.length1 + padding); + } + // Add one chunk for good luck. + padding += this.Patch_Margin; + + // Add the prefix. + var prefix = text.substring(patch.start2 - padding, patch.start2); + if (prefix) { + patch.diffs.unshift(new diff_match_patch.Diff(DIFF_EQUAL, prefix)); + } + // Add the suffix. + var suffix = text.substring(patch.start2 + patch.length1, + patch.start2 + patch.length1 + padding); + if (suffix) { + patch.diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, suffix)); + } + + // Roll back the start points. + patch.start1 -= prefix.length; + patch.start2 -= prefix.length; + // Extend the lengths. + patch.length1 += prefix.length + suffix.length; + patch.length2 += prefix.length + suffix.length; +}; + +/** + * Compute a list of patches to turn text1 into text2. + * Use diffs if provided, otherwise compute it ourselves. + * There are four ways to call this function, depending on what data is + * available to the caller: + * Method 1: + * a = text1, b = text2 + * Method 2: + * a = diffs + * Method 3 (optimal): + * a = text1, b = diffs + * Method 4 (deprecated, use method 3): + * a = text1, b = text2, c = diffs + * + * @param {string|!Array.} a text1 (methods 1,3,4) or + * Array of diff tuples for text1 to text2 (method 2). + * @param {string|!Array.=} opt_b text2 (methods 1,4) or + * Array of diff tuples for text1 to text2 (method 3) or undefined (method 2). + * @param {string|!Array.=} opt_c Array of diff tuples + * for text1 to text2 (method 4) or undefined (methods 1,2,3). + * @return {!Array.} Array of Patch objects. + */ +diff_match_patch.prototype.patch_make = function (a, opt_b, opt_c) { + var text1, diffs; + if (typeof a == 'string' && typeof opt_b == 'string' && + typeof opt_c == 'undefined') { + // Method 1: text1, text2 + // Compute diffs from text1 and text2. + text1 = /** @type {string} */(a); + diffs = this.diff_main(text1, /** @type {string} */(opt_b), true); + if (diffs.length > 2) { + this.diff_cleanupSemantic(diffs); + this.diff_cleanupEfficiency(diffs); + } + } else if (a && typeof a == 'object' && typeof opt_b == 'undefined' && + typeof opt_c == 'undefined') { + // Method 2: diffs + // Compute text1 from diffs. + diffs = /** @type {!Array.} */(a); + text1 = this.diff_text1(diffs); + } else if (typeof a == 'string' && opt_b && typeof opt_b == 'object' && + typeof opt_c == 'undefined') { + // Method 3: text1, diffs + text1 = /** @type {string} */(a); + diffs = /** @type {!Array.} */(opt_b); + } else if (typeof a == 'string' && typeof opt_b == 'string' && + opt_c && typeof opt_c == 'object') { + // Method 4: text1, text2, diffs + // text2 is not used. + text1 = /** @type {string} */(a); + diffs = /** @type {!Array.} */(opt_c); + } else { + throw new Error('Unknown call format to patch_make.'); + } - /** - * Split two texts into an array of strings. Reduce the texts to a string of - * hashes where each Unicode character represents one line. - * @param {string} text1 First string. - * @param {string} text2 Second string. - * @return {{chars1: string, chars2: string, lineArray: !Array.}} - * An object containing the encoded text1, the encoded text2 and - * the array of unique strings. - * The zeroth element of the array of unique strings is intentionally blank. - * @private - */ - DiffMatchPatch.prototype.diffLinesToChars = function (text1, text2) { - let lineArray = []; // E.g. lineArray[4] === 'Hello\n' - let lineHash = {}; // E.g. lineHash['Hello\n'] === 4 - - // '\x00' is a valid character, but various debuggers don't like it. - // So we'll insert a junk entry to avoid generating a null character. - lineArray[0] = ''; - - /** - * Split a text into an array of strings. Reduce the texts to a string of - * hashes where each Unicode character represents one line. - * Modifies linearray and linehash through being a closure. - * @param {string} text String to encode. - * @return {string} Encoded string. - * @private - */ - function diffLinesToCharsMunge (text) { - let chars = ''; - - // Walk the text, pulling out a substring for each line. - // text.split('\n') would would temporarily double our memory footprint. - // Modifying text would create many large strings to garbage collect. - let lineStart = 0; - let lineEnd = -1; - - // Keeping our own length variable is faster than looking it up. - let lineArrayLength = lineArray.length; - while (lineEnd < text.length - 1) { - lineEnd = text.indexOf('\n', lineStart); - if (lineEnd === -1) { - lineEnd = text.length - 1; + if (diffs.length === 0) { + return []; // Get rid of the null case. + } + var patches = []; + var patch = new diff_match_patch.patch_obj(); + var patchDiffLength = 0; // Keeping our own length var is faster in JS. + var char_count1 = 0; // Number of characters into the text1 string. + var char_count2 = 0; // Number of characters into the text2 string. + // Start with text1 (prepatch_text) and apply the diffs until we arrive at + // text2 (postpatch_text). We recreate the patches one by one to determine + // context info. + var prepatch_text = text1; + var postpatch_text = text1; + for (var x = 0; x < diffs.length; x++) { + var diff_type = diffs[x][0]; + var diff_text = diffs[x][1]; + + if (!patchDiffLength && diff_type !== DIFF_EQUAL) { + // A new patch starts here. + patch.start1 = char_count1; + patch.start2 = char_count2; + } + + switch (diff_type) { + case DIFF_INSERT: + patch.diffs[patchDiffLength++] = diffs[x]; + patch.length2 += diff_text.length; + postpatch_text = postpatch_text.substring(0, char_count2) + diff_text + + postpatch_text.substring(char_count2); + break; + case DIFF_DELETE: + patch.length1 += diff_text.length; + patch.diffs[patchDiffLength++] = diffs[x]; + postpatch_text = postpatch_text.substring(0, char_count2) + + postpatch_text.substring(char_count2 + + diff_text.length); + break; + case DIFF_EQUAL: + if (diff_text.length <= 2 * this.Patch_Margin && + patchDiffLength && diffs.length != x + 1) { + // Small equality inside a patch. + patch.diffs[patchDiffLength++] = diffs[x]; + patch.length1 += diff_text.length; + patch.length2 += diff_text.length; + } else if (diff_text.length >= 2 * this.Patch_Margin) { + // Time for a new patch. + if (patchDiffLength) { + this.patch_addContext_(patch, prepatch_text); + patches.push(patch); + patch = new diff_match_patch.patch_obj(); + patchDiffLength = 0; + // Unlike Unidiff, our patch lists have a rolling context. + // https://github.com/google/diff-match-patch/wiki/Unidiff + // Update prepatch text & pos to reflect the application of the + // just completed patch. + prepatch_text = postpatch_text; + char_count1 = char_count2; + } } - let line = text.substring(lineStart, lineEnd + 1); - lineStart = lineEnd + 1; + break; + } - if (hasOwn.call(lineHash, line)) { - chars += String.fromCharCode(lineHash[line]); + // Update the current character count. + if (diff_type !== DIFF_INSERT) { + char_count1 += diff_text.length; + } + if (diff_type !== DIFF_DELETE) { + char_count2 += diff_text.length; + } + } + // Pick up the leftover patch if not empty. + if (patchDiffLength) { + this.patch_addContext_(patch, prepatch_text); + patches.push(patch); + } + + return patches; +}; + +/** + * Given an array of patches, return another array that is identical. + * @param {!Array.} patches Array of Patch objects. + * @return {!Array.} Array of Patch objects. + */ +diff_match_patch.prototype.patch_deepCopy = function (patches) { + // Making deep copies is hard in JavaScript. + var patchesCopy = []; + for (var x = 0; x < patches.length; x++) { + var patch = patches[x]; + var patchCopy = new diff_match_patch.patch_obj(); + patchCopy.diffs = []; + for (var y = 0; y < patch.diffs.length; y++) { + patchCopy.diffs[y] = + new diff_match_patch.Diff(patch.diffs[y][0], patch.diffs[y][1]); + } + patchCopy.start1 = patch.start1; + patchCopy.start2 = patch.start2; + patchCopy.length1 = patch.length1; + patchCopy.length2 = patch.length2; + patchesCopy[x] = patchCopy; + } + return patchesCopy; +}; + +/** + * Merge a set of patches onto the text. Return a patched text, as well + * as a list of true/false values indicating which patches were applied. + * @param {!Array.} patches Array of Patch objects. + * @param {string} text Old text. + * @return {!Array.>} Two element Array, containing the + * new text and an array of boolean values. + */ +diff_match_patch.prototype.patch_apply = function (patches, text) { + if (patches.length == 0) { + return [text, []]; + } + + // Deep copy the patches so that no changes are made to originals. + patches = this.patch_deepCopy(patches); + + var nullPadding = this.patch_addPadding(patches); + text = nullPadding + text + nullPadding; + + this.patch_splitMax(patches); + // delta keeps track of the offset between the expected and actual location + // of the previous patch. If there are patches expected at positions 10 and + // 20, but the first patch was found at 12, delta is 2 and the second patch + // has an effective expected position of 22. + var delta = 0; + var results = []; + for (var x = 0; x < patches.length; x++) { + var expected_loc = patches[x].start2 + delta; + var text1 = this.diff_text1(patches[x].diffs); + var start_loc; + var end_loc = -1; + if (text1.length > this.Match_MaxBits) { + // patch_splitMax will only provide an oversized pattern in the case of + // a monster delete. + start_loc = this.match_main(text, text1.substring(0, this.Match_MaxBits), + expected_loc); + if (start_loc != -1) { + end_loc = this.match_main(text, + text1.substring(text1.length - this.Match_MaxBits), + expected_loc + text1.length - this.Match_MaxBits); + if (end_loc == -1 || start_loc >= end_loc) { + // Can't find valid trailing context. Drop this patch. + start_loc = -1; + } + } + } else { + start_loc = this.match_main(text, text1, expected_loc); + } + if (start_loc == -1) { + // No match found. :( + results[x] = false; + // Subtract the delta for this failed patch from subsequent patches. + delta -= patches[x].length2 - patches[x].length1; + } else { + // Found a match. :) + results[x] = true; + delta = start_loc - expected_loc; + var text2; + if (end_loc == -1) { + text2 = text.substring(start_loc, start_loc + text1.length); + } else { + text2 = text.substring(start_loc, end_loc + this.Match_MaxBits); + } + if (text1 == text2) { + // Perfect match, just shove the replacement text in. + text = text.substring(0, start_loc) + + this.diff_text2(patches[x].diffs) + + text.substring(start_loc + text1.length); + } else { + // Imperfect match. Run a diff to get a framework of equivalent + // indices. + var diffs = this.diff_main(text1, text2, false); + if (text1.length > this.Match_MaxBits && + this.diff_levenshtein(diffs) / text1.length > + this.Patch_DeleteThreshold) { + // The end points match, but the content is unacceptably bad. + results[x] = false; } else { - chars += String.fromCharCode(lineArrayLength); - lineHash[line] = lineArrayLength; - lineArray[lineArrayLength++] = line; + this.diff_cleanupSemanticLossless(diffs); + var index1 = 0; + var index2; + for (var y = 0; y < patches[x].diffs.length; y++) { + var mod = patches[x].diffs[y]; + if (mod[0] !== DIFF_EQUAL) { + index2 = this.diff_xIndex(diffs, index1); + } + if (mod[0] === DIFF_INSERT) { // Insertion + text = text.substring(0, start_loc + index2) + mod[1] + + text.substring(start_loc + index2); + } else if (mod[0] === DIFF_DELETE) { // Deletion + text = text.substring(0, start_loc + index2) + + text.substring(start_loc + this.diff_xIndex(diffs, + index1 + mod[1].length)); + } + if (mod[0] !== DIFF_DELETE) { + index1 += mod[1].length; + } + } } } - return chars; } + } + // Strip the padding off. + text = text.substring(nullPadding.length, text.length - nullPadding.length); + return [text, results]; +}; + +/** + * Add some padding on text start and end so that edges can match something. + * Intended to be called only from within patch_apply. + * @param {!Array.} patches Array of Patch objects. + * @return {string} The padding string added to each side. + */ +diff_match_patch.prototype.patch_addPadding = function (patches) { + var paddingLength = this.Patch_Margin; + var nullPadding = ''; + for (var x = 1; x <= paddingLength; x++) { + nullPadding += String.fromCharCode(x); + } - const chars1 = diffLinesToCharsMunge(text1); - const chars2 = diffLinesToCharsMunge(text2); - return { - chars1: chars1, - chars2: chars2, - lineArray: lineArray - }; - }; + // Bump all the patches forward. + for (var x = 0; x < patches.length; x++) { + patches[x].start1 += paddingLength; + patches[x].start2 += paddingLength; + } - /** - * Rehydrate the text in a diff from a string of line hashes to real lines of - * text. - * @param {!Array.} diffs Array of diff tuples. - * @param {!Array.} lineArray Array of unique strings. - * @private - */ - DiffMatchPatch.prototype.diffCharsToLines = function (diffs, lineArray) { - for (let x = 0; x < diffs.length; x++) { - const chars = diffs[x][1]; - const text = []; - for (let y = 0; y < chars.length; y++) { - text[y] = lineArray[chars.charCodeAt(y)]; - } - diffs[x][1] = text.join(''); - } - }; + // Add some padding on start of first diff. + var patch = patches[0]; + var diffs = patch.diffs; + if (diffs.length == 0 || diffs[0][0] != DIFF_EQUAL) { + // Add nullPadding equality. + diffs.unshift(new diff_match_patch.Diff(DIFF_EQUAL, nullPadding)); + patch.start1 -= paddingLength; // Should be 0. + patch.start2 -= paddingLength; // Should be 0. + patch.length1 += paddingLength; + patch.length2 += paddingLength; + } else if (paddingLength > diffs[0][1].length) { + // Grow first equality. + var extraLength = paddingLength - diffs[0][1].length; + diffs[0][1] = nullPadding.substring(diffs[0][1].length) + diffs[0][1]; + patch.start1 -= extraLength; + patch.start2 -= extraLength; + patch.length1 += extraLength; + patch.length2 += extraLength; + } - /** - * Reorder and merge like edit sections. Merge equalities. - * Any edit section can move as long as it doesn't cross an equality. - * @param {!Array.} diffs Array of diff tuples. - */ - DiffMatchPatch.prototype.diffCleanupMerge = function (diffs) { - diffs.push([DIFF_EQUAL, '']); // Add a dummy entry at the end. - let pointer = 0; - let countDelete = 0; - let countInsert = 0; - let textDelete = ''; - let textInsert = ''; - - while (pointer < diffs.length) { - switch (diffs[pointer][0]) { - case DIFF_INSERT: - countInsert++; - textInsert += diffs[pointer][1]; - pointer++; - break; - case DIFF_DELETE: - countDelete++; - textDelete += diffs[pointer][1]; - pointer++; - break; - case DIFF_EQUAL: - - // Upon reaching an equality, check for prior redundancies. - if (countDelete + countInsert > 1) { - if (countDelete !== 0 && countInsert !== 0) { - // Factor out any common prefixes. - let commonlength = this.diffCommonPrefix(textInsert, textDelete); - if (commonlength !== 0) { - if ((pointer - countDelete - countInsert) > 0 && - diffs[pointer - countDelete - countInsert - 1][0] === - DIFF_EQUAL) { - diffs[pointer - countDelete - countInsert - 1][1] += - textInsert.substring(0, commonlength); - } else { - diffs.splice(0, 0, [DIFF_EQUAL, - textInsert.substring(0, commonlength) - ]); - pointer++; - } - textInsert = textInsert.substring(commonlength); - textDelete = textDelete.substring(commonlength); - } + // Add some padding on end of last diff. + patch = patches[patches.length - 1]; + diffs = patch.diffs; + if (diffs.length == 0 || diffs[diffs.length - 1][0] != DIFF_EQUAL) { + // Add nullPadding equality. + diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, nullPadding)); + patch.length1 += paddingLength; + patch.length2 += paddingLength; + } else if (paddingLength > diffs[diffs.length - 1][1].length) { + // Grow last equality. + var extraLength = paddingLength - diffs[diffs.length - 1][1].length; + diffs[diffs.length - 1][1] += nullPadding.substring(0, extraLength); + patch.length1 += extraLength; + patch.length2 += extraLength; + } - // Factor out any common suffixies. - commonlength = this.diffCommonSuffix(textInsert, textDelete); - if (commonlength !== 0) { - diffs[pointer][1] = textInsert.substring(textInsert.length - - commonlength) + diffs[pointer][1]; - textInsert = textInsert.substring(0, textInsert.length - - commonlength); - textDelete = textDelete.substring(0, textDelete.length - - commonlength); - } - } + return nullPadding; +}; - // Delete the offending records and add the merged ones. - if (countDelete === 0) { - diffs.splice(pointer - countInsert, - countDelete + countInsert, [DIFF_INSERT, textInsert]); - } else if (countInsert === 0) { - diffs.splice(pointer - countDelete, - countDelete + countInsert, [DIFF_DELETE, textDelete]); - } else { - diffs.splice( - pointer - countDelete - countInsert, - countDelete + countInsert, - [DIFF_DELETE, textDelete], [DIFF_INSERT, textInsert] - ); - } - pointer = pointer - countDelete - countInsert + - (countDelete ? 1 : 0) + (countInsert ? 1 : 0) + 1; - } else if (pointer !== 0 && diffs[pointer - 1][0] === DIFF_EQUAL) { - // Merge this equality with the previous one. - diffs[pointer - 1][1] += diffs[pointer][1]; - diffs.splice(pointer, 1); +/** + * Look through the patches and break up any which are longer than the maximum + * limit of the match algorithm. + * Intended to be called only from within patch_apply. + * @param {!Array.} patches Array of Patch objects. + */ +diff_match_patch.prototype.patch_splitMax = function (patches) { + var patch_size = this.Match_MaxBits; + for (var x = 0; x < patches.length; x++) { + if (patches[x].length1 <= patch_size) { + continue; + } + var bigpatch = patches[x]; + // Remove the big old patch. + patches.splice(x--, 1); + var start1 = bigpatch.start1; + var start2 = bigpatch.start2; + var precontext = ''; + while (bigpatch.diffs.length !== 0) { + // Create one of several smaller patches. + var patch = new diff_match_patch.patch_obj(); + var empty = true; + patch.start1 = start1 - precontext.length; + patch.start2 = start2 - precontext.length; + if (precontext !== '') { + patch.length1 = patch.length2 = precontext.length; + patch.diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, precontext)); + } + while (bigpatch.diffs.length !== 0 && + patch.length1 < patch_size - this.Patch_Margin) { + var diff_type = bigpatch.diffs[0][0]; + var diff_text = bigpatch.diffs[0][1]; + if (diff_type === DIFF_INSERT) { + // Insertions are harmless. + patch.length2 += diff_text.length; + start2 += diff_text.length; + patch.diffs.push(bigpatch.diffs.shift()); + empty = false; + } else if (diff_type === DIFF_DELETE && patch.diffs.length == 1 && + patch.diffs[0][0] == DIFF_EQUAL && + diff_text.length > 2 * patch_size) { + // This is a large deletion. Let it pass in one chunk. + patch.length1 += diff_text.length; + start1 += diff_text.length; + empty = false; + patch.diffs.push(new diff_match_patch.Diff(diff_type, diff_text)); + bigpatch.diffs.shift(); + } else { + // Deletion or equality. Only take as much as we can stomach. + diff_text = diff_text.substring(0, + patch_size - patch.length1 - this.Patch_Margin); + patch.length1 += diff_text.length; + start1 += diff_text.length; + if (diff_type === DIFF_EQUAL) { + patch.length2 += diff_text.length; + start2 += diff_text.length; } else { - pointer++; + empty = false; } - countInsert = 0; - countDelete = 0; - textDelete = ''; - textInsert = ''; - break; + patch.diffs.push(new diff_match_patch.Diff(diff_type, diff_text)); + if (diff_text == bigpatch.diffs[0][1]) { + bigpatch.diffs.shift(); + } else { + bigpatch.diffs[0][1] = + bigpatch.diffs[0][1].substring(diff_text.length); + } + } + } + // Compute the head context for the next patch. + precontext = this.diff_text2(patch.diffs); + precontext = + precontext.substring(precontext.length - this.Patch_Margin); + // Append the end context for this patch. + var postcontext = this.diff_text1(bigpatch.diffs) + .substring(0, this.Patch_Margin); + if (postcontext !== '') { + patch.length1 += postcontext.length; + patch.length2 += postcontext.length; + if (patch.diffs.length !== 0 && + patch.diffs[patch.diffs.length - 1][0] === DIFF_EQUAL) { + patch.diffs[patch.diffs.length - 1][1] += postcontext; + } else { + patch.diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, postcontext)); + } + } + if (!empty) { + patches.splice(++x, 0, patch); } } - if (diffs[diffs.length - 1][1] === '') { - diffs.pop(); // Remove the dummy entry at the end. + } +}; + +/** + * Take a list of patches and return a textual representation. + * @param {!Array.} patches Array of Patch objects. + * @return {string} Text representation of patches. + */ +diff_match_patch.prototype.patch_toText = function (patches) { + var text = []; + for (var x = 0; x < patches.length; x++) { + text[x] = patches[x]; + } + return text.join(''); +}; + +/** + * Parse a textual representation of patches and return a list of Patch objects. + * @param {string} textline Text representation of patches. + * @return {!Array.} Array of Patch objects. + * @throws {!Error} If invalid input. + */ +diff_match_patch.prototype.patch_fromText = function (textline) { + var patches = []; + if (!textline) { + return patches; + } + var text = textline.split('\n'); + var textPointer = 0; + var patchHeader = /^@@ -(\d+),?(\d*) \+(\d+),?(\d*) @@$/; + while (textPointer < text.length) { + var m = text[textPointer].match(patchHeader); + if (!m) { + throw new Error('Invalid patch string: ' + text[textPointer]); + } + var patch = new diff_match_patch.patch_obj(); + patches.push(patch); + patch.start1 = parseInt(m[1], 10); + if (m[2] === '') { + patch.start1--; + patch.length1 = 1; + } else if (m[2] == '0') { + patch.length1 = 0; + } else { + patch.start1--; + patch.length1 = parseInt(m[2], 10); } - // Second pass: look for single edits surrounded on both sides by equalities - // which can be shifted sideways to eliminate an equality. - // e.g: ABAC -> ABAC - let changes = false; - pointer = 1; - - // Intentionally ignore the first and last element (don't need checking). - while (pointer < diffs.length - 1) { - if (diffs[pointer - 1][0] === DIFF_EQUAL && - diffs[pointer + 1][0] === DIFF_EQUAL) { - const diffPointer = diffs[pointer][1]; - const position = diffPointer.substring( - diffPointer.length - diffs[pointer - 1][1].length - ); - - // This is a single edit surrounded by equalities. - if (position === diffs[pointer - 1][1]) { - // Shift the edit over the previous equality. - diffs[pointer][1] = diffs[pointer - 1][1] + - diffs[pointer][1].substring(0, diffs[pointer][1].length - - diffs[pointer - 1][1].length); - diffs[pointer + 1][1] = - diffs[pointer - 1][1] + diffs[pointer + 1][1]; - diffs.splice(pointer - 1, 1); - changes = true; - } else if (diffPointer.substring(0, diffs[pointer + 1][1].length) === - diffs[pointer + 1][1]) { - // Shift the edit over the next equality. - diffs[pointer - 1][1] += diffs[pointer + 1][1]; - diffs[pointer][1] = - diffs[pointer][1].substring(diffs[pointer + 1][1].length) + - diffs[pointer + 1][1]; - diffs.splice(pointer + 1, 1); - changes = true; - } + patch.start2 = parseInt(m[3], 10); + if (m[4] === '') { + patch.start2--; + patch.length2 = 1; + } else if (m[4] == '0') { + patch.length2 = 0; + } else { + patch.start2--; + patch.length2 = parseInt(m[4], 10); + } + textPointer++; + + while (textPointer < text.length) { + var sign = text[textPointer].charAt(0); + try { + var line = decodeURI(text[textPointer].substring(1)); + } catch (ex) { + // Malformed URI sequence. + throw new Error('Illegal escape in patch_fromText: ' + line); } - pointer++; + if (sign == '-') { + // Deletion. + patch.diffs.push(new diff_match_patch.Diff(DIFF_DELETE, line)); + } else if (sign == '+') { + // Insertion. + patch.diffs.push(new diff_match_patch.Diff(DIFF_INSERT, line)); + } else if (sign == ' ') { + // Minor equality. + patch.diffs.push(new diff_match_patch.Diff(DIFF_EQUAL, line)); + } else if (sign == '@') { + // Start of next patch. + break; + } else if (sign === '') { + // Blank line? Whatever. + } else { + // WTF? + throw new Error('Invalid patch mode "' + sign + '" in: ' + line); + } + textPointer++; } + } + return patches; +}; - // If shifts were made, the diff needs reordering and another shift sweep. - if (changes) { - this.diffCleanupMerge(diffs); +/** + * Class representing one patch operation. + * @constructor + */ +diff_match_patch.patch_obj = function () { + /** @type {!Array.} */ + this.diffs = []; + /** @type {?number} */ + this.start1 = null; + /** @type {?number} */ + this.start2 = null; + /** @type {number} */ + this.length1 = 0; + /** @type {number} */ + this.length2 = 0; +}; + +/** + * Emulate GNU diff's format. + * Header: @@ -382,8 +481,9 @@ + * Indices are printed as 1-based, not 0-based. + * @return {string} The GNU diff string. + */ +diff_match_patch.patch_obj.prototype.toString = function () { + var coords1, coords2; + if (this.length1 === 0) { + coords1 = this.start1 + ',0'; + } else if (this.length1 == 1) { + coords1 = this.start1 + 1; + } else { + coords1 = (this.start1 + 1) + ',' + this.length1; + } + if (this.length2 === 0) { + coords2 = this.start2 + ',0'; + } else if (this.length2 == 1) { + coords2 = this.start2 + 1; + } else { + coords2 = (this.start2 + 1) + ',' + this.length2; + } + var text = ['@@ -' + coords1 + ' +' + coords2 + ' @@\n']; + var op; + // Escape the body of the patch with %xx notation. + for (var x = 0; x < this.diffs.length; x++) { + switch (this.diffs[x][0]) { + case DIFF_INSERT: + op = '+'; + break; + case DIFF_DELETE: + op = '-'; + break; + case DIFF_EQUAL: + op = ' '; + break; } - }; + text[x + 1] = op + encodeURI(this.diffs[x][1]) + '\n'; + } + return text.join('').replace(/%20/g, ' '); +}; + +/** + * Inline word diff, formatted as HTML. + * + * @param {string} before + * @param {string} after + * @return {string} HTML with , , and
+ */ +export function diffHtml (before, after) { + const dmp = new diff_match_patch(); + const diffs = dmp.diff_main(before, after); + dmp.diff_cleanupSemantic(diffs); + dmp.diff_cleanupEfficiency(diffs); - export default function diff (o, n) { - let diff, output, text; - diff = new DiffMatchPatch(); - output = diff.DiffMain(o, n); - diff.diffCleanupEfficiency(output); - text = diff.diffPrettyHtml(output); + return dmp.diff_prettyHtml(diffs); +} - return text; +/** + * Line diff, plain text. + * + * @param {string} before + * @param {string} after + * @return {string} Patch + */ +export function diffText (before, after) { + // https://github.com/google/diff-match-patch/wiki/Line-or-Word-Diffs + const dmp = new diff_match_patch(); + const x = dmp.diff_linesToChars_(before, after); + const lineText1 = x.chars1; + const lineText2 = x.chars2; + const lineArray = x.lineArray; + const diffs = dmp.diff_main(lineText1, lineText2, false); + dmp.diff_charsToLines_(diffs, lineArray); + + let text = ''; + for (let i = 0; i < diffs.length; i++) { + const op = diffs[i][0]; + const data = diffs[i][1]; + // Strip last newline since each block already ends with a newline. + // Otherwise, delete/insert boundaries would be formatted as: + // + // -foo + // -+bar + const isDataEndsLn = (data.slice(-1) === '\n'); + const dataWithoutLn = isDataEndsLn ? data.slice(0, -1) : data; + switch (op) { + case DIFF_INSERT: + text += '+' + dataWithoutLn.replace(/\n/g, '\n+'); + break; + case DIFF_DELETE: + text += '-' + dataWithoutLn.replace(/\n/g, '\n-'); + break; + case DIFF_EQUAL: + text += ' ' + dataWithoutLn.replace(/\n/g, '\n '); + break; + } + // Restore newline + text += isDataEndsLn ? '\n' : ''; } + return text; +} diff --git a/src/core/reporters/HtmlReporter.js b/src/core/reporters/HtmlReporter.js index fdeff09b5..263490de9 100644 --- a/src/core/reporters/HtmlReporter.js +++ b/src/core/reporters/HtmlReporter.js @@ -1,5 +1,5 @@ import { extend, errorString, escapeText } from '../utilities.js'; -import diff from '../diff.js'; +import { diffHtml as diff } from '../diff.js'; import dump from '../dump.js'; import { prioritySymbol } from '../events.js'; import { window, document, navigator, StringMap } from '../globals.js'; @@ -116,14 +116,6 @@ function getUrlConfigHtml (config) { return urlConfigHtml; } -function stripHtml (string) { - // Strip tags, html entity and whitespaces - return string - .replace(/<\/?[^>]+(>|$)/g, '') - .replace(/"/g, '') - .replace(/\s+/g, ''); -} - export default class HtmlReporter { /** * @param {QUnit} QUnit @@ -870,10 +862,8 @@ export default class HtmlReporter { ) { diffHtml = diff(expected, actual); - // don't show diff if there is zero overlap - showDiff = stripHtml(diffHtml).length !== - stripHtml(expected).length + - stripHtml(actual).length; + // don't show diff if there is zero overlap (i.e. no context 's) + showDiff = diff.indexOf('') !== -1; } if (showDiff) { diff --git a/src/core/reporters/TapReporter.js b/src/core/reporters/TapReporter.js index 326d784ff..65cb8dadc 100644 --- a/src/core/reporters/TapReporter.js +++ b/src/core/reporters/TapReporter.js @@ -1,9 +1,14 @@ import kleur from 'kleur'; -import { errorString } from '../utilities.js'; + +import { diffText } from '../diff.js'; +import dump from '../dump.js'; import { console } from '../globals.js'; +import { errorString } from '../utilities.js'; const hasOwn = Object.prototype.hasOwnProperty; +const TAP_YAML_INDENT = 4; + /** * Format a given value into YAML. * @@ -36,7 +41,7 @@ const hasOwn = Object.prototype.hasOwnProperty; * Objects with cyclical references will be stringifed as * "[Circular]" as they cannot otherwise be represented. */ -function prettyYamlValue (value, indent = 4) { +function prettyYamlValue (value, indent = TAP_YAML_INDENT) { if (value === undefined) { // Not supported in JSON/YAML, turn into string // and let the below output it as bare string. @@ -261,6 +266,26 @@ export default class TapReporter { out += `\n expected: ${prettyYamlValue(error.expected)}`; } + if (hasOwn.call(error, 'actual') && hasOwn.call(error, 'expected')) { + const diff = diffText(dump.parse(error.expected), dump.parse(error.actual)); + // Only show diff if there is both overlap and differences, + // otherwise the actual/expected blocks suffice. + if (/^ /m.test(diff) && /^[-+]/m.test(diff)) { + const prefix = (new Array(TAP_YAML_INDENT + 1)).join(' '); + // YAML does not permit simple multi-line block values to start with + // a space, as that makes for ambigious indentation. + const diffFormatted = '--- expected\n+++ actual\n' + diff + // Highlight + .replace(/^(-[^\n]+)/mg, kleur.red('$1')) + .replace(/^(\+[^\n]+)/mg, kleur.green('$1')); + const diffIndented = diffFormatted + .split('\n') + .map(line => prefix + line) + .join('\n'); + out += `\n diff: |\n${diffIndented}`; + } + } + if (error.stack) { // Since stacks aren't user generated, take a bit of liberty by // adding a trailing new line to allow a straight-forward YAML Blocks. diff --git a/test/main/TapReporter.js b/test/main/TapReporter.js index 9ec70355c..d9ea86cfa 100644 --- a/test/main/TapReporter.js +++ b/test/main/TapReporter.js @@ -6,7 +6,7 @@ function mockStack (error) { return error; } -function makeFailingTestEnd (actualValue) { +function makeFailingTestEnd (actualValue, expectedValue) { return { name: 'Failing', suiteName: null, @@ -16,7 +16,7 @@ function makeFailingTestEnd (actualValue) { errors: [{ passed: false, actual: actualValue, - expected: 'expected' + expected: expectedValue || 'expected' }], assertions: null }; @@ -418,6 +418,37 @@ QUnit.module('TapReporter', function (hooks) { ); }); + QUnit.test('output diff for non-trivial differences', function (assert) { + emitter.emit('testEnd', makeFailingTestEnd( + { a: true, b: 'x' }, + { c: false, b: 'x' } + )); + // eslint-disable-next-line no-control-regex + var lastDekleur = last.replace(/\x1b\[\d+m/g, ''); + assert.strictEqual(lastDekleur, ' ---\n' + +' message: failed\n' + +' severity: failed\n' + +' actual : {\n' + +' "a": true,\n' + +' "b": "x"\n' + +'}\n' + +' expected: {\n' + +' "c": false,\n' + +' "b": "x"\n' + +'}\n' + +' diff: |\n' + +' --- expected\n' + +' +++ actual\n' + +' {\n' + +' - "b": "x",\n' + +' - "c": false\n' + +' + "a": true,\n' + +' + "b": "x"\n' + +' }\n' + +' ...' + ); + }); + QUnit.test('output the total number of tests', function (assert) { emitter.emit('runEnd', { testCounts: { diff --git a/test/main/diff.js b/test/main/diff.js index c1f450a73..a1c12a22a 100644 --- a/test/main/diff.js +++ b/test/main/diff.js @@ -45,13 +45,13 @@ QUnit.test('different strings', function (assert) { assert.equal( QUnit.diff(a, 'acd'), - 'a<b/>cd', + 'ab<b/>cd', "QUnit.diff( 'abcd', 'acd' )" ); assert.equal( QUnit.diff(a, 'a<b/>cd'), - 'a&lt;b/&gt;cd', + 'ab&lt;b/&gt;cd', "QUnit.diff( 'abcd', 'a<b/>cd' )" ); }); @@ -104,7 +104,7 @@ QUnit.test('test with line mode on long strings', function (assert) { assert.equal( QUnit.diff(a, b), 'QUnit is a very powerful, easy-to-use ' + - 'JavaScript unit testing framework. It's used by the jQuery ' + + 'JavaScript unit testing framework. It\'s used by the jQuery ' + 'Core, jQuery UI and jQuery Mobile projects and is capable of' + ' testing any generic JavaScript code, including ' + 'itself!' + @@ -158,7 +158,7 @@ QUnit.test('simplified diffs', function (assert) { 'foobarbaz01' + 'foobarbaz01' + 'foobarbaz more than 100 chars\n' + - 'second_lifferent_end', + 'second_linedifferent_end', 'simplify longer multi-line diffs' ); }); @@ -188,9 +188,10 @@ QUnit.test('Edge cases', function (assert) { assert.equal( QUnit.diff( 'A\nB\n' + X + '\nD', - 'A\nCCC\nB\nCCC\n' + Y + '\nD'), - 'A\nCCC\nB\n' + - '' + X + 'CCC\n' + Y + '' + + 'A\nCCC\nB\nCCC\n' + Y + '\nD' + ), + 'A\nB\n' + X + '' + + 'CCC\nB\nCCC\n' + Y + '' + '\nD'); function repeat (substr, n) {