feat: new encoder (wip)

README.md

@@ -93,7 +93,8 @@ Mergely will emit an `updated` event when the editor is first initialized, and e
 |<a name="ignorews"></a>ignorews|boolean|`false`|Ignores white-space.|
 |<a name="ignorecase"></a>ignorecase|boolean|`false`|Ignores case.|
 |<a name="ignoreaccents"></a>ignoreaccents|boolean|`false`|Ignores accented characters.|
-|<a name="lcs"></a>lcs|boolean|`true`|Enables/disables LCS computation for paragraphs (char-by-char changes). Disabling can give a performance gain for large documents.|
+|<a name="inline"></a>inline|string|`chars`|The line-by-line (inline) type of diff. Valid values are: `none`, `chars`, `words`. When `none`, inline diff is disabled. When `chars` differientiation is done on a character-by-character basis. When `words` differentiation is done on a whitespace basis.|
+|<a name="lcs"></a>lcs|boolean|`true`|:warning: **Deprecated**, use [`inline`](#inline). Enables/disables LCS computation for paragraphs (char-by-char changes). Disabling can give a performance gain for large documents.|
 |<a name="lhs"></a>lhs|boolean,`function handler(setValue)`|`null`|Sets the value of the editor on the left-hand side.|
 |<a name="license"></a>license|string|`lgpl`|The choice of license to use with Mergely.  Valid values are: `lgpl`, `gpl`, `mpl` or `lgpl-separate-notice`, `gpl-separate-notice`, `mpl-separate-notice` (the license requirements are met in a separate notice file).|
 |<a name="line_numbers"></a>line_numbers|boolean|`true`|Enables/disables line numbers. Enabling line numbers will toggle the visibility of the line number margins.|

examples/app.js

@@ -4,15 +4,9 @@ require('codemirror/addon/selection/mark-selection.js');
 require('codemirror/lib/codemirror.css');
 require('../src/mergely.css');
 
-const lhs = `\
-the quick red fox
-jumped over the hairy dog
-`;
+const lhs = `hello`;
 
-const rhs = `\
-the quick brown fox
-jumped over the lazy dog
-`;
+const rhs = `hello\ngoodbye`;
 
 
 document.onreadystatechange = function () {
@@ -22,6 +16,7 @@ document.onreadystatechange = function () {
 
 	const mergely = new Mergely('#compare', {
 		license: 'lgpl',
+		inline: 'words',
 		lhs,
 		rhs
 	});

examples/change-styles.html

@@ -304,7 +304,6 @@ dog
 and the postman
 `
 }];
-		console.log(data.length);
 		for (let i = 0; i < data.length; ++i) {
 			const { lhs, rhs } = data[i];
 			const darkModeOptions = i === 11 ? {

examples/editor.html

@@ -298,7 +298,6 @@
 				rhs.style = 'color:initial';
 			});
 			jQuery('#search-text').on('keypress', (ev) => {
-				console.log(ev.which)
 				if (event.which === 13) {
 					ev.preventDefault();
 					jQuery('#search').click();

src/diff-view.js

@@ -941,7 +941,6 @@ CodeMirrorDiffView.prototype._markupLineChanges = function (changes) {
 	for (let i = 0; i < changes.length; ++i) {
 		const change = changes[i];
 		const isCurrent = current_diff === i;
-		const lineDiff = this.settings.lcs !== false;
 		const lhsInView = this._isChangeInView('lhs', lhsvp, change);
 		const rhsInView = this._isChangeInView('rhs', rhsvp, change);
 
@@ -952,7 +951,7 @@ CodeMirrorDiffView.prototype._markupLineChanges = function (changes) {
 
 			vdoc.addRender('lhs', change, i, {
 				isCurrent,
-				lineDiff,
+				lineDiff: this.settings.inline !== 'none',
 				// TODO: move out of loop
 				getMergeHandler: (change, side, oside) => {
 					return () => this._merge_change(change, side, oside);
@@ -969,7 +968,7 @@ CodeMirrorDiffView.prototype._markupLineChanges = function (changes) {
 
 			vdoc.addRender('rhs', change, i, {
 				isCurrent,
-				lineDiff,
+				lineDiff: this.settings.inline !== 'none',
 				// TODO: move out of loop
 				getMergeHandler: (change, side, oside) => {
 					return () => this._merge_change(change, side, oside);
@@ -979,13 +978,14 @@ CodeMirrorDiffView.prototype._markupLineChanges = function (changes) {
 			});
 		}
 
-		if (lineDiff
+		if (this.settings.inline !== 'none'
 			&& (lhsInView || rhsInView)
 			&& change.op === 'c') {
 			vdoc.addInlineDiff(change, i, {
 				ignoreaccents: this.settings.ignoreaccents,
 				ignorews: this.settings.ignorews,
 				ignorecase: this.settings.ignorecase,
+				split: this.settings.inline,
 				getText: (side, lineNum) => {
 					if (side === 'lhs') {
 						const text = led.getLine(lineNum);

src/diff.js

@@ -1,43 +1,52 @@
+const Encoder = require('./encoder.js');
+
 const SMS_TIMEOUT_SECONDS = 1.0;
 
-function diff(lhs, rhs, options = {}) {
+function diff(lhs, rhs, opts) {
 	const {
 		ignorews = false,
 		ignoreaccents = false,
 		ignorecase = false,
 		split = 'lines'
-	} = options;
-
-	this.codeify = new CodeifyText(lhs, rhs, {
+	} = opts || {};
+	const options = {
 		ignorews,
 		ignoreaccents,
 		ignorecase,
 		split
-	});
-	const lhs_ctx = {
-		codes: this.codeify.getCodes('lhs'),
+	};
+
+	const encoder = new Encoder();
+	const lhsCodes = encoder.encode(lhs, options);
+	const rhsCodes = encoder.encode(rhs, options);
+	const lhsCtx = {
+		codes: lhsCodes.codes,
+		length: lhsCodes.length,
+		parts: lhsCodes.parts,
 		modified: {}
 	};
-	const rhs_ctx = {
-		codes: this.codeify.getCodes('rhs'),
+	const rhsCtx = {
+		codes: rhsCodes.codes,
+		length: rhsCodes.length,
+		parts: rhsCodes.parts,
 		modified: {}
 	};
 	const vector_d = [];
 	const vector_u = [];
-	this._lcs(lhs_ctx, 0, lhs_ctx.codes.length, rhs_ctx, 0, rhs_ctx.codes.length, vector_u, vector_d);
-	this._optimize(lhs_ctx);
-	this._optimize(rhs_ctx);
-	this.items = this._create_diffs(lhs_ctx, rhs_ctx);
+	this._lcs(lhsCtx, 0, lhsCodes.length, rhsCtx, 0, rhsCodes.length, vector_u, vector_d);
+	this._optimize(lhsCtx);
+	this._optimize(rhsCtx);
+	this.items = this._create_diffs(lhsCtx, rhsCtx, options);
+	this.sides = {
+		lhs: lhsCtx,
+		rhs: rhsCtx
+	};
 };
 
 diff.prototype.changes = function() {
 	return this.items;
 };
 
-diff.prototype.getLines = function(side) {
-	return this.codeify.getLines(side);
-};
-
 diff.prototype.normal_form = function() {
 	let nf = '';
 	for (let index = 0; index < this.items.length; ++index) {
@@ -57,8 +66,8 @@ diff.prototype.normal_form = function() {
 		else rhs_str = (item.rhs_start + 1) + ',' + (item.rhs_start + item.rhs_inserted_count);
 		nf += lhs_str + change + rhs_str + '\n';
 
-		const lhs_lines = this.getLines('lhs');
-		const rhs_lines = this.getLines('rhs');
+		const lhs_lines = this.sides.lhs.parts;
+		const rhs_lines = this.sides.rhs.parts;
 		if (rhs_lines && lhs_lines) {
 			let i;
 			// if rhs/lhs lines have been retained, output contextual diff
@@ -102,7 +111,7 @@ diff.prototype._lcs = function(lhs_ctx, lhs_lower, lhs_upper, rhs_ctx, rhs_lower
 
 diff.prototype._sms = function(lhs_ctx, lhs_lower, lhs_upper, rhs_ctx, rhs_lower, rhs_upper, vector_u, vector_d) {
 	const timeout = Date.now() + SMS_TIMEOUT_SECONDS * 1000;
-	const max = lhs_ctx.codes.length + rhs_ctx.codes.length + 1;
+	const max = lhs_ctx.length + rhs_ctx.length + 1;
 	const kdown = lhs_lower - rhs_lower;
 	const kup = lhs_upper - rhs_upper;
 	const delta = (lhs_upper - lhs_lower) - (rhs_upper - rhs_lower);
@@ -115,12 +124,13 @@ diff.prototype._sms = function(lhs_ctx, lhs_lower, lhs_upper, rhs_ctx, rhs_lower
 	const ret = { x:0, y:0 }
 	let x;
 	let y;
+	let k;
 	for (let d = 0; d <= maxd; ++d) {
 		if (SMS_TIMEOUT_SECONDS && Date.now() > timeout) {
 			// bail if taking too long
 			return { x: lhs_lower, y: rhs_upper };
 		}
-		for (let k = kdown - d; k <= kdown + d; k += 2) {
+		for (k = kdown - d; k <= kdown + d; k += 2) {
 			if (k === kdown - d) {
 				x = vector_d[ offset_down + k + 1 ];//down
 			}
@@ -178,15 +188,15 @@ diff.prototype._sms = function(lhs_ctx, lhs_lower, lhs_upper, rhs_ctx, rhs_lower
 diff.prototype._optimize = function(ctx) {
 	let start = 0;
 	let end = 0;
-	while (start < ctx.codes.length) {
-		while ((start < ctx.codes.length) && (ctx.modified[start] === undefined || ctx.modified[start] === false)) {
+	while (start < ctx.length) {
+		while ((start < ctx.length) && (ctx.modified[start] === undefined || ctx.modified[start] === false)) {
 			start++;
 		}
 		end = start;
-		while ((end < ctx.codes.length) && (ctx.modified[end] === true)) {
+		while ((end < ctx.length) && (ctx.modified[end] === true)) {
 			end++;
 		}
-		if ((end < ctx.codes.length) && (ctx.codes[start] === ctx.codes[end])) {
+		if ((end < ctx.length) && (ctx.codes[start] === ctx.codes[end])) {
 			ctx.modified[start] = false;
 			ctx.modified[end] = true;
 		}
@@ -196,16 +206,16 @@ diff.prototype._optimize = function(ctx) {
 	}
 };
 
-diff.prototype._create_diffs = function(lhs_ctx, rhs_ctx) {
+diff.prototype._create_diffs = function(lhs_ctx, rhs_ctx, options) {
 	const items = [];
 	let lhs_start = 0;
 	let rhs_start = 0;
 	let lhs_line = 0;
 	let rhs_line = 0;
 
-	while (lhs_line < lhs_ctx.codes.length || rhs_line < rhs_ctx.codes.length) {
-		if ((lhs_line < lhs_ctx.codes.length) && (!lhs_ctx.modified[lhs_line])
-			&& (rhs_line < rhs_ctx.codes.length) && (!rhs_ctx.modified[rhs_line])) {
+	while (lhs_line < lhs_ctx.length || rhs_line < rhs_ctx.length) {
+		if ((lhs_line < lhs_ctx.length) && (!lhs_ctx.modified[lhs_line])
+			&& (rhs_line < rhs_ctx.length) && (!rhs_ctx.modified[rhs_line])) {
 			// equal lines
 			lhs_line++;
 			rhs_line++;
@@ -215,19 +225,50 @@ diff.prototype._create_diffs = function(lhs_ctx, rhs_ctx) {
 			lhs_start = lhs_line;
 			rhs_start = rhs_line;
 
-			while (lhs_line < lhs_ctx.codes.length && (rhs_line >= rhs_ctx.codes.length || lhs_ctx.modified[lhs_line]))
+			while (lhs_line < lhs_ctx.length && (rhs_line >= rhs_ctx.length || lhs_ctx.modified[lhs_line]))
 				lhs_line++;
 
-			while (rhs_line < rhs_ctx.codes.length && (lhs_line >= lhs_ctx.codes.length || rhs_ctx.modified[rhs_line]))
+			while (rhs_line < rhs_ctx.length && (lhs_line >= lhs_ctx.length || rhs_ctx.modified[rhs_line]))
 				rhs_line++;
 
 			if ((lhs_start < lhs_line) || (rhs_start < rhs_line)) {
 				// store a new difference-item
+				let deleted_count;
+				let inserted_count;
+				let lhs_start = lhs_line;
+				let rhs_start = rhs_line;
+				if (options.split === 'lines') {
+					lhs_start = lhs_line;
+					rhs_start = rhs_line;
+					deleted_count = lhs_line - lhs_start;
+					inserted_count = rhs_line - rhs_start;
+				} else {
+					const ditem_lhs_start = (lhs_start >= lhs_ctx.length)
+						? lhs_ctx.length
+						: lhs_ctx.parts[lhs_start].from;
+					const ditem_rhs_start = (rhs_start >= rhs_ctx.length)
+						? rhs_ctx.length
+						: rhs_ctx.parts[rhs_start].from;
+
+					const ditem_lhs_end = (lhs_line >= lhs_ctx.length)
+						? lhs_ctx.length
+						: lhs_ctx.parts[lhs_line].from;
+					const ditem_rhs_end = (rhs_line >= rhs_ctx.length)
+						? rhs_ctx.length
+						: rhs_ctx.parts[rhs_line];
+					// const delim_len = (options.split === 'words') ? 1 : 0;
+					// const ditemLhs
+
+					deleted_count = ditem_lhs_end - ditem_lhs_start;
+					inserted_count = ditem_rhs_end - ditem_rhs_start;
+					lhs_start = ditem_lhs_start;
+					rhs_start = ditem_rhs_start;
+				}
 				items.push({
 					lhs_start: lhs_start,
 					rhs_start: rhs_start,
-					lhs_deleted_count: lhs_line - lhs_start,
-					rhs_inserted_count: rhs_line - rhs_start
+					lhs_deleted_count: deleted_count,
+					rhs_inserted_count: inserted_count
 				});
 			}
 		}
@@ -235,77 +276,4 @@ diff.prototype._create_diffs = function(lhs_ctx, rhs_ctx) {
 	return items;
 };
 
-function CodeifyText(lhs, rhs, options) {
-    this._max_code = 0;
-    this._diff_codes = {};
-	this.ctxs = {};
-	this.options = options;
-	this.options.split = this.options.split || 'lines';
-
-	if (typeof lhs === 'string') {
-		if (this.options.split === 'chars') {
-			this.lhs = lhs.split('');
-		} else if (this.options.split === 'words') {
-			this.lhs = lhs.split(/\s/);
-		} else if (this.options.split === 'lines') {
-			this.lhs = lhs.split('\n');
-		}
-	} else {
-		this.lhs = lhs;
-	}
-	if (typeof rhs === 'string') {
-		if (this.options.split === 'chars') {
-			this.rhs = rhs.split('');
-		} else if (this.options.split === 'words') {
-			this.rhs = rhs.split(/\s/);
-		} else if (this.options.split === 'lines') {
-			this.rhs = rhs.split('\n');
-		}
-	} else {
-		this.rhs = rhs;
-	}
-};
-
-CodeifyText.prototype.getCodes = function(side) {
-	if (!this.ctxs.hasOwnProperty(side)) {
-		var ctx = this._diff_ctx(this[side]);
-		this.ctxs[side] = ctx;
-		ctx.codes.length = Object.keys(ctx.codes).length;
-	}
-	return this.ctxs[side].codes;
-}
-
-CodeifyText.prototype.getLines = function(side) {
-	return this.ctxs[side].lines;
-}
-
-CodeifyText.prototype._diff_ctx = function(lines) {
-	var ctx = {i: 0, codes: {}, lines: lines};
-	this._codeify(lines, ctx);
-	return ctx;
-}
-
-CodeifyText.prototype._codeify = function(lines, ctx) {
-	for (let i = 0; i < lines.length; ++i) {
-		let line = lines[i];
-		if (this.options.ignorews) {
-			line = line.replace(/\s+/g, '');
-		}
-		if (this.options.ignorecase) {
-			line = line.toLowerCase();
-		}
-		if (this.options.ignoreaccents) {
-			line = line.normalize('NFD').replace(/[\u0300-\u036f]/g, '');
-		}
-		const aCode = this._diff_codes[line];
-		if (aCode !== undefined) {
-			ctx.codes[i] = aCode;
-		} else {
-			++this._max_code;
-			this._diff_codes[line] = this._max_code;
-			ctx.codes[i] = this._max_code;
-		}
-	}
-}
-
 module.exports = diff;

src/encoder.js

@@ -0,0 +1,79 @@
+class Encoder {
+    constructor() {
+        this._maxCode = 0;
+        this._codes = {};
+    }
+
+    encode(text, options) {
+        let exp;
+        let fudge = 0;
+        if (options.split === 'chars') {
+            exp = /./g;
+            fudge = 1;
+        } else if (options.split === 'words') {
+            exp = /\s+/g;
+        } else {
+            exp = /\n/g;
+        }
+        let match;
+        let p0 = -1;
+        const parts = [];
+        while ((match = exp.exec(text)) !== null) {
+            const from = (options.split === 'lines') ? parts.length : p0 + 1;
+            const to = (options.split === 'lines') ? parts.length + 1 : match.index + fudge;
+            const item = {
+                from,
+                to,
+                text: text.substr(p0 + 1, match.index - p0 - 1 + fudge)
+            };
+            parts.push(item);
+            p0 = match.index;
+        }
+        const from = (options.split === 'lines') ? parts.length : p0 + 1;
+        const to = (options.split === 'lines') ? parts.length + 1 : text.length;
+        parts.push({
+            from,
+            to,
+            text: text.substr(p0 + 1)
+        });
+        const hash = this._hash(parts, options)
+        return {
+            codes: hash.codes,
+            parts: hash.parts,
+            length: Object.keys(hash.codes).length
+        };
+    }
+
+    _hash(parts, options) {
+        const codes = {};
+        let i = 0;
+        for (const part of parts) {
+            let text = part.text;
+
+            if (options.ignorews) {
+                text = text.replace(/\s+/g, '');
+            }
+            if (options.ignorecase) {
+                text = text.toLowerCase();
+            }
+            if (options.ignoreaccents) {
+                text = text.normalize('NFD').replace(/[\u0300-\u036f]/g, '');
+            }
+            const code = this._codes[text];
+            if (code !== undefined) {
+                codes[i] = code;
+            } else {
+                ++this._maxCode;
+                this._codes[text] = this._maxCode;
+                codes[i] = this._maxCode;
+            }
+            i += 1;
+        }
+        return {
+            codes,
+            parts
+        };
+    }
+}
+
+module.exports = Encoder;

src/mergely.js

@@ -12,6 +12,7 @@ const defaultOptions = {
 	wrap_lines: false,
 	line_numbers: true,
 	lcs: true,
+	inline: 'chars',
 	sidebar: true,
 	viewport: false,
 	ignorews: false,
@@ -119,13 +120,15 @@ class Mergely {
 		const colors = dom.getColors(this.el);
 		this._options = {
 			...defaultOptions,//lgpl
+			...{
+				// default inline based off `lcs`
+				inline: options && options.lcs === false ? 'none' : 'chars'
+			},
 			...this._initOptions,
-			...options//lgpl-separate-notice
+			...options,//lgpl-separate-notice
 		};
 		this._viewOptions = {
-			...defaultOptions,
-			...this._initOptions,
-			...options,
+			...this._options,
 			_colors: colors
 		};
 	}

src/vdoc.js

@@ -124,7 +124,7 @@ class VDoc {
 		this._setRenderedChange(side, changeId);
 	}
 
-	addInlineDiff(change, changeId, { getText, ignorews, ignoreaccents, ignorecase }) {
+	addInlineDiff(change, changeId, { getText, ignorews, ignoreaccents, ignorecase, split = 'chars' }) {
 		if (this.options._debug) {
 			trace('vdoc#addInlineDiff', changeId, change);
 		}
@@ -152,7 +152,7 @@ class VDoc {
 					ignoreaccents,
 					ignorews,
 					ignorecase,
-					split: 'chars'
+					split
 				});
 				for (const change of results.changes()) {
 					const {

test/diff.spec.js

@@ -0,0 +1,16 @@
+const diff = require('../src/diff');
+
+describe('diff', () => {
+    it('should insert one line when lhs is empty and rhs has no line ending', () => {
+        const _diff = new diff('', 'hello', { split: 'lines' });
+        const changes = _diff.changes();
+        console.log(changes);
+        // with lhs_start at 1, the insert is at the end
+        expect(changes).to.deep.equal([{
+            lhs_start: 1,
+            rhs_start: 1,
+            lhs_deleted_count: 0,
+            rhs_inserted_count: 0
+        }]);
+    });
+});

test/markup.spec.js

@@ -189,7 +189,6 @@ describe('markup', () => {
 				expect(editor.querySelectorAll('.mergely.rhs')).to.have.length(0);
 			}
 		},
-
 		{
 			name: 'Changed lines (lhs)',
 			lhs: 'the quick red fox\njumped over the hairy dog',
@@ -233,6 +232,46 @@ describe('markup', () => {
 				expect(rhs_spans[1].innerText).to.equal('h');
 				expect(rhs_spans[2].innerText).to.equal('ir');
 			}
+		},
+		{
+			name: 'Changed lines with inline words (lhs)',
+			lhs: 'the quick red fox\njumped over the hairy dog',
+			rhs: 'the quick brown fox\njumped over the lazy dog',
+			options: { inline: 'words' },
+			check: (editor) => {
+				expect(editor.querySelectorAll(LHS_CHANGE_START + '.cid-0')).to.have.length(1);
+				expect(editor.querySelectorAll(LHS_CHANGE_END + '.cid-0')).to.have.length(1);
+				expect(editor.querySelectorAll(RHS_CHANGE_START + '.cid-0')).to.have.length(1);
+				expect(editor.querySelectorAll(RHS_CHANGE_END + '.cid-0')).to.have.length(1);
+				const lhs_spans = editor.querySelectorAll(LHS_INLINE_TEXT + '.cid-0');
+				expect(lhs_spans).to.have.length(2);
+				expect(lhs_spans[0].innerText).to.equal('red');
+				expect(lhs_spans[1].innerText).to.equal('hairy');
+				const rhs_spans = editor.querySelectorAll(RHS_INLINE_TEXT + '.cid-0');
+				expect(rhs_spans).to.have.length(2);
+				expect(rhs_spans[0].innerText).to.equal('brown');
+				expect(rhs_spans[1].innerText).to.equal('lazy');
+			}
+		},
+		{
+			name: 'Changed lines (rhs)',
+			lhs: 'the quick brown fox\njumped over the lazy dog',
+			rhs: 'the quick red fox\njumped over the hairy dog',
+			options: { inline: 'words' },
+			check: (editor) => {
+				expect(editor.querySelectorAll(LHS_CHANGE_START + '.cid-0')).to.have.length(1);
+				expect(editor.querySelectorAll(LHS_CHANGE_END + '.cid-0')).to.have.length(1);
+				expect(editor.querySelectorAll(RHS_CHANGE_START + '.cid-0')).to.have.length(1);
+				expect(editor.querySelectorAll(RHS_CHANGE_END + '.cid-0')).to.have.length(1);
+				const lhs_spans = editor.querySelectorAll(LHS_INLINE_TEXT + '.cid-0');
+				expect(lhs_spans).to.have.length(2);
+				expect(lhs_spans[0].innerText).to.equal('brown');
+				expect(lhs_spans[1].innerText).to.equal('lazy');
+				const rhs_spans = editor.querySelectorAll(RHS_INLINE_TEXT + '.cid-0');
+				expect(rhs_spans).to.have.length(2);
+				expect(rhs_spans[0].innerText).to.equal('red');
+				expect(rhs_spans[1].innerText).to.equal('hairy');
+			}
 		}
 	];
 
@@ -246,8 +285,9 @@ describe('markup', () => {
 				license: 'lgpl-separate-notice',
 				change_timeout: 0,
 				_debug: debug,
-				lhs: (setValue) => setValue(opt.lhs),
-				rhs: (setValue) => setValue(opt.rhs)
+				lhs: opt.lhs,
+				rhs: opt.rhs,
+				...opt.options
 			});
 			const test = () => {
 				try {

test/mergely.spec.js

@@ -8,6 +8,7 @@ const defaultOptions = {
 	rhs_margin: 'right',
 	wrap_lines: false,
 	line_numbers: true,
+	inline: 'chars',
 	lcs: true,
 	sidebar: true,
 	viewport: false,