User:Cacycle/diff.js: Difference between revisions

Revision as of 12:00, 14 September 2014

Revision as of 00:59, 15 September 2014

Line 3:

// ==UserScript==

// @name wDiff

// @version 1.0.2021

// @date September 14, 2014

// @description improved word-based diff library with block move detection

Line 136:

// UniCode letter support for regexps, from http://xregexp.com/addons/unicode/unicode-base.js v1.0.0

if (wDiff.letters === undefined) { wDiff.letters = 'a-zA-Z0-9' + '00AA00B500BA00C0-00D600D8-00F600F8-02C102C6-02D102E0-02E402EC02EE0370-037403760377037A-037D03860388-038A038C038E-03A103A3-03F503F7-0481048A-05270531-055605590561-058705D0-05EA05F0-05F20620-064A066E066F0671-06D306D506E506E606EE06EF06FA-06FC06FF07100712-072F074D-07A507B107CA-07EA07F407F507FA0800-0815081A082408280840-085808A008A2-08AC0904-0939093D09500958-09610971-09770979-097F0985-098C098F09900993-09A809AA-09B009B209B6-09B909BD09CE09DC09DD09DF-09E109F009F10A05-0A0A0A0F0A100A13-0A280A2A-0A300A320A330A350A360A380A390A59-0A5C0A5E0A72-0A740A85-0A8D0A8F-0A910A93-0AA80AAA-0AB00AB20AB30AB5-0AB90ABD0AD00AE00AE10B05-0B0C0B0F0B100B13-0B280B2A-0B300B320B330B35-0B390B3D0B5C0B5D0B5F-0B610B710B830B85-0B8A0B8E-0B900B92-0B950B990B9A0B9C0B9E0B9F0BA30BA40BA8-0BAA0BAE-0BB90BD00C05-0C0C0C0E-0C100C12-0C280C2A-0C330C35-0C390C3D0C580C590C600C610C85-0C8C0C8E-0C900C92-0CA80CAA-0CB30CB5-0CB90CBD0CDE0CE00CE10CF10CF20D05-0D0C0D0E-0D100D12-0D3A0D3D0D4E0D600D610D7A-0D7F0D85-0D960D9A-0DB10DB3-0DBB0DBD0DC0-0DC60E01-0E300E320E330E40-0E460E810E820E840E870E880E8A0E8D0E94-0E970E99-0E9F0EA1-0EA30EA50EA70EAA0EAB0EAD-0EB00EB20EB30EBD0EC0-0EC40EC60EDC-0EDF0F000F40-0F470F49-0F6C0F88-0F8C1000-102A103F1050-1055105A-105D106110651066106E-10701075-1081108E10A0-10C510C710CD10D0-10FA10FC-1248124A-124D1250-12561258125A-125D1260-1288128A-128D1290-12B012B2-12B512B8-12BE12C012C2-12C512C8-12D612D8-13101312-13151318-135A1380-138F13A0-13F41401-166C166F-167F1681-169A16A0-16EA1700-170C170E-17111720-17311740-17511760-176C176E-17701780-17B317D717DC1820-18771880-18A818AA18B0-18F51900-191C1950-196D1970-19741980-19AB19C1-19C71A00-1A161A20-1A541AA71B05-1B331B45-1B4B1B83-1BA01BAE1BAF1BBA-1BE51C00-1C231C4D-1C4F1C5A-1C7D1CE9-1CEC1CEE-1CF11CF51CF61D00-1DBF1E00-1F151F18-1F1D1F20-1F451F48-1F4D1F50-1F571F591F5B1F5D1F5F-1F7D1F80-1FB41FB6-1FBC1FBE1FC2-1FC41FC6-1FCC1FD0-1FD31FD6-1FDB1FE0-1FEC1FF2-1FF41FF6-1FFC2071207F2090-209C21022107210A-211321152119-211D212421262128212A-212D212F-2139213C-213F2145-2149214E218321842C00-2C2E2C30-2C5E2C60-2CE42CEB-2CEE2CF22CF32D00-2D252D272D2D2D30-2D672D6F2D80-2D962DA0-2DA62DA8-2DAE2DB0-2DB62DB8-2DBE2DC0-2DC62DC8-2DCE2DD0-2DD62DD8-2DDE2E2F300530063031-3035303B303C3041-3096309D-309F30A1-30FA30FC-30FF3105-312D3131-318E31A0-31BA31F0-31FF3400-4DB54E00-9FCCA000-A48CA4D0-A4FDA500-A60CA610-A61FA62AA62BA640-A66EA67F-A697A6A0-A6E5A717-A71FA722-A788A78B-A78EA790-A793A7A0-A7AAA7F8-A801A803-A805A807-A80AA80C-A822A840-A873A882-A8B3A8F2-A8F7A8FBA90A-A925A930-A946A960-A97CA984-A9B2A9CFAA00-AA28AA40-AA42AA44-AA4BAA60-AA76AA7AAA80-AAAFAAB1AAB5AAB6AAB9-AABDAAC0AAC2AADB-AADDAAE0-AAEAAAF2-AAF4AB01-AB06AB09-AB0EAB11-AB16AB20-AB26AB28-AB2EABC0-ABE2AC00-D7A3D7B0-D7C6D7CB-D7FBF900-FA6DFA70-FAD9FB00-FB06FB13-FB17FB1DFB1F-FB28FB2A-FB36FB38-FB3CFB3EFB40FB41FB43FB44FB46-FBB1FBD3-FD3DFD50-FD8FFD92-FDC7FDF0-FDFBFE70-FE74FE76-FEFCFF21-FF3AFF41-FF5AFF66-FFBEFFC2-FFC7FFCA-FFCFFFD2-FFD7FFDA-FFDC'.replace(/(\w{4})/g, '\\u$1'); }

if (wDiff.smallSpaces === undefined) { wDiff.smallSpaces = ' \\u00a0\\u1680​\\u180e\\u2000​\\u2002\\u2004-\\u200a​\\u2028\\u2029​​\\u202f\\u205f​\\u3000'; }

if (wDiff.wideSpaces === undefined) { wDiff.wideSpaces = '\\u2001\\u2003'; }

// new line characters without and with '\n' and '\r'

// regExps for splitting text

if (wDiff.newLines === undefined) { wDiff.newLines = '\\u0085\\u2028'; }

if (wDiff.newLinesAll === undefined) { wDiff.newLinesAll = '\\n\\r\\u0085\\u2028'; }

// full stops without '.'

if (wDiff.fullStops === undefined) { wDiff.fullStops = '058906D40701070209640DF41362166E180318092CF92CFE2E3C3002A4FFA60EA6F3FE52FF0EFF61'.replace(/(\w{4})/g, '\\u$1'); }

// new paragraph characters without '\n' and '\r'

if (wDiff.newParagraph === undefined) { wDiff.newParagraph = '\\u2029'; }

// exclamation marks without '!'

if (wDiff.exclamationMarks === undefined) { wDiff.exclamationMarks = '01C301C301C3055C055C07F919441944203C203C20482048FE15FE57FF01'.replace(/(\w{4})/g, '\\u$1'); }

// question marks without '?'

if (wDiff.questionMarks === undefined) { wDiff.questionMarks = '037E055E061F13671945204720492CFA2CFB2E2EA60FA6F7FE56FF1F'.replace(/(\w{4})/g, '\\u$1') + '\\u11143'; }

// regExps for splitting text (included separators)

if (wDiff.regExpSplit === undefined) {

wDiff.regExpSplit = {

// paragraphs: after double newlines

paragraph: /new RegExp('(.|\\n)*?((\\r\\n|\\n|\\r){2,}/|[' + wDiff.newParagraph + '])+', 'g'),

// sentences: after newlines and .spaces

sentence: new RegExp('[^' + wDiff.newLinesAll + ']*?([.!?;]+[^\\S' + wDiff.newLinesAll + ']+|[' + wDiff.fullStops + wDiff.exclamationMarks + wDiff.questionMarks + ']+[^\\S' + wDiff.newLinesAll + ']*|[' + wDiff.newLines + ']|\\r\\n|\\n|\\r)', 'g'),

sentence: /.*?(\.\s+|\n)/g,

// inline chunks

Line 162 ⟶ 176:

// regExps for sliding gaps

if (wDiff.regExpSlideStop === undefined) { wDiff.regExpSlideStop = /new RegExp('[\\n\\r' + wDiff.newLines + ']$/'); }

if (wDiff.regExpSlideClosingregExpSlideBorder === undefined) { wDiff.regExpSlideClosingregExpSlideBorder = /^new RegExp('[\s)\]}>\-–—^' + wDiff.,:;?!’\/\\=letters + ']/$'); }

// regExp for counting words

if (wDiff.regExpWordCount === undefined) { wDiff.regExpWordCount = new RegExp('[' + wDiff.letters + ']+([\'’_]?[' + wDiff.letters + ']+)*', 'g'); }

// regExp fordetecting wikiblank-only codeand nonsingle-letterchar charactersblocks

if (wDiff.regExpWikiCodeCharsregExpBlankBlock === undefined) { wDiff.regExpWikiCodeCharsregExpBlankBlock = /^([ ^\t\n\[\S]{}+|+[^\-!*#:;=<>'\/_,.&?t]+)$/; }

// regExp detecting blank-only blocks

if (wDiff.regExpBlankBlock === undefined) { wDiff.regExpBlankBlock = new RegExp('^([' + wDiff.smallSpaces + ']{0,2}|[' + wDiff.wideSpaces + ']?|[' + wDiff.smallSpaces + ']?\\n([' + wDiff.smallSpaces + ']{0,2}|[' + wDiff.wideSpaces + ']?|[' + wDiff.smallSpaces + ']?\\n)?)$'); }

//

Line 214 ⟶ 225:

if (wDiff.stylesheet === undefined) {

wDiff.stylesheet =

'.wDiffTab:before { content: "→"; color: #bbb; font-size: smaller; }' +

// insert

'.wDiffNewline:before { content: "¶"; color: transparent; }' +

'.wDiffInsert:hover .wDiffNewline:before { color: #999; }' +

'.wDiffDelete:hover .wDiffNewline:before { color: #aaa; }' +

'.wDiffInsertBlank:hover .wDiffNewline:before { color: #888; }' +

'.wDiffDeleteBlank:hover .wDiffNewline:before { color: #999; }' +

'.wDiffBlockLeft:hover .wDiffNewline:before, .wDiffBlockRight:hover .wDiffNewline:before { color: #ccc; }' +

'.wDiffMarkRight:before { content: "' + wDiff.symbolMarkRight + '"; }' +

'.wDiffMarkLeft:before { content: "' + wDiff.symbolMarkLeft + '"; }' +

'.wDiffDelete { font-weight: bold; background-color: #ffe49c; color: #222; border-radius: 0.25em; padding: 0.2em 1px; }' +

'.wDiffInsert { font-weight: bold; background-color: #bbddff; color: #222; border-radius: 0.25em; padding: 0.2em 1px; }' +

'.wDiffDeleteBlankwDiffInsertBlank { background-color: #ffc84066bbff; }' +

'.wDiffFragment:hover .wDiffInsertBlank { background-color: #66b8ffbbddff; }' +

'.wDiffBlockLeft { font-weight: bold; background-color: #e8e8e8; border-radius: 0.25em; padding: 0.2em 1px; margin: 0 1px; }' +

// delete

'.wDiffBlockRight { font-weight: bold; background-color: #e8e8e8; border-radius: 0.25em; padding: 0.2em 1px; margin: 0 1px; }' +

'.wDiffMarkLeft wDiffDelete { font-weight: bold; background-color: #ffe49c; color: #666222; border-radius: 0.25em; padding: 0.2em; margin: 0 1px; }' +

'.wDiffMarkRightwDiffDeleteBlank { font-weight: bold; background-color: #ffe49c; color: #666; border-radius: 0.25em; padding: 0.2em; margin: 0 1pxffd064; }' +

'.wDiffFragment:hover .wDiffDeleteBlank { background-color: #ffe49c; }' +

// block

'.wDiffBlockLeft, .wDiffBlockRight { font-weight: bold; background-color: #e8e8e8; border-radius: 0.25em; padding: 0.2em 1px; margin: 0 1px; }' +

'.wDiffBlockHighlight { background-color: #777; color: #fff; border: solid #777; border-width: 1px 0; }' +

'.wDiffBlock { }' +

'.wDiffBlock0 { background-color: #ffff60; }' +

Line 241 ⟶ 249:

'.wDiffBlock7 { background-color: #ffbbbb; }' +

'.wDiffBlock8 { background-color: #a0e8a0; }' +

// mark

'.wDiffMarkLeft, .wDiffMarkRight { font-weight: bold; background-color: #ffe49c; color: #666; border-radius: 0.25em; padding: 0.2em; margin: 0 1px; }' +

'.wDiffMarkRight:before { content: "' + wDiff.symbolMarkRight + '"; }' +

'.wDiffMarkLeft:before { content: "' + wDiff.symbolMarkLeft + '"; }' +

'.wDiffMark { }' +

'.wDiffMarkHighlight { background-color: #777; color: #fff; }' +

'.wDiffMark0 { color: #ffff60; }' +

'.wDiffMark1 { color: #c0ff60; }' +

Line 251 ⟶ 265:

'.wDiffMark7 { color: #ff9999; }' +

'.wDiffMark8 { color: #90d090; }' +

'.wDiffBlockHighlight { background-color: #777; color: #fff; border: solid #777; border-width: 1px 0; }' +

// wrappers

'.wDiffMarkHighlight { background-color: #777; color: #fff; }' +

'.wDiffContainer { }' +

'.wDiffFragment { white-space: pre-wrap; background: #fff; border: #bbb solid; border-width: 1px 1px 1px 0.5em; border-radius: 0.5em; font-family: sans-serif; font-size: 88%; line-height: 1.6; box-shadow: 2px 2px 2px #ddd; padding: 1em; margin: 0; }' +

'.wDiffNoChange { white-space: pre-wrap; background: #f0f0f0; border: #bbb solid; border-width: 1px 1px 1px 0.5em; border-radius: 0.5em; font-family: sans-serif; font-size: 88%; line-height: 1.6; box-shadow: 2px 2px 2px #ddd; padding: 0.5em; margin: 1em 0; }' +

'.wDiffSeparator { margin-bottom: 1em; }' +

'.wDiffOmittedChars { }'; +

// newline

'.wDiffNewline:before { content: "¶"; color: transparent; }' +

'.wDiffBlockHighlight .wDiffNewline:before { color: transparent; }' +

'.wDiffBlockHighlight:hover .wDiffNewline:before { color: #ccc; }' +

'.wDiffBlockHighlight:hover .wDiffInsert .wDiffNewline:before, .wDiffInsert:hover .wDiffNewline:before { color: #999; }' +

'.wDiffBlockHighlight:hover .wDiffDelete .wDiffNewline:before, .wDiffDelete:hover .wDiffNewline:before { color: #aaa; }' +

// tab

'.wDiffTab { position: relative; }' +

'.wDiffTabSymbol { position: absolute; top: -0.2em; }' +

'.wDiffTabSymbol:before { content: "→"; font-size: smaller; color: transparent; color: #ccc; }' +

'.wDiffBlockLeft .wDiffTabSymbol:before, .wDiffBlockRight .wDiffTabSymbol:before { color: #aaa; }' +

'.wDiffBlockHighlight .wDiffTabSymbol:before { color: #aaa; }' +

'.wDiffInsert .wDiffTabSymbol:before { color: #aaa; }' +

'.wDiffDelete .wDiffTabSymbol:before { color: #bbb; }' +

// space

'.wDiffSpace { position: relative; }' +

'.wDiffSpaceSymbol { position: absolute; top: -0.2em; left: -0.05em; }' +

'.wDiffSpaceSymbol:before { content: "·"; color: transparent; }' +

'.wDiffBlockHighlight .wDiffSpaceSymbol:before { color: transparent; }' +

'.wDiffBlockHighlight:hover .wDiffSpaceSymbol:before { color: #ddd; }' +

'.wDiffBlockHighlight:hover .wDiffInsert .wDiffSpaceSymbol:before, .wDiffInsert:hover .wDiffSpaceSymbol:before { color: #888; }' +

'.wDiffBlockHighlight:hover .wDiffDelete .wDiffSpaceSymbol:before, .wDiffDelete:hover .wDiffSpaceSymbol:before { color: #999; }';

}

Line 264 ⟶ 303:

//

if (wDiff.styleDelete === undefined) { wDiff.styleDelete = ''; }

if (wDiff.styleInsert === undefined) { wDiff.styleInsert = ''; }

if (wDiff.styleDeleteBlankstyleDelete === undefined) { wDiff.styleDeleteBlankstyleDelete = ''; }

if (wDiff.styleInsertBlank === undefined) { wDiff.styleInsertBlank = ''; }

if (wDiff.styleDeleteBlank === undefined) { wDiff.styleDeleteBlank = ''; }

if (wDiff.styleBlockLeft === undefined) { wDiff.styleBlockLeft = ''; }

if (wDiff.styleBlockRight === undefined) { wDiff.styleBlockRight = ''; }

Line 280 ⟶ 319:

if (wDiff.styleSeparator === undefined) { wDiff.styleSeparator = ''; }

if (wDiff.styleOmittedChars === undefined) { wDiff.styleOmittedChars = ''; }

if (wDiff.styleTab === undefined) { wDiff.styleTab = ''; }

if (wDiff.styleNewline === undefined) { wDiff.styleNewline = ''; }

if (wDiff.styleTab === undefined) { wDiff.styleTab = ''; }

if (wDiff.styleTabSymbol === undefined) { wDiff.styleTabSymbol = ''; }

if (wDiff.styleSpace === undefined) { wDiff.styleSpace = ''; }

if (wDiff.styleSpaceSymbol === undefined) { wDiff.styleSpaceSymbol = ''; }

//

Line 308 ⟶ 350:

if (wDiff.htmlBlockRightEnd === undefined) { wDiff.htmlBlockRightEnd = '</span><!--wDiffBlockRight-->'; }

if (wDiff.htmlMarkLeft === undefined) { wDiff.htmlMarkLeft = '<span class="wDiffMarkLeft wDiffMark{class}" style="' + wDiff.styleMarkLeft + '{mark}"{title} id="wDiffMark{number}"' + wDiff.blockEvent + '></span><!--wDiffMarkLeft-->'; }

if (wDiff.htmlMarkRight === undefined) { wDiff.htmlMarkRight = '<span class="wDiffMarkRight wDiffMark{class}" style="' + wDiff.styleMarkRight + '{mark}"{title} id="wDiffMark{number}"' + wDiff.blockEvent + '></span><!--wDiffMarkRight-->'; }

if (wDiff.htmlMarkLeft === undefined) { wDiff.htmlMarkLeft = '<span class="wDiffMarkLeft wDiffMark{class}" style="' + wDiff.styleMarkLeft + '{mark}"{title} id="wDiffMark{number}"' + wDiff.blockEvent + '></span><!--wDiffMarkLeft-->'; }

if (wDiff.htmlNewline === undefined) { wDiff.htmlNewline = '<span class="wDiffNewline" style="' + wDiff.styleNewline + '">\n</span>\n'; }

if (wDiff.htmlTab === undefined) { wDiff.htmlTab = '<span class="wDiffTab" style="' + wDiff.styleTab + '"><span class="wDiffTabSymbol" style="' + wDiff.styleTabSymbol + '"></span>\t</span>'; }

if (wDiff.htmlSpace === undefined) { wDiff.htmlSpace = '<span class="wDiffSpace" style="' + wDiff.styleSpace + '"><span class="wDiffSpaceSymbol" style="' + wDiff.styleSpaceSymbol + '"></span> </span>'; }

//

Line 324 ⟶ 367:

if (wDiff.htmlSeparator === undefined) { wDiff.htmlSeparator = '<div class="wDiffSeparator" style="' + wDiff.styleSeparator + '"></div>'; }

if (wDiff.htmlOmittedChars === undefined) { wDiff.htmlOmittedChars = '<span class="wDiffOmittedChars" style="' + wDiff.styleOmittedChars + '">…</span>'; }

//

Line 512 ⟶ 554:

}

// parse and count count words in texts for later identification of unique words

wDiff.CountTextWords(text.newText);

wDiff.CountTextWords(text.oldText);

Line 961 ⟶ 1,003:

// wDiff.SlideGaps: move gaps with ambiguous identical fronts andto backslast upnewline or, if absent, last word border

// start ambiguous gap borders after line breaks and text section closing characters

// changes: text (text.newText or text.oldText) .tokens list

// called from: wDiff.Diff()

Line 984 ⟶ 1,025:

var front = gapStart;

var back = i;

var frontPrevfrontTest = null;

var backPrevbackTest = null;

while (

(front !== null) && (back !== null) &&

Line 994 ⟶ 1,035:

textLinked.tokens[ text.tokens[front].link ].link = front;

text.tokens[back].link = null;

frontPrevfrontTest = front;

backPrevbackTest = back;

front = text.tokens[front].next;

back = text.tokens[back].next;

}

// test baloonslide up, remember last line break or closingword textborder

var frontStop = null;

var frontTest = frontPrev;

var backTest = backPrev;

while (

(frontTest !== null) && (backTest !== null) &&

Line 1,013 ⟶ 1,052:

break;

}

else if ( (frontStop === null) && (wDiff.regExpSlideClosingregExpSlideBorder.test(text.tokens[frontTest].token) === true) ) {

frontStop = frontTest;

}

Line 1,020 ⟶ 1,059:

}

// actually slide up to last line break or, if absent, closingword textborder

if (frontStop !== null) {

while (

Line 1,337 ⟶ 1,376:

if ( (j !== null) && (text.oldText.tokens[j] !== null) && (text.newText.tokens[i].link === null) && (text.oldText.tokens[j].link === null) ) {

// determine the limits of of the unresolved new sequence

var iStart = i;

var iEnd = null;

Line 1,351 ⟶ 1,390:

}

// determine the limits of of the unresolved old sequence

var jStart = j;

var jEnd = null;

Line 1,407 ⟶ 1,446:

if ( (j !== null) && (text.oldText.tokens[j] !== null) && (text.newText.tokens[i].link === null) && (text.oldText.tokens[j].link === null) ) {

// determine the limits of of the unresolved new sequence

var iStart = null;

var iEnd = i;

Line 1,421 ⟶ 1,460:

}

// determine the limits of of the unresolved old sequence

var jStart = null;

var jEnd = j;

Line 2,578 ⟶ 2,617:

text.diff = text.diff.replace(/<\/(\w+)><!--wDiff(Delete|Insert)--><\1\b[^>]*\bclass="wDiff\2"[^>]*>/g, '');

text.diff = text.diff.replace(/\t/g, wDiff.htmlTab);

text.diff = text.diff.replace(/(<[^>]*>)|( )/g, function (p, p1, p2) {

if (p2 == ' ') {

return wDiff.htmlSpace;

}

return p1;

});

text.diff = wDiff.htmlContainerStart + wDiff.htmlFragmentStart + text.diff + wDiff.htmlFragmentEnd + wDiff.htmlContainerEnd;

return;

Line 3,015 ⟶ 3,060:

for (var i = 0; i < groups.length; i ++) {

dump += i + ' \t' + groups[i].blockStart + ' \t' + groups[i].blockEnd + ' \t' + groups[i].unique + ' \t' + groups[i].maxWords + ' \t' + groups[i].words + ' \t' + groups[i].chars + ' \t' + groups[i].fixed + ' \t' + groups[i].oldNumber + ' \t' + groups[i].movedFrom + ' \t' + groups[i].color + ' \t' + groups[i].moved.toString() + ' \t' + wDiff.DebugShortenString(groups[i].diff) + '\n';

}

return dump;

};

//

// wDiff.DebugGaps: dump gaps object

//

wDiff.DebugGaps = function (gaps) {

var dump = '\ni \tnFirs \tnLast \tnTok \toFirs \toLast \toTok \tcharSplit\n';

for (var i = 0; i < gaps.length; i ++) {

dump += i + ' \t' + gaps[i].newFirst + ' \t' + gaps[i].newLast + ' \t' + gaps[i].newTokens + ' \t' + gaps[i].oldFirst + ' \t' + gaps[i].oldLast + ' \t' + gaps[i].oldTokens + ' \t' + gaps[i].charSplit + '\n';

}

return dump;