infisical/frontend/src/ee/utilities/findTextDifferences.ts

/**
 * 
 * @param textOld - old secret
 * @param textNew - new (updated) secret
 * @param diffPlusFlag - a flag for whether we want to detect moving segments 
 * - doesn't work in some examples (e.g., when we have a full reverse ordering of the text)
 * @returns 
 */
function patienceDiff(textOld: string[], textNew: string[], diffPlusFlag?: boolean) {

  /**
   * findUnique finds all unique values in arr[lo..hi], inclusive. This
   * function is used in preparation for determining the longest common
   * subsequence. Specifically, it first reduces the array range in question
   * to unique values.
   * @param chars - an array of characters
   * @param lo 
   * @param hi 
   * @returns - an ordered Map, with the arr[i] value as the Map key and the
	 * array index i as the Map value.
   */
	function findUnique(chars: string[], lo: number, hi: number) {
		const characterMap = new Map();

		for (let i=lo; i<=hi; i++) {
			const character = chars[i];
      
			if (characterMap.has(character)) {
				characterMap.get(character).count++;
				characterMap.get(character).index = i;
			} else {
				characterMap.set(character, { count: 1, index: i });
			}
		}

		characterMap.forEach((val, key, map) => {
			if (val.count !== 1) {
				map.delete(key);
			} else {
				map.set(key, val.index);
			}
		});

		return characterMap;
	}

  /**
   * @param aArray 
   * @param aLo 
   * @param aHi 
   * @param bArray 
   * @param bLo 
   * @param bHi 
   * @returns an ordered Map, with the Map key as the common line between aArray
   * and bArray, with the Map value as an object containing the array indexes of
   * the matching unique lines.
   * 
   */
	function uniqueCommon(aArray: string[], aLo: number, aHi: number, bArray: string[], bLo: number, bHi: number) {
		const ma = findUnique(aArray, aLo, aHi);
		const mb = findUnique(bArray, bLo, bHi);

		ma.forEach((val, key, map) => {
			if (mb.has(key)) {
				map.set(key, {
					indexA: val,
					indexB: mb.get(key)
				});
			} else {
				map.delete(key);
			}
		});

		return ma;
	}

  /**
   * longestCommonSubsequence takes an ordered Map from the function uniqueCommon
   * and determines the Longest Common Subsequence (LCS).
   * @param abMap 
   * @returns an ordered array of objects containing the array indexes of the
   * matching lines for a LCS.
   */
	function longestCommonSubsequence(abMap: Map<number, { indexA: number, indexB: number, prev?: number }>) {
		const ja: any = [];

		// First, walk the list creating the jagged array.
		abMap.forEach((val, key, map) => {
			let i = 0;
      
			while (ja[i] && ja[i][ja[i].length - 1].indexB < val.indexB) {
				i++;
			}

			if (!ja[i]) {
				ja[i] = [];
			}

			if (0 < i) {
				val.prev = ja[i-1][ja[i - 1].length - 1];
			}
			ja[i].push(val);
		});

		// Now, pull out the longest common subsequence.
		let lcs: any[] = [];
    
		if (0 < ja.length) {
			const n = ja.length - 1;
			lcs = [ja[n][ja[n].length - 1]];
      
			while (lcs[lcs.length - 1].prev) {
				lcs.push(lcs[lcs.length - 1].prev);
			}
		}

		return lcs.reverse();
	}

	// "result" is the array used to accumulate the textOld that are deleted, the
	// lines that are shared between textOld and textNew, and the textNew that were
	// inserted.
  
	const result: any[] = [];
	let deleted = 0;
	let inserted = 0;

	// aMove and bMove will contain the lines that don't match, and will be returned
	// for possible searching of lines that moved.
  
	const aMove: any[] = [];
	const aMoveIndex: any[] = [];
	const bMove: any[] = [];
	const bMoveIndex: any[] = [];

  /**
   * addToResult simply pushes the latest value onto the "result" array. This
   * array captures the diff of the line, aIndex, and bIndex from the textOld
   * and textNew array.
   * @param aIndex 
   * @param bIndex 
   */
	function addToResult(aIndex: number, bIndex: number) {
		if (bIndex < 0) {
			aMove.push(textOld[aIndex]);
			aMoveIndex.push(result.length);
			deleted++;
		} else if (aIndex < 0) {
			bMove.push(textNew[bIndex]);
			bMoveIndex.push(result.length);
			inserted++;
		}

		result.push({
			line: 0 <= aIndex ? textOld[aIndex] : textNew[bIndex],
			aIndex: aIndex,
			bIndex: bIndex,
		});
	}
  
  /**
   * addSubMatch handles the lines between a pair of entries in the LCS. Thus,
   * this function might recursively call recurseLCS to further match the lines
   * between textOld and textNew.
   * @param aLo 
   * @param aHi 
   * @param bLo 
   * @param bHi 
   */
	function addSubMatch(aLo: number, aHi: number, bLo: number, bHi: number) {
		// Match any lines at the beginning of textOld and textNew.
		while (aLo <= aHi && bLo <= bHi && textOld[aLo] === textNew[bLo]) {
			addToResult(aLo++, bLo++);
		}

		// Match any lines at the end of textOld and textNew, but don't place them
		// in the "result" array just yet, as the lines between these matches at
		// the beginning and the end need to be analyzed first.
    
		const aHiTemp = aHi;
		while (aLo <= aHi && bLo <= bHi && textOld[aHi] === textNew[bHi]) {
			aHi--;
			bHi--;
		}

		// Now, check to determine with the remaining lines in the subsequence
		// whether there are any unique common lines between textOld and textNew.
		//
		// If not, add the subsequence to the result (all textOld having been
		// deleted, and all textNew having been inserted).
		//
		// If there are unique common lines between textOld and textNew, then let's
		// recursively perform the patience diff on the subsequence.
    
		const uniqueCommonMap = uniqueCommon(textOld, aLo, aHi, textNew, bLo, bHi);
    
		if (uniqueCommonMap.size === 0) {
			while (aLo <= aHi) {
				addToResult(aLo++, -1);
			}

			while (bLo <= bHi) {
				addToResult(-1, bLo++);
			}
		} else {
			recurseLCS(aLo, aHi, bLo, bHi, uniqueCommonMap);
		}

		// Finally, let's add the matches at the end to the result.
		while (aHi < aHiTemp) {
			addToResult(++aHi, ++bHi);
		}
	}
  
  /**
   * recurseLCS finds the longest common subsequence (LCS) between the arrays
   * textOld[aLo..aHi] and textNew[bLo..bHi] inclusive.  Then for each subsequence
   * recursively performs another LCS search (via addSubMatch), until there are
   * none found, at which point the subsequence is dumped to the result.
   * @param aLo 
   * @param aHi 
   * @param bLo 
   * @param bHi 
   * @param uniqueCommonMap 
   */
	function recurseLCS(aLo: number, aHi: number, bLo: number, bHi: number, uniqueCommonMap?: any) {
		const x = longestCommonSubsequence(uniqueCommonMap || uniqueCommon(textOld, aLo, aHi, textNew, bLo, bHi));
    
		if (x.length === 0) {
			addSubMatch(aLo, aHi, bLo, bHi);
		} else {
			if (aLo < x[0].indexA || bLo < x[0].indexB) {
				addSubMatch(aLo, x[0].indexA - 1, bLo, x[0].indexB - 1);
			}

			let i;
			for (i = 0; i < x.length - 1; i++) {
				addSubMatch(x[i].indexA, x[i+1].indexA - 1, x[i].indexB, x[i+1].indexB - 1);
			}

			if (x[i].indexA <= aHi || x[i].indexB <= bHi) {
				addSubMatch(x[i].indexA, aHi, x[i].indexB, bHi);
			}
		}
	}

	recurseLCS(0, textOld.length - 1, 0, textNew.length - 1);

	if (diffPlusFlag) {
		return {
			lines: result,
			lineCountDeleted: deleted,
			lineCountInserted: inserted,
			lineCountMoved: 0,
			aMove: aMove,
			aMoveIndex: aMoveIndex,
			bMove: bMove,
			bMoveIndex: bMoveIndex,
		};
	}

	return {
		lines: result,
		lineCountDeleted: deleted,
		lineCountInserted: inserted,
		lineCountMoved: 0,
	};
}

/**
 * use:  patienceDiffPlus( textOld[], textNew[] )
 *
 * where:
 *      textOld[] contains the original text lines.
 *      textNew[] contains the new text lines.
 *
 * returns an object with the following properties:
 *      lines[] with properties of:
 *          line containing the line of text from textOld or textNew.
 *          aIndex referencing the index in aLine[].
 *          bIndex referencing the index in textNew[].
 *              (Note:  The line is text from either textOld or textNew, with aIndex and bIndex
 *               referencing the original index. If aIndex === -1 then the line is new from textNew,
 *               and if bIndex === -1 then the line is old from textOld.)
 *          moved is true if the line was moved from elsewhere in textOld[] or textNew[].
 *      lineCountDeleted is the number of lines from textOld[] not appearing in textNew[].
 *      lineCountInserted is the number of lines from textNew[] not appearing in textOld[].
 *      lineCountMoved is the number of lines that moved.
 */

function patienceDiffPlus(textOld: string[], textNew: string[]) {

	const difference = patienceDiff(textOld, textNew, true);

	let aMoveNext = difference.aMove;
	let aMoveIndexNext = difference.aMoveIndex;
	let bMoveNext = difference.bMove;
	let bMoveIndexNext = difference.bMoveIndex;

	delete difference.aMove;
	delete difference.aMoveIndex;
	delete difference.bMove;
	delete difference.bMoveIndex;
  
	let lastLineCountMoved;

	do {
		const aMove = aMoveNext;
		const aMoveIndex = aMoveIndexNext;
		const bMove = bMoveNext;
		const bMoveIndex = bMoveIndexNext;

		aMoveNext = [];
		aMoveIndexNext = [];
		bMoveNext = [];
		bMoveIndexNext = [];

		const subDiff = patienceDiff(aMove!, bMove!);
    
		lastLineCountMoved = difference.lineCountMoved;

		subDiff.lines.forEach((v, i) => {

			if (0 <= v.aIndex && 0 <= v.bIndex) {

				difference.lines[aMoveIndex![v.aIndex]].moved = true;
				difference.lines[bMoveIndex![v.bIndex]].aIndex = aMoveIndex![v.aIndex];
				difference.lines[bMoveIndex![v.bIndex]].moved = true;
				difference.lineCountInserted--;
				difference.lineCountDeleted--;
				difference.lineCountMoved++;
			} else if (v.bIndex < 0) {
				aMoveNext!.push(aMove![v.aIndex]);
				aMoveIndexNext!.push(aMoveIndex![v.aIndex]);
			} else {
				bMoveNext!.push(bMove![v.bIndex]);
				bMoveIndexNext!.push(bMoveIndex![v.bIndex]);
			}
		});
	} while (0 < difference.lineCountMoved - lastLineCountMoved);

	return difference;

}

export default patienceDiff;