@odoo/o-spreadsheet 17.2.4 → 17.2.5

package/dist/o-spreadsheet.cjs.js

@@ -3,9 +3,9 @@
 /**
  * This file is generated by o-spreadsheet build tools. Do not edit it.
  * @see https://github.com/odoo/o-spreadsheet
- * @version 17.2.4
- * @date 2024-04-18T16:41:38.407Z
- * @hash 0c66038
+ * @version 17.2.5
+ * @date 2024-04-26T07:41:13.193Z
+ * @hash a730f5c
  */
 
 'use strict';
@@ -32073,17 +32073,13 @@ class Composer extends owl.Component {
                 this.DOMFocusableElementStore.setFocusableElement(el);
             }
             this.contentHelper.updateEl(el);
-            this.processTokenAtCursor();
         });
         owl.useEffect(() => {
             this.processContent();
         });
-        owl.onPatched(() => {
-            // Required because typing '=SUM' and double-clicking another cell leaves ShowProvider/ShowDescription true
-            if (this.composerStore.editionMode === "inactive") {
-                this.processTokenAtCursor();
-            }
-        });
+        owl.useEffect(() => {
+            this.processTokenAtCursor();
+        }, () => [this.composerStore.editionMode !== "inactive"]);
     }
     // ---------------------------------------------------------------------------
     // Handlers
@@ -40380,7 +40376,7 @@ class BordersPlugin extends CorePlugin {
                 this.clearBorders(cmd.sheetId, cmd.target);
                 break;
             case "REMOVE_COLUMNS_ROWS":
-                for (let el of cmd.elements) {
+                for (let el of [...cmd.elements].sort((a, b) => b - a)) {
                     if (cmd.dimension === "COL") {
                         this.shiftBordersHorizontally(cmd.sheetId, el + 1, -1);
                     }
@@ -45919,6 +45915,353 @@ class CompilationParametersBuilder {
     }
 }
 
+/**
+ * ####################################################
+ * # INTRODUCTION
+ * ####################################################
+ *
+ * This file contain the function recomputeZones.
+ * This function try to recompute in a performant way
+ * an ensemble of zones possibly overlapping to avoid
+ * overlapping and to reduce the number of zones.
+ *
+ * It also allows to remove some zones from the ensemble.
+ *
+ * In the following example, 2 zones are overlapping.
+ * Applying recomputeZones will return zones without
+ * overlapping:
+ *
+ * ["B3:D4", "D2:E3"]         ["B3:C4", "D2:D4", "E2:E3"]
+ *
+ *      A B C D E                    A B C D E
+ *    1       ___                  1       ___
+ *    2   ___|_  |                 2   ___| | |
+ *    3  |   |_|_|      --->       3  |   | |_|
+ *    4  |_____|                   4  |___|_|
+ *    6                            6
+ *    7                            7
+ *
+ *
+ * In the following example, 2 zones are contiguous.
+ * Applying recomputeZones will return only one zone:
+ *
+ *  ["B2:B3", "C2:D3"]               ["B2:D3"]
+ *
+ *       A B C D E                   A B C D E
+ *     1   _ ___                   1   _____
+ *     2  | |   |        --->      2  |     |
+ *     3  |_|___|                  3  |_____|
+ *     4                           4
+ *
+ *
+ * In the following example, we want to remove a zone
+ * from the ensemble. Applying recomputeZones will
+ * return the ensemble without the zone to remove:
+ *
+ *    remove ["C3:D3"]           ["B2:B4", "C2:D2",
+ *                                "C4:D4", "E2:E4"]
+ *
+ *       A B C D E F                 A B C D E F
+ *     1   _______                 1   _______
+ *     2  |       |       --->     2  | |___| |
+ *     3  |  xxx  |                3  | |___| |
+ *     4  |_______|                4  |_|___|_|
+ *     5                           5
+ *
+ *
+ * The exercise seems simple when we have only 2 zones.
+ * But with n zones and in a performant way, we want to
+ * avoid comparing each zone with all the others.
+ *
+ *
+ * ####################################################
+ * # Methodological approach
+ * ####################################################
+ *
+ * The methodological approach to avoid comparing each
+ * zone with all the others is to use a data structure
+ * that allow to quickly find which zones are
+ * overlapping with any other given zone.
+ *
+ * Here the idea is to profile the zones at the columns level.
+ *
+ * To do that, we propose to use a data structure
+ * composed of 2 parts:
+ * - profilesStartingPosition: a sorted number array
+ * indicating on which columns a new profile begins.
+ * - profiles: a map where the key is a column
+ * position (from profilesStartingPosition) and the
+ * value is a sorted number array representing a
+ * profile.
+ *
+ *
+ * See the following example:    here profileStartingPosition
+ *                               corresponds to [A,C,E,G,K]
+ *    A B C D E F G H I J K      so with number [0,2,4,6,10]
+ *  1    '   '   '       '
+ *  2    '   '   '_______'       here profile correspond
+ *  3    '___'   |_______|       for A to []
+ *  4    |   |                   for C to [3, 5]
+ *  5    |___|                   for E to []
+ *  6                            for G to [2, 3]
+ *  7                            for K to []
+ *
+ *
+ * Now we can easily find which zones are overlapping
+ * with a given zone. Suppose we want to add a new zone
+ * D5:H6 to the ensemble:
+ *
+ *                              With a binary search of left and right
+ *    A B C D E F G H I J K     on profilesStartingPosition, we can
+ *  1    '   '   '       '      find the indexes of the profiles on which
+ *  2    '   '   '_______'      to apply a modification.
+ *  3    '___'   |_______|
+ *  4    |  _|_______           Here we will:
+ *  5    |_|_|       |          - add a new profile in D   --> become [3, 6]
+ *  6      |_________|          - modify the profile in E  --> become [4, 6]
+ *  7                           - modify the profile in G  --> become [2, 3, 4, 6]
+ *                              - add a new profile in I   --> become [8, 10]
+ *
+ *  See below the result:
+ *
+ *                              Note the particularity of the profile
+ *    A B C D E F G H I J K     for G: it will correspond to [2, 3, 4, 6]
+ *  1    ' ' '   '   '   '
+ *  2    ' ' '   '___'___'      To know how to modify the profile (add a
+ *  3    '_'_'   |___|___|      zone or remove it) we do a binary
+ *  4    | | |___ ___           search of the top and bottom value on the
+ *  5    |_| |   |   |          profile array. Depending on the result index
+ *  6      |_|___|___|          parity (odd or even), because zone boundaries
+ *  7                           go by pairs, we know if we are in a zone or
+ *                              not and how operate.
+ */
+/**
+ * Recompute the zone without the cells in toRemoveZones and avoid overlapping.
+ * This compute is particularly useful because after this function:
+ * - you will find coordinate of a cell only once among all the zones
+ * - the number of zones will be reduced to the minimum
+ */
+function futureRecomputeZones(zones, zonesToRemove = []) {
+    const profilesStartingPosition = [0];
+    const profiles = new Map([[0, []]]);
+    modifyProfiles(profilesStartingPosition, profiles, zones, false);
+    modifyProfiles(profilesStartingPosition, profiles, zonesToRemove, true);
+    return constructZonesFromProfiles(profilesStartingPosition, profiles);
+}
+function modifyProfiles(// export for testing only
+profilesStartingPosition, profiles, zones, toRemove = false) {
+    for (const zone of zones) {
+        const leftValue = zone.left;
+        const rightValue = zone.right === undefined ? undefined : zone.right + 1;
+        const leftIndex = findIndexAndCreateProfile(profilesStartingPosition, profiles, leftValue, true, 0);
+        const rightIndex = findIndexAndCreateProfile(profilesStartingPosition, profiles, rightValue, false, leftIndex);
+        for (let i = leftIndex; i <= rightIndex; i++) {
+            const profile = profiles.get(profilesStartingPosition[i]);
+            modifyProfile(profile, zone, toRemove);
+        }
+        // maybe this part cost in performance, and maybe it's not necessary (depending on the use case). To be checked
+        removeContiguousProfiles(profilesStartingPosition, profiles, leftIndex, rightIndex);
+    }
+}
+function findIndexAndCreateProfile(profilesStartingPosition, profiles, value, searchLeft, startIndex) {
+    if (value === undefined) {
+        // this is only the case when the value correspond to a bottom value that could be undefined
+        return profilesStartingPosition.length - 1;
+    }
+    const predecessorIndex = binaryPredecessorSearch(profilesStartingPosition, value, startIndex);
+    if (value != profilesStartingPosition[predecessorIndex]) {
+        // mean that the value is not ending/starting at the same position as the previous/next profile
+        // --> it's a new profile
+        // --> we need to add it
+        profilesStartingPosition.splice(predecessorIndex + 1, 0, value);
+        // suppose the               we want to add the       for the left value
+        // following profile         following zone:          'C', the predecessor index
+        //   for B: [1, 3]                "C3:D4"             correspond to 'B'.
+        //                                                    The next line code will
+        //       A B C D                A B C D               copy the profile of 'B'
+        //     1  '___'               1  '___'                to 'C'. In the rest of the
+        //     2  |   |       --->    2  |  _|_               process the 'modifyProfile'
+        //     3  |___|               3  |_|_| |              function will adapt the waiting
+        //     4                      4    |___|              'C' profile [1, 3] to the
+        //                                                    correct 'C' profile [1, 4]
+        profiles.set(value, [...profiles.get(profilesStartingPosition[predecessorIndex])]);
+        return searchLeft ? predecessorIndex + 1 : predecessorIndex;
+    }
+    return searchLeft ? predecessorIndex : predecessorIndex - 1;
+}
+/**
+ *  Suppose the following        Suppose we want to add          We want to have the
+ *  profile:                     the following zone:             following profile:
+ *
+ *       A B C D E F                  A B C D E F                     A B C D E F
+ *     1    '___'                   1    '   '                      1    '___'
+ *     2    |___|                   2    '___'                      2    |   |
+ *     3    '   '                   3    |   |                      3    |   |
+ *     4    '___'          -->      4    |   |            -->       4    |   |
+ *     6    |   |                   6    |___|                      6    |   |
+ *     7    |___|                   7                               7    |___|
+ *     8                            8                               8
+ *
+ *  the profile for 'C'          the top zone correspond        Here [2, 3, 5, 8] with [3, 7]
+ *  corresponds to:              to 3 and the bottom zone       would be merged into [2, 8]
+ *   ____  ____                  correspond to 6
+ *  [2, 3, 5, 8]                 would be the profile:          The difficulty of modify profile
+ *                                ____                          is to know what must be deleted
+ *  Note that the 'filled        [3, 7]                         and what must be added to the
+ *  zone' are always between                                    existing profile.
+ *  an even index and its
+ *  next index
+ *
+ */
+function modifyProfile(profile, zone, toRemove = false) {
+    const topValue = zone.top;
+    const bottomValue = zone.bottom === undefined ? undefined : zone.bottom + 1;
+    const newPoints = [];
+    // Case we want to add a zone to the profile:
+    // - If the top predecessor index `topPredIndex` is even, it means the top of the zone is already positioned on a filled zone
+    // so we don't need to add it to the profile. we can keep in reference the index of the predecessor.
+    // - If it is odd, it means the top of the zone must be the beginning of a filled zone.
+    // so we can keep the index of the top position
+    // Case we want to remove a zone from the profile: it's the opposite of the previous case
+    const topPredIndex = binaryPredecessorSearch(profile, topValue, 0, false);
+    if ((topPredIndex % 2 !== 0 && !toRemove) || (topPredIndex % 2 === 0 && toRemove)) {
+        newPoints.push(topValue);
+    }
+    if (bottomValue === undefined) {
+        // The following two code lines will not impact the final result,
+        // but they will impact the intermediate profile.
+        // We keep them for performance reason
+        profile.splice(topPredIndex + 1);
+        profile.push(...newPoints);
+        return;
+    }
+    // Case we want to add a zone to the profile:
+    // - If the bottom successor index `bottomSuccIndex` is even, it means the bottom of the zone must be the ending of a filled zone
+    // so we can keep the index of the bottom position.
+    // - If it is odd, it means the bottom of the zone is already positioned on a filled zone
+    // so we don't need to add it to the profile. we can keep in reference the index of the successor
+    // Case we want to remove a zone from the profile: it's the opposite of the previous case
+    const bottomSuccIndex = binarySuccessorSearch(profile, bottomValue, 0, false);
+    if ((bottomSuccIndex % 2 === 0 && !toRemove) || (bottomSuccIndex % 2 !== 0 && toRemove)) {
+        newPoints.push(bottomValue);
+    }
+    // add the top and bottom value to the profile and
+    // remove all information between the top and bottom index
+    profile.splice(topPredIndex + 1, bottomSuccIndex - topPredIndex - 1, ...newPoints);
+}
+function removeContiguousProfiles(profilesStartingPosition, profiles, leftIndex, rightIndex) {
+    const start = leftIndex - 1 === -1 ? 0 : leftIndex - 1;
+    const end = rightIndex === profilesStartingPosition.length - 1 ? rightIndex : rightIndex + 1;
+    for (let i = end; i > start; i--) {
+        if (deepEqualsArray(profiles.get(profilesStartingPosition[i]), profiles.get(profilesStartingPosition[i - 1]))) {
+            profiles.delete(profilesStartingPosition[i]);
+            profilesStartingPosition.splice(i, 1);
+        }
+    }
+}
+function constructZonesFromProfiles(profilesStartingPosition, profiles) {
+    const mergedZone = [];
+    let pendingZones = [];
+    for (let colIndex = 0; colIndex < profilesStartingPosition.length; colIndex++) {
+        const left = profilesStartingPosition[colIndex];
+        const profile = profiles.get(left);
+        if (!profile || profile.length === 0) {
+            mergedZone.push(...pendingZones);
+            pendingZones = [];
+            continue;
+        }
+        let right = profilesStartingPosition[colIndex + 1];
+        if (right !== undefined) {
+            right--;
+        }
+        const nextPendingZones = [];
+        for (let i = 0; i < profile.length; i += 2) {
+            const top = profile[i];
+            let bottom = profile[i + 1];
+            if (bottom !== undefined) {
+                bottom--;
+            }
+            const profileZone = {
+                top,
+                left,
+                bottom,
+                right,
+                hasHeader: (bottom === undefined && top !== 0) || (right === undefined && left !== 0),
+            };
+            let findCorrespondingZone = false;
+            for (let j = pendingZones.length - 1; j >= 0; j--) {
+                const pendingZone = pendingZones[j];
+                if (pendingZone.top === profileZone.top && pendingZone.bottom === profileZone.bottom) {
+                    pendingZone.right = profileZone.right;
+                    pendingZones.splice(j, 1);
+                    nextPendingZones.push(pendingZone);
+                    findCorrespondingZone = true;
+                    break;
+                }
+            }
+            if (!findCorrespondingZone) {
+                nextPendingZones.push(profileZone);
+            }
+        }
+        mergedZone.push(...pendingZones);
+        pendingZones = nextPendingZones;
+    }
+    mergedZone.push(...pendingZones);
+    return mergedZone;
+}
+function binaryPredecessorSearch(arr, val, start = 0, matchEqual = true) {
+    let end = arr.length - 1;
+    let result = -1;
+    while (start <= end) {
+        const mid = Math.floor((start + end) / 2);
+        if (arr[mid] === val && matchEqual) {
+            return mid;
+        }
+        else if (arr[mid] < val) {
+            result = mid;
+            start = mid + 1;
+        }
+        else {
+            end = mid - 1;
+        }
+    }
+    return result;
+}
+function binarySuccessorSearch(arr, val, start = 0, matchEqual = true) {
+    let end = arr.length - 1;
+    let result = arr.length;
+    while (start <= end) {
+        const mid = Math.floor((start + end) / 2);
+        if (arr[mid] === val && matchEqual) {
+            return mid;
+        }
+        else if (arr[mid] > val) {
+            result = mid;
+            end = mid - 1;
+        }
+        else {
+            start = mid + 1;
+        }
+    }
+    return result;
+}
+/**
+ * Compares two arrays.
+ * For performance reasons, this function is to be preferred
+ * to 'deepEquals' in the case we know that the inputs are arrays.
+ */
+function deepEqualsArray(arr1, arr2) {
+    if (arr1.length !== arr2.length) {
+        return false;
+    }
+    for (let i = 0; i < arr1.length; i++) {
+        if (!deepEquals(arr1[i], arr2[i])) {
+            return false;
+        }
+    }
+    return true;
+}
+
 class PositionMap {
     map = {};
     set({ sheetId, col, row }, value) {
@@ -46717,7 +47060,7 @@ class FormulaDependencyGraph {
         }
     }
     /**
-     * Return the cell and all cells that depend on it,
+     * Return all the cells that depend on the provided ranges,
      * in the correct order they should be evaluated.
      * This is called a topological ordering (excluding cycles)
      */
@@ -46728,11 +47071,19 @@ class FormulaDependencyGraph {
             const range = queue.pop();
             visited.addMany(positions(range.zone).map((position) => ({ sheetId: range.sheetId, ...position })));
             const impactedPositions = this.rTree.search(range).map((dep) => dep.data);
+            const nextInQueue = {};
             for (const position of impactedPositions) {
                 if (!visited.has(position)) {
-                    queue.push({ sheetId: position.sheetId, zone: positionToZone(position) });
+                    if (!nextInQueue[position.sheetId]) {
+                        nextInQueue[position.sheetId] = [];
+                    }
+                    nextInQueue[position.sheetId].push(positionToZone(position));
                 }
             }
+            for (const sheetId in nextInQueue) {
+                const zones = futureRecomputeZones(nextInQueue[sheetId]);
+                queue.push(...zones.map((zone) => ({ sheetId, zone })));
+            }
         }
         visited.deleteMany(ranges.flatMap((r) => positions(r.zone).map((position) => ({ sheetId: r.sheetId, ...position }))));
         return visited;
@@ -47071,7 +47422,7 @@ class Evaluator {
             }
             if (!content) {
                 // The previous content could have blocked some array formulas
-                impactedPositions.addMany(this.getArrayFormulasBlockedByOrSpreadingOn(position));
+                impactedPositions.addMany(this.getArrayFormulasBlockedBy(position));
             }
         }
         return impactedPositions;
@@ -47114,14 +47465,23 @@ class Evaluator {
         positions.fillAllPositions();
         return positions;
     }
-    getArrayFormulasBlockedByOrSpreadingOn(position) {
+    /**
+     * Return the position of formulas blocked by the given position
+     * as well as all their dependencies.
+     */
+    getArrayFormulasBlockedBy(position) {
         if (!this.spreadingRelations.hasArrayFormulaResult(position)) {
             return [];
         }
         const arrayFormulas = this.spreadingRelations.getFormulaPositionsSpreadingOn(position);
         const positions = this.createEmptyPositionSet();
         positions.addMany(arrayFormulas);
-        positions.addMany(this.getCellsDependingOn(arrayFormulas));
+        const arrayFormulaPosition = this.getArrayFormulaSpreadingOn(position);
+        if (arrayFormulaPosition) {
+            // ignore the formula spreading on the position. Keep only the blocked ones
+            positions.delete(arrayFormulaPosition);
+        }
+        positions.addMany(this.getCellsDependingOn(positions));
         return positions;
     }
     nextPositionsToUpdate = new PositionSet({});
@@ -47261,7 +47621,7 @@ class Evaluator {
             }
             this.evaluatedCells.delete(child);
             this.nextPositionsToUpdate.addMany(this.getCellsDependingOn([child]));
-            this.nextPositionsToUpdate.addMany(this.getArrayFormulasBlockedByOrSpreadingOn(child));
+            this.nextPositionsToUpdate.addMany(this.getArrayFormulasBlockedBy(child));
         }
         this.spreadingRelations.removeNode(position);
     }
@@ -60240,6 +60600,6 @@ exports.tokenColors = tokenColors;
 exports.tokenize = tokenize;
 
 
-__info__.version = "17.2.4";
-__info__.date = "2024-04-18T16:41:38.407Z";
-__info__.hash = "0c66038";
+__info__.version = "17.2.5";
+__info__.date = "2024-04-26T07:41:13.193Z";
+__info__.hash = "a730f5c";