@epam/pdf-highlighter-kit 0.0.1

package/dist/core/text-segmentation.js

@@ -0,0 +1,338 @@
+export class TextSegmentation {
+    constructor(debugMode = false) {
+        this.debugMode = false;
+        this.debugMode = debugMode;
+    }
+    segmentText(textContent, highlights, pageNumber) {
+        const characterIndex = this.buildCharacterIndex(textContent.items);
+        const highlightRanges = this.mapHighlightsToCharacters(highlights, pageNumber, textContent.items);
+        const segmentationPoints = this.findSegmentationPoints(textContent.items, highlightRanges);
+        const segments = this.createSegments(textContent.items, characterIndex, segmentationPoints, highlightRanges);
+        return this.optimizeSegments(segments);
+    }
+    buildCharacterIndex(textItems) {
+        const index = [];
+        let globalIndex = 0;
+        textItems.forEach((item, itemIndex) => {
+            const itemBounds = this.getTextItemBounds(item);
+            const charWidth = item.width / item.str.length;
+            for (let charIndex = 0; charIndex < item.str.length; charIndex++) {
+                const char = item.str[charIndex];
+                const charBounds = {
+                    x1: itemBounds.x1 + charIndex * charWidth,
+                    y1: itemBounds.y1,
+                    x2: itemBounds.x1 + (charIndex + 1) * charWidth,
+                    y2: itemBounds.y2,
+                };
+                index.push({
+                    char,
+                    itemIndex,
+                    charIndex,
+                    globalIndex,
+                    bounds: charBounds,
+                });
+                globalIndex++;
+            }
+        });
+        return index;
+    }
+    mapHighlightsToCharacters(highlights, pageNumber, textItems) {
+        const ranges = [];
+        Object.entries(highlights).forEach(([category, categoryData]) => {
+            const pageHighlights = categoryData.pages[pageNumber.toString()];
+            if (!pageHighlights)
+                return;
+            pageHighlights.forEach((occurrence, occIndex) => {
+                occurrence.coordinates.forEach((coord, _coordIndex) => {
+                    const textMatches = this.findTextMatchesForCoordinate(coord, textItems);
+                    textMatches.forEach((match) => {
+                        ranges.push({
+                            startIndex: match.startIndex,
+                            endIndex: match.endIndex,
+                            termId: occurrence.termId,
+                            category,
+                            coordinates: coord,
+                            priority: this.calculateHighlightPriority(category, occIndex),
+                        });
+                    });
+                });
+            });
+        });
+        ranges.sort((a, b) => {
+            if (a.priority !== b.priority) {
+                return b.priority - a.priority;
+            }
+            return a.startIndex - b.startIndex;
+        });
+        return this.resolveOverlappingHighlights(ranges);
+    }
+    findTextMatchesForCoordinate(coord, textItems) {
+        const matches = [];
+        let globalIndex = 0;
+        textItems.forEach((item) => {
+            const itemBounds = this.getTextItemBounds(item);
+            if (this.boundsIntersect(itemBounds, coord)) {
+                const intersectionArea = this.calculateIntersectionArea(itemBounds, coord);
+                const itemArea = (itemBounds.x2 - itemBounds.x1) * (itemBounds.y2 - itemBounds.y1);
+                const overlapPercent = intersectionArea / itemArea;
+                if (overlapPercent > 0.5) {
+                    matches.push({
+                        startIndex: globalIndex,
+                        endIndex: globalIndex + item.str.length,
+                    });
+                }
+            }
+            globalIndex += item.str.length;
+        });
+        return this.mergeAdjacentMatches(matches);
+    }
+    calculateIntersectionArea(bounds1, bounds2) {
+        const left = Math.max(bounds1.x1, bounds2.x1);
+        const right = Math.min(bounds1.x2, bounds2.x2);
+        const top = Math.max(bounds1.y1, bounds2.y1);
+        const bottom = Math.min(bounds1.y2, bounds2.y2);
+        if (left >= right || top >= bottom) {
+            return 0;
+        }
+        return (right - left) * (bottom - top);
+    }
+    mergeAdjacentMatches(matches) {
+        if (matches.length <= 1)
+            return matches;
+        const merged = [];
+        let current = matches[0];
+        for (let i = 1; i < matches.length; i++) {
+            const next = matches[i];
+            if (current.endIndex >= next.startIndex) {
+                current.endIndex = Math.max(current.endIndex, next.endIndex);
+            }
+            else {
+                merged.push(current);
+                current = next;
+            }
+        }
+        merged.push(current);
+        return merged;
+    }
+    resolveOverlappingHighlights(ranges) {
+        const resolved = [];
+        for (const range of ranges) {
+            const overlapping = resolved.filter((existing) => this.rangesOverlap(range, existing));
+            if (overlapping.length === 0) {
+                resolved.push(range);
+            }
+            else {
+                const splitRanges = this.splitRangeAroundOverlaps(range, overlapping);
+                resolved.push(...splitRanges);
+            }
+        }
+        return resolved;
+    }
+    rangesOverlap(range1, range2) {
+        return !(range1.endIndex <= range2.startIndex || range1.startIndex >= range2.endIndex);
+    }
+    splitRangeAroundOverlaps(range, overlapping) {
+        const splits = [];
+        let currentStart = range.startIndex;
+        overlapping.sort((a, b) => a.startIndex - b.startIndex);
+        for (const overlap of overlapping) {
+            if (currentStart < overlap.startIndex) {
+                splits.push({
+                    ...range,
+                    startIndex: currentStart,
+                    endIndex: Math.min(range.endIndex, overlap.startIndex),
+                });
+            }
+            currentStart = Math.max(currentStart, overlap.endIndex);
+        }
+        if (currentStart < range.endIndex) {
+            splits.push({
+                ...range,
+                startIndex: currentStart,
+                endIndex: range.endIndex,
+            });
+        }
+        return splits.filter((split) => split.startIndex < split.endIndex);
+    }
+    calculateHighlightPriority(category, occurrenceIndex) {
+        const categoryPriorities = {
+            protein: 100,
+            gene: 90,
+            disease: 80,
+            chemical: 70,
+            species: 60,
+            default: 50,
+        };
+        const basePriority = categoryPriorities[category] || categoryPriorities['default'];
+        return basePriority - occurrenceIndex;
+    }
+    findSegmentationPoints(textItems, highlightRanges) {
+        const points = [];
+        let globalIndex = 0;
+        textItems.forEach((item, itemIndex) => {
+            if (itemIndex > 0) {
+                points.push({
+                    globalIndex,
+                    itemIndex,
+                    charIndex: 0,
+                    reason: 'font-change',
+                });
+            }
+            globalIndex += item.str.length;
+        });
+        highlightRanges.forEach((range) => {
+            points.push({
+                globalIndex: range.startIndex,
+                itemIndex: this.getItemIndexForGlobalIndex(range.startIndex, textItems),
+                charIndex: this.getCharIndexForGlobalIndex(range.startIndex, textItems),
+                reason: 'highlight-start',
+            });
+            points.push({
+                globalIndex: range.endIndex,
+                itemIndex: this.getItemIndexForGlobalIndex(range.endIndex, textItems),
+                charIndex: this.getCharIndexForGlobalIndex(range.endIndex, textItems),
+                reason: 'highlight-end',
+            });
+        });
+        points.sort((a, b) => a.globalIndex - b.globalIndex);
+        return this.removeDuplicatePoints(points);
+    }
+    getItemIndexForGlobalIndex(globalIndex, textItems) {
+        let currentIndex = 0;
+        for (let i = 0; i < textItems.length; i++) {
+            if (globalIndex < currentIndex + textItems[i].str.length) {
+                return i;
+            }
+            currentIndex += textItems[i].str.length;
+        }
+        return textItems.length - 1;
+    }
+    getCharIndexForGlobalIndex(globalIndex, textItems) {
+        let currentIndex = 0;
+        for (let i = 0; i < textItems.length; i++) {
+            if (globalIndex < currentIndex + textItems[i].str.length) {
+                return globalIndex - currentIndex;
+            }
+            currentIndex += textItems[i].str.length;
+        }
+        return 0;
+    }
+    removeDuplicatePoints(points) {
+        const unique = [];
+        let lastIndex = -1;
+        for (const point of points) {
+            if (point.globalIndex !== lastIndex) {
+                unique.push(point);
+                lastIndex = point.globalIndex;
+            }
+        }
+        return unique;
+    }
+    createSegments(textItems, characterIndex, segmentationPoints, highlightRanges) {
+        const segments = [];
+        for (let i = 0; i < segmentationPoints.length - 1; i++) {
+            const startPoint = segmentationPoints[i];
+            const endPoint = segmentationPoints[i + 1];
+            const segment = this.createSegment(startPoint.globalIndex, endPoint.globalIndex, textItems, characterIndex, highlightRanges);
+            if (segment) {
+                segments.push(segment);
+            }
+        }
+        return segments;
+    }
+    createSegment(startIndex, endIndex, textItems, characterIndex, highlightRanges) {
+        if (startIndex >= endIndex || startIndex >= characterIndex.length) {
+            return null;
+        }
+        const startChar = characterIndex[startIndex];
+        const endIdx = Math.min(endIndex - 1, characterIndex.length - 1);
+        const endChar = characterIndex[endIdx];
+        const text = characterIndex
+            .slice(startIndex, endIndex)
+            .map((char) => char.char)
+            .join('');
+        const bounds = {
+            x1: startChar.bounds.x1,
+            y1: Math.min(startChar.bounds.y1, endChar.bounds.y1),
+            x2: endChar.bounds.x2,
+            y2: Math.max(startChar.bounds.y2, endChar.bounds.y2),
+        };
+        const overlappingHighlight = highlightRanges.find((range) => startIndex >= range.startIndex && endIndex <= range.endIndex);
+        const textItem = textItems[startChar.itemIndex];
+        return {
+            text,
+            bounds,
+            hasHighlight: !!overlappingHighlight,
+            highlightInfo: overlappingHighlight
+                ? {
+                    termId: overlappingHighlight.termId,
+                    category: overlappingHighlight.category,
+                }
+                : undefined,
+            transform: textItem.transform,
+            fontName: textItem.fontName,
+        };
+    }
+    optimizeSegments(segments) {
+        const optimized = [];
+        let current = null;
+        for (const segment of segments) {
+            if (!current) {
+                current = segment;
+                continue;
+            }
+            if (this.canMergeSegments(current, segment)) {
+                current = this.mergeSegments(current, segment);
+            }
+            else {
+                optimized.push(current);
+                current = segment;
+            }
+        }
+        if (current) {
+            optimized.push(current);
+        }
+        return optimized;
+    }
+    canMergeSegments(segment1, segment2) {
+        if (segment1.hasHighlight !== segment2.hasHighlight) {
+            return false;
+        }
+        if (segment1.hasHighlight && segment2.hasHighlight) {
+            return (segment1.highlightInfo?.termId === segment2.highlightInfo?.termId &&
+                segment1.highlightInfo?.category === segment2.highlightInfo?.category);
+        }
+        return segment1.fontName === segment2.fontName;
+    }
+    mergeSegments(segment1, segment2) {
+        return {
+            text: segment1.text + segment2.text,
+            bounds: {
+                x1: Math.min(segment1.bounds.x1, segment2.bounds.x1),
+                y1: Math.min(segment1.bounds.y1, segment2.bounds.y1),
+                x2: Math.max(segment1.bounds.x2, segment2.bounds.x2),
+                y2: Math.max(segment1.bounds.y2, segment2.bounds.y2),
+            },
+            hasHighlight: segment1.hasHighlight,
+            highlightInfo: segment1.highlightInfo,
+            transform: segment1.transform,
+            fontName: segment1.fontName,
+        };
+    }
+    getTextItemBounds(item) {
+        const transform = item.transform;
+        return {
+            x1: transform[4],
+            y1: transform[5] - item.height,
+            x2: transform[4] + item.width,
+            y2: transform[5],
+        };
+    }
+    boundsIntersect(bounds1, bounds2) {
+        return !(bounds1.x2 < bounds2.x1 ||
+            bounds1.x1 > bounds2.x2 ||
+            bounds1.y2 < bounds2.y1 ||
+            bounds1.y1 > bounds2.y2);
+    }
+}
+//# sourceMappingURL=text-segmentation.js.map

package/dist/core/text-segmentation.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"text-segmentation.js","sourceRoot":"","sources":["../../core/text-segmentation.ts"],"names":[],"mappings":"AAkBA,MAAM,OAAO,gBAAgB;IAG3B,YAAY,SAAS,GAAG,KAAK;QAFrB,cAAS,GAAG,KAAK,CAAC;QAGxB,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC;IAC7B,CAAC;IAED,WAAW,CAAC,WAAwB,EAAE,UAAyB,EAAE,UAAkB;QACjF,MAAM,cAAc,GAAG,IAAI,CAAC,mBAAmB,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC;QAEnE,MAAM,eAAe,GAAG,IAAI,CAAC,yBAAyB,CACpD,UAAU,EACV,UAAU,EACV,WAAW,CAAC,KAAK,CAClB,CAAC;QAEF,MAAM,kBAAkB,GAAG,IAAI,CAAC,sBAAsB,CAAC,WAAW,CAAC,KAAK,EAAE,eAAe,CAAC,CAAC;QAE3F,MAAM,QAAQ,GAAG,IAAI,CAAC,cAAc,CAClC,WAAW,CAAC,KAAK,EACjB,cAAc,EACd,kBAAkB,EAClB,eAAe,CAChB,CAAC;QAEF,OAAO,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,CAAC;IACzC,CAAC;IAEO,mBAAmB,CAAC,SAAqB;QAO/C,MAAM,KAAK,GAMN,EAAE,CAAC;QAER,IAAI,WAAW,GAAG,CAAC,CAAC;QAEpB,SAAS,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,SAAS,EAAE,EAAE;YACpC,MAAM,UAAU,GAAG,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC;YAChD,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC;YAE/C,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,CAAC;gBACjE,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;gBAEjC,MAAM,UAAU,GAAgB;oBAC9B,EAAE,EAAE,UAAU,CAAC,EAAE,GAAG,SAAS,GAAG,SAAS;oBACzC,EAAE,EAAE,UAAU,CAAC,EAAE;oBACjB,EAAE,EAAE,UAAU,CAAC,EAAE,GAAG,CAAC,SAAS,GAAG,CAAC,CAAC,GAAG,SAAS;oBAC/C,EAAE,EAAE,UAAU,CAAC,EAAE;iBAClB,CAAC;gBAEF,KAAK,CAAC,IAAI,CAAC;oBACT,IAAI;oBACJ,SAAS;oBACT,SAAS;oBACT,WAAW;oBACX,MAAM,EAAE,UAAU;iBACnB,CAAC,CAAC;gBAEH,WAAW,EAAE,CAAC;YAChB,CAAC;QACH,CAAC,CAAC,CAAC;QAEH,OAAO,KAAK,CAAC;IACf,CAAC;IAEO,yBAAyB,CAC/B,UAAyB,EACzB,UAAkB,EAClB,SAAqB;QAErB,MAAM,MAAM,GAAqB,EAAE,CAAC;QAEpC,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,QAAQ,EAAE,YAAY,CAAC,EAAE,EAAE;YAC9D,MAAM,cAAc,GAAG,YAAY,CAAC,KAAK,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;YACjE,IAAI,CAAC,cAAc;gBAAE,OAAO;YAE5B,cAAc,CAAC,OAAO,CAAC,CAAC,UAAU,EAAE,QAAQ,EAAE,EAAE;gBAC9C,UAAU,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,WAAW,EAAE,EAAE;oBACpD,MAAM,WAAW,GAAG,IAAI,CAAC,4BAA4B,CAAC,KAAK,EAAE,SAAS,CAAC,CAAC;oBAExE,WAAW,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE;wBAC5B,MAAM,CAAC,IAAI,CAAC;4BACV,UAAU,EAAE,KAAK,CAAC,UAAU;4BAC5B,QAAQ,EAAE,KAAK,CAAC,QAAQ;4BACxB,MAAM,EAAE,UAAU,CAAC,MAAM;4BACzB,QAAQ;4BACR,WAAW,EAAE,KAAK;4BAClB,QAAQ,EAAE,IAAI,CAAC,0BAA0B,CAAC,QAAQ,EAAE,QAAQ,CAAC;yBAC9D,CAAC,CAAC;oBACL,CAAC,CAAC,CAAC;gBACL,CAAC,CAAC,CAAC;YACL,CAAC,CAAC,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;YACnB,IAAI,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,QAAQ,EAAE,CAAC;gBAC9B,OAAO,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC,QAAQ,CAAC;YACjC,CAAC;YACD,OAAO,CAAC,CAAC,UAAU,GAAG,CAAC,CAAC,UAAU,CAAC;QACrC,CAAC,CAAC,CAAC;QAEH,OAAO,IAAI,CAAC,4BAA4B,CAAC,MAAM,CAAC,CAAC;IACnD,CAAC;IAEO,4BAA4B,CAClC,KAAkB,EAClB,SAAqB;QAErB,MAAM,OAAO,GAAoD,EAAE,CAAC;QACpE,IAAI,WAAW,GAAG,CAAC,CAAC;QAEpB,SAAS,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE;YACzB,MAAM,UAAU,GAAG,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC;YAEhD,IAAI,IAAI,CAAC,eAAe,CAAC,UAAU,EAAE,KAAK,CAAC,EAAE,CAAC;gBAC5C,MAAM,gBAAgB,GAAG,IAAI,CAAC,yBAAyB,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC;gBAC3E,MAAM,QAAQ,GAAG,CAAC,UAAU,CAAC,EAAE,GAAG,UAAU,CAAC,EAAE,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE,GAAG,UAAU,CAAC,EAAE,CAAC,CAAC;gBACnF,MAAM,cAAc,GAAG,gBAAgB,GAAG,QAAQ,CAAC;gBAEnD,IAAI,cAAc,GAAG,GAAG,EAAE,CAAC;oBACzB,OAAO,CAAC,IAAI,CAAC;wBACX,UAAU,EAAE,WAAW;wBACvB,QAAQ,EAAE,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM;qBACxC,CAAC,CAAC;gBACL,CAAC;YACH,CAAC;YAED,WAAW,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC;QACjC,CAAC,CAAC,CAAC;QAEH,OAAO,IAAI,CAAC,oBAAoB,CAAC,OAAO,CAAC,CAAC;IAC5C,CAAC;IAEO,yBAAyB,CAAC,OAAoB,EAAE,OAAoB;QAC1E,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC;QAC9C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC;QAC/C,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC;QAC7C,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC;QAEhD,IAAI,IAAI,IAAI,KAAK,IAAI,GAAG,IAAI,MAAM,EAAE,CAAC;YACnC,OAAO,CAAC,CAAC;QACX,CAAC;QAED,OAAO,CAAC,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC;IACzC,CAAC;IAEO,oBAAoB,CAC1B,OAAwD;QAExD,IAAI,OAAO,CAAC,MAAM,IAAI,CAAC;YAAE,OAAO,OAAO,CAAC;QAExC,MAAM,MAAM,GAAoD,EAAE,CAAC;QACnE,IAAI,OAAO,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;QAEzB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACxC,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;YAExB,IAAI,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;gBACxC,OAAO,CAAC,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;YAC/D,CAAC;iBAAM,CAAC;gBACN,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;gBACrB,OAAO,GAAG,IAAI,CAAC;YACjB,CAAC;QACH,CAAC;QAED,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACrB,OAAO,MAAM,CAAC;IAChB,CAAC;IAEO,4BAA4B,CAAC,MAAwB;QAC3D,MAAM,QAAQ,GAAqB,EAAE,CAAC;QAEtC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,MAAM,WAAW,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,IAAI,CAAC,aAAa,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,CAAC;YAEvF,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBAC7B,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YACvB,CAAC;iBAAM,CAAC;gBACN,MAAM,WAAW,GAAG,IAAI,CAAC,wBAAwB,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC;gBACtE,QAAQ,CAAC,IAAI,CAAC,GAAG,WAAW,CAAC,CAAC;YAChC,CAAC;QACH,CAAC;QAED,OAAO,QAAQ,CAAC;IAClB,CAAC;IAEO,aAAa,CAAC,MAAsB,EAAE,MAAsB;QAClE,OAAO,CAAC,CAAC,MAAM,CAAC,QAAQ,IAAI,MAAM,CAAC,UAAU,IAAI,MAAM,CAAC,UAAU,IAAI,MAAM,CAAC,QAAQ,CAAC,CAAC;IACzF,CAAC;IAEO,wBAAwB,CAC9B,KAAqB,EACrB,WAA6B;QAE7B,MAAM,MAAM,GAAqB,EAAE,CAAC;QACpC,IAAI,YAAY,GAAG,KAAK,CAAC,UAAU,CAAC;QAEpC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC;QAExD,KAAK,MAAM,OAAO,IAAI,WAAW,EAAE,CAAC;YAClC,IAAI,YAAY,GAAG,OAAO,CAAC,UAAU,EAAE,CAAC;gBACtC,MAAM,CAAC,IAAI,CAAC;oBACV,GAAG,KAAK;oBACR,UAAU,EAAE,YAAY;oBACxB,QAAQ,EAAE,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,QAAQ,EAAE,OAAO,CAAC,UAAU,CAAC;iBACvD,CAAC,CAAC;YACL,CAAC;YAED,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,YAAY,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC;QAC1D,CAAC;QAED,IAAI,YAAY,GAAG,KAAK,CAAC,QAAQ,EAAE,CAAC;YAClC,MAAM,CAAC,IAAI,CAAC;gBACV,GAAG,KAAK;gBACR,UAAU,EAAE,YAAY;gBACxB,QAAQ,EAAE,KAAK,CAAC,QAAQ;aACzB,CAAC,CAAC;QACL,CAAC;QAED,OAAO,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,UAAU,GAAG,KAAK,CAAC,QAAQ,CAAC,CAAC;IACrE,CAAC;IAEO,0BAA0B,CAAC,QAAgB,EAAE,eAAuB;QAC1E,MAAM,kBAAkB,GAA2B;YACjD,OAAO,EAAE,GAAG;YACZ,IAAI,EAAE,EAAE;YACR,OAAO,EAAE,EAAE;YACX,QAAQ,EAAE,EAAE;YACZ,OAAO,EAAE,EAAE;YACX,OAAO,EAAE,EAAE;SACZ,CAAC;QAEF,MAAM,YAAY,GAAG,kBAAkB,CAAC,QAAQ,CAAC,IAAI,kBAAkB,CAAC,SAAS,CAAC,CAAC;QACnF,OAAO,YAAY,GAAG,eAAe,CAAC;IACxC,CAAC;IAEO,sBAAsB,CAC5B,SAAqB,EACrB,eAAiC;QAEjC,MAAM,MAAM,GAAwB,EAAE,CAAC;QACvC,IAAI,WAAW,GAAG,CAAC,CAAC;QAEpB,SAAS,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,SAAS,EAAE,EAAE;YACpC,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;gBAClB,MAAM,CAAC,IAAI,CAAC;oBACV,WAAW;oBACX,SAAS;oBACT,SAAS,EAAE,CAAC;oBACZ,MAAM,EAAE,aAAa;iBACtB,CAAC,CAAC;YACL,CAAC;YACD,WAAW,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC;QACjC,CAAC,CAAC,CAAC;QAEH,eAAe,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE;YAChC,MAAM,CAAC,IAAI,CAAC;gBACV,WAAW,EAAE,KAAK,CAAC,UAAU;gBAC7B,SAAS,EAAE,IAAI,CAAC,0BAA0B,CAAC,KAAK,CAAC,UAAU,EAAE,SAAS,CAAC;gBACvE,SAAS,EAAE,IAAI,CAAC,0BAA0B,CAAC,KAAK,CAAC,UAAU,EAAE,SAAS,CAAC;gBACvE,MAAM,EAAE,iBAAiB;aAC1B,CAAC,CAAC;YAEH,MAAM,CAAC,IAAI,CAAC;gBACV,WAAW,EAAE,KAAK,CAAC,QAAQ;gBAC3B,SAAS,EAAE,IAAI,CAAC,0BAA0B,CAAC,KAAK,CAAC,QAAQ,EAAE,SAAS,CAAC;gBACrE,SAAS,EAAE,IAAI,CAAC,0BAA0B,CAAC,KAAK,CAAC,QAAQ,EAAE,SAAS,CAAC;gBACrE,MAAM,EAAE,eAAe;aACxB,CAAC,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,GAAG,CAAC,CAAC,WAAW,CAAC,CAAC;QACrD,OAAO,IAAI,CAAC,qBAAqB,CAAC,MAAM,CAAC,CAAC;IAC5C,CAAC;IAEO,0BAA0B,CAAC,WAAmB,EAAE,SAAqB;QAC3E,IAAI,YAAY,GAAG,CAAC,CAAC;QACrB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAC1C,IAAI,WAAW,GAAG,YAAY,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC;gBACzD,OAAO,CAAC,CAAC;YACX,CAAC;YACD,YAAY,IAAI,SAAS,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC;QAC1C,CAAC;QACD,OAAO,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC;IAC9B,CAAC;IAEO,0BAA0B,CAAC,WAAmB,EAAE,SAAqB;QAC3E,IAAI,YAAY,GAAG,CAAC,CAAC;QACrB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAC1C,IAAI,WAAW,GAAG,YAAY,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC;gBACzD,OAAO,WAAW,GAAG,YAAY,CAAC;YACpC,CAAC;YACD,YAAY,IAAI,SAAS,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC;QAC1C,CAAC;QACD,OAAO,CAAC,CAAC;IACX,CAAC;IAEO,qBAAqB,CAAC,MAA2B;QACvD,MAAM,MAAM,GAAwB,EAAE,CAAC;QACvC,IAAI,SAAS,GAAG,CAAC,CAAC,CAAC;QAEnB,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,IAAI,KAAK,CAAC,WAAW,KAAK,SAAS,EAAE,CAAC;gBACpC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;gBACnB,SAAS,GAAG,KAAK,CAAC,WAAW,CAAC;YAChC,CAAC;QACH,CAAC;QAED,OAAO,MAAM,CAAC;IAChB,CAAC;IAEO,cAAc,CACpB,SAAqB,EACrB,cAA2D,EAC3D,kBAAuC,EACvC,eAAiC;QAEjC,MAAM,QAAQ,GAAc,EAAE,CAAC;QAE/B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,kBAAkB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;YACvD,MAAM,UAAU,GAAG,kBAAkB,CAAC,CAAC,CAAC,CAAC;YACzC,MAAM,QAAQ,GAAG,kBAAkB,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YAE3C,MAAM,OAAO,GAAG,IAAI,CAAC,aAAa,CAChC,UAAU,CAAC,WAAW,EACtB,QAAQ,CAAC,WAAW,EACpB,SAAS,EACT,cAAc,EACd,eAAe,CAChB,CAAC;YAEF,IAAI,OAAO,EAAE,CAAC;gBACZ,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YACzB,CAAC;QACH,CAAC;QAED,OAAO,QAAQ,CAAC;IAClB,CAAC;IAEO,aAAa,CACnB,UAAkB,EAClB,QAAgB,EAChB,SAAqB,EACrB,cAA2D,EAC3D,eAAiC;QAEjC,IAAI,UAAU,IAAI,QAAQ,IAAI,UAAU,IAAI,cAAc,CAAC,MAAM,EAAE,CAAC;YAClE,OAAO,IAAI,CAAC;QACd,CAAC;QAED,MAAM,SAAS,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QAC7C,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,CAAC,EAAE,cAAc,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACjE,MAAM,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;QAEvC,MAAM,IAAI,GAAG,cAAc;aACxB,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC;aAC3B,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC;aACxB,IAAI,CAAC,EAAE,CAAC,CAAC;QAEZ,MAAM,MAAM,GAAgB;YAC1B,EAAE,EAAE,SAAS,CAAC,MAAM,CAAC,EAAE;YACvB,EAAE,EAAE,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,MAAM,CAAC,EAAE,EAAE,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;YACpD,EAAE,EAAE,OAAO,CAAC,MAAM,CAAC,EAAE;YACrB,EAAE,EAAE,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,MAAM,CAAC,EAAE,EAAE,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;SACrD,CAAC;QAEF,MAAM,oBAAoB,GAAG,eAAe,CAAC,IAAI,CAC/C,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,IAAI,KAAK,CAAC,UAAU,IAAI,QAAQ,IAAI,KAAK,CAAC,QAAQ,CACxE,CAAC;QAEF,MAAM,QAAQ,GAAG,SAAS,CAAC,SAAS,CAAC,SAAS,CAAC,CAAC;QAEhD,OAAO;YACL,IAAI;YACJ,MAAM;YACN,YAAY,EAAE,CAAC,CAAC,oBAAoB;YACpC,aAAa,EAAE,oBAAoB;gBACjC,CAAC,CAAC;oBACE,MAAM,EAAE,oBAAoB,CAAC,MAAM;oBACnC,QAAQ,EAAE,oBAAoB,CAAC,QAAQ;iBACxC;gBACH,CAAC,CAAC,SAAS;YACb,SAAS,EAAE,QAAQ,CAAC,SAAS;YAC7B,QAAQ,EAAE,QAAQ,CAAC,QAAQ;SAC5B,CAAC;IACJ,CAAC;IAEO,gBAAgB,CAAC,QAAmB;QAC1C,MAAM,SAAS,GAAc,EAAE,CAAC;QAChC,IAAI,OAAO,GAAmB,IAAI,CAAC;QAEnC,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;YAC/B,IAAI,CAAC,OAAO,EAAE,CAAC;gBACb,OAAO,GAAG,OAAO,CAAC;gBAClB,SAAS;YACX,CAAC;YAED,IAAI,IAAI,CAAC,gBAAgB,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,CAAC;gBAC5C,OAAO,GAAG,IAAI,CAAC,aAAa,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;YACjD,CAAC;iBAAM,CAAC;gBACN,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;gBACxB,OAAO,GAAG,OAAO,CAAC;YACpB,CAAC;QACH,CAAC;QAED,IAAI,OAAO,EAAE,CAAC;YACZ,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC1B,CAAC;QAED,OAAO,SAAS,CAAC;IACnB,CAAC;IAEO,gBAAgB,CAAC,QAAiB,EAAE,QAAiB;QAC3D,IAAI,QAAQ,CAAC,YAAY,KAAK,QAAQ,CAAC,YAAY,EAAE,CAAC;YACpD,OAAO,KAAK,CAAC;QACf,CAAC;QAED,IAAI,QAAQ,CAAC,YAAY,IAAI,QAAQ,CAAC,YAAY,EAAE,CAAC;YACnD,OAAO,CACL,QAAQ,CAAC,aAAa,EAAE,MAAM,KAAK,QAAQ,CAAC,aAAa,EAAE,MAAM;gBACjE,QAAQ,CAAC,aAAa,EAAE,QAAQ,KAAK,QAAQ,CAAC,aAAa,EAAE,QAAQ,CACtE,CAAC;QACJ,CAAC;QAED,OAAO,QAAQ,CAAC,QAAQ,KAAK,QAAQ,CAAC,QAAQ,CAAC;IACjD,CAAC;IAEO,aAAa,CAAC,QAAiB,EAAE,QAAiB;QACxD,OAAO;YACL,IAAI,EAAE,QAAQ,CAAC,IAAI,GAAG,QAAQ,CAAC,IAAI;YACnC,MAAM,EAAE;gBACN,EAAE,EAAE,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,EAAE,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;gBACpD,EAAE,EAAE,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,EAAE,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;gBACpD,EAAE,EAAE,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,EAAE,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;gBACpD,EAAE,EAAE,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,EAAE,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;aACrD;YACD,YAAY,EAAE,QAAQ,CAAC,YAAY;YACnC,aAAa,EAAE,QAAQ,CAAC,aAAa;YACrC,SAAS,EAAE,QAAQ,CAAC,SAAS;YAC7B,QAAQ,EAAE,QAAQ,CAAC,QAAQ;SAC5B,CAAC;IACJ,CAAC;IAEO,iBAAiB,CAAC,IAAc;QACtC,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC;QACjC,OAAO;YACL,EAAE,EAAE,SAAS,CAAC,CAAC,CAAC;YAChB,EAAE,EAAE,SAAS,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,MAAM;YAC9B,EAAE,EAAE,SAAS,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK;YAC7B,EAAE,EAAE,SAAS,CAAC,CAAC,CAAC;SACjB,CAAC;IACJ,CAAC;IAEO,eAAe,CAAC,OAAoB,EAAE,OAAoB;QAChE,OAAO,CAAC,CACN,OAAO,CAAC,EAAE,GAAG,OAAO,CAAC,EAAE;YACvB,OAAO,CAAC,EAAE,GAAG,OAAO,CAAC,EAAE;YACvB,OAAO,CAAC,EAAE,GAAG,OAAO,CAAC,EAAE;YACvB,OAAO,CAAC,EAAE,GAAG,OAAO,CAAC,EAAE,CACxB,CAAC;IACJ,CAAC;CACF"}

package/dist/core/unified-layer-builder.d.ts

@@ -0,0 +1,27 @@
+import { TextContent, HighlightData } from '../types';
+export declare class UnifiedLayerBuilder {
+    private pageContainer;
+    private unifiedLayer;
+    constructor();
+    buildUnifiedLayer(pageContainer: HTMLElement, textContent: TextContent, highlights: HighlightData, pageNumber: number, scale?: number): Promise<HTMLElement>;
+    private analyzeContent;
+    private findIntersectingTextItems;
+    private getTextItemBounds;
+    private boundsIntersect;
+    private segmentTextWithHighlights;
+    private getHighlightsForTextItem;
+    private createHighlightedSegments;
+    private mergeAdjacentSegments;
+    private canMergeSegments;
+    private mergeSegments;
+    private buildDOM;
+    private createTextElement;
+    private createHighlightElement;
+    private applyTextPositioning;
+    private positionLayer;
+    updateHighlights(highlights: HighlightData, pageNumber: number, textContent?: TextContent): void;
+    private updateExistingHighlights;
+    clear(): void;
+    getUnifiedLayer(): HTMLElement | null;
+}
+//# sourceMappingURL=unified-layer-builder.d.ts.map

package/dist/core/unified-layer-builder.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"unified-layer-builder.d.ts","sourceRoot":"","sources":["../../core/unified-layer-builder.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,WAAW,EACX,aAAa,EAMd,MAAM,UAAU,CAAC;AAElB,qBAAa,mBAAmB;IAC9B,OAAO,CAAC,aAAa,CAA4B;IACjD,OAAO,CAAC,YAAY,CAA4B;;IAK1C,iBAAiB,CACrB,aAAa,EAAE,WAAW,EAC1B,WAAW,EAAE,WAAW,EACxB,UAAU,EAAE,aAAa,EACzB,UAAU,EAAE,MAAM,EAClB,KAAK,SAAM,GACV,OAAO,CAAC,WAAW,CAAC;IAevB,OAAO,CAAC,cAAc;IAyCtB,OAAO,CAAC,yBAAyB;IA8BjC,OAAO,CAAC,iBAAiB;IAazB,OAAO,CAAC,eAAe;IASvB,OAAO,CAAC,yBAAyB;IAmCjC,OAAO,CAAC,wBAAwB;IA4BhC,OAAO,CAAC,yBAAyB;IAwBjC,OAAO,CAAC,qBAAqB;IAsB7B,OAAO,CAAC,gBAAgB;IAexB,OAAO,CAAC,aAAa;IAgBrB,OAAO,CAAC,QAAQ;IA0BhB,OAAO,CAAC,iBAAiB;IAUzB,OAAO,CAAC,sBAAsB;IAoB9B,OAAO,CAAC,oBAAoB;IAmB5B,OAAO,CAAC,aAAa;IAyDrB,gBAAgB,CAAC,UAAU,EAAE,aAAa,EAAE,UAAU,EAAE,MAAM,EAAE,WAAW,CAAC,EAAE,WAAW,GAAG,IAAI;IAYhG,OAAO,CAAC,wBAAwB;IAyBhC,KAAK,IAAI,IAAI;IAQb,eAAe,IAAI,WAAW,GAAG,IAAI;CAGtC"}