3 * Copyright 2012 Google Inc.
5 * Use of this source code is governed by a BSD-style license that can be
6 * found in the LICENSE file.
9 #ifndef SkRTree_DEFINED
10 #define SkRTree_DEFINED
13 #include "SkTDArray.h"
14 #include "SkChunkAlloc.h"
15 #include "SkBBoxHierarchy.h"
18 * An R-Tree implementation. In short, it is a balanced n-ary tree containing a hierarchy of
19 * bounding rectangles.
21 * Much like a B-Tree it maintains balance by enforcing minimum and maximum child counts, and
22 * splitting nodes when they become overfull. Unlike B-trees, however, we're using spatial data; so
23 * there isn't a canonical ordering to use when choosing insertion locations and splitting
24 * distributions. A variety of heuristics have been proposed for these problems; here, we're using
25 * something resembling an R*-tree, which attempts to minimize area and overlap during insertion,
26 * and aims to minimize a combination of margin, overlap, and area when splitting.
28 * One detail that is thus far unimplemented that may improve tree quality is attempting to remove
29 * and reinsert nodes when they become full, instead of immediately splitting (nodes that may have
30 * been placed well early on may hurt the tree later when more nodes have been added; removing
31 * and reinserting nodes generally helps reduce overlap and make a better tree). Deletion of nodes
32 * is also unimplemented.
34 * For more details see:
36 * Beckmann, N.; Kriegel, H. P.; Schneider, R.; Seeger, B. (1990). "The R*-tree:
37 * an efficient and robust access method for points and rectangles"
39 * It also supports bulk-loading from a batch of bounds and values; if you don't require the tree
40 * to be usable in its intermediate states while it is being constructed, this is significantly
41 * quicker than individual insertions and produces more consistent trees.
43 class SkRTree : public SkBBoxHierarchy {
45 SK_DECLARE_INST_COUNT(SkRTree)
48 * Create a new R-Tree with specified min/max child counts.
49 * The child counts are valid iff:
50 * - (max + 1) / 2 >= min (splitting an overfull node must be enough to populate 2 nodes)
54 * If you have some prior information about the distribution of bounds you're expecting, you
55 * can provide an optional aspect ratio parameter. This allows the bulk-load algorithm to create
56 * better proportioned tiles of rectangles.
58 static SkRTree* Create(int minChildren, int maxChildren, SkScalar aspectRatio = 1,
59 bool orderWhenBulkLoading = true);
62 virtual void insert(SkAutoTMalloc<SkRect>* boundsArray, int N) SK_OVERRIDE;
63 virtual void search(const SkRect& query, SkTDArray<unsigned>* results) const SK_OVERRIDE;
66 // Return the depth of the tree structure.
67 int getDepth() const { return this->isEmpty() ? 0 : fRoot.fChild.subtree->fLevel + 1; }
68 // Insertion count (not overall node count, which may be greater).
69 int getCount() const { return fCount; }
72 bool isEmpty() const { return 0 == this->getCount(); }
77 * A branch of the tree, this may contain a pointer to another interior node, or a data value
88 * A node in the tree, has between fMinChildren and fMaxChildren (the root is a special case)
91 uint16_t fNumChildren;
93 bool isLeaf() { return 0 == fLevel; }
94 // Since we want to be able to pick min/max child counts at runtime, we assume the creator
95 // has allocated sufficient space directly after us in memory, and index into that space
96 Branch* child(size_t index) {
97 return reinterpret_cast<Branch*>(this + 1) + index;
101 typedef int32_t SkIRect::*SortSide;
103 // Helper for sorting our children arrays by sides of their rects
104 struct RectLessThan {
105 RectLessThan(SkRTree::SortSide side) : fSide(side) { }
106 bool operator()(const SkRTree::Branch lhs, const SkRTree::Branch rhs) const {
107 return lhs.fBounds.*fSide < rhs.fBounds.*fSide;
110 const SkRTree::SortSide fSide;
114 bool operator()(const SkRTree::Branch lhs, const SkRTree::Branch rhs) {
115 return ((lhs.fBounds.fRight - lhs.fBounds.fLeft) >> 1) <
116 ((rhs.fBounds.fRight - lhs.fBounds.fLeft) >> 1);
121 bool operator()(const SkRTree::Branch lhs, const SkRTree::Branch rhs) {
122 return ((lhs.fBounds.fBottom - lhs.fBounds.fTop) >> 1) <
123 ((rhs.fBounds.fBottom - lhs.fBounds.fTop) >> 1);
127 SkRTree(int minChildren, int maxChildren, SkScalar aspectRatio, bool orderWhenBulkLoading);
130 * Recursively descend the tree to find an insertion position for 'branch', updates
131 * bounding boxes on the way up.
133 Branch* insert(Node* root, Branch* branch, uint16_t level = 0);
135 int chooseSubtree(Node* root, Branch* branch);
136 SkIRect computeBounds(Node* n);
137 int distributeChildren(Branch* children);
138 void search(Node* root, const SkIRect query, SkTDArray<unsigned>* results) const;
141 * This performs a bottom-up bulk load using the STR (sort-tile-recursive) algorithm, this
142 * seems to generally produce better, more consistent trees at significantly lower cost than
143 * repeated insertions.
145 * This consumes the input array.
147 * TODO: Experiment with other bulk-load algorithms (in particular the Hilbert pack variant,
148 * which groups rects by position on the Hilbert curve, is probably worth a look). There also
149 * exist top-down bulk load variants (VAMSplit, TopDownGreedy, etc).
151 Branch bulkLoad(SkTDArray<Branch>* branches, int level = 0);
153 void validate() const;
154 int validateSubtree(Node* root, SkIRect bounds, bool isRoot = false) const;
156 const int fMinChildren;
157 const int fMaxChildren;
158 const size_t fNodeSize;
160 // This is the count of data elements (rather than total nodes in the tree)
165 SkScalar fAspectRatio;
166 bool fSortWhenBulkLoading;
168 Node* allocateNode(uint16_t level);
170 typedef SkBBoxHierarchy INHERITED;