예제 #1
0
SuffixTree buildTempSuffixTree(const vector <int> &input) {
    if (input.size() == 0) {
        return SuffixTree();
    }
    if (input.size() == 1) {
        SuffixTree result;
        unsigned int newNodeIndex = result.newNode(result.root, 0, 1, 0);
        result[result.root].push_back(newNodeIndex);
        return result;
    }

    SuffixTree compressed = buildTempSuffixTree(compressInput(input));
    decompress(compressed, input);

    SuffixTree &even = compressed;
    checkTree("EVEN:\n", even, input);

    SuffixTree odd = buildOddSuffixTree(even, input);
    checkTree("ODD:\n", odd, input);

    SuffixTree almostResult = mergeTrees(even, odd, input);
    checkTree("ALMOST:\n", almostResult, input);

    cleanTreeDfs(almostResult.root, -1, almostResult);
    checkTree("RESL:\n", almostResult, input);

    return almostResult;
}
예제 #2
0
unsigned int appendCopyNode(const SuffixTree &from, SuffixTree &to,
    unsigned int toStart, unsigned int u
) {
    unsigned int newStart = to.newNode(toStart);
    to[toStart].push_back(newStart);
    copyNodeExceptParentAndChildren(from[u], to[newStart]);
    copySubTree(from, to, u, newStart);
    return newStart;
}
예제 #3
0
void correctMerge(unsigned int v, unsigned int parentsPlace, SuffixTree &merged,
    IndexedPair<MergeTreesStruct> &updatedTrees,
    const vector <unsigned int> &trueLength, const vector <int> &input,
    unsigned int copyTree
) {
    if (merged[v].isHiddenInfo()) {
        doSomething(updatedTrees, [&merged, &v] (MergeTreesStruct &tree) {
            tree.evaluate(merged, v);
        });
        if (copyTree == 2 && merged[v].depth != trueLength[v]) {
            unsigned int commonLength = trueLength[v]
                - merged[merged[v].parent].depth;
            MergeTreesStruct& donor = minimal(updatedTrees,
                [&input, &commonLength] (MergeTreesStruct &tree)
                    -> int {
                    auto const &node = tree.tree[tree.info];
                    return input[tree.suffix[tree.info]
                        + (node.parent == -1 ? 0 : tree.tree[node.parent].depth)
                        + commonLength];
                }
            );
            copyTree = donor.number;
            copyNodeExceptParentAndChildren(donor.tree[donor.info], merged[v]);
            unsigned int newNodeIndex = merged.splitEdge(merged[v].parent,
                parentsPlace, commonLength
            );
            unsigned int newCopy = merged.newNode(newNodeIndex);
            merged[newNodeIndex].push_back(newCopy);
            MergeTreesStruct &notDonor = *xorPointers(&donor, &updatedTrees[0],
                &updatedTrees[1]
            );
            copyNodeExceptParentAndChildren(notDonor.tree[notDonor.info],
                merged[newCopy]
            );
            merged[newCopy].indexOfParentEdge += commonLength;
            copySubTree(notDonor.tree, merged, notDonor.info, newCopy);
        } else {
            MergeTreesStruct &donor = minimal(updatedTrees,
                [&copyTree] (MergeTreesStruct &tree) -> pair<bool, bool> {
                    return make_pair(copyTree == 2 || copyTree != tree.number,
                        !(tree.tree[tree.info].leaf != -1)
                    );
                }
            );
            copyNodeExceptParentAndChildren(donor.tree[donor.info], merged[v]);
        }
    }
    if (copyTree != 2 && merged[v].leaf != -1
            && merged[v].leaf % 2 != copyTree) {
        merged[v].leaf = -1;
    }
    for (unsigned int i = 0; i < merged[v].size(); ++i) {
        unsigned int u = merged[v][i];
        correctMerge(u, i, merged, updatedTrees, trueLength, input, copyTree);
    }
}
예제 #4
0
SuffixTree buildSuffixTreeFromSA(vector <unsigned int> &sa,
    vector <unsigned int> &lcp,
    unsigned int length
) {
    vector <int> tmp;
    SuffixTree result = buildTempSuffixTree(tmp);
    int newNodeIndex
        = result.newNode(result.root, sa[0], length - sa[0], sa[0]);
    result[result.root].push_back(newNodeIndex);
    unsigned int current = newNodeIndex;

    for (unsigned int i = 1; i < sa.size(); ++i) {
        while (result[current].parent != -1
            && result[result[current].parent].depth >= lcp[i - 1]
        ) {
            current = result[current].parent;
        }
        unsigned int parent;
        if (result[current].parent != -1 &&
            result[result[current].parent].depth == lcp[i - 1]
        ) {
            parent = result[current].parent;
        } else if (result[current].depth == lcp[i - 1]) {
            parent = current;
        } else {
            unsigned int currentParent = result[current].parent;
            parent = result.splitEdge(currentParent,
                result[currentParent].size() - 1,
                lcp[i - 1] - result[currentParent].depth
            );
            result[parent].leaf = (length - sa[i] == lcp[i - 1] ? sa[i] : -1);
        }
        if (lcp[i - 1] != length - sa[i]) {
            newNodeIndex = result.newNode(parent, sa[i] + lcp[i - 1],
                length - sa[i], sa[i]
            );
            result[parent].push_back(newNodeIndex);
            parent = newNodeIndex;
        }
        current = parent;
    }
    return result;
}
예제 #5
0
void mergeNodes(unsigned int first, unsigned int second, unsigned int to,
    SuffixTree &result, SuffixTree &tree1, SuffixTree &tree2,
    const vector <int> &input
) {
    IndexedPair<MergeNodesStruct> merging(MergeNodesStruct(tree1, first, input),
        MergeNodesStruct(tree2, second, input)
    );
    while (!merging[0].end() && !merging[1].end()) {
        doSomething(merging, [] (MergeNodesStruct &instance) {
            instance.evaluate();
        });
        if (merging[0].symbol == merging[1].symbol) {
            doSomething(merging, [] (MergeNodesStruct &instance) {
                instance.length
                    = instance.tree[instance.child].lengthOfEdge(instance.tree);
            });
            unsigned int minimalLength
                = min(merging[0].length, merging[1].length);
            if (merging[0].length != merging[1].length) {
                MergeNodesStruct &splitVictim = minimal(merging,
                    [] (const MergeNodesStruct &instance) -> int {
                        return -static_cast<int>(instance.length);
                    }
                );

                splitVictim.split(minimalLength);
            }
            unsigned int newNodeIndex = result.newNode(to);
            auto &newNode = result[newNodeIndex];
            result[to].push_back(newNodeIndex);

            if (minimalLength == 1) {
                MergeNodesStruct &recipient = minimal(merging,
                    [] (const MergeNodesStruct &instance)
                        -> pair<unsigned int, int> {
                        return make_pair(instance.length,
                            -instance.tree[instance.child].leaf);
                    }
                );
                copyNodeExceptParentAndChildren(recipient.tree[recipient.child],
                    newNode
                );
            } else {
                newNode.setHiddenInfo(merging[0].child, merging[1].child);
                newNode.depth = minimalLength + result[to].depth;
            }
            mergeNodes(merging[0].child, merging[1].child, newNodeIndex,
                result, tree1, tree2, input);
            doSomething(merging, [] (MergeNodesStruct &instance) {
                ++instance.childIndex;
            });
        } else {
            MergeNodesStruct &recipient = minimal(merging,
                [] (const MergeNodesStruct &instance) -> int {
                    return instance.symbol;
                }
            );
            ++recipient.childIndex;
            appendCopyNode(recipient.tree, result, to, recipient.child);
        }
    }
    MergeNodesStruct &recipient = minimal(merging,
        [] (const MergeNodesStruct & instance) -> bool {
            return instance.end();
        }
    );
    for (; !recipient.end(); ++recipient.childIndex) {
        recipient.evaluate();
        appendCopyNode(recipient.tree, result, to, recipient.child);
    }
}