Kylin/HttpProxy/UrlRouterPrivate.cpp

#include "UrlRouterPrivate.h"
#include "BoostLog.h"
#include "TemplateSegmentRule.h"
#include <boost/url/decode_view.hpp>
#include <boost/url/rfc/detail/path_rules.hpp>

UrlRouterPrivate::UrlRouterPrivate() {
    m_nodes.push_back(SegementNode{});
}

void UrlRouterPrivate::insertImpl(boost::core::string_view pattern, const std::shared_ptr<AnyResource> &resource) {
    if (pattern.starts_with("/")) pattern.remove_prefix(1);
    auto segements = boost::urls::grammar::parse(pattern, templatePathRule);
    if (!segements) {
        segements.value();
    }
    auto iterator = segements->begin();
    auto end = segements->end();
    auto currentNode = &m_nodes.front();
    int level = 0;
    while (iterator != end) {
        boost::core::string_view segement = (*iterator).string();
        if (segement == ".") {
            ++iterator;
            continue;
        } else if (segement == "..") {
            if (currentNode == &m_nodes.front()) {
                --level;
                ++iterator;
                continue;
            }
            std::size_t parentIndex = currentNode->parentIndex;
            if (currentNode == &m_nodes.back() && !currentNode->resource && currentNode->childIndexes.empty()) {
                auto p = &m_nodes[parentIndex];
                std::size_t currentIndex = currentNode - m_nodes.data();
                p->childIndexes.erase(std::remove(p->childIndexes.begin(), p->childIndexes.end(), currentIndex));
                m_nodes.pop_back();
            }
            currentNode = &m_nodes[parentIndex];
            ++iterator;
            continue;
        }
        if (level < 0) {
            ++level;
            ++iterator;
            continue;
        }
        auto cit = std::find_if(currentNode->childIndexes.begin(), currentNode->childIndexes.end(),
                                [this, &iterator](std::size_t ci) -> bool { return m_nodes[ci].segment == *iterator; });
        if (cit != currentNode->childIndexes.end()) {
            currentNode = &m_nodes[*cit];
        } else {
            SegementNode child;
            child.segment = *iterator;
            std::size_t currentIndex = currentNode - m_nodes.data();
            child.parentIndex = currentIndex;
            m_nodes.push_back(std::move(child));
            m_nodes[currentIndex].childIndexes.push_back(m_nodes.size() - 1);
            if (m_nodes[currentIndex].childIndexes.size() > 1) {
                // keep nodes sorted
                auto &cs = m_nodes[currentIndex].childIndexes;
                std::size_t n = cs.size() - 1;
                while (n) {
                    if (m_nodes[cs.begin()[n]].segment < m_nodes[cs.begin()[n - 1]].segment)
                        std::swap(cs.begin()[n], cs.begin()[n - 1]);
                    else
                        break;
                    --n;
                }
            }
            currentNode = &m_nodes.back();
        }
        ++iterator;
    }
    if (level != 0) {
        boost::urls::detail::throw_invalid_argument();
    }
    currentNode->resource = resource;
    currentNode->templatePath = pattern;
}

const AnyResource *UrlRouterPrivate::findImpl(boost::urls::segments_encoded_view path,
                                              boost::core::string_view *&matches,
                                              boost::core::string_view *&ids) const noexcept {
    if (path.empty()) path = boost::urls::segments_encoded_view("./");
    const SegementNode *p = tryMatch(path.begin(), path.end(), &m_nodes.front(), 0, matches, ids);
    if (p) return p->resource.get();
    return nullptr;
}

const SegementNode *UrlRouterPrivate::tryMatch(boost::urls::segments_encoded_base::const_iterator it,
                                               boost::urls::segments_encoded_base::const_iterator end,
                                               const SegementNode *current, int level,
                                               boost::core::string_view *&matches,
                                               boost::core::string_view *&ids) const {
    while (it != end) {
        boost::urls::pct_string_view s = *it;
        if (*s == ".") {
            // ignore segment
            ++it;
            continue;
        } else if (*s == "..") { // move back to the parent node
            ++it;
            if (level <= 0 && current != &m_nodes.front()) {
                if (!current->segment.isLiteral()) {
                    --matches;
                    --ids;
                }
                current = &m_nodes[current->parentIndex];
            } else
                // there's no parent, so we
                // discount that from the implicit
                // tree beyond terminals
                --level;
            continue;
        }
        if (level < 0) {
            ++level;
            ++it;
            continue;
        }

        bool branch = false;
        if (current->childIndexes.size() > 1) {
            int branches_lb = 0;
            for (auto i : current->childIndexes) {
                auto &c = m_nodes[i];
                if (c.segment.isLiteral() || !c.segment.hasModifier()) {
                    // a literal path counts only
                    // if it matches
                    branches_lb += c.segment.match(s);
                } else {
                    // everything not matching
                    // a single path counts as
                    // more than one path already
                    branches_lb = 2;
                }
                if (branches_lb > 1) {
                    // already know we need to
                    // branch
                    branch = true;
                    break;
                }
            }
        }
        const SegementNode *r = nullptr;
        bool matchAny = false;
        for (auto index : current->childIndexes) {
            auto &child = m_nodes[index];
            if (!child.segment.match(s)) continue;
            if (child.segment.isLiteral()) {
                if (branch) {
                    r = tryMatch(std::next(it), end, &child, level, matches, ids);
                    if (r) break;
                } else {
                    current = &child;
                    matchAny = true;
                    break;
                }
            } else if (!child.segment.hasModifier()) {
                if (branch) {
                    auto matches0 = matches;
                    auto ids0 = ids;
                    *matches++ = *it;
                    *ids++ = child.segment.id();
                    r = tryMatch(std::next(it), end, &child, level, matches, ids);
                    if (r) {
                        break;
                    } else {
                        // rewind
                        matches = matches0;
                        ids = ids0;
                    }
                } else {
                    // only path possible
                    *matches++ = *it;
                    *ids++ = child.segment.id();
                    current = &child;
                    matchAny = true;
                    break;
                }
            } else if (child.segment.isOptional()) {
                auto matches0 = matches;
                auto ids0 = ids;
                *matches++ = *it;
                *ids++ = child.segment.id();
                r = tryMatch(std::next(it), end, &child, level, matches, ids);
                if (r) break;
                matches = matches0;
                ids = ids0;
                *matches++ = {};
                *ids++ = child.segment.id();
                r = tryMatch(it, end, &child, level, matches, ids);
                if (r) break;
                matches = matches0;
                ids = ids0;
            } else {
                auto first = it;
                std::size_t ndotdot = 0;
                std::size_t nnondot = 0;
                auto it1 = it;
                while (it1 != end) {
                    if (*it1 == "..") {
                        ++ndotdot;
                        if (ndotdot >= (nnondot + child.segment.isStar())) break;
                    } else if (*it1 != ".") {
                        ++nnondot;
                    }
                    ++it1;
                }
                if (it1 != end) break;
                auto matches0 = matches;
                auto ids0 = ids;
                *matches++ = *it;
                *ids++ = child.segment.id();
                if (child.segment.isPlus()) {
                    ++first;
                }
                auto start = end;
                while (start != first) {
                    r = tryMatch(start, end, &child, level, matches, ids);
                    if (r) {
                        boost::core::string_view prev = *std::prev(start);
                        *matches0 = {matches0->data(), prev.data() + prev.size()};
                        break;
                    }
                    matches = matches0 + 1;
                    ids = ids0 + 1;
                    --start;
                }
                if (r) {
                    break;
                }
                matches = matches0 + 1;
                ids = ids0 + 1;
                r = tryMatch(start, end, &child, level, matches, ids);
                if (r) {
                    if (!child.segment.isPlus()) *matches0 = {};
                    break;
                }
            }
        }
        if (r) return r;
        if (!matchAny) ++level;
        ++it;
    }
    if (level != 0) { // the path ended below or above an existing node
        return nullptr;
    }
    if (!current->resource) {
        return findOptionalResource(current, m_nodes, matches, ids);
    }
    return current;
}

const SegementNode *UrlRouterPrivate::findOptionalResource(const SegementNode *root,
                                                           const std::vector<SegementNode> &ns,
                                                           boost::core::string_view *&matches,
                                                           boost::core::string_view *&ids) {
    BOOST_ASSERT(root);
    if (root->resource) return root;
    BOOST_ASSERT(!root->childIndexes.empty());
    for (auto index : root->childIndexes) {
        auto &child = ns[index];
        if (!child.segment.isOptional() && !child.segment.isStar()) continue;
        // Child nodes are also potentially optional.
        auto matches0 = matches;
        auto ids0 = ids;
        *matches++ = {};
        *ids++ = child.segment.id();
        auto n = findOptionalResource(&child, ns, matches, ids);
        if (n) return n;
        matches = matches0;
        ids = ids0;
    }
    return nullptr;
}

bool ChildIndexVector::empty() const {
    return m_size == 0;
}

size_t ChildIndexVector::size() const {
    return m_size;
}

size_t *ChildIndexVector::begin() {
    if (m_childIndexes) return m_childIndexes;
    return m_staticChildIndexes;
}

const size_t *ChildIndexVector::begin() const {
    return const_cast<ChildIndexVector *>(this)->begin();
}

size_t *ChildIndexVector::end() {
    return begin() + m_size;
}

const size_t *ChildIndexVector::end() const {
    return const_cast<ChildIndexVector *>(this)->end();
}

void ChildIndexVector::erase(size_t *it) {
    BOOST_ASSERT(it - begin() >= 0);
    std::memmove(it - 1, it, end() - it);
    --m_size;
}

void ChildIndexVector::push_back(size_t v) {
    if (m_size == N && !m_childIndexes) {
        m_childIndexes = new std::size_t[N * 2];
        m_capcity = N * 2;
        std::memcpy(m_childIndexes, m_staticChildIndexes, N * sizeof(std::size_t));
    } else if (m_childIndexes && m_size == m_capcity) {
        auto *tmp = new std::size_t[m_capcity * 2];
        std::memcpy(tmp, m_childIndexes, m_capcity * sizeof(std::size_t));
        delete[] m_childIndexes;
        m_childIndexes = tmp;
        m_capcity = m_capcity * 2;
    }
    begin()[m_size++] = v;
}