2015-09-25 19:38:45 +02:00
|
|
|
/*
|
|
|
|
* nghttp2 - HTTP/2 C Library
|
|
|
|
*
|
|
|
|
* Copyright (c) 2015 Tatsuhiro Tsujikawa
|
|
|
|
*
|
|
|
|
* Permission is hereby granted, free of charge, to any person obtaining
|
|
|
|
* a copy of this software and associated documentation files (the
|
|
|
|
* "Software"), to deal in the Software without restriction, including
|
|
|
|
* without limitation the rights to use, copy, modify, merge, publish,
|
|
|
|
* distribute, sublicense, and/or sell copies of the Software, and to
|
|
|
|
* permit persons to whom the Software is furnished to do so, subject to
|
|
|
|
* the following conditions:
|
|
|
|
*
|
|
|
|
* The above copyright notice and this permission notice shall be
|
|
|
|
* included in all copies or substantial portions of the Software.
|
|
|
|
*
|
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
|
|
|
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|
|
|
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
|
|
|
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
|
|
|
|
* LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
|
|
|
|
* OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
|
|
|
|
* WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
|
|
|
*/
|
|
|
|
#include "shrpx_router.h"
|
|
|
|
|
2015-09-26 17:51:32 +02:00
|
|
|
#include <algorithm>
|
|
|
|
|
2015-09-25 19:38:45 +02:00
|
|
|
#include "shrpx_config.h"
|
|
|
|
|
|
|
|
namespace shrpx {
|
|
|
|
|
|
|
|
RNode::RNode() : s(nullptr), len(0), index(-1) {}
|
|
|
|
|
|
|
|
RNode::RNode(const char *s, size_t len, size_t index)
|
|
|
|
: s(s), len(len), index(index) {}
|
|
|
|
|
2016-06-03 16:52:44 +02:00
|
|
|
Router::Router() : balloc_(1024, 1024), root_{} {}
|
|
|
|
|
|
|
|
Router::~Router() {}
|
2015-09-25 19:38:45 +02:00
|
|
|
|
|
|
|
namespace {
|
|
|
|
RNode *find_next_node(const RNode *node, char c) {
|
|
|
|
auto itr = std::lower_bound(std::begin(node->next), std::end(node->next), c,
|
|
|
|
[](const std::unique_ptr<RNode> &lhs,
|
|
|
|
const char c) { return lhs->s[0] < c; });
|
|
|
|
if (itr == std::end(node->next) || (*itr)->s[0] != c) {
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
return (*itr).get();
|
|
|
|
}
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
namespace {
|
|
|
|
void add_next_node(RNode *node, std::unique_ptr<RNode> new_node) {
|
|
|
|
auto itr = std::lower_bound(std::begin(node->next), std::end(node->next),
|
|
|
|
new_node->s[0],
|
|
|
|
[](const std::unique_ptr<RNode> &lhs,
|
|
|
|
const char c) { return lhs->s[0] < c; });
|
|
|
|
node->next.insert(itr, std::move(new_node));
|
|
|
|
}
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
void Router::add_node(RNode *node, const char *pattern, size_t patlen,
|
|
|
|
size_t index) {
|
2016-06-03 16:52:44 +02:00
|
|
|
auto pat = make_string_ref(balloc_, StringRef{pattern, patlen});
|
|
|
|
auto new_node = make_unique<RNode>(pat.c_str(), pat.size(), index);
|
2015-09-25 19:38:45 +02:00
|
|
|
add_next_node(node, std::move(new_node));
|
|
|
|
}
|
|
|
|
|
2017-02-04 10:59:06 +01:00
|
|
|
size_t Router::add_route(const StringRef &pattern, size_t index) {
|
2015-09-25 19:38:45 +02:00
|
|
|
auto node = &root_;
|
|
|
|
size_t i = 0;
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
auto next_node = find_next_node(node, pattern[i]);
|
|
|
|
if (next_node == nullptr) {
|
2016-02-14 10:55:53 +01:00
|
|
|
add_node(node, pattern.c_str() + i, pattern.size() - i, index);
|
2017-02-04 10:59:06 +01:00
|
|
|
return index;
|
2015-09-25 19:38:45 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
node = next_node;
|
|
|
|
|
2016-02-14 10:55:53 +01:00
|
|
|
auto slen = pattern.size() - i;
|
|
|
|
auto s = pattern.c_str() + i;
|
2015-09-25 19:38:45 +02:00
|
|
|
auto n = std::min(node->len, slen);
|
|
|
|
size_t j;
|
|
|
|
for (j = 0; j < n && node->s[j] == s[j]; ++j)
|
|
|
|
;
|
|
|
|
if (j == n) {
|
|
|
|
// The common prefix was matched
|
|
|
|
if (slen == node->len) {
|
|
|
|
// Complete match
|
|
|
|
if (node->index != -1) {
|
2017-02-04 10:59:06 +01:00
|
|
|
// Return the existing index for duplicates.
|
|
|
|
return node->index;
|
2015-09-25 19:38:45 +02:00
|
|
|
}
|
|
|
|
node->index = index;
|
2017-02-04 10:59:06 +01:00
|
|
|
return index;
|
2015-09-25 19:38:45 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (slen > node->len) {
|
|
|
|
// We still have pattern to add
|
|
|
|
i += j;
|
|
|
|
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (node->len > j) {
|
|
|
|
// node must be split into 2 nodes. new_node is now the child
|
|
|
|
// of node.
|
|
|
|
auto new_node =
|
|
|
|
make_unique<RNode>(&node->s[j], node->len - j, node->index);
|
|
|
|
std::swap(node->next, new_node->next);
|
|
|
|
|
|
|
|
node->len = j;
|
|
|
|
node->index = -1;
|
|
|
|
|
|
|
|
add_next_node(node, std::move(new_node));
|
|
|
|
|
|
|
|
if (slen == j) {
|
|
|
|
node->index = index;
|
2017-02-04 10:59:06 +01:00
|
|
|
return index;
|
2015-09-25 19:38:45 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
i += j;
|
|
|
|
|
2016-02-14 10:55:53 +01:00
|
|
|
assert(pattern.size() > i);
|
|
|
|
add_node(node, pattern.c_str() + i, pattern.size() - i, index);
|
2015-09-25 19:38:45 +02:00
|
|
|
|
2017-02-04 10:59:06 +01:00
|
|
|
return index;
|
2015-09-25 19:38:45 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
namespace {
|
|
|
|
const RNode *match_complete(size_t *offset, const RNode *node,
|
|
|
|
const char *first, const char *last) {
|
|
|
|
*offset = 0;
|
|
|
|
|
|
|
|
if (first == last) {
|
|
|
|
return node;
|
|
|
|
}
|
|
|
|
|
|
|
|
auto p = first;
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
auto next_node = find_next_node(node, *p);
|
|
|
|
if (next_node == nullptr) {
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
node = next_node;
|
|
|
|
|
|
|
|
auto n = std::min(node->len, static_cast<size_t>(last - p));
|
|
|
|
if (memcmp(node->s, p, n) != 0) {
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
p += n;
|
|
|
|
if (p == last) {
|
|
|
|
*offset = n;
|
|
|
|
return node;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
namespace {
|
|
|
|
const RNode *match_partial(const RNode *node, size_t offset, const char *first,
|
|
|
|
const char *last) {
|
|
|
|
if (first == last) {
|
|
|
|
if (node->len == offset) {
|
|
|
|
return node;
|
|
|
|
}
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
auto p = first;
|
|
|
|
|
|
|
|
const RNode *found_node = nullptr;
|
|
|
|
|
|
|
|
if (offset > 0) {
|
|
|
|
auto n = std::min(node->len - offset, static_cast<size_t>(last - first));
|
|
|
|
if (memcmp(node->s + offset, first, n) != 0) {
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
p += n;
|
|
|
|
|
|
|
|
if (p == last) {
|
|
|
|
if (node->len == offset + n) {
|
|
|
|
if (node->index != -1) {
|
|
|
|
return node;
|
|
|
|
}
|
|
|
|
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (node->index != -1 && offset + n + 1 == node->len &&
|
|
|
|
node->s[node->len - 1] == '/') {
|
|
|
|
return node;
|
|
|
|
}
|
|
|
|
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (node->index != -1 && node->s[node->len - 1] == '/') {
|
|
|
|
found_node = node;
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(node->len == offset + n);
|
|
|
|
}
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
auto next_node = find_next_node(node, *p);
|
|
|
|
if (next_node == nullptr) {
|
|
|
|
return found_node;
|
|
|
|
}
|
|
|
|
|
|
|
|
node = next_node;
|
|
|
|
|
|
|
|
auto n = std::min(node->len, static_cast<size_t>(last - p));
|
|
|
|
if (memcmp(node->s, p, n) != 0) {
|
|
|
|
return found_node;
|
|
|
|
}
|
|
|
|
|
|
|
|
p += n;
|
|
|
|
|
|
|
|
if (p == last) {
|
|
|
|
if (node->len == n) {
|
|
|
|
// Complete match with this node
|
|
|
|
if (node->index != -1) {
|
|
|
|
return node;
|
|
|
|
}
|
|
|
|
|
|
|
|
return found_node;
|
|
|
|
}
|
|
|
|
|
|
|
|
// We allow match without trailing "/" at the end of pattern.
|
|
|
|
// So, if pattern ends with '/', and pattern and path matches
|
|
|
|
// without that slash, we consider they match to deal with
|
|
|
|
// request to the directory without trailing slash. That is if
|
|
|
|
// pattern is "/foo/" and path is "/foo", we consider they
|
|
|
|
// match.
|
|
|
|
if (node->index != -1 && n + 1 == node->len && node->s[n] == '/') {
|
|
|
|
return node;
|
|
|
|
}
|
|
|
|
|
|
|
|
return found_node;
|
|
|
|
}
|
|
|
|
|
|
|
|
// This is the case when pattern which ends with "/" is included
|
|
|
|
// in query.
|
|
|
|
if (node->index != -1 && node->s[node->len - 1] == '/') {
|
|
|
|
found_node = node;
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(node->len == n);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} // namespace
|
|
|
|
|
2016-02-14 12:48:06 +01:00
|
|
|
ssize_t Router::match(const StringRef &host, const StringRef &path) const {
|
2015-09-25 19:38:45 +02:00
|
|
|
const RNode *node;
|
|
|
|
size_t offset;
|
|
|
|
|
2016-02-14 12:48:06 +01:00
|
|
|
node = match_complete(&offset, &root_, std::begin(host), std::end(host));
|
2015-09-25 19:38:45 +02:00
|
|
|
if (node == nullptr) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2016-02-14 11:07:22 +01:00
|
|
|
node = match_partial(node, offset, std::begin(path), std::end(path));
|
2015-09-25 19:38:45 +02:00
|
|
|
if (node == nullptr || node == &root_) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return node->index;
|
|
|
|
}
|
|
|
|
|
2016-06-24 17:28:15 +02:00
|
|
|
ssize_t Router::match(const StringRef &s) const {
|
|
|
|
const RNode *node;
|
|
|
|
size_t offset;
|
|
|
|
|
|
|
|
node = match_complete(&offset, &root_, std::begin(s), std::end(s));
|
|
|
|
if (node == nullptr) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (node->len != offset) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return node->index;
|
|
|
|
}
|
|
|
|
|
2016-06-10 16:13:40 +02:00
|
|
|
namespace {
|
2016-06-11 06:31:04 +02:00
|
|
|
const RNode *match_prefix(size_t *nread, const RNode *node, const char *first,
|
2016-06-10 16:13:40 +02:00
|
|
|
const char *last) {
|
|
|
|
if (first == last) {
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
auto p = first;
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
auto next_node = find_next_node(node, *p);
|
|
|
|
if (next_node == nullptr) {
|
2016-06-11 06:31:04 +02:00
|
|
|
return nullptr;
|
2016-06-10 16:13:40 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
node = next_node;
|
|
|
|
|
|
|
|
auto n = std::min(node->len, static_cast<size_t>(last - p));
|
|
|
|
if (memcmp(node->s, p, n) != 0) {
|
2016-06-11 06:31:04 +02:00
|
|
|
return nullptr;
|
2016-06-10 16:13:40 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
p += n;
|
|
|
|
|
|
|
|
if (p != last) {
|
|
|
|
if (node->index != -1) {
|
2016-06-11 06:31:04 +02:00
|
|
|
*nread = p - first;
|
|
|
|
return node;
|
2016-06-10 16:13:40 +02:00
|
|
|
}
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (node->len == n) {
|
2016-06-11 06:31:04 +02:00
|
|
|
*nread = p - first;
|
2016-06-10 16:13:40 +02:00
|
|
|
return node;
|
|
|
|
}
|
|
|
|
|
2016-06-11 06:31:04 +02:00
|
|
|
return nullptr;
|
2016-06-10 16:13:40 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
} // namespace
|
|
|
|
|
2016-06-11 06:31:04 +02:00
|
|
|
ssize_t Router::match_prefix(size_t *nread, const RNode **last_node,
|
|
|
|
const StringRef &s) const {
|
|
|
|
if (*last_node == nullptr) {
|
|
|
|
*last_node = &root_;
|
|
|
|
}
|
|
|
|
|
|
|
|
auto node =
|
|
|
|
::shrpx::match_prefix(nread, *last_node, std::begin(s), std::end(s));
|
2016-06-10 16:13:40 +02:00
|
|
|
if (node == nullptr) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2016-06-11 06:31:04 +02:00
|
|
|
*last_node = node;
|
|
|
|
|
2016-06-10 16:13:40 +02:00
|
|
|
return node->index;
|
|
|
|
}
|
|
|
|
|
2015-09-25 19:38:45 +02:00
|
|
|
namespace {
|
|
|
|
void dump_node(const RNode *node, int depth) {
|
|
|
|
fprintf(stderr, "%*ss='%.*s', len=%zu, index=%zd\n", depth, "",
|
|
|
|
(int)node->len, node->s, node->len, node->index);
|
|
|
|
for (auto &nd : node->next) {
|
|
|
|
dump_node(nd.get(), depth + 4);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
void Router::dump() const { dump_node(&root_, 0); }
|
|
|
|
|
|
|
|
} // namespace shrpx
|