nghttpx: Rewrite wildcard router

src/shrpx_client_handler.cc

@@ -910,20 +910,22 @@ ClientHandler::get_downstream_connection(Downstream *downstream) {
     group_idx = catch_all;
   } else {
     auto &router = downstreamconf.router;
+    auto &rev_wildcard_router = downstreamconf.rev_wildcard_router;
     auto &wildcard_patterns = downstreamconf.wildcard_patterns;
     if (!req.authority.empty()) {
-      group_idx =
-          match_downstream_addr_group(router, wildcard_patterns, req.authority,
-                                      req.path, groups, catch_all);
+      group_idx = match_downstream_addr_group(router, rev_wildcard_router,
+                                              wildcard_patterns, req.authority,
+                                              req.path, groups, catch_all);
     } else {
       auto h = req.fs.header(http2::HD_HOST);
       if (h) {
-        group_idx = match_downstream_addr_group(
-            router, wildcard_patterns, h->value, req.path, groups, catch_all);
+        group_idx = match_downstream_addr_group(router, rev_wildcard_router,
+                                                wildcard_patterns, h->value,
+                                                req.path, groups, catch_all);
       } else {
-        group_idx =
-            match_downstream_addr_group(router, wildcard_patterns, StringRef{},
-                                        req.path, groups, catch_all);
+        group_idx = match_downstream_addr_group(router, rev_wildcard_router,
+                                                wildcard_patterns, StringRef{},
+                                                req.path, groups, catch_all);
       }
     }
   }

src/shrpx_config.cc

@@ -828,6 +828,12 @@ int parse_mapping(Config *config, DownstreamAddrConfig addr,
 
         auto &router = wildcard_patterns.back().router;
         router.add_route(path, idx);
+
+        auto rev_host = host.str();
+        std::reverse(std::begin(rev_host), std::end(rev_host));
+
+        downstreamconf.rev_wildcard_router.add_route(
+            StringRef{rev_host}, wildcard_patterns.size() - 1);
       } else {
         (*it).router.add_route(path, idx);
       }
@@ -2831,21 +2837,6 @@ int configure_downstream_group(Config *config, bool http2_proxy,
     router = Router();
     router.add_route(StringRef{catch_all.pattern}, addr_groups.size());
     addr_groups.push_back(std::move(catch_all));
-  } else {
-    auto &wildcard_patterns = downstreamconf.wildcard_patterns;
-    std::sort(std::begin(wildcard_patterns), std::end(wildcard_patterns),
-              [](const WildcardPattern &lhs, const WildcardPattern &rhs) {
-                return std::lexicographical_compare(
-                    rhs.host.rbegin(), rhs.host.rend(), lhs.host.rbegin(),
-                    lhs.host.rend());
-              });
-    if (LOG_ENABLED(INFO)) {
-      LOG(INFO) << "Reverse sorted wildcard hosts (compared from tail to head, "
-                   "and sorted in reverse order):";
-      for (auto &wp : wildcard_patterns) {
-        LOG(INFO) << wp.host;
-      }
-    }
   }
 
   // backward compatibility: override all SNI fields with the option

src/shrpx_config.h

@@ -599,7 +599,7 @@ struct RateLimitConfig {
 };
 
 // Wildcard host pattern routing.  We strips left most '*' from host
-// field.  router includes all path pattern sharing same wildcard
+// field.  router includes all path patterns sharing the same wildcard
 // host.
 struct WildcardPattern {
   WildcardPattern(const StringRef &host)
@@ -616,6 +616,12 @@ struct DownstreamConfig {
     ev_tstamp idle_read;
   } timeout;
   Router router;
+  // Router for reversed wildcard hosts.  Since this router has
+  // wildcard hosts reversed without '*', one should call match()
+  // function with reversed host stripping last character.  This is
+  // because we require at least one character must match for '*'.
+  // The index stored in this router is index of wildcard_patterns.
+  Router rev_wildcard_router;
   std::vector<WildcardPattern> wildcard_patterns;
   std::vector<DownstreamAddrGroupConfig> addr_groups;
   // The index of catch-all group in downstream_addr_groups.

src/shrpx_router.cc

@@ -279,6 +279,57 @@ ssize_t Router::match(const StringRef &host, const StringRef &path) const {
   return node->index;
 }
 
+namespace {
+const RNode *match_prefix(const RNode *node, const char *first,
+                          const char *last) {
+  if (first == last) {
+    return nullptr;
+  }
+
+  auto p = first;
+
+  const RNode *ans = nullptr;
+
+  for (;;) {
+    auto next_node = find_next_node(node, *p);
+    if (next_node == nullptr) {
+      return ans;
+    }
+
+    node = next_node;
+
+    auto n = std::min(node->len, static_cast<size_t>(last - p));
+    if (memcmp(node->s, p, n) != 0) {
+      return ans;
+    }
+
+    p += n;
+
+    if (p != last) {
+      if (node->index != -1) {
+        ans = node;
+      }
+      continue;
+    }
+
+    if (node->len == n) {
+      return node;
+    }
+
+    return ans;
+  }
+}
+} // namespace
+
+ssize_t Router::match_prefix(const StringRef &s) const {
+  auto node = ::shrpx::match_prefix(&root_, std::begin(s), std::end(s));
+  if (node == nullptr) {
+    return -1;
+  }
+
+  return node->index;
+}
+
 namespace {
 void dump_node(const RNode *node, int depth) {
   fprintf(stderr, "%*ss='%.*s', len=%zu, index=%zd\n", depth, "",

src/shrpx_router.h

@@ -67,6 +67,9 @@ public:
   bool add_route(const StringRef &pattern, size_t index);
   // Returns the matched index of pattern.  -1 if there is no match.
   ssize_t match(const StringRef &host, const StringRef &path) const;
+  // Returns the matched index of pattern if a pattern is a suffix of
+  // |s|, otherwise -1.
+  ssize_t match_prefix(const StringRef &s) const;
 
   void add_node(RNode *node, const char *pattern, size_t patlen, size_t index);
 

src/shrpx_worker.cc

@@ -456,7 +456,8 @@ ConnectionHandler *Worker::get_connection_handler() const {
 
 namespace {
 size_t match_downstream_addr_group_host(
-    const Router &router, const std::vector<WildcardPattern> &wildcard_patterns,
+    const Router &router, const Router &rev_wildcard_router,
+    const std::vector<WildcardPattern> &wildcard_patterns,
     const StringRef &host, const StringRef &path,
     const std::vector<std::shared_ptr<DownstreamAddrGroup>> &groups,
     size_t catch_all) {
@@ -486,23 +487,24 @@ size_t match_downstream_addr_group_host(
     return group;
   }
 
-  for (auto it = std::begin(wildcard_patterns);
-       it != std::end(wildcard_patterns); ++it) {
-    /* left most '*' must match at least one character */
-    if (host.size() <= (*it).host.size() ||
-        !util::ends_with(std::begin(host), std::end(host),
-                         std::begin((*it).host), std::end((*it).host))) {
-      continue;
-    }
-    auto group = (*it).router.match(StringRef{}, path);
-    if (group != -1) {
-      // We sorted wildcard_patterns in a way that first match is the
-      // longest host pattern.
-      if (LOG_ENABLED(INFO)) {
-        LOG(INFO) << "Found wildcard pattern with query " << host << path
-                  << ", matched pattern=" << groups[group]->pattern;
+  if (!wildcard_patterns.empty() && !host.empty()) {
+    auto rev_host = std::string{std::begin(host) + 1, std::end(host)};
+    std::reverse(std::begin(rev_host), std::end(rev_host));
+
+    auto wcidx = rev_wildcard_router.match_prefix(
+        StringRef{std::begin(rev_host), std::end(rev_host)});
+    if (wcidx != -1) {
+      auto &wc = wildcard_patterns[wcidx];
+      auto group = wc.router.match(StringRef{}, path);
+      if (group != -1) {
+        // We sorted wildcard_patterns in a way that first match is the
+        // longest host pattern.
+        if (LOG_ENABLED(INFO)) {
+          LOG(INFO) << "Found wildcard pattern with query " << host << path
+                    << ", matched pattern=" << groups[group]->pattern;
+        }
+        return group;
       }
-      return group;
     }
   }
 
@@ -523,7 +525,8 @@ size_t match_downstream_addr_group_host(
 } // namespace
 
 size_t match_downstream_addr_group(
-    const Router &router, const std::vector<WildcardPattern> &wildcard_patterns,
+    const Router &router, const Router &rev_wildcard_router,
+    const std::vector<WildcardPattern> &wildcard_patterns,
     const StringRef &hostport, const StringRef &raw_path,
     const std::vector<std::shared_ptr<DownstreamAddrGroup>> &groups,
     size_t catch_all) {
@@ -539,8 +542,9 @@ size_t match_downstream_addr_group(
   auto path = StringRef{std::begin(raw_path), query};
 
   if (hostport.empty()) {
-    return match_downstream_addr_group_host(router, wildcard_patterns, hostport,
-                                            path, groups, catch_all);
+    return match_downstream_addr_group_host(router, rev_wildcard_router,
+                                            wildcard_patterns, hostport, path,
+                                            groups, catch_all);
   }
 
   StringRef host;
@@ -570,8 +574,9 @@ size_t match_downstream_addr_group(
     util::inp_strlower(low_host);
     host = StringRef{low_host};
   }
-  return match_downstream_addr_group_host(router, wildcard_patterns, host, path,
-                                          groups, catch_all);
+  return match_downstream_addr_group_host(router, rev_wildcard_router,
+                                          wildcard_patterns, host, path, groups,
+                                          catch_all);
 }
 
 void downstream_failure(DownstreamAddr *addr) {

src/shrpx_worker.h

@@ -280,7 +280,8 @@ private:
 // group.  The catch-all group index is given in |catch_all|.  All
 // patterns are given in |groups|.
 size_t match_downstream_addr_group(
-    const Router &router, const std::vector<WildcardPattern> &wildcard_patterns,
+    const Router &router, const Router &rev_wildcard_router,
+    const std::vector<WildcardPattern> &wildcard_patterns,
     const StringRef &hostport, const StringRef &path,
     const std::vector<std::shared_ptr<DownstreamAddrGroup>> &groups,
     size_t catch_all);

src/shrpx_worker_test.cc

@@ -50,6 +50,7 @@ void test_shrpx_worker_match_downstream_addr_group(void) {
   }
 
   Router router;
+  Router wcrouter;
 
   for (size_t i = 0; i < groups.size(); ++i) {
     auto &g = groups[i];
@@ -59,122 +60,124 @@ void test_shrpx_worker_match_downstream_addr_group(void) {
   std::vector<WildcardPattern> wp;
 
   CU_ASSERT(0 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/"), groups, 255));
 
   // port is removed
-  CU_ASSERT(0 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org:8080"),
-                     StringRef::from_lit("/"), groups, 255));
+  CU_ASSERT(0 ==
+            match_downstream_addr_group(router, wcrouter, wp,
+                                        StringRef::from_lit("nghttp2.org:8080"),
+                                        StringRef::from_lit("/"), groups, 255));
 
   // host is case-insensitive
   CU_ASSERT(4 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("WWW.nghttp2.org"),
+                     router, wcrouter, wp,
+                     StringRef::from_lit("WWW.nghttp2.org"),
                      StringRef::from_lit("/alpha"), groups, 255));
 
   CU_ASSERT(1 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/alpha/bravo/"), groups, 255));
 
   // /alpha/bravo also matches /alpha/bravo/
   CU_ASSERT(1 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/alpha/bravo"), groups, 255));
 
   // path part is case-sensitive
   CU_ASSERT(0 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/Alpha/bravo"), groups, 255));
 
   CU_ASSERT(1 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/alpha/bravo/charlie"), groups, 255));
 
   CU_ASSERT(2 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/alpha/charlie"), groups, 255));
 
   // pattern which does not end with '/' must match its entirely.  So
   // this matches to group 0, not group 2.
   CU_ASSERT(0 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/alpha/charlie/"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit("example.org"),
+                       router, wcrouter, wp, StringRef::from_lit("example.org"),
                        StringRef::from_lit("/"), groups, 255));
 
-  CU_ASSERT(255 ==
-            match_downstream_addr_group(router, wp, StringRef::from_lit(""),
-                                        StringRef::from_lit("/"), groups, 255));
+  CU_ASSERT(255 == match_downstream_addr_group(
+                       router, wcrouter, wp, StringRef::from_lit(""),
+                       StringRef::from_lit("/"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit(""),
+                       router, wcrouter, wp, StringRef::from_lit(""),
                        StringRef::from_lit("alpha"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit("foo/bar"),
+                       router, wcrouter, wp, StringRef::from_lit("foo/bar"),
                        StringRef::from_lit("/"), groups, 255));
 
   // If path is StringRef::from_lit("*", only match with host + "/").
   CU_ASSERT(0 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("*"), groups, 255));
 
-  CU_ASSERT(
-      5 == match_downstream_addr_group(router, wp, StringRef::from_lit("[::1]"),
-                                       StringRef::from_lit("/"), groups, 255));
   CU_ASSERT(5 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("[::1]:8080"),
+                     router, wcrouter, wp, StringRef::from_lit("[::1]"),
+                     StringRef::from_lit("/"), groups, 255));
+  CU_ASSERT(5 == match_downstream_addr_group(
+                     router, wcrouter, wp, StringRef::from_lit("[::1]:8080"),
                      StringRef::from_lit("/"), groups, 255));
-  CU_ASSERT(255 ==
-            match_downstream_addr_group(router, wp, StringRef::from_lit("[::1"),
-                                        StringRef::from_lit("/"), groups, 255));
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit("[::1]8000"),
+                       router, wcrouter, wp, StringRef::from_lit("[::1"),
+                       StringRef::from_lit("/"), groups, 255));
+  CU_ASSERT(255 == match_downstream_addr_group(
+                       router, wcrouter, wp, StringRef::from_lit("[::1]8000"),
                        StringRef::from_lit("/"), groups, 255));
 
   // Check the case where adding route extends tree
   CU_ASSERT(6 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/alpha/bravo/delta"), groups, 255));
 
   CU_ASSERT(1 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/alpha/bravo/delta/"), groups, 255));
 
   // Check the case where query is done in a single node
   CU_ASSERT(7 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("example.com"),
+                     router, wcrouter, wp, StringRef::from_lit("example.com"),
                      StringRef::from_lit("/alpha/bravo"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit("example.com"),
+                       router, wcrouter, wp, StringRef::from_lit("example.com"),
                        StringRef::from_lit("/alpha/bravo/"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit("example.com"),
+                       router, wcrouter, wp, StringRef::from_lit("example.com"),
                        StringRef::from_lit("/alpha"), groups, 255));
 
   // Check the case where quey is done in a single node
   CU_ASSERT(8 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("192.168.0.1"),
+                     router, wcrouter, wp, StringRef::from_lit("192.168.0.1"),
                      StringRef::from_lit("/alpha"), groups, 255));
 
   CU_ASSERT(8 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("192.168.0.1"),
+                     router, wcrouter, wp, StringRef::from_lit("192.168.0.1"),
                      StringRef::from_lit("/alpha/"), groups, 255));
 
   CU_ASSERT(8 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("192.168.0.1"),
+                     router, wcrouter, wp, StringRef::from_lit("192.168.0.1"),
                      StringRef::from_lit("/alpha/bravo"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit("192.168.0.1"),
+                       router, wcrouter, wp, StringRef::from_lit("192.168.0.1"),
                        StringRef::from_lit("/alph"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit("192.168.0.1"),
+                       router, wcrouter, wp, StringRef::from_lit("192.168.0.1"),
                        StringRef::from_lit("/"), groups, 255));
 
   // Test for wildcard hosts
@@ -187,34 +190,41 @@ void test_shrpx_worker_match_downstream_addr_group(void) {
   groups.push_back(std::move(g2));
 
   wp.emplace_back(StringRef::from_lit("git.nghttp2.org"));
+  wcrouter.add_route(StringRef::from_lit("gro.2ptthgn.tig"), 0);
   wp.back().router.add_route(StringRef::from_lit("/echo/"), 10);
 
   wp.emplace_back(StringRef::from_lit(".nghttp2.org"));
+  wcrouter.add_route(StringRef::from_lit("gro.2ptthgn."), 1);
   wp.back().router.add_route(StringRef::from_lit("/echo/"), 11);
   wp.back().router.add_route(StringRef::from_lit("/echo/foxtrot"), 12);
 
   CU_ASSERT(11 == match_downstream_addr_group(
-                      router, wp, StringRef::from_lit("git.nghttp2.org"),
+                      router, wcrouter, wp,
+                      StringRef::from_lit("git.nghttp2.org"),
                       StringRef::from_lit("/echo"), groups, 255));
 
   CU_ASSERT(10 == match_downstream_addr_group(
-                      router, wp, StringRef::from_lit("0git.nghttp2.org"),
+                      router, wcrouter, wp,
+                      StringRef::from_lit("0git.nghttp2.org"),
                       StringRef::from_lit("/echo"), groups, 255));
 
   CU_ASSERT(11 == match_downstream_addr_group(
-                      router, wp, StringRef::from_lit("it.nghttp2.org"),
+                      router, wcrouter, wp,
+                      StringRef::from_lit("it.nghttp2.org"),
                       StringRef::from_lit("/echo"), groups, 255));
 
   CU_ASSERT(255 == match_downstream_addr_group(
-                       router, wp, StringRef::from_lit(".nghttp2.org"),
+                       router, wcrouter, wp,
+                       StringRef::from_lit(".nghttp2.org"),
                        StringRef::from_lit("/echo/foxtrot"), groups, 255));
 
   CU_ASSERT(9 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("alpha.nghttp2.org"),
+                     router, wcrouter, wp,
+                     StringRef::from_lit("alpha.nghttp2.org"),
                      StringRef::from_lit("/golf"), groups, 255));
 
   CU_ASSERT(0 == match_downstream_addr_group(
-                     router, wp, StringRef::from_lit("nghttp2.org"),
+                     router, wcrouter, wp, StringRef::from_lit("nghttp2.org"),
                      StringRef::from_lit("/echo"), groups, 255));
 }