introduce safe regex matcher based on re2 engine

api/docs/BUILD

@@ -102,6 +102,7 @@ proto_library(
         "//envoy/type:range",
         "//envoy/type/matcher:metadata",
         "//envoy/type/matcher:number",
+        "//envoy/type/matcher:regex",
         "//envoy/type/matcher:string",
     ],
 )

api/envoy/api/v2/route/BUILD

@@ -10,6 +10,8 @@ api_proto_library_internal(
         "//envoy/api/v2/core:base",
         "//envoy/type:percent",
         "//envoy/type:range",
+        "//envoy/type/matcher:regex",
+        "//envoy/type/matcher:string",
     ],
 )
 
@@ -20,5 +22,7 @@ api_go_proto_library(
         "//envoy/api/v2/core:base_go_proto",
         "//envoy/type:percent_go_proto",
         "//envoy/type:range_go_proto",
+        "//envoy/type/matcher:regex_go_proto",
+        "//envoy/type/matcher:string_go_proto",
     ],
 )

api/envoy/api/v2/route/route.proto

@@ -9,6 +9,8 @@ option go_package = "route";
 option java_generic_services = true;
 
 import "envoy/api/v2/core/base.proto";
+import "envoy/type/matcher/regex.proto";
+import "envoy/type/matcher/string.proto";
 import "envoy/type/percent.proto";
 import "envoy/type/range.proto";
 
@@ -349,7 +351,25 @@ message RouteMatch {
     // * The regex */b[io]t* matches the path */bot*
     // * The regex */b[io]t* does not match the path */bite*
     // * The regex */b[io]t* does not match the path */bit/bot*
-    string regex = 3 [(validate.rules).string.max_bytes = 1024];
+    //
+    // .. attention::
+    //   This field has been deprecated in favor of `safe_regex` as it is not safe for use with
+    //   untrusted input in all cases.
+    string regex = 3 [(validate.rules).string.max_bytes = 1024, deprecated = true];
+
+    // If specified, the route is a regular expression rule meaning that the
+    // regex must match the *:path* header once the query string is removed. The entire path
+    // (without the query string) must match the regex. The rule will not match if only a
+    // subsequence of the *:path* header matches the regex.
+    //
+    // [#next-major-version: In the v3 API we should redo how path specification works such
+    // that we utilize StringMatcher, and additionally have consistent options around whether we
+    // strip query strings, do a case sensitive match, etc. In the interim it will be too disruptive
+    // to deprecate the existing options. We should even consider whether we want to do away with
+    // path_specifier entirely and just rely on a set of header matchers which can already match
+    // on :path, etc. The issue with that is it is unclear how to generically deal with query string
+    // stripping. This needs more thought.]
+    type.matcher.RegexMatcher safe_regex = 10 [(validate.rules).message.required = true];
   }
 
   // Indicates that prefix/path matching should be case insensitive. The default
@@ -404,12 +424,24 @@ message CorsPolicy {
   // Specifies the origins that will be allowed to do CORS requests.
   //
   // An origin is allowed if either allow_origin or allow_origin_regex match.
-  repeated string allow_origin = 1;
+  //
+  // .. attention::
+  //  This field has been deprecated in favor of `allow_origin_string_match`.
+  repeated string allow_origin = 1 [deprecated = true];
 
   // Specifies regex patterns that match allowed origins.
   //
   // An origin is allowed if either allow_origin or allow_origin_regex match.
-  repeated string allow_origin_regex = 8 [(validate.rules).repeated .items.string.max_bytes = 1024];
+  //
+  // .. attention::
+  //   This field has been deprecated in favor of `allow_origin_string_match` as it is not safe for
+  //   use with untrusted input in all cases.
+  repeated string allow_origin_regex = 8
+      [(validate.rules).repeated .items.string.max_bytes = 1024, deprecated = true];
+
+  // Specifies string patterns that match allowed origins. An origin is allowed if any of the
+  // string matchers match.
+  repeated type.matcher.StringMatcher allow_origin_string_match = 11;
 
   // Specifies the content for the *access-control-allow-methods* header.
   string allow_methods = 2;
@@ -1077,18 +1109,28 @@ message VirtualCluster {
   // * The regex */rides/\d+* matches the path */rides/0*
   // * The regex */rides/\d+* matches the path */rides/123*
   // * The regex */rides/\d+* does not match the path */rides/123/456*
-  string pattern = 1 [(validate.rules).string = {min_bytes: 1, max_bytes: 1024}];
+  //
+  // .. attention::
+  //   This field has been deprecated in favor of `headers` as it is not safe for use with
+  //   untrusted input in all cases.
+  string pattern = 1 [(validate.rules).string.max_bytes = 1024, deprecated = true];
+
+  // Specifies a list of header matchers to use for matching requests. Each specified header must
+  // match. The pseudo-headers `:path` and `:method` can be used to match the request path and
+  // method, respectively.
+  repeated HeaderMatcher headers = 4;
 
-  //  Specifies the name of the virtual cluster. The virtual cluster name as well
+  // Specifies the name of the virtual cluster. The virtual cluster name as well
   // as the virtual host name are used when emitting statistics. The statistics are emitted by the
   // router filter and are documented :ref:`here <config_http_filters_router_stats>`.
   string name = 2 [(validate.rules).string.min_bytes = 1];
 
   // Optionally specifies the HTTP method to match on. For example GET, PUT,
   // etc.
-  // [#comment:TODO(htuch): add (validate.rules).enum.defined_only = true once
-  // https://github.com/lyft/protoc-gen-validate/issues/42 is resolved.]
-  core.RequestMethod method = 3;
+  //
+  // .. attention::
+  //   This field has been deprecated in favor of `headers`.
+  core.RequestMethod method = 3 [deprecated = true];
 }
 
 // Global rate limiting :ref:`architecture overview <arch_overview_rate_limit>`.
@@ -1248,6 +1290,7 @@ message RateLimit {
 //   <envoy_api_field_route.HeaderMatcher.name>` header will match, regardless of the header's
 //   value.
 //
+//  [#next-major-version: HeaderMatcher should be refactored to use StringMatcher.]
 message HeaderMatcher {
   // Specifies the name of the header in the request.
   string name = 1 [(validate.rules).string.min_bytes = 1];
@@ -1273,7 +1316,16 @@ message HeaderMatcher {
     // * The regex *\d{3}* matches the value *123*
     // * The regex *\d{3}* does not match the value *1234*
     // * The regex *\d{3}* does not match the value *123.456*
-    string regex_match = 5 [(validate.rules).string.max_bytes = 1024];
+    //
+    // .. attention::
+    //   This field has been deprecated in favor of `safe_regex_match` as it is not safe for use
+    //   with untrusted input in all cases.
+    string regex_match = 5 [(validate.rules).string.max_bytes = 1024, deprecated = true];
+
+    // If specified, this regex string is a regular expression rule which implies the entire request
+    // header value must match the regex. The rule will not match if only a subsequence of the
+    // request header value matches the regex.
+    type.matcher.RegexMatcher safe_regex_match = 11;
 
     // If specified, header match will be performed based on range.
     // The rule will match if the request header value is within this range.
@@ -1328,11 +1380,25 @@ message QueryParameterMatcher {
   // Specifies the value of the key. If the value is absent, a request
   // that contains the key in its query string will match, whether the
   // key appears with a value (e.g., "?debug=true") or not (e.g., "?debug")
-  string value = 3;
+  //
+  // ..attention::
+  //   This field is deprecated. Use an `exact` match inside the `string_match` field.
+  string value = 3 [deprecated = true];
 
   // Specifies whether the query parameter value is a regular expression.
   // Defaults to false. The entire query parameter value (i.e., the part to
   // the right of the equals sign in "key=value") must match the regex.
   // E.g., the regex "\d+$" will match "123" but not "a123" or "123a".
-  google.protobuf.BoolValue regex = 4;
+  //
+  // ..attention::
+  //   This field is deprecated. Use a `safe_regex` match inside the `string_match` field.
+  google.protobuf.BoolValue regex = 4 [deprecated = true];
+
+  oneof query_parameter_match_specifier {
+    // Specifies whether a query parameter value should match against a string.
+    type.matcher.StringMatcher string_match = 5 [(validate.rules).message.required = true];
+
+    // Specifies whether a query parameter should be present.
+    bool present_match = 6;
+  }
 }

api/envoy/type/matcher/BUILD

@@ -40,11 +40,17 @@ api_proto_library_internal(
     name = "string",
     srcs = ["string.proto"],
     visibility = ["//visibility:public"],
+    deps = [
+        ":regex",
+    ],
 )
 
 api_go_proto_library(
     name = "string",
     proto = ":string",
+    deps = [
+        ":regex_go_proto",
+    ],
 )
 
 api_proto_library_internal(
@@ -65,3 +71,14 @@ api_go_proto_library(
         ":string_go_proto",
     ],
 )
+
+api_proto_library_internal(
+    name = "regex",
+    srcs = ["regex.proto"],
+    visibility = ["//visibility:public"],
+)
+
+api_go_proto_library(
+    name = "regex",
+    proto = ":regex",
+)

api/envoy/type/matcher/regex.proto

@@ -0,0 +1,37 @@
+syntax = "proto3";
+
+package envoy.type.matcher;
+
+option java_outer_classname = "StringProto";
+option java_multiple_files = true;
+option java_package = "io.envoyproxy.envoy.type.matcher";
+option go_package = "matcher";
+
+import "google/protobuf/wrappers.proto";
+import "validate/validate.proto";
+
+// [#protodoc-title: RegexMatcher]
+
+// A regex matcher designed for safety when used with untrusted input.
+message RegexMatcher {
+  // Google's `RE2 <https://github.com/google/re2>`_ regex engine. The regex string must adhere to
+  // the documented `syntax <https://github.com/google/re2/wiki/Syntax>`_. The engine is designed
+  // to complete execution in linear time as well as limit the amount of memory used.
+  message GoogleRE2 {
+    // This field controls the RE2 "program size" which is a rough estimate of how complex a
+    // compiled regex is to evaluate. A regex that has a program size greater than the configured
+    // value will fail to compile. In this case, the configured max program size can be increased
+    // or the regex can be simplified. If not specified, the default is 100.
+    google.protobuf.UInt32Value max_program_size = 1;
+  }
+
+  oneof engine_type {
+    option (validate.required) = true;
+
+    // Google's RE2 regex engine.
+    GoogleRE2 google_re2 = 1 [(validate.rules).message.required = true];
+  }
+
+  // The regex match string. The string must be supported by the configured engine.
+  string regex = 2 [(validate.rules).string.min_bytes = 1];
+}

api/envoy/type/matcher/string.proto

@@ -7,6 +7,8 @@ option java_multiple_files = true;
 option java_package = "io.envoyproxy.envoy.type.matcher";
 option go_package = "matcher";
 
+import "envoy/type/matcher/regex.proto";
+
 import "validate/validate.proto";
 
 // [#protodoc-title: StringMatcher]
@@ -48,7 +50,14 @@ message StringMatcher {
     // * The regex *\d{3}* matches the value *123*
     // * The regex *\d{3}* does not match the value *1234*
     // * The regex *\d{3}* does not match the value *123.456*
-    string regex = 4 [(validate.rules).string.max_bytes = 1024];
+    //
+    // .. attention::
+    //   This field has been deprecated in favor of `safe_regex` as it is not safe for use with
+    //   untrusted input in all cases.
+    string regex = 4 [(validate.rules).string.max_bytes = 1024, deprecated = true];
+
+    // The input string must match the regular expression specified here.
+    RegexMatcher safe_regex = 5 [(validate.rules).message.required = true];
   }
 }
 

bazel/repository_locations.bzl

@@ -253,8 +253,8 @@ REPOSITORY_LOCATIONS = dict(
         urls = ["https://github.com/google/cel-cpp/archive/d9d02b20ab85da2444dbdd03410bac6822141364.tar.gz"],
     ),
     com_googlesource_code_re2 = dict(
-        sha256 = "f31db9cd224d018a7e4fe88ef84aaa874b0b3ed91d4d98ee5a1531101d3fdc64",
-        strip_prefix = "re2-87e2ad45e7b18738e1551474f7ee5886ff572059",
-        urls = ["https://github.com/google/re2/archive/87e2ad45e7b18738e1551474f7ee5886ff572059.tar.gz"],
+        sha256 = "38bc0426ee15b5ed67957017fd18201965df0721327be13f60496f2b356e3e01",
+        strip_prefix = "re2-2019-08-01",
+        urls = ["https://github.com/google/re2/archive/2019-08-01.tar.gz"],
     ),
 )

docs/build.sh

@@ -156,6 +156,7 @@ PROTO_RST="
   /envoy/type/matcher/metadata/envoy/type/matcher/metadata.proto.rst
   /envoy/type/matcher/value/envoy/type/matcher/value.proto.rst
   /envoy/type/matcher/number/envoy/type/matcher/number.proto.rst
+  /envoy/type/matcher/regex/envoy/type/matcher/regex.proto.rst
   /envoy/type/matcher/string/envoy/type/matcher/string.proto.rst
 "
 

docs/root/api-v2/types/types.rst

@@ -10,5 +10,6 @@ Types
   ../type/range.proto
   ../type/matcher/metadata.proto
   ../type/matcher/number.proto
+  ../type/matcher/regex.proto
   ../type/matcher/string.proto
   ../type/matcher/value.proto

docs/root/intro/deprecated.rst

@@ -12,8 +12,25 @@ Deprecated items below are listed in chronological order.
 
 Version 1.12.0 (pending)
 ========================
-* The ORIGINAL_DST_LB :ref:`load balancing policy <envoy_api_field_Cluster.lb_policy>` is deprecated, use CLUSTER_PROVIDED policy instead when configuring an :ref:`original destination cluster <envoy_api_field_Cluster.type>`.
-* The :option:`--allow-unknown-fields` command-line option, use :option:`--allow-unknown-static-fields` instead.
+* The ORIGINAL_DST_LB :ref:`load balancing policy <envoy_api_field_Cluster.lb_policy>` is
+  deprecated, use CLUSTER_PROVIDED policy instead when configuring an :ref:`original destination
+  cluster <envoy_api_field_Cluster.type>`.
+* The `regex` field in :ref:`StringMatcher <envoy_api_msg_type.matcher.StringMatcher>` has been
+  deprecated in favor of the `safe_regex` field.
+* The `regex` field in :ref:`RouteMatch <envoy_api_msg_route.RouteMatch>` has been
+  deprecated in favor of the `safe_regex` field.
+* The `allow_origin` and `allow_origin_regex` fields in :ref:`CorsPolicy
+  <envoy_api_msg_route.CorsPolicy>` have been deprecated in favor of the
+  `allow_origin_string_match` field.
+* The `pattern` and `method` fields in :ref:`VirtualCluster <envoy_api_msg_route.VirtualCluster>`
+  have been deprecated in favor of the `headers` field.
+* The `regex_match` field in :ref:`HeaderMatcher <envoy_api_msg_route.HeaderMatcher>` has been
+  deprecated in favor of the `safe_regex_match` field.
+* The `value` and `regex` fields in :ref:`QueryParameterMatcher
+  <envoy_api_msg_route.QueryParameterMatcher>` has been deprecated in favor of the `string_match`
+  and `present_match` fields.
+* The :option:`--allow-unknown-fields` command-line option,
+  use :option:`--allow-unknown-static-fields` instead.
 
 Version 1.11.0 (July 11, 2019)
 ==============================

docs/root/intro/version_history.rst

@@ -7,8 +7,8 @@ Version history
 * admin: added ability to configure listener :ref:`socket options <envoy_api_field_config.bootstrap.v2.Admin.socket_options>`.
 * admin: added config dump support for Secret Discovery Service :ref:`SecretConfigDump <envoy_api_msg_admin.v2alpha.SecretsConfigDump>`.
 * api: added ::ref:`set_node_on_first_message_only <envoy_api_field_core.ApiConfigSource.set_node_on_first_message_only>` option to omit the node identifier from the subsequent discovery requests on the same stream.
-* config: enforcing that terminal filters (e.g. HttpConnectionManager for L4, router for L7) be the last in their respective filter chains.
 * buffer filter: the buffer filter populates content-length header if not present, behavior can be disabled using the runtime feature `envoy.reloadable_features.buffer_filter_populate_content_length`.
+* config: enforcing that terminal filters (e.g. HttpConnectionManager for L4, router for L7) be the last in their respective filter chains.
 * config: added access log :ref:`extension filter<envoy_api_field_config.filter.accesslog.v2.AccessLogFilter.extension_filter>`.
 * config: added support for :option:`--reject-unknown-dynamic-fields`, providing independent control
   over whether unknown fields are rejected in static and dynamic configuration. By default, unknown
@@ -29,11 +29,15 @@ Version history
 * http: added the ability to :ref:`merge adjacent slashes<envoy_api_field_config.filter.network.http_connection_manager.v2.HttpConnectionManager.merge_slashes>` in the path.
 * listeners: added :ref:`continue_on_listener_filters_timeout <envoy_api_field_Listener.continue_on_listener_filters_timeout>` to configure whether a listener will still create a connection when listener filters time out.
 * listeners: added :ref:`HTTP inspector listener filter <config_listener_filters_http_inspector>`.
-* performance: stats symbol table implementation (disabled by default; to test it, add "--use-fake-symbol-table 0" to the command-line arguments when starting Envoy).
-* redis: added :ref:`read_policy <envoy_api_field_config.filter.network.redis_proxy.v2.RedisProxy.ConnPoolSettings.read_policy>` to allow reading from redis replicas for Redis Cluster deployments.
-* rbac: added support for DNS SAN as :ref:`principal_name <envoy_api_field_config.rbac.v2.Principal.Authenticated.principal_name>`.
 * lua: extended `httpCall()` and `respond()` APIs to accept headers with entry values that can be a string or table of strings.
 * performance: new buffer implementation enabled by default (to disable add "--use-libevent-buffers 1" to the command-line arguments when starting Envoy).
+* performance: stats symbol table implementation (disabled by default; to test it, add "--use-fake-symbol-table 0" to the command-line arguments when starting Envoy).
+* rbac: added support for DNS SAN as :ref:`principal_name <envoy_api_field_config.rbac.v2.Principal.Authenticated.principal_name>`.
+* redis: added :ref:`read_policy <envoy_api_field_config.filter.network.redis_proxy.v2.RedisProxy.ConnPoolSettings.read_policy>` to allow reading from redis replicas for Redis Cluster deployments.
+* regex: introduce new :ref:`RegexMatcher <envoy_api_msg_type.matcher.RegexMatcher>` type that
+  provides a safe regex implementation for untrusted user input. This type is now used in all
+  configuration that processes user provided input. See :ref:`deprecated configuration details
+  <deprecated>` for more information.
 * rbac: added conditions to the policy, see :ref:`condition <envoy_api_field_config.rbac.v2.Policy.condition>`.
 * router: added :ref:`rq_retry_skipped_request_not_complete <config_http_filters_router_stats>` counter stat to router stats.
 * router check tool: add coverage reporting & enforcement.

include/envoy/common/BUILD

@@ -29,6 +29,19 @@ envoy_cc_library(
     hdrs = ["time.h"],
 )
 
+envoy_cc_library(
+    name = "matchers_interface",
+    hdrs = ["matchers.h"],
+)
+
+envoy_cc_library(
+    name = "regex_interface",
+    hdrs = ["regex.h"],
+    deps = [
+        ":matchers_interface",
+    ],
+)
+
 envoy_cc_library(
     name = "token_bucket_interface",
     hdrs = ["token_bucket.h"],