feat: endpoint renaming (#260)

* Fix bin/format to work on mac os

For reasons I didn't explore further, /mnt/host/Users is shown in the
directory listing of /mnt/host, but can't be traversed

* Implement endpoint guessing

Author: cataphract

BUILD.bazel

@@ -17,6 +17,7 @@ cc_library(
       "src/datadog/datadog_agent_config.cpp",
       "src/datadog/datadog_agent.cpp",
       "src/datadog/default_http_client_null.cpp",
+      "src/datadog/endpoint_inferral.cpp",
       "src/datadog/environment.cpp",
       "src/datadog/error.cpp",
       "src/datadog/extraction_util.cpp",
@@ -60,6 +61,7 @@ cc_library(
       "src/datadog/datadog_agent.h",
       "src/datadog/default_http_client.h",
       "src/datadog/extracted_data.h",
+      "src/datadog/endpoint_inferral.h",
       "src/datadog/extraction_util.h",
       "src/datadog/glob.h",
       "src/datadog/hex.h",
@@ -97,6 +99,7 @@ cc_library(
       "include/datadog/event_scheduler.h",
       "include/datadog/expected.h",
       "include/datadog/http_client.h",
+      "include/datadog/http_endpoint_calculation_mode.h",
       "include/datadog/id_generator.h",
       "include/datadog/injection_options.h",
       "include/datadog/logger.h",

CMakeLists.txt

@@ -173,6 +173,7 @@ target_sources(dd-trace-cpp-objects
     src/datadog/collector_response.cpp
     src/datadog/datadog_agent_config.cpp
     src/datadog/datadog_agent.cpp
+    src/datadog/endpoint_inferral.cpp
     src/datadog/environment.cpp
     src/datadog/error.cpp
     src/datadog/extraction_util.cpp

bin/format

@@ -2,8 +2,12 @@
 
 set -e
 
+# Determine the script directory and repository root
+SCRIPT_DIR=$(cd "$(dirname "$0")" && pwd)
+REPO_ROOT=$(cd "$SCRIPT_DIR/.." && pwd)
+
 # Go to the repository root directory.
-cd "$(dirname "$0")"/..
+cd "$REPO_ROOT"
 
 # clang-format's behavior changes between versions, even given the same
 # configuration.
@@ -41,16 +45,16 @@ if [ "$(docker image ls --quiet $image | wc -l)" -eq 0 ]; then
     esac
 fi
 
-mount_path=/mnt/host
+mount_path=/mnt/repo
 
-# File paths passed to the dockerized clang-format need to be resolved and
-# prefixed with the bind mount path.
+# File paths passed to the dockerized clang-format need to be relative to
+# the repository root and prefixed with the bind mount path.
 #
 # Non-path arguments (flags) are left alone.
 process_arg() {
   case "$1" in
     -*) printf '%s\0' "$1" ;;
-    *) printf '%s/%s\0' "$mount_path" "$(realpath "$1")" ;;
+    *) printf '%s/%s\0' "$mount_path" "$1" ;;
   esac
 }
 
@@ -65,7 +69,7 @@ docker_clang_format() {
       --volume /etc/passwd:/etc/passwd:ro \
       --volume /etc/group:/etc/group:ro \
       --user "$(id -u):$(id -g)" \
-      --mount "type=bind,source=/,destination=$mount_path" \
+      --mount "type=bind,source=$REPO_ROOT,destination=$mount_path" \
       "$image" \
       clang-format $formatter_options
 }

include/datadog/config.h

@@ -28,6 +28,8 @@ enum class ConfigName : char {
   TRACE_BAGGAGE_MAX_BYTES,
   TRACE_BAGGAGE_MAX_ITEMS,
   APM_TRACING_ENABLED,
+  TRACE_RESOURCE_RENAMING_ENABLED,
+  TRACE_RESOURCE_RENAMING_ALWAYS_SIMPLIFIED_ENDPOINT,
 };
 
 // Represents metadata for configuration parameters

include/datadog/environment.h

@@ -23,44 +23,46 @@ namespace environment {
 
 // To enforce correspondence between `enum Variable` and `variable_names`, the
 // preprocessor is used so that the DD_* symbols are listed exactly once.
-#define LIST_ENVIRONMENT_VARIABLES(MACRO)            \
-  MACRO(DD_AGENT_HOST)                               \
-  MACRO(DD_ENV)                                      \
-  MACRO(DD_INSTRUMENTATION_TELEMETRY_ENABLED)        \
-  MACRO(DD_PROPAGATION_STYLE_EXTRACT)                \
-  MACRO(DD_PROPAGATION_STYLE_INJECT)                 \
-  MACRO(DD_REMOTE_CONFIGURATION_ENABLED)             \
-  MACRO(DD_REMOTE_CONFIG_POLL_INTERVAL_SECONDS)      \
-  MACRO(DD_SERVICE)                                  \
-  MACRO(DD_SPAN_SAMPLING_RULES)                      \
-  MACRO(DD_SPAN_SAMPLING_RULES_FILE)                 \
-  MACRO(DD_TRACE_PROPAGATION_STYLE_EXTRACT)          \
-  MACRO(DD_TRACE_PROPAGATION_STYLE_INJECT)           \
-  MACRO(DD_TRACE_PROPAGATION_STYLE)                  \
-  MACRO(DD_TAGS)                                     \
-  MACRO(DD_TRACE_AGENT_PORT)                         \
-  MACRO(DD_TRACE_AGENT_URL)                          \
-  MACRO(DD_TRACE_DEBUG)                              \
-  MACRO(DD_TRACE_ENABLED)                            \
-  MACRO(DD_TRACE_RATE_LIMIT)                         \
-  MACRO(DD_TRACE_REPORT_HOSTNAME)                    \
-  MACRO(DD_TRACE_SAMPLE_RATE)                        \
-  MACRO(DD_TRACE_SAMPLING_RULES)                     \
-  MACRO(DD_TRACE_STARTUP_LOGS)                       \
-  MACRO(DD_TRACE_TAGS_PROPAGATION_MAX_LENGTH)        \
-  MACRO(DD_VERSION)                                  \
-  MACRO(DD_TRACE_128_BIT_TRACEID_GENERATION_ENABLED) \
-  MACRO(DD_TELEMETRY_HEARTBEAT_INTERVAL)             \
-  MACRO(DD_TELEMETRY_METRICS_ENABLED)                \
-  MACRO(DD_TELEMETRY_METRICS_INTERVAL_SECONDS)       \
-  MACRO(DD_TELEMETRY_DEBUG)                          \
-  MACRO(DD_TRACE_BAGGAGE_MAX_ITEMS)                  \
-  MACRO(DD_TRACE_BAGGAGE_MAX_BYTES)                  \
-  MACRO(DD_TELEMETRY_LOG_COLLECTION_ENABLED)         \
-  MACRO(DD_INSTRUMENTATION_INSTALL_ID)               \
-  MACRO(DD_INSTRUMENTATION_INSTALL_TYPE)             \
-  MACRO(DD_INSTRUMENTATION_INSTALL_TIME)             \
-  MACRO(DD_APM_TRACING_ENABLED)                      \
+#define LIST_ENVIRONMENT_VARIABLES(MACRO)                      \
+  MACRO(DD_AGENT_HOST)                                         \
+  MACRO(DD_ENV)                                                \
+  MACRO(DD_INSTRUMENTATION_TELEMETRY_ENABLED)                  \
+  MACRO(DD_PROPAGATION_STYLE_EXTRACT)                          \
+  MACRO(DD_PROPAGATION_STYLE_INJECT)                           \
+  MACRO(DD_REMOTE_CONFIGURATION_ENABLED)                       \
+  MACRO(DD_REMOTE_CONFIG_POLL_INTERVAL_SECONDS)                \
+  MACRO(DD_SERVICE)                                            \
+  MACRO(DD_SPAN_SAMPLING_RULES)                                \
+  MACRO(DD_SPAN_SAMPLING_RULES_FILE)                           \
+  MACRO(DD_TRACE_PROPAGATION_STYLE_EXTRACT)                    \
+  MACRO(DD_TRACE_PROPAGATION_STYLE_INJECT)                     \
+  MACRO(DD_TRACE_PROPAGATION_STYLE)                            \
+  MACRO(DD_TAGS)                                               \
+  MACRO(DD_TRACE_AGENT_PORT)                                   \
+  MACRO(DD_TRACE_AGENT_URL)                                    \
+  MACRO(DD_TRACE_DEBUG)                                        \
+  MACRO(DD_TRACE_ENABLED)                                      \
+  MACRO(DD_TRACE_RATE_LIMIT)                                   \
+  MACRO(DD_TRACE_REPORT_HOSTNAME)                              \
+  MACRO(DD_TRACE_SAMPLE_RATE)                                  \
+  MACRO(DD_TRACE_SAMPLING_RULES)                               \
+  MACRO(DD_TRACE_STARTUP_LOGS)                                 \
+  MACRO(DD_TRACE_TAGS_PROPAGATION_MAX_LENGTH)                  \
+  MACRO(DD_VERSION)                                            \
+  MACRO(DD_TRACE_128_BIT_TRACEID_GENERATION_ENABLED)           \
+  MACRO(DD_TELEMETRY_HEARTBEAT_INTERVAL)                       \
+  MACRO(DD_TELEMETRY_METRICS_ENABLED)                          \
+  MACRO(DD_TELEMETRY_METRICS_INTERVAL_SECONDS)                 \
+  MACRO(DD_TELEMETRY_DEBUG)                                    \
+  MACRO(DD_TRACE_BAGGAGE_MAX_ITEMS)                            \
+  MACRO(DD_TRACE_BAGGAGE_MAX_BYTES)                            \
+  MACRO(DD_TELEMETRY_LOG_COLLECTION_ENABLED)                   \
+  MACRO(DD_INSTRUMENTATION_INSTALL_ID)                         \
+  MACRO(DD_INSTRUMENTATION_INSTALL_TYPE)                       \
+  MACRO(DD_INSTRUMENTATION_INSTALL_TIME)                       \
+  MACRO(DD_APM_TRACING_ENABLED)                                \
+  MACRO(DD_TRACE_RESOURCE_RENAMING_ENABLED)                    \
+  MACRO(DD_TRACE_RESOURCE_RENAMING_ALWAYS_SIMPLIFIED_ENDPOINT) \
   MACRO(DD_EXTERNAL_ENV)
 
 #define WITH_COMMA(ARG) ARG,

include/datadog/http_client.h

@@ -26,6 +26,7 @@ class HTTPClient {
     std::string scheme;     // http, https, or unix
     std::string authority;  // domain:port or /path/to/socket
     std::string path;       // resource, e.g. /v0.4/traces
+    std::string query;      // query string without '?'
 
     static Expected<HTTPClient::URL> parse(StringView input);
   };

include/datadog/http_endpoint_calculation_mode.h

@@ -0,0 +1,34 @@
+#pragma once
+
+// This component provides an enumeration that controls how the http.endpoint
+// tag is calculated for HTTP spans.
+
+#include <cstdint>
+
+namespace datadog {
+namespace tracing {
+
+// `HttpEndpointCalculationMode` determines when and how the http.endpoint tag
+// is inferred from http.url for HTTP spans.
+//
+// The http.endpoint tag provides a normalized, parameterized version of the
+// HTTP path (e.g., "/users/{param:int}" instead of "/users/123"). This helps
+// aggregate similar requests and reduce cardinality in monitoring systems.
+enum class HttpEndpointCalculationMode : std::uint8_t {
+  // Do not calculate http.endpoint. The tag will not be set unless explicitly
+  // provided by the user.
+  DISABLED,
+
+  // Calculate http.endpoint from http.url only when http.route is not present.
+  // This mode acts as a fallback - if instrumentation provides http.route,
+  // use that; otherwise, infer http.endpoint from the URL path.
+  FALLBACK,
+
+  // Always calculate http.endpoint from http.url, even when http.route is
+  // present. Both tags will be set, allowing for comparison between
+  // user-provided routes and automatically inferred endpoints.
+  ALWAYS_CALCULATE,
+};
+
+}  // namespace tracing
+}  // namespace datadog

include/datadog/trace_segment.h

@@ -37,6 +37,7 @@
 #include "runtime_id.h"
 #include "sampling_decision.h"
 #include "sampling_priority.h"
+#include "tracer_config.h"
 
 namespace datadog {
 namespace telemetry {
@@ -79,6 +80,8 @@ class TraceSegment {
 
   std::shared_ptr<ConfigManager> config_manager_;
 
+  HttpEndpointCalculationMode resource_renaming_mode_;
+
   bool tracing_enabled_;
 
  public:
@@ -97,6 +100,7 @@ class TraceSegment {
                Optional<std::string> additional_w3c_tracestate,
                Optional<std::string> additional_datadog_w3c_tracestate,
                std::unique_ptr<SpanData> local_root,
+               HttpEndpointCalculationMode resource_renaming_mode,
                bool tracing_enabled = true);
 
   const SpanDefaults& defaults() const;

include/datadog/tracer.h

@@ -55,6 +55,7 @@ class Tracer {
   bool baggage_injection_enabled_;
   bool baggage_extraction_enabled_;
   bool tracing_enabled_;
+  HttpEndpointCalculationMode resource_renaming_mode_;
 
  public:
   // Create a tracer configured using the specified `config`, and optionally:

include/datadog/tracer_config.h

@@ -15,6 +15,7 @@
 #include "clock.h"
 #include "datadog_agent_config.h"
 #include "expected.h"
+#include "http_endpoint_calculation_mode.h"
 #include "propagation_style.h"
 #include "runtime_id.h"
 #include "span_defaults.h"
@@ -178,6 +179,15 @@ struct TracerConfig {
   /// Overridden by the `DD_APM_TRACING_ENABLED` environment variable. Defaults
   /// to `true`.
   Optional<bool> tracing_enabled;
+
+  // Whether generation of http.endpoint is enabled. This is disabled by
+  // default.
+  Optional<bool> resource_renaming_enabled;
+
+  // Whether http.endpoint is always calculated, even when http.route is
+  // present. This is disabled by default.
+  // This option is ignored if `resource_renaming_enabled` is not `true`.
+  Optional<bool> resource_renaming_always_simplified_endpoint;
 };
 
 // `FinalizedTracerConfig` contains `Tracer` implementation details derived from
@@ -218,6 +228,7 @@ class FinalizedTracerConfig final {
   std::shared_ptr<EventScheduler> event_scheduler;
   std::shared_ptr<HTTPClient> http_client;
   bool tracing_enabled;
+  HttpEndpointCalculationMode resource_renaming_mode;
 };
 
 // Return a `FinalizedTracerConfig` from the specified `config` and from any