Adds process_tags to the first span of each tracing payload (#3566)

* init process tags for APM

Co-Authored-By: PROFeNoM <[email protected]>

* feat(process_tags): add process_tags to tracing payloads

* small auto review and fix test

* bwoebi review

* fix test

* Move process tags assign

Signed-off-by: Bob Weinand <[email protected]>

---------

Signed-off-by: Bob Weinand <[email protected]>
Co-authored-by: PROFeNoM <[email protected]>
Co-authored-by: Bob Weinand <[email protected]>

Author: 3 people

components-rs/ddtrace.h

@@ -239,4 +239,8 @@ void ddog_add_zstr_span_meta_struct_CharSlice(ddog_SpanBytes *ptr,
                                               struct _zend_string *key,
                                               ddog_CharSlice val);
 
+const char *ddog_normalize_process_tag_value(ddog_CharSlice val);
+
+void ddog_free_normalized_tag_value(const char* normalized_val);
+
 #endif  /* DDTRACE_PHP_H */

components-rs/lib.rs

@@ -132,3 +132,54 @@ pub unsafe extern "C" fn posix_spawn_file_actions_addchdir_np(
         -libc::ENOSYS
     }
 }
+
+const MAX_TAG_VALUE_LENGTH: usize = 100;
+
+#[no_mangle]
+pub extern "C" fn ddog_normalize_process_tag_value(
+    tag_value: CharSlice,
+) -> *const c_char {
+    let value = tag_value.to_utf8_lossy();
+
+    let mut out = String::new();
+    let mut prev_underscore = false;
+    let mut started = false;
+    for c in value.chars().take(MAX_TAG_VALUE_LENGTH) {
+        if c.is_alphanumeric() || matches!(c, '/' | '.' | '-') {
+            for lc in c.to_lowercase() {
+                out.push(lc);
+            }
+            started = true;
+            prev_underscore = false;
+        } else {
+            if started && !prev_underscore {
+                out.push('_');
+                prev_underscore = true;
+            }
+        }
+    }
+
+    // trim trailing underscores
+    if out.ends_with('_') {
+        out.pop();
+    }
+
+    match std::ffi::CString::new(out) {
+        Ok(c_string) => {
+            let out_ptr = c_string.as_ptr();
+            std::mem::forget(c_string);
+            out_ptr
+        }
+        Err(_) => std::ptr::null(),
+    }
+}
+
+#[no_mangle]
+pub extern "C" fn ddog_free_normalized_tag_value(ptr: *const c_char) {
+    if ptr.is_null() {
+        return;
+    }
+    unsafe {
+        drop(std::ffi::CString::from_raw(ptr as *mut c_char));
+    }
+}

config.m4

@@ -203,6 +203,7 @@ if test "$PHP_DDTRACE" != "no"; then
     ext/memory_limit.c \
     ext/otel_config.c \
     ext/priority_sampling/priority_sampling.c \
+    ext/process_tags.c \
     ext/profiling.c \
     ext/random.c \
     ext/remote_config.c \

config.w32

@@ -49,6 +49,7 @@ if (PHP_DDTRACE != 'no') {
 	DDTRACE_EXT_SOURCES += " logging.c";
 	DDTRACE_EXT_SOURCES += " memory_limit.c";
 	DDTRACE_EXT_SOURCES += " otel_config.c";
+	DDTRACE_EXT_SOURCES += " process_tags.c";
 	DDTRACE_EXT_SOURCES += " profiling.c";
 	DDTRACE_EXT_SOURCES += " random.c";
 	DDTRACE_EXT_SOURCES += " remote_config.c";

ext/configuration.h

@@ -263,6 +263,7 @@ enum ddtrace_sampling_rules_format {
     CONFIG(INT, DD_CODE_ORIGIN_MAX_USER_FRAMES, "8")                                                           \
     CONFIG(BOOL, DD_TRACE_RESOURCE_RENAMING_ENABLED, "false")                                                  \
     CONFIG(BOOL, DD_TRACE_RESOURCE_RENAMING_ALWAYS_SIMPLIFIED_ENDPOINT, "false")                               \
+    CONFIG(BOOL, DD_EXPERIMENTAL_PROPAGATE_PROCESS_TAGS_ENABLED, "false")                                      \
     DD_INTEGRATIONS
 
 #ifndef _WIN32

ext/ddtrace.c

@@ -68,6 +68,7 @@
 #include "limiter/limiter.h"
 #include "standalone_limiter.h"
 #include "priority_sampling/priority_sampling.h"
+#include "process_tags.h"
 #include "random.h"
 #include "autoload_php_files.h"
 #include "remote_config.h"
@@ -1615,6 +1616,7 @@ static PHP_MSHUTDOWN_FUNCTION(ddtrace) {
     ddtrace_sidecar_shutdown();
 
     ddtrace_live_debugger_mshutdown();
+    ddtrace_process_tags_mshutdown();
 
 #if PHP_VERSION_ID >= 80000 && PHP_VERSION_ID < 80100
     // See dd_register_span_data_ce for explanation
@@ -1637,6 +1639,11 @@ static void dd_rinit_once(void) {
      */
     ddtrace_startup_logging_first_rinit();
 
+    // Collect process tags now that script path is available
+    if (get_global_DD_EXPERIMENTAL_PROPAGATE_PROCESS_TAGS_ENABLED()) {
+        ddtrace_process_tags_first_rinit();
+    }
+
     // Uses config, cannot run earlier
 #ifndef _WIN32
     ddtrace_signals_first_rinit();
@@ -2610,6 +2617,26 @@ PHP_FUNCTION(DDTrace_Testing_trigger_error) {
     }
 }
 
+PHP_FUNCTION(DDTrace_Testing_normalize_tag_value) {
+    ddtrace_string value;
+    if (zend_parse_parameters(ZEND_NUM_ARGS(), "s", &value.ptr, &value.len) != SUCCESS) {
+        RETURN_EMPTY_STRING();
+    }
+
+    const char* normalized = ddog_normalize_process_tag_value((ddog_CharSlice){
+        .ptr = value.ptr,
+        .len = value.len
+    });
+
+    if (normalized) {
+        zend_string *result = zend_string_init(normalized, strlen(normalized), 0);
+        ddog_free_normalized_tag_value(normalized);
+        RETURN_STR(result);
+    } else {
+        RETURN_EMPTY_STRING();
+    }
+}
+
 PHP_FUNCTION(DDTrace_Internal_add_span_flag) {
     zend_object *span;
     zend_long flag;

ext/ddtrace.stub.php

@@ -904,6 +904,13 @@ function trigger_error(string $message, int $errorType): void {}
      * Emits an asm event
      */
     function emit_asm_event(): void {}
+    /**
+     * Normalizes a process tag value
+     *
+     * @param string $value The tag value to normalize
+     * @return string The normalized tag value
+     */
+    function normalize_tag_value(string $value): string {}
 }
 
 namespace DDTrace\Internal {
@@ -939,24 +946,24 @@ function handle_fork(): void {}
 
 namespace datadog\appsec\v2 {
     /**
- * Track a user login success event.
- *
- * @param string $login is the data used by the user to authenticate
- * @param string|array $user when string, it represents the user id. When array it represents the user information.
- *  The array should at least contain the following keys:
- * - id: string, Unique identifier of the user. Should be the same id and format used on set_user
- * @param array $metadata User metadata added to the root span
- */
-function track_user_login_success(string $login, string|array|null $user = null, array $metadata = []): void {}
-
-/**
- * Track a user login failure event.
- *
- * @param string $login is the data used by the user to authenticate
- * @param bool $exists Whether the user exists in the system
- * @param array $metadata User metadata added to the root span
- */
-function track_user_login_failure(string $login, bool $exists, array $metadata = []): void {}
+     * Track a user login success event.
+     *
+     * @param string $login is the data used by the user to authenticate
+     * @param string|array $user when string, it represents the user id. When array it represents the user information.
+     *  The array should at least contain the following keys:
+     * - id: string, Unique identifier of the user. Should be the same id and format used on set_user
+     * @param array $metadata User metadata added to the root span
+     */
+    function track_user_login_success(string $login, string|array|null $user = null, array $metadata = []): void {}
+
+    /**
+     * Track a user login failure event.
+     *
+     * @param string $login is the data used by the user to authenticate
+     * @param bool $exists Whether the user exists in the system
+     * @param array $metadata User metadata added to the root span
+     */
+    function track_user_login_failure(string $login, bool $exists, array $metadata = []): void {}
 }
 
 namespace {

ext/ddtrace_arginfo.h

@@ -203,6 +203,10 @@ ZEND_END_ARG_INFO()
 
 #define arginfo_DDTrace_Testing_emit_asm_event arginfo_DDTrace_flush
 
+ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX(arginfo_DDTrace_Testing_normalize_tag_value, 0, 1, IS_STRING, 0)
+	ZEND_ARG_TYPE_INFO(0, value, IS_STRING, 0)
+ZEND_END_ARG_INFO()
+
 ZEND_BEGIN_ARG_WITH_RETURN_TYPE_INFO_EX(arginfo_DDTrace_Internal_add_span_flag, 0, 2, IS_VOID, 0)
 	ZEND_ARG_OBJ_INFO(0, span, DDTrace\\SpanData, 0)
 	ZEND_ARG_TYPE_INFO(0, flag, IS_LONG, 0)
@@ -383,6 +387,7 @@ ZEND_FUNCTION(DDTrace_UserRequest_notify_commit);
 ZEND_FUNCTION(DDTrace_UserRequest_set_blocking_function);
 ZEND_FUNCTION(DDTrace_Testing_trigger_error);
 ZEND_FUNCTION(DDTrace_Testing_emit_asm_event);
+ZEND_FUNCTION(DDTrace_Testing_normalize_tag_value);
 ZEND_FUNCTION(DDTrace_Internal_add_span_flag);
 ZEND_FUNCTION(DDTrace_Internal_handle_fork);
 ZEND_FUNCTION(datadog_appsec_v2_track_user_login_success);
@@ -473,6 +478,7 @@ static const zend_function_entry ext_functions[] = {
 	ZEND_RAW_FENTRY(ZEND_NS_NAME("DDTrace\\UserRequest", "set_blocking_function"), zif_DDTrace_UserRequest_set_blocking_function, arginfo_DDTrace_UserRequest_set_blocking_function, 0, NULL, NULL)
 	ZEND_RAW_FENTRY(ZEND_NS_NAME("DDTrace\\Testing", "trigger_error"), zif_DDTrace_Testing_trigger_error, arginfo_DDTrace_Testing_trigger_error, 0, NULL, NULL)
 	ZEND_RAW_FENTRY(ZEND_NS_NAME("DDTrace\\Testing", "emit_asm_event"), zif_DDTrace_Testing_emit_asm_event, arginfo_DDTrace_Testing_emit_asm_event, 0, NULL, NULL)
+	ZEND_RAW_FENTRY(ZEND_NS_NAME("DDTrace\\Testing", "normalize_tag_value"), zif_DDTrace_Testing_normalize_tag_value, arginfo_DDTrace_Testing_normalize_tag_value, 0, NULL, NULL)
 	ZEND_RAW_FENTRY(ZEND_NS_NAME("DDTrace\\Internal", "add_span_flag"), zif_DDTrace_Internal_add_span_flag, arginfo_DDTrace_Internal_add_span_flag, 0, NULL, NULL)
 	ZEND_RAW_FENTRY(ZEND_NS_NAME("DDTrace\\Internal", "handle_fork"), zif_DDTrace_Internal_handle_fork, arginfo_DDTrace_Internal_handle_fork, 0, NULL, NULL)
 	ZEND_RAW_FENTRY(ZEND_NS_NAME("datadog\\appsec\\v2", "track_user_login_success"), zif_datadog_appsec_v2_track_user_login_success, arginfo_datadog_appsec_v2_track_user_login_success, 0, NULL, NULL)