Merge pull request #2384 from DataDog/bob/extended-sampling

Implement extended sampling
DataDog · Dec 12, 2023 · b7a55c5 · b7a55c5
2 parents e33deef + ef2686c
commit b7a55c5
Show file tree

Hide file tree

Showing 40 changed files with 784 additions and 200 deletions.
diff --git a/ext/compatibility.h b/ext/compatibility.h
@@ -65,6 +65,15 @@ static inline zval *ddtrace_assign_variable(zval *variable_ptr, zval *value) {
 #define MAY_BE_ARRAY 0
 
 #define Z_EXTRA_P(z) Z_NEXT_P(z)
+
+#undef zval_get_long
+#define zval_get_long ddtrace_zval_get_long
+static inline zend_long zval_get_long(zval *op) {
+    if (Z_ISUNDEF_P(op)) {
+        return 0;
+    }
+    return _zval_get_long(op);
+}
 #endif
 
 #if PHP_VERSION_ID < 70200

diff --git a/ext/configuration.c b/ext/configuration.c
@@ -67,6 +67,19 @@ static bool dd_parse_dbm_mode(zai_str value, zval *decoded_value, bool persisten
     return true;
 }
 
+static bool dd_parse_sampling_rules_format(zai_str value, zval *decoded_value, bool persistent) {
+    UNUSED(persistent);
+    if (zai_str_eq_ci_cstr(value, "regex")) {
+        ZVAL_LONG(decoded_value, DD_TRACE_SAMPLING_RULES_FORMAT_REGEX);
+    } else if (zai_str_eq_ci_cstr(value, "glob")) {
+        ZVAL_LONG(decoded_value, DD_TRACE_SAMPLING_RULES_FORMAT_GLOB);
+    } else {
+        return false;
+    }
+
+    return true;
+}
+
 // Allow for partially defined struct initialization here
 #pragma GCC diagnostic ignored "-Wmissing-field-initializers"
 

diff --git a/ext/configuration.h b/ext/configuration.h
@@ -23,6 +23,12 @@ enum ddtrace_dbm_propagation_mode {
     DD_TRACE_DBM_PROPAGATION_FULL,
 };
 
+// To remove in 1.0
+enum ddtrace_sampling_rules_format {
+    DD_TRACE_SAMPLING_RULES_FORMAT_REGEX,
+    DD_TRACE_SAMPLING_RULES_FORMAT_GLOB
+};
+
 /* From the curl docs on CONNECT_TIMEOUT_MS:
  *     If libcurl is built to use the standard system name resolver, that
  *     portion of the transfer will still use full-second resolution for
@@ -116,6 +122,7 @@ enum ddtrace_dbm_propagation_mode {
     CONFIG(INT, DD_TRACE_RATE_LIMIT, "0", .ini_change = zai_config_system_ini_change)                          \
     CALIAS(DOUBLE, DD_TRACE_SAMPLE_RATE, "-1", CALIASES("DD_SAMPLING_RATE"))                                   \
     CONFIG(JSON, DD_TRACE_SAMPLING_RULES, "[]")                                                                \
+    CONFIG(CUSTOM(INT), DD_TRACE_SAMPLING_RULES_FORMAT, "regex", .parser = dd_parse_sampling_rules_format)     \
     CONFIG(JSON, DD_SPAN_SAMPLING_RULES, "[]")                                                                 \
     CONFIG(STRING, DD_SPAN_SAMPLING_RULES_FILE, "", .ini_change = ddtrace_alter_sampling_rules_file_config)    \
     CONFIG(SET_LOWERCASE, DD_TRACE_HEADER_TAGS, "")                                                            \

diff --git a/ext/ddshared.c b/ext/ddshared.c
@@ -1,9 +1,11 @@
 #include <components-rs/ddtrace.h>
 #include <Zend/zend_API.h>
+#include <components/log/log.h>
 
 #include "ddshared.h"
+#include "configuration.h"
 #include "ddtrace.h"
-#include <components/log/log.h>
+#include "uri_normalization.h"
 
 ZEND_EXTERN_MODULE_GLOBALS(ddtrace);
 
@@ -14,3 +16,82 @@ void ddshared_minit(void) {
 
     ddtrace_php_version = Z_STR_P(zend_get_constant_str(ZEND_STRL("PHP_VERSION")));
 }
+
+ bool dd_rule_matches(zval *pattern, zval *prop, int rulesFormat) {
+    if (Z_TYPE_P(pattern) != IS_STRING) {
+        return false;
+    }
+    if (Z_TYPE_P(prop) != IS_STRING) {
+        return true;  // default case unset or null must be true, everything else is too then...
+    }
+
+    if (rulesFormat == DD_TRACE_SAMPLING_RULES_FORMAT_GLOB) {
+        return dd_glob_rule_matches(pattern, Z_STR_P(prop));
+    }
+    else {
+        return zai_match_regex(Z_STR_P(pattern), Z_STR_P(prop));
+    }
+}
+
+bool dd_glob_rule_matches(zval *pattern, zend_string* value) {
+    if (Z_TYPE_P(pattern) != IS_STRING) {
+        return false;
+    }
+
+    char *p = Z_STRVAL_P(pattern);
+    char *s = ZSTR_VAL(value);
+
+    int wildcards = 0;
+    int patternLength = 0;
+    int stringLength = ZSTR_LEN(value);
+    while (*p) {
+        if (*(p++) == '*') {
+            ++wildcards;
+        }
+        patternLength++;
+    }
+
+    // If there are no wildcards, no need to go through the whole string if pattern is shorter than the input string
+    // Indeed wildcards (ie '*') can replace multiple characters while '?' canonly replace one
+    if (wildcards == 0 && patternLength < stringLength) {
+        return false;
+    }
+
+    p = Z_STRVAL_P(pattern);
+
+    ALLOCA_FLAG(use_heap)
+    char **backtrack_points = do_alloca(wildcards * 2 * sizeof(char *), use_heap);
+    int backtrack_idx = 0;
+
+    while (*p) {
+        if (!*s) {
+            while (*p == '*') {
+                ++p;
+            }
+            free_alloca(backtrack_points, use_heap);
+            return !*p;
+        }
+        if (*s == *p || *p == '?') {
+            ++s, ++p;
+        } else if (*p == '*') {
+            backtrack_points[backtrack_idx++] = ++p;
+            backtrack_points[backtrack_idx++] = s;
+        } else {
+            do {
+                if (backtrack_idx > 0) {
+                    backtrack_idx -= 2;
+                    p = backtrack_points[backtrack_idx];
+                    s = ++backtrack_points[backtrack_idx + 1];
+                } else {
+                    free_alloca(backtrack_points, use_heap);
+                    return false;
+                }
+            } while (!*s);
+            backtrack_idx += 2;
+        }
+    }
+
+    free_alloca(backtrack_points, use_heap);
+
+    return true;
+}
diff --git a/ext/ddshared.h b/ext/ddshared.h
@@ -6,5 +6,7 @@
 extern zend_string *ddtrace_php_version;
 
 void ddshared_minit(void);
+bool dd_rule_matches(zval *pattern, zval *prop, int rulesFormat);
+bool dd_glob_rule_matches(zval *pattern, zend_string* value);
 
 #endif  // DD_TRACE_SHARED_H
diff --git a/ext/ddtrace.c b/ext/ddtrace.c
@@ -632,14 +632,16 @@ static zval *ddtrace_root_span_data_write(zend_object *object, zend_string *memb
             }
         }
     } else if (zend_string_equals_literal(prop_name, "traceId")) {
-        span->trace_id = Z_TYPE_P(value) == IS_STRING ? ddtrace_parse_userland_trace_id(Z_STR_P(value)) : (ddtrace_trace_id){ 0 };
+        span->trace_id = Z_TYPE_P(value) == IS_STRING ? ddtrace_parse_hex_trace_id(Z_STRVAL_P(value), Z_STRLEN_P(value)) : (ddtrace_trace_id){ 0 };
         if (!span->trace_id.low && !span->trace_id.high) {
             span->trace_id = (ddtrace_trace_id) {
                 .low = span->span_id,
                 .time = get_DD_TRACE_128_BIT_TRACEID_GENERATION_ENABLED() ? span->start / UINT64_C(1000000000) : 0,
             };
             value = &span->property_id;
         }
+    } else if (zend_string_equals_literal(prop_name, "samplingPriority")) {
+        span->explicit_sampling_priority = zval_get_long(value) != DDTRACE_PRIORITY_SAMPLING_UNKNOWN;
     }
 
 #if PHP_VERSION_ID >= 70400

diff --git a/ext/distributed_tracing_headers.c b/ext/distributed_tracing_headers.c
@@ -7,13 +7,6 @@
 
 ZEND_EXTERN_MODULE_GLOBALS(ddtrace);
 
-static ddtrace_trace_id dd_parse_b3_trace_id(char *trace_id, ssize_t trace_id_len) {
-    return (ddtrace_trace_id){
-        .high = trace_id_len > 16 ? ddtrace_parse_hex_span_id_str(trace_id, MIN(16, trace_id_len - 16)) : 0,
-        .low = ddtrace_parse_hex_span_id_str(trace_id + MAX(0, trace_id_len - 16), MIN(16, trace_id_len)),
-    };
-}
-
 static inline bool dd_is_hex_char(char chr) {
     return (chr >= '0' && chr <= '9') || (chr >= 'a' && chr <= 'f');
 }
@@ -99,7 +92,7 @@ static ddtrace_distributed_tracing_result ddtrace_read_distributed_tracing_ids_b
             ++b3_ptr;
         }
 
-        result.trace_id = dd_parse_b3_trace_id(b3_traceid, b3_ptr - b3_traceid);
+        result.trace_id = ddtrace_parse_hex_trace_id(b3_traceid, b3_ptr - b3_traceid);
 
         char *b3_spanid = ++b3_ptr;
         while (b3_ptr < b3_end && *b3_ptr != '-') {
@@ -137,7 +130,7 @@ static ddtrace_distributed_tracing_result ddtrace_read_distributed_tracing_ids_b
     ddtrace_distributed_tracing_result result = dd_init_empty_result();
 
     if (read_header(ZAI_STRL("X_B3_TRACEID"), "x-b3-traceid", &trace_id_str, data)) {
-        result.trace_id = dd_parse_b3_trace_id(ZSTR_VAL(trace_id_str), ZSTR_LEN(trace_id_str));
+        result.trace_id = ddtrace_parse_hex_trace_id(ZSTR_VAL(trace_id_str), ZSTR_LEN(trace_id_str));
         zend_string_release(trace_id_str);
     }
 
@@ -482,14 +475,10 @@ void ddtrace_apply_distributed_tracing_result(ddtrace_distributed_tracing_result
         if (!span) {
             DDTRACE_G(propagated_priority_sampling) = DDTRACE_G(default_priority_sampling) = result->priority_sampling;
         } else {
-            ddtrace_set_priority_sampling_on_span(span, result->priority_sampling, DD_MECHANISM_DEFAULT);
-
-            if (result->priority_sampling == DDTRACE_PRIORITY_SAMPLING_UNSET) {
-                ZVAL_UNDEF(&zv);
-            } else {
-                ZVAL_LONG(&zv, result->priority_sampling);
-            }
+            ZVAL_LONG(&zv, result->priority_sampling);
             ddtrace_assign_variable(&span->property_propagated_sampling_priority, &zv);
+
+            ddtrace_set_priority_sampling_on_span(span, result->priority_sampling, DD_MECHANISM_DEFAULT);
         }
     }
 }