[k2] dump k2 built-in invocations (#1554)

Author: Shamzik

runtime-light/stdlib/diagnostics/functions-call-stats.h

@@ -6,4 +6,4 @@
 
 #include "runtime-light/stdlib/diagnostics/logs.h"
 
-#define DUMP_BUILTIN_CALL_STATS(builtin_name, builtin_call) (kphp::log::debug("built-in called: " builtin_name), builtin_call)
+#define DUMP_BUILTIN_CALL_STATS(builtin_name, builtin_call) (kphp::log::debug("std function called: " builtin_name), builtin_call)

tests/aggregate_std_function_invocations.py

@@ -0,0 +1,63 @@
+import json
+import csv
+import argparse
+import pathlib
+import collections
+
+SCRIPT_DIR = pathlib.Path(__file__).parent
+DEFAULT_OUTPUT = SCRIPT_DIR / 'tmp' / 'std_function_invocations.csv'
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Aggregates std function invocations from multiple JSON files into a single CSV report."
+    )
+
+    parser.add_argument(
+        'inputs',
+        nargs='+',
+        type=pathlib.Path,
+        help='Paths to the JSON files generated by test groups',
+    )
+
+    parser.add_argument(
+        '--output', '-o',
+        type=pathlib.Path,
+        default=DEFAULT_OUTPUT,
+        help=f'Path to the output CSV file (default: {DEFAULT_OUTPUT})',
+    )
+    
+    args = parser.parse_args()
+    total_counts = collections.defaultdict(int)
+
+    for file_path in args.inputs:
+        if file_path.exists():
+            try:
+                data = json.loads(file_path.read_text(encoding='utf-8'))
+                if isinstance(data, dict):
+                    for func_name, count in data.items():
+                        total_counts[func_name] += count
+                else:
+                    print(f"Warning: {file_path} content is not a dictionary. Skipping.")
+            except json.JSONDecodeError:
+                print(f"Error: {file_path} is not a valid JSON file. Skipping.")
+        else:
+            print(f"Info: {file_path} not found. It might have been disabled.")
+
+    args.output.parent.mkdir(parents=True, exist_ok=True)
+
+    try:
+        with args.output.open('w', newline='', encoding='utf-8') as f:
+            writer = csv.writer(f)
+            writer.writerow(['Built-in', 'Call count'])
+
+            sorted_results = sorted(total_counts.items(), key=lambda item: (item[1], item[0]))
+
+            for func, count in sorted_results:
+                writer.writerow([func, count])
+        
+        print(f"Success! Aggregated report saved to: {args.output.absolute()}")
+    except Exception as e:
+        print(f"Error saving report: {e}")
+
+if __name__ == "__main__":
+    main()

…on/lib/k2_kphp_tracked_builtins_list.txt tests/k2_kphp_tracked_builtins_list.txttests/python/lib/k2_kphp_tracked_builtins_list.txt renamed to tests/k2_kphp_tracked_builtins_list.txt tests/python/lib/k2_kphp_tracked_builtins_list.txt renamed to tests/k2_kphp_tracked_builtins_list.txt

@@ -353,15 +353,11 @@ rpc_queue_next
 rpc_server_fetch_request
 rpc_server_store_response
 rpc_tl_pending_queries_count
-rpc_tl_query
-rpc_tl_query_result
-rpc_tl_query_result_synchronously
 rsort
 rtrim
 sched_yield
 sched_yield_sleep
 serialize
-set_detect_incorrect_encoding_names_warning
 set_fail_rpc_on_int32_overflow
 set_json_log_on_timeout_mode
 set_migration_php8_warning
@@ -413,9 +409,6 @@ tan
 time
 to_array_debug
 trim
-typed_rpc_tl_query
-typed_rpc_tl_query_result
-typed_rpc_tl_query_result_synchronously
 uasort
 UberH3$$geoToH3
 UberH3$$h3ToParent
@@ -454,3 +447,81 @@ warning
 wordwrap
 zstd_compress
 zstd_uncompress
+flush
+getopt
+is_readable
+kphp_extended_instance_cache_metrics_init
+md5_file
+fgetcsv
+fgets
+file_exists
+file_put_contents
+prepare_search_query
+set_json_log_demangle_stacktrace
+strftime
+ini_set
+get_webserver_stats
+thread_pool_test_load
+chmod
+getimagesize
+copy
+dirname
+feof
+filemtime
+fputcsv
+fseek
+ftell
+is_dir
+mkdir
+rename
+rewind
+scandir
+tempnam
+stream_context_create
+kphp_job_worker_fetch_request
+kphp_job_worker_start
+kphp_job_worker_start_multi
+kphp_job_worker_start_no_reply
+kphp_job_worker_store_response
+kphp_backtrace
+kphp_turn_on_host_tag_in_inner_statshouse_metrics_toggle
+memory_get_static_usage
+openssl_x509_checkpurpose
+openssl_x509_verify
+preg_last_error
+new_rpc_connection
+store_finish
+fetch_lookup_int
+gethostbynamel
+localtime
+system
+kphp_tracing_func_enter_branch
+kphp_tracing_get_current_active_span
+kphp_tracing_get_level
+kphp_tracing_get_root_span
+kphp_tracing_init
+kphp_tracing_register_enums_provider
+kphp_tracing_register_on_finish
+kphp_tracing_register_rpc_details_provider
+msgpack_deserialize_safe
+kphp_tracing_set_level
+kphp_tracing_start_span
+KphpDiv$$assignTraceCtx
+KphpDiv$$generateTraceCtxForChild
+preg_split
+KphpSpan$$addAttributeBool
+KphpSpan$$addAttributeFloat
+KphpSpan$$addAttributeInt
+KphpSpan$$addAttributeString
+KphpSpan$$addEvent
+KphpSpan$$exclude
+KphpSpan$$finish
+KphpSpan$$finishWithError
+KphpSpan$$updateName
+KphpSpanEvent$$addAttributeInt
+KphpSpanEvent$$addAttributeString
+ignore_user_abort
+send_http_103_early_hints
+profiler_set_function_label
+vk_dot_product
+vk_flex

tests/kphp_ci_pipeline_runner.py

@@ -4,6 +4,8 @@
 import math
 import multiprocessing
 import os
+import pathlib
+import shlex
 import signal
 import subprocess
 import sys
@@ -13,6 +15,10 @@
 from python.lib.nocc_for_kphp_tester import nocc_env
 
 
+K2_KPHP_TRACKED_BUILTINS_LIST_PATH = pathlib.Path(__file__).parent / "k2_kphp_tracked_builtins_list.txt"
+K2_KPHP_TRACKED_BUILTINS_LIST = K2_KPHP_TRACKED_BUILTINS_LIST_PATH.read_text()
+
+
 class TestStatus(Enum):
     FAILED = red("failed")
     PASSED = green("passed")
@@ -348,9 +354,11 @@ def _calculate_pytest_jobs_count(default_percent: int = 95) -> int:
         name="k2-kphp-tests",
         description="run k2-kphp tests with cxx={}".format(args.cxx_name),
         cmd="KPHP_TESTS_POLYFILLS_REPO={kphp_polyfills_repo} "
+        "KPHP_TRACKED_BUILTINS_LIST={K2_KPHP_TRACKED_BUILTINS_LIST} "
         "{kphp_runner} -j{jobs} --cxx-name {cxx_name} --k2-bin {k2_bin}".format(
             jobs=n_cpu,
             kphp_polyfills_repo=kphp_polyfills_repo,
+            K2_KPHP_TRACKED_BUILTINS_LIST=shlex.quote(K2_KPHP_TRACKED_BUILTINS_LIST),
             kphp_runner=kphp_test_runner,
             cxx_name=args.cxx_name,
             k2_bin=args.k2_bin,
@@ -427,10 +435,17 @@ def _calculate_pytest_jobs_count(default_percent: int = 95) -> int:
     runner.add_test_group(
         name="k2-functional-tests",
         description="run k2-kphp functional tests with cxx={}".format(args.cxx_name),
-        cmd="KPHP_TESTS_POLYFILLS_REPO={kphp_polyfills_repo} KPHP_CXX={cxx_name} K2_BIN={k2_bin} K2_MONITORING_ABORT_HANGING_GLOBAL_TASK=false K2_MONITORING_ABORT_HANGING_REQUEST_TASK=false python3 -m pytest --basetemp={base_tempdir} --tb=native -n{jobs} {functional_tests_dir}".format(
+        cmd="KPHP_TESTS_POLYFILLS_REPO={kphp_polyfills_repo} "
+        "KPHP_CXX={cxx_name} "
+        "K2_BIN={k2_bin} "
+        "K2_MONITORING_ABORT_HANGING_GLOBAL_TASK=false "
+        "K2_MONITORING_ABORT_HANGING_REQUEST_TASK=false "
+        "KPHP_TRACKED_BUILTINS_LIST={K2_KPHP_TRACKED_BUILTINS_LIST} "
+        "python3 -m pytest --basetemp={base_tempdir} --tb=native -n{jobs} {functional_tests_dir}".format(
             kphp_polyfills_repo=kphp_polyfills_repo,
             cxx_name=args.cxx_name,
             k2_bin=args.k2_bin,
+            K2_KPHP_TRACKED_BUILTINS_LIST=shlex.quote(K2_KPHP_TRACKED_BUILTINS_LIST),
             jobs=n_cpu,
             functional_tests_dir=functional_tests_dir,
             base_tempdir=os.path.expanduser(
@@ -474,6 +489,7 @@ def _calculate_pytest_jobs_count(default_percent: int = 95) -> int:
             "KPHP_TESTS_INTERGRATION_TESTS_ENABLED=1 "
             "KPHP_CXX={cxx_name} "
             "K2_BIN={k2_bin} "
+            "KPHP_TRACKED_BUILTINS_LIST={K2_KPHP_TRACKED_BUILTINS_LIST} "
             "python3 -m pytest --tb=native -n{jobs} {tests_dir}".format(
                 jobs=n_cpu,
                 lib_dir=os.path.join(runner_dir, "python"),
@@ -482,6 +498,7 @@ def _calculate_pytest_jobs_count(default_percent: int = 95) -> int:
                 kphp_polyfills_repo=kphp_polyfills_repo,
                 cxx_name=args.cxx_name,
                 k2_bin=args.k2_bin,
+                K2_KPHP_TRACKED_BUILTINS_LIST=shlex.quote(K2_KPHP_TRACKED_BUILTINS_LIST),
                 tests_dir=" ".join([
                     os.path.join(args.kphp_tests_repo, "python/tests/k2_rpc_client/"),
                     os.path.join(args.kphp_tests_repo, "python/tests/k2_rpc_server/"),

tests/kphp_tester.py

@@ -3,20 +3,26 @@
 import math
 import multiprocessing
 import os
+import pathlib
 import re
 import signal
 import sys
 from functools import partial
 from multiprocessing.dummy import Pool as ThreadPool
+import typing
 
 from python.lib.colors import red, green, yellow, blue, cyan
 from python.lib.file_utils import search_php_bin
 from python.lib.nocc_for_kphp_tester import nocc_start_daemon_in_background
 from python.lib.kphp_run_once import KphpRunOnce
+from python.lib import std_function
 from python.lib import tcp
 
 TCP_SERVER_TAG_PREFIX = "tcp_server:"
 
+FILE = pathlib.Path(__file__)
+TMP_DIR = FILE.with_name("{}_tmp".format(FILE.stem))
+
 
 class TestFile:
     def __init__(self, file_path, test_tmp_dir, tags, env_vars: dict, out_regexps=None, forbidden_regexps=None):
@@ -55,13 +61,16 @@ def is_php8(self):
     def is_available_for_k2(self):
         return "k2_skip" not in self.tags
 
-    def make_kphp_once_runner(self, use_nocc, cxx_name, k2_bin):
+    def make_kphp_once_runner(
+            self, use_nocc, cxx_name, k2_bin, std_function_invocations: typing.Optional[std_function.Invocations]
+    ):
         tester_dir = os.path.abspath(os.path.dirname(__file__))
         return KphpRunOnce(
             php_script_path=self.file_path,
             working_dir=os.path.abspath(os.path.join(self.test_tmp_dir, "working_dir")),
             artifacts_dir=os.path.abspath(os.path.join(self.test_tmp_dir, "artifacts")),
             php_bin=search_php_bin(php_version=self.php_version),
+            std_function_invocations=std_function_invocations,
             extra_include_dirs=[os.path.join(tester_dir, "php_include")],
             vkext_dir=os.path.abspath(os.path.join(tester_dir, os.path.pardir, "objs", "vkext")),
             use_nocc=use_nocc,
@@ -75,7 +84,6 @@ def set_up_env_for_k2(self):
         self.env_vars["KPHP_ENABLE_GLOBAL_VARS_MEMORY_STATS"] = "0"
         self.env_vars["KPHP_PROFILER"] = "0"
         self.env_vars["KPHP_FORCE_LINK_RUNTIME"] = "1"
-        self.env_vars["KPHP_TRACKED_BUILTINS_LIST"] = KphpRunOnce.K2_KPHP_TRACKED_BUILTINS_LIST
 
 
 def make_test_file(file_path, test_tmp_dir, test_tags):
@@ -152,12 +160,11 @@ def test_files_from_list(tests_dir, test_list):
 
 def collect_tests(tests_dir, test_tags, test_list):
     tests = []
-    tmp_dir = "{}_tmp".format(__file__[:-3])
     file_it = test_files_from_list(tests_dir, test_list) if test_list else test_files_from_dir(tests_dir)
     for root, file in file_it:
         if file.endswith(".php") or file.endswith(".phpt"):
             test_file_path = os.path.join(root, file)
-            test_tmp_dir = os.path.join(tmp_dir, os.path.relpath(test_file_path, os.path.dirname(tests_dir)))
+            test_tmp_dir = os.path.join(TMP_DIR, os.path.relpath(test_file_path, os.path.dirname(tests_dir)))
             test_tmp_dir = test_tmp_dir[:-4] if test_tmp_dir.endswith(".php") else test_tmp_dir[:-5]
             test_file = make_test_file(test_file_path, test_tmp_dir, test_tags)
             if test_file:
@@ -350,11 +357,13 @@ def run_ok_test(test: TestFile, runner):
     return TestResult.passed(test, runner.artifacts)
 
 
-def run_test(use_nocc, cxx_name, k2_bin, test: TestFile):
+def run_test(
+        use_nocc, cxx_name, k2_bin, std_function_invocations: typing.Optional[std_function.Invocations], test: TestFile
+):
     if not os.path.exists(test.file_path):
         return TestResult.failed(test, None, "can't find test file")
 
-    runner = test.make_kphp_once_runner(use_nocc, cxx_name, k2_bin)
+    runner = test.make_kphp_once_runner(use_nocc, cxx_name, k2_bin, std_function_invocations)
     runner.remove_artifacts_dir()
     if k2_bin is not None:
         test.set_up_env_for_k2()
@@ -393,17 +402,32 @@ def run_all_tests(tests_dir, jobs, test_tags, no_report, passed_list, test_list,
             "tag" if len(test_tags) == 1 else "tags"))
         sys.exit(1)
 
+    if "KPHP_TRACKED_BUILTINS_LIST" in os.environ:
+        std_function_invocations = std_function.Invocations(os.environ["KPHP_TRACKED_BUILTINS_LIST"])
+    else:
+        std_function_invocations = None
+
     results = []
     with ThreadPool(jobs) as pool:
         tests_completed = 0
-        for test_result in pool.imap_unordered(partial(run_test, use_nocc, cxx_name, k2_bin), tests):
+        for test_result in pool.imap_unordered(partial(run_test, use_nocc, cxx_name, k2_bin, std_function_invocations), tests):
             if hack_reference_exit:
                 print(yellow("Testing process was interrupted"), flush=True)
                 break
             tests_completed = tests_completed + 1
             test_result.print_short_report(len(tests), tests_completed)
             results.append(test_result)
 
+    if std_function_invocations:
+        std_function_invocations_output_dir = TMP_DIR / "artifacts"
+        std_function_invocations_output_dir.mkdir(parents=True, exist_ok=True)
+
+        std_function_invocations_filename = "std_function_invocations.json"
+        std_function_invocations_output_path = std_function_invocations_output_dir / std_function_invocations_filename
+
+        with open(std_function_invocations_output_path, "w", encoding="utf-8") as f:
+            std_function_invocations.dump(f)
+
     print("\nTesting results:", flush=True)
 
     skipped = len(tests) - len(results)