Add tests

tests/python/lib/kphp_run_once.py

@@ -93,14 +93,14 @@ def run_with_php(self, extra_options=[], runs_cnt=1):
 
         return php_proc.returncode == 0
 
-    def run_with_kphp(self, runs_cnt=1):
+    def run_with_kphp(self, runs_cnt=1, args=[]):
         self._clear_working_dir(self._kphp_runtime_tmp_dir)
 
         sanitizer_log_name = "kphp_runtime_sanitizer_log"
         env, sanitizer_glob_mask = self._prepare_sanitizer_env(self._kphp_runtime_tmp_dir, sanitizer_log_name)
 
         cmd = [self._kphp_runtime_bin, "--once={}".format(runs_cnt), "--profiler-log-prefix", "profiler.log",
-               "--worker-queries-to-reload", "1"]
+               "--worker-queries-to-reload", "1"] + args
         if not os.getuid():
             cmd += ["-u", "root", "-g", "root"]
         kphp_server_proc = subprocess.Popen(cmd,

tests/python/tests/kml/catboost/php/catboost_multiclass_tutorial_ht_catnum.php

@@ -0,0 +1,77 @@
+<?php
+
+/** @return shape(features_map:float[], ans:float[])[] */
+function getInput() {
+return [
+   shape([
+    "features_map" => [
+      "year" => 1996,
+      "count" => 197,
+      "season" => 1
+    ],
+    "ans" => [
+      -0.4315705250918395,
+      -0.07602514583990287,
+      0.5075956709317426
+    ]
+  ]),
+  shape([
+    "features_map" => [
+      "year" => 1968,
+      "count" => 37,
+      "season" => 1
+    ],
+    "ans" => [
+      -0.7547556359399779,
+      -0.9511000859169567,
+      1.7058557218569348
+    ]
+  ]),
+  shape([
+    "features_map" => [
+      "year" => 2002,
+      "count" => 77,
+      "season" => 0
+    ],
+    "ans" => [
+      -0.1531870128758389,
+      0.3682398885336728,
+      -0.21505287565783404
+    ]
+  ]),
+  shape([
+    "features_map" => [
+      "year" => 1948,
+      "count" => 59,
+      "season" => 0
+    ],
+    "ans" => [
+      -0.04081236490074001,
+      -0.7956756034401067,
+      0.8364879683408469
+    ]
+  ])
+];
+}
+
+const eps = 1e-5;
+
+function ensure($x)
+{
+    if (!$x) {
+        die(1);
+    }
+}
+
+function main()
+{
+    $input = getInput();
+    foreach ($input as $row) {
+        $predictions = kml_catboost_predict_ht_multi("catboost_multiclass_tutorial_ht_catnum", $row['features_map']);
+        ensure(count($predictions) == count($row['ans']));
+        for ($i = 0; $i < count($predictions); $i++) {
+            ensure(abs($predictions[$i] - $row['ans'][$i]) < eps);
+        }
+    }
+}
+main();

tests/python/tests/kml/catboost/php/catboost_tiny_1float_1hot_10trees.php

@@ -0,0 +1,71 @@
+<?php
+
+/** @return shape(float_features:float[], cat_features:string[], ans:float)[] */
+function getInput() {
+return [
+    shape([
+        "float_features" => [
+            1
+        ],
+        "cat_features" => [
+            "a"
+        ],
+        "ans" => 0.7363564860973357
+    ]),
+    shape([
+        "float_features" => [
+            2
+        ],
+        "cat_features" => [
+            "a"
+        ],
+        "ans" => 1.4672203311538694
+    ]),
+    shape([
+        "float_features" => [
+            3
+        ],
+        "cat_features" => [
+            "a"
+        ],
+        "ans" => -2.276542165549472
+    ]),
+    shape([
+        "float_features" => [
+            3
+        ],
+        "cat_features" => [
+            "b"
+        ],
+        "ans" => -0.9489540606737137
+    ]),
+    shape([
+        "float_features" => [
+            5
+        ],
+        "cat_features" => [
+            "b"
+        ],
+        "ans" => -0.9489540606737137
+    ]),
+];
+}
+
+const eps = 1e-5;
+
+function ensure($x)
+{
+    if (!$x) {
+        die(1);
+    }
+}
+
+function main()
+{
+    $input = getInput();
+    foreach ($input as $row) {
+        $prediction = kml_catboost_predict_vectors("catboost_tiny_1float_1hot_10trees", $row['float_features'], $row['cat_features']);
+        ensure(abs($prediction - $row['ans']) < eps);
+    }
+}
+main();

tests/python/tests/kml/catboost/php/unrelated_to_inference.php

@@ -0,0 +1,20 @@
+<?php
+
+function ensure($x) {
+    if (!$x) {
+        die(1);
+    }
+}
+
+function main() {
+    $model_name = "catboost_tiny_1float_1hot_10trees";
+
+    ensure(kml_model_exists($model_name));
+    ensure(["x", "y"] === kml_get_feature_names($model_name));
+    ensure("val_1" == kml_get_custom_property($model_name, "property_1"));
+    ensure("val_2, val_3" === kml_get_custom_property($model_name, "property_2"));
+    ensure(null === kml_get_custom_property($model_name, "property_absent"));
+    ensure(null === kml_xgboost_predict_matrix("absent_model", []));
+}
+
+main()

tests/python/tests/kml/test_kml.py

@@ -0,0 +1,36 @@
+import os
+
+from python.lib.testcase import KphpCompilerAutoTestCase
+
+
+class TestKML(KphpCompilerAutoTestCase):
+    def compile_and_check_success(self, php_script_path, kml_dir):
+        kml_dir = os.path.join(self.test_dir, kml_dir)
+        args = ["--kml-dir={}".format(kml_dir)]
+
+        once_runner = self.make_kphp_once_runner(php_script_path)
+
+        self.assertTrue(once_runner.compile_with_kphp({}))
+        self.assertTrue(once_runner.run_with_kphp(args=args), "Got KPHP runtime error")
+
+    def test_xgboost_tiny_ht_direct_int_keys_to_fvalue(self):
+        self.compile_and_check_success(php_script_path="xgboost/php/xgb_tiny_ht_direct_int_keys_to_fvalue.php",
+                                       kml_dir="xgboost/kml")
+        self.compile_and_check_success(php_script_path="xgboost/php/xgb_tiny_ht_direct_int_keys_to_fvalue.php",
+                                       kml_dir="xgboost")  # Traverse directories recursively
+
+    def test_xgb_tiny_ht_remap_str_keys_to_fvalue(self):
+        self.compile_and_check_success(php_script_path="xgboost/php/xgb_tiny_ht_remap_str_keys_to_fvalue.php",
+                                       kml_dir="xgboost/kml")
+
+    def test_catboost_tiny_1float_1hot_10trees(self):
+        self.compile_and_check_success(php_script_path="catboost/php/catboost_tiny_1float_1hot_10trees.php",
+                                       kml_dir="catboost/kml")
+
+    def test_catboost_multiclass_tutorial_ht_catnum(self):
+        self.compile_and_check_success(php_script_path="catboost/php/catboost_multiclass_tutorial_ht_catnum.php",
+                                       kml_dir="catboost/kml")
+
+    def test_unrelated_to_inference(self):
+        self.compile_and_check_success(php_script_path="catboost/php/unrelated_to_inference.php",
+                                       kml_dir="catboost/kml")

tests/python/tests/kml/xgboost/php/xgb_tiny_ht_direct_int_keys_to_fvalue.php

@@ -0,0 +1,36 @@
+<?php
+const eps = 1e-5;
+
+function ensure($x)
+{
+    if (!$x) {
+        die(1);
+    }
+}
+
+function main()
+{
+    $inputs = array(
+        array("0" => 1, "1" => 1),
+        array("0" => 2, "1" => 2),
+        array("0" => 100, "1" => 100),
+        array("0" => 4, "1" => -10),
+        array("0" => 5, "1" => -555),
+    );
+    $expected = array(
+        0.745589,
+        0.437018,
+        0.437018,
+        0.156840,
+        0.156840,
+    );
+
+    $ans = kml_xgboost_predict_matrix("xgb_tiny_ht_direct_int_keys_to_fvalue", $inputs);
+
+    $length = count($expected);
+    ensure($length == count($ans));
+    for ($i = 0; $i < $length; $i++) {
+        ensure(abs($expected[$i] - $ans[$i]) < eps);
+    }
+}
+main();

tests/python/tests/kml/xgboost/php/xgb_tiny_ht_remap_str_keys_to_fvalue.php

@@ -0,0 +1,36 @@
+<?php
+const eps = 1e-5;
+
+function ensure($x)
+{
+    if (!$x) {
+        die(1);
+    }
+}
+
+function main()
+{
+    $inputs = array(
+        array("height" => 1, "weight" => 1, "useless_feature" => 0.1),
+        array("height" => 2, "weight" => 2, "useless_feature" => 0.12),
+        array("height" => 100, "weight" => 100, "useless_feature" => 0.123),
+        array("height" => 4, "weight" => -10, "useless_feature" => 0.1234),
+        array("height" => 5, "weight" => -555, "useless_feature" => 0.12345),
+    );
+    $expected = array(
+        0.745589,
+        0.437018,
+        0.437018,
+        0.156840,
+        0.156840,
+    );
+
+    $ans = kml_xgboost_predict_matrix("xgb_tiny_ht_remap_str_keys_to_fvalue", $inputs);
+
+    $length = count($expected);
+    ensure($length == count($ans));
+    for ($i = 0; $i < $length; $i++) {
+        ensure(abs($expected[$i] - $ans[$i]) < eps);
+    }
+}
+main();