fix: [6880760974] normalize unstable win vtable output.

Author: Simon9997

test/cases/05-VirtualTables/test_vtable_join.py

@@ -197,6 +197,6 @@ def test_vtable_join(self):
         self.sqlFile = etool.curFile(__file__, f"in/{testCase}.in")
         self.ansFile = etool.curFile(__file__, f"ans/{testCase}.ans")
 
-        tdCom.compare_testcase_result(self.sqlFile, self.ansFile, testCase)
-
-
+        tdCom.compare_testcase_result(
+            self.sqlFile, self.ansFile, testCase, float_tolerance=2e-5
+        )

test/new_test_framework/utils/common.py

@@ -31,6 +31,7 @@
 from .constant import *
 from .epath import *
 from dataclasses import dataclass, field
+from decimal import Decimal, InvalidOperation
 from typing import List
 from datetime import datetime, timedelta
 import re
@@ -3039,7 +3040,119 @@ def generate_query_result(self, inputfile, test_case):
                 )
             return self.query_result_file
 
-    def compare_result_files(self, file1, file2):
+    def _get_numeric_compare_tolerance(self, token1, token2, float_tolerance):
+        if float_tolerance > 0.0:
+            return Decimal(str(float_tolerance))
+
+        def count_decimal_places(token):
+            mantissa = token.lower().split("e", 1)[0]
+            if "." not in mantissa:
+                return 0
+            return len(mantissa.split(".", 1)[1])
+
+        precision = max(count_decimal_places(token1), count_decimal_places(token2))
+        if precision <= 0:
+            return Decimal("0")
+        return Decimal(1).scaleb(-precision)
+
+    def _normalize_result_line_for_compare(self, line):
+        """Normalize CLI-only suffixes before answer/result file comparison.
+
+        Args:
+            line: A single line from an answer or result file.
+
+        Returns:
+            The normalized line with runtime-only Windows suffixes removed.
+        """
+
+        normalized = line.rstrip()
+        normalized = re.sub(r"\s*\([0-9]+\.[0-9]+s\)$", "", normalized)
+        normalized = re.sub(r"cost=[0-9]+\.[0-9]+\.\.[0-9]+\.[0-9]+", "", normalized)
+        normalized = re.sub(r"Planning Time: [0-9]+\.[0-9]+ ms", "", normalized)
+        normalized = re.sub(r"Execution Time: [0-9]+\.[0-9]+ ms", "", normalized)
+        normalized = re.sub(r"max_row_task=[0-9]+, ", "", normalized)
+        return normalized.rstrip()
+
+    def _compare_normalized_result_lines(self, file1, file2):
+        """Compare result files after stripping platform-specific CLI noise.
+
+        Args:
+            file1: Expected result file path.
+            file2: Actual result file path.
+
+        Returns:
+            True when the normalized result lines are identical.
+        """
+
+        with open(file1, "r", encoding="utf-8", errors="ignore") as f1:
+            lines1 = f1.read().splitlines()
+        with open(file2, "r", encoding="utf-8", errors="ignore") as f2:
+            lines2 = f2.read().splitlines()
+
+        if len(lines1) != len(lines2):
+            return False
+
+        for line1, line2 in zip(lines1, lines2):
+            if self._normalize_result_line_for_compare(
+                line1
+            ) != self._normalize_result_line_for_compare(line2):
+                return False
+
+        return True
+
+    def _compare_file_lines_with_float_tolerance(self, file1, file2, float_tolerance):
+        number_pattern = re.compile(r"[-+]?(?:\d+\.\d+|\d+|\.\d+)(?:[eE][-+]?\d+)?")
+
+        with open(file1, "r", encoding="utf-8", errors="ignore") as f1:
+            lines1 = f1.read().splitlines()
+        with open(file2, "r", encoding="utf-8", errors="ignore") as f2:
+            lines2 = f2.read().splitlines()
+
+        if len(lines1) != len(lines2):
+            return False
+
+        for line1, line2 in zip(lines1, lines2):
+            line1 = self._normalize_result_line_for_compare(line1)
+            line2 = self._normalize_result_line_for_compare(line2)
+
+            if line1 == line2:
+                continue
+
+            matches1 = list(number_pattern.finditer(line1))
+            matches2 = list(number_pattern.finditer(line2))
+            if len(matches1) != len(matches2):
+                return False
+
+            cursor1 = 0
+            cursor2 = 0
+            for match1, match2 in zip(matches1, matches2):
+                if line1[cursor1:match1.start()] != line2[cursor2:match2.start()]:
+                    return False
+
+                token1 = match1.group(0)
+                token2 = match2.group(0)
+                try:
+                    value1 = Decimal(token1)
+                    value2 = Decimal(token2)
+                except InvalidOperation:
+                    if token1 != token2:
+                        return False
+                else:
+                    tolerance = self._get_numeric_compare_tolerance(
+                        token1, token2, float_tolerance
+                    )
+                    if abs(value1 - value2) > tolerance:
+                        return False
+
+                cursor1 = match1.end()
+                cursor2 = match2.end()
+
+            if line1[cursor1:] != line2[cursor2:]:
+                return False
+
+        return True
+
+    def compare_result_files(self, file1, file2, float_tolerance=0.0):
         try:
             # use subprocess.run to execute  diff/fc commands
             # print(file1, file2)
@@ -3067,6 +3180,19 @@ def compare_result_files(self, file1, file2):
                     return True
             # if result is not empty, print the differences and files name. Otherwise, the files are identical.
             if result.returncode != 0:
+                if self._compare_normalized_result_lines(file1, file2):
+                    tdLog.info("Result files matched after output normalization.")
+                    return True
+                if platform.system().lower() == "windows" and self._compare_file_lines_with_float_tolerance(
+                    file1, file2, float_tolerance
+                ):
+                    tdLog.info(
+                        "Result files matched after Windows output normalization."
+                        if float_tolerance <= 0.0
+                        else "Result files matched after Windows output normalization "
+                        f"with float tolerance {float_tolerance}."
+                    )
+                    return True
                 tdLog.info(f"{cmd} result.returncode: {result.returncode}")
                 tdLog.info(f"{cmd} result.stdout: {result.stdout}")
                 tdLog.info(f"{cmd} result.stderr: {result.stderr}")
@@ -3087,9 +3213,13 @@ def compare_result_files(self, file1, file2):
         except Exception as e:
             tdLog.debug(f"An error occurred: {e}")
 
-    def compare_query_with_result_file(self, idx, sql, resultFile, test_case):
+    def compare_query_with_result_file(
+        self, idx, sql, resultFile, test_case, float_tolerance=0.0
+    ):
         self.generate_query_result_file(test_case, idx, sql)
-        if self.compare_result_files(resultFile, self.query_result_file):
+        if self.compare_result_files(
+            resultFile, self.query_result_file, float_tolerance=float_tolerance
+        ):
             tdLog.info("Test passed: Result files are identical.")
             # os.system(f"rm -f {self.query_result_file}")
         else:
@@ -3098,10 +3228,14 @@ def compare_query_with_result_file(self, idx, sql, resultFile, test_case):
                 f"{caller.lineno}(line:{caller.lineno}) failed: expect_file:{resultFile}  != reult_file:{self.query_result_file} "
             )
 
-    def compare_testcase_result(self, inputfile, expected_file, test_case):
+    def compare_testcase_result(
+        self, inputfile, expected_file, test_case, float_tolerance=0.0
+    ):
         test_reulst_file = self.generate_query_result(inputfile, test_case)
 
-        if self.compare_result_files(expected_file, test_reulst_file):
+        if self.compare_result_files(
+            expected_file, test_reulst_file, float_tolerance=float_tolerance
+        ):
             tdLog.info("Test passed: Result files are identical.")
             os.system(f"rm -f {test_reulst_file}")
         else: