update median

paconvert/api_mapping.json

@@ -8494,7 +8494,7 @@
     }
   },
   "torch.median": {
-    "Matcher": "MedianMatcher",
+    "Matcher": "DoubleAssignMatcher",
     "paddle_api": "paddle.median",
     "args_list": [
       "input",
@@ -8506,6 +8506,9 @@
     "kwargs_change": {
       "input": "x",
       "dim": "axis"
+    },
+    "paddle_default_kwargs": {
+      "mode": "'min'"
     }
   },
   "torch.meshgrid": {
@@ -8711,7 +8714,7 @@
     }
   },
   "torch.nanmedian": {
-    "Matcher": "MedianMatcher",
+    "Matcher": "DoubleAssignMatcher",
     "paddle_api": "paddle.nanmedian",
     "args_list": [
       "input",
@@ -8723,6 +8726,9 @@
     "kwargs_change": {
       "input": "x",
       "dim": "axis"
+    },
+    "paddle_default_kwargs": {
+      "mode": "'min'"
     }
   },
   "torch.nanquantile": {

paconvert/api_matcher.py

@@ -3305,6 +3305,12 @@ def generate_code(self, kwargs):
             if k in kwargs:
                 kwargs[kwargs_change[k]] = kwargs.pop(k)
 
+        paddle_default_kwargs = self.api_mapping.get("paddle_default_kwargs", {})
+
+        for k in paddle_default_kwargs:
+            if k not in kwargs:
+                kwargs[k] = paddle_default_kwargs[k]
+
         if "out" in kwargs:
             out_v = kwargs.pop("out")
             API_TEMPLATE = textwrap.dedent(
@@ -3336,6 +3342,12 @@ def generate_code(self, kwargs):
                 else:
                     kwargs.pop(k)
 
+        paddle_default_kwargs = self.api_mapping.get("paddle_default_kwargs", {})
+
+        for k in paddle_default_kwargs:
+            if k not in kwargs:
+                kwargs[k] = paddle_default_kwargs[k]
+
         if "out" in kwargs:
             out_v = kwargs.pop("out")
             API_TEMPLATE = textwrap.dedent(
@@ -4328,11 +4340,3 @@ def generate_code(self, kwargs):
         else:
             code = "misidentify"
         return code
-
-
-class MedianMatcher(BaseMatcher):
-    def generate_code(self, kwargs):
-        if kwargs.get("dim", -1) != "-1":
-            kwargs["mode"] = "'min'"
-
-        return GenericMatcher.generate_code(self, kwargs)

tests/test_median.py

@@ -72,7 +72,7 @@ def test_case_4():
     )
 
 
-def _test_case_5():
+def test_case_5():
     pytorch_code = textwrap.dedent(
         """
         import torch
@@ -81,7 +81,7 @@ def _test_case_5():
         result = torch.median(input, dim=1, keepdim=True, out=out)
         """
     )
-    obj.run(pytorch_code, ["result"])
+    obj.run(pytorch_code, ["result", "out"])
 
 
 def test_case_6():
@@ -96,3 +96,14 @@ def test_case_6():
         pytorch_code,
         ["result"],
     )
+
+
+def test_case_7():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        input = torch.tensor([[1.4907, 1.0593, 1.5696], [1.4907, 1.0593, 1.5696]])
+        result = torch.median(input, dim=1, keepdim=True)
+        """
+    )
+    obj.run(pytorch_code, ["result"])

tests/test_nanmedian.py

@@ -44,8 +44,18 @@ def test_case_2():
     )
 
 
-# 会引发段错误，先屏蔽
-def _test_case_4():
+def test_case_3():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        input = torch.tensor([[1.4907, 1.0593, 1.5696], [1.4907, 1.0593, 1.5696]])
+        result = torch.nanmedian(input, 1, keepdim=True)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_4():
     pytorch_code = textwrap.dedent(
         """
         import torch
@@ -59,8 +69,7 @@ def _test_case_4():
     )
 
 
-# 会引发段错误，先屏蔽
-def _test_case_5():
+def test_case_5():
     pytorch_code = textwrap.dedent(
         """
         import torch
@@ -75,8 +84,7 @@ def _test_case_5():
     )
 
 
-# 会引发段错误，先屏蔽
-def _test_case_6():
+def test_case_6():
     pytorch_code = textwrap.dedent(
         """
         import torch