refactor: Added raise from clause to exceptions for improved tracebacks

ivy/__init__.py

@@ -443,8 +443,8 @@ def unknown_shape(rank=None, **kwargs):
     def with_rank(self, rank):
         try:
             return self.merge_with(self.unknown_shape(rank=rank))
-        except ValueError:
-            raise ValueError(f"Shape {self} must have rank {rank}")
+        except ValueError as e:
+            raise ValueError(f"Shape {self} must have rank {rank}") from e
 
     def with_rank_at_least(self, rank):
         if self.rank is not None and self.rank < rank:

ivy/functional/backends/jax/data_type.py

@@ -133,7 +133,7 @@ def broadcast_arrays(*arrays: JaxArray) -> List[JaxArray]:
     try:
         return jnp.broadcast_arrays(*arrays)
     except ValueError as e:
-        raise ivy.utils.exceptions.IvyBroadcastShapeError(e)
+        raise ivy.utils.exceptions.IvyBroadcastShapeError(e) from e
 
 
 def broadcast_to(

ivy/functional/backends/jax/experimental/manipulation.py

@@ -442,11 +442,11 @@ def take(
         if ivy.exists(axis):
             try:
                 x_shape = x.shape[axis]
-            except Exception:
+            except Exception as e:
                 raise ValueError(
                     f"axis {axis} is out of bounds for array of dimension"
                     f" {len(x.shape)}"
-                )
+                ) from e
         else:
             x_shape = jnp.prod(x.shape)
 

ivy/functional/backends/jax/manipulation.py

@@ -39,7 +39,7 @@ def concat(
     try:
         return jnp.concatenate(xs, axis)
     except ValueError as error:
-        raise ivy.utils.exceptions.IvyIndexError(error)
+        raise ivy.utils.exceptions.IvyIndexError(error) from error
 
 
 def expand_dims(
@@ -54,7 +54,7 @@ def expand_dims(
         ret = jnp.expand_dims(x, axis)
         return ret
     except ValueError as error:
-        raise ivy.utils.exceptions.IvyIndexError(error)
+        raise ivy.utils.exceptions.IvyIndexError(error) from error
 
 
 def flip(
@@ -143,7 +143,7 @@ def stack(
     try:
         return jnp.stack(arrays, axis=axis)
     except ValueError as error:
-        raise ivy.utils.exceptions.IvyIndexError(error)
+        raise ivy.utils.exceptions.IvyIndexError(error) from error
 
 
 # Extra #

ivy/functional/backends/jax/utility.py

@@ -19,7 +19,7 @@ def all(
     try:
         return jnp.all(x, axis, keepdims=keepdims)
     except ValueError as error:
-        raise ivy.utils.exceptions.IvyIndexError(error)
+        raise ivy.utils.exceptions.IvyIndexError(error) from error
 
 
 def any(
@@ -34,4 +34,4 @@ def any(
     try:
         return jnp.any(x, axis, keepdims=keepdims, out=out)
     except ValueError as error:
-        raise ivy.utils.exceptions.IvyIndexError(error)
+        raise ivy.utils.exceptions.IvyIndexError(error) from error

ivy/functional/backends/numpy/data_type.py

@@ -130,7 +130,7 @@ def broadcast_arrays(*arrays: np.ndarray) -> List[np.ndarray]:
     try:
         return np.broadcast_arrays(*arrays)
     except ValueError as e:
-        raise ivy.utils.exceptions.IvyBroadcastShapeError(e)
+        raise ivy.utils.exceptions.IvyBroadcastShapeError(e) from e
 
 
 @with_unsupported_dtypes({"1.26.3 and below": ("complex",)}, backend_version)

ivy/functional/backends/numpy/utility.py

@@ -17,7 +17,7 @@ def all(
     try:
         return np.asarray(np.all(x, axis=axis, keepdims=keepdims, out=out))
     except np.AxisError as error:
-        raise ivy.utils.exceptions.IvyIndexError(error)
+        raise ivy.utils.exceptions.IvyIndexError(error) from error
 
 
 all.support_native_out = True

ivy/functional/backends/paddle/experimental/manipulation.py

@@ -774,7 +774,7 @@ def take(
         if ivy.exists(axis):
             try:
                 x_shape = x.shape[axis]
-            except Exception:
+            except Exception as e:
                 rank = len(x.shape)
                 raise IndexError(
                     "(OutOfRange) Attr(axis) is out of range, "
@@ -784,7 +784,7 @@ def take(
                     "(0 - input_dim.size()) == true, "
                     "but received axis < input_dim.size() && axis >= "
                     "(0 - input_dim.size()):0 != true:1.]"
-                )
+                ) from e
         else:
             x_shape = paddle.prod(paddle.to_tensor(x.shape))
 

ivy/functional/backends/paddle/experimental/norms.py

@@ -55,11 +55,11 @@ def batch_norm(
             if data_format[-1] == "C"
             else data_formats[0:4][x.ndim - 2]
         )
-    except IndexError:
+    except IndexError as e:
         raise IndexError(
             "data_format must be one of 'NC', 'NCL', 'NCHW', 'NCDHW', 'NLC', 'NHWC',"
             f" 'NDHWC' but receive {data_format}"
-        )
+        ) from e
 
     with ivy.ArrayMode(False):
         if training:

ivy/functional/backends/tensorflow/data_type.py

@@ -117,15 +117,15 @@ def broadcast_arrays(
         try:
             desired_shape = tf.broadcast_dynamic_shape(arrays[0].shape, arrays[1].shape)
         except tf.errors.InvalidArgumentError as e:
-            raise ivy.utils.exceptions.IvyBroadcastShapeError(e)
+            raise ivy.utils.exceptions.IvyBroadcastShapeError(e) from e
         if len(arrays) > 2:
             for i in range(2, len(arrays)):
                 try:
                     desired_shape = tf.broadcast_dynamic_shape(
                         desired_shape, arrays[i].shape
                     )
                 except tf.errors.InvalidArgumentError as e:
-                    raise ivy.utils.exceptions.IvyBroadcastShapeError(e)
+                    raise ivy.utils.exceptions.IvyBroadcastShapeError(e) from e
     else:
         return [arrays[0]]
     result = []

ivy/functional/backends/tensorflow/manipulation.py

@@ -66,7 +66,7 @@ def expand_dims(
         ret = tf.reshape(x, shape=out_shape)
         return ret
     except (tf.errors.InvalidArgumentError, np.AxisError) as error:
-        raise ivy.utils.exceptions.IvyIndexError(error)
+        raise ivy.utils.exceptions.IvyIndexError(error) from error
 
 
 def flip(
@@ -196,7 +196,7 @@ def stack(
     try:
         return tf.experimental.numpy.stack(arrays, axis)
     except ValueError as e:
-        raise ivy.utils.exceptions.IvyIndexError(e)
+        raise ivy.utils.exceptions.IvyIndexError(e) from e
 
 
 # Extra #

ivy/functional/backends/tensorflow/utility.py

@@ -24,7 +24,7 @@ def all(
     try:
         return tf.reduce_all(tf.cast(x, tf.bool), axis=axis, keepdims=keepdims)
     except tf.errors.InvalidArgumentError as e:
-        raise ivy.utils.exceptions.IvyIndexError(e)
+        raise ivy.utils.exceptions.IvyIndexError(e) from e
 
 
 def any(
@@ -44,4 +44,4 @@ def any(
     try:
         return tf.reduce_any(tf.cast(x, tf.bool), axis=axis, keepdims=keepdims)
     except tf.errors.InvalidArgumentError as e:
-        raise ivy.utils.exceptions.IvyIndexError(e)
+        raise ivy.utils.exceptions.IvyIndexError(e) from e

ivy/functional/backends/torch/data_type.py

@@ -91,7 +91,7 @@ def broadcast_arrays(*arrays: torch.Tensor) -> List[torch.Tensor]:
     try:
         return list(torch.broadcast_tensors(*arrays))
     except RuntimeError as e:
-        raise ivy.utils.exceptions.IvyBroadcastShapeError(e)
+        raise ivy.utils.exceptions.IvyBroadcastShapeError(e) from e
 
 
 def broadcast_to(

ivy/functional/backends/torch/experimental/manipulation.py

@@ -550,13 +550,13 @@ def take(
         if ivy.exists(axis):
             try:
                 x_shape = x.shape[axis]
-            except Exception:
+            except Exception as e:
                 rank = len(x.shape)
                 raise IndexError(
                     "IndexError: Dimension out of range"
                     f"(expected to be in range of[-{rank}, {rank-1}]"
                     f", but got {axis})"
-                )
+                ) from e
         else:
             x_shape = torch.prod(torch.tensor(x.shape))
 

ivy/functional/backends/torch/general.py

@@ -387,10 +387,10 @@ def scatter_flat(
     if torch_scatter is None:
         try:
             import torch_scatter as torch_scatter
-        except ImportError:
+        except ImportError as e:
             raise ivy.utils.exceptions.IvyException(
                 "Unable to import torch_scatter, verify this is correctly installed."
-            )
+            ) from e
     if reduction == "replace":
         output[indices.type(torch.int64)] = updates
         res = output
@@ -478,10 +478,10 @@ def scatter_nd(
     if torch_scatter is None:
         try:
             import torch_scatter as torch_scatter
-        except ImportError:
+        except ImportError as e:
             raise ivy.utils.exceptions.IvyException(
                 "Unable to import torch_scatter, verify this is correctly installed."
-            )
+            ) from e
     if reduction == "replace":
         flat_output[flat_indices_for_flat] = flat_updates
         flat_scatter = flat_output

ivy/functional/frontends/jax/numpy/__init__.py

@@ -407,8 +407,10 @@ def promote_types_jax(
     """
     try:
         ret = jax_promotion_table[(ivy.as_ivy_dtype(type1), ivy.as_ivy_dtype(type2))]
-    except KeyError:
-        raise ivy.utils.exceptions.IvyException("these dtypes are not type promotable")
+    except KeyError as e:
+        raise ivy.utils.exceptions.IvyException(
+            "these dtypes are not type promotable"
+        ) from e
     return ret
 
 

ivy/functional/frontends/mxnet/numpy/__init__.py

@@ -121,8 +121,10 @@ def promote_types_mxnet(
     """
     try:
         ret = mxnet_promotion_table[(ivy.as_ivy_dtype(type1), ivy.as_ivy_dtype(type2))]
-    except KeyError:
-        raise ivy.utils.exceptions.IvyException("these dtypes are not type promotable")
+    except KeyError as e:
+        raise ivy.utils.exceptions.IvyException(
+            "these dtypes are not type promotable"
+        ) from e
     return ret
 
 

ivy/functional/frontends/numpy/__init__.py

@@ -425,8 +425,10 @@ def promote_numpy_dtypes(
     type1, type2 = ivy.as_ivy_dtype(type1), ivy.as_ivy_dtype(type2)
     try:
         return numpy_promotion_table[(type1, type2)]
-    except KeyError:
-        raise ivy.utils.exceptions.IvyException("these dtypes are not type promotable")
+    except KeyError as e:
+        raise ivy.utils.exceptions.IvyException(
+            "these dtypes are not type promotable"
+        ) from e
 
 
 @handle_exceptions

ivy/functional/frontends/numpy/data_type_routines/creating_data_types.py

@@ -19,11 +19,11 @@ def __repr__(self):
     def __ge__(self, other):
         try:
             other = dtype(other)
-        except TypeError:
+        except TypeError as e:
             raise ivy.utils.exceptions.IvyException(
                 "Attempted to compare a dtype with something which"
                 "couldn't be interpreted as a dtype"
-            )
+            ) from e
 
         return self == np_frontend.promote_numpy_dtypes(
             self._ivy_dtype, other._ivy_dtype
@@ -32,22 +32,22 @@ def __ge__(self, other):
     def __gt__(self, other):
         try:
             other = dtype(other)
-        except TypeError:
+        except TypeError as e:
             raise ivy.utils.exceptions.IvyException(
                 "Attempted to compare a dtype with something which"
                 "couldn't be interpreted as a dtype"
-            )
+            ) from e
 
         return self >= other and self != other
 
     def __lt__(self, other):
         try:
             other = dtype(other)
-        except TypeError:
+        except TypeError as e:
             raise ivy.utils.exceptions.IvyException(
                 "Attempted to compare a dtype with something which"
                 "couldn't be interpreted as a dtype"
-            )
+            ) from e
 
         return self != np_frontend.promote_numpy_dtypes(
             self._ivy_dtype, other._ivy_dtype
@@ -56,11 +56,11 @@ def __lt__(self, other):
     def __le__(self, other):
         try:
             other = dtype(other)
-        except TypeError:
+        except TypeError as e:
             raise ivy.utils.exceptions.IvyException(
                 "Attempted to compare a dtype with something which"
                 "couldn't be interpreted as a dtype"
-            )
+            ) from e
 
         return self < other or self == other
 

ivy/functional/frontends/onnx/__init__.py

@@ -208,8 +208,10 @@ def promote_types_onnx(
     """
     try:
         ret = onnx_promotion_table[(ivy.as_ivy_dtype(type1), ivy.as_ivy_dtype(type2))]
-    except KeyError:
-        raise ivy.utils.exceptions.IvyException("these dtypes are not type promotable")
+    except KeyError as e:
+        raise ivy.utils.exceptions.IvyException(
+            "these dtypes are not type promotable"
+        ) from e
     return ret
 
 

ivy/functional/frontends/paddle/__init__.py

@@ -173,8 +173,10 @@ def promote_types_paddle(
     """
     try:
         ret = paddle_promotion_table[(ivy.as_ivy_dtype(type1), ivy.as_ivy_dtype(type2))]
-    except KeyError:
-        raise ivy.utils.exceptions.IvyException("these dtypes are not type promotable")
+    except KeyError as e:
+        raise ivy.utils.exceptions.IvyException(
+            "these dtypes are not type promotable"
+        ) from e
     return ret
 
 

ivy/functional/frontends/torch/__init__.py

@@ -210,8 +210,10 @@ def promote_types_torch(
         ret = torch_frontend.torch_promotion_table[
             (ivy.as_ivy_dtype(type1), ivy.as_ivy_dtype(type2))
         ]
-    except KeyError:
-        raise ivy.utils.exceptions.IvyException("these dtypes are not type promotable")
+    except KeyError as e:
+        raise ivy.utils.exceptions.IvyException(
+            "these dtypes are not type promotable"
+        ) from e
     return ret
 
 

ivy/functional/frontends/torch/linalg.py

@@ -23,7 +23,7 @@ def cholesky_ex(input, *, upper=False, check_errors=False, out=None):
         return matrix, info
     except RuntimeError as e:
         if check_errors:
-            raise RuntimeError(e)
+            raise RuntimeError(e) from e
         else:
             matrix = input * math.nan
             info = ivy.ones(input.shape[:-2], dtype=ivy.int32)
@@ -292,7 +292,7 @@ def solve_ex(A, B, *, left=True, check_errors=False, out=None):
         return result, info
     except RuntimeError as e:
         if check_errors:
-            raise RuntimeError(e)
+            raise RuntimeError(e) from e
         else:
             result = A * math.nan
             info = ivy.ones(A.shape[:-2], dtype=ivy.int32)