emmit type name when failing to deserialize json

Author: rok

cpp/src/arrow/extension/fixed_shape_tensor.cc

@@ -122,7 +122,8 @@ Result<std::shared_ptr<DataType>> FixedShapeTensorType::Deserialize(
   std::vector<int64_t> shape;
   for (const auto& x : document["shape"].GetArray()) {
     if (!x.IsInt64()) {
-      return Status::Invalid("shape must contain integers");
+      return Status::Invalid("shape must contain integers, got ",
+                             internal::JsonTypeName(x));
     }
     shape.emplace_back(x.GetInt64());
   }
@@ -131,11 +132,13 @@ Result<std::shared_ptr<DataType>> FixedShapeTensorType::Deserialize(
   if (document.HasMember("permutation")) {
     const auto& json_permutation = document["permutation"];
     if (!json_permutation.IsArray()) {
-      return Status::Invalid("permutation must be an array");
+      return Status::Invalid("permutation must be an array, got ",
+                             internal::JsonTypeName(json_permutation));
     }
     for (const auto& x : json_permutation.GetArray()) {
       if (!x.IsInt64()) {
-        return Status::Invalid("permutation must contain integers");
+        return Status::Invalid("permutation must contain integers, got ",
+                               internal::JsonTypeName(x));
       }
       permutation.emplace_back(x.GetInt64());
     }
@@ -148,11 +151,13 @@ Result<std::shared_ptr<DataType>> FixedShapeTensorType::Deserialize(
   if (document.HasMember("dim_names")) {
     const auto& json_dim_names = document["dim_names"];
     if (!json_dim_names.IsArray()) {
-      return Status::Invalid("dim_names must be an array");
+      return Status::Invalid("dim_names must be an array, got ",
+                             internal::JsonTypeName(json_dim_names));
     }
     for (const auto& x : json_dim_names.GetArray()) {
       if (!x.IsString()) {
-        return Status::Invalid("dim_names must contain strings");
+        return Status::Invalid("dim_names must contain strings, got ",
+                               internal::JsonTypeName(x));
       }
       dim_names.emplace_back(x.GetString());
     }

cpp/src/arrow/extension/tensor_internal.cc

@@ -30,6 +30,20 @@
 
 namespace arrow::internal {
 
+namespace {
+
+// Names indexed by rapidjson::Type enum value:
+// kNullType=0, kFalseType=1, kTrueType=2, kObjectType=3,
+// kArrayType=4, kStringType=5, kNumberType=6.
+constexpr const char* kJsonTypeNames[] = {"Null",  "False",  "True",  "Object",
+                                          "Array", "String", "Number"};
+
+}  // namespace
+
+const char* JsonTypeName(const ::arrow::rapidjson::Value& v) {
+  return kJsonTypeNames[v.GetType()];
+}
+
 Result<int64_t> ComputeShapeProduct(std::span<const int64_t> shape) {
   int64_t product = 1;
   for (const auto dim : shape) {

cpp/src/arrow/extension/tensor_internal.h

@@ -21,11 +21,18 @@
 #include <span>
 #include <vector>
 
+#include "arrow/json/rapidjson_defs.h"  // IWYU pragma: keep
 #include "arrow/result.h"
 #include "arrow/type_fwd.h"
 
+#include <rapidjson/document.h>
+
 namespace arrow::internal {
 
+/// \brief Return the name of a RapidJSON value's type (e.g., "Null", "Array", "Number").
+ARROW_EXPORT
+const char* JsonTypeName(const ::arrow::rapidjson::Value& v);
+
 /// \brief Compute the product of the given shape dimensions.
 ///
 /// Returns Status::Invalid if the product would overflow int64_t.

cpp/src/arrow/extension/variable_shape_tensor.cc

@@ -159,12 +159,14 @@ Result<std::shared_ptr<DataType>> VariableShapeTensorType::Deserialize(
   if (document.HasMember("permutation")) {
     const auto& json_permutation = document["permutation"];
     if (!json_permutation.IsArray()) {
-      return Status::Invalid("permutation must be an array");
+      return Status::Invalid("permutation must be an array, got ",
+                             internal::JsonTypeName(json_permutation));
     }
     permutation.reserve(ndim);
     for (const auto& x : json_permutation.GetArray()) {
       if (!x.IsInt64()) {
-        return Status::Invalid("permutation must contain integers");
+        return Status::Invalid("permutation must contain integers, got ",
+                               internal::JsonTypeName(x));
       }
       permutation.emplace_back(x.GetInt64());
     }
@@ -174,12 +176,14 @@ Result<std::shared_ptr<DataType>> VariableShapeTensorType::Deserialize(
   if (document.HasMember("dim_names")) {
     const auto& json_dim_names = document["dim_names"];
     if (!json_dim_names.IsArray()) {
-      return Status::Invalid("dim_names must be an array");
+      return Status::Invalid("dim_names must be an array, got ",
+                             internal::JsonTypeName(json_dim_names));
     }
     dim_names.reserve(ndim);
     for (const auto& x : json_dim_names.GetArray()) {
       if (!x.IsString()) {
-        return Status::Invalid("dim_names must contain strings");
+        return Status::Invalid("dim_names must contain strings, got ",
+                               internal::JsonTypeName(x));
       }
       dim_names.emplace_back(x.GetString());
     }
@@ -189,7 +193,8 @@ Result<std::shared_ptr<DataType>> VariableShapeTensorType::Deserialize(
   if (document.HasMember("uniform_shape")) {
     const auto& json_uniform_shape = document["uniform_shape"];
     if (!json_uniform_shape.IsArray()) {
-      return Status::Invalid("uniform_shape must be an array");
+      return Status::Invalid("uniform_shape must be an array, got ",
+                             internal::JsonTypeName(json_uniform_shape));
     }
     uniform_shape.reserve(ndim);
     for (const auto& x : json_uniform_shape.GetArray()) {
@@ -198,7 +203,8 @@ Result<std::shared_ptr<DataType>> VariableShapeTensorType::Deserialize(
       } else if (x.IsInt64()) {
         uniform_shape.emplace_back(x.GetInt64());
       } else {
-        return Status::Invalid("uniform_shape must contain integers or nulls");
+        return Status::Invalid("uniform_shape must contain integers or nulls, got ",
+                               internal::JsonTypeName(x));
       }
     }
   }