feat: respect iox::column_type::field metadata when mapping query

[NguyenHoangSon96]

Closes #

Proposed Changes

Response from the query will now respect the iox::column_type::field metadata

Checklist

• CHANGELOG.md updated
• Rebased/mergeable
• A test has been added if appropriate
• Tests pass
• Commit messages are conventional
• Sign CLA (if not already signed)

[codecov]

Codecov Report

Attention: Patch coverage is 90.38462% with 10 lines in your changes missing coverage. Please review.

Project coverage is 89.71%. Comparing base (d098ff7) to head (fcca2df).
Report is 3 commits behind head on main.

Files with missing lines	Patch %	Lines
.../v3/client/internal/VectorSchemaRootConverter.java	88.70%	1 Missing and 6 partials ⚠️
...a/com/influxdb/v3/client/internal/TypeCasting.java	81.81%	0 Missing and 2 partials ⚠️
...fluxdb/v3/client/internal/NanosecondConverter.java	96.15%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #200      +/-   ##
==========================================
+ Coverage   89.59%   89.71%   +0.12%     
==========================================
  Files          17       18       +1     
  Lines         903      963      +60     
  Branches      137      150      +13     
==========================================
+ Hits          809      864      +55     
- Misses         38       39       +1     
- Partials       56       60       +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[bednar]

The VectorSchemaRootConverter has two similar methods: toPointValues and setFieldWithMetaType.

What do you think about reusing the inner part of getArrayObjectFromVectorSchemaRoot in toPointValues? We should prepare a helper method to obtain the correctly typed value for each column and reuse this in both methods.

[NguyenHoangSon96]

The VectorSchemaRootConverter has two similar methods: toPointValues and setFieldWithMetaType.

What do you think about reusing the inner part of getArrayObjectFromVectorSchemaRoot in toPointValues? We should prepare a helper method to obtain the correctly typed value for each column and reuse this in both methods.

So with this suggestion, we will not need setFieldWithMetaType, right?
I still don't understand how we can make a function that can be used in both toPointValues and getArrayObjectFromVectorSchemaRoot because one set data for an array, and one set data for a PointValues 🤔

[bednar]

The VectorSchemaRootConverter has two similar methods: toPointValues and setFieldWithMetaType.
What do you think about reusing the inner part of getArrayObjectFromVectorSchemaRoot in toPointValues? We should prepare a helper method to obtain the correctly typed value for each column and reuse this in both methods.

So with this suggestion, we will not need setFieldWithMetaType, right? I still don't understand how we can make a function that can be used in both toPointValues and getArrayObjectFromVectorSchemaRoot because one set data for an array, and one set data for a PointValues 🤔

We can have something like (not production ready, just quick snippet ;)):

    private static Object getMappedValue(String valueType, String metaType, Object value, String fieldName, Field field) {
        if ("field".equals(valueType)) {
            switch (metaType) {
                case "iox::column_type::field::integer":
                case "iox::column_type::field::uinteger":
                    if (value instanceof Number) {
                        return TypeCasting.toLongValue(value);
                    }
                case "iox::column_type::field::float":
                    if (value instanceof Number) {
                        return TypeCasting.toDoubleValue(value);
                    }
                case "iox::column_type::field::string":
                    if (value instanceof Text || value instanceof String) {
                        return TypeCasting.toStringValue(value);
                    }
                case "iox::column_type::field::boolean":
                    if (value instanceof Boolean) {
                        return TypeCasting.toBoolValue(value);
                    }
                default:
                    return value;
            }
        } else if ("timestamp".equals(valueType) || Objects.equals(fieldName, "time")) {
            return NanosecondConverter.getTimestampNano(value, field);
        } else if ("tag".equals(valueType)) {
            return TypeCasting.toStringValue(value);
        } else {
            return value;
        }
    }

and this can be reused:

Object mappedValue = getMappedValue(valueType, metaType, value, fieldName, field);
row.add(mappedValue);

and with little refactoring also in toPointValues.

[NguyenHoangSon96]

The VectorSchemaRootConverter has two similar methods: toPointValues and setFieldWithMetaType.
What do you think about reusing the inner part of getArrayObjectFromVectorSchemaRoot in toPointValues? We should prepare a helper method to obtain the correctly typed value for each column and reuse this in both methods.

So with this suggestion, we will not need setFieldWithMetaType, right? I still don't understand how we can make a function that can be used in both toPointValues and getArrayObjectFromVectorSchemaRoot because one set data for an array, and one set data for a PointValues 🤔

We can have something like (not production ready, just quick snippet ;)):

    private static Object getMappedValue(String valueType, String metaType, Object value, String fieldName, Field field) {
        if ("field".equals(valueType)) {
            switch (metaType) {
                case "iox::column_type::field::integer":
                case "iox::column_type::field::uinteger":
                    if (value instanceof Number) {
                        return TypeCasting.toLongValue(value);
                    }
                case "iox::column_type::field::float":
                    if (value instanceof Number) {
                        return TypeCasting.toDoubleValue(value);
                    }
                case "iox::column_type::field::string":
                    if (value instanceof Text || value instanceof String) {
                        return TypeCasting.toStringValue(value);
                    }
                case "iox::column_type::field::boolean":
                    if (value instanceof Boolean) {
                        return TypeCasting.toBoolValue(value);
                    }
                default:
                    return value;
            }
        } else if ("timestamp".equals(valueType) || Objects.equals(fieldName, "time")) {
            return NanosecondConverter.getTimestampNano(value, field);
        } else if ("tag".equals(valueType)) {
            return TypeCasting.toStringValue(value);
        } else {
            return value;
        }
    }

and this can be reused:

Object mappedValue = getMappedValue(valueType, metaType, value, fieldName, field);
row.add(mappedValue);

and with little refactoring also in toPointValues.

Got it, thank you for your detail snipped Jakub

[bednar]

@NguyenHoangSon96, thanks for the PR! We’re almost done; there are just a few minor requirements before we can merge it.

[bednar]

@NguyenHoangSon96, thanks again for the PR 👍. There’s one last small requirement, just to make future maintenance easier:

[bednar]

LGTM 🚀