feat: auto deduce content type when requesting data

[Kybxd]

• close Data: handle any type gracefully based on "Content-Type" of headers #48

[codecov]

Codecov Report

❌ Patch coverage is 90.90909% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.05%. Comparing base (cbe3a6b) to head (4e17b20).

Files with missing lines	Patch %	Lines
request.go	90.90%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #49      +/-   ##
==========================================
+ Coverage   80.19%   83.05%   +2.85%     
==========================================
  Files           7        7              
  Lines         409      360      -49     
==========================================
- Hits          328      299      -29     
+ Misses         59       38      -21     
- Partials       22       23       +1     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[wenchy]

Add more details for deducing different content types based on data types:

1. "application/json": struct, slice(except []byte), and map
2. auto deduce by [http.DetectContentType]: []byte, io.Reader
3. "text/plain": others

[Kybxd]

done

[wenchy]

If the data type is io.Reader, we should also process it well. just like: https://github.com/go-resty/resty/blob/v3/middleware.go#L428C7-L428C16

[Kybxd]

Done. We check if the input data implements io.Reader first before checking its reflect.Kind.

[wenchy]

Code review

Found 2 issues:

1. *[]byte is JSON/base64-marshaled instead of being treated as raw bytes. reflect.Indirect dereferences a *[]byte into a reflect.Slice, so the Slice case is entered, but the []byte type assertion is performed on the original (still *[]byte) data value and fails. The code then falls through to json.Marshal(data), which base64-encodes the bytes and returns application/json — contradicting the documented []byte behavior (use http.DetectContentType). The test suite has no *[]byte case, so this is uncaught. The assertion should be against the dereferenced value.

requests/request.go

Lines 189 to 199 in c6790f3

	}
	bodyValue := reflect.Indirect(reflect.ValueOf(data))
	switch bodyValue.Kind() {
	case reflect.Struct, reflect.Map, reflect.Slice:
	// check slice here to differentiate between any slice vs byte slice
	if body, ok := data.([]byte); ok {
	return http.DetectContentType(body), body, nil
	} else {
	body, err := json.Marshal(data)
	return jsonContentType, body, err
	}

2. Passing a typed nil pointer (e.g. Data((*MyStruct)(nil))) panics. A typed nil pointer assigned to any is a non-nil interface, so opts.Data != nil passes and deduceContentTypeAndBody runs. reflect.Indirect on a nil pointer returns an invalid zero reflect.Value (Kind() == reflect.Invalid), so the switch falls through to default, where bodyValue.Interface() panics with reflect: call of reflect.Value.Interface on zero Value. The pre-PR code used fmt.Sprintf("%v", opts.Data) which safely returned <nil>. Add a !bodyValue.IsValid() guard (or a nil check) before reaching bodyValue.Interface().

requests/request.go

Lines 189 to 202 in c6790f3

	}
	bodyValue := reflect.Indirect(reflect.ValueOf(data))
	switch bodyValue.Kind() {
	case reflect.Struct, reflect.Map, reflect.Slice:
	// check slice here to differentiate between any slice vs byte slice
	if body, ok := data.([]byte); ok {
	return http.DetectContentType(body), body, nil
	} else {
	body, err := json.Marshal(data)
	return jsonContentType, body, err
	}
	default:
	return plainTextType, fmt.Appendf(nil, "%v", bodyValue.Interface()), nil
	}

🤖 Generated with Claude Code

_{- If this code review was useful, please react with 👍. Otherwise, react with 👎.}