Protobuf Text to JSON Converter

TextProto is a human-readable text representation of Protocol Buffer messages, primarily used for configuration files, test fixtures, and debugging output. Unlike the binary wire format used for transport, TextProto uses field names and is readable without a schema: field_name: value for scalars, and field_name { ... } for nested messages. It is not a standard serialisation format for production use but is common in Google’s internal tooling and TensorFlow config files.

Protobuf binary format uses field numbers (not names) and varint encoding to produce compact, fast-to-parse messages. A JSON equivalent of a Protobuf message is typically 3–5x larger. However, Protobuf binary is not self-describing — you need the .proto schema to decode it. JSON is self-describing and human-readable. Protobuf is preferred for high-throughput internal APIs; JSON is preferred for public APIs and configuration.

Common scenarios include: migrating configuration files from proto-based systems (TensorFlow, Bazel) to JSON-based systems, debugging Protobuf messages by converting to readable JSON, writing test fixtures in JSON for systems that consume Protobuf, and inspecting ML model configurations stored in TextProto format. The conversion is lossy for types with no JSON equivalent (like Protobuf Any or raw bytes), so review the output carefully.

Without the .proto schema, a converter cannot know field types (int32 vs int64 vs string) or distinguish between optional, required, and repeated fields from structure alone. Repeated fields with a single value look identical to optional fields. Enums appear as integers without symbolic names. This tool performs a best-effort structural conversion; for production use, use the official protoc JSON mapping with the schema available.