astralis/src/Components/Compressor.vala

using Invercargill;
using Invercargill.DataStructures;
using Inversion;

namespace Astralis {

    public abstract class Compressor : Object, PipelineComponent {

        protected uint64 max_buffer_size = 1024 * 1024 * 16;

        protected Compressor(uint64 max_buffer_size = 1024 * 1024 * 16) {
            this.max_buffer_size = max_buffer_size;
        }

        // The encoding token that identifies this compression format (e.g., "gzip")
        public abstract string encoding_token { get; }

        public async HttpResult process_request(HttpContext http_context, PipelineContext pipeline_context) throws Error {
            var result = yield pipeline_context.next();
            
            // Check existing encoding on http result (to avoid double encoding)
            // Do not compress if: DO_NOT_COMPRESS flag set, or `Content-Encoding` is set and is not "identity"
            var existing_encoding = http_context.request.headers.get_any_or_default("Content-Encoding");
            if (result.flag_is_set(HttpResultFlag.DO_NOT_COMPRESS) || (existing_encoding != null && existing_encoding != "identity")) {
                return result;
            }
            
            var accept_encoding = http_context.request.headers.get_any_or_default("Accept-Encoding");
            
            // Check if the client accepts this encoding
            if (accept_encoding == null || !accept_encoding.contains(encoding_token)) {
                return result;
            }

            // Don't event bother compressing if the length is less than 10 bytes (length of GZip header)
            if(result.content_length != null && result.content_length <= 10) {
                return result;
            }

            // Case 1: Content length known and within threshold -> buffer and compress
            if (result.content_length != null && result.content_length <= max_buffer_size) {
                var buffered_result = yield buffer_and_compress(result);
                if (buffered_result != null) {
                    buffered_result.set_header("Content-Encoding", encoding_token);
                    buffered_result.set_header("Vary", "Accept-Encoding");
                    return buffered_result;
                }
                return result;
            }

            // Case 2: Content length known and above threshold, and `DO_NOT_CHUNK` is set -> do nothing
            else if(result.content_length != null && result.flag_is_set(HttpResultFlag.DO_NOT_CHUNK)) {
                return result;
            }

            // Case 3: Content length above threshold or is unknown -> send chunked response (no Content-Length header)
            var streaming_result = compress_chunked(result);
            streaming_result.set_header("Content-Encoding", encoding_token);
            streaming_result.set_header("Vary", "Accept-Encoding");
            return streaming_result;

        }

        /// Compress a ByteBuffer of data, returning null if compression doesn't reduce size
        /// @param data The data to compress
        /// @param content_type Optional content type hint for compression optimization
        /// @return Compressed data as ByteBuffer, or null if compression doesn't reduce size
        public abstract ByteBuffer compress_buffer(ByteBuffer data, string? content_type) throws Error;

        /// Create a streaming compression result for the given inner result, or null if not compressible
        public abstract HttpResult compress_chunked(HttpResult inner_result) throws Error;

        /// Buffer and compress the result, returning null if compression doesn't reduce size
        private async HttpResult? buffer_and_compress(HttpResult inner_result) throws Error {
            // Buffer the input data
            var input_buffer = new BufferAsyncOutput();
            yield inner_result.send_body(input_buffer);
            var input_data = input_buffer.get_buffer();
            
            // Get content type for potential compression optimization
            string? content_type = null;
            if (inner_result.headers.has("Content-Type")) {
                content_type = inner_result.headers["Content-Type"];
            }
            
            // Compress the data
            ByteBuffer? compressed_data = compress_buffer(input_data, content_type);
            if (compressed_data == null) {
                return null;
            }
            
            // Check if compressed size is larger than or equal to original
            if (compressed_data.length >= input_data.length) {
                return null;
            }
            
            // Construct HttpDataResult and copy headers
            var http_result = new HttpDataResult(compressed_data, inner_result.status);
            copy_headers(inner_result, http_result, {"content-length", "content-encoding"});
            return http_result;
        }

        protected static void copy_headers(HttpResult? source, HttpResult dest, string[] skip_headers) {
            if (source == null) {
                return;
            }
            foreach (var header in source.headers) {
                var lower_key = header.key.down();
                bool skip = false;
                foreach (var skip_key in skip_headers) {
                    if (lower_key == skip_key) {
                        skip = true;
                        break;
                    }
                }
                if (!skip) {
                    dest.set_header(header.key, header.value);
                }
            }
        }

    }

}