如何使用flask替换pipeline_raw_http示例中的server.py?

您好，我正在尝试将pipeline_raw_http示例对接到flask 服务器上，按照server.py重构后的代码如下：
import datetimeimport wavefrom flask import Flask, request, Responseimport sysapp = Flask(__name__)def _get_chunk_size():    data = request.stream.read(2)    print(type(data))        while data[-2:] != b"rn":        data += request.stream.read(1)    return int(data[:-2], 16)def _get_chunk_data( chunk_size):    data = request.stream.read(chunk_size)    request.stream.read(2)    return datadef _write_wav( data, rates, bits, ch):    t = datetime.datetime.utcnow()    time = t.strftime('%Y%m%dT%H%M%SZ')    filename = str.format('{}_{}_{}_{}.wav', time, rates, bits, ch)    wavfile = wave.open(filename, 'wb')    wavfile.setparams((ch, int(bits/8), rates, 0, 'NONE', 'NONE'))    wavfile.writeframesraw(bytearray(data))    wavfile.close()    return filename@app.route('/')def hello_world():    return 'Hello, World!'@app.route("/upload", methods=["POST"])def upload():    total_bytes = 0    sample_rates = 0    bits = 0    channel = 0    if request.headers.get('Transfer-Encoding').lower() == 'chunked':        data = []        sample_rates = request.headers.get('x-audio-sample-rates', '').lower()        bits = request.headers.get('x-audio-bits', '').lower()        channel = request.headers.get('x-audio-channel', '').lower()        sample_rates = request.headers.get('x-audio-sample-rates', '').lower()        print("Audio information, sample rates: {}, bits: {}, channel(s): {}".format(sample_rates, bits, channel))        while True:            chunk_size = _get_chunk_size()            total_bytes += chunk_size            print("Total bytes received: {}".format(total_bytes))            sys.stdout.write("�33[F")            if (chunk_size == 0):                break            else:                chunk_data = _get_chunk_data(chunk_size)                data += chunk_data        filename = _write_wav(data, int(sample_rates), int(bits), int(channel))        def generate():            yield 'File {} was written, size {}'.format(filename, total_bytes).encode('utf-8')        response = Response(generate(), mimetype='text/html')        response.status_code = 200        response.headers['Content-type'] = 'text/html;charset=utf-8'        response.headers['Content-Length'] = str(total_bytes)                return response                if __name__ == '__main__':    app.run(host='0.0.0.0', port=8000, debug=True)
然而当代码运行到 _get_chunk_size() 时，使用request.stream.read()并不能像server.py中使用rfile.read()一样获取到正确的chunksize，而是获取到了与server.py中_get_chunk_data( chunk_size)类似的16bit音频流，似乎在request.stream.read()运行时有一部分数据流被跳过了。
请问这是什么原因？除了使用http stream 以外还有什么办法能将音频传输到后端？