measurements.py 文件源码

def get_measurement(measurement_id):
    # XXX this query is SUPER slow
    m = RE_MSM_ID.match(measurement_id)
    if not m:
        raise BadRequest("Invalid measurement_id")
    msm_no = int(m.group(1))

    q = current_app.db_session.query(
            Measurement.report_no.label('report_no'),
            Measurement.frame_off.label('frame_off'),
            Measurement.frame_size.label('frame_size'),
            Measurement.intra_off.label('intra_off'),
            Measurement.intra_size.label('intra_size'),
            Report.report_no.label('r_report_no'),
            Report.autoclaved_no.label('r_autoclaved_no'),
            Autoclaved.filename.label('a_filename'),
            Autoclaved.autoclaved_no.label('a_autoclaved_no'),

    ).filter(Measurement.msm_no == msm_no)\
        .join(Report, Report.report_no == Measurement.report_no)\
        .join(Autoclaved, Autoclaved.autoclaved_no == Report.autoclaved_no)
    try:
        msmt = q.one()
    except exc.MultipleResultsFound:
        current_app.logger.warning("Duplicate rows for measurement_id: %s" % measurement_id)
        msmt = q.first()
    except exc.NoResultFound:
        # XXX we should actually return a 404 here
        raise BadRequest("No measurement found")

    # Usual size of LZ4 frames is 256kb of decompressed text.
    # Largest size of LZ4 frame was ~55Mb compressed and ~56Mb decompressed. :-/
    range_header = "bytes={}-{}".format(msmt.frame_off, msmt.frame_off + msmt.frame_size - 1)
    r = requests.get(urljoin(current_app.config['AUTOCLAVED_BASE_URL'], msmt.a_filename),
            headers={"Range": range_header})
    r.raise_for_status()
    blob = r.content
    if len(blob) != msmt.frame_size:
        raise RuntimeError('Failed to fetch LZ4 frame', len(blob), msmt.frame_size)
    blob = lz4framed.decompress(blob)[msmt.intra_off:msmt.intra_off+msmt.intra_size]
    if len(blob) != msmt.intra_size or blob[:1] != b'{' or blob[-1:] != b'}':
        raise RuntimeError('Failed to decompress LZ4 frame to measurement.json', len(blob), msmt.intra_size, blob[:1], blob[-1:])
    # There is no replacement of `measurement_id` with `msm_no` or anything
    # else to keep sanity. Maybe it'll happen as part of orchestration update.
    # Also, blob is not decoded intentionally to save CPU
    return current_app.response_class(
        blob,
        mimetype=current_app.config['JSONIFY_MIMETYPE']
    )