def post(self):
fig_data = self.get_fig_data()
if fig_data is None:
return self.visible_error(403, 'Broken connection to server.')
if not self.request.files:
return self.visible_error(403, 'No file uploaded.')
f = self.request.files['query'][0]
fname = f['filename']
logging.info('Parsing file: %s', fname)
fh = BytesIO(f['body'])
try:
query = parse_spectrum(fh)
except Exception:
try:
fh = StringIO(f['body'].decode('utf-8', 'ignore'),
newline=None)
query = parse_spectrum(fh)
except Exception:
logging.exception('Spectrum parse failed.')
# XXX: save failed uploads for debugging purposes
open('logs/badupload-' + fname, 'w').write(f['body'])
return self.visible_error(415, 'Spectrum upload failed.')
ds = UploadedSpectrumDataset(fname, query)
fig_data.set_selected(ds.view(), title=fname)
axlimits = fig_data.plot()
return self.write_json(axlimits)
def main():
ap = ArgumentParser()
ap.add_argument('--algorithm', default='als', choices=BL_CLASSES.keys(),
help='Algorithm type. [%(default)s]')
ap.add_argument('-s', '--segment', action='store_true',
help='Auto-detect band segments and run on each separately.')
ap.add_argument('-o', '--out-dir', help='If provided, write corrected files'
' to disk instead of plotting them.')
ap.add_argument('files', type=open, nargs='+')
args = ap.parse_args()
for f in args.files:
S = parse_spectrum(f)
bands, intensities = S.T
bl = BL_CLASSES[args.algorithm]()
corrected = bl.fit_transform(bands, intensities, segment=args.segment)
if args.out_dir:
out_file = os.path.join(args.out_dir, os.path.basename(f.name))
if os.path.exists(out_file):
print 'Skipping existing file:', out_file
continue
header = ('Baseline corrected data from original file %s\n'
'BL Algorithm: %s (default params)' % (f.name, args.algorithm))
traj = np.column_stack((bands, corrected))
np.savetxt(out_file, traj, fmt='%f', header=header)
else:
_, (ax1,ax2) = pyplot.subplots(nrows=2)
ax1.plot(bands, intensities, 'r-')
ax1.plot(bands, bl.baseline, 'k-')
ax2.plot(bands, corrected, '-')
pyplot.suptitle(f.name)
pyplot.show()
def _traj_ds(fh, ds_name, ds_kind, meta_kwargs, meta_pkeys, resample,
description):
zf = ZipFile(fh)
traj_data = {}
for subfile in zf.infolist():
if subfile.file_size <= 0:
continue
# ignore directory prefixes
fname = os.path.basename(subfile.filename)
# ignore hidden files
if fname.startswith('.'):
continue
# read and wrap, because the ZipExtFile object isn't seekable
sub_fh = BytesIO(zf.open(subfile).read())
try:
# TODO: ensure each traj has wavelengths in increasing order
traj_data[fname] = parse_spectrum(sub_fh)
except Exception:
logging.exception('bad spectrum subfile: ' + fname)
return (415, 'Unable to parse spectrum file: %s' % fname)
num_meta = len(meta_pkeys)
num_traj = len(traj_data)
if num_meta == 0:
meta_pkeys = traj_data.keys()
elif num_meta != num_traj:
return (415, 'Failed: %d metadata entries for %d spectra' %
(num_meta, num_traj))
else:
for pkey in meta_pkeys:
if pkey not in traj_data:
return (415, 'Failed: %r not in spectra.' % pkey)
if resample is None:
_load = _make_loader_function(description, meta_pkeys, traj_data,
**meta_kwargs)
WebTrajDataset(ds_name, ds_kind, _load)
else:
lb, ub, step = map(_maybe_float, resample)
waves = [t[:, 0] for t in traj_data.values()]
if lb is None:
lb = max(w[0] for w in waves)
if ub is None:
ub = min(w[-1] for w in waves)
if step is None:
step = min(np.diff(w).min() for w in waves)
wave = np.arange(lb, ub + step / 2, step, dtype=waves[0].dtype)
spectra = np.zeros((len(waves), len(wave)), dtype=wave.dtype)
for i, key in enumerate(meta_pkeys):
traj = traj_data[key]
spectra[i] = np.interp(wave, traj[:, 0], traj[:, 1])
pkey = PrimaryKeyMetadata(meta_pkeys)
_load = _make_loader_function(description,
wave,
spectra,
pkey=pkey,
**meta_kwargs)
WebVectorDataset(ds_name, ds_kind, _load)
return None
def _parse(infile):
return infile, parse_spectrum(infile)
dest='legend',
help='Turn off figure legend')
ap.add_argument('--pp', default='normalize:max', help='Preprocess string')
ap.add_argument('--baseline',
default='none',
help='Baseline algorithm',
choices=BL_CLASSES.keys() + ['none'])
ap.add_argument('--type',
choices=PARSERS,
help=('Parser type to use for all files. If not provided, '
'all parsers will be attempted.'))
ap.add_argument('file', nargs='+', help='Spectrum file(s).')
args = ap.parse_args()
fig, ax = pyplot.subplots(figsize=(12, 6))
for f in args.file:
print('Parsing', f)
bands, intensities = parse_spectrum(f, filetype=args.type).T
if args.baseline != 'none':
bl_alg = BL_CLASSES[args.baseline]()
intensities = bl_alg.fit_transform(bands, intensities)
intensities = preprocess(intensities[None], args.pp).ravel()
ax.plot(bands, intensities, label=basename(f))
ax.yaxis.set_ticklabels([])
ax.set_xlabel('Wavenumber (1/cm)')
ax.set_ylabel('Intensity')
if args.legend:
pyplot.legend(loc='best')
pyplot.tight_layout()
pyplot.show()