Remove debugging printout from JPK driver
[hooke.git] / hooke / driver / jpk.py
index 28887789203b315a84eb8503b19cc85978332a8b..7b33c5b19bbce5b083703d50069222e513e221c6 100644 (file)
-import string
-from .. import curve as lhc
-
-class DataChunk(list):
-    #Dummy class to provide ext and ret methods to the data list.
-
-    def ext(self):
-        halflen=(len(self)/2)
-        return self[0:halflen]
-
-    def ret(self):
-        halflen=(len(self)/2)
-        return self[halflen:]
-
-class jpkDriver(lhc.Driver):
-
-    def __init__(self, filename):
-        self.filename=filename #self.filename can always be useful, and should be defined
-        self.filedata = open(filename,'r') #We open the file
-        self.filelines=self.filedata.readlines()
-        self.filedata.close()
-        '''These are two strings that can be used by Hooke commands/plugins to understand what they are looking at. They have no other
-        meaning. They have to be somehow defined however - commands often look for those variables.
-
-        self.filetype should contain the name of the exact filetype defined by the driver (so that filetype-specific commands can know
-                      if they're dealing with the correct filetype)
-        self.experiment should contain instead the type of data involved (for example, various drivers can be used for force-clamp experiments,
-                      but hooke commands could like to know if we're looking at force clamp data, regardless of their origin, and not other
-                      kinds of data)
-
-        Of course, all other variables you like can be defined in the class.
-        '''
-        self.filetype = 'jpk'
-        self.experiment = 'smfs'
-
-
-
-    def __del__(self):
-        self.filedata.close()
-
-    def is_me(self):
-        '''
-        we define our magic heuristic for jpk files
-        '''
-        myfile=file(self.filename)
-        headerlines=myfile.readlines()[0:3]
-        myfile.close()
-        if headerlines[0][0:11]=='# xPosition' and headerlines[1][0:11]=='# yPosition':
-            return True
-        else:
+# Copyright (C) 2008-2010 Massimo Sandal <devicerandom@gmail.com>
+#                         W. Trevor King <wking@drexel.edu>
+#
+# This file is part of Hooke.
+#
+# Hooke is free software: you can redistribute it and/or modify it
+# under the terms of the GNU Lesser General Public License as
+# published by the Free Software Foundation, either version 3 of the
+# License, or (at your option) any later version.
+#
+# Hooke is distributed in the hope that it will be useful, but WITHOUT
+# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+# or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General
+# Public License for more details.
+#
+# You should have received a copy of the GNU Lesser General Public
+# License along with Hooke.  If not, see
+# <http://www.gnu.org/licenses/>.
+
+"""Driver for JPK ForceRobot's velocity clamp data format.
+"""
+
+import os.path
+import pprint
+import zipfile
+
+import numpy
+
+from .. import curve as curve
+from .. import experiment as experiment
+from ..util.util import Closing as Closing
+from ..util.si import join_data_label, split_data_label
+from . import Driver as Driver
+
+
+class JPKDriver (Driver):
+    """Handle JPK ForceRobot's data format.
+    """
+    def __init__(self):
+        super(JPKDriver, self).__init__(name='jpk')
+
+    def is_me(self, path):
+        if os.path.isdir(path):
             return False
-
-    def close_all(self):
-        self.filedata.close()
-
-    def _read_data_segment(self):
-        #routine that actually reads the data
-
-        height_ms=[]
-        height_m=[]
-        height=[]
-        v_deflection=[]
-        h_deflection=[]
-
-        self.springconstant=0 #if we don't meet any spring constant, use deflection...
-
-        for line in self.filelines:
-            #we meet the segment defining the order of data columns
-
-            if line[0:9]=='# columns':
-                splitline=line.split()[2:]
-                height_ms_index=splitline.index('smoothedStrainGaugeHeight')
-                height_m_index=splitline.index('strainGaugeHeight')
-                height_index=splitline.index('height')
-                v_deflection_index=splitline.index('vDeflection')
-                #h_deflection=splitline.index('hDeflection')
-
-            if line[0:16]=='# springConstant':
-                self.springconstant=float(line.split()[2])
-
-            if line[0] != '#' and len(line.split())>1:
-                dataline=line.split()
-                height_ms.append(float(dataline[height_ms_index]))
-                height_m.append(float(dataline[height_m_index]))
-                height.append(float(dataline[height_index]))
-                v_deflection.append(float(dataline[v_deflection_index]))
-                #h_deflection.append(float(dataline[h_deflection_index]))
-
-        if self.springconstant != 0:
-            force=[item*self.springconstant for item in v_deflection]
-        else: #we have measured no spring constant :(
-            force=v_deflection
-
-        height_ms=DataChunk([item*-1 for item in height_ms])
-        height_m=DataChunk([item*-1 for item in height_m])
-        height=DataChunk([item*-1 for item in height])
-        deflection=DataChunk(v_deflection)
-        force=DataChunk(force)
-
-        return height_ms,height_m,height,deflection,force
-
-    def deflection(self):
-        height_ms,height_m,height,deflection,force=self._read_data_segment()
-        deflection_ext=deflection.ext()
-        deflection_ret=deflection.ret()
-        deflection_ret.reverse()
-        return deflection_ext,deflection_ret
-
-    def default_plots(self):
-
-        height_ms,height_m,height,deflection,force=self._read_data_segment()
-
-        height_ms_ext=height_ms.ext()
-        height_ms_ret=height_ms.ret()
-        force_ext=force.ext()
-        force_ret=force.ret()
-        #reverse the return data, to make it coherent with hooke standard
-        height_ms_ret.reverse()
-        force_ret.reverse()
-
-        main_plot=lhc.PlotObject()
-        main_plot.add_set(height_ms_ext,force_ext)
-        main_plot.add_set(height_ms_ret,force_ret)
-
-
-
-        if self.springconstant != 0:
-            main_plot.units=['meters','force']
+        if zipfile.is_zipfile(path):  # JPK file versions since at least 0.5
+            with Closing(zipfile.ZipFile(path, 'r')) as f:
+                if 'header.properties' not in f.namelist():
+                    return False
+                with Closing(f.open('header.properties')) as h:
+                    if 'jpk-data-file' in h.read():
+                        return True
         else:
-            main_plot.units=['meters','meters']
-
-        main_plot.normalize_vectors()
-
-        main_plot.destination=0
-        main_plot.title=self.filename
-
-        return [main_plot]
+            with Closing(open(path, 'r')) as f:
+                headlines = []
+                for i in range(3):
+                    headlines.append(f.readline())
+            if headlines[0].startswith('# xPosition') \
+                    and headlines[1].startswith('# yPosition'):
+                return True
+        return False
+
+    def read(self, path, info=None):
+        if info == None:
+            info = {}
+        if zipfile.is_zipfile(path):  # JPK file versions since at least 0.5
+            return self._read_zip(path, info)
+        else:
+            return self._read_old(path, info)
+
+    def _read_zip(self, path, info):
+        with Closing(zipfile.ZipFile(path, 'r')) as f:
+            f.path = path
+            zip_info = self._zip_info(f)
+            segments = []
+            for i in range(len([p for p in f.namelist()
+                                if p.endswith('segment-header.properties')])):
+                segments.append(self._zip_segment(f, path, info, zip_info, i))
+        if zip_info['file-format-version'] not in ['0.5']:
+            raise NotImplementedError(
+                'JPK file version %s not supported (yet).'
+                % zip_info['file-format-version'])
+        for name in ['approach', 'retract']:
+            if len([s for s in segments if s.info['name'] == name]) == 0:
+                raise ValueError(
+                    'No segment for %s in %s, only %s'
+                    % (name, path, [s.info['name'] for s in segments]))
+        curve_info = self._zip_translate_params(zip_info,
+                                                segments[0].info['raw info'])
+        for segment in segments:
+            segment.info['spring constant (N/m)'] = \
+                curve_info['spring constant (N/m)']
+        return (segments, curve_info)
+
+    def _zip_info(self, zipfile):
+        with Closing(zipfile.open('header.properties')) as f:
+            info = self._parse_params(f.readlines())
+            return info
+
+    def _zip_segment(self, zipfile, path, info, zip_info, index):
+        prop_file = zipfile.open(os.path.join(
+                'segments', str(index), 'segment-header.properties'))
+        prop = self._parse_params(prop_file.readlines())
+        prop_file.close()
+        expected_shape = (int(prop['force-segment-header']['num-points']),)
+        channels = []
+        for chan in prop['channels']['list']:
+            chan_info = prop['channel'][chan]
+            channels.append(self._zip_channel(zipfile, index, chan, chan_info))
+            if channels[-1].shape != expected_shape:
+                    raise NotImplementedError(
+                        'Channel %d:%s in %s has strange shape %s != %s'
+                        % (index, chan, zipfile.path,
+                           channels[-1].shape, expected_shape))
+        d = curve.Data(
+            shape=(len(channels[0]), len(channels)),
+            dtype=channels[0].dtype,
+            info=self._zip_translate_segment_params(prop))
+        for i,chan in enumerate(channels):
+            d[:,i] = chan
+        return self._zip_scale_segment(d, path, info)
+
+    def _zip_channel(self, zipfile, segment_index, channel_name, chan_info):
+        f = zipfile.open(os.path.join(
+                'segments', str(segment_index),
+                chan_info['data']['file']['name']), 'r')
+        assert chan_info['data']['file']['format'] == 'raw', \
+            'Non-raw data format:\n%s' % pprint.pformat(chan_info)
+        assert chan_info['data']['type'] == 'float-data', \
+            'Non-float data format:\n%s' % pprint.pformat(chan_info)
+        data = numpy.frombuffer(
+            buffer(f.read()),
+            dtype=numpy.dtype(numpy.float32).newbyteorder('>'))
+        # '>' (big endian) byte order.
+        # From version 0.3 of JPKForceSpec.txt in the "Binary data" section:
+        #    All forms of raw data are stored in chronological order
+        #    (the order in which they were collected), and the
+        #    individual values are stored in network byte order
+        #    (big-endian). The data type used to store the data is
+        #    specified by the "channel.*.data.type" property, and is
+        #    either short (2 bytes per value), integer (4 bytes), or
+        #    float (4 bytes, IEEE format).
+        f.close()
+        return data
+
+    def _zip_translate_params(self, params, chan_info):
+        info = {
+            'raw info':params,
+            'filetype':self.name,
+            #'time':self._time_from_TODO(raw_info[]),
+            }
+        # TODO: distinguish between force clamp and velocity clamp
+        # experiments.  Note that the JPK file format is flexible
+        # enough to support mixed experiments (i.e. both force clamp
+        # and velocity clamp segments in a single experiment), but I
+        # have no idea what sort of analysis such experiments would
+        # require ;).
+        info['experiment'] = experiment.VelocityClamp
+        force_unit = chan_info['channel']['vDeflection']['conversion-set']['conversion']['force']['scaling']['unit']['unit']
+        assert force_unit == 'N', force_unit
+        force_base = chan_info['channel']['vDeflection']['conversion-set']['conversion']['force']['base-calibration-slot']
+        assert force_base == 'distance', force_base
+        dist_unit = chan_info['channel']['vDeflection']['conversion-set']['conversion']['distance']['scaling']['unit']['unit']
+        assert dist_unit == 'm', dist_unit
+        force_mult = float(
+            chan_info['channel']['vDeflection']['conversion-set']['conversion']['force']['scaling']['multiplier'])
+        info['spring constant (N/m)'] = force_mult
+        return info
+
+    def _zip_translate_segment_params(self, params):
+        info = {
+            'raw info':params,
+            'columns':list(params['channels']['list']),
+            'name':params['force-segment-header']['name']['name'],
+            }
+        if info['name'] in ['extend-spm', 'retract-spm', 'pause-at-end-spm']:
+            info['name'] = info['name'][:-len('-spm')]
+            if info['name'] == 'extend':
+                info['name'] = 'approach'
+        else:
+            raise NotImplementedError(
+                'Unrecognized segment type %s' % info['name'])
+        return info
+
+    def _zip_scale_segment(self, segment, path, info):
+        data = curve.Data(
+            shape=segment.shape,
+            dtype=segment.dtype,
+            info={})
+        data[:,:] = segment
+        segment.info['raw data'] = data
+
+        # raw column indices
+        channels = segment.info['raw info']['channels']['list']
+        for i,channel in enumerate(channels):
+            conversion = None
+            if channel == 'vDeflection':
+                conversion = 'distance'
+            segment = self._zip_scale_channel(
+                segment, channel, conversion=conversion, path=path, info=info)
+            name,unit = split_data_label(segment.info['columns'][i])
+            if name == 'vDeflection':
+                assert unit == 'm', segment.info['columns'][i]
+                segment.info['columns'][i] = join_data_label('deflection', 'm')
+            elif name == 'height':
+                assert unit == 'm', segment.info['columns'][i]
+                segment.info['columns'][i] = join_data_label('z piezo', 'm')
+        return segment
+
+    def _zip_scale_channel(self, segment, channel_name, conversion=None,
+                           path=None, info={}):
+        channel = segment.info['raw info']['channels']['list'].index(
+            channel_name)
+        conversion_set = segment.info['raw info']['channel'][channel_name]['conversion-set']
+        if conversion == None:
+            conversion = conversion_set['conversions']['default']
+        if conversion == conversion_set['conversions']['base']:
+            # Our conversion is the base data.
+            if conversion != 'volts':
+                raise NotImplementedError(
+                    'unknown units for base channel: %s' % conversion)
+            segment.info['columns'][channel] = join_data_label(
+                channel_name, 'V')
+            return segment
+        conversion_info = conversion_set['conversion'][conversion]
+        if conversion_info['base-calibration-slot'] \
+                != conversion_set['conversions']['base']:
+            # Our conversion is stacked on a previous conversion.  Do
+            # the previous conversion first.
+            segment = self._zip_scale_channel(
+                segment, channel_name,
+                conversion_info['base-calibration-slot'],
+                path=path, info=info)
+        if conversion_info['type'] == 'file':
+            # Michael Haggerty at JPK points out that the conversion
+            # information stored in the external file is reproduced in
+            # the force curve file.  So there is no need to actually
+            # read `conversion_info['file']`.  In fact, the data there
+            # may have changed with future calibrations, while the
+            # information stored directly in conversion_info retains
+            # the calibration information as it was when the experiment
+            # was performed.
+            pass  # Fall through to 'simple' conversion processing.
+        else:
+            assert conversion_info['type'] == 'simple', conversion_info['type']
+        assert conversion_info['scaling']['type'] == 'linear', \
+            conversion_info['scaling']['type']
+        assert conversion_info['scaling']['style'] == 'offsetmultiplier', \
+            conversion_info['scaling']['style']
+        multiplier = float(conversion_info['scaling']['multiplier'])
+        offset = float(conversion_info['scaling']['offset'])
+        unit = conversion_info['scaling']['unit']['unit']
+        segment[:,channel] = segment[:,channel] * multiplier + offset
+        segment.info['columns'][channel] = join_data_label(channel_name, unit)
+        return segment
+
+    def _parse_params(self, lines):
+        info = {}
+        for line in lines:
+            line = line.strip()
+            if line.startswith('#'):
+                continue
+            else:
+                # e.g.: force-segment-header.type=xy-position-segment-header
+                fields = line.split('=', 1)
+                assert len(fields) == 2, line
+                setting = fields[0].split('.')
+                sub_info = info  # drill down, e.g. info['force-s..']['type']
+                for s in setting[:-1]:
+                    if s not in sub_info:
+                        sub_info[s] = {}
+                    sub_info = sub_info[s]
+                if setting[-1] == 'list':  # split a space-delimited list
+                    sub_info[setting[-1]] = fields[1].split(' ')
+                else:
+                    sub_info[setting[-1]] = fields[1]
+        return info
+
+    def _read_old(self, path, info):
+        raise NotImplementedError('No old-style JPK files were available for testing, please send us yours: %s' % path)