import pandas as pd import logging import os import datetime from pybis import Openbis import hidden admissible_props_list = ['$name', 'filenumber', 'default_experiment.experimental_results', 'dataquality', '$xmlcomments', '$annotations_state', 'sample_name', 'position_x', 'position_y', 'position_z', 'temp', 'cell_pressure', 'gas_flow_setting', 'sample_notes', 'beamline', 'photon_energy', 'slit_entrance_v', 'slit_exit_v', 'izero', 'slit_exit_h', 'hos', 'cone', 'endstation', 'hof', 'method_name', 'region', 'lens_mode', 'acq_mode', 'dwell_time', 'frames', 'passenergy', 'iterations', 'sequenceiterations', 'ke_range_center', 'ke_step'] def initialize_openbis_obj(): # TODO: implement a more secure authentication method. openbis_obj = Openbis('https://openbis-psi.labnotebook.ch/openbis/webapp/eln-lims/?menuUniqueId=null&viewName=showBlancPage&viewData=null', verify_certificates=False) openbis_obj.login(hidden.username,hidden.password) return openbis_obj def align_datetime_observation_windows(df_h5: pd.DataFrame, df_openbis: pd.DataFrame, h5_datetime_var: str = 'lastModifiedDatestr', ob_datetime_var: str = 'registrationDate') -> pd.DataFrame: """ returns filtered/reduced versions of 'df' and 'df_ref' with aligned datetime observation windows. That is, the datetime variable range is the same for the returned dataframes.""" #""returns a filtered or reduced version of 'df' by removing all rows that are outside the datetime variable overlapping region between 'df' and 'df_ref'. #""" #df_h5['lastModifiedDatestr'] = df_h5['lastModifiedDatestr'].astype('datetime64[ns]') #df_h5 = df_h5.sort_values(by='lastModifiedDatestr') if not (h5_datetime_var in df_h5.columns.to_list() and ob_datetime_var in df_openbis.columns.to_list()): #TODO: Check if ValueError is the best type of error to raise here raise ValueError("Dataframes 'df' and 'df_ref' must contain columns 'datetime_var' and 'datetime_var_ref', storing values in suitable datetime string format (e.g., yyyy-mm-dd hh:mm:ss).") df_h5[h5_datetime_var] = df_h5[h5_datetime_var].astype('datetime64[ns]') df_openbis[ob_datetime_var] = df_openbis[ob_datetime_var].astype('datetime64[ns]') min_timestamp = max([df_openbis[ob_datetime_var].min(), df_h5[h5_datetime_var].min()]) max_timestamp = min([df_openbis[ob_datetime_var].max(), df_h5[h5_datetime_var].max()]) # Determine overlap between df and df_ref, and filters out all rows from df with datetime variable outside the overlapping datetime region. datetime_overlap_indicator = (df_h5[h5_datetime_var] >= min_timestamp) & (df_h5[h5_datetime_var] <= max_timestamp) df_h5 = df_h5.loc[datetime_overlap_indicator,:] datetime_overlap_indicator = (df_openbis[ob_datetime_var] >= min_timestamp) & (df_openbis[ob_datetime_var] <= max_timestamp) df_openbis = df_openbis.loc[datetime_overlap_indicator,:] df_h5 = df_h5.sort_values(by=h5_datetime_var) df_openbis = df_openbis.sort_values(by=ob_datetime_var) return df_h5, df_openbis def reformat_openbis_dataframe_filenumber(df_openbis): if not 'FILENUMBER' in df_openbis.columns: raise ValueError('df_openbis does not contain the column "FILENUMBER". Make sure you query (e.g., o.get_samples(props=["filenumbe"])) that before creating df_openbis.') #if not 'name' in df.columns: # raise ValueError("df does not contain the column 'name'. Ensure df complies with Throsten's Table's format.") # Augment df_openbis with 'name' column consitent with Thorsten's naming convention name_list = ['0' + item.zfill(3) + item.zfill(3) for item in df_openbis['FILENUMBER']] df_openbis['REFORMATED_FILENUMBER'] = pd.Series(name_list, index=df_openbis.index) return df_openbis def pair_openbis_and_h5_dataframes(df_openbis, df_h5, pairing_ob_var: str, pairing_h5_var: str): """ Pairs every row (or openbis sample) in 'df_openbis' with a set of rows (or measurements) in 'df_h5' by matching the i-th row in 'df_h5' with the rows of 'df_h5' that satisfy the string df_openbis.loc[i,pairing_var_1] is contained in the string df_h5[i,pairing_var_2] Example: pairing_var_1, pairing_var_2 = reformated 'REFORMATED_FILENUMBER', 'name' """ # Reformat openbis dataframe filenumber so that it can be used to find associated measurements in h5 dataframe df_openbis = reformat_openbis_dataframe_filenumber(df_openbis) related_indices_list = [] for sample_idx in df_openbis.index: sample_value = df_openbis.loc[sample_idx,pairing_ob_var] tmp_list = [sample_value in item[0:item.find('_')] for item in df_h5[pairing_h5_var]] related_indices_list.append(df_h5.index[tmp_list]) print('Paring openbis sample: ' + df_openbis.loc[sample_idx,pairing_ob_var]) print('with reformated FILENUMBER: ' + sample_value) print('to following measurements in h5 dataframe:') print(df_h5.loc[df_h5.index[tmp_list],'name']) print('\n') df_openbis['related_h5_indices'] = pd.Series(related_indices_list, index=df_openbis.index) return df_openbis def range_cols_2_string(df,lb_var,ub_var): if not sum(df.loc[:,ub_var]-df.loc[:,lb_var])==0: #tmp_list = ['-'.join([str(round(df.loc[i,lb_var],2)),str(round(df.loc[i,ub_var],1))]) for i in df.index] tmp_list = ['-'.join(["{:.1f}".format(df.loc[i,lb_var]),"{:.1f}".format(df.loc[i,ub_var])]) for i in df.index] elif len(df.loc[:,lb_var].unique())>1: # check if values are different #tmp_list = [str(round(df.loc[i,lb_var],2)) for i in df.index] tmp_list = ["{:.1f}".format(df.loc[i,lb_var]) for i in df.index] else: #tmp_list = [str(round(df.loc[0,lb_var],2))] tmp_list = ["{:.1f}".format(df.loc[0,lb_var])] return '/'.join(tmp_list) def col_2_string(df,column_var): if not column_var in df.columns: raise ValueError("'column var must belong in df.columns") #tmp_list = [str(round(item,1)) for item in df[column_var]] tmp_list = ["{:.2f}".format(item) for item in df[column_var]] if len(df[column_var].unique())==1: tmp_list = [tmp_list[0]] return '/'.join(tmp_list) def compute_openbis_sample_props_from_h5(df_openbis, df_h5, sample_idx): prop2attr = {'sample_name':'sample', # ask Throsten whether this assignment is correct or not 'position_x':'smplX_mm', 'position_y':'smplY_mm', 'position_z':'smplZ_mm', 'temp':'sampleTemp_dC', 'cell_pressure':'cellPressure_mbar', #'gas_flow_setting': '', 'method_name':'regionName', # measurement type: XPS or NEXAFS 'region':'regionName', # VB/N1s/C1s 'passenergy':'regionName', # REAL 'photon_energy':'xRayEkinRange_eV', 'dwell_time':'scientaDwellTime_ms', 'acq_mode':'scientaAcquisitionMode', 'ke_range_center':'scientaEkinRange_eV', 'ke_step':'scientaEkinStep_eV', 'lens_mode':'scientaLensMode' } sample_identifier = df_openbis.loc[sample_idx,'identifier'] props_dict = {'FILENUMBER' : df_openbis.loc[sample_idx,'FILENUMBER']} #props_dict = {} if not len(df_openbis.loc[sample_idx,'related_h5_indices']): props_dict['identifier'] = sample_identifier return props_dict reduced_df_h5 = df_h5.loc[df_openbis.loc[sample_idx,'related_h5_indices'],:] reduced_df_h5 = reduced_df_h5.reset_index() # include related_samples key for validation purposes. Related samples are used to compute average and/or combined openbis properties. related_sample_list = [reduced_df_h5['name'][index] for index in reduced_df_h5['name'].index] related_samples = ' / '.join(related_sample_list) props_dict['Subject_samples'] = related_samples props_dict['sample_name'] = reduced_df_h5['sample'].unique()[0] if len(reduced_df_h5['sample'].unique())==1 else '/'.join(reduced_df_h5['sample'].tolist()) if not 'NEXAFS' in reduced_df_h5['regionName'].iloc[0]: props_dict['identifier'] = sample_identifier props_dict['method_name'] = 'XPS' for item_idx in reduced_df_h5.index: item = reduced_df_h5.loc[item_idx,'regionName'] if item_idx > 0: props_dict['region'] = props_dict['region'] + '/' + item[0:item.find('_')] #props_dict['dwell_time'] = props_dict['dwell_time'] + '/' + str(reduced_df_h5.loc[item_idx,'scientaDwellTime_ms']) #props_dict['ke_range_center'] = props_dict['ke_range_center'] + '/' + str(round(reduced_df_h5.loc[item_idx,['scientaEkinRange_eV_1','scientaEkinRange_eV_2']].mean(),2)) #props_dict['ke_step_center'] = props_dict['ke_step_center'] + '/' + str(reduced_df_h5.loc[item_idx,'scientaEkinStep_eV']) #props_dict['passenergy'].append(float(item[item.find('_')+1:item.find('eV')])) else: props_dict['region'] = item[0:item.find('_')] #props_dict['dwell_time'] = str(reduced_df_h5.loc[item_idx,'scientaDwellTime_ms']) #props_dict['ke_range_center'] = str(round(reduced_df_h5.loc[item_idx,['scientaEkinRange_eV_1','scientaEkinRange_eV_2']].mean(),2)) #props_dict['ke_step_center'] = str(reduced_df_h5.loc[item_idx,'scientaEkinStep_eV']) #props_dict['passenergy'] = reduced_df_h5.loc[:,'scientaPassEnergy_eV'].min() else: props_dict = {'identifier':sample_identifier,'method_name':'NEXAFS'} #props_dict['temp'] = round(reduced_df_h5['sampleTemp_dC'].mean(),2) #props_dict['cell_pressure'] = round(reduced_df_h5['cellPressure_mbar'].mean(),2) props_dict['temp'] = "{:.2f}".format(reduced_df_h5['sampleTemp_dC'].mean()) props_dict['cell_pressure'] = "{:.2f}".format(reduced_df_h5['cellPressure_mbar'].mean()) reduced_df_h5['scientaDwellTime_ms'] = reduced_df_h5['scientaDwellTime_ms']*1e-3 # covert ms to seconds props_dict['dwell_time'] = col_2_string(reduced_df_h5,'scientaDwellTime_ms') props_dict['passenergy'] = col_2_string(reduced_df_h5,'scientaPassEnergy_eV') props_dict['ke_step_center'] = col_2_string(reduced_df_h5,'scientaEkinStep_eV') #props_dict['photon_energy'] =round(reduced_df_h5[['xRayEkinRange_eV_1','xRayEkinRange_eV_2']].mean(axis=1)[0],2) props_dict['photon_energy'] = range_cols_2_string(reduced_df_h5,'xRayEkinRange_eV_1','xRayEkinRange_eV_2') props_dict['ke_range_center'] = range_cols_2_string(reduced_df_h5,'scientaEkinRange_eV_1','scientaEkinRange_eV_2') props_dict['lens_mode'] = reduced_df_h5['scientaLensMode'][0] props_dict['acq_mode'] = reduced_df_h5['scientaAcquisitionMode'][0] props_dict['position_x'] = "{:.2f}".format(reduced_df_h5.loc[:,'smplX_mm'].mean()) # round(reduced_df_h5.loc[:,'smplX_mm'].mean(),2) props_dict['position_y'] = "{:.2f}".format(reduced_df_h5.loc[:,'smplY_mm'].mean()) props_dict['position_z'] = "{:.2f}".format(reduced_df_h5.loc[:,'smplZ_mm'].mean()) return props_dict def single_sample_update(sample_props_dict,sample_collection,props_include_list): """ Updates sample in openbis database specified in sample_props_dict, which must belong in sample_collection (i.e., result of openbis_obj.get_samples(...)) """ try: sample_path_identifier = sample_props_dict['identifier'] #path-like index sample = sample_collection[sample_path_identifier] for prop in sample_props_dict.keys(): if (prop in admissible_props_list) and (prop in props_include_list): sample.props[prop] = sample_props_dict[prop] sample.save() except Exception: logging.error(Exception) return 0 def sample_batch_update(openbis_obj,sample_collection,df_openbis,df_h5,props_include_list): """ See """ if not 'related_h5_indices' in df_openbis.columns: raise ValueError("Input dataframe 'df_openbis' must contain a column named 'related_h5_indeces', resulting from suitable proprocessing steps.") # TODO: as a safeguard, create exclude list containing properties that must not be changed exclude_list = ['filenumber','FILENUMBER','identifier'] for item in props_include_list: if item in exclude_list: props_include_list.remove(item) trans = openbis_obj.new_transaction() for sample_idx in len(range(df_openbis['identifier'])): props_dict = compute_openbis_sample_props_from_h5(df_openbis, df_h5, sample_idx) sample_path_identifier = props_dict['identifier'] #path-like index sample = sample_collection[sample_path_identifier] for prop in props_dict.keys(): if prop in props_include_list: sample.props[prop] = props_dict[prop] trans.add(sample) trans.commit() return 0 def conduct_dataframe_preprocessing_steps(df_h5, df_openbis): if not 'lastModifiedDatestr'in df_h5.columns: raise ValueError('') df_h5, df_openbis = align_datetime_observation_windows(df_h5, df_openbis, 'lastModifiedDatestr' , 'registrationDate') df_openbis = pair_openbis_and_h5_dataframes(df_openbis, df_h5, 'REFORMATED_FILENUMBER', 'name') return df_h5, df_openbis