Skip to content

Overview

The Pythonic version of the CommonDataModel is built object-orientated in the subfolder carrot/cdm/:

carrot/cdm/
├── __init__.py
├── decorators.py
├── model.py
├── objects
│   ├── __init__.py
│   ├── common.py
│   └── versions
│       ├── __init__.py
│       └── v5_3_1
│           ├── __init__.py
│           ├── condition_occurrence.py
│           ├── drug_exposure.py
│           ├── measurement.py
│           ├── observation.py
│           ├── person.py
│           ├── procedure_occurrence.py
│           ├── specimen.py
│           └── visit_occurrence.py

Destination Tables

All CDM destination tables are formed as objects and are defined in carrot/cdm/objects, inheriting from a base class (DestinationTable, defined in common.py):

Generating More Tables

The package contains .csv dumps taken from BCLink that give descriptions of what fields are contained within each CDM table.

At present are only dumps from version 5.3.1 of the CDM:

$ ls $(carrot info data_folder)/cdm/BCLINK_EXPORT/5.3.1
export-CONDITION_OCCURRENCE.csv export-MEASUREMENT.csv          export-PERSON.csv
export-DRUG_EXPOSURE.csv        export-OBSERVATION.csv

To help generate a pythonic template for a CDM template, the CLI can be used to do this

carrot generate cdm drug_exposure 5.3.1
This command tool outputs the following code that you could use to copy, paster and edit to create a new table for drug_exposure
self.drug_exposure_id = DestinationField(dtype="Integer", required=False , pk=True)
self.person_id = DestinationField(dtype="Integer", required=False )
self.drug_concept_id = DestinationField(dtype="Integer", required=False )
self.drug_exposure_start_date = DestinationField(dtype="Date", required=False )
self.drug_exposure_start_datetime = DestinationField(dtype="Timestamp", required=False )
self.drug_exposure_end_date = DestinationField(dtype="Date", required=False )
self.drug_exposure_end_datetime = DestinationField(dtype="Timestamp", required=False )
self.verbatim_end_date = DestinationField(dtype="Date", required=False )
self.drug_type_concept_id = DestinationField(dtype="Integer", required=False )
self.stop_reason = DestinationField(dtype="Text20", required=False )
self.refills = DestinationField(dtype="Integer", required=False )
self.quantity = DestinationField(dtype="Float", required=False )
self.days_supply = DestinationField(dtype="Integer", required=False )
self.sig = DestinationField(dtype="Integer", required=False )
self.route_concept_id = DestinationField(dtype="Integer", required=False )
self.lot_number = DestinationField(dtype="Text50", required=False )
self.provider_id = DestinationField(dtype="Integer", required=False )
self.visit_occurrence_id = DestinationField(dtype="Integer", required=False )
self.drug_source_value = DestinationField(dtype="Text50", required=False )
self.drug_source_concept_id = DestinationField(dtype="Integer", required=False )
self.route_source_value = DestinationField(dtype="Text50", required=False )
self.dose_unit_source_value = DestinationField(dtype="Text50", required=False )

Destination Fields

In common.py a class called DestinationField defines how to handle an input pandas series. This pandas series is effectively a column in the output of the CDM Tables, in other words, DestinationField is an object for the destination_field, e.g. person_id in the destination_table person.

class DestinationField(object):
    def __init__(self, dtype: str, required: bool, pk=False):
        self.series = None
        self.dtype = dtype
        self.required = required
        self.pk = pk
  • self.series: initialises as None and will hold a pandas.Series object if the column is to be filled in the output.
  • self.dtype: specifies a string for handling how to format the output of this column so it's in the right format when saved to the final .csv to be uploaded successfully to a BCLink.
  • self.required: if the destination_field is required (True), any rows of the final table that do not have this column filled (None, NaN), are removed (dropped) from the output.
  • self.pk: specifies if this column is the primary key for its associated table. This can be used to order the tables based on this.

Bases: Logger

Pythonic Version of the OHDSI CDM.

This class controls and manages CDM Table objects that are added to it

When self.process() is executed by the user, all added objects are defined, merged, formatted and finalised, before being dumped to an output file (.tsv file by default).

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
class CommonDataModel(Logger):
    """Pythonic Version of the OHDSI CDM.

    This class controls and manages CDM Table objects that are added to it

    When self.process() is executed by the user, all added objects are defined, merged, formatted and finalised, before being dumped to an output file (.tsv file by default).

    """


    @classmethod
    def load(cls,inputs,**kwargs):
        default_kwargs = {'save_files':False,'do_mask_person_id':False,'format_level':0}
        default_kwargs.update(kwargs)
        cdm = cls(**default_kwargs)
        cdm._load_inputs(inputs)
        return cdm

    @classmethod
    def from_rules(cls,rules,**kwargs):
        cdm = cls(**kwargs)
        cdm.create_and_add_objects(rules)
        return cdm


    def __init__(self, name=None, omop_version='5.3.1',
                 outputs = None,
                 save_files=True,
                 inputs=None,
                 use_profiler=False,
                 format_level=None,
                 do_mask_person_id=True,
                 drop_duplicates=True,
                 automatically_fill_missing_columns=True):
        """
        CommonDataModel class initialisation 
        Args:
            name (str): Give a name for the class to appear in the logging
            output_folder (str): Path of where the output tsv/csv files should be written to.
                                 The default is to save to a folder in the current directory
                                 called 'output_data'.
            inputs (dict or DataCollection): inputs can be a dictionary mapping file names to pandas dataframes,
                                        or can be a DataCollection object
            use_profiler (bool): Turn on/off profiling of the CPU/Memory of running the current process. 
                                 The default is set to false.
        """
        self.profiler = None
        name = self.__class__.__name__ if name is None else self.__class__.__name__ + "::" + name

        self.logger.info(f"CommonDataModel ({omop_version}) created with co-connect-tools version {carrot_version}")

        self.omop_version = omop_version

        self.drop_duplicates = drop_duplicates
        self.do_mask_person_id = do_mask_person_id
        self.execution_order = None

        if format_level == None:
            format_level = 1
        try:
            format_level = int(format_level)
        except ValueError:
            self.logger.error(f"You as specifying format_level='{format_level}' -- this should be an integer!")
            raise ValueError("format_level not set as an int ")

        self.format_level = FormatterLevel(format_level)
        self.profiler = None

        self.outputs = outputs
        self.save_files = save_files

        if use_profiler:
            self.logger.debug(f"Turning on cpu/memory profiling")
            self.profiler = Profiler(name=name)
            self.profiler.start()

        #perform some checks on the input data
        if isinstance(inputs,dict):
            self.logger.info("Running with an DataCollection object")
        elif isinstance(inputs,DataCollection):
            self.logger.info("Running with an DataCollection object")
        elif inputs is not None:
            _type = type(inputs).__name__
            raise BadInputObject(f"input object {inputs} is of type {_type}, which is not a valid input object")

        if inputs is not None:
            if hasattr(self,'inputs'):
                self.logger.warning("overwriting inputs")
            self.inputs = inputs
        elif not hasattr(self,'inputs'):
            self.inputs = None

        #register opereation tools
        self.tools = OperationTools()

        #allow rules to be generated automatically or not
        self.automatically_fill_missing_columns = automatically_fill_missing_columns
        if self.automatically_fill_missing_columns:
            self.logger.info(f"Turning on automatic cdm column filling")

        #define a person_id masker, if the person_id are to be masked
        if self.outputs:
            self.person_id_masker = self.outputs.load_global_ids()
            self.indexing_conf = self.outputs.load_indexing()
        else:
            self.person_id_masker = None
            self.indexing_conf = None

        #stores the final pandas dataframe for each CDM object
        # {
        #   'person':pandas.DataFrame,
        #   'measurement':pandas.DataFrame.
        #   ....
        #}
        self.__df_map = {}

        #stores the invididual objects associated to this model
        # {
        #     'observation':
        #     {
        #         'observation_0': <carrot.cdm.objects.observation.Observation object 0x000>,
        #         'observation_1': <carrot.cdm.objects.observation.Observation object 0x001>,
        #     },
        #     'measurement':
        #     {
        #         'measurement_0': <carrot.cdm.objects.measurement.Measurement object 0x000>,
        #         ...
        #     }
        #     ...
        # }
        self.__objects = {}
        #check if objects have already been registered with this class
        #via the decorator methods

        registered_objects = [
            getattr(self,name)
            for name in dir(self)
            if isinstance(getattr(self,name),DestinationTable)
        ]
        #if they have, then include them in this model
        for obj in registered_objects:
            obj.inputs = self.inputs
            self.add(obj)

        self.__analyses = {
            name:getattr(self,name)
            for name in dir(self)
            if isinstance(getattr(self,name),analysis)
        }

        #bookkeep some logs
        self.logs = {
            'meta':{
                'version': carrot_version,
                'created_by': getpass.getuser(),
                'created_at': strftime("%Y-%m-%dT%H%M%S", gmtime()),
                'dataset':name,
                'total_data_processed':{}
            }
        }

    def _load_inputs(self,inputs):
        for fname in inputs.keys():
            destination_table,_ = os.path.splitext(fname)
            try:
                obj = get_cdm_class(destination_table).from_df(inputs[fname],destination_table)
            except KeyError:
                self.logger.warning(f"Not loading {fname}, this is not a valid CDM Table")
                continue

            df = obj.get_df(force_rebuild=False)
            self[destination_table] = df.set_index(df.columns[0])

    def reset(self):
        self.__df_map.clear()
        [x.reset() for x in self.get_all_objects()]
        self.inputs.reset()

        if self.outputs:
            self.person_id_masker = self.outputs.load_global_ids()
            self.indexing_conf = self.outputs.load_indexing()
        else:
            self.person_id_masker = None
            self.indexing_conf = None



    def close(self):
        """
        Class destructor:
              Stops the profiler from running before deleting self
        """


        self.logger.info(json.dumps(self.logs['meta'],indent=6))
        if self.outputs:
            self.outputs.write_meta(self.logs)
            self.outputs.finalise()

        if not hasattr(self,'profiler'):
            return
        if self.profiler:
            self.profiler.stop()
            df_profile = self.profiler.get_df()
            f_out = self.output_folder
            f_out = f'{f_out}{os.path.sep}logs{os.path.sep}'
            if not os.path.exists(f'{f_out}'):
                self.logger.info(f'making output folder {f_out}')
                os.makedirs(f'{f_out}')

            date = self.logs['meta']['created_at']
            fname = f'{f_out}{os.path.sep}statistics_{date}.csv'
            df_profile.to_csv(fname)
            self.logger.info(f"Writen the memory/cpu statistics to {fname}")
            self.logger.info("Finished")

    @classmethod
    def from_existing(cls,**kwargs):
        """
        Initialise the CDM model from existing data in the CDM format
        """
        cdm = cls(**kwargs)
        if 'inputs' not in kwargs:
            raise NoInputFiles("you need to specify some inputs")
        inputs = kwargs['inputs']
        #loop over all input names
        for fname in inputs.keys():
            #obtain the name of the destination table
            #e.g fname='person.tsv' we want 'person'
            destination_table,_ = os.path.splitext(fname)
            name = destination_table
            if '.' in destination_table:
                destination_table = destination_table.split('.')[0]
            try:
                obj = get_cdm_class(destination_table).from_df(inputs[fname],name=name)
            except KeyError:
                cdm.logger.warning(f"Not loading {fname}, this is not a valid CDM Table")
                continue
            cdm.add(obj)
        return cdm

    def __del__(self):
        self.__df_map.clear()
        del self.__df_map
        self.__objects.clear()
        del self.__objects


    def __getitem__(self,key):
        """
        Ability lookup processed objects from the CDM
        Example:
            cdm = CommonDataModel()
            ...
            cdm.process()
            ...
            person = cdm['person']
        Args:
            key (str): The name of the cdm table to be returned
        Returns:
            pandas.DataFrame if a processed object is found, otherwise returns None
        """
        if key not in self.__df_map.keys():
            return None
        else:
            return self.__df_map[key]

    def __setitem__(self,key,obj):
        """
        Registration of a new dataframe for a new object
        Args:
            key (str) : name of the CDM table (e.g. "person")
            obj (pandas.DataFrame) : dataframe to refer to 
        """
        self.logger.debug(f"creating {obj} for {key}")
        self.__df_map[key] = obj

    def print(self):
        for name in self.keys():
            print (self[name].dropna(axis=1))

    def add(self,obj):
        """
        Function to add a new CDM table (object) to the current model
        Args:
            obj (DestinationTable) : CDM Table to be registered with the class
        """
        if obj._type not in self.__objects:
            self.__objects[obj._type] = {}

        if obj.name in self.__objects[obj._type].keys():
            raise Exception(f"Object called {obj.name} already exists")

        obj.cdm = self
        obj.format_level = self.format_level

        self.__objects[obj._type][obj.name] = obj
        self.logger.info(f"Added {obj.name} of type {obj._type}")


    def create_and_add_objects(self,config):
        #loop over the cdm object types defined in the configuration
        #e.g person, measurement etc..
        for destination_table,rules_set in config['cdm'].items():
            #loop over each object instance in the rule set
            #for example, condition_occurrence may have multiple rulesx
            #for multiple condition_ocurrences e.g. Headache, Fever ..
            for name,rules in rules_set.items():
                #make a new object for the cdm object
                #Example:
                # destination_table : person
                # get_cdm_class returns <Person>
                # obj : Person()
                obj = get_cdm_class(destination_table)()
                #set the name of the object
                obj.set_name(name)

                #Build a lambda function that will get executed during run time
                #and will be able to apply these rules to the inputs that are loaded
                #(this is useful when chunk)
                obj.define = lambda x,rules=rules : carrot.tools.apply_rules(x,rules,inputs=self.inputs)

                #register this object with the CDM model, so it can be processed
                self.add(obj)


    def add_analysis(self,func,_id=None):
        if _id is None:
            _id = hex(id(func))
        self.__analyses[_id] = func

    def get_analyses(self):
        return self.__analyses
    def get_analysis(self,key):
        return self.__analyses[key]

    def run_analysis(self,f):
        return f(self)

    def run_analyses(self,analyses=None,max_workers=4):

        def msg(x):
            self.logger.info(f"finished with {x}")
            self.logger.debug(x.result())

        start = time()

        with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
            futures = {}
            for name,f in self.__analyses.items():
                self.logger.info(f"Start thread for {f} ")
                future = executor.submit(f, self)
                future.add_done_callback(msg)
                futures[future] = name

            while True:
                status = {futures[f]:{'status':'running' if f.running() else 'done' if f.done() else 'waiting'} for f in futures}
                self.logger.debug(json.dumps(status,indent=6))
                if all([f.done() for f in futures]):
                    break
                sleep(1)


        results = {}
        for future in concurrent.futures.as_completed(futures):
            _id = futures[future]
            results[_id] = future.result()

        end = time() - start
        self.logger.info(f"Running analyses took {end} seconds")
        return results

    def find_one(self,config,cols=None,dropna=False):
        return self.filter(config,cols,dropna).sample(frac=1).iloc[0]

    def find(self,config,cols=None,dropna=False):
        return self.filter(config,cols,dropna)

    def _filter(self,df,filters):
        ops = {
            '>': operator.gt,
            '<': operator.lt,
            '>=': operator.ge,
            '<=': operator.le,
            '==': operator.eq
        }


        if not isinstance(filters,dict):
            raise NotImplementedError("filter must be a 'dict' .")

        for col,value in filters.items():
            if isinstance(value,dict):
                for op_str,val in value.items():
                    df = df[ops[op_str](df[col],val)]                        
            else:
                df = df[df[col] == value]
        return df


    def filter(self,config,cols=None,dropna=False):
        retval = copy.deepcopy(self)
        for table,spec in config.items():
            df = retval[table]
            if isinstance(df,DestinationTable):
                df = df.get_df()
            for col,func in spec.items():
                df = df[df[col].apply(func)]

            retval[table] = df

        if dropna:
            retval = retval.dropna(axis=1)
        if cols is not None:
            index = retval.index.name
            retval = retval.reset_index()
            retval = retval[[col for col,keep in cols.items() if keep]]
            if index in retval.columns:
                retval = retval.set_index(index)

        return retval


        # for obj in config:
        #     if isinstance(obj,str):
        #         df = self[obj].get_df()
        #         if df.index.name != 'person_id':
        #             df = df.set_index('person_id')
        #     elif isinstance(obj,dict):
        #         for key,value in obj.items():
        #             print (self[key])
        #             df = self[key]
        #             df = self._filter(df,value)
        #             if df.index.name != 'person_id':
        #                 df = df.set_index('person_id')
        #             if retval is None:
        #                 retval = df
        #             else:
        #                 retval = retval.merge(df,left_index=True,right_index=True)
        #     else:
        #         raise NotImplementedError("need to pass a json object to filter()")


    def get_all_objects(self):
        return [ obj for collection in self.__objects.values() for obj in collection.values()]


    def get_start_index(self,destination_table):
        self.logger.debug(f'getting start index for {destination_table}')

        if self.indexing_conf == None or not self.indexing_conf :
            self.logger.debug(f"no indexing specified, so starting the index for {destination_table} from 1")
            return 1

        if destination_table in self.indexing_conf:
            return int(self.indexing_conf[destination_table])
        else:
            self.logger.warning(self.indexing_conf)
            self.logger.warning("indexing configuration has be parsed "
                                f"but this table ({destination_table}) "
                                "has not be passed, so starting from 1")
            return 1

    def clear_objects(self,destination_table=None):
        if destination_table:
            self.__objects[destination_table].clear()
        else:
            for destination_table in self.__objects:
                self.__objects[destination_table].clear()

    def get_objects(self,destination_table=None):
        """
        For a given destination table:
        * Retrieve all associated objects that have been registered with the class

        Args:
            destination_table (str) : name of a destination (CDM) table e.g. "person"
        Returns:
            list : a list of destination table objects 
                   e.g. [<person_0>, <person_1>] 
                   which would be objects for male and female mapping
        """
        if destination_table == None:
            return self.__objects

        self.logger.debug(f"looking for {destination_table}")
        if destination_table not in self.__objects.keys():
            self.logger.error(f"Trying to obtain the table '{destination_table}', but cannot find any objects")
            raise Exception("Something wrong!")

        return [
            obj
            for obj in self.__objects[destination_table].values()
        ]


    def mask_person_id(self,df,destination_table):
        """
        Given a dataframe object, apply a masking map on the person_id, if one has been created
        Args:
            df (pandas.Dataframe) : input pandas dataframe
        Returns:
            pandas.Dataframe: modified dataframe with person id masked
        """

        if 'person_id' in df.columns:
            #if masker has not been defined, define it
            if destination_table == 'person':
                if self.person_id_masker is not None:
                    start_index = int(list(self.person_id_masker.values())[-1]) + 1
                    new = False
                else:
                    self.person_id_masker = {}
                    start_index = self.get_start_index(destination_table)
                    new = True

                person_id_masker = {}
                for i,x in enumerate(df['person_id'].unique()):
                    index = i+start_index
                    if x in self.person_id_masker:
                        existing_index = self.person_id_masker[x]
                        self.logger.error(f"'{x}' already found in the person_id_masker")
                        self.logger.error(f"'{existing_index}' assigned to this already")
                        self.logger.error(f"was trying to set '{index}'")
                        self.logger.error(f"Most likely cause is this is duplicate data!")
                        raise PersonExists('Duplicate person found!')
                    person_id_masker[x] = index

                self.person_id_masker.update(person_id_masker)

                if self.outputs:
                    dfp = pd.DataFrame(((s,t)
                                        for t,s in person_id_masker.items()),
                                       columns=['SOURCE_SUBJECT','TARGET_SUBJECT'])

                    mode = 'w' if new else 'a'
                    self.outputs.write(f"person_ids",dfp,mode)

            #apply the masking
            if self.person_id_masker is None:
                raise Exception(f"Person ID masking cannot be performed on"
                                f" {destination_table} as no masker based on a person table has been defined!")

            nbefore = len(df['person_id'])
            df['person_id'] = df['person_id'].map(self.person_id_masker)

            self.logger.debug(f"Just masked person_id using integers")
            if destination_table != 'person':
                df.dropna(subset=['person_id'],inplace=True)
                nafter = len(df['person_id'])
                ndiff = nbefore - nafter
                df.attrs['valid_person_id'] = {'before':nbefore,'after':nafter}
                #if rows have been removed
                if ndiff>0:
                    self.logger.error("There are person_ids in this table that are not in the output person table!")
                    self.logger.error("Either they are not in the original data, or while creating the person table, ")
                    self.logger.error("studies have been removed due to lack of required fields, such as birthdate.")
                    self.logger.error(f"{nafter}/{nbefore} were good, {ndiff} studies are removed.")

        return df

    def count_objects(self):
        """
        For each CDM (destination) table, count the number of objects associated
        e.g.
        {
           "observation": 6,
           "condition_occurrence": 1,
           "person": 2
        }
        """
        count_map = json.dumps({
            key: len(obj.keys())
            for key,obj in self.__objects.items()
        },indent=6)
        self.logger.info(f"Number of objects to process for each table...\n{count_map}")

    def keys(self):
        """
        For cdm.keys(), return the keys of which objects have been mapped.
        Hence which CDM table dataframes have been created.
        This should be used AFTER cdm.process() has been run, which creates the dataframes.
        """
        return self.__df_map.keys()

    def objects(self):
        """
        Method to retrieve the input objects to the CDM 
        """
        return self.__objects


    def process(self,object_list=None,conserve_memory=False):
        """
        Process chunked data, processes as follows
        * While the chunking of is not yet finished
        * Loop over all CDM tables (via execution order) 
          and process the table (see process_table), returning a dataframe
        * Register the retrieve dataframe with the model  
        * For the current chunk slice, save the data/logs to files
        * Retrieve the next chunk of data

        """
        self.execution_order = self.get_execution_order()
        self.logger.info(f"Starting processing in order: {self.execution_order}")
        self.count_objects()

        for destination_table in self.execution_order:
            first = True
            i = 0
            while True:                
                df_generator = self.process_table(destination_table,object_list=object_list)
                ntables = 0
                nrows = 0
                dfs = []
                #print (self.drop_duplicates)
                for j,obj in enumerate(df_generator):

                    df = obj.get_df()
                    ntables +=1
                    nrows += len(df)
                    if conserve_memory and self.save_files:
                        mode = None if first else 'a'
                        self.save_dataframe(destination_table,df,mode=mode)
                        first = False
                        obj.clear()
                        del df
                        df = None
                    else:
                        dfs.append(df)

                if not conserve_memory:
                    df = pd.concat(dfs,ignore_index=True)#.sort_values(df.columns[0])
                    if self.save_files:
                        if self.drop_duplicates and destination_table != 'person':
                            nbefore = len(df)
                            df_hash = pd.util.hash_pandas_object(df.drop(df.columns[0],axis=1),index=False)
                            df_temp = df[df_hash.duplicated(keep=False)].head(10).dropna(axis=1)
                            df = df[~df_hash.duplicated()]
                            nafter = len(df)
                            ndiff = nbefore - nafter
                            if ndiff>0:
                                self.logger.error(f"Removed {ndiff} row(s) due to duplicates found when merging {destination_table}")
                                self.logger.warning("Example duplicates...")
                                self.logger.warning(df_temp.set_index(df_temp.columns[0]))

                        mode = None if first else 'a'
                        self.save_dataframe(destination_table,df,mode=mode)
                        first = False

                    for col in df.columns:
                        if col.endswith("_id"):
                            df[col] = df[col].astype(float).astype(pd.Int64Dtype())
                    if df.index.name == 'index' or df.index.name is None:
                        df = df.set_index(df.columns[0])

                    self[destination_table] = df


                self.logger.info(f'finalised {destination_table} on iteration {i} producing {nrows} rows from {ntables} tables')

                #move onto the next iteration                
                i+=1

                if self.inputs:
                    try:
                        self.inputs.next()
                    except StopIteration:
                        break
                else:
                    break

            #if inputs are defined, and we havent just finished the last table,
            #reset the inputs
            if self.inputs and not destination_table == self.execution_order[-1]:
                self.inputs.reset()

        #for destination_table in self.execution_order:
        #    index = self.get_start_index(destination_table)
        #    print (index)


    def process_simult(self,object_list=None,conserve_memory=False):
        """
        process simulataneously
        """
        self.execution_order = self.get_execution_order()
        self.logger.info(f"Starting processing in order: {self.execution_order}")
        self.count_objects()
        i=0
        while True:
            for destination_table in self.execution_order:
                df_generator = self.process_table(destination_table,object_list=object_list)
                ntables = 0
                nrows = 0
                dfs = []
                for j,obj in enumerate(df_generator):
                    df = obj.get_df()

                    ntables +=1
                    nrows += len(df)
                    if self.save_files:
                        mode = None if j==0 else 'a'
                        self.save_dataframe(destination_table,df,mode=mode)
                        first = False
                    if not conserve_memory:
                        dfs.append(df)
                    #else:
                    #    obj.clear()
                    #    del df
                    #    df = None
                if not conserve_memory:
                    self[destination_table] = pd.concat(dfs,ignore_index=True)

            if self.inputs:
                try:
                    self.inputs.next()
                except StopIteration:
                    break
            else:
                break
            i+=1



    def get_tables(self):
        return list(self.__objects.keys())

    def get_execution_order(self):
        if not self.execution_order:
            self.execution_order = sorted(self.__objects.keys(), key=lambda x: x != 'person')
        return self.execution_order

    def set_execution_order(self,order):
        self.execution_order = order

    def process_table(self,destination_table,object_list=None):
        """
        Process a CDM (destination) table. The method proceeds as follows:
        * Given a destination table name e.g. 'person' 
        * Retrieve all objects belonging to the given CDM table (e.g. <person_0, person_1>)
        * Loop over each object
        * Retrieve a dataframe for that object given it's definition/rules (see get_df)
        * Concatenate all retrieve dataframes together by stacking on top of each other vertically
        * Create new indexes for the primary column so the go from 1-N 

        Args:
            destination_table (str) : name of a destination table to process (e.g. 'person')
            object_list (list) : [optional] list of objects to process
        Returns:
            list(pandas.Dataframe): a dataframes in the CDM format for this destination table
        """
        objects = self.get_objects(destination_table)
        if object_list:
            objects = [obj for obj in object_list if obj in objects]

        nobjects = len(objects)
        extra = ""
        if nobjects>1:
            extra="s"
        self.logger.info(f"for {destination_table}: found {nobjects} object{extra}")

        if len(objects) == 0:
            yield None

        #execute them all
        dfs = []
        self.logger.info(f"working on {destination_table}")
        logs = {'objects':{}}

        if destination_table not in self.logs['meta']['total_data_processed']:
            self.logs['meta']['total_data_processed'][destination_table] = 0

        nrows_processed = self.logs['meta']['total_data_processed'][destination_table]

        for i,obj in enumerate(objects):
            self.logger.info(f"starting on {obj.name}")

            start_index = self.get_start_index(destination_table)
            start_index += nrows_processed

            #force_rebuild=True,
            df = obj.get_df(start_index=start_index)

            self.logger.info(f"finished {obj.name} ({hex(id(df))}) "
                             f"... {i+1}/{len(objects)} completed, {len(df)} rows") 
            if len(df) == 0:
                self.logger.warning(f".. no outputs were found ")
                continue

            if self.do_mask_person_id:
                df = self.mask_person_id(df,destination_table)

            obj._meta.update(df.attrs)
            nrows_processed += len(df)
            self.logs['meta']['total_data_processed'][destination_table] = nrows_processed
            if destination_table not in self.logs:
                self.logs[destination_table] = {}

            self.logs[destination_table][hex(id(df))] = obj._meta

            obj.set_df(df)
            yield obj

    def save_dataframe(self,table,df=None,mode=None):
        if self.outputs:
            _id = hex(id(df))
            self.logger.info(f"saving dataframe ({_id}) to {self.outputs}")
            self.outputs.write(table,df,mode)
        else:
            self.logger.info(f"called save_dateframe but outputs are not defined. save_files: {self.save_files}")

    def set_person_id_map(self,person_id_map):
        self.person_id_masker = person_id_map

    def set_indexing_map(self,indexing):
        self.indexing_conf = indexing

    def set_outfile_separator(self,sep):
        """
        Set which separator to use, e.g. ',' or '\t' 

        Args:
            sep (str): which separator to use when writing csv (tsv) files
        """
        self._outfile_separator = sep

    def set_indexing(self,index_map,strict_check=False):
        """
        Create indexes on input files which would allow rules to use data from 
        different input tables.

        Args:
            index_map (dict): a map between the filename and what should be the column used for indexing 

        """
        if self.inputs == None:
            raise NoInputFiles('Trying to indexing before any inputs have been setup')

        for key,index in index_map.items():
            if key not in self.inputs.keys():
                self.logger.warning(f"trying to set index '{index}' for '{key}' but this has not been loaded as an inputs!")
                continue

            if index not in self.inputs[key].columns:
                self.logger.error(f"trying to set index '{index}' on dataset '{key}', but this index is not in the columns! something really wrong!")
                continue
            self.inputs[key].index = self.inputs[key][index].rename('index') 

__getitem__(key)

Ability lookup processed objects from the CDM Example: cdm = CommonDataModel() ... cdm.process() ... person = cdm['person'] Args: key (str): The name of the cdm table to be returned Returns: pandas.DataFrame if a processed object is found, otherwise returns None

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
def __getitem__(self,key):
    """
    Ability lookup processed objects from the CDM
    Example:
        cdm = CommonDataModel()
        ...
        cdm.process()
        ...
        person = cdm['person']
    Args:
        key (str): The name of the cdm table to be returned
    Returns:
        pandas.DataFrame if a processed object is found, otherwise returns None
    """
    if key not in self.__df_map.keys():
        return None
    else:
        return self.__df_map[key]

__init__(name=None, omop_version='5.3.1', outputs=None, save_files=True, inputs=None, use_profiler=False, format_level=None, do_mask_person_id=True, drop_duplicates=True, automatically_fill_missing_columns=True)

CommonDataModel class initialisation Args: name (str): Give a name for the class to appear in the logging output_folder (str): Path of where the output tsv/csv files should be written to. The default is to save to a folder in the current directory called 'output_data'. inputs (dict or DataCollection): inputs can be a dictionary mapping file names to pandas dataframes, or can be a DataCollection object use_profiler (bool): Turn on/off profiling of the CPU/Memory of running the current process. The default is set to false.

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
def __init__(self, name=None, omop_version='5.3.1',
             outputs = None,
             save_files=True,
             inputs=None,
             use_profiler=False,
             format_level=None,
             do_mask_person_id=True,
             drop_duplicates=True,
             automatically_fill_missing_columns=True):
    """
    CommonDataModel class initialisation 
    Args:
        name (str): Give a name for the class to appear in the logging
        output_folder (str): Path of where the output tsv/csv files should be written to.
                             The default is to save to a folder in the current directory
                             called 'output_data'.
        inputs (dict or DataCollection): inputs can be a dictionary mapping file names to pandas dataframes,
                                    or can be a DataCollection object
        use_profiler (bool): Turn on/off profiling of the CPU/Memory of running the current process. 
                             The default is set to false.
    """
    self.profiler = None
    name = self.__class__.__name__ if name is None else self.__class__.__name__ + "::" + name

    self.logger.info(f"CommonDataModel ({omop_version}) created with co-connect-tools version {carrot_version}")

    self.omop_version = omop_version

    self.drop_duplicates = drop_duplicates
    self.do_mask_person_id = do_mask_person_id
    self.execution_order = None

    if format_level == None:
        format_level = 1
    try:
        format_level = int(format_level)
    except ValueError:
        self.logger.error(f"You as specifying format_level='{format_level}' -- this should be an integer!")
        raise ValueError("format_level not set as an int ")

    self.format_level = FormatterLevel(format_level)
    self.profiler = None

    self.outputs = outputs
    self.save_files = save_files

    if use_profiler:
        self.logger.debug(f"Turning on cpu/memory profiling")
        self.profiler = Profiler(name=name)
        self.profiler.start()

    #perform some checks on the input data
    if isinstance(inputs,dict):
        self.logger.info("Running with an DataCollection object")
    elif isinstance(inputs,DataCollection):
        self.logger.info("Running with an DataCollection object")
    elif inputs is not None:
        _type = type(inputs).__name__
        raise BadInputObject(f"input object {inputs} is of type {_type}, which is not a valid input object")

    if inputs is not None:
        if hasattr(self,'inputs'):
            self.logger.warning("overwriting inputs")
        self.inputs = inputs
    elif not hasattr(self,'inputs'):
        self.inputs = None

    #register opereation tools
    self.tools = OperationTools()

    #allow rules to be generated automatically or not
    self.automatically_fill_missing_columns = automatically_fill_missing_columns
    if self.automatically_fill_missing_columns:
        self.logger.info(f"Turning on automatic cdm column filling")

    #define a person_id masker, if the person_id are to be masked
    if self.outputs:
        self.person_id_masker = self.outputs.load_global_ids()
        self.indexing_conf = self.outputs.load_indexing()
    else:
        self.person_id_masker = None
        self.indexing_conf = None

    #stores the final pandas dataframe for each CDM object
    # {
    #   'person':pandas.DataFrame,
    #   'measurement':pandas.DataFrame.
    #   ....
    #}
    self.__df_map = {}

    #stores the invididual objects associated to this model
    # {
    #     'observation':
    #     {
    #         'observation_0': <carrot.cdm.objects.observation.Observation object 0x000>,
    #         'observation_1': <carrot.cdm.objects.observation.Observation object 0x001>,
    #     },
    #     'measurement':
    #     {
    #         'measurement_0': <carrot.cdm.objects.measurement.Measurement object 0x000>,
    #         ...
    #     }
    #     ...
    # }
    self.__objects = {}
    #check if objects have already been registered with this class
    #via the decorator methods

    registered_objects = [
        getattr(self,name)
        for name in dir(self)
        if isinstance(getattr(self,name),DestinationTable)
    ]
    #if they have, then include them in this model
    for obj in registered_objects:
        obj.inputs = self.inputs
        self.add(obj)

    self.__analyses = {
        name:getattr(self,name)
        for name in dir(self)
        if isinstance(getattr(self,name),analysis)
    }

    #bookkeep some logs
    self.logs = {
        'meta':{
            'version': carrot_version,
            'created_by': getpass.getuser(),
            'created_at': strftime("%Y-%m-%dT%H%M%S", gmtime()),
            'dataset':name,
            'total_data_processed':{}
        }
    }

__setitem__(key, obj)

Registration of a new dataframe for a new object Args: key (str) : name of the CDM table (e.g. "person") obj (pandas.DataFrame) : dataframe to refer to

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
299
300
301
302
303
304
305
306
307
def __setitem__(self,key,obj):
    """
    Registration of a new dataframe for a new object
    Args:
        key (str) : name of the CDM table (e.g. "person")
        obj (pandas.DataFrame) : dataframe to refer to 
    """
    self.logger.debug(f"creating {obj} for {key}")
    self.__df_map[key] = obj

add(obj)

Function to add a new CDM table (object) to the current model Args: obj (DestinationTable) : CDM Table to be registered with the class

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
def add(self,obj):
    """
    Function to add a new CDM table (object) to the current model
    Args:
        obj (DestinationTable) : CDM Table to be registered with the class
    """
    if obj._type not in self.__objects:
        self.__objects[obj._type] = {}

    if obj.name in self.__objects[obj._type].keys():
        raise Exception(f"Object called {obj.name} already exists")

    obj.cdm = self
    obj.format_level = self.format_level

    self.__objects[obj._type][obj.name] = obj
    self.logger.info(f"Added {obj.name} of type {obj._type}")

close()

Class destructor

Stops the profiler from running before deleting self

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
def close(self):
    """
    Class destructor:
          Stops the profiler from running before deleting self
    """


    self.logger.info(json.dumps(self.logs['meta'],indent=6))
    if self.outputs:
        self.outputs.write_meta(self.logs)
        self.outputs.finalise()

    if not hasattr(self,'profiler'):
        return
    if self.profiler:
        self.profiler.stop()
        df_profile = self.profiler.get_df()
        f_out = self.output_folder
        f_out = f'{f_out}{os.path.sep}logs{os.path.sep}'
        if not os.path.exists(f'{f_out}'):
            self.logger.info(f'making output folder {f_out}')
            os.makedirs(f'{f_out}')

        date = self.logs['meta']['created_at']
        fname = f'{f_out}{os.path.sep}statistics_{date}.csv'
        df_profile.to_csv(fname)
        self.logger.info(f"Writen the memory/cpu statistics to {fname}")
        self.logger.info("Finished")

count_objects()

For each CDM (destination) table, count the number of objects associated e.g. { "observation": 6, "condition_occurrence": 1, "person": 2 }

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
def count_objects(self):
    """
    For each CDM (destination) table, count the number of objects associated
    e.g.
    {
       "observation": 6,
       "condition_occurrence": 1,
       "person": 2
    }
    """
    count_map = json.dumps({
        key: len(obj.keys())
        for key,obj in self.__objects.items()
    },indent=6)
    self.logger.info(f"Number of objects to process for each table...\n{count_map}")

from_existing(**kwargs) classmethod

Initialise the CDM model from existing data in the CDM format

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
@classmethod
def from_existing(cls,**kwargs):
    """
    Initialise the CDM model from existing data in the CDM format
    """
    cdm = cls(**kwargs)
    if 'inputs' not in kwargs:
        raise NoInputFiles("you need to specify some inputs")
    inputs = kwargs['inputs']
    #loop over all input names
    for fname in inputs.keys():
        #obtain the name of the destination table
        #e.g fname='person.tsv' we want 'person'
        destination_table,_ = os.path.splitext(fname)
        name = destination_table
        if '.' in destination_table:
            destination_table = destination_table.split('.')[0]
        try:
            obj = get_cdm_class(destination_table).from_df(inputs[fname],name=name)
        except KeyError:
            cdm.logger.warning(f"Not loading {fname}, this is not a valid CDM Table")
            continue
        cdm.add(obj)
    return cdm

get_objects(destination_table=None)

For a given destination table: * Retrieve all associated objects that have been registered with the class

Parameters:

Name Type Description Default
destination_table str)

name of a destination (CDM) table e.g. "person"

None

Returns: list : a list of destination table objects e.g. [, ] which would be objects for male and female mapping

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
def get_objects(self,destination_table=None):
    """
    For a given destination table:
    * Retrieve all associated objects that have been registered with the class

    Args:
        destination_table (str) : name of a destination (CDM) table e.g. "person"
    Returns:
        list : a list of destination table objects 
               e.g. [<person_0>, <person_1>] 
               which would be objects for male and female mapping
    """
    if destination_table == None:
        return self.__objects

    self.logger.debug(f"looking for {destination_table}")
    if destination_table not in self.__objects.keys():
        self.logger.error(f"Trying to obtain the table '{destination_table}', but cannot find any objects")
        raise Exception("Something wrong!")

    return [
        obj
        for obj in self.__objects[destination_table].values()
    ]

keys()

For cdm.keys(), return the keys of which objects have been mapped. Hence which CDM table dataframes have been created. This should be used AFTER cdm.process() has been run, which creates the dataframes.

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
609
610
611
612
613
614
615
def keys(self):
    """
    For cdm.keys(), return the keys of which objects have been mapped.
    Hence which CDM table dataframes have been created.
    This should be used AFTER cdm.process() has been run, which creates the dataframes.
    """
    return self.__df_map.keys()

mask_person_id(df, destination_table)

Given a dataframe object, apply a masking map on the person_id, if one has been created Args: df (pandas.Dataframe) : input pandas dataframe Returns: pandas.Dataframe: modified dataframe with person id masked

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
def mask_person_id(self,df,destination_table):
    """
    Given a dataframe object, apply a masking map on the person_id, if one has been created
    Args:
        df (pandas.Dataframe) : input pandas dataframe
    Returns:
        pandas.Dataframe: modified dataframe with person id masked
    """

    if 'person_id' in df.columns:
        #if masker has not been defined, define it
        if destination_table == 'person':
            if self.person_id_masker is not None:
                start_index = int(list(self.person_id_masker.values())[-1]) + 1
                new = False
            else:
                self.person_id_masker = {}
                start_index = self.get_start_index(destination_table)
                new = True

            person_id_masker = {}
            for i,x in enumerate(df['person_id'].unique()):
                index = i+start_index
                if x in self.person_id_masker:
                    existing_index = self.person_id_masker[x]
                    self.logger.error(f"'{x}' already found in the person_id_masker")
                    self.logger.error(f"'{existing_index}' assigned to this already")
                    self.logger.error(f"was trying to set '{index}'")
                    self.logger.error(f"Most likely cause is this is duplicate data!")
                    raise PersonExists('Duplicate person found!')
                person_id_masker[x] = index

            self.person_id_masker.update(person_id_masker)

            if self.outputs:
                dfp = pd.DataFrame(((s,t)
                                    for t,s in person_id_masker.items()),
                                   columns=['SOURCE_SUBJECT','TARGET_SUBJECT'])

                mode = 'w' if new else 'a'
                self.outputs.write(f"person_ids",dfp,mode)

        #apply the masking
        if self.person_id_masker is None:
            raise Exception(f"Person ID masking cannot be performed on"
                            f" {destination_table} as no masker based on a person table has been defined!")

        nbefore = len(df['person_id'])
        df['person_id'] = df['person_id'].map(self.person_id_masker)

        self.logger.debug(f"Just masked person_id using integers")
        if destination_table != 'person':
            df.dropna(subset=['person_id'],inplace=True)
            nafter = len(df['person_id'])
            ndiff = nbefore - nafter
            df.attrs['valid_person_id'] = {'before':nbefore,'after':nafter}
            #if rows have been removed
            if ndiff>0:
                self.logger.error("There are person_ids in this table that are not in the output person table!")
                self.logger.error("Either they are not in the original data, or while creating the person table, ")
                self.logger.error("studies have been removed due to lack of required fields, such as birthdate.")
                self.logger.error(f"{nafter}/{nbefore} were good, {ndiff} studies are removed.")

    return df

objects()

Method to retrieve the input objects to the CDM

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
617
618
619
620
621
def objects(self):
    """
    Method to retrieve the input objects to the CDM 
    """
    return self.__objects

process(object_list=None, conserve_memory=False)

Process chunked data, processes as follows * While the chunking of is not yet finished * Loop over all CDM tables (via execution order) and process the table (see process_table), returning a dataframe * Register the retrieve dataframe with the model
* For the current chunk slice, save the data/logs to files * Retrieve the next chunk of data

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
def process(self,object_list=None,conserve_memory=False):
    """
    Process chunked data, processes as follows
    * While the chunking of is not yet finished
    * Loop over all CDM tables (via execution order) 
      and process the table (see process_table), returning a dataframe
    * Register the retrieve dataframe with the model  
    * For the current chunk slice, save the data/logs to files
    * Retrieve the next chunk of data

    """
    self.execution_order = self.get_execution_order()
    self.logger.info(f"Starting processing in order: {self.execution_order}")
    self.count_objects()

    for destination_table in self.execution_order:
        first = True
        i = 0
        while True:                
            df_generator = self.process_table(destination_table,object_list=object_list)
            ntables = 0
            nrows = 0
            dfs = []
            #print (self.drop_duplicates)
            for j,obj in enumerate(df_generator):

                df = obj.get_df()
                ntables +=1
                nrows += len(df)
                if conserve_memory and self.save_files:
                    mode = None if first else 'a'
                    self.save_dataframe(destination_table,df,mode=mode)
                    first = False
                    obj.clear()
                    del df
                    df = None
                else:
                    dfs.append(df)

            if not conserve_memory:
                df = pd.concat(dfs,ignore_index=True)#.sort_values(df.columns[0])
                if self.save_files:
                    if self.drop_duplicates and destination_table != 'person':
                        nbefore = len(df)
                        df_hash = pd.util.hash_pandas_object(df.drop(df.columns[0],axis=1),index=False)
                        df_temp = df[df_hash.duplicated(keep=False)].head(10).dropna(axis=1)
                        df = df[~df_hash.duplicated()]
                        nafter = len(df)
                        ndiff = nbefore - nafter
                        if ndiff>0:
                            self.logger.error(f"Removed {ndiff} row(s) due to duplicates found when merging {destination_table}")
                            self.logger.warning("Example duplicates...")
                            self.logger.warning(df_temp.set_index(df_temp.columns[0]))

                    mode = None if first else 'a'
                    self.save_dataframe(destination_table,df,mode=mode)
                    first = False

                for col in df.columns:
                    if col.endswith("_id"):
                        df[col] = df[col].astype(float).astype(pd.Int64Dtype())
                if df.index.name == 'index' or df.index.name is None:
                    df = df.set_index(df.columns[0])

                self[destination_table] = df


            self.logger.info(f'finalised {destination_table} on iteration {i} producing {nrows} rows from {ntables} tables')

            #move onto the next iteration                
            i+=1

            if self.inputs:
                try:
                    self.inputs.next()
                except StopIteration:
                    break
            else:
                break

        #if inputs are defined, and we havent just finished the last table,
        #reset the inputs
        if self.inputs and not destination_table == self.execution_order[-1]:
            self.inputs.reset()

process_simult(object_list=None, conserve_memory=False)

process simulataneously

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
def process_simult(self,object_list=None,conserve_memory=False):
    """
    process simulataneously
    """
    self.execution_order = self.get_execution_order()
    self.logger.info(f"Starting processing in order: {self.execution_order}")
    self.count_objects()
    i=0
    while True:
        for destination_table in self.execution_order:
            df_generator = self.process_table(destination_table,object_list=object_list)
            ntables = 0
            nrows = 0
            dfs = []
            for j,obj in enumerate(df_generator):
                df = obj.get_df()

                ntables +=1
                nrows += len(df)
                if self.save_files:
                    mode = None if j==0 else 'a'
                    self.save_dataframe(destination_table,df,mode=mode)
                    first = False
                if not conserve_memory:
                    dfs.append(df)
                #else:
                #    obj.clear()
                #    del df
                #    df = None
            if not conserve_memory:
                self[destination_table] = pd.concat(dfs,ignore_index=True)

        if self.inputs:
            try:
                self.inputs.next()
            except StopIteration:
                break
        else:
            break
        i+=1

process_table(destination_table, object_list=None)

Process a CDM (destination) table. The method proceeds as follows: * Given a destination table name e.g. 'person' * Retrieve all objects belonging to the given CDM table (e.g. ) * Loop over each object * Retrieve a dataframe for that object given it's definition/rules (see get_df) * Concatenate all retrieve dataframes together by stacking on top of each other vertically * Create new indexes for the primary column so the go from 1-N

Parameters:

Name Type Description Default
destination_table str)

name of a destination table to process (e.g. 'person')

required
object_list list)

[optional] list of objects to process

None

Returns: list(pandas.Dataframe): a dataframes in the CDM format for this destination table

Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
def process_table(self,destination_table,object_list=None):
    """
    Process a CDM (destination) table. The method proceeds as follows:
    * Given a destination table name e.g. 'person' 
    * Retrieve all objects belonging to the given CDM table (e.g. <person_0, person_1>)
    * Loop over each object
    * Retrieve a dataframe for that object given it's definition/rules (see get_df)
    * Concatenate all retrieve dataframes together by stacking on top of each other vertically
    * Create new indexes for the primary column so the go from 1-N 

    Args:
        destination_table (str) : name of a destination table to process (e.g. 'person')
        object_list (list) : [optional] list of objects to process
    Returns:
        list(pandas.Dataframe): a dataframes in the CDM format for this destination table
    """
    objects = self.get_objects(destination_table)
    if object_list:
        objects = [obj for obj in object_list if obj in objects]

    nobjects = len(objects)
    extra = ""
    if nobjects>1:
        extra="s"
    self.logger.info(f"for {destination_table}: found {nobjects} object{extra}")

    if len(objects) == 0:
        yield None

    #execute them all
    dfs = []
    self.logger.info(f"working on {destination_table}")
    logs = {'objects':{}}

    if destination_table not in self.logs['meta']['total_data_processed']:
        self.logs['meta']['total_data_processed'][destination_table] = 0

    nrows_processed = self.logs['meta']['total_data_processed'][destination_table]

    for i,obj in enumerate(objects):
        self.logger.info(f"starting on {obj.name}")

        start_index = self.get_start_index(destination_table)
        start_index += nrows_processed

        #force_rebuild=True,
        df = obj.get_df(start_index=start_index)

        self.logger.info(f"finished {obj.name} ({hex(id(df))}) "
                         f"... {i+1}/{len(objects)} completed, {len(df)} rows") 
        if len(df) == 0:
            self.logger.warning(f".. no outputs were found ")
            continue

        if self.do_mask_person_id:
            df = self.mask_person_id(df,destination_table)

        obj._meta.update(df.attrs)
        nrows_processed += len(df)
        self.logs['meta']['total_data_processed'][destination_table] = nrows_processed
        if destination_table not in self.logs:
            self.logs[destination_table] = {}

        self.logs[destination_table][hex(id(df))] = obj._meta

        obj.set_df(df)
        yield obj

set_indexing(index_map, strict_check=False)

Create indexes on input files which would allow rules to use data from different input tables.

Parameters:

Name Type Description Default
index_map dict

a map between the filename and what should be the column used for indexing

required
Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
def set_indexing(self,index_map,strict_check=False):
    """
    Create indexes on input files which would allow rules to use data from 
    different input tables.

    Args:
        index_map (dict): a map between the filename and what should be the column used for indexing 

    """
    if self.inputs == None:
        raise NoInputFiles('Trying to indexing before any inputs have been setup')

    for key,index in index_map.items():
        if key not in self.inputs.keys():
            self.logger.warning(f"trying to set index '{index}' for '{key}' but this has not been loaded as an inputs!")
            continue

        if index not in self.inputs[key].columns:
            self.logger.error(f"trying to set index '{index}' on dataset '{key}', but this index is not in the columns! something really wrong!")
            continue
        self.inputs[key].index = self.inputs[key][index].rename('index') 

set_outfile_separator(sep)

Set which separator to use, e.g. ',' or ' '

Parameters:

Name Type Description Default
sep str

which separator to use when writing csv (tsv) files

required
Source code in docs/CaRROT-CDM/source_code/carrot/cdm/model.py
850
851
852
853
854
855
856
857
def set_outfile_separator(self,sep):
    """
    Set which separator to use, e.g. ',' or '\t' 

    Args:
        sep (str): which separator to use when writing csv (tsv) files
    """
    self._outfile_separator = sep