encyclopedia.py 46.7 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
# Copyright 2018 Markus Scheidgen
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an"AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

15
"""
16
The encyclopedia API of the nomad@FAIRDI APIs.
17
"""
18
import re
19
import math
20
import numpy as np
21

22
23
from flask_restplus import Resource, abort, fields, marshal
from flask import request
24
from elasticsearch_dsl import Search, Q, A
25
from elasticsearch_dsl.utils import AttrDict
26

27
from nomad import config, files
28
from nomad.units import ureg
Lauri Himanen's avatar
Lauri Himanen committed
29
from nomad.atomutils import get_hill_decomposition
30
from nomad.datamodel.datamodel import EntryArchive
31
from .api import api
32

33
ns = api.namespace("encyclopedia", description="Access encyclopedia metadata.")
34
35
re_formula = re.compile(r"([A-Z][a-z]?)(\d*)")

36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
material_prop_map = {
    # General
    "material_id": "encyclopedia.material.material_id",
    "formula": "encyclopedia.material.formula",
    "formula_reduced": "encyclopedia.material.formula_reduced",
    "system_type": "encyclopedia.material.material_type",
    # Bulk only
    "has_free_wyckoff_parameters": "encyclopedia.material.bulk.has_free_wyckoff_parameters",
    "strukturbericht_designation": "encyclopedia.material.bulk.strukturbericht_designation",
    "material_name": "encyclopedia.material.material_name",
    "bravais_lattice": "encyclopedia.material.bulk.bravais_lattice",
    "crystal_system": "encyclopedia.material.bulk.crystal_system",
    "point_group": "encyclopedia.material.bulk.point_group",
    "space_group_number": "encyclopedia.material.bulk.space_group_number",
    "space_group_international_short_symbol": "encyclopedia.material.bulk.space_group_international_short_symbol",
    "structure_prototype": "encyclopedia.material.bulk.structure_prototype",
    "structure_type": "encyclopedia.material.bulk.structure_type",
}
54
55


56
57
58
59
60
61
62
63
64
65
66
67
def rgetattr(obj, attr_name):
    """Used to perform attribute access based on a (possibly nested) attribute
    name given as string.
    """
    try:
        for attr in attr_name.split("."):
            obj = obj[attr]
    except KeyError:
        return None
    return obj


68
def get_es_doc_values(es_doc, mapping, keys=None):
69
70
    """Used to form a material definition for "materials/<material_id>" from
    the given ElasticSearch root document.
71
    """
72
73
74
    if keys is None:
        keys = mapping.keys()

75
    result = {}
76
    for key in keys:
77
        es_key = mapping[key]
78
        value = rgetattr(es_doc, es_key)
79
        result[key] = value
80
81
82
83
84

    return result


material_query = api.parser()
85
86
87
88
89
90
91
92
material_query.add_argument(
    "property",
    type=str,
    choices=tuple(material_prop_map.keys()),
    help="Optional single property to retrieve for the given material. If not specified, all properties will be returned.",
    location="args"
)
material_result = api.model("material_result", {
93
94
    # General
    "material_id": fields.String,
95
96
    "formula": fields.String,
    "formula_reduced": fields.String,
97
    "system_type": fields.String,
98
    "n_matches": fields.Integer,
99
    # Bulk only
100
    "has_free_wyckoff_parameters": fields.Boolean,
101
    "strukturbericht_designation": fields.String,
102
    "material_name": fields.String,
103
104
    "bravais_lattice": fields.String,
    "crystal_system": fields.String,
105
    "point_group": fields.String,
106
107
108
    "space_group_number": fields.Integer,
    "space_group_international_short_symbol": fields.String,
    "structure_prototype": fields.String,
109
110
    "structure_type": fields.String,
})
111
112


113
@ns.route("/materials/<string:material_id>")
114
class EncMaterialResource(Resource):
115
116
117
    @api.response(404, "The material does not exist")
    @api.response(200, "Metadata send", fields.Raw)
    @api.doc("material/<material_id>")
118
    @api.expect(material_query)
119
    @api.marshal_with(material_result, skip_none=True)
120
121
122
    def get(self, material_id):
        """Used to retrive basic information related to the specified material.
        """
123
124
125
126
        # Parse request arguments
        args = material_query.parse_args()
        prop = args.get("property", None)
        if prop is not None:
127
128
            keys = [prop]
            es_keys = [material_prop_map[prop]]
129
130
        else:
            keys = list(material_prop_map.keys())
131
            es_keys = list(material_prop_map.values())
132

133
134
135
136
137
138
139
140
141
        # Find the first public entry with this material id and take
        # information from there. In principle all other entries should have
        # the same information.
        s = Search(index=config.elastic.index_name)

        # Since we are looking for an exact match, we use filter context
        # together with term search for speed (instead of query context and
        # match search)
        query = Q(
142
            "bool",
143
            filter=[
144
145
146
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
147
148
149
            ]
        )
        s = s.query(query)
150

151
152
153
154
155
156
157
158
        # If a representative calculation is requested, all calculations are
        # returned in order to perform the scoring with a custom loop.
        # Otherwise, only one representative entry is returned.
        s = s.extra(**{
            "_source": {"includes": es_keys},
            "size": 10000,
            "collapse": {"field": "encyclopedia.material.material_id"},
        })
159
160
        response = s.execute()

161
        # No such material
162
        if len(response) == 0:
163
            abort(404, message="There is no material {}".format(material_id))
164

165
        # Add values from ES entry
166
        entry = response[0]
167
        result = get_es_doc_values(entry, material_prop_map, keys)
168

169
170
171
        return result, 200


172
range_query = api.model("range_query", {
173
174
175
    "max": fields.Float,
    "min": fields.Float,
})
176
177
178
materials_after = api.model("materials_after", {
    "materials": fields.String,
})
179
180
materials_query = api.model("materials_input", {
    "search_by": fields.Nested(api.model("search_query", {
181
182
        "exclusive": fields.Boolean(default=False),
        "formula": fields.String,
Lauri Himanen's avatar
Lauri Himanen committed
183
        "element": fields.String,
184
        "page": fields.Integer(default=1),
185
        "after": fields.Nested(materials_after, allow_null=True),
186
187
        "per_page": fields.Integer(default=25),
        "pagination": fields.Boolean,
188
        "mode": fields.String(default="aggregation"),
189
    })),
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
    "material_name": fields.List(fields.String),
    "structure_type": fields.List(fields.String),
    "space_group_number": fields.List(fields.Integer),
    "system_type": fields.List(fields.String),
    "crystal_system": fields.List(fields.String),
    "band_gap": fields.Nested(range_query, description="Band gap range in eV."),
    "band_gap_direct": fields.Boolean,
    "has_band_structure": fields.Boolean,
    "has_dos": fields.Boolean,
    "has_fermi_surface": fields.Boolean,
    "has_thermal_properties": fields.Boolean,
    "functional_type": fields.List(fields.String),
    "basis_set_type": fields.List(fields.String),
    "code_name": fields.List(fields.String),
    "mass_density": fields.Nested(range_query, description="Mass density range in kg / m ** 3."),
205
})
206
207
208
209
210
pages_result = api.model("page_info", {
    "per_page": fields.Integer,
    "total": fields.Integer,
    "page": fields.Integer,
    "pages": fields.Integer,
211
    "after": fields.Nested(materials_after),
212
213
})

214
215
materials_result = api.model("materials_result", {
    "total_results": fields.Integer(allow_null=False),
216
217
    "results": fields.List(fields.Nested(material_result, skip_none=True)),
    "pages": fields.Nested(pages_result, skip_none=True),
218
    "es_query": fields.String(allow_null=False),
219
220
221
})


222
@ns.route("/materials")
223
class EncMaterialsResource(Resource):
224
225
226
    @api.response(404, "No materials found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
227
    @api.expect(materials_query, validate=False)
228
    @api.marshal_with(materials_result, skip_none=True)
229
    @api.doc("materials")
230
231
232
233
234
235
236
237
238
239
240
241
242
243
    def post(self):
        """Used to query a list of materials with the given search options.
        """
        # Get query parameters as json
        try:
            data = marshal(request.get_json(), materials_query)
        except Exception as e:
            abort(400, message=str(e))

        filters = []
        must_nots = []
        musts = []

        # Add term filters
244
245
        filters.append(Q("term", published=True))
        filters.append(Q("term", with_embargo=False))
246
247
248
249
250
251
252

        def add_terms_filter(source, target, query_type="terms"):
            if data[source]:
                filters.append(Q(query_type, **{target: data[source]}))

        add_terms_filter("material_name", "encyclopedia.material.material_name")
        add_terms_filter("structure_type", "encyclopedia.material.bulk.structure_type")
253
        add_terms_filter("space_group_number", "encyclopedia.material.bulk.space_group_number")
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
        add_terms_filter("system_type", "encyclopedia.material.material_type")
        add_terms_filter("crystal_system", "encyclopedia.material.bulk.crystal_system")
        add_terms_filter("band_gap_direct", "encyclopedia.properties.band_gap_direct", query_type="term")
        add_terms_filter("functional_type", "encyclopedia.method.functional_type")
        add_terms_filter("basis_set_type", "dft.basis_set")
        add_terms_filter("code_name", "dft.code_name")

        # Add exists filters
        def add_exists_filter(source, target):
            param = data[source]
            if param is not None:
                query = Q("exists", field=target)
                if param is True:
                    filters.append(query)
                elif param is False:
                    must_nots.append(query)

        add_exists_filter("has_thermal_properties", "encyclopedia.properties.thermodynamical_properties")
        add_exists_filter("has_band_structure", "encyclopedia.properties.electronic_band_structure")
        add_exists_filter("has_dos", "encyclopedia.properties.electronic_dos")
        add_exists_filter("has_fermi_surface", "encyclopedia.properties.fermi_surface")

        # Add range filters
        def add_range_filter(source, target, source_unit=None, target_unit=None):
            param = data[source]
            query_dict = {}
            if param["min"] is not None:
                if source_unit is None and target_unit is None:
                    gte = param["min"]
                else:
                    gte = (param["min"] * source_unit).to(target_unit).magnitude
                query_dict["gte"] = gte
            if param["max"] is not None:
                if source_unit is None and target_unit is None:
                    lte = param["max"]
                else:
                    lte = (param["max"] * source_unit).to(target_unit).magnitude
                query_dict["lte"] = lte
            if len(query_dict) != 0:
                query = Q("range", **{target: query_dict})
                filters.append(query)

        add_range_filter("band_gap", "encyclopedia.properties.band_gap", ureg.eV, ureg.J)
        add_range_filter("mass_density", "encyclopedia.properties.mass_density")
298

299
300
        # Create query for elements or formula
        search_by = data["search_by"]
301
        mode = search_by["mode"]
302
        formula = search_by["formula"]
Lauri Himanen's avatar
Lauri Himanen committed
303
        elements = search_by["element"]
304
305
306
        exclusive = search_by["exclusive"]

        if formula is not None:
Lauri Himanen's avatar
Lauri Himanen committed
307
308
            # Here we determine a list of atom types. The types may occur
            # multiple times and at multiple places.
309
310
311
312
313
314
315
316
317
318
            element_list = []
            matches = re_formula.finditer(formula)
            for match in matches:
                groups = match.groups()
                symbol = groups[0]
                count = groups[1]
                if symbol != "":
                    if count == "":
                        element_list.append(symbol)
                    else:
Lauri Himanen's avatar
Lauri Himanen committed
319
320
321
322
323
                        element_list += [symbol] * int(count)

            # The given list of species is reformatted with the Hill system
            # into a query string. The counts are reduced by the greatest
            # common divisor.
324
            names, reduced_counts = get_hill_decomposition(element_list, reduced=True)
Lauri Himanen's avatar
Lauri Himanen committed
325
326
327
328
329
            query_string = []
            for name, count in zip(names, reduced_counts):
                if count == 1:
                    query_string.append(name)
                else:
330
                    query_string.append("{}{}".format(name, int(count)))
Lauri Himanen's avatar
Lauri Himanen committed
331
            query_string = " ".join(query_string)
332
333
334

            # With exclusive search we look for exact match
            if exclusive:
Lauri Himanen's avatar
Lauri Himanen committed
335
                filters.append(Q("term", **{"encyclopedia.material.species_and_counts.keyword": query_string}))
336
337
338
339
340
            # With non-exclusive search we look for match that includes at
            # least all parts of the formula, possibly even more.
            else:
                musts.append(Q(
                    "match",
Lauri Himanen's avatar
Lauri Himanen committed
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
                    encyclopedia__material__species_and_counts={"query": query_string, "operator": "and"}
                ))
        elif elements is not None:
            # The given list of species is reformatted with the Hill system into a query string
            species, _ = get_hill_decomposition(elements.split(","))
            query_string = " ".join(species)

            # With exclusive search we look for exact match
            if exclusive:
                filters.append(Q("term", **{"encyclopedia.material.species.keyword": query_string}))
            # With non-exclusive search we look for match that includes at
            # least all species, possibly even more.
            else:
                musts.append(Q(
                    "match",
                    encyclopedia__material__species={"query": query_string, "operator": "and"}
357
358
                ))

359
360
        page = search_by["page"]
        per_page = search_by["per_page"]
361
        after = search_by["after"]
362
        bool_query = Q(
363
            "bool",
364
365
366
367
            filter=filters,
            must_not=must_nots,
            must=musts,
        )
368

369
        # 1: The paginated approach: No way to know the amount of materials,
370
        # but can return aggregation results in a quick fashion including
371
        # the number of calculation entries per material.
372
        if mode == "aggregation":
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
            # The top query filters out entries based on the user query
            s = Search(index=config.elastic.index_name)
            s = s.query(bool_query)

            # The materials are grouped by using three aggregations:
            # "Composite" to enable scrolling, "Terms" to enable selecting
            # by material_id and "Top Hits" to fetch a single
            # representative material document. Unnecessary fields are
            # filtered to reduce data transfer.
            terms_agg = A("terms", field="encyclopedia.material.material_id")
            composite_kwargs = {"sources": {"materials": terms_agg}, "size": per_page}
            if after is not None:
                composite_kwargs["after"] = after
            composite_agg = A("composite", **composite_kwargs)
            composite_agg.metric("representative", A(
                "top_hits",
                size=1,
                _source={"includes": list(material_prop_map.values())},
            ))
            s.aggs.bucket("materials", composite_agg)

            # We ignore the top level hits
            s = s.extra(**{
                "size": 0,
            })
399

400
401
402
403
404
            response = s.execute()
            materials = response.aggs.materials.buckets
            if len(materials) == 0:
                abort(404, message="No materials found for the given search criteria or pagination.")
            after = response.aggs.materials["after_key"]
405
406
407
408

            # Gather results from aggregations
            result_list = []
            materials = response.aggs.materials.buckets
409
            keys = list(material_prop_map.keys())
410
411
            for material in materials:
                representative = material["representative"][0]
412
413
                mat_dict = get_es_doc_values(representative, material_prop_map, keys)
                mat_dict["n_matches"] = material.doc_count
414
415
416
417
418
419
                result_list.append(mat_dict)

            # Page information is incomplete for aggregations
            pages = {
                "page": page,
                "per_page": per_page,
420
                "after": after,
421
422
            }
        # 2. Collapse approach. Quickly provides a list of materials
423
        # corresponding to the query, offers full pagination, doesn"t include
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
        # the number of matches per material.
        elif mode == "collapse":
            s = Search(index=config.elastic.index_name)
            s = s.query(bool_query)
            s = s.extra(**{
                "collapse": {"field": "encyclopedia.material.material_id"},
                "size": per_page,
                "from": (page - 1) * per_page,
            })

            # Execute query
            response = s.execute()

            # No matches
            if len(response) == 0:
439
                abort(404, message="No materials found for the given search criteria or pagination.")
440
441
442

            # Loop over materials
            result_list = []
443
            keys = list(material_prop_map.keys())
444
            for material in response:
445
                mat_result = get_es_doc_values(material, material_prop_map, keys)
446
447
448
449
450
451
452
453
454
                result_list.append(mat_result)

            # Full page information available for collapse
            pages = {
                "page": page,
                "per_page": per_page,
                "pages": math.ceil(response.hits.total / per_page),
                "total": response.hits.total,
            }
455
456
457

        result = {
            "results": result_list,
458
            "pages": pages,
459
        }
460
        return result, 200
461
462


463
groups_result = api.model("groups_result", {
464
465
    "groups_eos": fields.Raw,
    "groups_par": fields.Raw,
466
467
468
})


469
@ns.route("/materials/<string:material_id>/groups")
Lauri Himanen's avatar
Lauri Himanen committed
470
class EncGroupsResource(Resource):
471
472
473
    @api.response(404, "Material not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
474
    @api.marshal_with(groups_result)
475
    @api.doc("enc_materials")
Lauri Himanen's avatar
Lauri Himanen committed
476
    def get(self, material_id):
477
478
479
        """Returns a summary of the calculation groups that were identified for
        this material.
        """
480
481
482
        # Find entries for the given material, which have EOS or parameter
        # variation hashes set.
        bool_query = Q(
483
            "bool",
484
            filter=[
485
486
487
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
488
489
490
            ],
            must=[
                Q("exists", field="encyclopedia.properties.energies.energy_total"),
491
                Q("exists", field="encyclopedia.material.idealized_structure.cell_volume"),
492
493
            ],
            should=[
494
495
                Q("exists", field="encyclopedia.method.group_eos_id"),
                Q("exists", field="encyclopedia.method.group_parametervariation_id"),
496
497
498
            ],
            minimum_should_match=1,  # At least one of the should query must match
        )
Lauri Himanen's avatar
Lauri Himanen committed
499
500

        s = Search(index=config.elastic.index_name)
501
502
503
504
        s = s.query(bool_query)

        # Bucket the calculations by the group hashes. Only create a bucket if an
        # above-minimum number of documents are found.
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
        group_eos_bucket = A("terms", field="encyclopedia.method.group_eos_id", min_doc_count=4)
        group_param_bucket = A("terms", field="encyclopedia.method.group_parametervariation_id", min_doc_count=2)
        calc_aggregation = A(
            "top_hits",
            _source={"includes": ["calc_id"]},
            sort=[{"encyclopedia.properties.energies.energy_total": {"order": "asc"}}],
            size=100,
        )
        group_eos_bucket.bucket("calculations", calc_aggregation)
        group_param_bucket.bucket("calculations", calc_aggregation)
        s.aggs.bucket("groups_eos", group_eos_bucket)
        s.aggs.bucket("groups_param", group_param_bucket)

        # We ignore the top level hits
        s = s.extra(**{
            "size": 0,
        })

        # Collect information for each group from the aggregations
        response = s.execute()
        groups_eos = {group.key: [calc.calc_id for calc in group.calculations.hits] for group in response.aggs.groups_eos.buckets}
        groups_param = {group.key: [calc.calc_id for calc in group.calculations.hits] for group in response.aggs.groups_param.buckets}

        # Return results
        result = {
            "groups_eos": groups_eos,
            "groups_par": groups_param,
        }

        return result, 200


group_result = api.model("group_result", {
    "calculations": fields.List(fields.String),
    "energies": fields.List(fields.Float),
    "volumes": fields.List(fields.Float),
})
group_source = {
    "includes": [
        "calc_id",
        "encyclopedia.properties.energies.energy_total",
        "encyclopedia.material.idealized_structure.cell_volume",
    ]
}


@ns.route("/materials/<string:material_id>/groups/<string:group_type>/<string:group_id>")
class EncGroupResource(Resource):
    @api.response(404, "Group not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
    @api.marshal_with(group_result)
    @api.doc("enc_group")
    def get(self, material_id, group_type, group_id):
        """Used to query detailed information for a specific calculation group.
        """
        # Find entries for the given material, which have EOS or parameter
        # variation hashes set.
        if group_type == "eos":
            group_id_source = "encyclopedia.method.group_eos_id"
        elif group_type == "par":
            group_id_source = "encyclopedia.method.group_parametervariation_id"
        else:
            abort(400, message="Unsupported group type.")

        bool_query = Q(
            "bool",
            filter=[
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
                Q("term", **{group_id_source: group_id}),
            ],
        )

        s = Search(index=config.elastic.index_name)
        s = s.query(bool_query)
582
583
584
585
586

        # calc_id and energy should be extracted for each matched document. The
        # documents are sorted by energy so that the minimum energy one can be
        # easily extracted. A maximum request size is set in order to limit the
        # result size. ES also has an index-level property
587
        # "index.max_inner_result_window" that limits the number of results
588
589
590
591
592
593
594
        # that an inner result can contain.
        energy_aggregation = A(
            "top_hits",
            _source=group_source,
            sort=[{"encyclopedia.properties.energies.energy_total": {"order": "asc"}}],
            size=100,
        )
595
        s.aggs.bucket("groups_eos", energy_aggregation)
596

597
598
599
600
        # We ignore the top level hits
        s = s.extra(**{
            "size": 0,
        })
601

602
        # Collect information for each group from the aggregations
603
        response = s.execute()
604

605
606
607
608
609
610
611
612
        hits = response.aggs.groups_eos.hits
        calculations = [doc.calc_id for doc in hits]
        energies = [doc.encyclopedia.properties.energies.energy_total for doc in hits]
        volumes = [doc.encyclopedia.material.idealized_structure.cell_volume for doc in hits]
        group_dict = {
            "calculations": calculations,
            "energies": energies,
            "volumes": volumes,
613
        }
614
615

        return group_dict, 200
616
617


618
619
620
621
suggestions_map = {
    "code_name": "dft.code_name",
    "structure_type": "encyclopedia.material.bulk.structure_type",
}
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
suggestions_query = api.parser()
suggestions_query.add_argument(
    "property",
    type=str,
    choices=("code_name", "structure_type"),
    help="The property name for which suggestions are returned.",
    location="args"
)
suggestions_result = api.model("suggestions_result", {
    "code_name": fields.List(fields.String),
    "structure_type": fields.List(fields.String),
})


@ns.route("/suggestions")
class EncSuggestionsResource(Resource):
    @api.response(404, "Suggestion not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
    @api.expect(suggestions_query, validate=False)
    @api.marshal_with(suggestions_result, skip_none=True)
    @api.doc("enc_suggestions")
    def get(self):

        # Parse request arguments
        args = suggestions_query.parse_args()
        prop = args.get("property", None)

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
        # Use aggregation to return all unique terms for the requested field.
        # Without using composite aggregations there is a size limit for the
        # number of aggregation buckets. This should, however, not be a problem
        # since the number of unique values is low for all supported properties.
        s = Search(index=config.elastic.index_name)
        query = Q(
            "bool",
            filter=[
                Q("term", published=True),
                Q("term", with_embargo=False),
            ]
        )
        s = s.query(query)
        s = s.extra(**{
            "size": 0,
        })

        terms_agg = A("terms", field=suggestions_map[prop])
        s.aggs.bucket("suggestions", terms_agg)

        # Gather unique values into a list
        response = s.execute()
        suggestions = [x.key for x in response.aggs.suggestions.buckets]

        return {prop: suggestions}, 200
675
676
677
678
679
680
681
682


calc_prop_map = {
    "calc_id": "calc_id",
    "code_name": "dft.code_name",
    "code_version": "dft.code_version",
    "functional_type": "encyclopedia.method.functional_type",
    "basis_set_type": "dft.basis_set",
683
    "core_electron_treatment": "encyclopedia.method.core_electron_treatment",
684
685
686
687
    "run_type": "encyclopedia.calculation.calculation_type",
    "has_dos": "encyclopedia.properties.electronic_dos",
    "has_band_structure": "encyclopedia.properties.electronic_band_structure",
    "has_thermal_properties": "encyclopedia.properties.thermodynamical_properties",
688
689
    "has_phonon_dos": "encyclopedia.properties.phonon_dos",
    "has_phonon_band_structure": "encyclopedia.properties.phonon_band_structure",
690
691
692
693
694
695
696
}
calculation_result = api.model("calculation_result", {
    "calc_id": fields.String,
    "code_name": fields.String,
    "code_version": fields.String,
    "functional_type": fields.String,
    "basis_set_type": fields.String,
697
    "core_electron_treatment": fields.String,
698
699
700
701
    "run_type": fields.String,
    "has_dos": fields.Boolean,
    "has_band_structure": fields.Boolean,
    "has_thermal_properties": fields.Boolean,
702
703
    "has_phonon_dos": fields.Boolean,
    "has_phonon_band_structure": fields.Boolean,
704
})
705
706
707
708
709
710
representatives_result = api.model("representatives_result", {
    "idealized_structure": fields.String,
    "electronic_band_structure": fields.String,
    "electronic_dos": fields.String,
    "thermodynamical_properties": fields.String,
})
711
712
713
714
calculations_result = api.model("calculations_result", {
    "total_results": fields.Integer,
    "pages": fields.Nested(pages_result),
    "results": fields.List(fields.Nested(calculation_result)),
715
    "representatives": fields.Nested(representatives_result, skip_none=True),
716
717
718
719
})


@ns.route("/materials/<string:material_id>/calculations")
720
class EncCalculationsResource(Resource):
721
722
723
724
725
    @api.response(404, "Suggestion not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
    @api.doc("enc_calculations")
    def get(self, material_id):
726
727
728
        """Used to return all calculations related to the given material. Also
        returns a representative calculation for each property shown in the
        overview page.
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
        """
        s = Search(index=config.elastic.index_name)
        query = Q(
            "bool",
            filter=[
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
            ]
        )
        s = s.query(query)

        # The query is filtered already on the ES side so we don"t need to
        # transfer so much data.
        s = s.extra(**{
744
745
746
            "_source": {"includes": list(calc_prop_map.values()) + ["dft.xc_functional"]},
            "size": 10000,
            "from": 0,
747
748
749
750
751
752
753
        })
        response = s.execute()

        # No such material
        if len(response) == 0:
            abort(404, message="There is no material {}".format(material_id))

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
        # Add representative properties. It might be possible to write a custom
        # ES scoring mechanism or aggregation to also perform the selection.
        representatives = {}

        def calc_score(entry):
            """Custom scoring function used to sort results by their
            "quality". Currently built to mimic the scoring that was used
            in the old Encyclopedia GUI.
            """
            score = 0
            functional_score = {
                "GGA": 100
            }
            code_score = {
                "FHI-aims": 3,
                "VASP": 2,
                "Quantum Espresso": 1,
            }
            code_name = entry.dft.code_name
            functional = entry.dft.xc_functional
            has_dos = rgetattr(entry, "encyclopedia.properties.electronic_band_structure") is not None
            has_bs = rgetattr(entry, "encyclopedia.properties.electronic_dos") is not None
            score += functional_score.get(functional, 0)
            score += code_score.get(code_name, 0)
            if has_dos and has_bs:
                score += 10

            return score

        # The calculations are first sorted by "quality"
        sorted_calc = sorted(response, key=lambda x: calc_score(x), reverse=True)

        # Get the requested representative properties
        representatives["idealized_structure"] = sorted_calc[0].calc_id
        thermo_found = False
        bs_found = False
        dos_found = False
        for calc in sorted_calc:
            if rgetattr(calc, "encyclopedia.properties.thermodynamical_properties") is not None:
                representatives["thermodynamical_properties"] = calc.calc_id
                thermo_found = True
            if rgetattr(calc, "encyclopedia.properties.electronic_band_structure") is not None:
                representatives["electronic_band_structure"] = calc.calc_id
                bs_found = True
            if rgetattr(calc, "encyclopedia.properties.electronic_dos") is not None:
                representatives["electronic_dos"] = calc.calc_id
                dos_found = True
            if thermo_found and bs_found and dos_found:
                break

804
805
806
807
808
        # Create result JSON
        results = []
        for entry in response:
            calc_dict = get_es_doc_values(entry, calc_prop_map)
            calc_dict["has_dos"] = calc_dict["has_dos"] is not None
809
            calc_dict["has_band_structure"] = calc_dict["has_band_structure"] is not None
810
            calc_dict["has_thermal_properties"] = calc_dict["has_thermal_properties"] is not None
811
812
            calc_dict["has_phonon_dos"] = calc_dict["has_phonon_dos"] is not None
            calc_dict["has_phonon_band_structure"] = calc_dict["has_phonon_band_structure"] is not None
813
814
815
816
817
            results.append(calc_dict)

        result = {
            "total_results": len(results),
            "results": results,
818
            "representatives": representatives,
819
820
821
822
823
        }

        return result, 200


824
825
826
827
histogram = api.model("histogram", {
    "occurrences": fields.List(fields.Integer),
    "values": fields.List(fields.Float),
})
828
829
statistics_query = api.model("statistics_query", {
    "calculations": fields.List(fields.String),
830
    "properties": fields.List(fields.String),
831
    "n_histogram_bins": fields.Integer,
832
833
834
835
836
})
statistics = api.model("statistics", {
    "min": fields.Float,
    "max": fields.Float,
    "avg": fields.Float,
837
    "histogram": fields.Nested(histogram, skip_none=True)
838
839
})
statistics_result = api.model("statistics_result", {
840
841
842
843
844
845
846
847
848
849
    "cell_volume": fields.Nested(statistics, skip_none=True),
    "atomic_density": fields.Nested(statistics, skip_none=True),
    "mass_density": fields.Nested(statistics, skip_none=True),
    "lattice_a": fields.Nested(statistics, skip_none=True),
    "lattice_b": fields.Nested(statistics, skip_none=True),
    "lattice_c": fields.Nested(statistics, skip_none=True),
    "alpha": fields.Nested(statistics, skip_none=True),
    "beta": fields.Nested(statistics, skip_none=True),
    "gamma": fields.Nested(statistics, skip_none=True),
    "band_gap": fields.Nested(statistics, skip_none=True),
850
})
851
852
853
854
855
856
857
858
859
860
property_map = {
    "cell_volume": "encyclopedia.material.idealized_structure.cell_volume",
    "atomic_density": "encyclopedia.properties.atomic_density",
    "mass_density": "encyclopedia.properties.mass_density",
    "lattice_a": "encyclopedia.material.idealized_structure.lattice_parameters.a",
    "lattice_b": "encyclopedia.material.idealized_structure.lattice_parameters.b",
    "lattice_c": "encyclopedia.material.idealized_structure.lattice_parameters.c",
    "alpha": "encyclopedia.material.idealized_structure.lattice_parameters.alpha",
    "beta": "encyclopedia.material.idealized_structure.lattice_parameters.beta",
    "gamma": "encyclopedia.material.idealized_structure.lattice_parameters.gamma",
Lauri Himanen's avatar
Lauri Himanen committed
861
    "band_gap": "encyclopedia.properties.band_gap",
862
}
863
864
865


@ns.route("/materials/<string:material_id>/statistics")
866
class EncStatisticsResource(Resource):
867
868
869
    @api.response(404, "Suggestion not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
870
871
    @api.expect(statistics_query, validate=False)
    @api.marshal_with(statistics_result, skip_none=True)
872
873
    @api.doc("enc_statistics")
    def post(self, material_id):
874
875
        """Used to return statistics related to the specified material and
        calculations.
876
        """
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
        # Get query parameters as json
        try:
            data = marshal(request.get_json(), statistics_query)
        except Exception as e:
            abort(400, message=str(e))

        # Find entries for the given material.
        bool_query = Q(
            "bool",
            filter=[
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
                Q("terms", calc_id=data["calculations"]),
            ]
        )

        s = Search(index=config.elastic.index_name)
        s = s.query(bool_query)
        s = s.extra(**{
            "size": 0,
        })

900
901
902
903
904
905
        # Add statistics aggregations for each requested property
        properties = data["properties"]
        for prop in properties:
            stats_agg = A("stats", field=property_map[prop])
            s.aggs.bucket("{}_stats".format(prop), stats_agg)

906
907
908
909
910
        # No hits on the top query level
        response = s.execute()
        if response.hits.total == 0:
            abort(404, message="Could not find matching calculations.")

911
912
913
914
915
916
917
918
        # Run a second query that creates histograms with fixed size buckets
        # based on the min and max from previous query. Might make more sense
        # to use the mean and sigma to define the range?
        s = Search(index=config.elastic.index_name)
        s = s.query(bool_query)
        s = s.extra(**{
            "size": 0,
        })
919
        n_bins = data["n_histogram_bins"]
920
921
        for prop in properties:
            stats = getattr(response.aggs, "{}_stats".format(prop))
922
923
            if stats.count == 0:
                continue
924
            interval = (stats.max * 1.001 - stats.min) / n_bins
925
926
            if interval == 0:
                interval = 1
927
            hist_agg = A("histogram", field=property_map[prop], interval=interval, offset=stats.min, min_doc_count=0)
928
929
930
            s.aggs.bucket("{}_hist".format(prop), hist_agg)
        response_hist = s.execute()

931
        # Return results
932
933
934
        result = {}
        for prop in properties:
            stats = getattr(response.aggs, "{}_stats".format(prop))
935
936
            if stats.count == 0:
                continue
937
938
939
940
941
942
943
            hist = getattr(response_hist.aggs, "{}_hist".format(prop))
            occurrences = [x.doc_count for x in hist.buckets]
            values = [x.key for x in hist.buckets]
            result[prop] = {
                "min": stats.min,
                "max": stats.max,
                "avg": stats.avg,
944
                "histogram": {
945
946
947
                    "occurrences": occurrences,
                    "values": values,
                }
948
            }
949

950
        return result, 200
951
952
953
954
955
956
957
958
959
960
961


wyckoff_variables_result = api.model("wyckoff_variables_result", {
    "x": fields.Float,
    "y": fields.Float,
    "z": fields.Float,
})
wyckoff_set_result = api.model("wyckoff_set_result", {
    "wyckoff_letter": fields.String,
    "indices": fields.List(fields.Integer),
    "element": fields.String,
962
    "variables": fields.List(fields.Nested(wyckoff_variables_result, skip_none=True)),
963
})
964
965
966
967
968
969
970
971
lattice_parameters = api.model("lattice_parameters", {
    "a": fields.Float,
    "b": fields.Float,
    "c": fields.Float,
    "alpha": fields.Float,
    "beta": fields.Float,
    "gamma": fields.Float,
})
972
973
974
975
976
977

idealized_structure_result = api.model("idealized_structure_result", {
    "atom_labels": fields.List(fields.String),
    "atom_positions": fields.List(fields.List(fields.Float)),
    "lattice_vectors": fields.List(fields.List(fields.Float)),
    "lattice_vectors_primitive": fields.List(fields.List(fields.Float)),
978
    "lattice_parameters": fields.Nested(lattice_parameters, skip_none=True),
979
980
981
    "periodicity": fields.List(fields.Boolean),
    "number_of_atoms": fields.Integer,
    "cell_volume": fields.Float,
982
    "wyckoff_sets": fields.List(fields.Nested(wyckoff_set_result, skip_none=True)),
983
984
})

985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
calculation_property_map = {
    "lattice_parameters": {
        "es_source": "encyclopedia.material.idealized_structure.lattice_parameters"
    },
    "energies": {
        "es_source": "encyclopedia.properties.energies",
    },
    "mass_density": {
        "es_source": "encyclopedia.properties.mass_density",
    },
    "atomic_density": {
        "es_source": "encyclopedia.properties.atomic_density",
    },
    "cell_volume": {
        "es_source": "encyclopedia.material.idealized_structure.cell_volume"
    },
Lauri Himanen's avatar
Lauri Himanen committed
1001
1002
1003
    "band_gap": {
        "es_source": "encyclopedia.properties.band_gap"
    },
1004
1005
1006
1007
1008
1009
    "electronic_band_structure": {
        "es_source": "encyclopedia.properties.electronic_band_structure"
    },
    "electronic_dos": {
        "es_source": "encyclopedia.properties.electronic_dos"
    },
1010
1011
1012
1013
1014
1015
1016
1017
1018
    "phonon_band_structure": {
        "es_source": "encyclopedia.properties.phonon_band_structure"
    },
    "phonon_dos": {
        "es_source": "encyclopedia.properties.phonon_dos"
    },
    "thermodynamical_properties": {
        "es_source": "encyclopedia.properties.thermodynamical_properties"
    },
1019
1020
1021
    "wyckoff_sets": {
        "arch_source": "section_metadata/encyclopedia/material/idealized_structure/wyckoff_sets"
    },
1022
1023
1024
    "idealized_structure": {
        "arch_source": "section_metadata/encyclopedia/material/idealized_structure"
    },
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
}

calculation_property_query = api.model("calculation_query", {
    "properties": fields.List(fields.String),
})
energies = api.model("energies", {
    "energy_total": fields.Float,
    "energy_total_T0": fields.Float,
    "energy_free": fields.Float,
})
Lauri Himanen's avatar
Lauri Himanen committed
1035
1036
1037
1038
electronic_band_structure = api.model("electronic_band_structure", {
    "reciprocal_cell": fields.List(fields.List(fields.Float)),
    "brillouin_zone": fields.Raw,
    "section_k_band_segment": fields.Raw,
1039
    "section_band_gap": fields.Raw,
Lauri Himanen's avatar
Lauri Himanen committed
1040
1041
1042
})
electronic_dos = api.model("electronic_dos", {
    "dos_energies": fields.List(fields.Float),
1043
    "dos_values": fields.List(fields.List(fields.Float)),
Lauri Himanen's avatar
Lauri Himanen committed
1044
})
1045
1046
1047
calculation_property_result = api.model("calculation_property_result", {
    "lattice_parameters": fields.Nested(lattice_parameters, skip_none=True),
    "energies": fields.Nested(energies, skip_none=True),
1048
1049
1050
    "mass_density": fields.Float,
    "atomic_density": fields.Float,
    "cell_volume": fields.Float,
1051
    "wyckoff_sets": fields.Nested(wyckoff_set_result, skip_none=True),
1052
    "idealized_structure": fields.Nested(idealized_structure_result, skip_none=True),
1053
1054
1055
    "band_gap": fields.Float,
    "electronic_band_structure": fields.Nested(electronic_band_structure, skip_none=True),
    "electronic_dos": fields.Nested(electronic_dos, skip_none=True),
1056
1057
1058
    "phonon_band_structure": fields.Raw,
    "phonon_dos": fields.Raw,
    "thermodynamical_properties": fields.Raw,
1059
1060
1061
})


1062
1063
1064
1065
1066
@ns.route("/materials/<string:material_id>/calculations/<string:calc_id>")
class EncCalculationResource(Resource):
    @api.response(404, "Material or calculation not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
1067
    @api.expect(calculation_property_query, validate=False)
1068
    @api.marshal_with(calculation_property_result, skip_none=True)
1069
    @api.doc("enc_calculation")
1070
1071
1072
1073
    def post(self, material_id, calc_id):
        """Used to return calculation details. Some properties are not
        available in the ES index and are instead read from the Archive
        directly.
1074
        """
1075
1076
1077
1078
1079
1080
        # Get query parameters as json
        try:
            data = marshal(request.get_json(), calculation_property_query)
        except Exception as e:
            abort(400, message=str(e))

1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
        s = Search(index=config.elastic.index_name)
        query = Q(
            "bool",
            filter=[
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
                Q("term", calc_id=calc_id),
            ]
        )
        s = s.query(query)

1093
        # Create dictionaries for requested properties
1094
        references = []
1095
1096
1097
        properties = data["properties"]
        arch_properties = {}
        es_properties = {}
1098
1099
1100
1101
1102
1103
1104
        ref_properties = set((
            "electronic_dos",
            "electronic_band_structure",
            "thermodynamical_properties",
            "phonon_dos",
            "phonon_band_structure",
        ))
1105
1106
1107
1108
        for prop in properties:
            es_source = calculation_property_map[prop].get("es_source")
            if es_source is not None:
                es_properties[prop] = es_source
1109
                if prop in ref_properties:
1110
                    references.append(prop)
1111
1112
1113
1114
            arch_source = calculation_property_map[prop].get("arch_source")
            if arch_source is not None:
                arch_properties[prop] = arch_source

1115
        # The query is filtered already on the ES side so we don't need to
1116
        # transfer so much data.
1117
1118
1119
        sources = [
            "upload_id",
            "calc_id",
1120
            "encyclopedia",
1121
1122
1123
        ]
        sources += list(es_properties.values())

1124
        s = s.extra(**{
1125
            "_source": {"includes": sources},
1126
1127
1128
1129
1130
1131
1132
1133
1134
            "size": 1,
        })

        response = s.execute()

        # No such material
        if len(response) == 0:
            abort(404, message="There is no material {} with calculation {}".format(material_id, calc_id))

1135
1136
1137
        # Add references that are to be read from the archive
        for ref in references:
            arch_path = response[0]
1138
1139
            arch_path = rgetattr(arch_path, es_properties[ref])
            if arch_path is not None:
Lauri Himanen's avatar
Lauri Himanen committed
1140
                arch_properties[ref] = arch_path
1141
1142
            del es_properties[ref]

1143
1144
1145
1146
1147
1148
1149
        # If any of the requested properties require data from the Archive, the
        # file is opened and read.
        result = {}
        if len(arch_properties) != 0:
            entry = response[0]
            upload_id = entry.upload_id
            calc_id = entry.calc_id
1150
            root = read_archive(
1151
1152
1153
1154
1155
                upload_id,
                calc_id,
            )

            # Add results from archive
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
            for key, arch_path in arch_properties.items():
                value = root[arch_path]

                # Save derived properties and turn into dict
                if key == "thermodynamical_properties":
                    specific_heat_capacity = value.specific_heat_capacity.magnitude.tolist()
                    specific_free_energy = value.specific_vibrational_free_energy_at_constant_volume.magnitude.tolist()
                value = value.m_to_dict()
                if key == "thermodynamical_properties":
                    value["specific_heat_capacity"] = specific_heat_capacity
                    value["specific_vibrational_free_energy_at_constant_volume"] = specific_free_energy
1167

1168
                # DOS results are simplified.
1169
                if key == "electronic_dos":
Lauri Himanen's avatar
Lauri Himanen committed
1170
1171
1172
1173
1174
1175
                    if "dos_energies_normalized" in value:
                        value["dos_energies"] = value["dos_energies_normalized"]
                        del value["dos_energies_normalized"]
                    if "dos_values_normalized" in value:
                        value["dos_values"] = value["dos_values_normalized"]
                        del value["dos_values_normalized"]
1176

1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
                # Pre-calculate k-path length to be used as x-coordinate in
                # plots. If the VBM and CBM information is needed later, it
                # can be added as indices along the path. The exact
                # k-points and occupations are removed to save band width.
                if key == "electronic_band_structure":
                    segments = value["section_k_band_segment"]
                    k_path_length = 0
                    for segment in segments:
                        k_points = np.array(segment["band_k_points"])
                        segment_length = np.linalg.norm(k_points[-1, :] - k_points[0, :])
                        k_path_distances = k_path_length + np.linalg.norm(k_points - k_points[0, :], axis=1)
                        k_path_length += segment_length
                        segment["k_path_distances"] = k_path_distances.tolist()
                        del segment["band_k_points"]
                        if "band_occupations" in segment:
                            del segment["band_occupations"]

1194
1195
1196
                result[key] = value

        # Add results from ES
1197
        for prop, es_source in es_properties.items():
1198
1199
            value = rgetattr(response[0], es_source)
            if value is not None:
Lauri Himanen's avatar
Lauri Himanen committed
1200
1201
1202
                if isinstance(value, AttrDict):
                    value = value.to_dict()
                result[prop] = value
1203
1204
1205
1206

        return result, 200


1207
def read_archive(upload_id: str, calc_id: str) -> EntryArchive:
1208
    """Used to read data from the archive.
1209
1210
1211
1212
1213
1214

    Args:
        upload_id: Upload id.
        calc_id: Calculation id.

    Returns:
1215
        MSection: The section_run as MSection
1216
1217
        For each path, a dictionary containing the path as key and the returned
        section as value.
1218
1219
1220
1221
    """
    upload_files = files.UploadFiles.get(upload_id)
    with upload_files.read_archive(calc_id) as archive:
        data = archive[calc_id]
1222
        root = EntryArchive.m_from_dict(data.to_dict())
1223

1224
    return root