encyclopedia.py 45.6 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

21
22
from flask_restplus import Resource, abort, fields, marshal
from flask import request
23
from elasticsearch_dsl import Search, Q, A
24
from elasticsearch_dsl.utils import AttrDict
25

26
from nomad import config, files
27
from nomad.units import ureg
Lauri Himanen's avatar
Lauri Himanen committed
28
from nomad.atomutils import get_hill_decomposition
29
from nomad.datamodel.datamodel import EntryArchive
30
from .api import api
31

32
ns = api.namespace("encyclopedia", description="Access encyclopedia metadata.")
33
34
re_formula = re.compile(r"([A-Z][a-z]?)(\d*)")

35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
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",
}
53
54


55
56
57
58
59
60
61
62
63
64
65
66
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


67
def get_es_doc_values(es_doc, mapping, keys=None):
68
69
    """Used to form a material definition for "materials/<material_id>" from
    the given ElasticSearch root document.
70
    """
71
72
73
    if keys is None:
        keys = mapping.keys()

74
    result = {}
75
    for key in keys:
76
        es_key = mapping[key]
77
        value = rgetattr(es_doc, es_key)
78
        result[key] = value
79
80
81
82
83

    return result


material_query = api.parser()
84
85
86
87
88
89
90
91
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", {
92
93
    # General
    "material_id": fields.String,
94
95
    "formula": fields.String,
    "formula_reduced": fields.String,
96
    "system_type": fields.String,
97
    "n_matches": fields.Integer,
98
    # Bulk only
99
    "has_free_wyckoff_parameters": fields.Boolean,
100
    "strukturbericht_designation": fields.String,
101
    "material_name": fields.String,
102
103
    "bravais_lattice": fields.String,
    "crystal_system": fields.String,
104
    "point_group": fields.String,
105
106
107
    "space_group_number": fields.Integer,
    "space_group_international_short_symbol": fields.String,
    "structure_prototype": fields.String,
108
109
    "structure_type": fields.String,
})
110
111


112
@ns.route("/materials/<string:material_id>")
113
class EncMaterialResource(Resource):
114
115
116
    @api.response(404, "The material does not exist")
    @api.response(200, "Metadata send", fields.Raw)
    @api.doc("material/<material_id>")
117
    @api.expect(material_query)
118
    @api.marshal_with(material_result, skip_none=True)
119
120
121
    def get(self, material_id):
        """Used to retrive basic information related to the specified material.
        """
122
123
124
125
        # Parse request arguments
        args = material_query.parse_args()
        prop = args.get("property", None)
        if prop is not None:
126
127
            keys = [prop]
            es_keys = [material_prop_map[prop]]
128
129
        else:
            keys = list(material_prop_map.keys())
130
            es_keys = list(material_prop_map.values())
131

132
133
134
135
136
137
138
139
140
        # 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(
141
            "bool",
142
            filter=[
143
144
145
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
146
147
148
            ]
        )
        s = s.query(query)
149

150
151
152
153
154
155
156
157
        # 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"},
        })
158
159
        response = s.execute()

160
        # No such material
161
        if len(response) == 0:
162
            abort(404, message="There is no material {}".format(material_id))
163

164
        # Add values from ES entry
165
        entry = response[0]
166
        result = get_es_doc_values(entry, material_prop_map, keys)
167

168
169
170
        return result, 200


171
range_query = api.model("range_query", {
172
173
174
    "max": fields.Float,
    "min": fields.Float,
})
175
176
177
materials_after = api.model("materials_after", {
    "materials": fields.String,
})
178
179
materials_query = api.model("materials_input", {
    "search_by": fields.Nested(api.model("search_query", {
180
181
        "exclusive": fields.Boolean(default=False),
        "formula": fields.String,
Lauri Himanen's avatar
Lauri Himanen committed
182
        "element": fields.String,
183
        "page": fields.Integer(default=1),
184
        "after": fields.Nested(materials_after, allow_null=True),
185
186
        "per_page": fields.Integer(default=25),
        "pagination": fields.Boolean,
187
        "mode": fields.String(default="aggregation"),
188
    })),
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
    "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."),
204
})
205
206
207
208
209
pages_result = api.model("page_info", {
    "per_page": fields.Integer,
    "total": fields.Integer,
    "page": fields.Integer,
    "pages": fields.Integer,
210
    "after": fields.Nested(materials_after),
211
212
})

213
214
materials_result = api.model("materials_result", {
    "total_results": fields.Integer(allow_null=False),
215
216
    "results": fields.List(fields.Nested(material_result, skip_none=True)),
    "pages": fields.Nested(pages_result, skip_none=True),
217
    "es_query": fields.String(allow_null=False),
218
219
220
})


221
@ns.route("/materials")
222
class EncMaterialsResource(Resource):
223
224
225
    @api.response(404, "No materials found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
226
    @api.expect(materials_query, validate=False)
227
    @api.marshal_with(materials_result, skip_none=True)
228
    @api.doc("materials")
229
230
231
232
233
234
235
236
237
238
239
240
241
242
    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
243
244
        filters.append(Q("term", published=True))
        filters.append(Q("term", with_embargo=False))
245
246
247
248
249
250
251

        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")
252
        add_terms_filter("space_group_number", "encyclopedia.material.bulk.space_group_number")
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
        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")
297

298
299
        # Create query for elements or formula
        search_by = data["search_by"]
300
        mode = search_by["mode"]
301
        formula = search_by["formula"]
Lauri Himanen's avatar
Lauri Himanen committed
302
        elements = search_by["element"]
303
304
305
        exclusive = search_by["exclusive"]

        if formula is not None:
Lauri Himanen's avatar
Lauri Himanen committed
306
307
            # Here we determine a list of atom types. The types may occur
            # multiple times and at multiple places.
308
309
310
311
312
313
314
315
316
317
            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
318
319
320
321
322
                        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.
323
            names, reduced_counts = get_hill_decomposition(element_list, reduced=True)
Lauri Himanen's avatar
Lauri Himanen committed
324
325
326
327
328
            query_string = []
            for name, count in zip(names, reduced_counts):
                if count == 1:
                    query_string.append(name)
                else:
329
                    query_string.append("{}{}".format(name, int(count)))
Lauri Himanen's avatar
Lauri Himanen committed
330
            query_string = " ".join(query_string)
331
332
333

            # With exclusive search we look for exact match
            if exclusive:
Lauri Himanen's avatar
Lauri Himanen committed
334
                filters.append(Q("term", **{"encyclopedia.material.species_and_counts.keyword": query_string}))
335
336
337
338
339
            # 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
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
                    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"}
356
357
                ))

358
359
        page = search_by["page"]
        per_page = search_by["per_page"]
360
        after = search_by["after"]
361
        bool_query = Q(
362
            "bool",
363
364
365
366
            filter=filters,
            must_not=must_nots,
            must=musts,
        )
367

368
        # 1: The paginated approach: No way to know the amount of materials,
369
        # but can return aggregation results in a quick fashion including
370
        # the number of calculation entries per material.
371
        if mode == "aggregation":
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
            # 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,
            })
398

399
400
401
402
403
            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"]
404
405
406
407

            # Gather results from aggregations
            result_list = []
            materials = response.aggs.materials.buckets
408
            keys = list(material_prop_map.keys())
409
410
            for material in materials:
                representative = material["representative"][0]
411
412
                mat_dict = get_es_doc_values(representative, material_prop_map, keys)
                mat_dict["n_matches"] = material.doc_count
413
414
415
416
417
418
                result_list.append(mat_dict)

            # Page information is incomplete for aggregations
            pages = {
                "page": page,
                "per_page": per_page,
419
                "after": after,
420
421
            }
        # 2. Collapse approach. Quickly provides a list of materials
422
        # corresponding to the query, offers full pagination, doesn"t include
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
        # 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:
438
                abort(404, message="No materials found for the given search criteria or pagination.")
439
440
441

            # Loop over materials
            result_list = []
442
            keys = list(material_prop_map.keys())
443
            for material in response:
444
                mat_result = get_es_doc_values(material, material_prop_map, keys)
445
446
447
448
449
450
451
452
453
                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,
            }
454
455
456

        result = {
            "results": result_list,
457
            "pages": pages,
458
        }
459
        return result, 200
460
461


462
groups_result = api.model("groups_result", {
463
464
    "groups_eos": fields.Raw,
    "groups_par": fields.Raw,
465
466
467
})


468
@ns.route("/materials/<string:material_id>/groups")
Lauri Himanen's avatar
Lauri Himanen committed
469
class EncGroupsResource(Resource):
470
471
472
    @api.response(404, "Material not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
473
    @api.marshal_with(groups_result)
474
    @api.doc("enc_materials")
Lauri Himanen's avatar
Lauri Himanen committed
475
    def get(self, material_id):
476
477
478
        """Returns a summary of the calculation groups that were identified for
        this material.
        """
479
480
481
        # Find entries for the given material, which have EOS or parameter
        # variation hashes set.
        bool_query = Q(
482
            "bool",
483
            filter=[
484
485
486
                Q("term", published=True),
                Q("term", with_embargo=False),
                Q("term", encyclopedia__material__material_id=material_id),
487
488
489
            ],
            must=[
                Q("exists", field="encyclopedia.properties.energies.energy_total"),
490
                Q("exists", field="encyclopedia.material.idealized_structure.cell_volume"),
491
492
            ],
            should=[
493
494
                Q("exists", field="encyclopedia.method.group_eos_id"),
                Q("exists", field="encyclopedia.method.group_parametervariation_id"),
495
496
497
            ],
            minimum_should_match=1,  # At least one of the should query must match
        )
Lauri Himanen's avatar
Lauri Himanen committed
498
499

        s = Search(index=config.elastic.index_name)
500
501
502
503
        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.
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
        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)
581
582
583
584
585

        # 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
586
        # "index.max_inner_result_window" that limits the number of results
587
588
589
590
591
592
593
        # that an inner result can contain.
        energy_aggregation = A(
            "top_hits",
            _source=group_source,
            sort=[{"encyclopedia.properties.energies.energy_total": {"order": "asc"}}],
            size=100,
        )
594
        s.aggs.bucket("groups_eos", energy_aggregation)
595

596
597
598
599
        # We ignore the top level hits
        s = s.extra(**{
            "size": 0,
        })
600

601
        # Collect information for each group from the aggregations
602
        response = s.execute()
603

604
605
606
607
608
609
610
611
        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,
612
        }
613
614

        return group_dict, 200
615
616


617
618
619
620
suggestions_map = {
    "code_name": "dft.code_name",
    "structure_type": "encyclopedia.material.bulk.structure_type",
}
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
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)

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
        # 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
674
675
676
677
678
679
680
681


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",
682
    "core_electron_treatment": "encyclopedia.method.core_electron_treatment",
683
684
685
686
    "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",
687
688
    "has_phonon_dos": "encyclopedia.properties.phonon_dos",
    "has_phonon_band_structure": "encyclopedia.properties.phonon_band_structure",
689
690
691
692
693
694
695
}
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,
696
    "core_electron_treatment": fields.String,
697
698
699
700
    "run_type": fields.String,
    "has_dos": fields.Boolean,
    "has_band_structure": fields.Boolean,
    "has_thermal_properties": fields.Boolean,
701
702
    "has_phonon_dos": fields.Boolean,
    "has_phonon_band_structure": fields.Boolean,
703
})
704
705
706
707
708
709
representatives_result = api.model("representatives_result", {
    "idealized_structure": fields.String,
    "electronic_band_structure": fields.String,
    "electronic_dos": fields.String,
    "thermodynamical_properties": fields.String,
})
710
711
712
713
calculations_result = api.model("calculations_result", {
    "total_results": fields.Integer,
    "pages": fields.Nested(pages_result),
    "results": fields.List(fields.Nested(calculation_result)),
714
    "representatives": fields.Nested(representatives_result, skip_none=True),
715
716
717
718
})


@ns.route("/materials/<string:material_id>/calculations")
719
class EncCalculationsResource(Resource):
720
721
722
723
724
    @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):
725
726
727
        """Used to return all calculations related to the given material. Also
        returns a representative calculation for each property shown in the
        overview page.
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
        """
        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(**{
743
744
745
            "_source": {"includes": list(calc_prop_map.values()) + ["dft.xc_functional"]},
            "size": 10000,
            "from": 0,
746
747
748
749
750
751
752
        })
        response = s.execute()

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

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
        # 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

803
804
805
806
807
        # 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
808
            calc_dict["has_band_structure"] = calc_dict["has_band_structure"] is not None
809
            calc_dict["has_thermal_properties"] = calc_dict["has_thermal_properties"] is not None
810
811
            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
812
813
814
815
816
            results.append(calc_dict)

        result = {
            "total_results": len(results),
            "results": results,
817
            "representatives": representatives,
818
819
820
821
822
        }

        return result, 200


823
824
825
826
histogram = api.model("histogram", {
    "occurrences": fields.List(fields.Integer),
    "values": fields.List(fields.Float),
})
827
828
statistics_query = api.model("statistics_query", {
    "calculations": fields.List(fields.String),
829
    "properties": fields.List(fields.String),
830
    "n_histogram_bins": fields.Integer,
831
832
833
834
835
})
statistics = api.model("statistics", {
    "min": fields.Float,
    "max": fields.Float,
    "avg": fields.Float,
836
    "histogram": fields.Nested(histogram, skip_none=True)
837
838
})
statistics_result = api.model("statistics_result", {
839
840
841
842
843
844
845
846
847
848
    "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),
849
})
850
851
852
853
854
855
856
857
858
859
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
860
    "band_gap": "encyclopedia.properties.band_gap",
861
}
862
863
864


@ns.route("/materials/<string:material_id>/statistics")
865
class EncStatisticsResource(Resource):
866
867
868
    @api.response(404, "Suggestion not found")
    @api.response(400, "Bad request")
    @api.response(200, "Metadata send", fields.Raw)
869
870
    @api.expect(statistics_query, validate=False)
    @api.marshal_with(statistics_result, skip_none=True)
871
872
    @api.doc("enc_statistics")
    def post(self, material_id):
873
874
        """Used to return statistics related to the specified material and
        calculations.
875
        """
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
        # 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,
        })

899
900
901
902
903
904
        # 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)

905
906
907
908
909
        # No hits on the top query level
        response = s.execute()
        if response.hits.total == 0:
            abort(404, message="Could not find matching calculations.")

910
911
912
913
914
915
916
917
        # 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,
        })
918
        n_bins = data["n_histogram_bins"]
919
920
        for prop in properties:
            stats = getattr(response.aggs, "{}_stats".format(prop))
921
922
            if stats.count == 0:
                continue
923
            interval = (stats.max * 1.001 - stats.min) / n_bins
924
925
            if interval == 0:
                interval = 1
926
            hist_agg = A("histogram", field=property_map[prop], interval=interval, offset=stats.min, min_doc_count=0)
927
928
929
            s.aggs.bucket("{}_hist".format(prop), hist_agg)
        response_hist = s.execute()

930
        # Return results
931
932
933
        result = {}
        for prop in properties:
            stats = getattr(response.aggs, "{}_stats".format(prop))
934
935
            if stats.count == 0:
                continue
936
937
938
939
940
941
942
            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,
943
                "histogram": {
944
945
946
                    "occurrences": occurrences,
                    "values": values,
                }
947
            }
948

949
        return result, 200
950
951
952
953
954
955
956
957
958
959
960


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,
961
    "variables": fields.List(fields.Nested(wyckoff_variables_result, skip_none=True)),
962
})
963
964
965
966
967
968
969
970
lattice_parameters = api.model("lattice_parameters", {
    "a": fields.Float,
    "b": fields.Float,
    "c": fields.Float,
    "alpha": fields.Float,
    "beta": fields.Float,
    "gamma": fields.Float,
})
971
972
973
974
975
976

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)),
977
    "lattice_parameters": fields.Nested(lattice_parameters),
978
979
980
981
982
983
    "periodicity": fields.List(fields.Boolean),
    "number_of_atoms": fields.Integer,
    "cell_volume": fields.Float,
    "wyckoff_sets": fields.List(fields.Nested(wyckoff_set_result)),
})

984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
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
1000
1001
1002
    "band_gap": {
        "es_source": "encyclopedia.properties.band_gap"
    },
1003
1004
1005
1006
1007
1008
    "electronic_band_structure": {
        "es_source": "encyclopedia.properties.electronic_band_structure"
    },
    "electronic_dos": {
        "es_source": "encyclopedia.properties.electronic_dos"
    },
1009
1010
1011
1012
1013
1014
1015
1016
1017
    "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"
    },
1018
1019
1020
    "wyckoff_sets": {
        "arch_source": "section_metadata/encyclopedia/material/idealized_structure/wyckoff_sets"
    },
1021
1022
1023
    "idealized_structure": {
        "arch_source": "section_metadata/encyclopedia/material/idealized_structure"
    },
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
}

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
1034
1035
1036
1037
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,
1038
    "section_band_gap": fields.Raw,
Lauri Himanen's avatar
Lauri Himanen committed
1039
1040
1041
})
electronic_dos = api.model("electronic_dos", {
    "dos_energies": fields.List(fields.Float),
1042
    "dos_values": fields.List(fields.List(fields.Float)),
Lauri Himanen's avatar
Lauri Himanen committed
1043
})
1044
1045
1046
calculation_property_result = api.model("calculation_property_result", {
    "lattice_parameters": fields.Nested(lattice_parameters, skip_none=True),
    "energies": fields.Nested(energies, skip_none=True),
1047
1048
1049
    "mass_density": fields.Float,
    "atomic_density": fields.Float,
    "cell_volume": fields.Float,
1050
    "wyckoff_sets": fields.Nested(wyckoff_set_result, skip_none=True),
1051
    "idealized_structure": fields.Nested(idealized_structure_result, skip_none=True),
1052
1053
1054
    "band_gap": fields.Float,
    "electronic_band_structure": fields.Nested(electronic_band_structure, skip_none=True),
    "electronic_dos": fields.Nested(electronic_dos, skip_none=True),
1055
1056
1057
    "phonon_band_structure": fields.Raw,
    "phonon_dos": fields.Raw,
    "thermodynamical_properties": fields.Raw,
1058
1059
1060
})


1061
1062
1063
1064
1065
@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)
1066
    @api.expect(calculation_property_query, validate=False)
1067
    @api.marshal_with(calculation_property_result, skip_none=True)
1068
    @api.doc("enc_calculation")
1069
1070
1071
1072
    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.
1073
        """
1074
1075
1076
1077
1078
1079
        # Get query parameters as json
        try:
            data = marshal(request.get_json(), calculation_property_query)
        except Exception as e:
            abort(400, message=str(e))

1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
        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)

1092
        # Create dictionaries for requested properties
1093
        references = []
1094
1095
1096
        properties = data["properties"]
        arch_properties = {}
        es_properties = {}
1097
1098
1099
1100
1101
1102
1103
        ref_properties = set((
            "electronic_dos",
            "electronic_band_structure",
            "thermodynamical_properties",
            "phonon_dos",
            "phonon_band_structure",
        ))
1104
1105
1106
1107
        for prop in properties:
            es_source = calculation_property_map[prop].get("es_source")
            if es_source is not None:
                es_properties[prop] = es_source
1108
                if prop in ref_properties:
1109
                    references.append(prop)
1110
1111
1112
1113
            arch_source = calculation_property_map[prop].get("arch_source")
            if arch_source is not None:
                arch_properties[prop] = arch_source

1114
        # The query is filtered already on the ES side so we don't need to
1115
        # transfer so much data.
1116
1117
1118
        sources = [
            "upload_id",
            "calc_id",
1119
            "encyclopedia",
1120
1121
1122
        ]
        sources += list(es_properties.values())

1123
        s = s.extra(**{
1124
            "_source": {"includes": sources},
1125
1126
1127
1128
1129
1130
1131
1132
1133
            "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))

1134
1135
1136
        # Add references that are to be read from the archive
        for ref in references:
            arch_path = response[0]
1137
1138
            arch_path = rgetattr(arch_path, es_properties[ref])
            if arch_path is not None:
Lauri Himanen's avatar
Lauri Himanen committed
1139
                arch_properties[ref] = arch_path
1140
1141
            del es_properties[ref]

1142
1143
1144
1145
1146
1147
1148
        # 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
1149
            root = read_archive(
1150
1151
1152
1153
1154
                upload_id,
                calc_id,
            )

            # Add results from archive
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
            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
1166

1167
                # DOS results are simplified.
1168
                if key == "electronic_dos":
Lauri Himanen's avatar
Lauri Himanen committed
1169
1170
1171
1172
1173
1174
                    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"]
1175

1176
1177
1178
                result[key] = value

        # Add results from ES
1179
        for prop, es_source in es_properties.items():
1180
1181
            value = rgetattr(response[0], es_source)
            if value is not None:
Lauri Himanen's avatar
Lauri Himanen committed
1182
1183
1184
                if isinstance(value, AttrDict):
                    value = value.to_dict()
                result[prop] = value
1185
1186
1187
1188

        return result, 200


1189
def read_archive(upload_id: str, calc_id: str) -> EntryArchive:
1190
    """Used to read data from the archive.
1191
1192
1193
1194
1195
1196

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

    Returns:
1197
        MSection: The section_run as MSection
1198
1199
        For each path, a dictionary containing the path as key and the returned
        section as value.
1200
1201
1202
1203
    """
    upload_files = files.UploadFiles.get(upload_id)
    with upload_files.read_archive(calc_id) as archive:
        data = archive[calc_id]
1204
        root = EntryArchive.m_from_dict(data.to_dict())
1205

1206
    return root