encyclopedia.py 22.4 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
# 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.

'''
The encyclopedia API of the nomad@FAIRDI APIs.
'''
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
25

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

ns = api.namespace('encyclopedia', description='Access encyclopedia metadata.')
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
re_formula = re.compile(r"([A-Z][a-z]?)(\d*)")


def add_result(result, key, function, default=""):
    """Convenience function that attempts to add a value from the ElasticSearch
    result into the given result object. Upon failing returns the specified
    default value.
    """
    try:
        value = function()
    except Exception:
        value = default
    result[key] = value


def get_material(es_doc):
47
48
    """Used to form a material definition for "materials/<material_id>" from
    the given ElasticSearch root document.
49
50
    """
    result = {}
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
    # General
    add_result(result, "material_id", lambda: es_doc.encyclopedia.material.material_id, None),
    add_result(result, "formula", lambda: es_doc.encyclopedia.material.formula, None)
    add_result(result, "formula_reduced", lambda: es_doc.encyclopedia.material.formula_reduced, None)
    add_result(result, "system_type", lambda: es_doc.encyclopedia.material.material_type, None)

    # Bulk only
    add_result(result, "has_free_wyckoff_parameters", lambda: es_doc.encyclopedia.material.bulk.has_free_wyckoff_parameters, None)
    add_result(result, "strukturbericht_designation", lambda: es_doc.encyclopedia.material.bulk.strukturbericht_designation, None)
    add_result(result, "material_name", lambda: es_doc.encyclopedia.material.material_name, None)
    add_result(result, "bravais_lattice", lambda: es_doc.encyclopedia.material.bulk.bravais_lattice, None),
    add_result(result, "crystal_system", lambda: es_doc.encyclopedia.material.bulk.crystal_system, None)
    add_result(result, "point_group", lambda: es_doc.encyclopedia.material.bulk.point_group, None)
    add_result(result, "space_group_number", lambda: es_doc.encyclopedia.material.bulk.space_group_number, None)
    add_result(result, "space_group_international_short_symbol", lambda: es_doc.encyclopedia.material.bulk.space_group_international_short_symbol, None)
    add_result(result, "structure_prototype", lambda: es_doc.encyclopedia.material.bulk.structure_prototype, None)
    add_result(result, "structure_type", lambda: es_doc.encyclopedia.material.bulk.structure_type, None)
68
69
70
71
72
73
74

    return result


material_query = api.parser()
material_query.add_argument('material_id', type=str, help='Identifier for the searched material.', location='args')
material_result = api.model('material_result', {
75
76
    # General
    "material_id": fields.String,
77
78
    "formula": fields.String,
    "formula_reduced": fields.String,
79
80
81
82
    "system_type": fields.String,
    # Bulk only
    "has_free_wyckoff_parameters": fields.String,
    "strukturbericht_designation": fields.String,
83
    "material_name": fields.String,
84
85
    "bravais_lattice": fields.String,
    "crystal_system": fields.String,
86
    "point_group": fields.String,
87
88
89
    "space_group_number": fields.Integer,
    "space_group_international_short_symbol": fields.String,
    "structure_prototype": fields.String,
90
91
    "structure_type": fields.String,
})
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
material_source = {
    "includes": [
        "encyclopedia.material.material_id",
        "encyclopedia.material.formula",
        "encyclopedia.material.formula_reduced",
        "encyclopedia.material.material_type",
        "encyclopedia.material.bulk.has_free_wyckoff_parameters",
        "encyclopedia.material.bulk.strukturbericht_designation",
        "encyclopedia.material.material_name",
        "encyclopedia.material.bulk.bravais_lattice",
        "encyclopedia.material.bulk.crystal_system",
        "encyclopedia.material.bulk.point_group",
        "encyclopedia.material.bulk.space_group_number",
        "encyclopedia.material.bulk.space_group_international_short_symbol",
        "encyclopedia.material.bulk.structure_prototype",
        "encyclopedia.material.bulk.structure_type",
    ]
}
110
111
112
113
114
115


@ns.route('/materials/<string:material_id>')
class EncMaterialResource(Resource):
    @api.response(404, 'The material does not exist')
    @api.response(200, 'Metadata send', fields.Raw)
116
117
    @api.doc('material/<material_id>')
    @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
126
127
128
129
130
131
132
133
134
135
136
137
138
        # 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(
            'bool',
            filter=[
                Q('term', published=True),
                Q('term', with_embargo=False),
                Q('term', encyclopedia__material__material_id=material_id),
            ]
        )
        s = s.query(query)
139
140
141
142
143

        # The query is collapsed already on the ES side so we don't need to
        # transfer so much data.
        s = s.extra(**{
            "collapse": {"field": "encyclopedia.material.material_id"},
144
            "_source": material_source
145
146
        })

147
148
        response = s.execute()

149
        # No such material
150
151
152
        if len(response) == 0:
            abort(404, message='There is no material {}'.format(material_id))

153
        # Create result JSON
154
        entry = response[0]
155
        result = get_material(entry)
156

157
158
159
160
161
162
163
164
165
166
167
        return result, 200


range_query = api.model('range_query', {
    "max": fields.Float,
    "min": fields.Float,
})
materials_query = api.model('materials_input', {
    'search_by': fields.Nested(api.model('search_query', {
        "exclusive": fields.Boolean(default=False),
        "formula": fields.String,
Lauri Himanen's avatar
Lauri Himanen committed
168
        "element": fields.String,
169
170
171
        "page": fields.Integer(default=1),
        "per_page": fields.Integer(default=25),
        "pagination": fields.Boolean,
172
        "mode": fields.String(default="collapse"),
173
174
175
    })),
    'material_name': fields.List(fields.String),
    'structure_type': fields.List(fields.String),
176
    'space_group_number': fields.List(fields.Integer),
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
    '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."),
})
materials_result = api.model('materials_result', {
    'total_results': fields.Integer(allow_null=False),
192
193
194
195
196
197
198
199
    'results': fields.List(fields.Nested(material_result)),
    'pages': fields.Nested(api.model("page_info", {
        "per_page": fields.Integer,
        "total": fields.Integer,
        "page": fields.Integer,
        "pages": fields.Integer,
    })),
    'es_query': fields.String(allow_null=False),
200
201
202
203
204
205
206
207
208
})


@ns.route('/materials')
class EncMaterialsResource(Resource):
    @api.response(404, 'No materials found')
    @api.response(400, 'Bad request')
    @api.response(200, 'Metadata send', fields.Raw)
    @api.expect(materials_query, validate=False)
209
    @api.marshal_with(materials_result, skip_none=True)
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
    @api.doc('materials')
    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
        filters.append(Q('term', published=True))
        filters.append(Q('term', with_embargo=False))

        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")
234
        add_terms_filter("space_group_number", "encyclopedia.material.bulk.space_group_number")
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
        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")
279

280
281
        # Create query for elements or formula
        search_by = data["search_by"]
282
        mode = search_by["mode"]
283
        formula = search_by["formula"]
Lauri Himanen's avatar
Lauri Himanen committed
284
        elements = search_by["element"]
285
286
287
        exclusive = search_by["exclusive"]

        if formula is not None:
Lauri Himanen's avatar
Lauri Himanen committed
288
289
            # Here we determine a list of atom types. The types may occur
            # multiple times and at multiple places.
290
291
292
293
294
295
296
297
298
299
            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
300
301
302
303
304
                        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.
305
            names, reduced_counts = get_hill_decomposition(element_list, reduced=True)
Lauri Himanen's avatar
Lauri Himanen committed
306
307
308
309
310
            query_string = []
            for name, count in zip(names, reduced_counts):
                if count == 1:
                    query_string.append(name)
                else:
311
                    query_string.append("{}{}".format(name, int(count)))
Lauri Himanen's avatar
Lauri Himanen committed
312
            query_string = " ".join(query_string)
313
314
315

            # With exclusive search we look for exact match
            if exclusive:
Lauri Himanen's avatar
Lauri Himanen committed
316
                filters.append(Q("term", **{"encyclopedia.material.species_and_counts.keyword": query_string}))
317
318
319
320
321
            # 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
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
                    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"}
338
339
                ))

340
341
342
343
344
345
346
347
        page = search_by["page"]
        per_page = search_by["per_page"]
        bool_query = Q(
            'bool',
            filter=filters,
            must_not=must_nots,
            must=musts,
        )
348

349
350
351
352
353
354
355
356
357
358
359
360
361
        # 1: The paginated approach: No way to know the amount of matches,
        # but can return aggregation results in a quick fashion including
        # the number of matches entries per material.
        if mode == "aggregate":
            after = None
            # The loop is awkward, but emulates the old behaviour until the GUI is adapted.
            for _ in range(page):

                # 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:
362
363
364
365
                # '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.
366
367
368
369
370
                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)
371
372
373
374
375
                composite_agg.metric('representative', A(
                    'top_hits',
                    size=1,
                    _source=material_source,
                ))
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
                s.aggs.bucket("materials", composite_agg)

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

                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"]

            # Gather results from aggregations
            result_list = []
            materials = response.aggs.materials.buckets
            for material in materials:
                representative = material["representative"][0]
                mat_dict = get_material(representative)
                mat_dict["n_of_calculations"] = material.doc_count
                result_list.append(mat_dict)

            # Page information is incomplete for aggregations
            pages = {
                "page": page,
                "per_page": per_page,
            }
        # 2. Collapse approach. Quickly provides a list of materials
        # corresponding to the query, offers full pagination, doesn't include
        # 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:
                abort(404, message='No materials found for the given search criteria or pagination.')

            # Loop over materials
            result_list = []
            for material in response:
                mat_result = get_material(material)
                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,
            }
435
436
437

        result = {
            "results": result_list,
438
439
440
            "total_results": len(result_list),
            "es_query": s.to_dict(),
            "pages": pages,
441
        }
442
        return result, 200
443
444
445
446
447
448
449
450
451
452
453
454
455
456


group_result = api.model('group_result', {
    "calculation_list": fields.List(fields.String),
    "energy_minimum": fields.Float,
    "group_hash": fields.String,
    "group_type": fields.String,
    "nr_of_calculations": fields.Integer,
    "representative_calculation_id": fields.String,
})
groups_result = api.model('groups_result', {
    'total_groups': fields.Integer(allow_null=False),
    'groups': fields.List(fields.Nested(group_result)),
})
457
458
459
460
461
462
group_source = {
    "includes": [
        "calc_id",
        "encyclopedia.properties.energies.energy_total",
    ]
}
463
464


Lauri Himanen's avatar
Lauri Himanen committed
465
466
467
468
469
470
471
@ns.route('/materials/<string:material_id>/groups')
class EncGroupsResource(Resource):
    @api.response(404, 'Material not found')
    @api.response(400, 'Bad request')
    @api.response(200, 'Metadata send', fields.Raw)
    @api.expect(material_query, validate=False)
    @api.marshal_with(groups_result, skip_none=True)
472
    @api.doc('enc_materials')
Lauri Himanen's avatar
Lauri Himanen committed
473
474
    def get(self, material_id):

475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
        # Find entries for the given material, which have EOS or parameter
        # variation hashes set.
        bool_query = Q(
            'bool',
            filter=[
                Q('term', published=True),
                Q('term', with_embargo=False),
                Q('term', encyclopedia__material__material_id=material_id),
            ],
            must=[
                Q("exists", field="encyclopedia.properties.energies.energy_total"),
            ],
            should=[
                Q("exists", field="encyclopedia.method.group_eos_hash"),
                Q("exists", field="encyclopedia.method.group_parametervariation_hash"),
            ],
            minimum_should_match=1,  # At least one of the should query must match
        )
Lauri Himanen's avatar
Lauri Himanen committed
493
494

        s = Search(index=config.elastic.index_name)
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
        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.
        group_eos_bucket = A("terms", field="encyclopedia.method.group_eos_hash", min_doc_count=4)
        group_param_bucket = A("terms", field="encyclopedia.method.group_parametervariation_hash", min_doc_count=2)

        # 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
        # 'index.max_inner_result_window' that limits the number of results
        # that an inner result can contain.
        energy_aggregation = A(
            "top_hits",
            _source=group_source,
            sort=[{"encyclopedia.properties.energies.energy_total": {"order": "asc"}}],
            size=100,
        )
        group_eos_bucket.bucket("energies", energy_aggregation)
        group_param_bucket.bucket("energies", energy_aggregation)
        s.aggs.bucket("groups_eos", group_eos_bucket)
        s.aggs.bucket("groups_param", group_param_bucket)
518

519
520
521
522
        # We ignore the top level hits
        s = s.extra(**{
            "size": 0,
        })
523

524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
        # No hits on the top query level
        response = s.execute()
        n_hits = response.hits.total
        if n_hits == 0:
            abort(404, message='The specified material could not be found.')

        # Collect information for each group from the aggregations
        groups = []
        groups_eos = response.aggs.groups_eos.buckets
        groups_param = response.aggs.groups_param.buckets

        def get_group(group, group_type, group_hash):
            hits = group.energies.hits
            calculations = [doc.calc_id for doc in hits]
            group_dict = {
                "group_hash": group_hash,
                "group_type": group_type,
                "nr_of_calculations": len(calculations),
                "representative_calculation_id": hits[0].calc_id,
                "calculation_list": calculations,
                "energy_minimum": hits[0].encyclopedia.properties.energies.energy_total,
            }
            return group_dict
Lauri Himanen's avatar
Lauri Himanen committed
547

548
549
550
551
552
553
554
555
556
557
558
        for group in groups_eos:
            groups.append(get_group(group, "equation of state", group.key))
        for group in groups_param:
            groups.append(get_group(group, "parameter variation", group.key))

        # Return results
        result = {
            "groups": groups,
            "total_groups": len(groups),
        }
        return result, 200