automaton.py 22.6 KB
Newer Older
1
# -*- coding: utf-8 -*-
2
""" harvest_tools.automaton
3 4 5 6 7

"""
import re
import traceback

8

9 10 11 12
from base import (MSG_FIX_ORIGIN,
                  MSG_IN_DB,
                  search_synonym,
                  ToolException)
13
from gluon.storage import Storage
14 15 16
from invenio_tools import (CheckAndFix,
                           InvenioStore,
                           Marc12,
17
                           OAI_URL)
18 19
from msg import Msg
from msgcollection import MsgCollection
20
from plugin_dbui import CALLBACK_ERRORS, get_id
21

22

23 24 25
MSG_NO_CAT = 'Select a "category" !!!'
MSG_NO_PROJECT = 'Select a "project" !!!'
MSG_NO_TEAM = 'Select a "team" !!!'
26

27
MSG_NSERT_FAIL = "Fail to insert the new record in the database."
28
MSG_NO_OAI = "Reject no OAI identifier"
29
MSG_WELL_FORM_OAI = "Reject OAI is not well formed"
30

31

32
class Automaton(object):
33
    """Base class to search and process publications:
34

35 36 37 38 39
        * Decode the selector defining user criteria.
        * Search for publications in the store, according to
          user criteria
        * Decode the XML string return by the store.
        * Insert new records in the database.
40

41 42
    Note:
        The parameters of the search are defined by the current ``request``.
43

44 45 46 47 48
    The logic implements in the ``Automaton`` class is the following:

        #. Ask to the store, all the `record_id` satisfying the user request.
        #. Reject `record_id` matching the `origin` field of database entry.
        #. Request to the store, the XML description of the publication
49
           and decode it.
50 51 52
        #. Check that the *oai* of the publication is defined and well formed.
           Recover it, if it is not the case. From time to time, the `id`
           encoded in the `oai` field is different from the `record_id`.
53 54 55
           This happens when an old record is redirected to new one
           for obscure reasons. The record is ignore if a database entry
           is found with the bad OAI.
56 57
        #. Reject temporarily publication.
        #. Check that *authors* are defined.
58
           Reject the publication if it is not the case.
59
        #. Check that *my institute* is in the list of the institutes
60 61 62 63 64 65
           signing the publication. Reject the publication if it is
           not the case. When the affiliation are not defined,
           try to recover this case, by finding the author of my institute
           signing the publication. This recovery procedure uses
           the *author rescue list*. Reject the record when the recovery
           procedure failed.
66
        #. Check that the *collaboration*, if defined, is well formed.
67
           Reject the publication if it is not the case
68 69 70 71 72 73 74 75
        #. Several check are applied depending on the publication type.
        #. At the end of this process, the publisher, the authors are
           formatted and the list of signatories of my institute extracted.

    Args:
        db (gluon.DAL): the database connection.
        id_team (int): the identifier of the team in the database.
        id_project (int): the identifier of the project in the database.
76
        automaton (unicode): the name of the automaton which
77 78 79
            will be used to process the data. Possible values are:
            ``articles``, ``notes``, ``preprints``, ``proceedings``,
            ``reports``, ``talks`` and ``theses``.
80 81 82 83 84 85 86 87 88 89
        id_category (int): the identifier of the category of publication
        year_start (int): starting year for the scan
        year_end (int): ending year of the scan
        dry_run (bool): new records are not inserted in the database
            when ``True``.
        debug (bool): activate the verbose mode when ``True``.

    Raises:
        ToolException: when the team or the project or the publication category
            is not defined.
90

91 92 93 94 95
    """
    def __init__(self,
                 db,
                 id_team,
                 id_project,
96
                 automaton,
97 98 99 100 101 102 103 104 105 106
                 id_category,
                 year_start=None,
                 year_end=None,
                 dry_run=True,
                 debug=False):

        self.collection_logs = []
        self.db = db
        self.id_team = id_team
        self.id_project = id_project
107
        self.controller = automaton
108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134
        self.id_category = id_category
        self.year_start = year_start
        self.year_end = year_end
        self.dry_run = dry_run
        self.dbg = debug
        self.logs = []

        self.check = CheckAndFix()
        self.marc12 = Marc12()

        # check parameters
        # protection team, project and/or category have to be defined
        if not self.id_team:
            raise ToolException(MSG_NO_TEAM)

        if not self.id_project:
            raise ToolException(MSG_NO_PROJECT)

        if not self.id_category:
            raise ToolException(MSG_NO_CAT)

        # Construct harvester Storage needed for the log
        self.harvester = Storage(id_teams=self.id_team,
                                 id_projects=self.id_project,
                                 controller=self.controller,
                                 id_categories=self.id_category)

135 136 137
    def _insert_in_db(self, log_year="", **fields):
        """Insert the record in the database, handling database exception.

138 139
        Args:
            log_year (unicode): year of the record for the log
140

141 142 143
        Keyword Args:
            **fields: keyword arguments defining the record values to be
                inserted in the database.
144

145 146 147
        Returns:
            int: one when the record is inserted / updated in the database,
                zero otherwise.
148 149 150 151 152

        """
        db = self.db

        try:
LE GAC Renaud's avatar
LE GAC Renaud committed
153
            rec_id = db.publications.insert(**fields)
LE GAC Renaud's avatar
LE GAC Renaud committed
154 155
            if rec_id:
                return 1
156

LE GAC Renaud's avatar
LE GAC Renaud committed
157 158 159 160 161
            # operation can be reject by callbacks table._before_insert
            else:
                msg = MSG_NSERT_FAIL
                if CALLBACK_ERRORS in db.publications:
                    msg = db.publications._callback_errors
162

LE GAC Renaud's avatar
LE GAC Renaud committed
163 164 165
                # reduce the error message
                if isinstance(msg, list):
                    msg = "%s %s" % (msg[0], msg[-1])
166

LE GAC Renaud's avatar
LE GAC Renaud committed
167 168
                self.logs[-1].reject(msg, log_year)
                return 0
169

LE GAC Renaud's avatar
LE GAC Renaud committed
170 171 172 173
        # operation can be rejected by the database
        except Exception as dbe:
            self.logs[-1].reject(dbe.message, log_year)
            return 0
174

175
    def _is_record_in_db(self, title, host=None, rec_id=None, oai_url=None):
176
        """Return ``True`` when the record is already in the database.
177
        The search is based on the origin field and on the primary OAI.
178

179 180
        Note:
            A new log entry is created when a record is found.
181

182
        Args:
183 184 185 186 187
            title (unicode): the title of the publication.

        Keyword Args:
            host (unicode): the store. possible values are ``cds.cern.ch`` or
                ``inspirehep.net``. To be used with *rec_id*.
188
            rec_id (int): the record identifier in the store
189 190 191 192
            oai_url (unicode): the URL of the record in the store.

                Note:
                    Either use *host* and *rec_id* or *oai_url*
193

194
        Returns:
195 196
            int: the id of the record in the database when a record is found,
                0 otherwise.
197

198 199 200
        Raises:
            ValueError: when keyword arguments are not defined properly.

201 202
        """
        db = self.db
203
        harvester = self.harvester
204

205 206 207 208 209 210 211 212 213 214 215
        # build the OAI URL
        if host is not None and rec_id is not None and oai_url is None:
            url = OAI_URL % (host, rec_id)
        elif host is None and rec_id is None and oai_url is not None:
            url = oai_url
        else:
            raise ValueError

        # check the OAI
        query = db.publications.origin.contains(url)
        setrows = db(query)
216

217
        if setrows.count() == 0:
218
            return 0
219

220
        # one record found
221 222
        columns = [db.publications.id,
                   db.publications.id_categories,
223 224 225
                   db.publications.title,
                   db.publications.year]
        publication = setrows.select(*columns).first()
226

227 228 229 230
        # Note:
        # The category for the publication and the harvester have to be equal.
        # However, keep the record if it is not the case.
        # This is required to transform a preprint into article
231
        if publication.id_categories != harvester.id_categories:
232
            return 0
233 234

        # log
235
        self.logs.append(Msg(harvester=harvester,
236 237 238 239 240 241
                             collection=title,
                             record_id=rec_id,
                             title=publication.title))

        self.logs[-1].idle(MSG_IN_DB, publication.year)

242
        return publication.id
243

244 245 246 247 248
    def _search_parameters(self, collection):
        """Build the keywords to steer the URL search in invenio store.
        The main parameter is the collection and the date range defined
        in the selector.

249 250 251 252 253 254
        Args:
            collection (unicode): string defining the collection in the
                store. The syntax depends on the invenio store:

                    * ``"find cn d0 and tc p and not tc c"``
                    * ``"LHCb Papers"``.
255

256 257 258
        Returns:
            dict: the key are a sub-set of those defined in
                :py:meth:invenio_tools.InvenioStore.get_ids.
259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304

        """

        # INSPIREHEP store
        if collection.startswith('find'):

            query = collection

            if self.year_start and not self.year_end:
                query += " and date %s" % self.year_start

            elif not self.year_start and self.year_end:
                query += " and date %s" % self.year_end

            elif self.year_start and self.year_end:
                query += " and date > %s and date < %s " \
                         % (self.year_start - 1, self.year_end + 1)

            dic = dict(p=query,  # query à la spires
                       rg=1000,  # maximum number of records returned
                       sf='year',  # sort by date
                       so='d')  # descending order

        # CERN INVENIO store
        else:

            if self.year_start and not self.year_end:
                rex = self.year_start

            elif not self.year_start and self.year_end:
                rex = self.year_end

            elif self.year_start and self.year_end:
                li = []
                for year in range(self.year_start, self.year_end + 1):
                    li.append(str(year))
                rex = '|'.join(li)

            dic = dict(cc=collection,  # collection
                       f1='year',  # search on year
                       m1='r',  # use regular expression
                       p1=rex,  # regular expression defining year
                       sf='year',  # sort by date
                       so='d')  # descending order
        return dic

LE GAC Renaud's avatar
LE GAC Renaud committed
305
    def check_record(self, record):
306 307
        """Check the content of the record in order to fix non-conformities.
        Return ``False`` when non-conformities are found and can not be
308 309
        corrected.

310 311 312
        Note:
            Some checks depend on the type of publications and have to be
            implemented in inherited class.
313

314 315 316 317
        Note:
            The order of the checks matter. It should be oai,
            temporary record, authors, my authors and then a series of checks
            specific to the publication type.
318

319 320
        Args:
            record (Record): Marc12 record describing the publication.
321

322 323 324
        Returns:
            bool: ``False`` when a non-conformity is found and can not be
                corrected.
325 326 327 328 329 330

        """
        if self.dbg:
            print "check record"

        try:
331 332
            self.check.recover_oai(record, self.harvester.host)

333 334 335 336
            if self.check.is_bad_oai_used(record):
                self.logs[-1].idle(MSG_IN_DB, record.year())
                return False

337 338
            self.check.temporary_record(record)
            self.check.authors(record)
339
            self.check.my_affiliation(record, self.id_project, self.id_team)
340 341 342
            self.check.collaboration(record)

        except Exception as e:
343
            self.logs[-1].reject(e, record=record)
344 345 346 347
            return False

        return True

348
    def get_record_by_fields(self, oai_url, year, **kwargs):
349 350
        """Get database record matching fields values defined
        in the keyword arguments.
351

352
        Note:
353 354
            This method is required to deal with publication entered by hand
            and found later by an harvester.
355

356 357 358 359 360 361 362 363 364 365 366 367
        Args:
            oai_url (unicode): the oai_url, *e.g*
                ``http://cds.cern.ch/record/123456``. The origin field
                of the existing database record is update to **oai_url**
                when a match is found.

            year (int): the year of the publication. It is used
                by the search algorithm and by the logger.

        Keyword Args:
            kwargs (unicode): a series of key, value pair where the
                key is the name of a publications database field.
368

369 370 371 372 373
        Returns:
            tuple: ``(id, status)`` which contains the ``id`` of the record.
            The ``id`` is equal to ``None`` when there is no matching.
            The ``status`` is equal to one when the existing record was
            modified zero otherwise.
374 375 376

        """
        if self.dbg:
377
            print "get existing record by fields"
378

379
        # alias
380
        db = self.db
381
        logs = self.logs
382

383 384 385
        # add the publication year to search criteria
        if year:
            kwargs["year"] = year
386 387 388 389 390 391 392

        # look for an existing record
        rec_id = get_id(db.publications, **kwargs)
        if not rec_id:
            return (None, 0)

        # fix origin field
393 394
        publication = db.publications[rec_id]
        ok = publication.origin and publication.origin == oai_url
395 396
        if not ok:
            if not self.dry_run:
397
                publication = dict(origin=oai_url)
398

399
            logs[-1].modify(MSG_FIX_ORIGIN, year)
400 401
            return (rec_id, 1)

402
        logs[-1].idle(MSG_IN_DB, year)
403 404
        return (rec_id, 0)

405 406
    def insert_record(self, record):
        """Insert the record in the database.
407

408 409 410
        Note:
            This method depend on the type of publications.
            It has to be implemented for each inherited class.
411

412 413
        Args:
            record (Record): Marc12 record describing the publication.
414

415 416 417
        Returns:
            int: one when the record is inserted / updated in the database,
                zero otherwise.
418 419 420 421 422

        """
        return 0

    def process_xml(self, xml):
423 424 425 426 427 428 429
        """Decode the XML string and insert the corresponding records
        in the database.


        Args:
            xml (unicode): XML string encoding the publication records.
                The format follows the MARC12 standard.
430

431 432
        Raises:
           ToolException: when project, team or category identifier
433
              are not defined.
434
           StoreException: when something goes wrong interrogating the
435
              store.
436 437 438 439
           Marc12Exception: when something goes wrong decoding the XML
              string return by the store.
           CheckException: when the record has non-conformities.
           Exception: when the python code crashes.
440 441 442 443 444 445 446 447 448 449 450 451 452 453

        """
        if self.dbg:
            print "start processing", self.__class__.__name__
            print "decode request"

        if self.dbg:
            print "get harvest parameters"

        # decode the XML request
        self.collection_logs.append(MsgCollection(found=1))
        self.decode_xml(xml)

    def process_url(self, host, collections):
454 455
        """Retrieve the XML string from the invenio store and
        insert corresponding records in the database.
456

457 458 459
        Args:
            host (unicode): host name to query for publications, either
                ``cds.cern.ch`` or ``inspirehep.net``.
460

461
            collections (unicode): list of collection to be interrogated.
462

463 464 465 466 467 468 469
        Raises:
           StoreException: when something goes wrong interrogating the
              store.
           Marc12Exception: when something goes wrong decoding the XML
              string return by the store.
           CheckException: when the record has non-conformities.
           Exception: when the python code crashes.
470 471 472 473 474 475 476 477 478 479 480 481 482 483 484

        """
        if self.dbg:
            print "process URL search"

        # extend harvester for logs
        self.harvester.host = host
        self.harvester.collections = collections

        store = InvenioStore(host)

        # list of collections
        collections = re.sub(' *, *', ',', collections).split(',')

        # alias
485
        collection_logs = self.collection_logs
486
        controller = self.controller
487 488 489
        decode_xml = self.decode_xml
        is_record_in_db = self._is_record_in_db
        logs = self.logs
490 491 492 493 494 495
        project = self.db.projects[self.id_project].project

        # extract the list of publications from the store for each collection
        # the search is perform on a range of creation date
        # if not defined all element are return
        #
496
        # The method uses here minimise the memory usage
497 498 499 500 501 502 503
        # on the server as well as on the client side

        for collection in collections:

            # log collection information
            # A collection is identified as "Project Controller collection"
            title = "%s / %s / %s" % (project, controller, collection)
504
            collection_logs.append(MsgCollection(title=title))
505 506 507 508 509 510 511 512

            # search record in the harvester repository
            kwargs = self._search_parameters(collection)

            try:
                rec_ids = store.get_ids(**kwargs)

            except Exception as error:
513 514
                collection_logs[-1].url = store.last_search_url()
                collection_logs[-1].error = error
515 516
                continue

517 518
            collection_logs[-1].url = store.last_search_url()
            collection_logs[-1].found = len(rec_ids)
519 520 521 522 523 524 525 526 527 528 529 530 531

            if not rec_ids:
                continue

            if self.dbg:
                print '%i records found in %s' % (len(rec_ids), collection)

            for rec_id in rec_ids:

                if self.dbg:
                    print "\nprocessing record", rec_id

                try:
532
                    if is_record_in_db(title, host=host, rec_id=rec_id):
533 534
                        continue

535
                    xml = store.get_record(rec_id)
536
                    decode_xml(xml)
537

538
                except Exception as e:
539 540
                    print traceback.format_exc()
                    url = OAI_URL % (host, rec_id)
541 542 543 544 545
                    logs.append(Msg(harvester=self.harvester,
                                    collection=title,
                                    record_id=rec_id,
                                    title=url))
                    logs[-1].reject(e)
546 547

    def decode_xml(self, xml):
548
        """Decode the MARC XML string and insert records in the database.
549

550 551
        Args:
            xml (unicode): MARC XML string
552 553

        """
554

555 556 557
        if self.dbg:
            print "process xml record"

558
        # alias
559
        db = self.db
560 561
        check_record = self.check_record
        insert_record = self.insert_record
562
        is_record_in_db = self._is_record_in_db
563 564
        logs = self.logs

565 566 567
        # NOTE
        # BaseException and inherited class
        # are caught by the previous stage
568 569 570 571 572 573 574 575
        records = self.marc12(xml)

        # process individual record
        for record in records:

            if self.dbg:
                print "record decoded"

576 577
            # reject the record using the secondary OAI
            # require to cover the case:
578
            # - scan store A. Only OAI_A is defined.
579
            # - later scan store B. OAI_B and OAI_A are found.
580 581 582 583 584 585 586 587
            rec_id = is_record_in_db(record.title(),
                                     oai_url=record.secondary_oai_url())
            if rec_id:
                # update the oai_url for later uses.
                if not self.dry_run:
                    tp = (record.secondary_oai_url(), record.primary_oai_url())
                    oai_url = "%s, %s" % tp
                    db.publications[rec_id] = dict(origin=oai_url)
588 589
                continue

590
            # start the log for the record
591 592 593 594
            logs.append(Msg(harvester=self.harvester,
                            collection=self.collection_logs[-1].title,
                            record_id=record.id(),
                            title=record.title()))
595

596 597
            # check that the record is well formed
            # repair non-conformity as far as possible
598
            if not check_record(record):
599 600 601
                continue

            if self.dbg:
602
                print "insert record in the database"
603

604
            # insert the record in the database
605
            insert_record(record)
606 607

            if self.dbg:
608
                print logs[-1].action.upper(), logs[-1].txt
609 610 611 612

    def report(self):
        """Build the processing report.

613 614 615 616 617 618
        Returns:
            dict:
                * ``collection_logs`` list of :class:MsgCollection
                * ``controller`` unicode
                * ``logs`` list of :class:Msg
                * ``selector`` :py:class:plugin_dbui.Selector
619 620 621 622 623 624

        """

        return dict(collection_logs=self.collection_logs,
                    controller=self.controller,
                    logs=self.logs)
625 626

    def search_collaboration(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
627
        """Get the database collaboration identifier using synonyms.
628
        Create the collaboration, if it is well formed and does not exist.
629 630 631 632 633 634

        Args:
            value (unicode): the name of the collaboration.

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
635
                * the id of the collaboration record.
636 637
                * UNDEF_ID if value is not defined.

638
        Raises:
639 640 641
            ToolException: when more than one synonym is found.

        """
642 643 644 645 646

        return search_synonym(self.db.collaborations,
                              "collaboration",
                              value,
                              True)
647 648

    def search_country(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
649
        """Get the database country identifier using synonyms.
650 651 652 653 654 655

        Args:
            value (unicode): the name of the country.

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
656
                * the id of the country record.
657 658
                * UNDEF_ID if value is not defined.

659
        Raises:
660 661
            ToolException: when more than one synonym is found ot when
                the country is not defined.
662 663

        """
664
        return search_synonym(self.db.countries, "country", value)
665 666

    def search_publisher(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
667
        """Get the database publisher identifier using synonyms.
668 669 670 671 672 673

        Args:
            value (unicode): the abbreviation of the publisher.

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
674
                * the id of the publisher record.
675 676
                * UNDEF_ID if value is not defined.

677
        Raises:
678 679
            ToolException: when more than one synonym is found or when
                the publisher is not defined.
680 681

        """
682
        return search_synonym(self.db.publishers, "abbreviation", value)