automaton.py 21.9 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:
183 184 185 186 187 188 189 190 191 192
            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*.
            rec_id (int): the record identifier.
            oai_url (unicode): the URL of the record in the store.

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

194 195
        Returns:
            bool: ``True`` when a record if found, ``False`` otherwise.
196

197 198 199
        Raises:
            ValueError: when keyword arguments are not defined properly.

200 201
        """
        db = self.db
202
        harvester = self.harvester
203

204 205 206 207 208 209 210 211 212 213 214
        # 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)
215

216
        if setrows.count() == 0:
217 218
            return False

219 220 221 222 223
        # one record found
        columns = [db.publications.id_categories,
                   db.publications.title,
                   db.publications.year]
        publication = setrows.select(*columns).first()
224

225 226 227 228
        # 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
229 230 231 232
        if publication.id_categories != harvester.id_categories:
            return False

        # log
233
        self.logs.append(Msg(harvester=harvester,
234 235 236 237 238 239 240 241
                             collection=title,
                             record_id=rec_id,
                             title=publication.title))

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

        return True

242 243 244 245 246
    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.

247 248 249 250 251 252
        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"``.
253

254 255 256
        Returns:
            dict: the key are a sub-set of those defined in
                :py:meth:invenio_tools.InvenioStore.get_ids.
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302

        """

        # 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
303
    def check_record(self, record):
304 305
        """Check the content of the record in order to fix non-conformities.
        Return ``False`` when non-conformities are found and can not be
306 307
        corrected.

308 309 310
        Note:
            Some checks depend on the type of publications and have to be
            implemented in inherited class.
311

312 313 314 315
        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.
316

317 318
        Args:
            record (Record): Marc12 record describing the publication.
319

320 321 322
        Returns:
            bool: ``False`` when a non-conformity is found and can not be
                corrected.
323 324 325 326 327 328

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

        try:
329 330
            self.check.recover_oai(record, self.harvester.host)

331 332 333 334
            if self.check.is_bad_oai_used(record):
                self.logs[-1].idle(MSG_IN_DB, record.year())
                return False

335 336
            self.check.temporary_record(record)
            self.check.authors(record)
337
            self.check.my_affiliation(record, self.id_project, self.id_team)
338 339 340
            self.check.collaboration(record)

        except Exception as e:
341
            self.logs[-1].reject(e, record=record)
342 343 344 345
            return False

        return True

346
    def get_record_by_fields(self, **kwargs):
347 348
        """Get database record matching fields values defined
        in the keyword arguments.
349

350
        Keyword Args:
351 352
            oai_url (unicode): *e.g* ``"http://cds.cern.ch/record/123456"``
            year (int): the year of the publication.
353

354 355
        Note:
            Fix the field origin when a match is found.
356

357 358
        Note:
            The year is only used by the logger.
359

360 361 362 363 364
        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.
365 366 367

        """
        if self.dbg:
368
            print "get existing record by fields"
369

370
        # alias
371
        db = self.db
372
        logs = self.logs
373 374 375 376 377 378 379 380 381 382 383

        # origin can't be used for the search
        oai_url = kwargs["oai_url"]
        del kwargs["oai_url"]

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

        # fix origin field
384 385
        publication = db.publications[rec_id]
        ok = publication.origin and publication.origin == oai_url
386 387
        if not ok:
            if not self.dry_run:
388
                publication = dict(origin=oai_url)
389

390
            logs[-1].modify(MSG_FIX_ORIGIN, kwargs["year"])
391 392
            return (rec_id, 1)

393
        logs[-1].idle(MSG_IN_DB, kwargs["year"])
394 395
        return (rec_id, 0)

396 397
    def insert_record(self, record):
        """Insert the record in the database.
398

399 400 401
        Note:
            This method depend on the type of publications.
            It has to be implemented for each inherited class.
402

403 404
        Args:
            record (Record): Marc12 record describing the publication.
405

406 407 408
        Returns:
            int: one when the record is inserted / updated in the database,
                zero otherwise.
409 410 411 412 413

        """
        return 0

    def process_xml(self, xml):
414 415 416 417 418 419 420
        """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.
421

422 423
        Raises:
           ToolException: when project, team or category identifier
424
              are not defined.
425
           StoreException: when something goes wrong interrogating the
426
              store.
427 428 429 430
           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.
431 432 433 434 435 436 437 438 439 440 441 442 443 444

        """
        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):
445 446
        """Retrieve the XML string from the invenio store and
        insert corresponding records in the database.
447

448 449 450
        Args:
            host (unicode): host name to query for publications, either
                ``cds.cern.ch`` or ``inspirehep.net``.
451

452
            collections (unicode): list of collection to be interrogated.
453

454 455 456 457 458 459 460
        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.
461 462 463 464 465 466 467 468 469 470 471 472 473 474 475

        """
        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
476
        collection_logs = self.collection_logs
477
        controller = self.controller
478 479 480
        decode_xml = self.decode_xml
        is_record_in_db = self._is_record_in_db
        logs = self.logs
481 482 483 484 485 486
        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
        #
487
        # The method uses here minimise the memory usage
488 489 490 491 492 493 494
        # 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)
495
            collection_logs.append(MsgCollection(title=title))
496 497 498 499 500 501 502 503

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

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

            except Exception as error:
504 505
                collection_logs[-1].url = store.last_search_url()
                collection_logs[-1].error = error
506 507
                continue

508 509
            collection_logs[-1].url = store.last_search_url()
            collection_logs[-1].found = len(rec_ids)
510 511 512 513 514 515 516 517 518 519 520 521 522

            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:
523
                    if is_record_in_db(title, host=host, rec_id=rec_id):
524 525
                        continue

526
                    xml = store.get_record(rec_id)
527
                    decode_xml(xml)
528

529
                except Exception as e:
530 531
                    print traceback.format_exc()
                    url = OAI_URL % (host, rec_id)
532 533 534 535 536
                    logs.append(Msg(harvester=self.harvester,
                                    collection=title,
                                    record_id=rec_id,
                                    title=url))
                    logs[-1].reject(e)
537 538

    def decode_xml(self, xml):
539
        """Decode the MARC XML string and insert records in the database.
540

541 542
        Args:
            xml (unicode): MARC XML string
543 544

        """
545

546 547 548
        if self.dbg:
            print "process xml record"

549 550 551 552 553 554
        # alias
        is_record_in_db = self._is_record_in_db
        check_record = self.check_record
        insert_record = self.insert_record
        logs = self.logs

555 556 557
        # NOTE
        # BaseException and inherited class
        # are caught by the previous stage
558 559 560 561 562 563 564 565
        records = self.marc12(xml)

        # process individual record
        for record in records:

            if self.dbg:
                print "record decoded"

566 567 568 569 570 571 572 573 574
            # reject the record using the secondary OAI
            # require to cover the case:
            # - san store A. Only OAI_A is defined.
            # - later scan store B. OAI_B and OAI_A are found.
            ok = is_record_in_db(record.title(),
                                 oai_url=record.secondary_oai_url())
            if ok:
                continue

575
            # start the log for the record
576 577 578 579
            logs.append(Msg(harvester=self.harvester,
                            collection=self.collection_logs[-1].title,
                            record_id=record.id(),
                            title=record.title()))
580

581 582
            # check that the record is well formed
            # repair non-conformity as far as possible
583
            if not check_record(record):
584 585 586
                continue

            if self.dbg:
587
                print "insert record in the database"
588

589
            # insert the record in the database
590
            insert_record(record)
591 592

            if self.dbg:
593
                print logs[-1].action.upper(), logs[-1].txt
594 595 596 597

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

598 599 600 601 602 603
        Returns:
            dict:
                * ``collection_logs`` list of :class:MsgCollection
                * ``controller`` unicode
                * ``logs`` list of :class:Msg
                * ``selector`` :py:class:plugin_dbui.Selector
604 605 606 607 608 609

        """

        return dict(collection_logs=self.collection_logs,
                    controller=self.controller,
                    logs=self.logs)
610 611

    def search_collaboration(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
612
        """Get the database collaboration identifier using synonyms.
613
        Create the collaboration, if it is well formed and does not exist.
614 615 616 617 618 619

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
620
                * the id of the collaboration record.
621 622
                * UNDEF_ID if value is not defined.

623
        Raises:
624 625 626
            ToolException: when more than one synonym is found.

        """
627 628 629 630 631

        return search_synonym(self.db.collaborations,
                              "collaboration",
                              value,
                              True)
632 633

    def search_country(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
634
        """Get the database country identifier using synonyms.
635 636 637 638 639 640

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
641
                * the id of the country record.
642 643
                * UNDEF_ID if value is not defined.

644
        Raises:
645 646
            ToolException: when more than one synonym is found ot when
                the country is not defined.
647 648

        """
649
        return search_synonym(self.db.countries, "country", value)
650 651

    def search_publisher(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
652
        """Get the database publisher identifier using synonyms.
653 654 655 656 657 658

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
659
                * the id of the publisher record.
660 661
                * UNDEF_ID if value is not defined.

662
        Raises:
663 664
            ToolException: when more than one synonym is found or when
                the publisher is not defined.
665 666

        """
667
        return search_synonym(self.db.publishers, "abbreviation", value)