automaton.py 24 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 checkandfix import CheckAndFix
14
from gluon.storage import Storage
15
from invenio_tools import (InvenioStore,
16
                           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 33 34
# search collection when using inspirehep
# require for "Hal Hidden"
REG_COLLECTION = re.compile(r"cc([A-Za-z ]+)(and|$)")

35

36
class Automaton(object):
37
    """Base class to search and process publications:
38

39
        * Decode the selector defining user criteria.
LE GAC Renaud's avatar
LE GAC Renaud committed
40
        * Search in the store publications matching user criteria.
41 42
        * Decode the XML string return by the store.
        * Insert new records in the database.
43

44 45
    Note:
        The parameters of the search are defined by the current ``request``.
46

47 48 49
    The logic implements in the ``Automaton`` class is the following:

        #. Ask to the store, all the `record_id` satisfying the user request.
LE GAC Renaud's avatar
LE GAC Renaud committed
50 51 52 53 54 55 56
        #. Reject `record_id` contains in the *origin* field of a
           database entry.
        #. Request to the store, the XML description of the publications
           and decode them.
        #. Reject the record for which the *secondary_oai_url* is contained in
           the *origin* field of a database entry. Update the *origin* field
           of the database record.
57
        #. Check that the *oai* of the publication is defined and well formed.
LE GAC Renaud's avatar
LE GAC Renaud committed
58 59
           Recover it, if it is not the case. At this stage the OAI is always
           defined.
60 61
        #. Reject temporarily publication.
        #. Check that *authors* are defined.
62
           Reject the publication if it is not the case.
63
        #. Check that *my institute* is in the list of the institutes
64 65 66 67 68 69
           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.
70
        #. Check that the *collaboration*, if defined, is well formed.
71
           Reject the publication if it is not the case
72 73 74 75 76 77 78 79
        #. 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.
80
        automaton (unicode): the name of the automaton which
81 82 83
            will be used to process the data. Possible values are:
            ``articles``, ``notes``, ``preprints``, ``proceedings``,
            ``reports``, ``talks`` and ``theses``.
84 85 86 87 88 89 90 91 92 93
        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.
94

95 96 97 98 99
    """
    def __init__(self,
                 db,
                 id_team,
                 id_project,
100
                 automaton,
101 102 103 104 105 106 107 108 109 110
                 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
111
        self.controller = automaton
112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138
        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)

139 140 141 142 143
        # Identifier of the categories preprint and articles
        # Used by the method _is_record_in_db
        self._id_preprint = get_id(db.categories, code="PRE")
        self._id_article = get_id(db.categories, code="ACL")

144 145 146
    def _insert_in_db(self, log_year="", **fields):
        """Insert the record in the database, handling database exception.

147 148
        Args:
            log_year (unicode): year of the record for the log
149

150 151 152
        Keyword Args:
            **fields: keyword arguments defining the record values to be
                inserted in the database.
153

154 155 156
        Returns:
            int: one when the record is inserted / updated in the database,
                zero otherwise.
157 158 159 160 161

        """
        db = self.db

        try:
LE GAC Renaud's avatar
LE GAC Renaud committed
162
            rec_id = db.publications.insert(**fields)
LE GAC Renaud's avatar
LE GAC Renaud committed
163 164
            if rec_id:
                return 1
165

LE GAC Renaud's avatar
LE GAC Renaud committed
166 167 168 169 170
            # 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
171

LE GAC Renaud's avatar
LE GAC Renaud committed
172 173 174
                # reduce the error message
                if isinstance(msg, list):
                    msg = "%s %s" % (msg[0], msg[-1])
175

LE GAC Renaud's avatar
LE GAC Renaud committed
176 177
                self.logs[-1].reject(msg, log_year)
                return 0
178

LE GAC Renaud's avatar
LE GAC Renaud committed
179 180 181 182
        # operation can be rejected by the database
        except Exception as dbe:
            self.logs[-1].reject(dbe.message, log_year)
            return 0
183

184
    def _is_record_in_db(self, title, host=None, rec_id=None, oai_url=None):
185
        """Return ``True`` when the record is already in the database.
186
        The search is based on the origin field and on the primary OAI.
187

188 189
        Note:
            A new log entry is created when a record is found.
190

191
        Args:
192 193 194 195 196
            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*.
197
            rec_id (int): the record identifier in the store
198 199 200 201
            oai_url (unicode): the URL of the record in the store.

                Note:
                    Either use *host* and *rec_id* or *oai_url*
202

203
        Returns:
204 205
            int: the id of the record in the database when a record is found,
                0 otherwise.
206

207 208 209
        Raises:
            ValueError: when keyword arguments are not defined properly.

210 211
        """
        db = self.db
212
        harvester = self.harvester
213

214 215 216 217 218 219 220 221
        # 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

222 223 224 225
        # protection empty url
        if len(url) == 0:
            return 0

226 227 228
        # check the OAI
        query = db.publications.origin.contains(url)
        setrows = db(query)
229

230
        if setrows.count() == 0:
231
            return 0
232

233
        # one record found
234 235
        columns = [db.publications.id,
                   db.publications.id_categories,
236 237 238
                   db.publications.title,
                   db.publications.year]
        publication = setrows.select(*columns).first()
239

240 241
        # Note:
        # The category for the publication and the harvester have to be equal.
242 243 244 245 246 247 248
        # However, keep the record if it is a preprint when the harvester
        # looks for articles. This is required to transform a preprint
        # into article
        #
        # Category can disagree when the publication is an article and
        # the harvester look for preprint. In that case, keep the article
        #
249
        if publication.id_categories != harvester.id_categories:
250 251 252 253 254 255 256

            is_preprint_to_article = \
                publication.id_categories == self._id_preprint \
                and harvester.id_categories == self._id_article

            if is_preprint_to_article:
                return 0
257 258

        # log
259
        self.logs.append(Msg(harvester=harvester,
260 261 262 263 264 265
                             collection=title,
                             record_id=rec_id,
                             title=publication.title))

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

266
        return publication.id
267

268 269 270 271 272
    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.

273 274 275 276 277 278
        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"``.
279

280 281 282
        Returns:
            dict: the key are a sub-set of those defined in
                :py:meth:invenio_tools.InvenioStore.get_ids.
283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305

        """

        # 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

306 307 308 309 310 311 312 313 314
            # handle the cc keyword (true inspirehep collection)
            match = REG_COLLECTION.search(query)
            if match:
                dic["cc"] = match.group(1).strip()
                dic["p"] = REG_COLLECTION.sub("", query).strip()
                dic["p"] = dic["p"].replace("  ", " ")
                if dic["p"] == "find":
                    del dic["p"]

315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337
        # 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
338
    def check_record(self, record):
339 340
        """Check the content of the record in order to fix non-conformities.
        Return ``False`` when non-conformities are found and can not be
341 342
        corrected.

343 344 345
        Note:
            Some checks depend on the type of publications and have to be
            implemented in inherited class.
346

347 348 349 350
        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.
351

352 353
        Args:
            record (Record): Marc12 record describing the publication.
354

355 356 357
        Returns:
            bool: ``False`` when a non-conformity is found and can not be
                corrected.
358 359 360 361 362 363

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

        try:
364 365
            self.check.recover_oai(record, self.harvester.host)

366 367 368 369
            if self.check.is_bad_oai_used(record):
                self.logs[-1].idle(MSG_IN_DB, record.year())
                return False

370 371
            self.check.temporary_record(record)
            self.check.authors(record)
372
            self.check.my_affiliation(record, self.id_project, self.id_team)
373 374 375
            self.check.collaboration(record)

        except Exception as e:
376
            self.logs[-1].reject(e, record=record)
377 378 379 380
            return False

        return True

381
    def get_record_by_fields(self, oai_url, year, **kwargs):
382 383
        """Get database record matching fields values defined
        in the keyword arguments.
384

385
        Note:
386 387
            This method is required to deal with publication entered by hand
            and found later by an harvester.
388

389 390 391 392 393 394 395 396 397 398 399 400
        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.
401

402 403 404 405 406
        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.
407 408 409

        """
        if self.dbg:
410
            print "get existing record by fields"
411

412
        # alias
413
        db = self.db
414
        logs = self.logs
415

416 417 418
        # add the publication year to search criteria
        if year:
            kwargs["year"] = year
419 420 421 422 423 424 425

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

        # fix origin field
426 427
        publication = db.publications[rec_id]
        ok = publication.origin and publication.origin == oai_url
428 429
        if not ok:
            if not self.dry_run:
430
                publication = dict(origin=oai_url)
431

432
            logs[-1].modify(MSG_FIX_ORIGIN, year)
433 434
            return (rec_id, 1)

435
        logs[-1].idle(MSG_IN_DB, year)
436 437
        return (rec_id, 0)

438 439
    def insert_record(self, record):
        """Insert the record in the database.
440

441 442 443
        Note:
            This method depend on the type of publications.
            It has to be implemented for each inherited class.
444

445 446
        Args:
            record (Record): Marc12 record describing the publication.
447

448 449 450
        Returns:
            int: one when the record is inserted / updated in the database,
                zero otherwise.
451 452 453 454 455

        """
        return 0

    def process_xml(self, xml):
456 457 458 459 460 461 462
        """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.
463

464 465
        Raises:
           ToolException: when project, team or category identifier
466
              are not defined.
467
           StoreException: when something goes wrong interrogating the
468
              store.
469 470 471 472
           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.
473 474 475 476 477 478 479 480 481 482 483 484 485 486

        """
        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):
487 488
        """Retrieve the XML string from the invenio store and
        insert corresponding records in the database.
489

490 491 492
        Args:
            host (unicode): host name to query for publications, either
                ``cds.cern.ch`` or ``inspirehep.net``.
493

494
            collections (unicode): list of collection to be interrogated.
495

496 497 498 499 500 501 502
        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.
503 504 505 506 507 508 509 510 511 512 513 514 515 516 517

        """
        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
518
        collection_logs = self.collection_logs
519
        controller = self.controller
520 521 522
        decode_xml = self.decode_xml
        is_record_in_db = self._is_record_in_db
        logs = self.logs
523 524 525 526 527 528
        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
        #
529
        # The method uses here minimise the memory usage
530 531 532 533 534 535 536
        # 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)
537
            collection_logs.append(MsgCollection(title=title))
538 539 540 541 542 543 544 545

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

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

            except Exception as error:
546 547
                collection_logs[-1].url = store.last_search_url()
                collection_logs[-1].error = error
548 549
                continue

550 551
            collection_logs[-1].url = store.last_search_url()
            collection_logs[-1].found = len(rec_ids)
552 553 554 555 556 557 558 559 560 561 562 563 564

            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:
565 566 567 568
                    db_id = is_record_in_db(title, host=host, rec_id=rec_id)
                    if db_id:
                        if self.dbg:
                            print "record in db", rec_id, "->", db_id
569 570
                        continue

571
                    xml = store.get_record(rec_id)
572
                    decode_xml(xml)
573

574
                except Exception as e:
575 576
                    print traceback.format_exc()
                    url = OAI_URL % (host, rec_id)
577 578 579 580 581
                    logs.append(Msg(harvester=self.harvester,
                                    collection=title,
                                    record_id=rec_id,
                                    title=url))
                    logs[-1].reject(e)
582 583

    def decode_xml(self, xml):
584
        """Decode the MARC XML string and insert records in the database.
585

586 587
        Args:
            xml (unicode): MARC XML string
588 589

        """
590

591 592 593
        if self.dbg:
            print "process xml record"

594
        # alias
595
        db = self.db
596 597
        check_record = self.check_record
        insert_record = self.insert_record
598
        is_record_in_db = self._is_record_in_db
599 600
        logs = self.logs

601 602 603
        # NOTE
        # BaseException and inherited class
        # are caught by the previous stage
604 605 606 607 608 609 610
        records = self.marc12(xml)

        # process individual record
        for record in records:

            if self.dbg:
                print "record decoded"
611
                print record.title()
612

613 614
            # reject the record using the secondary OAI
            # require to cover the case:
615
            # - scan store A. Only OAI_A is defined.
616
            # - later scan store B. OAI_B and OAI_A are found.
617 618 619 620 621 622 623 624
            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)
625 626 627 628

                if self.dbg:
                    print "record in db (secondary oai) ->", rec_id

629 630
                continue

631
            # start the log for the record
632 633 634 635
            logs.append(Msg(harvester=self.harvester,
                            collection=self.collection_logs[-1].title,
                            record_id=record.id(),
                            title=record.title()))
636

637 638
            # check that the record is well formed
            # repair non-conformity as far as possible
639
            if not check_record(record):
640 641
                if self.dbg:
                    print "record rejected", logs[-1].txt
642 643 644
                continue

            if self.dbg:
645
                print "insert record in the database"
646

647
            # insert the record in the database
648
            insert_record(record)
649 650

            if self.dbg:
651
                print logs[-1].action.upper(), logs[-1].txt
652 653 654 655

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

656 657 658 659 660 661
        Returns:
            dict:
                * ``collection_logs`` list of :class:MsgCollection
                * ``controller`` unicode
                * ``logs`` list of :class:Msg
                * ``selector`` :py:class:plugin_dbui.Selector
662 663 664 665 666 667

        """

        return dict(collection_logs=self.collection_logs,
                    controller=self.controller,
                    logs=self.logs)
668 669

    def search_collaboration(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
670
        """Get the database collaboration identifier using synonyms.
671 672 673 674 675 676

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
677
                * the id of the collaboration record.
678 679
                * UNDEF_ID if value is not defined.

680
        Raises:
681 682
            ToolException: when more than one synonym is found or when the
                collaboration is not defined.
683 684

        """
685

686
        return search_synonym(self.db.collaborations, "collaboration", value)
687 688

    def search_country(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
689
        """Get the database country identifier using synonyms.
690 691 692 693 694 695

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
696
                * the id of the country record.
697 698
                * UNDEF_ID if value is not defined.

699
        Raises:
700 701
            ToolException: when more than one synonym is found ot when
                the country is not defined.
702 703

        """
704
        return search_synonym(self.db.countries, "country", value)
705 706

    def search_publisher(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
707
        """Get the database publisher identifier using synonyms.
708 709 710 711 712 713

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
714
                * the id of the publisher record.
715 716
                * UNDEF_ID if value is not defined.

717
        Raises:
718 719
            ToolException: when more than one synonym is found or when
                the publisher is not defined.
720 721

        """
722
        return search_synonym(self.db.publishers, "abbreviation", value)