automaton.py 21.7 KB
Newer Older
1
# -*- coding: utf-8 -*-
2
""" harvest_tools.automaton
3 4 5 6 7

"""
import re
import traceback

8

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

19

20 21 22
MSG_NO_CAT = 'Select a "category" !!!'
MSG_NO_PROJECT = 'Select a "project" !!!'
MSG_NO_TEAM = 'Select a "team" !!!'
23

24
MSG_TOOMANY_SYNONYM = "Reject too many %s synonyms."
25
MSG_NSERT_FAIL = "Fail to insert the new record in the database."
26
MSG_NO_OAI = "Reject no OAI identifier"
27
MSG_WELL_FORM_OAI = "Reject OAI is not well formed"
28

29

30
class Automaton(object):
31
    """Base class to search and process publications:
32

33 34 35 36 37
        * 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.
38

39 40
    Note:
        The parameters of the search are defined by the current ``request``.
41

42 43 44 45 46
    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
47
           and decode it.
48 49 50
        #. 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`.
51 52 53
           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.
54 55
        #. Reject temporarily publication.
        #. Check that *authors* are defined.
56
           Reject the publication if it is not the case.
57
        #. Check that *my institute* is in the list of the institutes
58 59 60 61 62 63
           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.
64
        #. Check that the *collaboration*, if defined, is well formed.
65
           Reject the publication if it is not the case
66 67 68 69 70 71 72 73
        #. 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.
74
        automaton (unicode): the name of the automaton which
75 76 77
            will be used to process the data. Possible values are:
            ``articles``, ``notes``, ``preprints``, ``proceedings``,
            ``reports``, ``talks`` and ``theses``.
78 79 80 81 82 83 84 85 86 87
        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.
88

89 90 91 92 93
    """
    def __init__(self,
                 db,
                 id_team,
                 id_project,
94
                 automaton,
95 96 97 98 99 100 101 102 103 104
                 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
105
        self.controller = automaton
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
        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)

133 134 135
    def _insert_in_db(self, log_year="", **fields):
        """Insert the record in the database, handling database exception.

136 137
        Args:
            log_year (unicode): year of the record for the log
138

139 140 141
        Keyword Args:
            **fields: keyword arguments defining the record values to be
                inserted in the database.
142

143 144 145
        Returns:
            int: one when the record is inserted / updated in the database,
                zero otherwise.
146 147 148 149 150

        """
        db = self.db

        try:
LE GAC Renaud's avatar
LE GAC Renaud committed
151
            rec_id = db.publications.insert(**fields)
LE GAC Renaud's avatar
LE GAC Renaud committed
152 153
            if rec_id:
                return 1
154

LE GAC Renaud's avatar
LE GAC Renaud committed
155 156 157 158 159
            # 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
160

LE GAC Renaud's avatar
LE GAC Renaud committed
161 162 163
                # reduce the error message
                if isinstance(msg, list):
                    msg = "%s %s" % (msg[0], msg[-1])
164

LE GAC Renaud's avatar
LE GAC Renaud committed
165 166
                self.logs[-1].reject(msg, log_year)
                return 0
167

LE GAC Renaud's avatar
LE GAC Renaud committed
168 169 170 171
        # operation can be rejected by the database
        except Exception as dbe:
            self.logs[-1].reject(dbe.message, log_year)
            return 0
172

173
    def _is_record_in_db(self, rec_id, title):
174
        """Return ``True`` when the record is already in the database.
175 176
        The search is based on the origin field.

177 178
        Note:
            A new log entry is created when a record is found.
179

180 181 182
        Args:
            rec_id (int): record identifier
            title (unicode): title of the collection
183

184 185
        Returns:
            bool: ``True`` when a record if found, ``False`` otherwise.
186 187 188

        """
        db = self.db
189
        harvester = self.harvester
190

191 192
        # check
        url = OAI_URL % (harvester.host, rec_id)
193 194 195 196 197 198 199
        db_id = get_id(db.publications, origin=url)

        if db_id is None:
            return False

        publication = db.publications[db_id]

200 201 202 203 204 205 206
        # same category for the publication and the harvester
        # keep the record if it is not the case
        # this is required to transform a preprint into article
        if publication.id_categories != harvester.id_categories:
            return False

        # log
207 208 209 210 211 212 213 214 215
        self.logs.append(Msg(harvester=self.harvester,
                             collection=title,
                             record_id=rec_id,
                             title=publication.title))

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

        return True

216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243
    def _search(self, tablename, fieldname, value, create=False):
        """Get the database identifier for the record having the database field
        matching the value. The database field is defined by the argument
        *tablename* and *fieldname*.

        Note:
            The search is also perform using the synonym field
            when nothing is found.

        Args:
            tablename (unicode): name of the database table.
            fieldname (unicode): name of the database field.
            value (unicode): the search value.
            create(bool): create a new entry in the database table when
                it is ``True``

        Returns:
            int:
                * the id of the database record.
                * UNDEF_ID if value is not defined.

        Raise:
            ToolException: when more than one synonym is found.

        """
        if not value:
            return UNDEF_ID

244 245 246 247 248 249 250
        db = self.db
        table = db[tablename]

        kwargs = {}
        kwargs[fieldname] = value

        id_rec = get_id(table, **kwargs)
251 252 253 254 255 256 257 258 259 260
        if id_rec is not None:
            return id_rec

        # nothing found, have a look to the synonyms
        query = table.synonyms.contains(value)
        setrows = db(query)

        # no synonym found, create the entry
        ncount = setrows.count()
        if ncount == 0 and create:
261
            return table.insert(**kwargs)
262 263 264 265 266 267 268 269 270 271

        # one synonym found
        elif ncount == 1:
            return setrows.select(table.id).first().id

        # more than one synonyms - don't know how to choose
        else:
            msg = MSG_TOOMANY_SYNONYM % tablename
            raise ToolException(msg)

272 273 274 275 276
    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.

277 278 279 280 281 282
        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"``.
283

284 285 286
        Returns:
            dict: the key are a sub-set of those defined in
                :py:meth:invenio_tools.InvenioStore.get_ids.
287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332

        """

        # 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
333
    def check_record(self, record):
334 335
        """Check the content of the record in order to fix non-conformities.
        Return ``False`` when non-conformities are found and can not be
336 337
        corrected.

338 339 340
        Note:
            Some checks depend on the type of publications and have to be
            implemented in inherited class.
341

342 343 344 345
        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.
346

347 348
        Args:
            record (Record): Marc12 record describing the publication.
349

350 351 352
        Returns:
            bool: ``False`` when a non-conformity is found and can not be
                corrected.
353 354 355 356 357 358

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

        try:
359 360
            self.check.recover_oai(record, self.harvester.host)

361 362 363 364
            if self.check.is_bad_oai_used(record):
                self.logs[-1].idle(MSG_IN_DB, record.year())
                return False

365 366
            self.check.temporary_record(record)
            self.check.authors(record)
367
            self.check.my_affiliation(record, self.id_project, self.id_team)
368 369 370 371 372 373 374 375
            self.check.collaboration(record)

        except Exception as e:
            self.logs[-1].reject(e, record.year())
            return False

        return True

376
    def get_record_by_fields(self, **kwargs):
377 378
        """Get database record matching fields values defined
        in the keyword arguments.
379

380 381 382
        Args:
            oai_url (unicode): *e.g* ``"http://cds.cern.ch/record/123456"``
            year (int): the year of the publication.
383

384 385
        Note:
            Fix the field origin when a match is found.
386

387 388
        Note:
            The year is only used by the logger.
389

390 391 392 393 394
        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.
395 396 397

        """
        if self.dbg:
398
            print "get existing record by fields"
399 400 401 402 403 404 405 406 407 408 409 410 411 412

        db = self.db

        # 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
        ok = db.publications[rec_id].origin and \
LE GAC Renaud's avatar
LE GAC Renaud committed
413
            db.publications[rec_id].origin == oai_url
414 415 416 417 418 419 420 421 422 423
        if not ok:
            if not self.dry_run:
                db.publications[rec_id] = dict(origin=oai_url)

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

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

424 425
    def insert_record(self, record):
        """Insert the record in the database.
426

427 428 429
        Note:
            This method depend on the type of publications.
            It has to be implemented for each inherited class.
430

431 432
        Args:
            record (Record): Marc12 record describing the publication.
433

434 435 436
        Returns:
            int: one when the record is inserted / updated in the database,
                zero otherwise.
437 438 439 440 441

        """
        return 0

    def process_xml(self, xml):
442 443 444 445 446 447 448
        """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.
449

450 451
        Raises:
           ToolException: when project, team or category identifier
452
              are not defined.
453
           StoreException: when something goes wrong interrogating the
454
              store.
455 456 457 458
           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.
459 460 461 462 463 464 465 466 467 468 469 470 471 472

        """
        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):
473 474
        """Retrieve the XML string from the invenio store and
        insert corresponding records in the database.
475

476 477 478
        Args:
            host (unicode): host name to query for publications, either
                ``cds.cern.ch`` or ``inspirehep.net``.
479

480
            collections (unicode): list of collection to be interrogated.
481

482 483 484 485 486 487 488
        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.
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510

        """
        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
        controller = self.controller
        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
        #
511
        # The method use here minimize the memory usage
512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546
        # 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)
            self.collection_logs.append(MsgCollection(title=title))

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

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

            except Exception as error:
                self.collection_logs[-1].url = store.last_search_url()
                self.collection_logs[-1].error = error
                continue

            self.collection_logs[-1].url = store.last_search_url()
            self.collection_logs[-1].found = len(rec_ids)

            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:
547
                    if self._is_record_in_db(rec_id, title):
548 549
                        continue

550 551 552
                    xml = store.get_record(rec_id)
                    self.decode_xml(xml)

553
                except Exception as e:
554 555 556 557 558 559 560 561 562
                    print traceback.format_exc()
                    url = OAI_URL % (host, rec_id)
                    self.logs.append(Msg(harvester=self.harvester,
                                         collection=title,
                                         record_id=rec_id,
                                         title=url))
                    self.logs[-1].reject(e)

    def decode_xml(self, xml):
563
        """Decode the MARC XML string and insert records in the database.
564

565 566
        Args:
            xml (unicode): MARC XML string
567 568 569 570 571

        """
        if self.dbg:
            print "process xml record"

572 573 574
        # NOTE
        # BaseException and inherited class
        # are caught by the previous stage
575 576 577 578 579 580 581 582 583 584 585 586 587 588
        records = self.marc12(xml)

        # process individual record
        for record in records:

            if self.dbg:
                print "record decoded"

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

589 590 591
            # check that the record is well formed
            # repair non-conformity as far as possible
            if not self.check_record(record):
592 593 594
                continue

            if self.dbg:
595
                print "insert record in the database"
596

597 598
            # insert the record in the database
            self.insert_record(record)
599 600 601 602 603 604 605

            if self.dbg:
                print self.logs[-1].action.upper(), self.logs[-1].txt

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

606 607 608 609 610 611
        Returns:
            dict:
                * ``collection_logs`` list of :class:MsgCollection
                * ``controller`` unicode
                * ``logs`` list of :class:Msg
                * ``selector`` :py:class:plugin_dbui.Selector
612 613 614 615 616 617

        """

        return dict(collection_logs=self.collection_logs,
                    controller=self.controller,
                    logs=self.logs)
618 619

    def search_collaboration(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
620 621
        """Get the database collaboration identifier using synonyms.
        Create the collaboration, if it does not exist.
622 623 624 625 626 627

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
628
                * the id of the collaboration record.
629 630 631 632 633 634 635 636 637
                * UNDEF_ID if value is not defined.

        Raise:
            ToolException: when more than one synonym is found.

        """
        return self._search("collaborations", "collaboration", value, True)

    def search_country(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
638
        """Get the database country identifier using synonyms.
639 640 641 642 643 644

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
645
                * the id of the country record.
646 647 648 649 650 651 652 653 654
                * UNDEF_ID if value is not defined.

        Raise:
            ToolException: when more than one synonym is found.

        """
        return self._search("countries", "country", value)

    def search_publisher(self, value):
LE GAC Renaud's avatar
LE GAC Renaud committed
655
        """Get the database publisher identifier using synonyms.
656 657 658 659 660 661

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

        Returns:
            int:
LE GAC Renaud's avatar
LE GAC Renaud committed
662
                * the id of the publisher record.
663 664 665 666 667 668 669
                * UNDEF_ID if value is not defined.

        Raise:
            ToolException: when more than one synonym is found.

        """
        return self._search("publishers", "abbreviation", value)