checkandfix.py 29.5 KB
Newer Older
1
# -*- coding: utf-8 -*-
2
""" harvest_tools.checkandfix
3 4 5 6 7

"""
import re
import regex

8
from base import search_synonym
9
from exception import CheckException
10 11 12 13 14 15 16 17 18
from invenio_tools import (load_record,
                           MSG_NO_CONF,
                           MSG_NO_THESIS,
                           OAI_URL,
                           RecordConf,
                           RecordThesis,
                           REG_AUTHOR,
                           REG_OAI,
                           REG_YEAR)
19
from gluon import current
20
from plugin_dbui import CLEAN_SPACES, get_id
21 22


23
DECODE_ARXIV = re.compile(r"arXiv:(\d{2})(\d{2})\.")
24 25 26 27

# Decode submitted date: DD MMM YYYY or DD MM YYY
DECODE_DD_MMM_YYYY = re.compile(r"(\d{1,2}) *([A-Za-z]{3}) *(\d{4})")
DECODE_DD_MM_YYYY = re.compile(r"(\d{1,2}) +(\d{1,2}) +(\d{4})")
28
DECODE_YYYY = re.compile(r"^(\d{4})$")
29 30 31 32

# Decode publication reference:
#  Phys. Rev. Lett. 113, 032001 (2014)
#  Eur. Phys. J. C (2014) 74:2883
33 34 35
_ref1 = r"(?P<p>[A-Za-z\. ]+) +(?P<v>\d+),? +(?P<c>[\d-]+) +\((?P<y>[\d]+)\)"
_ref2 = r"(?P<p>[A-Za-z\. ]+) +\((?P<y>\d+)\) +(?P<v>[\d]+):(?P<c>[\d-]+)"
DECODE_REF = [re.compile(_ref1), re.compile(_ref2)]
36

LE GAC Renaud's avatar
LE GAC Renaud committed
37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
MONTHS = {u'Jan': '01',
          u'Feb': '02',
          u'Fev': '02',
          u'Mar': '03',
          u'Apr': '04',
          u'Avr': '04',
          u'May': '05',
          u'Mai': '05',
          u'Jun': '06',
          u'Jul': '07',
          u'Aug': '08',
          u'Sep': '09',
          u'Oct': '10',
          u'Nov': '11',
          u'Dec': '12'}
52

53 54
MSG_INVALID_HOST = "Invalid host"

55
MSG_NO_AUTHOR = "Reject no author(s)"
56
MSG_NO_CONF_DATE = "Reject no conference date"
57
MSG_NO_DATE = "Reject no submission date"
58
MSG_NO_MY_AUTHOR = "Reject no authors of my institute"
59 60
MSG_NO_REF = "Reject incomplete paper reference"
MSG_NO_YEAR = "Reject no publication year"
61

62
MSG_TEMPORARY_RECORD = "Temporary record"
63

64 65 66
MSG_TO_MANY_DATE = "Reject to many submit date"
MSG_TO_MANY_FAUTHOR = "Reject to many first author"
MSG_TO_MANY_YEAR = "Reject to many year"
67

68 69
MSG_WELL_FORMED_CONF_DATES = "Reject conference dates is not well formed"
MSG_WELL_FORMED_DATE = "Reject submission date is not well formed"
70

71
MSG_WELL_FORMED_EDITOR = "Reject editor is not well formed"
72

73 74
OAI_INVENIO = "oai:%s:%s"

75 76
REG_COLLABORATION = re.compile(regex.REG_COLLABORATION)
REG_CONF_DATES_1 = re.compile("(\d+) *-? *(\d+) *([A-Z][a-z]{2}) *(\d{4})")
77 78

REG_CONF_DATES_2 = \
LE GAC Renaud's avatar
LE GAC Renaud committed
79
    re.compile("(\d+) *([A-Z][a-z]{2}) *-? *(\d+) *([A-Z][a-z]{2}) *(\d{4})")
80

81 82 83
REG_CONF_DATES = re.compile(regex.REG_CONF_DATES)
REG_SUBMITTED = re.compile(regex.REG_SUBMITTED)

84 85
UNIVERSITY = "University"

86 87

class CheckAndFix(object):
88 89
    """A collection of tools to check and repair the content
    of the Marc12 record.
90

91 92
    """
    def __init__(self):
93

94 95 96 97 98 99 100 101 102 103 104 105
        self.db = current.globalenv['db']
        self.reg_institute = self._get_reg_institute()

        # private cache for my_author rescue list
        self.__par = None
        self.__reference = None

        # private cache for my authors list
        self.__my_authors = {}

    def _get_reg_institute(self):
        """
106 107
        Returns:
            unicode: the regular expression defining my institute
108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130

        """
        # alias
        app = current.app
        reg_institute = app.reg_institute

        # regular expression for the institute is not defined
        # find it using the institute definition in inspirehep
        # store the regular expression in current.app for a later use
        if not reg_institute:
            institute_id = app.inspirehep_institute_id
            institute = load_record("inspirehep.net", institute_id)

            reg_institute = institute.rex()

            app.institute = institute
            app.reg_institute = reg_institute

        return reg_institute

    def _get_author_rescue_list(self, record, id_project, id_team):
        """Get the rescue list for my authors.

131 132 133 134
        Args:
            record (RecordPubli): record describing a publication.
            id_project (int): identifier of the project in the database.
            id_team (int): identifier of the team in the database.
135

136 137
        Returns:
            list: empty when not defined
138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180

        """
        year = record.year()

        # try to recover year when not defined
        if not year:
            # published article, proceeding
            if "773" in record and "y" in record["773"]:
                year = record["773"]["y"]

            # start date of a conference
            elif "111" in record and "x" in record["111"]:
                year = record["111"]["x"]

            # end date of a conference
            elif "111" in record and "z" in record["111"]:
                year = record["111"]["z"]

            # submitted date
            elif "269" in record and "c" in record["269"]:
                year = record["269"]["c"]

            else:
                return []

        #
        # NOTE
        # keep in mind that the CheckAndfix mechanism is not yet run
        # therefore year can be a list due to erratum, ...
        #
        if isinstance(year, list):
            year.sort()
            year = year[0]

        # the value can have several format 1992, 1992-12-31, ....
        m = REG_YEAR.search(year)
        if m:
            year = m.group(1)

        else:
            return []

        # caching
LE GAC Renaud's avatar
LE GAC Renaud committed
181
        t = (year, id_project, id_team)
182 183 184 185 186
        if t == self.__par:
            return self.__reference

        # extract the list from the database
        row = self.db.my_authors(year=year,
LE GAC Renaud's avatar
LE GAC Renaud committed
187 188
                                 id_projects=id_project,
                                 id_teams=id_team)
189 190 191 192 193 194 195

        if row:
            self.__reference = row['authors'].split(', ')
        else:
            self.__reference = []

        return self.__reference
196

197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216
    def _is_synonym(self, tablename, value):
        """Check that the synonym field contains *value*.

        Args:
            tablename (unicode): name of the database table
            value (unicode): value to be searched

        Returns:
            bool: ``True`` if *one* row is found, ``False`` otherwise.

        """
        db = self.db
        table = db[tablename]

        query = table.synonyms.contains(value)
        if db(query).count() == 1:
            return True

        return False

217 218 219 220
    def _recover_submitted(self, record):
        """Recover submitted date using conference, preprint or thesis
        information.

221 222
        Args:
            record (RecordPubli): record describing a publication.
223

224 225
        Returns:
            unicode: empty when procedure failed
226 227 228

        """
        val = u''
229
        if isinstance(record, RecordConf):
230 231

            # INSPIREHEP start date encoded as 2014-12-31
232 233
            if "x" in record[u"111"]:
                val = record[u"111"]["x"]
234 235

            # CDS end date encoded as 20141231
236 237
            elif "z" in record[u"111"]:
                val = record[u"111"]["z"]
238 239
                val = "%s-%s-%s" % (val[0:4], val[4:6], val[6:8])

240
        elif isinstance(record, RecordThesis):
241 242 243 244 245 246 247 248 249 250 251 252 253
            val = record.these_defense()

        else:
            report = record.preprint_number()
            if report:
                m_arxiv = DECODE_ARXIV.match(report)
                if m_arxiv:
                    val = "20%s-%s" % (m_arxiv.group(1), m_arxiv.group(2))

        return val

    def _repair_paper_reference(self, record):
        """Repair paper reference.
254
        The recovery procedure use the "o" field (invenio)::
255

256 257
            Eur. Phys. J. C (2014) 74:2883
            Phys. Rev. Lett. 113, 032001 (2014)
258 259 260

        in order to extract editor, volume, year and page data.

261 262
        Args:
            record (RecordPubli): record describing a publication.
263

264 265
        Raises:
            CheckException: when the repair failed.
266 267 268

        """
        # standard case
269
        if isinstance(record[u"773"], dict):
270

271
            if "o" in record[u"773"]:
272
                for reg in DECODE_REF:
273
                    m = reg.match(record[u"773"]["o"])
274
                    if m:
275 276 277 278
                        record[u"773"]["p"] = m.group("p")
                        record[u"773"]["v"] = m.group("v")
                        record[u"773"]["y"] = m.group("y")
                        record[u"773"]["c"] = m.group("c")
279 280 281 282 283
                        return

            raise CheckException(MSG_NO_REF)

        # list case -- paper with erratum
284
        elif isinstance(record[u"773"], list):
285

286
            for i in range(len(record[u"773"])):
287

288
                if "o" in record[u"773"][i]:
289 290
                    fixed = False
                    for reg in DECODE_REF:
291
                        m = reg.match(record[u"773"][i]["o"])
292
                        if m:
293 294 295 296
                            record[u"773"][i]["p"] = m.group("p")
                            record[u"773"][i]["v"] = m.group("v")
                            record[u"773"][i]["y"] = m.group("y")
                            record[u"773"][i]["c"] = m.group("c")
297 298 299 300 301 302 303 304 305 306 307 308 309
                            fixed = True
                            break

                    if not fixed:
                        raise CheckException(MSG_NO_REF)
                else:
                    raise CheckException(MSG_NO_REF)

        # case not expected
        else:
            raise CheckException(MSG_NO_REF)

    def authors(self, record):
310
        """Check that author fields are defined.
311

312 313
        Args:
            record (RecordPubli): record describing a publication.
314

315 316 317
        Raises:
            CheckException: when there is no authors or more than
                one *first author*.
318 319 320

        """

321
        if not record.is_authors():
322 323
            raise CheckException(MSG_NO_AUTHOR)

324 325
        if isinstance(record.first_author(), list):
            raise CheckException(MSG_TO_MANY_FAUTHOR)
326 327 328 329

    def clean_erratum(self, record):
        """Clean record with erratum by removing them.

330 331 332
        Note:
            After this check the editor, volume, page and
            paper year field contains one entry.
333

334 335
        Args:
            record (RecordPubli): record describing a publication.
336 337 338 339 340 341 342

        """
        if not record.is_with_erratum():
            return

        # use the simplest algorithm by selecting the first entry in the list
        # fare to assume that the article is published first.
343
        record[u"773"] = record[u"773"][0]
344 345

        # treat year and submitted date
346
        for k in (u"260", u"269"):
347 348 349 350
            if k in record and isinstance(record[k], list):
                record[k] = record[k][0]

    def collaboration(self, record):
351 352
        """Check the collaboration.
        Have a look to the synonyms when the collaboration is not well formed.
353

354 355
        Args:
            record (RecordPubli): record describing a publication.
356

357
        Raises:
358 359
            ToolException: when the collaboration value is defined
                nor entered as a synonym.
360 361 362

        """
        val = record.collaboration()
363 364 365
        if not val:
            return

366 367
        db = self.db
        search_synonym(db.collaborations, "collaboration", val)
368

369
    def country(self, record):
370
        """Check conference country.
371
        Have a look to the synonyms when the country does not exist.
372

373 374
        Args:
            record (RecordConf): record describing a talk or a proceeding.
375

376
        Raises:
377 378
            ToolException: when the country is not defined
                nor entered as a synonym.
379 380

        """
381
        if not isinstance(record, RecordConf):
382 383
            return

384
        db = self.db
385
        val = record.conference_country()
386
        search_synonym(db.countries, "country", val)
387

388
    def conference_date(self, record):
389
        """Check conference date.
390

391 392
        Args:
            record (RecordConf): record describing a talk or a proceeding.
393

394 395
        Raises:
            CheckException: when dates are not found or not well formed.
396 397 398 399 400 401

        """
        # conference information are available, i.e proceeding
        if not isinstance(record, RecordConf):
            return

402
        # check and fix conference date
403 404 405
        if not (u"111" in record and "d" in record[u"111"]):
            raise CheckException(MSG_NO_CONF_DATE)

406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422
        value = record[u"111"]["d"]
        m = REG_CONF_DATES.match(value)
        if not m:

            # 12 - 15 Mar 2014 or 29 Feb - 1 Mar 2014
            m1 = REG_CONF_DATES_1.match(value)
            m2 = REG_CONF_DATES_2.match(value)

            if m1:
                record[u"111"]["d"] = "%s-%s %s %s" % m1.groups()

            elif m2:
                record[u"111"]["d"] = "%s %s - %s %s %s" % m2.groups()

            else:
                raise CheckException(MSG_WELL_FORMED_CONF_DATES)

423
    def is_bad_oai_used(self, record):
424 425 426
        """Bad OAI is when the ``id`` in the OAI field is different from
        the ``record id``. This happens when an old record is redirected
        to new one.
427

428 429
        Args:
            record (RecordPubli): record describing a publication.
430

431 432 433
        Returns:
            bool: ``True`` when a record is found in the database with
                the bad OAI.
434

435 436 437 438
        """
        value = record.oai()
        match = REG_OAI.match(value)

439
        if match.group(2) != record.id():
440 441
            db = self.db

442
            # a record with the bad OAI exists in the database
443 444 445
            bad_oai_url = OAI_URL % (match.group(1), match.group(2))
            if get_id(db.publications, origin=bad_oai_url):
                return True
446

447
        return False
448

449 450 451
    def format_authors(self, record, func):
        """Format the author names using the function func.

452 453 454
        Args:
            record (RecordPubli): record describing a publication.
            func (reference): function used to format the author names.
455 456

        """
457
        for key in (u"100", u"700"):
458 459
            if key in record:
                if isinstance(record[key], list):
460
                    for i in xrange(len(record[key])):
461
                        if "a" in record[key][i]:
462 463 464 465 466 467 468 469 470 471 472

                            # PROTECTION
                            # see RecordPubli.author_as_list
                            value = record[key][i]["a"]
                            if isinstance(value, unicode):
                                record[key][i]["a"] = func(value)

                            elif isinstance(value, list):
                                for elt in value:
                                    if REG_AUTHOR.match(elt):
                                        record[key][i]["a"] = func(elt)
473 474

                else:
475
                    if "a" in record[key]:
476 477 478 479 480 481 482 483 484 485 486
                        value = record[key]["a"]

                        # PROTECTION
                        # see RecordPubli.authors_as_list
                        if isinstance(value, unicode):
                            record[key]["a"] = func(value)

                        elif isinstance(value, list):
                            for elt in value:
                                if REG_AUTHOR.match(elt):
                                    record[key][i]["a"] = func(elt)
487 488

    def format_editor(self, record):
489 490
        """Format the editor abbreviation. The encoding
        depends on the store::
491

492 493
            INVENIO:    Phys. Lett. B + volume 673
            INSPIREHEP: Phys.Lett + volume B673
494

495
        Standardise the answer as ``Phys. Lett. B``.
496

497 498
        Note:
            It is recommended to call this method when erratum are removed.
499

500 501
        Args:
            record (RecordPubli): record describing a publication.
502

503 504
        Raises:
            CheckException: when the editor is not well formed.
505 506 507 508 509 510

        """
        if not record.is_published():
            return

        # standard case
511 512
        if isinstance(record[u"773"], dict):
            if "p" in record[u"773"] and "v" in record[u"773"]:
513

514 515
                editor = record[u"773"]["p"]
                volume = record[u"773"]["v"]
516 517 518 519 520 521 522 523 524 525 526

                # add space after the dot  Phys.Rev -> Phys. Rev
                editor = re.sub(r'\.([A-Z])', r'. \1', editor)

                # get the volume letter
                m = re.match(r'([A-Z]+) *(\d+)', volume)
                if m and m.group(1) != editor[-1]:
                    editor = "%s %s" % (editor, m.group(1))
                    volume = m.group(2)

                # remove stupid mistake
527
                editor = CLEAN_SPACES(editor)
528

529 530
                record[u"773"]["p"] = editor
                record[u"773"]["v"] = volume
531 532

        # list case -- publication with erratum
533
        elif isinstance(record[u"773"], list):
534

535 536
            editors = record._get(u"773", 'p', force_list=True)
            volumes = record._get(u"773", 'v', force_list=True)
537 538 539 540 541 542 543 544 545 546 547 548 549 550

            if len(editors) != len(volumes):
                raise CheckException(MSG_WELL_FORMED_EDITOR)

            for i in range(len(editors)):
                # add space after the dot  Phys.Rev -> Phys. Rev
                editor = re.sub(r'\.([A-Z])', r'. \1', editors[i])

                # get the volume letter
                m = re.match(r'([A-Z]+) *(\d+)', volumes[i])
                if m and m.group(1) != editor[-1]:
                    editor = "%s %s" % (editor, m.group(1))
                    volume = m.group(2)

551
                editor = CLEAN_SPACES(editor)
552

553 554
                record[u"773"][i]["p"] = editor
                record[u"773"][i]["v"] = volume
555

556 557 558
    def format_universities(self, record):
        """Format the name of the university for PhD:

559 560
            * Fix the name of Aix-Marseille University
            * Replace U. by University
561

562 563
        Args:
            record (RecordThesis): record describing a thesis.
564 565 566

        """
        # protection
567
        if not isinstance(record, RecordThesis):
568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606
            return

        # CPPM: fix the name of Aix-Marseille university
        if current.app.inspirehep_institute_id == 902989:

            year = REG_YEAR.search(record.these_defense()).group(1)
            if int(year) < 2012:
                university = u"Université de la Méditerrannée Aix-Marseille II"
            else:
                university = u"Aix Marseille Université"

            if u'502' in record and "b" in record[u'502']:
                if isinstance(record[u'502']['b'], unicode):
                    if "Marseille" in record[u'502']['b']:
                        record[u'502']['b'] = university

                elif isinstance(record[u'502']['b'], list):
                    for i in xrange(len(record[u'502']['b'])):
                        if "Marseille" in record[u'502']['b'][i]:
                            record[u'502']['b'][i] = university

        # Other: replace U. by University
        else:
            university = current.T(UNIVERSITY).decode("utf8")

            if u'502' in record and "b" in record[u'502']:
                if isinstance(record[u'502']['b'], unicode):
                    value = record[u'502']['b']
                    if "U." in value:
                        value = value.replace('U.', university)
                        record[u'502']['b'] = value

                elif isinstance(record[u'502']['b'], list):
                    for i in xrange(len(record[u'502']['b'])):
                        value = record[u'502']['b'][i]
                        if "U." in value:
                            value = value.replace('U.', university)
                            record[u'502']['b'][i] = value

607 608
    def get_my_authors(self, record, cmpFct=None):
        """Get authors of my institutes signing the record.
609 610
        The information is append to the Record object via the attribute
        ``my_authors``.
611

612 613 614 615 616
        Args:
            record (RecordPubli): record describing a publication.
            cmpFct (reference): extract the family name from the full name.
                It is used to sort my author list according to the
                author family name.
617

618 619
        Returns:
            unicode: the list of authors separated by comma
620

621 622
        Raises:
            CheckException: when the list is empty
623 624 625 626

        """
        # might have been computed when affiliation is checked
        rec_id = record.id()
LE GAC Renaud's avatar
LE GAC Renaud committed
627
        if rec_id in self.__my_authors:
628 629 630 631 632 633 634 635 636 637 638 639
            li = self.__my_authors[rec_id]
            li.sort(key=cmpFct)
            value = u', '.join(li)

        # find authors of my institute signing the record
        else:
            reg_institute = self.reg_institute
            value = record.find_authors_by_affiliation(reg_institute, cmpFct)

        if not value:
            raise CheckException(MSG_NO_MY_AUTHOR)

LE GAC Renaud's avatar
LE GAC Renaud committed
640
        record.my_authors = value
641

642 643 644
    def is_conference(self, record):
        """Check that the record described a conference talk / proceeding.

645 646
        Args:
            record (RecordPubli): record describing a publication.
647

648 649
        Raises:
            CheckException: when the record is not associated to a conference.
650 651 652 653 654 655

        """
        if not isinstance(record, RecordConf):
            raise CheckException(MSG_NO_CONF)

    def is_thesis(self, record):
656
        """Check that the record described a thesis.
657

658 659
        Args:
            record (RecordPubli): record describing a publication.
660

661 662
        Raises:
            CheckException: when the record does not describe a thesis.
663 664 665 666 667

        """
        if not isinstance(record, RecordThesis):
            raise CheckException(MSG_NO_THESIS)

668 669 670 671 672
    def my_affiliation(self, record, id_project, id_team):
        """Check that authors of my institute are signatories.

        Launch a recovery procedure when affiliations are not defined.
        It is based on the author rescue list stored in the database.
673

674 675 676 677
        Args:
            record (RecordPubli): record describing a publication.
            id_project (int): identifier of the project in the database
            id_team (int): identifier of the team in the database
678

679 680
        Raises:
            CheckException: when there is no authors from my institute.
681 682

        """
683 684 685
        value = record.find_affiliation(self.reg_institute)
        if value:
            return value
686

687 688 689 690 691
        # affiliation is not defined
        # try to recover using the authors rescue list
        rescue_list = self._get_author_rescue_list(record, id_project, id_team)
        if not rescue_list:
            raise CheckException(MSG_NO_MY_AUTHOR)
692

693 694 695
        # compute the intersection between the authors and the rescue list
        set_1 = set(record.authors_as_list())
        set_2 = set(rescue_list)
696

697 698 699
        li = list(set_1.intersection(set_2))
        if not li:
            raise CheckException(MSG_NO_MY_AUTHOR)
700

701 702
        # cache the result for a latter use
        self.__my_authors[record.id()] = li
703 704 705 706 707

    def paper_reference(self, record):
        """Check that editor, page, volume and paper year are defined
        for a published paper. Repair it as far as possible.

708 709 710
        Note:
            It is recommended to call this method when the
            erratum are removed.
711

712 713
        Args:
            record (RecordPubli): record describing a publication.
714

715 716
        Raises:
            CheckException: when the paper reference is not well formed.
717 718 719 720 721 722

        """
        if not record.is_published():
            return

        # list of reference (paper with erratum)
723 724
        refs = record[u"773"]
        if not isinstance(record[u"773"], list):
725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752
            refs = [refs]

        # INSPIREHEP
        # fix the following case:
        #    {'p': u'Phys.Lett.B'},
        #    {'y': u'2013', 'p': u'Phys.Lett.B', 'c': u'6-12', 'v': u'B720'}

        if len(refs) == 2 and record.host().startswith('inspirehep'):
            nkeys = [len(di) for di in refs]
            if nkeys.count(1):
                index = nkeys.index(1)
                if 'p' in refs[index]:
                    refs.pop(index)

        # do we have all fields defining a reference
        # editor, volume, year and page
        for ref in refs:

            # proceeding can only contains "w" and "y" keys
            if "w" in ref:
                continue

            # check full reference
            for k in ("p", "v", "y", "c"):
                if k not in ref:
                    self._repair_paper_reference(record)
                    return

753 754 755 756 757 758 759 760
    def publisher(self, record):
        """Check publisher.
        Have a look to the synonyms when the publisher does not exist.

        Args:
            record (RecordPubli): record describing a publication.

        Raises:
761 762
            ToolException: when the publisher is not defined
                nor entered as a synonym.
763 764 765 766 767 768 769 770

        """
        db = self.db
        val = record.paper_editor()
        if not val:
            return

        # erratum -- check the first entry
771
        if isinstance(val, list):
772 773
            val = val[0]

774
        search_synonym(db.publishers, "abbreviation", val)
775

776 777 778 779
    def recover_oai(self, record, host):
        """Recover the OAI identifier when it is not defined
        or not well form.

780 781 782 783
        Args:
            record (RecordPubli): record describing a publication.
            host (unicode): possible values ares ``cds.cern.ch``
                or ``inspirehep.net``
784 785

        """
LE GAC Renaud's avatar
LE GAC Renaud committed
786 787 788 789 790 791 792
        # Note:
        # For the record cds 1951625, possible values are:
        # oai:cds.cern.ch:1951625 (if it does not exist in inspirehep)
        # oai:cds.cern.ch:1951625, oai:inspirehep.net:1319638 (if it exist
        # in both store)
        # In all the case the first OAI corresponds to the record.id()
        #
793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810
        oai = record.oai()
        if oai is not None and REG_OAI.match(oai):
            return

        if host == "cds.cern.ch":
            field, subfield = u"0248", "a"

        elif host == "inspirehep.net":
            field, subfield = u"909CO", "o"

        else:
            raise ValueError(MSG_INVALID_HOST)

        if field not in record:
            record[field] = dict()

        record[field][subfield] = OAI_INVENIO % (host, record.id())

811
    def submitted(self, record):
812
        """Standardise the submitted date as ``YYYY-MM`` or ``YYYY-MM-DD``.
813 814
        Look for alternative when it is not defined.

815 816
        Note:
            After this check the year submitted contains one entry.
817

818 819
        Args:
            record (RecordPubli): record describing a publication.
820

821 822 823
        Raises:
            CheckException: when the date is not well formed or when more
                than one date are found.
824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840

        """
        dates = record.submitted()

        # recover missing date using conference, preprint, thesis information
        if not dates:
            val = self._recover_submitted(record)
            if not val:
                raise CheckException(MSG_NO_DATE)
            dates.append(val)

        # check that date are well formed
        for i in range(len(dates)):

            # 22 Mar 2011
            m = DECODE_DD_MMM_YYYY.match(dates[i])
            if m:
841 842
                data = (m.group(3), MONTHS[m.group(2)], int(m.group(1)))
                dates[i] = '%s-%s-%02i' % data
843 844 845 846 847
                continue

            # 22 03 2011
            m = DECODE_DD_MM_YYYY.match(dates[i])
            if m:
LE GAC Renaud's avatar
LE GAC Renaud committed
848
                data = (m.group(3), int(m.group(2)), int(m.group(1)))
849
                dates[i] = '%s-%02i-%02i' % data
850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868
                continue

            # 2011
            m_year = DECODE_YYYY.match(dates[i])
            if m_year:
                dates[i] = self._recover_submitted(record)

            # check the minimum requirement is 2001-05
            if not REG_SUBMITTED.match(dates[i]):
                raise CheckException(MSG_WELL_FORMED_DATE)

        # protection against list of identical date
        if len(dates) > 1:
            dates = list(set(dates))

        # Only one date
        if len(dates) != 1:
            raise CheckException(MSG_TO_MANY_DATE)

869
        if u"269" not in record or isinstance(record[u"269"], list):
870 871 872 873 874 875 876
            record[u"269"] = dict()

        record[u"269"]["c"] = dates[0]

    def temporary_record(self, record):
        """Some records are marked temporary.

877 878
        Args:
            record (RecordPubli): record describing a publication.
879

880 881
        Raises:
            CheckException: when the record is marked temporary
882 883 884

        """
        # found on INSPIREHEP (see record 1317573)
885 886
        if u"500" in record and "a" in record[u"500"]:
            if record[u"500"]["a"] == u"*Temporary record*":
887 888 889
                raise CheckException(MSG_TEMPORARY_RECORD)

    def year(self, record):
890
        """Standardise the the year as ``YYYY`` and look for alternative when
891 892
        it is not defined.

893 894
        Note:
            The method assumes that the erratum are removed.
895

896 897 898 899
        Note:
            It is recommended to call this method after
            the ``submitted`` one, since the submitted field is used in
            the recovery procedure.
900

901 902
        Note:
            After this check the year field contains one entry.
903

904 905
        Args:
            record (RecordPubli): record describing a publication.
906

907 908
        Raises:
            CheckException: when there is a list of year or when empty.
909 910 911 912 913

        """
        val = ''

        # protection against "publication date YYYY-MM-DD"
914
        li = record._get(u"260", 'c', force_list=True)
915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940
        li = [el for el in li if REG_YEAR.match(el)]

        if len(li) == 1:
            val = li[0]
        elif len(li) > 1:
            raise CheckException(MSG_TO_MANY_YEAR)

        # alternative: look to the year in the publication reference
        if not val:
            val = record.paper_year()

        # alternative: look to the submitted date
        if not val:
            val = record.submitted()

        # protection against remaining list
        if isinstance(val, list):
            if len(val) == 1:
                val = val[0]
            else:
                raise CheckException(MSG_TO_MANY_YEAR)

        # the value can have several format 1992, 1992-12-31, .... standardize
        if val:
            m = REG_YEAR.search(val)
            if m:
941 942
                if u"260" in record and isinstance(record[u"260"], dict):
                    record[u"260"]["c"] = m.group(1)
943
                else:
944
                    record[u"260"] = {"c": m.group(1)}
945 946 947
                return

        raise CheckException(MSG_NO_YEAR)