thesis.py 3.84 KB
Newer Older
1 2 3 4 5
""" harvest_tools.thesis

"""
import re

6
from .automaton import Automaton
7
from .base import MSG_CRASH, MSG_LOAD, T4
8
from plugin_dbui import get_id, UNDEF_ID
LE GAC Renaud's avatar
LE GAC Renaud committed
9
from store_tools import CheckException
10

11
MSG_NOT_THESIS = "Reject publication is not a thesis"
12 13


14
class Thesis(Automaton):
15
    """Automaton for thesis.
16 17

    """
18

19 20 21
    def check_record(self, record):
        """Check the content of the thesis in order to fix non conformities.

22
        Args:
LE GAC Renaud's avatar
LE GAC Renaud committed
23 24
            record (RecordThesis):
                record describing a thesis.
25

26
        Returns:
LE GAC Renaud's avatar
LE GAC Renaud committed
27 28 29
            bool:
                ``False`` when a non conformity is found and can not be
                corrected.
30 31

        """
32
        self.logger.debug(f"{T4}check and fix record (thesis)")
33

34
        if record.subtype() == "thesis":
35 36
            self.logs[-1].reject(MSG_NOT_THESIS, record)
            return False
LE GAC Renaud's avatar
LE GAC Renaud committed
37

38
        try:
39 40 41 42 43
            # is with authors form my institute
            # standardise name of collaboration
            # format authors according to my format
            # extract authors form my institute signing the publication
            # is submitted date well formed
44
            record.check_and_fix(db=self.db,
45
                                 fmt_author="F. Last",
46
                                 rex_institute=self.rex_institute,
47 48 49 50
                                 sep_author=", ",
                                 sort_author=True)

            record.format_universities()
51

52
        except CheckException as e:
53
            self.logs[-1].reject(e, record=record)
54 55 56
            return False

        except Exception as e:
57
            self.logs[-1].reject(MSG_CRASH % e, record=record, translate=False)
58 59
            return False

LE GAC Renaud's avatar
LE GAC Renaud committed
60
        return True
61

62 63
    def insert_record(self, record):
        """Insert a thesis in the database.
64

65
        Args:
LE GAC Renaud's avatar
LE GAC Renaud committed
66 67
            record (RecordThesis):
                record describing a thesis.
68

69
        Returns:
LE GAC Renaud's avatar
LE GAC Renaud committed
70 71
            int:
                one when the record is inserted / updated in the database
72
                zero otherwise.
73 74 75 76 77 78 79

        """
        db = self.db

        # alias
        defense_date = record.these_defense()
        first_author = record.first_author()
LE GAC Renaud's avatar
LE GAC Renaud committed
80
        id_category = get_id(db.categories, code="PHD")
81 82
        oai_url = record.oai_url()
        title = record.title()
LE GAC Renaud's avatar
LE GAC Renaud committed
83
        universities = ", ".join(record.these_universities())
84

85
        # extract the year from the defence date
86 87 88
        # this approach seems the most reliable
        year = re.search(r"(\d\d\d\d)", defense_date).group(1)

89
        # get an already published thesis
LE GAC Renaud's avatar
LE GAC Renaud committed
90 91 92 93
        fields = dict(first_author=first_author,
                      defense=defense_date,
                      id_projects=self.id_project,
                      id_teams=self.id_team,
94
                      title=title)
LE GAC Renaud's avatar
LE GAC Renaud committed
95

96
        rec_id, status = self.get_record_by_fields(oai_url, year, **fields)
97 98 99 100
        if rec_id:
            return status

        # eventually insert a new thesis
101
        ret = 1
102
        if not self.dry_run:
LE GAC Renaud's avatar
LE GAC Renaud committed
103
            fields = dict(authors=first_author,
104 105 106 107 108 109 110 111 112 113
                          authors_institute=first_author,
                          defense=defense_date,
                          directors=record.these_directors(),
                          first_author=first_author,
                          id_categories=id_category,
                          id_teams=self.id_team,
                          id_projects=self.id_project,
                          id_status=UNDEF_ID,
                          origin=oai_url,
                          publication_url=record.paper_url(),
114
                          submitted=record.submitted(),
115 116 117
                          title=title,
                          universities=universities,
                          year=year)
118

LE GAC Renaud's avatar
LE GAC Renaud committed
119 120
            ret = self._insert_in_db(log_year=year, **fields)

121 122 123 124 125
        if ret == 1:
            self.logs[-1].load(MSG_LOAD, year)
            return 1

        return 0