Docker-in-Docker (DinD) capabilities of public runners deactivated. More info

thesis.py 3.43 KB
Newer Older
1 2 3 4 5 6
""" harvest_tools.thesis

"""
import re


7 8 9
from .automaton import Automaton
from .base import MSG_CRASH, MSG_LOAD
from .checkandfix import CheckException
10 11 12
from plugin_dbui import get_id, UNDEF_ID


13
class Thesis(Automaton):
14
    """Automaton for thesis.
15 16

    """
17 18 19
    def check_record(self, record):
        """Check the content of the thesis in order to fix non conformities.

20
        Args:
LE GAC Renaud's avatar
LE GAC Renaud committed
21 22
            record (RecordThesis):
                record describing a thesis.
23

24
        Returns:
LE GAC Renaud's avatar
LE GAC Renaud committed
25 26 27
            bool:
                ``False`` when a non conformity is found and can not be
                corrected.
28 29 30 31 32

        """
        if not Automaton.check_record(self, record):
            return False

LE GAC Renaud's avatar
LE GAC Renaud committed
33 34 35
        if self.dbg:
            print("check thesis record")

36 37 38
        try:
            self.check.is_thesis(record)
            self.check.submitted(record)
LE GAC Renaud's avatar
LE GAC Renaud committed
39

40 41
            self.check.format_universities(record)

42 43
            self.check.format_authors(record, fmt="F. Last")
            self.check.get_my_authors(record, sort=True)
44

45
        except CheckException as e:
46
            self.logs[-1].reject(e, record=record)
47 48 49
            return False

        except Exception as e:
50
            self.logs[-1].reject(MSG_CRASH % e, record=record, translate=False)
51 52
            return False

LE GAC Renaud's avatar
LE GAC Renaud committed
53
        return True
54

55 56
    def insert_record(self, record):
        """Insert a thesis in the database.
57

58
        Args:
LE GAC Renaud's avatar
LE GAC Renaud committed
59 60
            record (RecordThesis):
                record describing a thesis.
61

62
        Returns:
LE GAC Renaud's avatar
LE GAC Renaud committed
63 64
            int:
                one when the record is inserted / updated in the database
65
                zero otherwise.
66 67 68 69 70 71 72

        """
        db = self.db

        # alias
        defense_date = record.these_defense()
        first_author = record.first_author()
LE GAC Renaud's avatar
LE GAC Renaud committed
73
        id_category = get_id(db.categories, code="PHD")
74 75
        oai_url = record.oai_url()
        title = record.title()
LE GAC Renaud's avatar
LE GAC Renaud committed
76
        universities = ", ".join(record.these_universities())
77

78
        # extract the year from the defence date
79 80 81
        # this approach seems the most reliable
        year = re.search(r"(\d\d\d\d)", defense_date).group(1)

82
        # get an already published thesis
LE GAC Renaud's avatar
LE GAC Renaud committed
83 84 85 86
        fields = dict(first_author=first_author,
                      defense=defense_date,
                      id_projects=self.id_project,
                      id_teams=self.id_team,
87
                      title=title)
LE GAC Renaud's avatar
LE GAC Renaud committed
88

89
        rec_id, status = self.get_record_by_fields(oai_url, year, **fields)
90 91 92 93
        if rec_id:
            return status

        # eventually insert a new thesis
94
        ret = 1
95
        if not self.dry_run:
LE GAC Renaud's avatar
LE GAC Renaud committed
96
            fields = dict(authors=first_author,
97 98 99 100 101 102 103 104 105 106 107 108 109 110
                          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(),
                          submitted=record.submitted()[0],
                          title=title,
                          universities=universities,
                          year=year)
111

LE GAC Renaud's avatar
LE GAC Renaud committed
112 113
            ret = self._insert_in_db(log_year=year, **fields)

114 115 116 117 118
        if ret == 1:
            self.logs[-1].load(MSG_LOAD, year)
            return 1

        return 0