summaryrefslogtreecommitdiffstats
path: root/python/fatcat_tools/importers/grobid_metadata.py
blob: 4d3b41bc3cb6d1556407c930aa9d9915ece9809c (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
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
133
134
135
136
137
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
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
#!/usr/bin/env python3

import sys
import json
import base64
import datetime
import fatcat_client
from .common import EntityImporter

MAX_ABSTRACT_BYTES=4096


class GrobidMetadataImporter(EntityImporter):
    """
    This is a complex case: we need to parse and create both file and release entities.

    The "primary" entity here is really File, not Release. If a matching File
    exists, we bail in want(); if not we insert the Release during parsing, and
    insert both.

    TODO: should instead check if the File has any releases; if not, insert and update.
    TODO: relaxing 'None' constraint on parse_record() might make this refactor-able.
    """

    def __init__(self, api, **kwargs):

        eg_desc = kwargs.get('editgroup_description',
            "Import of release and file metadata, as extracted from PDFs by GROBID.")
        eg_extra = kwargs.get('editgroup_extra', dict())
        eg_extra['agent'] = eg_extra.get('agent', 'fatcat_tools.GrobidMetadataImporter')
        super().__init__(api,
            editgroup_description=eg_desc,
            editgroup_extra=eg_extra)
        self.default_link_rel = kwargs.get("default_link_rel", "web")

    def want(self, raw_record):
        return True

    def parse_record(self, row):

        fields = row.split('\t')
        sha1_key = fields[0]
        cdx = json.loads(fields[1])
        mimetype = fields[2]
        file_size = int(fields[3])
        grobid_meta = json.loads(fields[4])
        fe = self.parse_file_metadata(sha1_key, cdx, mimetype, file_size)
        re = self.parse_grobid_json(grobid_meta)

        if not (fe and re):
            return None

        # lookup existing file SHA1
        existing = None
        try:
            existing = self.api.lookup_file(sha1=fe.sha1)
        except fatcat_client.rest.ApiException as err:
            if err.status != 404:
                raise err

        # if file is already in here, presumably not actually long-tail
        # HACK: this is doing an exists check in parse_record(), which is weird
        # TODO: this is where we should check if the file actually has
        # release_ids and/or URLs associated with it
        if existing and not self.bezerk_mode:
            self.counts['exists'] += 1
            self.counts['skip'] -= 1
            return None

        release_edit = self.create_release(re)
        fe.release_ids.append(release_edit.ident)
        return fe

    def parse_grobid_json(self, obj):

        if not obj.get('title'):
            return None

        extra = dict()

        if obj.get('abstract') and len(obj.get('abstract')) < MAX_ABSTRACT_BYTES:
            abobj = dict(
                mimetype="text/plain",
                language=None,
                content=obj.get('abstract').strip())
            abstracts = [abobj]
        else:
            abstracts = None

        contribs = []
        for i, a in enumerate(obj.get('authors', [])):
            contribs.append(fatcat_client.ReleaseContrib(
                index=i,
                raw_name=a['name'],
                role="author",
                extra=None))

        refs = []
        for raw in obj.get('citations', []):
            cite_extra = dict()
            ref = dict()
            ref['key'] = raw.get('id')
            if raw.get('title'):
                ref['title'] = raw['title'].strip()
            if raw.get('date'):
                try:
                    year = int(raw['date'].strip()[:4])
                    ref['year'] = year
                except:
                    pass
            for key in ('volume', 'url', 'issue', 'publisher'):
                if raw.get(key):
                    cite_extra[key] = raw[key].strip()
            if raw.get('authors'):
                cite_extra['authors'] = [a['name'] for a in raw['authors']]
            if cite_extra:
                cite_extra = dict(grobid=cite_extra)
            else:
                cite_extra = None
            ref['extra'] = cite_extra
            refs.append(ref)

        release_date = None
        release_year = None
        if obj.get('date'):
            # only returns year, ever?
            release_year = int(obj['date'][:4])

        if obj.get('doi'):
            extra['doi'] = obj['doi']
        if obj['journal'] and obj['journal'].get('name'):
            extra['container_name'] = obj['journal']['name']

        extra['is_longtail_oa'] = True

        # TODO: ISSN/eISSN handling? or just journal name lookup?

        if extra:
            extra = dict(grobid=extra)
        else:
            extra = None

        re = fatcat_client.ReleaseEntity(
            title=obj['title'].strip(),
            release_type="article-journal",
            release_date=release_date,
            release_year=release_year,
            contribs=contribs,
            refs=refs,
            publisher=obj['journal'].get('publisher'),
            volume=obj['journal'].get('volume'),
            issue=obj['journal'].get('issue'),
            abstracts=abstracts,
            extra=extra)
        return re

    # TODO: make this a common function somewhere
    def make_url(self, raw):
        rel = self.default_link_rel
        # TODO: this is where we could map specific domains to rel types,
        # and also filter out bad domains, invalid URLs, etc
        if "//archive.org/" in raw or "//arxiv.org/" in raw:
            # TODO: special-case the arxiv.org bulk mirror?
            rel = "repository"
        elif "//web.archive.org/" in raw or "//archive.is/" in raw:
            rel = "webarchive"
        return fatcat_client.FileEntityUrls(url=raw, rel=rel)

    def parse_file_metadata(self, sha1_key, cdx, mimetype, file_size):

        sha1 = base64.b16encode(base64.b32decode(sha1_key.replace('sha1:', ''))).decode('ascii').lower()

        fe = fatcat_client.FileEntity(
            sha1=sha1,
            size=int(file_size),
            mimetype=mimetype,
            release_ids=[],
            urls=[],
        )

        # parse URLs and CDX
        original = cdx['url']
        assert len(cdx['dt']) >= 8
        wayback = "https://web.archive.org/web/{}/{}".format(
            cdx['dt'],
            original)
        fe.urls.append(
            fatcat_client.FileEntityUrls(url=wayback, rel="webarchive"))
        original_url = self.make_url(original)
        if original_url is not None:
            fe.urls.append(original_url)

        return fe

    def try_update(self, entity):
        # did the exists check in 'parse_record()', because we needed to create a release
        return True

    def insert_batch(self, batch):
        self.api.create_file_batch(batch,
            autoaccept=True,
            description=self.editgroup_description,
            extra=json.dumps(self.editgroup_extra))