aboutsummaryrefslogtreecommitdiffstats
path: root/python/parse_arxivraw_xml.py
blob: 16def821f712dd0efabfe57ced46710814e2ab28 (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

import sys
import json
import datetime
from bs4 import BeautifulSoup
from bs4.element import NavigableString
from pylatexenc.latex2text import LatexNodes2Text


latex2text = LatexNodes2Text()

def parse_arxiv_authors(raw):
    if not raw:
        return []
    authors = raw.split(', ')
    if authors:
        last = authors[-1].split(" and ")
        if len(last) == 2:
            authors[-1] = last[0]
            authors.append(last[1])
    authors = [latex2text.latex_to_text(a).strip() for a in authors]
    return authors

def test_parse_arxiv_authors():

    assert parse_arxiv_authors("Raphael Chetrite, Shamik Gupta, Izaak Neri and \\'Edgar Rold\\'an") == [
        "Raphael Chetrite",
        "Shamik Gupta",
        "Izaak Neri",
        "Édgar Roldán",
    ]
    assert parse_arxiv_authors("Izaak Neri and \\'Edgar Rold\\'an") == [
        "Izaak Neri",
        "Édgar Roldán",
    ]
    assert parse_arxiv_authors("Raphael Chetrite Shamik Gupta") == [
        "Raphael Chetrite Shamik Gupta",
    ]

class ArxivRawXmlParser():
    """
    Converts arxiv.org "arXivRaw" OAI-PMH XML records to fatcat release entities

    TODO: this will require a special importer that keeps works together
    TODO: arxiv_id lookup in API (rust) with no version specified should select
          the "most recent" version; can be a simple sort?
    """

    def __init__(self):
        pass

    def parse_file(self, handle):

        # 1. open with beautiful soup
        soup = BeautifulSoup(handle, "xml")

        # 2. iterate over articles, call parse_article on each
        for article in soup.find_all("record"):
            resp = self.parse_record(article)
            print(json.dumps(resp))
            #sys.exit(-1)


    def parse_record(self, record):

        metadata = record.arXivRaw
        extra = dict()
        extra_arxiv = dict()

        base_id = metadata.id.string
        doi = None
        if metadata.doi and metadata.doi.string:
            doi = metadata.doi.string.lower().strip()
            assert doi.startswith('10.')
        title = latex2text.latex_to_text(metadata.title.string)
        authors = parse_arxiv_authors(metadata.authors.string)
        contribs = [dict(raw_name=a, role='author') for a in authors]

        lang = "en"     # the vast majority in english
        if metadata.comments and metadata.comments.string:
            comments = metadata.comments.string.strip()
            extra_arxiv['comments'] = comments
            if 'in french' in comments.lower():
                lang = 'fr'
            elif 'in spanish' in comments.lower():
                lang = 'es'
            elif 'in portuguese' in comments.lower():
                lang = 'pt'
            elif 'in hindi' in comments.lower():
                lang = 'hi'
            elif 'in japanese' in comments.lower():
                lang = 'ja'
            elif 'in german' in comments.lower():
                lang = 'de'
            elif 'simplified chinese' in comments.lower():
                lang = 'zh'
            elif 'in russian' in comments.lower():
                lang = 'ru'
            # more languages?

        release_type = "article-journal"

        if metadata.find('journal-ref') and metadata.find('journal-ref').string:
            journal_ref = metadata.find('journal-ref').string.strip()
            extra_arxiv['journal_ref'] = journal_ref
            if "conf." in journal_ref.lower() or "proc." in journal_ref.lower():
                release_type = "conference-paper"
        if metadata.find('report-no') and metadata.find('report-no').string:
            extra['number'] = metadata.find('report-no').string.strip()
            release_type = "report"
        if metadata.find('acm-class') and metadata.find('acm-class').string:
            extra_arxiv['acm_class'] = metadata.find('acm_class').string.strip()
        if metadata.categories and metadata.categories.string:
            extra_arxiv['categories'] = metadata.categories.string.split()
        license_slug = None
        if metadata.license and metadata.license.string:
            # XXX: convert URL to slug
            license_slug = metadata.license.string.strip()
        abstracts = None
        if metadata.abstract:
            # TODO: test for this multi-abstract code path
            abstracts = []
            abst = metadata.abstract.string.strip()
            orig = None
            if '-----' in abst:
                both = abst.split('-----')
                abst = both[0].strip()
                orig = both[1].strip()
            if '$' in abst or '{' in abstr:
                mime = "application/x-latex"
                abst_plain = latex2text.latex_to_text(abst)
                abstracts.append(dict(content=abst_plain, mime="text/plain", lang="en"))
            else:
                mime = "text/plain"
            abstracts.append(dict(content=abst, mime=mime, lang="en"))
            if orig:
                abstracts.append(dict(content=orig, mime=mime))

        if extra_arxiv:
            extra['arxiv'] = extra_arxiv
        if not extra:
            extra = None

        versions = []
        for version in metadata.find_all('version'):
            arxiv_id = base_id + version['version']
            release_date = version.date.string.strip()
            release_date = datetime.datetime.strptime(release_date, "%a, %d %b %Y %H:%M:%S %Z").date()
            versions.append(dict(
                work_id=None,
                title=title,
                #original_title
                release_type="article-journal",
                release_status='submitted', # XXX: source_type?
                release_date=release_date.isoformat(),
                release_year=release_date.year,
                arxiv_id=arxiv_id,
                #doi (see below)
                #pmid
                #pmcid
                #isbn13     # never in Article
                #volume
                #issue
                #pages
                #publisher
                language=lang,
                #license_slug   # not in MEDLINE

                # content, mimetype, lang
                abstracts=abstracts,

                # raw_name, role, raw_affiliation, extra
                contribs=contribs,

                #   name, type, publisher, issnl
                #   extra: issnp, issne, original_name, languages, country
                #container=container,   # very little/none; resolve via DOI?

                # extra:
                #   withdrawn_date
                #   translation_of
                #   subtitle
                #   aliases
                #   container_name
                #   group-title
                #   pubmed: retraction refs
                extra=extra,
            ))

        # only apply DOI to most recent version (HACK)
        if doi:
            versions[-1]['doi'] = doi
            versions[-1]['release_status'] = "published"
        return base_id, versions

if __name__=='__main__':
    parser = ArxivRawXmlParser()
    parser.parse_file(open(sys.argv[1]))