File size: 8,203 Bytes
a5d94b7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""(A publicly available subsample of) a reference corpus of Slovene texts."""


import glob
import logging
import os
import os.path
import re
import xml.etree.ElementTree as ET
from copy import deepcopy

import datasets

XML_NAMESPACE = "{http://www.w3.org/XML/1998/namespace}"


def namespace(element):
    # https://stackoverflow.com/a/12946675
    m = re.match(r'\{.*\}', element.tag)
    return m.group(0) if m else ''


_CITATION = """\
@misc{ccGigafida,
    title = {Written corpus {ccGigafida} 1.0},
    author = {Logar, Nata{\v s}a and Erjavec, Toma{\v z} and Krek, Simon and Gr{\v c}ar, Miha and Holozan, Peter},
    url = {http://hdl.handle.net/11356/1035},
    note = {Slovenian language resource repository {CLARIN}.{SI}},
    copyright = {Creative Commons - Attribution-{NonCommercial}-{ShareAlike} 4.0 International ({CC} {BY}-{NC}-{SA} 4.0)},
    issn = {2820-4042},
    year = {2013}
}
"""

_DESCRIPTION = """\
The ccGigafida corpus contains a subsample of the Gigafida corpus. The Gigafida corpus is an extensive collection of 
Slovene text of various genres, from daily newspapers, magazines, all kinds of books (fiction, non-fiction, textbooks), 
web pages, transcriptions of parliamentary debates and similar.
"""

_HOMEPAGE = "http://eng.slovenscina.eu/korpusi/proste-zbirke"

_LICENSE = "Creative Commons - Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)"

_URLS = {
    "ccGigafida": "https://www.clarin.si/repository/xmlui/bitstream/handle/11356/1035/ccGigafidaV1_0.zip"
}


class CcGigafida(datasets.GeneratorBasedBuilder):
    """(A publicly available subsample of) a reference corpus of Slovene texts."""

    VERSION = datasets.Version("1.0.0")

    def _info(self):
        features = datasets.Features(
            {
                "id_doc": datasets.Value("string"),
                "doc_title": datasets.Value("string"),
                "authors": datasets.Sequence(datasets.Value("string")),
                "publish_date": datasets.Value("string"),
                "publisher": datasets.Value("string"),
                "genres": datasets.Sequence(datasets.Value("string")),
                "doc_tokenized": datasets.Sequence(datasets.Sequence(datasets.Sequence(datasets.Value("string")))),
                "doc_string": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
                "id_sents": datasets.Sequence(datasets.Sequence(datasets.Value("string")))
            }
        )

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        # Allow user to specify path to the full Gigafida directory: `load_dataset(..., data_dir=...)`
        if dl_manager.manual_dir is not None:
            data_dir = dl_manager.manual_dir
        else:
            urls = _URLS["ccGigafida"]
            data_dir = dl_manager.download_and_extract(urls)
            data_dir = os.path.join(data_dir, "ccGigafidaV1_0")

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={"data_dir": data_dir}
            )
        ]

    def _generate_examples(self, data_dir):
        GENRE_MAPPING = {
            "SSJ.T": "tisk", "SSJ.T.K": "tisk/knjižno", "SSJ.T.K.L": "tisk/knjižno/leposlovno",
            "SSJ.T.K.S": "tisk/knjižno/strokovno", "SSJ.T.P": "tisk/periodično", "SSJ.T.P.C": "tisk/periodično/časopis",
            "SSJ.T.P.R": "tisk/periodično/revija", "SSJ.T.D": "tisk/drugo", "SSJ.I": "internet"
        }
        # genres are prefixed by "ssj:" in Gigafida 2.0
        for genre, description in deepcopy(GENRE_MAPPING).items():
            GENRE_MAPPING[f"ssj:{genre}"] = description

        # Recursively search for xml files in subdirectories
        all_files = [os.path.join(data_dir, file_name)
                     for file_name in glob.glob(os.path.join(data_dir, "**", "*.xml"), recursive=True)
                     if os.path.isfile(os.path.join(data_dir, file_name))]
        all_files = sorted(all_files)  # fix order

        for _idx_file, file_path in enumerate(all_files):
            curr_doc = ET.parse(file_path)
            root = curr_doc.getroot()
            NAMESPACE = namespace(root)
            id_doc = root.attrib[f"{XML_NAMESPACE}id"]

            # Document metadata
            bibl_el = root.find(f".//{NAMESPACE}bibl")
            doc_title = bibl_el.find(f"{NAMESPACE}title").text.strip()
            authors = list(map(lambda _tag: _tag.text.strip(), bibl_el.findall(f"{NAMESPACE}author")))
            publish_date = bibl_el.find(f"{NAMESPACE}date").text.strip()
            publisher = bibl_el.find(f"{NAMESPACE}publisher").text.strip()

            category_tags = root.findall(f".//{NAMESPACE}catRef")
            genres = []
            for _tag in category_tags:
                # in ccGigafida, the genres are noted with a "#" prefix
                __tag = _tag.attrib["target"][1:] if _tag.attrib["target"].startswith("#") else _tag.attrib["target"]
                mapped_tag = GENRE_MAPPING.get(__tag, None)
                # In addition to the genre of the document, there is sometimes a category assigned by the deduplication tool (dedup:nodup)
                if mapped_tag is None:
                    continue

                genres.append(mapped_tag)

            # Tokenized and raw string version - raw string version preserves spaces
            body_tag = root.find(f".//{NAMESPACE}body")
            tokenized_doc, doc_str = [], []
            doc_sent_ids = []

            for para_tag in body_tag.findall(f".//{NAMESPACE}p"):
                id_para = para_tag.attrib[f"{XML_NAMESPACE}id"]
                tokenized_para, para_str = [], []
                para_sent_ids = []

                for _idx_sent, sent_tag in enumerate(para_tag.findall(f".//{NAMESPACE}s")):
                    # ccGigafida does not have sentence IDs:
                    #   construct ID by taking the paragraph ID + their index in the paragraph
                    id_sent = sent_tag.attrib.get(f"{XML_NAMESPACE}id", None)
                    if id_sent is None:
                        id_sent = f"{id_para}.{_idx_sent}"

                    tokenized_sent, str_sent = [], []
                    for child_tag in sent_tag:
                        tag_str = child_tag.tag[len(NAMESPACE):]
                        if tag_str not in {"w", "S", "c", "pc"}:
                            logging.warning(f"Found unexpected tag in a sentence: '{tag_str}', skipping it.")
                            continue

                        # Tag for whitespace in ccGigafida
                        if tag_str == "S":
                            str_sent.append(" ")

                        # Tag for:
                        # - single-letter characters in ccGigafida;
                        # - whitespace in Gigafida
                        elif tag_str == "c":
                            str_sent.append(child_tag.text)
                            if child_tag.text != " ":
                                tokenized_sent.append(child_tag.text)

                        # word or punctuation character
                        else:
                            str_sent.append(child_tag.text)
                            tokenized_sent.append(child_tag.text)

                    str_sent = "".join(str_sent)
                    tokenized_para.append(tokenized_sent)
                    para_str.append(str_sent)
                    para_sent_ids.append(id_sent)

                tokenized_doc.append(tokenized_para)
                doc_str.append(para_str)
                doc_sent_ids.append(para_sent_ids)

            yield _idx_file, {
                "id_doc": id_doc,
                "doc_title": doc_title,
                "authors": authors,
                "publish_date": publish_date,
                "publisher": publisher,
                "genres": genres,
                "doc_tokenized": tokenized_doc,
                "doc_string": doc_str,
                "id_sents": doc_sent_ids
            }