Datasets:
Tasks:
Text Generation
Formats:
parquet
Sub-tasks:
language-modeling
Languages:
Danish
Size:
10M - 100M
ArXiv:
DOI:
License:
File size: 6,759 Bytes
566156e 3007163 566156e f1205a6 566156e ebc5129 566156e 3007163 566156e 060c443 566156e ebc5129 3007163 566156e 060c443 566156e 060c443 ebc5129 3007163 ebc5129 3007163 ebc5129 3007163 ebc5129 566156e a5ac9e2 3007163 566156e a5ac9e2 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 |
from pathlib import Path
from typing import Literal
import pandas as pd
from dynaword.datasheet import DataSheet, human_readable_large_int
from dynaword.paths import repo_path
main_sheet = DataSheet.load_from_path(repo_path / "README.md")
_datasets = [
cfg["config_name"] # type: ignore
for cfg in main_sheet.frontmatter["configs"] # type: ignore
if cfg["config_name"] != "default" # type: ignore
]
DEFAULT_LICENSE_REFERENCES = """[CC-0]: https://creativecommons.org/publicdomain/zero/1.0/legalcode.en
[CC-BY-SA 4.0]: https://creativecommons.org/licenses/by-sa/4.0/deed.en
[Apache 2.0]: https://www.apache.org/licenses/LICENSE-2.0
"""
def create_license_references() -> str:
license_references = DEFAULT_LICENSE_REFERENCES
for dataset in _datasets:
dataset_path = repo_path / "data" / dataset
readme_path = dataset_path / f"{dataset_path.name}.md"
sheet = DataSheet.load_from_path(readme_path)
if sheet.license == "other":
license_name = sheet.frontmatter["license_name"]
license_references += f"[{license_name}]: ./data/{dataset_path.name}/{dataset_path.name}.md#license-information\n"
return license_references
def create_dataset_readme_references():
readme_references = ""
for dataset in _datasets:
dataset_path = repo_path / "data" / dataset
readme_references += (
f"[{dataset_path.name}]: data/{dataset_path.name}/{dataset_path.name}.md\n"
)
return readme_references
def create_overview_table(
repo_path: Path = repo_path,
add_readable_tokens: bool = True,
add_total_row: bool = True,
add_readme_references: bool = True,
) -> pd.DataFrame:
table = {
"Source": [],
"Sources": [],
"Description": [],
"Domain": [],
"N. Tokens": [],
"License": [],
}
for dataset in _datasets:
dataset_path = repo_path / "data" / dataset
readme_path = dataset_path / f"{dataset_path.name}.md"
sheet = DataSheet.load_from_path(readme_path)
desc_stats = sheet.get_descritive_stats()
main_domain = sheet.domains[0] if sheet.domains else ""
table["Source"] += [f"{dataset_path.name}"]
table["Sources"] += [f"[{dataset_path.name}]"]
table["License"] += [f"[{sheet.license_name}]"]
table["Domain"] += [main_domain]
table["Description"] += [sheet.short_description]
table["N. Tokens"] += [desc_stats.number_of_tokens]
df = pd.DataFrame.from_dict(table)
df = df.sort_values("N. Tokens", ascending=False)
if add_total_row:
total_row = {
"Source": "**Total**",
"Sources": "**Total**",
"Domain": "",
"License": "",
"Description": "",
"N. Tokens": sum(table["N. Tokens"]),
}
df = pd.concat(
[
df,
pd.DataFrame([total_row]),
],
ignore_index=True,
)
if add_readme_references:
# replace Source with Sources
df["Source"] = df["Sources"]
df = df.drop(columns=["Sources"])
else:
# remove Sources
df = df.drop(columns=["Sources"])
if add_readable_tokens:
df["N. Tokens"] = df["N. Tokens"].apply(human_readable_large_int)
return df
def _get_normalized_license(ds: DataSheet) -> str:
non_standard_license_names = {
"Apache 2.0": "Other (Attribution required)",
"NLOD 2.0": "Other (Attribution required)",
"DanNet 1.0": "Other (Attribution required)",
"Gutenberg": "Other (Attribution required)",
"Danish Copyright Law": "Other (No attribution required)",
}
if (
ds.license_name not in non_standard_license_names
and ds.license_name is not None
):
return ds.license_name
if ds.license_name is None:
raise ValueError(
f"Datasheet {ds.pretty_name} has no license name specified in the frontmatter."
)
return non_standard_license_names[ds.license_name]
def _get_feature_by_string(
datasheet: DataSheet, feature_name: Literal["Domain", "Language", "License"]
) -> str:
"""Get a specific feature from the frontmatter."""
match feature_name:
case "Domain":
return datasheet.domains[0] if datasheet.domains else "N/A"
case "Language":
return ", ".join(datasheet.language)
case "License":
return _get_normalized_license(datasheet)
case _:
raise ValueError(f"Unknown feature: {feature_name}")
def create_grouped_table(
group: Literal["Domain", "Language", "License"] = "Domain",
repo_path: Path = repo_path,
add_readable_tokens: bool = True,
add_total_row: bool = True,
) -> pd.DataFrame:
table = {
"Sources": [],
group: [],
"N. Tokens": [],
}
for dataset in _datasets:
dataset_path = repo_path / "data" / dataset
readme_path = dataset_path / f"{dataset_path.name}.md"
sheet = DataSheet.load_from_path(readme_path)
desc_stats = sheet.get_descritive_stats()
feature = _get_feature_by_string(sheet, group)
table["Sources"] += [f"[{dataset_path.name}]"]
table[group] += [feature]
table["N. Tokens"] += [desc_stats.number_of_tokens]
if add_total_row:
table["Sources"] += [""]
table[group] += ["**Total**"]
table["N. Tokens"] += [sum(table["N. Tokens"])]
df = pd.DataFrame.from_dict(table)
df = df.groupby(group).agg({"Sources": lambda x: ", ".join(x), "N. Tokens": "sum"})
df = df.sort_values("N. Tokens", ascending=False)
df.index.name = group
df = df.reset_index()
# Trick the Total row to be at the bottom.
new_index = list(df.index.drop(0)) + [0]
df = df.reindex(new_index)
if add_readable_tokens:
df["N. Tokens"] = df["N. Tokens"].apply(human_readable_large_int)
return df
def create_grouped_table_str(
repo_path: Path = repo_path,
group: Literal["Domain", "Language", "License"] = "Domain",
) -> str:
table = create_grouped_table(group=group, repo_path=repo_path)
readme_references = create_dataset_readme_references()
package = f"{table.to_markdown(index=False, maxcolwidths=[None, None, None])}\n\n{readme_references}\n\n"
return package
def create_overview_table_str(repo_path: Path = repo_path) -> str:
main_table = create_overview_table(repo_path)
readme_references = create_dataset_readme_references()
license_references = create_license_references()
package = f"{main_table.to_markdown(index=False)}\n\n{readme_references}\n\n{license_references}\n\n"
return package
|