File size: 8,426 Bytes
07423df |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 |
import os
from abc import abstractmethod
from dataclasses import dataclass
from typing import Any, Callable, List, Optional, Sequence, Set, Tuple, Union
from llm_studio.src.nesting import Dependency
def _scan_dirs(dirname) -> List[str]:
"""Scans a directory for subfolders
Args:
dirname: directory name
Returns:
List of subfolders
"""
subfolders = [f.path for f in os.scandir(dirname) if f.is_dir()]
for dirname in list(subfolders):
subfolders.extend(_scan_dirs(dirname))
subfolders = [x + "/" if x[-1] != "/" else x for x in subfolders]
return subfolders
def _scan_files(
dirname, extensions: Tuple[str, ...] = (".csv", ".pq", ".parquet", ".json")
) -> List[str]:
"""Scans a directory for files with given extension
Args:
dirname: directory name
extensions: extensions to consider
Returns:
List of files
"""
path_list = [
os.path.join(dirpath, filename)
for dirpath, _, filenames in os.walk(dirname)
for filename in filenames
if any(map(filename.__contains__, extensions))
and not filename.startswith("__meta_info__")
]
return sorted(path_list)
def strip_prefix(paths: Sequence[str], ignore_set: Set[str] = set()) -> Tuple[str, ...]:
"""
Strips the common prefix of all the given paths.
Args:
paths: the paths to strip
ignore_set: set of path names to ignore when computing the prefix.
Returns:
List with the same length as `paths` without common prefixes.
"""
paths_to_check = [
os.path.split(os.path.normpath(path))[0]
for path in paths
if path not in ignore_set
]
if len(paths_to_check) == 0:
return tuple(paths)
prefix = os.path.commonpath(paths_to_check)
stripped = tuple(
[
path if path in ignore_set else os.path.relpath(path, prefix)
for path in paths
]
)
return stripped
class Value:
pass
@dataclass
class Number:
min: Optional[float] = None
max: Optional[float] = None
step: Union[str, float] = 1.0
@dataclass
class String:
# Each element of the tuple can be either:
# - a tuple of (value, name)
# - a string. In that case the same value will be used for name and value
values: Any = None
allow_custom: bool = False
placeholder: Optional[str] = None
class DatasetValue:
pass
@abstractmethod
def get_value(
self, dataset: Any, value: Any, type_annotation: type, mode: str
) -> Tuple[String, Any]:
pass
@staticmethod
def _compute_current_values(
current_values: List[str],
possible_values: List[str],
prefer_with: Optional[Callable[[str], bool]] = None,
) -> List[str]:
"""
Compute current values.
Args:
current_values: The preliminary current values.
possible_values: All possible values.
prefer_with: Function determining which values to prefer as default.
Returns:
A list
"""
if len(possible_values) == 0:
return [""]
# allow only values which are in the possible values
current_values = list(
filter(lambda value: value in possible_values, current_values)
)
if len(current_values) == 0:
# if the values are empty, take all the values where `prefer_with` is true
for c in possible_values:
if prefer_with is not None and prefer_with(c):
current_values.append(c)
# if they are still empty, just take the first possible value
if len(current_values) == 0:
current_values = [possible_values[0]]
return current_values
@dataclass
class Directories(DatasetValue):
add_none: Union[bool, Callable[[str], bool]] = False
prefer_with: Optional[Callable[[str], bool]] = None
prefer_none: bool = True
def get_value(self, dataset, value, type_annotation, mode) -> Tuple[String, Any]:
if dataset is None:
return String(tuple()), value
available_dirs = _scan_dirs(dataset["path"])
if (isinstance(self.add_none, bool) and self.add_none) or (
callable(self.add_none) and self.add_none(mode)
):
if self.prefer_none:
available_dirs.insert(0, "None")
else:
available_dirs.insert(len(available_dirs), "None")
if isinstance(value, str):
value = [value]
value = DatasetValue._compute_current_values(
value, available_dirs, self.prefer_with
)
return (
String(
tuple(
zip(
available_dirs,
strip_prefix(available_dirs, ignore_set={"None"}),
)
)
),
value if type_annotation == Tuple[str, ...] else value[0],
)
@dataclass
class Files(DatasetValue):
add_none: Union[bool, Callable[[str], bool]] = False
prefer_with: Optional[Callable[[str], bool]] = None
# For the case where no match found, whether to prioritize
# selecting any file or selecting no file
prefer_none: bool = True
def get_value(self, dataset, value, type_annotation, mode) -> Tuple[String, Any]:
if dataset is None:
return String(tuple()), value
available_files = _scan_files(dataset["path"])
if (isinstance(self.add_none, bool) and self.add_none) or (
callable(self.add_none) and self.add_none(mode)
):
if self.prefer_none:
available_files.insert(0, "None")
else:
available_files.insert(len(available_files), "None")
if isinstance(value, str):
value = [value]
value = DatasetValue._compute_current_values(
value, available_files, self.prefer_with
)
return (
String(
tuple(
zip(
available_files,
strip_prefix(available_files, ignore_set={"None"}),
)
)
),
value if type_annotation == Tuple[str, ...] else value[0],
)
@dataclass
class Columns(DatasetValue):
add_none: Union[bool, Callable[[str], bool]] = False
prefer_with: Optional[Callable[[str], bool]] = None
def get_value(self, dataset, value, type_annotation, mode) -> Tuple[String, Any]:
if dataset is None:
return String(tuple()), value
try:
columns = list(dataset["dataframe"].columns)
except KeyError:
columns = []
if (isinstance(self.add_none, bool) and self.add_none) or (
callable(self.add_none) and self.add_none(mode)
):
columns.insert(0, "None")
if isinstance(value, str):
value = [value]
if value is None:
value = [columns[0]]
value = DatasetValue._compute_current_values(value, columns, self.prefer_with)
return (
String(tuple(columns)),
value if type_annotation == Tuple[str, ...] else value[0],
)
@dataclass
class ColumnValue(DatasetValue):
column: str
default: List[str]
prefer_with: Optional[Callable[[str], bool]] = None
dependency: Optional[Dependency] = None
def get_value(self, dataset, value, type_annotation, mode) -> Tuple[String, Any]:
if dataset is None:
return String(tuple()), value
try:
df = dataset["dataframe"]
except KeyError:
df = None
if df is not None:
if self.dependency is not None and not self.dependency.check(
[dataset[self.dependency.key]]
):
values = self.default
elif self.column in df:
values = [str(v) for v in sorted(list(df[self.column].unique()))]
else:
values = self.default
else:
values = self.default
value = DatasetValue._compute_current_values(value, values, self.prefer_with)
return (
String(tuple(values)),
value if type_annotation == Tuple[str, ...] else value[0],
)
|