Exceptions and logging¶

Exceptions¶

All library exceptions inherit the base class RutsError and one of the built-in Python classes, so they can be caught both by their ruTS name and by the built-in type, for example except ValueError.

Exception	Built-in class	When raised
`RutsError`	`Exception`	Base class, never raised itself
`SourceTypeError`	`TypeError`	An argument of a wrong type: the data source is neither a string nor a `Doc`, the frequency counter is not a `Counter`, the tokenizer is not callable and the like
`SourceError`	`ValueError`	The source has no words, sentences, texts, collocations or dependency parse, has fewer texts than the measure needs, or nothing is left after culling
`ParameterError`	`ValueError`	A parameter value is out of range; an unknown measure, preset, layer, dataset category and the like
`UnknownStatError`	`ParameterError`, `KeyError`	An unknown statistic is requested by name (`describe`, `DiversityStats.windowed`, `MorphStats.get_stats` and others)
`DatasetNotFoundError`	`OSError`	The dataset is not downloaded; the message shows the download command
`DataFileError`	`ValueError`	A dataset file is corrupted, has an unexpected format or cannot be decoded
`DownloadError`	`RuntimeError`	The file could not be downloaded or failed the checksum verification

The classes are available from ruts and ruts.exceptions. They are the very classes of the anyTS core (RutsError is its AnyTSError), not subclasses: except RutsError catches the errors of the core code too, and a traceback shows anyts.exceptions.SourceError. The messages are in Russian, except the argument type checks in the core code and some core functions.

CodeResult

from ruts import BasicStats, RutsError, SourceError
from ruts.datasets import PoetryCorpus

try:
    BasicStats("...")
except SourceError as e:
    print(e)

try:
    list(PoetryCorpus(data_dir="/nowhere").get_texts(limit=1))
except RutsError as e:
    print(type(e).__name__)

В источнике данных отсутствуют слова
DatasetNotFoundError

Logging¶

The anyTS core writes messages about downloading and extracting datasets (download()) to the anyts logger at the INFO level. They are silent by default; to see them, configure logging in your application:

CodeResult

import logging

from ruts.datasets import SovChLit

logging.basicConfig(level=logging.INFO, format="%(name)s: %(message)s")
SovChLit(data_dir="/tmp/ruts_data").download()

anyts.datasets: Downloading the file https://github.com/SergeyShk/ruTS/raw/master/ruts/datasets/data/sov_chrest_lit_v1.tar.xz
anyts.datasets: The file is downloaded: /tmp/ruts_data/sov_chrest_lit_v1.tar.xz
anyts.datasets: Extracting the archive /tmp/ruts_data/sov_chrest_lit_v1.tar.xz

print_stats() and print_kwic() write to standard output; logging does not affect them.