neuralib.persistence.cli_persistence

Persistence Class for argp supporting

author:

Ta-Shun Su

class neuralib.persistence.cli_persistence.PersistenceOptions[source]

The Option class that handle one kind of cache class T, including load cache, create cache, and save cache.

GROUP_CACHE = 'Persistence options'
invalid_cache: bool

invalid persistence data

property persistence_class: type[T]
persistence_handler(dest)[source]
Parameters:

dest (Path) – save root directory.

Returns:

Return type:

PersistenceHandler[T]

abstract empty_cache()[source]

create an empty cache which only initialize required fields.

Returns:

cache instance

Return type:

T

find_cache(result, dest=None, validator=False)[source]

Find the persistence.

for all fields

>>> template = self.empty_cache()
>>> template.a = 1  # want to find all cache whose `a` equals to 1
>>> template.b = field_missing # want to find all cache and don't matter what `b` is
>>> found = self.find_cache(template)
Parameters:

  • result (T)

  • dest (Path | None) – save root directory.

  • validator

Returns:

Return type:

list[T]

save_cache(result, dest, force=True)[source]
Parameters:

  • result (T)

  • dest (Path) – save root directory.

  • force

Returns:

load_cache(result=None, error_when_missing=False, dest=None, **kwargs)[source]

load persistence from disk according to result’s required fields.

Parameters:

  • result (T | None) – persistence instance with necessary fields filled.

  • error_when_missing – do not try to generate the cache when cache missing.

  • dest (Path | None) – save root directory.

  • kwargs – overwrite field value in result.

Returns:

persistence instance.

Raises:

FileNotFoundError – error_when_missing and file not found.

Return type:

T

validate_cache(result_path, result)[source]

Validating loaded cache instance.

Once validating fail (return False), goto _compute_cache().

Parameters:

  • result_path (Path)

  • result (T)

Returns:

False if validating fail.

Raises:

  • TypeError – if validating fail.

  • ValueError – if validating fail.

  • KeyError – if validating fail.

  • AttributeError – if validating fail.

  • RuntimeError – if validating fail.

Return type:

bool

abstract compute_cache(result)[source]

Compute cache according to cache’s required fields.

Parameters:

result (T)

Returns:

computed result

Return type:

T

neuralib.persistence.cli_persistence.get_options_and_cache(opt_cls, ref, error_when_missing=False, **kwargs)[source]

copy the arguments from PersistenceOpt (class that compute the cache) to Apply*Opt Can be used for analysis apply two different cached files

Example
>>> @persistence.persistence_class
>>> class SortIdxCache:
...     # for cache attributes declare
...     pass

>>> class SortIdxOptions(PersistenceOptions[SortIdxCache]):
...     # for cache computing
...     def empty_cache(self) -> SortIdxCache:
...         pass
...     def _compute_cache(self, cache: SortIdxCache) -> SortIdxCache:
...         pass

>>> class ApplySortIdxOptions:
...     # for cache applying, suppose to be the one layer before parent class for analysis class
...     def apply_sort_idx_cache(self) -> SortIdxCache:
...         return get_options_and_cache(SortIdxOptions, self)

>>> class CPBeltSortOptions(ApplyPosBinActOptions, ApplySortIdxOptions):
...     pass
Parameters:
Returns:

Return type:

T