diff options
Diffstat (limited to 'bsie/extractor/generic/path.py')
-rw-r--r-- | bsie/extractor/generic/path.py | 40 |
1 files changed, 22 insertions, 18 deletions
diff --git a/bsie/extractor/generic/path.py b/bsie/extractor/generic/path.py index c39bbd2..f358a79 100644 --- a/bsie/extractor/generic/path.py +++ b/bsie/extractor/generic/path.py @@ -8,11 +8,10 @@ Author: Matthias Baumgartner, 2022 import os import typing -# inner-module imports +# bsie imports from bsie.base import extractor from bsie.utils import node, ns -from bsie.utils.bsfs import URI -import bsie.reader.path +from bsie.utils.bsfs import schema # exports __all__: typing.Sequence[str] = ( @@ -27,30 +26,31 @@ class Path(extractor.Extractor): CONTENT_READER = bsie.reader.path.Path - def __init__(self): - self.__callmap = { - ns.bse.filename: self.__filename, - } + # mapping from predicate to handler function. + _callmap: typing.Dict[schema.Predicate, typing.Callable[[str], typing.Any]] - def schema(self) -> str: - return ''' - bse:filename a bsfs:Predicate ; + def __init__(self): + super().__init__(schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + ''' + bse:filename rdfs:subClassOf bsfs:Predicate ; rdfs:domain bsfs:Entity ; rdfs:range xsd:string ; - rdf:label "File name"^^xsd:string ; + rdfs:label "File name"^^xsd:string ; schema:description "Filename of entity in some filesystem."^^xsd:string ; owl:maxCardinality "INF"^^xsd:number . - ''' + ''')) + self._callmap = { + self.schema.predicate(ns.bse.filename): self.__filename, + } def extract( self, subject: node.Node, content: CONTENT_READER.CONTENT_TYPE, - predicates: typing.Iterable[URI], - ) -> typing.Iterator[typing.Tuple[node.Node, URI, typing.Any]]: + predicates: typing.Iterable[schema.Predicate], + ) -> typing.Iterator[typing.Tuple[node.Node, schema.Predicate, typing.Any]]: for pred in predicates: # find callback - clbk = self.__callmap.get(pred) + clbk = self._callmap.get(pred) if clbk is None: continue # get value @@ -60,11 +60,15 @@ class Path(extractor.Extractor): # produce triple yield subject, pred, value - def __filename(self, path: str) -> str: + def __filename(self, path: str) -> typing.Optional[str]: try: return os.path.basename(path) - except Exception: - # FIXME: some kind of error reporting (e.g. logging) + except Exception: # some error, skip. + # FIXME: some kind of error reporting (e.g. logging)? + # Options: (a) Fail silently (current); (b) Skip and report to log; + # (c) Raise ExtractorError (aborts extraction); (d) separate content type + # checks from basename errors (report content type errors, skip basename + # errors) return None ## EOF ## |