diff options
Diffstat (limited to 'bsie/lib')
| -rw-r--r-- | bsie/lib/__init__.py | 4 | ||||
| -rw-r--r-- | bsie/lib/bsie.py | 6 | ||||
| -rw-r--r-- | bsie/lib/builder.py | 85 | ||||
| -rw-r--r-- | bsie/lib/pipeline.py | 151 |
4 files changed, 243 insertions, 3 deletions
diff --git a/bsie/lib/__init__.py b/bsie/lib/__init__.py index 578c2c4..4239d3b 100644 --- a/bsie/lib/__init__.py +++ b/bsie/lib/__init__.py @@ -4,15 +4,17 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import typing # inner-module imports from .bsie import BSIE +from .builder import PipelineBuilder # exports __all__: typing.Sequence[str] = ( 'BSIE', + 'PipelineBuilder', ) ## EOF ## diff --git a/bsie/lib/bsie.py b/bsie/lib/bsie.py index e087fa9..668783d 100644 --- a/bsie/lib/bsie.py +++ b/bsie/lib/bsie.py @@ -4,13 +4,15 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import typing # bsie imports -from bsie.tools import Pipeline from bsie.utils import bsfs, node, ns +# inner-module imports +from .pipeline import Pipeline + # exports __all__: typing.Sequence[str] = ( 'BSIE', diff --git a/bsie/lib/builder.py b/bsie/lib/builder.py new file mode 100644 index 0000000..c2abffe --- /dev/null +++ b/bsie/lib/builder.py @@ -0,0 +1,85 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import logging +import typing + +# bsie imports +from bsie.extractor import ExtractorBuilder +from bsie.reader import ReaderBuilder +from bsie.utils import bsfs, errors + +# inner-module imports +from . import pipeline + +# exports +__all__: typing.Sequence[str] = ( + 'PipelineBuilder', + ) + + +## code ## + +logger = logging.getLogger(__name__) + +class PipelineBuilder(): + """Build `bsie.tools.pipeline.Pipeline` instances.""" + + # Prefix to be used in the Pipeline. + prefix: bsfs.Namespace + + # builder for Readers. + rbuild: ReaderBuilder + + # builder for Extractors. + ebuild: ExtractorBuilder + + def __init__( + self, + prefix: bsfs.Namespace, + reader_builder: ReaderBuilder, + extractor_builder: ExtractorBuilder, + ): + self.prefix = prefix + self.rbuild = reader_builder + self.ebuild = extractor_builder + + def build(self) -> pipeline.Pipeline: + """Return a Pipeline instance.""" + ext2rdr = {} + + for eidx in self.ebuild: + # build extractor + try: + ext = self.ebuild.build(eidx) + + except errors.LoaderError as err: # failed to load extractor; skip + logger.error('failed to load extractor: %s', err) + continue + + except errors.BuilderError as err: # failed to build instance; skip + logger.error(str(err)) + continue + + try: + # get reader required by extractor + if ext.CONTENT_READER is not None: + rdr = self.rbuild.build(ext.CONTENT_READER) + else: + rdr = None + # store extractor + ext2rdr[ext] = rdr + + except errors.LoaderError as err: # failed to load reader + logger.error('failed to load reader: %s', err) + + except errors.BuilderError as err: # failed to build reader + logger.error(str(err)) + + return pipeline.Pipeline(self.prefix, ext2rdr) + +## EOF ## diff --git a/bsie/lib/pipeline.py b/bsie/lib/pipeline.py new file mode 100644 index 0000000..44685ba --- /dev/null +++ b/bsie/lib/pipeline.py @@ -0,0 +1,151 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +from collections import defaultdict +import logging +import typing + +# bsie imports +from bsie.extractor import Extractor +from bsie.reader import Reader +from bsie.utils import bsfs, errors, node, ns + +# exports +__all__: typing.Sequence[str] = ( + 'Pipeline', + ) + +# constants +FILE_PREFIX = 'file#' + +## code ## + +logger = logging.getLogger(__name__) + +class Pipeline(): + """Extraction pipeline to generate triples from files. + + The Pipeline binds readers and extractors, and performs + the necessary operations to produce triples from a file. + It takes a best-effort approach to extract as many triples + as possible. Errors during the extraction are passed over + and reported to the log. + + """ + + # combined extractor schemas. + _schema: bsfs.schema.Schema + + # node prefix. + _prefix: bsfs.Namespace + + # extractor -> reader mapping + _ext2rdr: typing.Dict[Extractor, typing.Optional[Reader]] + + def __init__( + self, + prefix: bsfs.Namespace, + ext2rdr: typing.Dict[Extractor, typing.Optional[Reader]] + ): + # store core members + self._prefix = prefix + FILE_PREFIX + self._ext2rdr = ext2rdr + # compile schema from all extractors + self._schema = bsfs.schema.Schema.Union(ext.schema for ext in ext2rdr) + + def __str__(self) -> str: + return bsfs.typename(self) + + def __repr__(self) -> str: + return f'{bsfs.typename(self)}(...)' + + def __hash__(self) -> int: + return hash((type(self), self._prefix, self._schema, tuple(self._ext2rdr), tuple(self._ext2rdr.values()))) + + def __eq__(self, other: typing.Any) -> bool: + return isinstance(other, type(self)) \ + and self._schema == other._schema \ + and self._prefix == other._prefix \ + and self._ext2rdr == other._ext2rdr + + @property + def schema(self) -> bsfs.schema.Schema: + """Return the pipeline's schema (combined from all extractors).""" + return self._schema + + @property + def principals(self) -> typing.Iterator[bsfs.schema.Predicate]: + """Return the principal predicates that can be extracted.""" + return iter({pred for ext in self._ext2rdr for pred in ext.principals}) + + def subschema(self, principals: typing.Iterable[bsfs.schema.Predicate]) -> bsfs.schema.Schema: + """Return the subset of the schema that supports the given *principals*.""" + # materialize principals + principals = set(principals) + # collect and combine schemas from extractors + return bsfs.schema.Schema.Union({ + ext.schema + for ext + in self._ext2rdr + if not set(ext.principals).isdisjoint(principals) + }) + + def __call__( + self, + path: bsfs.URI, + principals: typing.Optional[typing.Iterable[bsfs.schema.Predicate]] = None, + ) -> typing.Iterator[typing.Tuple[node.Node, bsfs.schema.Predicate, typing.Any]]: + """Extract triples from the file at *path*. Optionally, limit triples to *principals*.""" + # get principals + principals = set(principals) if principals is not None else set(self.schema.predicates()) + + # get extractors + extractors = {ext for ext in self._ext2rdr if not set(ext.principals).isdisjoint(principals)} + + # corner-case short-cut + if len(extractors) == 0: + return + + # get readers -> extractors mapping + rdr2ext = defaultdict(set) + for ext in extractors: + rdr = self._ext2rdr[ext] + rdr2ext[rdr].add(ext) + + # create subject for file + uuid = bsfs.uuid.UCID.from_path(path) + subject = node.Node(ns.bsfs.File, self._prefix[uuid]) + + # extract information + for rdr, extrs in rdr2ext.items(): + try: + # get content + content = rdr(path) if rdr is not None else None + #logger.info('extracted %s from %s', rdr, path) + + # apply extractors on this content + for ext in extrs: + try: + # get predicate/value tuples + for subject, pred, value in ext.extract(subject, content, principals): + yield subject, pred, value + + except errors.ExtractorError as err: + # critical extractor failure. + logger.error('%s failed to extract triples from content: %s', ext, err) + + except errors.UnsupportedFileFormatError: + # failed to read the file format. skip. + #logger.warning('%s could not process the file format of %s', rdr, err) + pass + + except errors.ReaderError as err: + # failed to read any content. skip. + logger.error('%s failed to read content: %s', rdr, err) + + +## EOF ## |