diff options
Diffstat (limited to 'bsie')
37 files changed, 1152 insertions, 315 deletions
diff --git a/bsie/__init__.py b/bsie/__init__.py index 8d2308c..c253f39 100644 --- a/bsie/__init__.py +++ b/bsie/__init__.py @@ -4,7 +4,7 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import collections import typing diff --git a/bsie/apps/__init__.py b/bsie/apps/__init__.py index a548c3c..1c3d0f9 100644 --- a/bsie/apps/__init__.py +++ b/bsie/apps/__init__.py @@ -4,7 +4,7 @@ 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 diff --git a/bsie/apps/index.py b/bsie/apps/index.py index 1dbfdd8..21c2318 100644 --- a/bsie/apps/index.py +++ b/bsie/apps/index.py @@ -4,16 +4,16 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import argparse import os import typing # bsie imports -from bsie.base import errors -from bsie.lib import BSIE -from bsie.tools import builder -from bsie.utils import bsfs +from bsie.extractor import ExtractorBuilder +from bsie.lib import BSIE, PipelineBuilder +from bsie.reader import ReaderBuilder +from bsie.utils import bsfs, errors # exports __all__: typing.Sequence[str] = ( @@ -44,9 +44,9 @@ def main(argv): # FIXME: Read reader/extractor configs from a config file # reader builder - rbuild = builder.ReaderBuilder({}) + rbuild = ReaderBuilder() # extractor builder - ebuild = builder.ExtractorBuilder([ + ebuild = ExtractorBuilder([ {'bsie.extractor.generic.path.Path': {}}, {'bsie.extractor.generic.stat.Stat': {}}, {'bsie.extractor.generic.constant.Constant': dict( @@ -58,9 +58,14 @@ def main(argv): bsfs:unique "true"^^xsd:boolean . ''', )}, + {'bsie.extractor.image.colors_spatial.ColorsSpatial': { + 'width': 2, + 'height': 2, + 'exp': 2, + }}, ]) # pipeline builder - pbuild = builder.PipelineBuilder( + pbuild = PipelineBuilder( bsfs.Namespace(args.user + ('/' if not args.user.endswith('/') else '')), rbuild, ebuild, @@ -82,7 +87,9 @@ def main(argv): # index input paths for path in args.input_file: - if os.path.isdir(path) and args.recursive: + if not os.path.exists(path): + pass # FIXME: notify the user + elif os.path.isdir(path) and args.recursive: for dirpath, _, filenames in os.walk(path, topdown=True, followlinks=args.follow): for filename in filenames: for node, pred, value in bsie.from_file(os.path.join(dirpath, filename)): diff --git a/bsie/apps/info.py b/bsie/apps/info.py index eaf1f71..64a4eba 100644 --- a/bsie/apps/info.py +++ b/bsie/apps/info.py @@ -4,15 +4,16 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import argparse import sys import typing # bsie imports -from bsie.base import errors -from bsie.tools import builder -from bsie.utils import bsfs +from bsie.extractor import ExtractorBuilder +from bsie.lib import PipelineBuilder +from bsie.reader import ReaderBuilder +from bsie.utils import bsfs, errors # exports __all__: typing.Sequence[str] = ( @@ -25,15 +26,15 @@ __all__: typing.Sequence[str] = ( def main(argv): """Show information from BSIE.""" parser = argparse.ArgumentParser(description=main.__doc__, prog='info') - parser.add_argument('what', choices=('predicates', ), + parser.add_argument('what', choices=('predicates', 'schema'), help='Select what information to show.') args = parser.parse_args(argv) # FIXME: Read reader/extractor configs from a config file # reader builder - rbuild = builder.ReaderBuilder({}) + rbuild = ReaderBuilder() # extractor builder - ebuild = builder.ExtractorBuilder([ + ebuild = ExtractorBuilder([ {'bsie.extractor.generic.path.Path': {}}, {'bsie.extractor.generic.stat.Stat': {}}, {'bsie.extractor.generic.constant.Constant': dict( @@ -45,9 +46,14 @@ def main(argv): bsfs:unique "true"^^xsd:boolean . ''', )}, + {'bsie.extractor.image.colors_spatial.ColorsSpatial': { + 'width': 2, + 'height': 2, + 'exp': 2, + }}, ]) # pipeline builder - pbuild = builder.PipelineBuilder( + pbuild = PipelineBuilder( bsfs.Namespace('http://example.com/me/'), # not actually used rbuild, ebuild, @@ -61,6 +67,9 @@ def main(argv): # show predicates for pred in pipeline.schema.predicates(): print(pred.uri) + elif args.what == 'schema': + # show schema + print(bsfs.schema.to_string(pipeline.schema)) else: # args.what is already checked by argparse raise errors.UnreachableError() diff --git a/bsie/base/__init__.py b/bsie/base/__init__.py deleted file mode 100644 index 0d362cd..0000000 --- a/bsie/base/__init__.py +++ /dev/null @@ -1,24 +0,0 @@ -"""The base module defines the BSIE interfaces. - -You'll mostly find abstract classes here. - -Part of the bsie module. -A copy of the license is provided with the project. -Author: Matthias Baumgartner, 2022 -""" -# imports -import typing - -# inner-module imports -from . import errors -from .extractor import Extractor -from .reader import Reader - -# exports -__all__: typing.Sequence[str] = ( - 'Extractor', - 'Reader', - 'errors', - ) - -## EOF ## diff --git a/bsie/extractor/__init__.py b/bsie/extractor/__init__.py index ef31343..5f385ee 100644 --- a/bsie/extractor/__init__.py +++ b/bsie/extractor/__init__.py @@ -6,10 +6,17 @@ 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 .base import Extractor +from .builder import ExtractorBuilder + # exports -__all__: typing.Sequence[str] = [] +__all__: typing.Sequence[str] = ( + 'Extractor', + 'ExtractorBuilder', + ) ## EOF ## diff --git a/bsie/base/extractor.py b/bsie/extractor/base.py index c44021b..7401244 100644 --- a/bsie/base/extractor.py +++ b/bsie/extractor/base.py @@ -4,7 +4,7 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import abc import typing @@ -31,13 +31,22 @@ SCHEMA_PREAMBLE = ''' prefix bsfs: <http://bsfs.ai/schema/> prefix bse: <http://bsfs.ai/schema/Entity#> + # default definitions + bsfs:Array rdfs:subClassOf bsfs:Literal . + bsfs:Number rdfs:subClassOf bsfs:Literal . + bsfs:Time rdfs:subClassOf bsfs:Literal . + bsfs:Feature rdfs:subClassOf bsfs:Array ; + bsfs:dimension "1"^^xsd:integer ; + bsfs:dtype bsfs:f16 ; + bsfs:distance bsfs:euclidean . + # essential nodes bsfs:Entity rdfs:subClassOf bsfs:Node . bsfs:File rdfs:subClassOf bsfs:Entity . # common definitions xsd:string rdfs:subClassOf bsfs:Literal . - xsd:integer rdfs:subClassOf bsfs:Literal . + xsd:integer rdfs:subClassOf bsfs:Number . ''' @@ -99,5 +108,6 @@ class Extractor(abc.ABC): principals: typing.Iterable[bsfs.schema.Predicate], ) -> typing.Iterator[typing.Tuple[node.Node, bsfs.schema.Predicate, typing.Any]]: """Return (node, predicate, value) triples.""" + # FIXME: type annotation could be more strict: value is Hashable ## EOF ## diff --git a/bsie/extractor/builder.py b/bsie/extractor/builder.py new file mode 100644 index 0000000..0fd3685 --- /dev/null +++ b/bsie/extractor/builder.py @@ -0,0 +1,77 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import typing + +# bsie imports +from bsie.utils import bsfs, errors, safe_load, unpack_qualified_name + +# inner-module imports +from . import base + +# exports +__all__: typing.Sequence[str] = ( + 'ExtractorBuilder', + ) + + +## code ## + +class ExtractorBuilder(): + """Build `bsie.base.Extractor instances. + + It is permissible to build multiple instances of the same extractor + (typically with different arguments), hence the ExtractorBuilder + receives a list of build specifications. Each specification is + a dict with a single key (extractor's qualified name) and a dict + to be used as keyword arguments. + Example: [{'bsie.extractor.generic.path.Path': {}}, ] + + """ + + # build specifications + _specs: typing.List[typing.Dict[str, typing.Dict[str, typing.Any]]] + + def __init__(self, specs: typing.List[typing.Dict[str, typing.Dict[str, typing.Any]]]): + self._specs = specs + + def __iter__(self) -> typing.Iterator[int]: + """Iterate over extractor specifications.""" + return iter(range(len(self._specs))) + + def build(self, index: int) -> base.Extractor: + """Return an instance of the n'th extractor (n=*index*).""" + # get build instructions + specs = self._specs[index] + + # check specs structure. expecting[{name: {kwargs}}] + if not isinstance(specs, dict): + raise TypeError(f'expected a dict, found {bsfs.typename(specs)}') + if len(specs) != 1: + raise TypeError(f'expected a dict of length one, found {len(specs)}') + + # get name and args from specs + name = next(iter(specs.keys())) + kwargs = specs[name] + + # check kwargs structure + if not isinstance(kwargs, dict): + raise TypeError(f'expected a dict, found {bsfs.typename(kwargs)}') + + # check name and get module/class components + module_name, class_name = unpack_qualified_name(name) + + # import extractor class + cls = safe_load(module_name, class_name) + + try: # build and return instance + return cls(**kwargs) + + except Exception as err: + raise errors.BuilderError(f'failed to build extractor {name} due to {bsfs.typename(err)}: {err}') from err + +## EOF ## diff --git a/bsie/extractor/generic/__init__.py b/bsie/extractor/generic/__init__.py index 0cb7e7f..4783949 100644 --- a/bsie/extractor/generic/__init__.py +++ b/bsie/extractor/generic/__init__.py @@ -7,7 +7,7 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import typing # exports diff --git a/bsie/extractor/generic/constant.py b/bsie/extractor/generic/constant.py index 11384e6..938e20c 100644 --- a/bsie/extractor/generic/constant.py +++ b/bsie/extractor/generic/constant.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.base import extractor from bsie.utils import bsfs, node +# inner-module imports +from .. import base + # exports __all__: typing.Sequence[str] = ( 'Constant', @@ -19,7 +21,7 @@ __all__: typing.Sequence[str] = ( ## code ## -class Constant(extractor.Extractor): +class Constant(base.Extractor): """Extract information from file's path.""" CONTENT_READER = None @@ -32,7 +34,7 @@ class Constant(extractor.Extractor): schema: str, tuples: typing.Iterable[typing.Tuple[bsfs.URI, typing.Any]], ): - super().__init__(bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + schema)) + super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + schema)) # NOTE: Raises a KeyError if the predicate is not part of the schema self._tuples = tuple((self.schema.predicate(p_uri), value) for p_uri, value in tuples) # TODO: use schema instance for value checking diff --git a/bsie/extractor/generic/path.py b/bsie/extractor/generic/path.py index 7018e12..c984515 100644 --- a/bsie/extractor/generic/path.py +++ b/bsie/extractor/generic/path.py @@ -4,12 +4,12 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import os import typing # bsie imports -from bsie.base import extractor +from bsie.extractor import base from bsie.utils import bsfs, node, ns # exports @@ -20,7 +20,7 @@ __all__: typing.Sequence[str] = ( ## code ## -class Path(extractor.Extractor): +class Path(base.Extractor): """Extract information from file's path.""" CONTENT_READER = 'bsie.reader.path.Path' @@ -29,7 +29,7 @@ class Path(extractor.Extractor): _callmap: typing.Dict[bsfs.schema.Predicate, typing.Callable[[str], typing.Any]] def __init__(self): - super().__init__(bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + ''' + super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + ''' bse:filename rdfs:subClassOf bsfs:Predicate ; rdfs:domain bsfs:File ; rdfs:range xsd:string ; diff --git a/bsie/extractor/generic/stat.py b/bsie/extractor/generic/stat.py index 0b9ce29..9394456 100644 --- a/bsie/extractor/generic/stat.py +++ b/bsie/extractor/generic/stat.py @@ -4,14 +4,16 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import os import typing # bsie imports -from bsie.base import extractor from bsie.utils import bsfs, node, ns +# inner-module imports +from .. import base + # exports __all__: typing.Sequence[str] = ( 'Stat', @@ -20,7 +22,7 @@ __all__: typing.Sequence[str] = ( ## code ## -class Stat(extractor.Extractor): +class Stat(base.Extractor): """Extract information from the file system.""" CONTENT_READER = 'bsie.reader.stat.Stat' @@ -29,7 +31,7 @@ class Stat(extractor.Extractor): _callmap: typing.Dict[bsfs.schema.Predicate, typing.Callable[[os.stat_result], typing.Any]] def __init__(self): - super().__init__(bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + ''' + super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + ''' bse:filesize rdfs:subClassOf bsfs:Predicate ; rdfs:domain bsfs:File ; rdfs:range xsd:integer ; diff --git a/bsie/extractor/image/__init__.py b/bsie/extractor/image/__init__.py new file mode 100644 index 0000000..75b118d --- /dev/null +++ b/bsie/extractor/image/__init__.py @@ -0,0 +1,13 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import typing + +# exports +__all__: typing.Sequence[str] = [] + +## EOF ## diff --git a/bsie/extractor/image/colors_spatial.py b/bsie/extractor/image/colors_spatial.py new file mode 100644 index 0000000..ce5b9f2 --- /dev/null +++ b/bsie/extractor/image/colors_spatial.py @@ -0,0 +1,154 @@ +"""Spatial color features. + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import typing + +# external imports +import PIL.Image +import numpy as np + +# bsie imports +from bsie.utils import bsfs, node, ns + +# inner-module imports +from .. import base + +# constants +FEATURE_NAME = ns.bsf + 'ColorsSpatial' +PREDICATE_NAME = ns.bse + 'colors_spatial' + +# exports +__all__: typing.Sequence[str] = ( + 'ColorsSpatial', + ) + + +## code ## + +class ColorsSpatial(base.Extractor): + """Determine dominant colors of subregions in the image. + + Computes the domiant color of increasingly smaller subregions of the image. + """ + + CONTENT_READER = 'bsie.reader.image.Image' + + # Initial subregion width. + width: int + + # Initial subregion height. + height: int + + # Decrement exponent. + exp: float + + # Principal predicate's URI. + _predicate_name: bsfs.URI + + def __init__( + self, + width: int = 32, + height: int = 32, + exp: float = 4., + ): + # instance identifier + uuid = bsfs.uuid.UCID.from_dict({ + 'width': width, + 'height': height, + 'exp': exp, + }) + # determine symbol names + instance_name = FEATURE_NAME[uuid] + predicate_name = PREDICATE_NAME[uuid] + # get vector dimension + dimension = self.dimension(width, height, exp) + # initialize parent with the schema + super().__init__(bsfs.schema.from_string(base.SCHEMA_PREAMBLE + f''' + <{FEATURE_NAME}> rdfs:subClassOf bsfs:Feature ; + # annotations + rdfs:label "Spatially dominant colors"^^xsd:string ; + schema:description "Domiant colors of subregions in an image."^^xsd:string ; + bsfs:dtype xsd:integer . + + <{instance_name}> rdfs:subClassOf <{FEATURE_NAME}> ; + bsfs:dimension "{dimension}"^^xsd:integer ; + # annotations + <{FEATURE_NAME}/args#width> "{width}"^^xsd:integer ; + <{FEATURE_NAME}/args#height> "{height}"^^xsd:integer ; + <{FEATURE_NAME}/args#exp> "{exp}"^^xsd:float . + + <{predicate_name}> rdfs:subClassOf bsfs:Predicate ; + rdfs:domain bsfs:File ; + rdfs:range <{instance_name}> ; + bsfs:unique "true"^^xsd:boolean . + + ''')) + # assign extra members + self.width = width + self.height = height + self.exp = exp + self._predicate_name = predicate_name + + def __repr__(self) -> str: + return f'{bsfs.typename(self)}({self.width}, {self.height}, {self.exp})' + + def __eq__(self, other: typing.Any) -> bool: + return super().__eq__(other) \ + and self.width == other.width \ + and self.height == other.height \ + and self.exp == other.exp + + def __hash__(self) -> int: + return hash((super().__hash__(), self.width, self.height, self.exp)) + + @staticmethod + def dimension(width: int, height: int, exp: float) -> int: + """Return the feature vector dimension.""" + # FIXME: replace with a proper formula + dim = 0 + while width >= 1 and height >= 1: + dim += width * height + width = np.floor(width / exp) + height = np.floor(height / exp) + dim *= 3 # per band + return int(dim) + + def extract( + self, + subject: node.Node, + content: PIL.Image, + principals: typing.Iterable[bsfs.schema.Predicate], + ) -> typing.Iterator[typing.Tuple[node.Node, bsfs.schema.Predicate, typing.Any]]: + # check principals + if self.schema.predicate(self._predicate_name) not in principals: + # nothing to do; abort + return + + # convert to HSV + content = content.convert('HSV') + + # get dimensions + width, height = self.width, self.height + num_bands = len(content.getbands()) # it's three since we converted to HSV before + + features = [] + while width >= 1 and height >= 1: + # downsample + img = content.resize((width, height), resample=PIL.Image.Resampling.BOX) + # feature vector + features.append( + np.array(img.getdata()).reshape((width * height, num_bands))) + # iterate + width = int(np.floor(width / self.exp)) + height = int(np.floor(height / self.exp)) + + # combine bands and convert features to tuple + value = tuple(np.vstack(features).reshape(-1)) + # return triple with feature vector as value + yield subject, self.schema.predicate(self._predicate_name), value + +## EOF ## 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/tools/pipeline.py b/bsie/lib/pipeline.py index 20e8ddf..44685ba 100644 --- a/bsie/tools/pipeline.py +++ b/bsie/lib/pipeline.py @@ -4,14 +4,15 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports from collections import defaultdict import logging import typing # bsie imports -from bsie import base -from bsie.utils import bsfs, node, ns +from bsie.extractor import Extractor +from bsie.reader import Reader +from bsie.utils import bsfs, errors, node, ns # exports __all__: typing.Sequence[str] = ( @@ -43,12 +44,12 @@ class Pipeline(): _prefix: bsfs.Namespace # extractor -> reader mapping - _ext2rdr: typing.Dict[base.extractor.Extractor, typing.Optional[base.reader.Reader]] + _ext2rdr: typing.Dict[Extractor, typing.Optional[Reader]] def __init__( self, prefix: bsfs.Namespace, - ext2rdr: typing.Dict[base.extractor.Extractor, typing.Optional[base.reader.Reader]] + ext2rdr: typing.Dict[Extractor, typing.Optional[Reader]] ): # store core members self._prefix = prefix + FILE_PREFIX @@ -124,6 +125,7 @@ class Pipeline(): 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: @@ -132,11 +134,16 @@ class Pipeline(): for subject, pred, value in ext.extract(subject, content, principals): yield subject, pred, value - except base.errors.ExtractorError as err: + except errors.ExtractorError as err: # critical extractor failure. logger.error('%s failed to extract triples from content: %s', ext, err) - except base.errors.ReaderError as 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) diff --git a/bsie/reader/__init__.py b/bsie/reader/__init__.py index a45f22b..4163d1c 100644 --- a/bsie/reader/__init__.py +++ b/bsie/reader/__init__.py @@ -15,5 +15,18 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ +# standard imports +import typing +# inner-module imports +from .base import Reader +from .builder import ReaderBuilder + +# exports +__all__: typing.Sequence[str] = ( + 'Reader', + 'ReaderBuilder', + ) + +## EOF ## ## EOF ## diff --git a/bsie/base/reader.py b/bsie/reader/base.py index cbabd36..099a327 100644 --- a/bsie/base/reader.py +++ b/bsie/reader/base.py @@ -8,7 +8,7 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import abc import typing @@ -39,7 +39,7 @@ class Reader(abc.ABC): return hash(type(self)) @abc.abstractmethod - def __call__(self, path: bsfs.URI) -> typing.Any: + def __call__(self, path: str) -> typing.Any: """Return some content of the file at *path*. Raises a `ReaderError` if the reader cannot make sense of the file format. """ diff --git a/bsie/reader/builder.py b/bsie/reader/builder.py new file mode 100644 index 0000000..8699e75 --- /dev/null +++ b/bsie/reader/builder.py @@ -0,0 +1,78 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import typing + +# bsie imports +from bsie.utils import bsfs, errors, safe_load, unpack_qualified_name + +# inner-module imports +from . import base + +# exports +__all__: typing.Sequence[str] = ( + 'ReaderBuilder', + ) + + +## code ## + +class ReaderBuilder(): + """Build `bsie.base.Reader` instances. + + Readers are defined via their qualified class name + (e.g., bsie.reader.path.Path) and optional keyword + arguments that are passed to the constructor via + the *kwargs* argument (name as key, kwargs as value). + The ReaderBuilder keeps a cache of previously built + reader instances, as they are anyway built with + identical keyword arguments. + + """ + + # keyword arguments + _kwargs: typing.Dict[str, typing.Dict[str, typing.Any]] + + # cached readers + _cache: typing.Dict[str, base.Reader] + + def __init__( + self, + kwargs: typing.Optional[typing.Dict[str, typing.Dict[str, typing.Any]]] = None): + if kwargs is None: + kwargs = {} + self._kwargs = kwargs + self._cache = {} + + def build(self, name: str) -> base.Reader: + """Return an instance for the qualified class name.""" + # return cached instance + if name in self._cache: + return self._cache[name] + + # check name and get module/class components + module_name, class_name = unpack_qualified_name(name) + + # import reader class + cls = safe_load(module_name, class_name) + + # get kwargs + kwargs = self._kwargs.get(name, {}) + if not isinstance(kwargs, dict): + raise TypeError(f'expected a kwargs dict, found {bsfs.typename(kwargs)}') + + try: # build, cache, and return instance + obj = cls(**kwargs) + # cache instance + self._cache[name] = obj + # return instance + return obj + + except Exception as err: + raise errors.BuilderError(f'failed to build reader {name} due to {bsfs.typename(err)}: {err}') from err + +## EOF ## diff --git a/bsie/reader/chain.py b/bsie/reader/chain.py new file mode 100644 index 0000000..5e9e0d5 --- /dev/null +++ b/bsie/reader/chain.py @@ -0,0 +1,88 @@ +""" + +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.utils import bsfs, errors + +# inner-module imports +from . import base +from . import builder + +# exports +__all__: typing.Sequence[str] = ( + 'ReaderChain', + ) + + +## code ## + +logger = logging.getLogger(__name__) + +# Content type. +T_CONTENT = typing.TypeVar('T_CONTENT') # pylint: disable=invalid-name + +class ReaderChain(base.Reader, typing.Generic[T_CONTENT]): + """Read an image.""" + + # sub-readers for specific file formats. + _children: typing.Tuple[base.Reader, ...] + + def __init__( + self, + subreader_names: typing.Iterable[str], + cfg: typing.Optional[typing.Any] = None, + ): + rbuild = builder.ReaderBuilder(cfg) + children = [] + for name in subreader_names: + try: + # build sub-reader + children.append(rbuild.build(name)) + except (ValueError, + TypeError, + errors.LoaderError, + errors.BuilderError) as err: + # failed to build a child; skip and notify + logger.warning('failed to load reader: %s', err) + + if len(children) == 0: + logger.warning('%s failed to load any sub-readers.', bsfs.typename(self)) + + # copy children to member + self._children = tuple(children) + + def __str__(self) -> str: + substr = ', '.join(str(child) for child in self._children) + return f'{bsfs.typename(self)}({substr})' + + def __repr__(self) -> str: + return f'{bsfs.typename(self)}({self._children})' + + def __eq__(self, other: typing.Any) -> bool: + return super().__eq__(other) \ + and self._children == other._children + + def __hash__(self) -> int: + return hash((super().__hash__(), self._children)) + + def __call__(self, path: str) -> T_CONTENT: + raise_error = errors.UnsupportedFileFormatError + for child in self._children: + try: + return child(path) + except errors.UnsupportedFileFormatError: + pass + except errors.ReaderError: + # child cannot read the file, skip. + raise_error = errors.ReaderError # type: ignore [assignment] # mypy is confused + + raise raise_error(path) + +## EOF ## diff --git a/bsie/reader/image/__init__.py b/bsie/reader/image/__init__.py new file mode 100644 index 0000000..1f290b5 --- /dev/null +++ b/bsie/reader/image/__init__.py @@ -0,0 +1,37 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import typing + +# external imports +import PIL.Image + +# inner-module imports +from .. import chain + +# constants +_FILE_FORMAT_READERS: typing.Sequence[str] = ( + __package__ + '._raw.RawImage', + __package__ + '._pillow.PillowImage', + ) + +# exports +__all__: typing.Sequence[str] = ( + 'Image', + ) + + +## code ## + +# FIXME: Check if PIL.Image or PIL.Image.Image, or if version-dependent +class Image(chain.ReaderChain[PIL.Image.Image]): # pylint: disable=too-few-public-methods + """Read an image file.""" + + def __init__(self, cfg: typing.Optional[typing.Any] = None): + super().__init__(_FILE_FORMAT_READERS, cfg) + +## EOF ## diff --git a/bsie/reader/image/_pillow.py b/bsie/reader/image/_pillow.py new file mode 100644 index 0000000..3144509 --- /dev/null +++ b/bsie/reader/image/_pillow.py @@ -0,0 +1,39 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import typing + +# external imports +import PIL.Image + +# bsie imports +from bsie.utils import errors + +# inner-module imports +from .. import base + +# exports +__all__: typing.Sequence[str] = ( + 'PillowImage', + ) + + +## code ## + +class PillowImage(base.Reader): + """Use PIL to read content of a variety of image file types.""" + + def __call__(self, path: str) -> PIL.Image: + try: + # open file with PIL + return PIL.Image.open(path) + except PIL.UnidentifiedImageError as err: + raise errors.UnsupportedFileFormatError(path) from err + except IOError as err: + raise errors.ReaderError(path) from err + +# EOF ## diff --git a/bsie/reader/image/_raw.py b/bsie/reader/image/_raw.py new file mode 100644 index 0000000..cd60453 --- /dev/null +++ b/bsie/reader/image/_raw.py @@ -0,0 +1,61 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import typing + +# external imports +import PIL.Image +import rawpy + +# bsie imports +from bsie.utils import errors, filematcher + +# inner-module imports +from .. import base + +# constants +MATCH_RULE = 'mime={image/x-nikon-nef} | extension={nef}' + +# exports +__all__: typing.Sequence[str] = ( + 'RawImage', + ) + + +## code ## + +class RawImage(base.Reader): + """Use rawpy to read content of raw image file types.""" + + # file matcher + match: filematcher.Matcher + + # additional kwargs to rawpy's postprocess + rawpy_kwargs: typing.Dict[str, typing.Any] + + def __init__(self, **rawpy_kwargs): + match_rule = rawpy_kwargs.pop('file_match_rule', MATCH_RULE) + self._match = filematcher.parse(match_rule) + self._rawpy_kwargs = rawpy_kwargs + + def __call__(self, path: str) -> PIL.Image: + # perform quick checks first + if not self._match(path): + raise errors.UnsupportedFileFormatError(path) + + try: + # open file with rawpy + ary = rawpy.imread(path).postprocess(**self._rawpy_kwargs) + # convert to PIL.Image + return PIL.Image.fromarray(ary) + except (rawpy.LibRawFatalError, # pylint: disable=no-member # pylint doesn't find the errors + rawpy.NotSupportedError, # pylint: disable=no-member + rawpy.LibRawNonFatalError, # pylint: disable=no-member + ) as err: + raise errors.ReaderError(path) from err + +## EOF ## diff --git a/bsie/reader/path.py b/bsie/reader/path.py index d60f187..1ca05a0 100644 --- a/bsie/reader/path.py +++ b/bsie/reader/path.py @@ -4,11 +4,11 @@ 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.base import reader +# inner-module imports +from . import base # exports __all__: typing.Sequence[str] = ( @@ -18,7 +18,7 @@ __all__: typing.Sequence[str] = ( ## code ## -class Path(reader.Reader): +class Path(base.Reader): """Return the path.""" def __call__(self, path: str) -> str: diff --git a/bsie/reader/stat.py b/bsie/reader/stat.py index fc5fb24..706dc47 100644 --- a/bsie/reader/stat.py +++ b/bsie/reader/stat.py @@ -4,12 +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 os import typing # bsie imports -from bsie.base import errors, reader +from bsie.utils import errors + +# inner-module imports +from . import base # exports __all__: typing.Sequence[str] = ( @@ -19,7 +22,7 @@ __all__: typing.Sequence[str] = ( ## code ## -class Stat(reader.Reader): +class Stat(base.Reader): """Read and return the filesystem's stat infos.""" def __call__(self, path: str) -> os.stat_result: diff --git a/bsie/tools/builder.py b/bsie/tools/builder.py deleted file mode 100644 index 190d9bf..0000000 --- a/bsie/tools/builder.py +++ /dev/null @@ -1,226 +0,0 @@ -""" - -Part of the bsie module. -A copy of the license is provided with the project. -Author: Matthias Baumgartner, 2022 -""" -# imports -import importlib -import logging -import typing - -# bsie imports -from bsie import base -from bsie.base import errors -from bsie.utils import bsfs - -# inner-module imports -from . import pipeline - -# exports -__all__: typing.Sequence[str] = ( - 'ExtractorBuilder', - 'PipelineBuilder', - 'ReaderBuilder', - ) - - -## code ## - -logger = logging.getLogger(__name__) - -def _safe_load(module_name: str, class_name: str): - """Get a class from a module. Raise BuilderError if anything goes wrong.""" - try: - # load the module - module = importlib.import_module(module_name) - except Exception as err: - # cannot import module - raise errors.LoaderError(f'cannot load module {module_name}') from err - - try: - # get the class from the module - cls = getattr(module, class_name) - except Exception as err: - # cannot find the class - raise errors.LoaderError(f'cannot load class {class_name} from module {module_name}') from err - - return cls - - -def _unpack_name(name): - """Split a name into its module and class component (dot-separated).""" - if not isinstance(name, str): - raise TypeError(name) - if '.' not in name: - raise ValueError('name must be a qualified class name.') - module_name, class_name = name[:name.rfind('.')], name[name.rfind('.')+1:] - if module_name == '': - raise ValueError('name must be a qualified class name.') - return module_name, class_name - - -class ReaderBuilder(): - """Build `bsie.base.Reader` instances. - - Readers are defined via their qualified class name - (e.g., bsie.reader.path.Path) and optional keyword - arguments that are passed to the constructor via - the *kwargs* argument (name as key, kwargs as value). - The ReaderBuilder keeps a cache of previously built - reader instances, as they are anyway built with - identical keyword arguments. - - """ - - # keyword arguments - _kwargs: typing.Dict[str, typing.Dict[str, typing.Any]] - - # cached readers - _cache: typing.Dict[str, base.Reader] - - def __init__(self, kwargs: typing.Dict[str, typing.Dict[str, typing.Any]]): - self._kwargs = kwargs - self._cache = {} - - def build(self, name: str) -> base.Reader: - """Return an instance for the qualified class name.""" - # return cached instance - if name in self._cache: - return self._cache[name] - - # check name and get module/class components - module_name, class_name = _unpack_name(name) - - # import reader class - cls = _safe_load(module_name, class_name) - - # get kwargs - kwargs = self._kwargs.get(name, {}) - if not isinstance(kwargs, dict): - raise TypeError(f'expected a kwargs dict, found {bsfs.typename(kwargs)}') - - try: # build, cache, and return instance - obj = cls(**kwargs) - # cache instance - self._cache[name] = obj - # return instance - return obj - - except Exception as err: - raise errors.BuilderError(f'failed to build reader {name} due to {bsfs.typename(err)}: {err}') from err - - -class ExtractorBuilder(): - """Build `bsie.base.Extractor instances. - - It is permissible to build multiple instances of the same extractor - (typically with different arguments), hence the ExtractorBuilder - receives a list of build specifications. Each specification is - a dict with a single key (extractor's qualified name) and a dict - to be used as keyword arguments. - Example: [{'bsie.extractor.generic.path.Path': {}}, ] - - """ - - # build specifications - _specs: typing.List[typing.Dict[str, typing.Dict[str, typing.Any]]] - - def __init__(self, specs: typing.List[typing.Dict[str, typing.Dict[str, typing.Any]]]): - self._specs = specs - - def __iter__(self) -> typing.Iterator[int]: - """Iterate over extractor specifications.""" - return iter(range(len(self._specs))) - - def build(self, index: int) -> base.Extractor: - """Return an instance of the n'th extractor (n=*index*).""" - # get build instructions - specs = self._specs[index] - - # check specs structure. expecting[{name: {kwargs}}] - if not isinstance(specs, dict): - raise TypeError(f'expected a dict, found {bsfs.typename(specs)}') - if len(specs) != 1: - raise TypeError(f'expected a dict of length one, found {len(specs)}') - - # get name and args from specs - name = next(iter(specs.keys())) - kwargs = specs[name] - - # check kwargs structure - if not isinstance(kwargs, dict): - raise TypeError(f'expected a dict, found {bsfs.typename(kwargs)}') - - # check name and get module/class components - module_name, class_name = _unpack_name(name) - - # import extractor class - cls = _safe_load(module_name, class_name) - - try: # build and return instance - return cls(**kwargs) - - except Exception as err: - raise errors.BuilderError(f'failed to build extractor {name} due to {bsfs.typename(err)}: {err}') from err - - -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/utils/__init__.py b/bsie/utils/__init__.py index bd22236..9cb60ed 100644 --- a/bsie/utils/__init__.py +++ b/bsie/utils/__init__.py @@ -4,19 +4,24 @@ 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 . import bsfs +from . import filematcher from . import namespaces as ns from . import node +from .loading import safe_load, unpack_qualified_name # exports __all__: typing.Sequence[str] = ( 'bsfs', + 'filematcher', 'node', 'ns', + 'safe_load', + 'unpack_qualified_name', ) ## EOF ## diff --git a/bsie/utils/bsfs.py b/bsie/utils/bsfs.py index 0b88479..ef5db31 100644 --- a/bsie/utils/bsfs.py +++ b/bsie/utils/bsfs.py @@ -4,7 +4,7 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import typing # bsfs imports diff --git a/bsie/base/errors.py b/bsie/utils/errors.py index dc3c30e..8133cd4 100644 --- a/bsie/base/errors.py +++ b/bsie/utils/errors.py @@ -4,7 +4,7 @@ Part of the bsie module. A copy of the license is provided with the project. Author: Matthias Baumgartner, 2022 """ -# imports +# standard imports import typing # exports @@ -39,4 +39,10 @@ class ProgrammingError(_BSIEError): class UnreachableError(ProgrammingError): """Bravo, you've reached a point in code that should logically not be reachable.""" +class ParserError(_BSIEError): + """Failed to parse due to invalid syntax or structures.""" + +class UnsupportedFileFormatError(ReaderError): + """Failed to read a file format.""" + ## EOF ## diff --git a/bsie/tools/__init__.py b/bsie/utils/filematcher/__init__.py index 803c321..1e23e4e 100644 --- a/bsie/tools/__init__.py +++ b/bsie/utils/filematcher/__init__.py @@ -4,17 +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 . import builder -from .pipeline import Pipeline +from .matcher import Matcher +from .parser import parse # exports __all__: typing.Sequence[str] = ( - 'builder', - 'Pipeline', + 'Matcher', + 'parse', ) ## EOF ## diff --git a/bsie/utils/filematcher/matcher.py b/bsie/utils/filematcher/matcher.py new file mode 100644 index 0000000..a279a4b --- /dev/null +++ b/bsie/utils/filematcher/matcher.py @@ -0,0 +1,179 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2021 +""" +# standard imports +from collections.abc import Callable, Collection, Hashable +import abc +import os +import typing + +# external imports +import magic + +# exports +__all__: typing.Sequence[str] = [] + + +## code ## + +# abstract nodes + +class Matcher(abc.ABC, Hashable, Callable, Collection): # type: ignore [misc] # Invalid base class Callable + """Matcher node base class.""" + + # child expressions or terminals + _childs: typing.Set[typing.Any] + + def __init__(self, *childs: typing.Any): + if len(childs) == 1 and isinstance(childs[0], (list, tuple, set)): + self._childs = set(childs[0]) + else: + self._childs = set(childs) + + def __contains__(self, needle: typing.Any) -> bool: + return needle in self._childs + + def __iter__(self) -> typing.Iterator[typing.Any]: + return iter(self._childs) + + def __len__(self) -> int: + return len(self._childs) + + def __repr__(self) -> str: + return f'{type(self).__name__}({self._childs})' + + def __hash__(self) -> int: + return hash((type(self), tuple(set(self._childs)))) + + def __eq__(self, other: typing.Any) -> bool: + return isinstance(other, type(self)) \ + and self._childs == other._childs + + @abc.abstractmethod + def __call__(self, path: str) -> bool: # pylint: disable=arguments-differ + """Check if *path* satisfies the conditions set by the Matcher instance.""" + +class NOT(Matcher): + """Invert a matcher result.""" + def __init__(self, expr: Matcher): + super().__init__(expr) + def __call__(self, path: str) -> bool: + return not next(iter(self._childs))(path) + +# aggregate nodes + +class Aggregate(Matcher): # pylint: disable=too-few-public-methods # Yeah, it's an interface... + """Aggregation function base class (And, Or).""" + +class And(Aggregate): + """Accept only if all conditions are satisfied.""" + def __call__(self, path: str) -> bool: + for itm in self: + if not itm(path): + return False + return True + +class Or(Aggregate): + """Accept only if at least one condition is satisfied.""" + def __call__(self, path: str) -> bool: + for itm in self: + if itm(path): + return True + return False + + +# criteria nodes + +class Criterion(Matcher): + """Criterion base class. Limits acceptance to certain values.""" + def accepted(self) -> typing.Set[typing.Any]: + """Return a set of accepted values.""" + return self._childs + +# criteria w/o value (valueless) + +class Any(Criterion): + """Accepts anything.""" + def __call__(self, path: str) -> bool: + return True + +class Nothing(Criterion): + """Accepts nothing.""" + def __call__(self, path: str) -> bool: + return False + +class Exists(Criterion): + """Filters by existence.""" + def __call__(self, path: str) -> bool: + return os.path.exists(path) + +class IsFile(Criterion): + """Checks if the path is a regular file.""" + def __call__(self, path: str) -> bool: + return os.path.isfile(path) + +class IsDir(Criterion): + """Checks if the path is a directory.""" + def __call__(self, path: str) -> bool: + return os.path.isdir(path) + +class IsLink(Criterion): + """Checks if the path is a link.""" + def __call__(self, path: str) -> bool: + return os.path.islink(path) + +class IsAbs(Criterion): + """Checks if the path is an absolute path.""" + def __call__(self, path: str) -> bool: + return os.path.isabs(path) + +class IsRel(Criterion): + """Checks if the path is a relative path.""" + def __call__(self, path: str) -> bool: + return not os.path.isabs(path) + +class IsMount(Criterion): + """Checks if the path is a mount point.""" + def __call__(self, path: str) -> bool: + return os.path.ismount(path) + +class IsEmpty(Criterion): + """Checks if the path is an empty file.""" + def __call__(self, path: str) -> bool: + return os.path.exists(path) and os.stat(path).st_size == 0 + +class IsReadable(Criterion): + """Checks if the path is readable.""" + def __call__(self, path: str) -> bool: + return os.path.exists(path) and os.access(path, os.R_OK) + +class IsWritable(Criterion): + """Checks if the path is writable.""" + def __call__(self, path: str) -> bool: + return os.path.exists(path) and os.access(path, os.W_OK) + +class IsExecutable(Criterion): + """Checks if the path is executable.""" + def __call__(self, path: str) -> bool: + return os.path.exists(path) and os.access(path, os.X_OK) + +# criteria w/ value + +class Extension(Criterion): + """Filters by file extension (without the dot).""" + def __call__(self, path: str) -> bool: + _, ext = os.path.splitext(path) + return ext[1:] in self.accepted() + +class Mime(Criterion): + """Filters by mime type.""" + def __call__(self, path: str) -> bool: + try: + return magic.from_file(path, mime=True).lower() in self.accepted() + except FileNotFoundError: + return False + +## EOF ## diff --git a/bsie/utils/filematcher/parser.py b/bsie/utils/filematcher/parser.py new file mode 100644 index 0000000..2f82875 --- /dev/null +++ b/bsie/utils/filematcher/parser.py @@ -0,0 +1,146 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2021 +""" +# standard imports +import typing + +# external imports +import pyparsing +from pyparsing import printables, alphas8bit, punc8bit, QuotedString, Word, \ + delimitedList, Or, CaselessKeyword, Group, oneOf, Optional + +# inner-module imports +from . import matcher +from .. import errors + +# exports +__all__: typing.Sequence[str] = ( + 'parse', + ) + + +## code ## + +class FileMatcherParser(): + """ + EXPR := RULES | RULES "|" RULES + RULESET := RULE | RULE, RULE + RULE := CRITERION OP VALUE | CRITERION OP {VALUES} | VALUELESS + OP := != | = + VALUES := VALUE | VALUE, VALUE + VALUE := [word] + CRITERION := mime | extension | ... + """ + + # criteria matcher nodes w/ arguments + _CRITERIA: typing.Dict[str, typing.Type[matcher.Matcher]] = { + 'extension': matcher.Extension, + 'mime': matcher.Mime, + } + + # criteria matcher nodes w/o arguments + _VALUELESS: typing.Dict[str, typing.Type[matcher.Matcher]] = { + 'any': matcher.Any, + 'nothing': matcher.Nothing, + 'exists': matcher.Exists, + 'isfile': matcher.IsFile, + 'isdir': matcher.IsDir, + 'islink': matcher.IsLink, + 'isabs': matcher.IsAbs, + 'isrel': matcher.IsRel, + 'ismount': matcher.IsMount, + 'emtpy': matcher.IsEmpty, + 'readable': matcher.IsReadable, + 'writable': matcher.IsWritable, + 'executable': matcher.IsExecutable, + } + + # pyparsing parser instance. + _parser: pyparsing.ParseExpression + + def __init__(self): + # build the parser + # VALUE := [word] + alphabet = (printables + alphas8bit + punc8bit).translate(str.maketrans('', '', ',{}|=')) + value = QuotedString(quoteChar='"', escChar='\\') ^ Word(alphabet) + # CRITERION := mime | extension | ... + criterion = Or([CaselessKeyword(p) for p in self._CRITERIA]).setResultsName('criterion') + valueless = Or([CaselessKeyword(p) for p in self._VALUELESS]).setResultsName('criterion') + # VALUES := VALUE | VALUE, VALUE + values = delimitedList(value, delim=',').setResultsName('value') + # OP := '=' | '!=' + eqop = oneOf('= !=').setResultsName('op') + # RULE := CRITERION OP VALUE | CRITERION OP {VALUES} | VALUELESS + rule_none = Group(Optional('!').setResultsName('op') + valueless).setResultsName('rule_none') + rule_one = Group(criterion + eqop + value.setResultsName('value')).setResultsName('rule_one') + rule_few = Group(criterion + eqop + '{' + values + '}').setResultsName('rule_few') + # RULESET := RULE | RULE, RULE + ruleset = Group(delimitedList(rule_none ^ rule_one ^ rule_few, delim=',')) + # EXPR := RULESET | RULESET \| RULESET + self._parser = delimitedList(ruleset, delim='|') + + def parse(self, query: str) -> matcher.Matcher: # pylint: disable=too-many-branches + """Build a file matcher from a rule definition.""" + # preprocess the query + query = query.strip() + + # empty query + if len(query) == 0: + return matcher.Any() + + try: + parsed = self._parser.parseString(query, parseAll=True) + except pyparsing.ParseException as err: + raise errors.ParserError(f'Cannot parse query {err}') + + # convert to Matcher + rules = [] + for exp in parsed: + tokens = [] + for rule in exp: + # fetch accepted values + if rule.getName() == 'rule_none': + accepted = [] + elif rule.getName() == 'rule_one': + accepted = [rule.value] + elif rule.getName() == 'rule_few': + accepted = list(rule.value) + else: # prevented by grammar + raise errors.UnreachableError('Invalid rule definition') + + # build criterion + if rule.criterion in self._VALUELESS: + cls = self._VALUELESS[rule.criterion] + if rule.op == '!': + tokens.append(matcher.NOT(cls())) + else: + tokens.append(cls()) + elif rule.criterion in self._CRITERIA: + cls = self._CRITERIA[rule.criterion] + if rule.op == '!=': + tokens.append(matcher.NOT(cls(accepted))) + else: + tokens.append(cls(accepted)) + else: # prevented by grammar + raise errors.UnreachableError(f'Invalid condition "{rule.criterion}"') + + # And-aggregate rules in one ruleset (if needed) + tokens = matcher.And(tokens) if len(tokens) > 1 else tokens[0] + rules.append(tokens) + + # Or-aggregate rulesets + expr = matcher.Or(rules) if len(rules) > 1 else rules[0] + + return expr + +# build default instance +file_match_parser = FileMatcherParser() + +def parse(query: str) -> matcher.Matcher: + """Shortcut for FileMatcherParser()(query).""" + return file_match_parser.parse(query) + +## EOF ## diff --git a/bsie/utils/loading.py b/bsie/utils/loading.py new file mode 100644 index 0000000..eb05c35 --- /dev/null +++ b/bsie/utils/loading.py @@ -0,0 +1,54 @@ +""" + +Part of the bsie module. +A copy of the license is provided with the project. +Author: Matthias Baumgartner, 2022 +""" +# standard imports +import importlib +import typing + +# inner-module imports +from . import errors + +# exports +__all__: typing.Sequence[str] = ( + 'safe_load', + 'unpack_qualified_name', + ) + + +## code ## + +def safe_load(module_name: str, class_name: str): + """Get a class from a module. Raise BuilderError if anything goes wrong.""" + try: + # load the module + module = importlib.import_module(module_name) + except Exception as err: + # cannot import module + raise errors.LoaderError(f'cannot load module {module_name}') from err + + try: + # get the class from the module + cls = getattr(module, class_name) + except Exception as err: + # cannot find the class + raise errors.LoaderError(f'cannot load class {class_name} from module {module_name}') from err + + return cls + + +def unpack_qualified_name(name): + """Split a name into its module and class component (dot-separated).""" + if not isinstance(name, str): + raise TypeError(name) + if '.' not in name: + raise ValueError('name must be a qualified class name.') + module_name, class_name = name[:name.rfind('.')], name[name.rfind('.')+1:] + if module_name == '': + raise ValueError('name must be a qualified class name.') + return module_name, class_name + + +## EOF ## diff --git a/bsie/utils/namespaces.py b/bsie/utils/namespaces.py index a29fc1b..393b436 100644 --- a/bsie/utils/namespaces.py +++ b/bsie/utils/namespaces.py @@ -4,7 +4,7 @@ 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 @@ -15,6 +15,7 @@ bse = _bsfs.Namespace('http://bsfs.ai/schema/Entity') bsfs = _bsfs.Namespace('http://bsfs.ai/schema', fsep='/') bsm = _bsfs.Namespace('http://bsfs.ai/schema/Meta') xsd = _bsfs.Namespace('http://www.w3.org/2001/XMLSchema') +bsf = _bsfs.Namespace('http://ie.bsfs.ai/schema/Feature') # export __all__: typing.Sequence[str] = ( diff --git a/bsie/utils/node.py b/bsie/utils/node.py index ecf39cd..91e4f37 100644 --- a/bsie/utils/node.py +++ b/bsie/utils/node.py @@ -4,7 +4,7 @@ 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 |